The Challenges of the New Spanish Multipartism: Government Formation Failure and the 2016 General Election

Abstract

The 2016 general election was the result of the incapacity of Spanish political parties to agree on the formation of a government following the previous parliamentary election six months earlier. The paper begins by examining how and why this process failed, focusing on institutional and contextual factors, especially those related to the emergence of Podemos and Ciudadanos in the Spanish political arena. Also explained are the negotiation strategies of the different parties, the electoral campaign and the results of the subsequent 2016 general election. These results show that, on the right, most vote changes went from Ciudadanos to the Popular Party while, in the case of the new pre-electoral coalition of UP, former Izquierda Unida voters were more prone to desert.

Keywords:

• Government formation
• polarisation
• party system change
• Podemos
• Ciudadanos
• PSOE
• PP

The 26 June (26J) 2016 Spanish general election was held because, for the first time in Spain, it had proved impossible to form a government. This failure was due to the totally new political environment created by the 2015 general election. The two traditional parties, the Partido Popular (Popular Party – PP), led by the Prime Minister (PM), Mariano Rajoy, and the PSOE (Partido Socialista Obrero Español – Spanish Socialist Workers' Party – ‘the Socialists’), led by Pedro Sánchez, together won only 50 per cent of the votes. Moreover, despite the PP being the first party, it only won 123 seats out of 350 (35.1 per cent), the narrowest majority of any winning party since 1977.

Two new parties challenged the traditional Spanish two-party system (Orriols & Cordero 2016). On the one hand, the main challenger was Podemos (We Can – Ps), led by Pablo Iglesias. This leftist anti-establishment party first emerged in the 2014 European Parliament election (Cordero & Montero 2015) and managed to forge pre-electoral coalitions with other parties in Catalonia (En Comú Podem), Valencia (Compromís Podemos Es el Moment) and Galicia (En Marea) which, together, obtained 69 seats and 20.6 per cent of the votes, only two points less than the PSOE (Fernández-Albertos 2015). On the other hand, Ciudadanos (Citizens – Cs) led by Albert Rivera, a centre-right party that emerged first in Catalonia but expanded throughout Spain, was able to gain 40 seats and 13.9 per cent of the votes (Rodríguez Teruel & Barrio 2015).

In contrast with previous decades, this new multiparty environment obliged the main national parties to reach agreements in order to form a government. Despite having a tradition of coalitions at local and regional levels, only absolute majority governments or minority governments with the external support of regional parties had ever been formed at the national level (Falcó & Verge 2013). However, it was not obvious that the new multiparty structure would result in political paralysis: old and new parties reached agreements in the regional and local arenas following the May 2015 regional and municipal elections (Rodón & Hierro 2016). In almost all regions, the PSOE reached investiture agreements with Podemos – or the regional parties – and the PP with Ciudadanos. Regional presidents of the PP and the PSOE were elected, despite both the new parties declining to enter governments and hold portfolios (Ramos & Simón 2015).

In contrast with the regional elections, in the 20 December (20D) 2015 general election neither the right-wing bloc (163 seats; PP and Ciudadanos) nor the left-wing bloc (161 seats; PSOE, Podemos and the traditional post-communist party, Izquierda Unida [United Left – IU]) obtained an absolute majority. Therefore, any government formation required the participation or abstention of parties from the other ideological bloc or of regional parties such as the Partido Nacionalista Vasco (Basque Nationalist Party – PNV), Coalición Canaria (Coalition for the Canary Islands – CC), Democràcia and Llibertat (Democracy and Freedom – DiL, the former regional Catalan conservative party Convergència i Unió [Convergence and Union – CiU]) and/or Esquerra Republicana de Catalunya (Republican Left of Catalonia – ERC). This was a completely new departure point for negotiations. Moreover, a failure in the process of the investiture vote would open up the possibility of new elections – as finally happened on 26 June.

This paper is structured as follows. The first section explains the procedure for forming a government at the national level and the constraints created by the results of the 2015 general election. Next, the negotiation process is explained and the behaviour of key actors is reviewed, with the resulting call for a new election on 26 June. The third section describes the electoral campaign, and contrasts it with the 2015 election and the impact of external factors on it. The last two sections focus on explaining the results of the 2016 general election and the formation of a new government.

Constraints on government formation after 20D

The process of government formation in Spain is regulated by the Constitution. According to article 99.1, after a general election, the king shall, after consultation with political groups, nominate a candidate for presidente del gobierno, the term for the Spanish PM. Spain has a weak bicameral system; only the lower chamber, the Congress, plays a role in investing the government.

The nomination of a candidate by the king requires the countersignature of the president of Congress, so the head of state is not free to designate any candidate but only the individual who has the support of Congress. Article 99.3 specifies that an absolute majority is required for the candidate to pass the investiture vote. If the candidate does not obtain this threshold, a new vote must take place within 48 hours and, in this case, the investiture only requires a simple majority (more yes votes than no votes). Moreover, the Spanish Constitution includes a provision to avoid political deadlock. According to article 99.5, if no candidate receives the support of Congress within two months of the first investiture vote, the king will call new elections.

According to the Constitution, Spain is a case of positive parliamentarianism. This means that the proposed candidate requires the explicit support of a majority of legislators before he or she can take office (Bergman et al. 2003). Scholars have suggested that positive investiture rules make the formation of minority governments more difficult. Parties not represented in the cabinet may be less disposed to support a government they are not part of, but can find it acceptable to allow a minority government if they do not have to openly support an investiture vote (Strøm 1990; De Winter 2003). Therefore, it can be argued that having a formal investiture vote increases the difficulties of forming a government, especially if a minority government is the preferred way to gain the support of various parties in Congress. Nevertheless, minority government has been the rule when the PP or the PSOE has not obtained 176 seats (50 per cent plus one of the 350 seats in the Congress). Other authors have argued that this in fact makes it more complex and lengthy to form a minority government after elections, but the empirical evidence is not conclusive (Golder 2010, p. 24).

Another institutional factor that can affect negotiations is that in Spain there is a ‘constructive’ vote of no confidence. This requires an absolute majority of seats for a new PM and is outlined in article 113 of the Spanish Constitution. Other systems, i.e. simple majority, can easily dismiss a prime minister from office (Louwerse 2014). However, the constructive system requires an absolute majority to agree on a common candidate; thus, in a fragmented and polarised context it is difficult to forge an alternative coalition, making the position of the already appointed PM quite safe. Finally, for those parties outside the cabinet, parliament can be a decisive arena for negotiating legislative coalitions and policies, especially in the case of minority governments (Strøm 1990). However, Spain has a weak parliamentary system, where the legislative committees are not independent, well funded or powerful, so, despite a minority government being a feasible possibility, the potential influence of other parties on policies may be rather limited once a government is in place.

Nevertheless, these three institutional features (positive parliamentarianism, constructive vote of no confidence, and weak parliament) have not changed in Spain since the restoration of democracy in 1977, so they alone cannot be considered sufficient conditions to complicate government formation. It is the interaction with two other dimensions that can make government formation a more challenging process: bargaining complexity and uncertainty.

The challenge of bargaining complexity

The first dimension is bargaining complexity, which comprises two key elements: number of parties in the legislature and ideological polarisation (Diermeier & van Roozendal 1998; Golder 2006; Strøm & Müller 1999). On the one hand, an increase in the effective number of parliamentary parties will increase the number of potential veto players involved in any bargaining process (Tsebelis 2002). On the other hand, ideological polarisation increases complexity because there are more potential policy domains in which there are disagreements among the parties (Martin & Vanberg 2003). It can be argued that the 2015 general elections involved a change in both dimensions.

As can be seen in Figure 1, the effective number of electoral parties (ENEP)¹ The effective number of parliamentary parties replaces p with the share of seats obtained by each party. (Laakso & Taagepera 1979), a traditional measure of party system fragmentation, rose significantly in Congress after the 2015 election. In 2008, the ENEP was 2.8, its lowest level in any election. However, in 2011, it rose to 3.3 and in 2015, with the emergence of Podemos and Ciudadanos, it increased to 5. There had never been such a high effective number of parliamentary parties in modern Spain. Moreover, it is important to note that fragmentation increased not only across parties, but also within them. On the one hand, Podemos is not a unitary actor but includes pre-electoral coalitions with regional parties.² Those parties were promised that they could operate in separate parliamentary groups and have their own leaders, goals and organisations. On the other hand, the PSOE was divided about the leadership of their candidate, Pedro Sánchez, and the strategy to be followed after the 20D election. Both elements were important to the bargaining process, as explained later.

Figure 1. Party system fractionalisation in Spain. Source: Own elaboration with data from Spanish Interior Ministry.

Note: Podemos′s regional coalitions have been conflated.

Table 1 shows, on the one hand, the mean self-placement of each party’s voters on the left–right continuum and, on the other hand, the electorate’s placement of each party in the 2011, 2015 and 2016 elections. It also includes a standard measure of polarisation (see Orriols & Balcells 2012, p.398).³ Party fragmentation does not entail polarisation per se if new parties do not emerge on the extremes of the ideological scale. Only if this happens is the margin of manoeuvre of actors reduced, since additional veto players reduce the set of common policies shared by the parties, thus making agreement more complicated (Tsebelis 2002). Nevertheless, as can be seen in Table 1, since 2011 both the PP and the PSOE have tended to be placed more on the right by the Spanish electorate, while Podemos emerged as the party furthest to the left in 2015. Ciudadanos was viewed as a centre-right party. The discrete polarisation index also increased as new elections took place and, when none of the blocs had an absolute majority, it created additional difficulties for approaching other actors.

Table 1. Left–right mean position of parties 2011–16.

	Self-placement by party′s voters			Voters’ placement of each party
	2011	2015	2016	2011	2015	2016
PP	7.4	7.6	7.7	7.9	8.3	8.4
PSOE	3.9	4.0	3.9	4.1	4.4	4.6
Podemos	n/a	2.6	2.6	n/a	2.3	2.2
Ciudadanos	n/a	5.9	5.9	n/a	6.4	6.4
IU	2.7	2.5	n/a	2.5	2.5	n/a
CiU/DiL	6.1	5.3	5.7	6.5	6.5	6.6
PNV	5.3	5.8	5.7	6.2	6.2	6.2
ERC	3.0	2.6	2.7	3.1	3.3	3.2
CC	5.8	6.0	6.3	5.9	5.7	5.8
Amaiur/Bildu	2.3	2.2	2.2	2.2	2.1	2.3
Polarisation	2.0	2.1	2.2	2.2	2.5	2.7

Source: Elaboration with CIS 2008 post-electoral survey (No. 2,757), 2015 post-electoral survey (No. 3,126) and 2016 pre-electoral survey (No. 3,141). Individuals are requested to place the different parties on a scale that ranges from 1 (extreme left) to 10 (extreme right).

Moreover, polarisation was related not only to the left–right scale, but also to the territorial dimension. After the September 2015 Catalan elections, framed as a de facto plebiscite on independence, the government of Catalonia, formed by the pre-electoral coalition Junts Pel Si (the sum of DiL and ERC), confirmed an 18-month roadmap to secession from Spain (Orriols & Rodon 2016). The goal is a referendum on a new Catalan constitution in an independent country, and thus anticipates political and judiciary tensions between the Spanish and the Catalan governments. Catalan parties’ vigorous demands for independence and demands for a referendum on self-determination were rejected by the PP, the PSOE and Ciudadanos. Only Podemos was in favour of holding a referendum in Catalonia. Thus, ERC and DiL (17 seats) were excluded from negotiations by three of the four main parties, despite their crucial roles in supporting PP and PSOE governments in the past. This introduced additional complexity to the negotiations, as is explained later.

The challenge of uncertainty

The second dimension that could complicate the negotiation process was the role played by uncertainty. Following an election, the composition of parliament changes and political leaders need time to learn which potential cabinets are acceptable to the legislative majority (Diermeier & Roozendaal 1998). Consequently, the process takes more time. However, inter-election uncertainty may be especially high when a de-institutionalisation process is in place, understood as ‘the degree to which competitive political regimes develop (or not) stable patterns of interparty competition’ (Mainwaring & Zoco 2007, p. 156). When party competition is stable, both voters and parties are more predictable from one election to another, and coalition formation outcomes should therefore be easier to anticipate. Conversely, when the political landscape changes considerably from one election to another, politics becomes less predictable.

Spain was widely thought to have a well-institutionalised party system (Mainwaring & Torcal 2006). However, the 2015 general election was an unusual contest. In Table 2, three elements are represented. Firstly, total electoral volatility from one election to another, according to the Pedersen Index.⁴ Secondly, a new-party volatility index, which reflects only the share of votes for parties that compete in election t but did not run in election t ‒ 1, and the volatility of traditional parties that competed in both elections (Nuñez, Simón & Pilet 2016).⁵ It can be expected that when volatility is due to the emergence of new parties, the instability of the system is higher, since new actors enter the legislative arena and interactions are less predictable.

Table 2. Electoral volatility in Spanish general elections.

Year	Volatility to traditional parties	Volatility to new parties	Total Volatility
1979	2.9	2.7	5.6
1982	34.4	1.5	35.8
1986	7.8	4.5	12.2
1989	10.2	0.3	10.5
1993	5.8	0.5	6.2
1996	3.8	0.0	3.8
2000	7.9	0.4	8.3
2004	9.9	2.2	12.1
2008	3.6	0.6	4.2
2011	16.6	4.0	20.6
2015	17.1	19.7	36.8
2016	1.2	2.5	3.7

Source: Own elaboration with data from Spanish Interior Ministry.

The 2015 general election had the second highest level of electoral volatility in recent Spanish history, second only to 1982, when the UCD (Unión de Centro Democrático), the party that had governed Spain since 1977, collapsed. However, the crucial point is that this election had greater volatility due to the emergence of two new medium-sized parties, with a combined 34 per cent of the total vote share. Political change in Spain therefore came very quickly, and the political parties do not know if the political system will stabilise over time. Moreover, the emergence of new political actors also increases the difficulty of agreeing on government formation, especially in the case of Podemos, a party with an anti-establishment platform (Cordero & Montero 2015). The routines and priorities of new parties are more difficult to anticipate: this again increases the challenges of post-electoral negotiations and strategic calculations about how parties will perform if a new election is held. Finally, for the first time in Spain, parties that compete with each other in the national arena have to reach an agreement to forge a parliamentary majority. This novelty, in a context of volatility in which electorates are not crystallised, makes any agreement under consideration more threatening to the integrity of the party.

Spanish stalemate: the process of negotiations after 20D

According to the results of the 20D election, various different investiture votes were possible, as Table 3 shows.

Table 3. Alternatives for an investiture vote after 20D elections.

Candidate		Yes		No		Abstention
Mariano Rajoy		PP+PSOE	213	The rest	137	n/a
		PP+PSOE+Cs	253	The rest	97	n/a
		PP+Cs	163	All but PSOE	97	PSOE	90
		PP	123	All but PSOE and Cs	97	PSOE and Cs	130
Pedro Sánchez		PSOE+Ps+Comp+IU	161	The rest but Cs	149	Cs	40
(with Podemos)		PSOE+Ps+Comp+IU + PNV	167	PP+Cs	163	All but PP and Cs	20
		PSOE+Ps+Comp+IU+ERC+DiL	178	PP+Cs	163	All but PP and Cs	9
		PSOE+Ps+Comp+IU+PNV+ERC+DiL	184	PP+Cs	163	All but PP and Cs	3
Pedro Sánchez		PSOE+Cs	130	All but PP	97	PP	123
(with Ciudadanos)		PSOE+Cs+Comp.+CC+PNV+IU	143	All but Ps	142	Ps	65
Pedro Sánchez		PSOE+Ps+Cs	199	The rest	151	n/a
(three-way alliance)

Source: Own compilation considering alternative possibilities according to press and party official statements.

Note: Size of the Congress: 350 seats. Absolute majority: 176 seats.

First, Mariano Rajoy could have formed a grand coalition with the PSOE (213 seats) or could even have incorporated Ciudadanos into an oversized coalition government (253 seats). Rajoy could have agreed to form a government with Ciudadanos (163) but that would have required the abstention of the PSOE. If those alternatives were not feasible, Rajoy could have tried to form a minority government (123 seats) with the abstention of the PSOE and Ciudadanos. Second, Pedro Sánchez could have tried to form a left-wing government with Podemos, Compromís and IU (161 seats), but that would have required the abstention of Ciudadanos or, if the PP and Ciudadanos voted against it, the affirmative vote of the Basque regional PNV and the abstention of the Catalan regional parties DiL/ERC. Third, Pedro Sánchez could have made a pact with Ciudadanos (130 seats) to forge a majority, but he would have required the abstention of the PP or the abstention of Podemos and the affirmative vote of IU, Compromís, Coalición Canaria and PNV (143 seats). Finally, Pedro Sánchez could have attempted a triple alliance with Podemos and Ciudadanos (199 seats).

Table 4 presents the different events in the negotiation process, taking into consideration the role played by the King, negotiations between parties, and intra-party dynamics that could affect the process as well as the role played by parliament:

Table 4. The negotiation process.

Date	King’s interventions	Negotiations	Intra-party dynamics	Parliament
Stage 1: Red lines and institutional paralysis
28 December 2015			PSOE Federal Committee (negotiations)
13 January 2016				Opening of the Congress and its presidency
18 January 2016	Round of consultations
22 January 2016	Rajoy rejects the appointment from the King	Iglesias in favour of a coalition government with PSOE
27 Januar 2016	Second round of consultations
31 January 2016			PSOE Federal Committee (consultation announcement)
Stage 2: Pedro Sánchez, the formateur
2 February 2016	Sánchez, formateur
6 Feburary 2016		Sánchez’s exploratory meetings
15 February 2016		Podemos repeats its proposal of a coalition government
22 February 2016		The table of the Four (PSOE, Ps, Comp, IU)
24 February 2016		PSOE–Cs agreement
27 Feburary 2016			PSOE consultation about agreement with Cs
1 March 2015				Defeat of Sánchez in parliament (130 vs. 220)
4 March 2016				Second defeat of Sánchez in parliament (131 vs. 219)
Stage 3: The road to new elections
5 March 2015	No new candidate appointed
9 March 2016			Resignations in Podemos Madrid
15 March 2016		Meeting Sánchez: Puigdemont (DiL), Junqueras (ERC) and Colau (En Comú)	Sergio Pascual (Podemos) dismissed
30 March 2016		Pedro Sánchez and Pablo Iglesias meeting
7 April 2016		PSOE, Podemos and Ciudadanos meeting
18 April 2016			Podemos consultation
25 Aprill 2016	Final round before calling new elections
26 April 2016		Proposal of minimum agreement (Compromís)
3 May 2016	The King signs a call for new elections

Source: Own compilation according to press records.

First stage: red lines and institutional paralysis (20 December to 27 January)

Two events took place immediately after the results of the 20D general election were announced. On the one hand, Podemos announced their conditions (they called them ‘red lines’) for supporting any government. The most significant was constitutional reform that would allow an independence referendum in Catalonia. Moreover, Pablo Iglesias, leader of Podemos, announced that the party would be ‘pleased’ to see new elections if their conditions were not met (Manetto 2015). On the other hand, the PSOE started to disclose their internal disputes. The poor results of the party, which fell from 110 to 90 seats, opened up the fight between the Socialist candidate, Pedro Sánchez, and the regional leaders (informally called ‘barons), and especially Susana Díaz, President of the Andalusia region and head of the strongest territorial branch of the PSOE. They had had a very bad relationship ever since Pedro Sánchez – whom Díaz had supported as leader of the PSOE but not as candidate for prime minister – had run without her support, and had won the primaries of the party uncontested (Sanz 2014).

On 28 December in a Federal Committee – the most important internal institution of the party aside from the Federal Congress – the regional leaders of the PSOE established the conditions to be met by Sánchez in any negotiations to form a government. Specifically, they vetoed any discussions with the PP, but also with the Catalan pro-independence parties (ERC and DiL) and with Podemos if these parties did not retract their demands for a self-determination referendum (Sanz 2015). Moreover, they declared that the PP, led by Mariano Rajoy, should initially be the formateur party. In sum, the Socialists greatly restricted the margin of negotiations of their leader because of their own internal divisions.

The inaugural session of the Congress of Deputies, the lower chamber of the Spanish parliament, took place on 13 January with a renewal of 62 per cent of MPs (Coller 2015). The first issue discussed was the election of the president of Congress, which was important because among other duties, he or she decides the date of the investiture vote (article 170 Congress Regulation) and countersigns the candidate appointed by the king. As a result of an agreement between the PP, the PSOE and Ciudadanos, Patxi López (PSOE), former president of the Basque Country, was appointed as president of Congress. Traditionally, this post had been assigned to the party most voted for.

Another element to be negotiated was the formation of the parliamentary groups. This was significant for Podemos because it had promised its territorial allies – En Comú Podem (Catalonia), Compromís Podemos Es el Moment (Valencia) and En Marea (Galicia) – that each of them would have their independent parliamentary group, which would have given them additional resources and political initiative in the chamber. The Congress presidium rejected this possibility and finally Podemos accepted that it had to form one single parliamentary group with its territorial allies, labelled a ‘confederal’ group (Garea & Manetto 2016). Nevertheless, four Compromís MPs did not merge with Podemos and instead entered the non-affiliated group (Sanjuan 2016) and operated as an independent political actor during negotiations.

In accordance with the Spanish Constitution, King Felipe VI began a round of consultations with the different groups. On 22 January, the King met Iglesias, Sánchez and Rajoy one by one. When Pablo Iglesias finished his meeting with the King, without consulting his potential partners, he publicly proposed a coalition government with the PSOE and IU. He specifically requested the vice-presidency and five ministerial portfolios for Podemos: Economy, Defence, Interior, Justice and Foreign Affairs (Alberola & Manetto 2016). The big surprise was that Rajoy, despite leading the winning party, rejected the mandate although, 48 hours earlier, he had said that he would accept it. Rajoy argued that he had insufficient parliamentary support because of the PSOE’s aversion to a grand coalition government, especially after the offer made by Pablo Iglesias (Casqueiro 2016). Thus, one month after the election, there was still no formal candidate for prime minister.

Second stage: Pedro Sánchez, the formateur (27 January to 5 March)

The King began a second round of consultations with the parties between 27 January and 2 February. Meanwhile, Pedro Sánchez announced two initiatives to the 31 January Federal Committee of the PSOE. First he would try to build a majority in Congress if Rajoy again declined to do so. And second he would ask party members to confirm any agreements that he might sign with other parties (Díez 2016a). Following his last meeting with Mariano Rajoy on 2 February, the King appointed Pedro Sánchez as candidate PM. It was the first time since the restoration of democracy in Spain that the second party in Congress was being asked to try to govern the country.

The PSOE was the formateur party and this meant that its preferences and its work to build a parliamentary coalition within party system constraints would be crucial (Warwick 1996). According to Table 3, three main workable majorities could be built by Pedro Sánchez: (a) an alliance with Podemos with the abstention of Ciudadanos and/or pro-independence parties (161 seats), (b) an alliance with Ciudadanos and the support of other parties but the abstention of Podemos (143 seats) and, finally, (c) the support of Podemos and Ciudadanos (199 seats). However, each of those paths had difficulties. On the one hand, the PSOE had internally vetoed any agreement with pro-independence parties (ERC and DiL) and they directly competed with Podemos to be the leading party on the left. On the other hand, Podemos and Ciudadanos vetoed each other during the negotiation process, so the support for or the abstention of one in favour of the other seemed unlikely.

Sánchez had bilateral meetings with all the parties. On 15 February, Iglesias repeated his offer of a coalition government with the PSOE and his preference for the abstention of Catalan pro-independence parties (ERC and DiL) instead of opening negotiations with Ciudadanos (Díez & Manetto 2016). Days later, Alberto Garzón, the leader of the IU, successfully proposed a round table of the four left-wing parties in order to open communications between Podemos and the PSOE. Meanwhile, the PSOE was progressing in its negotiations with Ciudadanos and, as a result, they presented an official public agreement (García de Blas, Díez & Manetto 2016). The parties agreed on 200 measures in seven broad groupings: the fiscal model and productive investment, the labour market, science, social spending and welfare state, civil rights, foreign affairs and constitutional reform (Piña 2016).⁶ These measures were presented in a formal document signed by the leaders and in return Albert Rivera, leader of Ciudadanos, promised to support Pedro Sánchez in his investiture vote, although the form of the government (minority or coalition) had not been defined.

This agreement would give 130 seats to Sánchez, insufficient to win the investiture vote without the support of other parties. In response to this agreement, Podemos considered the negotiations over, and IU and Compromís followed suit. They argued that the proposed government’s policies were right-wing and had been designed to incorporate the PP into the coalition. They also declared that they would not abstain in the investiture vote and, following the probable defeat of Pedro Sánchez, they would ‘offer their hand’ and negotiate again (Carvajal 2016). Over the bank holiday of 26 and 27 February, an internal PSOE consultation was held about the agreement with Ciudadanos. The final result was a 79 per cent vote in favour, with a 51.9 per cent turnout.

The President of Congress established that Sánchez’s investiture speech would take place on 1 March. The vote was planned to take place the following day and, if the candidate were to be defeated, the new vote would be held 48 hour later. As expected, only the PSOE and Ciudadanos voted in favour of the candidate, who was defeated in the first round. The result 48 hours later only added one single seat in favour (Coalición Canaria). Again, Sánchez was defeated. Nevertheless, after the first vote, the countdown to the new elections began. If an agreement were not reached in two months, the King would sign the new election decree on 3 May.

Third stage: the road to a new election (5 March to 3 May)

After the failure of the investiture vote, the King decided not to appoint a new candidate for PM (Alcazar 2016). The PSOE and Ciudadanos decided to consider their pact still valid and said that they were seeking the support of other parties. Thus, as shown in Table 2, the coalition could only succeed in a new investiture vote if (a) the PP abstained (130 seats), (b) the PSOE gained the support of other parties and the abstention of Podemos (143 seats) or (c) the PSOE and Ciudadanos joined Podemos in a three-way alliance (199 seats).

Suddenly, an internal crisis emerged in Podemos, with the resignation of various regional leaders in Madrid, all considered close to Iñigo Errejón, the deputy leader of the party (Carvajal & Bécares 2016). This conflict was related to the disagreement between those who preferred a less ideological and more pragmatic party and others who preferred a more anti-establishment and left-oriented party. The former thesis was publicly supported by Iñigo Errejón, with the latter supported by Pablo Iglesias, the party leader. On 15 March, Iglesias dismissed Sergio Pascual, the organisational Secretary of Podemos, close to Iñigo Errejón, because of his ‘unsatisfactory role’ in preventing the internal conflicts (Manetto 2016a). The internal fighting was only resolved on 2 April when Pablo Echenique, the leader of the Aragon branch, was nominated as organisation secretary.

This infighting and the risk of not being able to repeat regional coalitions in Valencia, Catalonia or Galicia (because of the incapacity to give them each a parliamentary group) were reasons put forward to argue against the wisdom of a new election for Podemos. Meanwhile, Sánchez had a meeting with Carles Puigdemont, President of the Catalan regional government and leader of DiL (eight seats). Weeks later, it also became known that Sánchez had had secret meetings with Oriol Junqueras, leader of ERC and Vice-President of the Catalan regional government (nine seats), and Ada Colau, Mayor of Barcelona and leader of En Comú Podem (Sanz 2016a). However, no progress was made public.

Following the Easter holidays, on 30 March, Iglesias and Sánchez had a meeting in Congress. Iglesias stated that he was still committed to a minority coalition government led by Pedro Sánchez (161 seats). He also said that he would forgo being part of the new government and accept having a meeting with the PSOE and Ciudadanos, but only to build a new agreement (Manetto & García de Blas 2016). The meeting between the three parties took place on 7 April and lasted two-and-a-half hours. Podemos presented a document with 20 policy measures and insisted on forming a PSOE–Podemos coalition government with the abstention of Ciudadanos (El Mundo 2016). Ciudadanos said it was impossible to come to any agreement with Podemos because of the huge differences in their programmes. The parties vetoed each other, while the PSOE insisted that an agreement was in fact feasible.

The next day, Podemos announced that they considered the negotiations to have ‘broken down’. Pablo Iglesias also announced that an internal consultation would take place among Podemos’s 400,000 registered members regarding their support for either the ‘Sánchez–Rivera pact’ or a ‘progressive government’. The final result of the consultation was made public ten days later. The turnout was of 40 per cent (representing 150,000 members) and 88.2 per cent rejected the PSOE–Ciudadanos agreement, while 91.7 per cent supported a PSOE–Podemos coalition government (Manetto 2016b).

The King opened a new round of consultations the following week, thus confirming that no agreement had been possible. On 26 April, Compromís suggested a new minimum agreement, including 30 measures, with the PSOE and without Ciudadanos, as a potential basis to avoid new elections. However, the counteroffer of the PSOE – to accept 27 of the 30 measures but with a minority government including independents and a no-confidence motion paged for two years later – was rejected by Compromís (Terrasa 2016). Meanwhile, Podemos and IU opened negotiations for a possible pre-electoral coalition. Following this process, and given the impossibility of appointing a candidate, the King, with the countersignature of the President of Congress and in accordance with article 99.5 of the Spanish Constitution, dissolved parliament and called a new election for 26 June.

Towards new elections

Differences from the 20D election

There were two crucial differences between the 26J electoral campaign and the previous one (Orriols & Cordero 2016).

The first key contrast was a pre-electoral coalition of Podemos and IU. This agreement was officially presented on 9 May by Pablo Iglesias and Alberto Garzón. The alliance was approved by the party affiliates of both parties (García de Blas 2016) and the pre-electoral coalition was registered under the name of Unidos Podemos (Together We Can – UP). The agreement incorporated 50 measures focused on economic, social, institutional, environmental and international issues⁷ and incorporated 11 different parties.⁸ The agreement had potential benefits both for IU and for Podemos. On the one hand, as IU was a party traditionally punished by the electoral system (Lago 2008), the alliance potentially encouraged a more efficient translation of votes into seats (in the 2015 general election, 80 per cent of IU votes ended up with no representation). On the other hand, Podemos expected to grow further with the additional votes of IU, now a member of the coalition. The party’s aim was to win second place in the election, overtaking the PSOE to become the main party of the Spanish left. Moreover, it was hoped that adding the IU voters would cancel out any potential electoral penalty for the failed negotiations with the Socialists which had provoked the new elections.

Podemos and IU decided to coordinate their campaigns but to run them separately. Their agreement also involved modifying Podemos’s electoral lists to ensure safe seats for the new partner. However, this also meant internal debate in Asturias or Malaga, where the relationship between Podemos and IU was not very good. Unidos Podemos also offered the PSOE a coalition for the Senate, but the Socialists rejected the idea. The plan was also rejected in Valencia, where the possibility had been considered in advance by the Socialist president of the region (Sanz 2016b).

The immediate implication of the pre-electoral agreement was that UP became the second party according to opinion polls and it remained in that position for the whole electoral campaign, as can be seen in Figure 2. This contrasted with the 2015 campaign, where Ciudadanos overtook Podemos in the polls and began a close fight for second place with the PSOE (Orriols & Cordero 2016). Moreover, considering that Ciudadanos had less than 15 per cent of the vote, in this new context it could be penalised by strategic voting in favour of the PP. In sum, the departure point was quite different for the new parties.

Figure 2. Average of survey polls before 26J general election. Source: Kiko Llaneras – pollster, El País.

Note: Polls are weighted based on sample size and date. The weight of each poll decays exponentially such that each poll has a half-life of seven days.

The second important change was the anomaly of repeating an election after a process of failed negotiations. In the context of a multiparty scenario in which a coalition agreement is necessary to govern, two elements are crucial in determining voters’ choices: retrospective voting based on the performance of the previous government, and the capacity to anticipate specific coalition outcomes (Hobolt & Karp 2010). It can be argued that the political deadlock between January and June in Spain affected both these elements. On the one hand, the effect of retrospective voting was softened. The idea of a vote based on the past performance of the incumbent was moderated, since six months of negotiations had blurred direct evaluation of the previous four-year legislative term. The caretaker government of the PP, which took no relevant role in negotiations, was not the central element discussed in the face of the upcoming election.

On the other hand, aside from sincere voting based on policy programmes, voters ‘vote prospectively’ by employing policy-maximising strategic voting. This means that a voter may cast a vote for their second preferred party to maximise the chances of a favourite coalition (Hobolt & Karp 2010, p.304). In the 20D electoral campaign, the voters might have anticipated the results of negotiations by taking into consideration the local and regional election results. In all cases, the PSOE had forged agreements with Podemos while the PP had allied with Ciudadanos. However, after 20D the only agreement reached was between the PSOE and Ciudadanos; and the animosity between the PSOE and Podemos was clearly highlighted. The consequence was that voters could easily anticipate the government formation that each party would favour. Therefore, it was clear that the campaign was going to deal predominantly with two issues: whether UP could become the largest opposition party and thus overtake the Socialists and, subsequently, the potential government that could be formed.

The context and the 26J electoral campaign

The campaign for the new election in Spain was held in a context of additional pressure applied by the European Union. Spain finished 2015 with a deficit of 5.1 per cent of gross domestic product (GDP), exceeding the 4.2 per cent target agreed by the Commission (Pérez 2016a). The country would require an additional budget cut of €8,000 million in 2016–17 if the Commission goal were to be met. Moreover, this deviation from the Stability Pact (whose goal is a deficit of three per cent of GDP) could have entailed a fine of a maximum of 0.2 per cent of GDP, some €2,200 million. However, on 19 May the European Commission announced that sanctions against Spain and Portugal would be postponed until after elections were held. Despite the potential European Commission fine, Mariano Rajoy insisted that he would lower taxes if he was re-elected as PM and that this was not incompatible with meeting the targets of the Stability Pact (Barber & Bruck 2016).

Days later, El País published a letter from Rajoy to Jean-Claude Juncker, President of the European Commission, dated 5 May. In the letter, the PM said that, after the election and as soon as a new government was in place, Spain would take additional measures, if necessary, to address the deficit (Pérez 2016b). He also stressed the need to consider reforms during the whole legislative term and the political complexity of what was to come. This letter was criticised by the opposition because it meant that the new government, whatever its composition, would have a very restricted margin of manoeuvre with the Commission. Those criticisms paved the way for the beginning of the campaign.

The 15-day electoral campaign started on 10 June. Two important electoral debates were held over the following days. The first was on 12 June and was the first time since 2008 that an economy minister held a televised debate with shadow ministers from other parties. The day after, there was a debate between the four candidates of the nationwide parties. In contrast with the December debate, Mariano Rajoy decided to take part. According to both internet and phone surveys, the winner of the debate was Pablo Iglesias, followed by Mariano Rajoy (De Miguel 2016). Over the following days, the electoral campaign led to no significant changes in the polls and the discussion centred on possible PSOE alliances after the election. The discussion was revitalised when Jordi Sevilla, shadow economy minister of the PSOE, suggested in a tweet that ‘in order to avoid a third election, the candidate with most parliamentary support should be allow to govern’ (Segovia 2016). It was assumed then that the PSOE might abstain in an investiture vote, and allow the PP to govern.

Two crucial events took place during the second week of the campaign. First, the online newspaper Publico.es leaked a secret conversation between the Minister of the Interior, Jorge Fernández Díaz, and the boss of the Anti-fraud Office in Catalonia, Daniel de Alfonso. In the transcript it was revealed that both were harbour the idea of promoting or creating corruption scandals that could incriminate DiL and ERC, the pro-independence parties (Publico.es 2016). This leak led to all opposition parties demanding the resignation of the Minister. Second, on 23 June the European Union membership referendum was held in the UK. The victory of the Leave campaign, with 51.9 per cent of the vote, was announced the following day, the last day of the Spanish campaign, and dominated all discussions. Rajoy made an official statement and insisted on the importance of political and economic stability just two days before the election (González 2016).

The 2016 general election and the Gordian knot

Turnout fell by 3.2 points in comparison with the 20D general election, as can be seen in Table 5. The final result was quite different to what the polls had forecast. The PP won the election with 33.1 per cent of the vote, gaining 690,655 votes and 14 seats compared with the previous general election. Another significant difference from the polls was that the PSOE retained second place. In fact, the PSOE lost only five seats and 120,606 votes. This represented a slight recovery in the two mainstream parties’ total share of the vote, which went from 50 per cent in 2015 to 55 per cent in 2016. UP, the new electoral coalition, was instead unable to increase its support to achieve the sorpasso (overtaking) of the Socialists. The coalition lost 1,089,760 votes compared to the sum of IU and Podemos in 2015. Finally, Ciudadanos lost eight seats and about 390,759 votes.

Table 5. Main results of 26J general election in Spain.

	Vote share			Seat share
	2016	2015	Change 2015–16	2016	2015	Change 2015–16
PP	33.0	28.7	4.3	137	123	14
PSOE	22.6	22.1	0.5	85	90	−5
Ps	21.2	24.2*	−3.0	71	71	0
Cs	13.1	13.9	−0.8	32	40	−8
ERC	2.6	2.3	0.3	9	0	9
DiL	2.0	2.2	−0.2	8	0	8
PNV	1.2	1.2	0	5	6	−1
Bildu	0.8	0.8	0	2	2	0
CC	0.3	0.3	0	1	1	0
Turnout	66.4	69.6	−3.2

Source: Own compilation with data from Spanish Interior Ministry.

Note:

* The changes in vote share have been considered according to the sum of IU (3.6) and Podemos (20.6) in the 2015 general election.

ERC gained 20,000 votes and maintained the same nine seats, while DiL gained almost 80,000 but also kept its eight seats. The PNV lost 20,000 votes and one seat, while the other parties had similar results to the previous general election. Similarly to 2015, the results of the parties were not uniform across the territory (Orriols & Cordero 2016). However, in contrast with the previous election, the PP won in Andalusia and Extremadura, two traditional Socialist strongholds. The PP was the preferred party in 40 of the 50 districts and the PSOE won only in Jaen, Seville and Huelva. Podemos was again the preferred party in Catalonia and actually increased its support in the Basque Country against the PNV, becoming the leading party in votes and seats.

This distribution also had implications for the transfer of votes into seats. The 350 seats are elected by proportional representation but the small district size, on average 6.4, has significant mechanical effects. A third of the districts have less than five seats, creating important disproportionality (Penades & Santiuste 2013) and in general benefiting the first and second parties. In this election, the electoral system over-represented the PP by 6.1 points and the PSOE by 1.6 points. UP received a similar share of votes and seats and was only under-represented by 0.8 points. The electoral system especially punished Ciudadanos, which was under-represented by 3.9 points. Finally, electoral volatility – at 3.8 – was far less than in the previous election, and the effective number of parties also fell in comparison with the 2015 general election.

Table 6 shows vote transfers compared with the results of the previous general election. The data are taken from the Metroscopia post-electoral survey,⁹ conducted using cellphone computer-assisted telephone interviewing (CATI) on 5 and 6 July with a total sample of 1,678 respondents, weighted by region (Metroscopia 2016).

Table 6. Percentage of vote transfers in the 2016 general election.

		Vote choice on 26J
Voters on 20D		PP	PSOE	UP	Cs	Others	Abs/b	Total
	PP	85.0	0.9	0.3	4.8	3.0	6.0	100
	PSOE	0.9	76.9	6.8	3.8	4.0	7.6	100
	Ps	0.9	5.7	73.8	3.3	3.6	8.1	100
	IU	1.5	17.4	62.3	1.5	8.7	8.7	100
	Cs	15.6	2.7	1.8	65.3	3.6	11.1	100

Source: Metroscopia post-electoral survey.

Note: Abs/b; Abstention/blank vote.

Unsurprisingly, the PP had the largest percentage of voters who remained loyal to the party. Only minor changes took place among the voters who moved to Ciudadanos or who abstained. Conversely, only 65 per cent of Ciudadanos’s voters repeated their vote. As expected, one significant shift was that voters from this party shifted to the PP (about 15 per cent) and a significant 11 per cent of Ciudadanos’s voters said that they preferred to abstain. On the left, the PSOE had slightly more loyal voters than Podemos: 76 per cent against 73 per cent. Despite that, their level of transfers to other parties was similar. However, the situation of IU voters was quite different. Only 62 per cent of those who voted IU on 20D decided to vote for the pre-electoral coalition of UP. In fact, 17 per cent of IU voters shifted to the PSOE, while 8.7 per cent voted for other parties. Thus, the aggregation of Podemos and IU voters was far from successful, but especially because of the latter party.

In order to see the extent to which the data confirm this idea at macro level, the vote share of each party has been considered for each of the 8,114 municipalities in Spain, with information provided by the Ministry of the Interior. In the appendix it is shown the correlation matrix of parties vote share. One of the main vote transfers was that of Ciudadanos voters in the December general election who then shifted to the PP. One potential mechanism behind this transfer is strategic voting (Lago 2008). It could be that Ciudadanos voters deserted their first option and voted for the PP in order to maximise representation. If this argument is sustained, the difference in vote share of the PP and Ciudadanos from one election to the next will be negatively correlated, especially in the municipalities of those provinces where the district magnitude is lower (Penades & Santiuste 2013). However, exploratory evidence is rather weak.¹⁰

The second significant change in vote share concerned the desertion of December 2015 IU voters to the PSOE in 2016. Thus, it could be expected that, in those municipalities where IU had more votes in 2015, the negative correlation between the difference in UP and PSOE votes would also be more intense. In order to see if this correlation holds at macro level, the sample has been split into four sections, depending on the votes for IU in the previous election. According to the available evidence, the relationship is in the expected direction but is less intense only in those municipalities where IU had less than three per cent of the votes. Conversely, where IU had more than nine per cent of votes in the 2015 general election, the negative correlation became more intense.¹¹

Thus, two potential hypotheses are central in explaining the 26J results. First, the PP may have taken advantage of the erosion of retrospective voting after the impasse of nine months (Hobolt & Karp 2010). Many voters could have considered that they had already punished this party in the 20D elections; this could explain the voters whom the PP attracted from Ciudadanos and other voters’ abstentions. However, as shown above, the evidence of strategic voting seems rather weak. Second, the UP pre-electoral coalition was unable to overtake the PSOE, despite the predictions of the polls. It is clear that IU voters deserted this party in favour of the PSOE. Organisational disputes and mutual animosity between the electorates of the parties may have played a role. Nevertheless, many voters moved to abstention, a reaction especially intense among young voters, a sector more prone to doing so (Orriols & Cordero 2016). Also the blame game about the new election may have eroded their prospects; some voters may have been disappointed by Podemos after 20D.

Finally, as can be deduced from the results, the new election did not solve the problem of forming a government. In Table 7 the different potential combinations for government formation are presented.

Table 7. Alternatives for an investiture vote after 26J election.

Candidate		Yes		No		Abstention
Mariano Rajoy		PP+PSOE	222	All the rest	128	n/a
		PP	137	All but PSOE and Cs	96	PSOE and Cs	117
Mariano Rajoy		PP+Cs	169	All but the PSOE	91	PSOE	90
(with Ciudadanos)		PP+Cs+CC	170	All but the PSOE	90	PSOE	90
		PP+Cs+CC	170	All but 11 MPs of PSOE	169	11 MPs of the PSOE	11
		PP+Cs+CC+PNV	175	All but one MP of PSOE	174	One MP of the PSOE*	1
Pedro Sánchez		PSOE+UP+Comp	156	PP	137	All the rest	57
(with Podemos)		PSOE+UP+Comp+PNV+DiL+ERC	179	All the rest	171	n/a
Pedro Sánchez		PSOE+UP+Cs	188	All the rest	162	n/a
(three way alliance)

Source: Own compilation considering alternative possibilities according to press and party official statements.

Note: Size of the Congress 350 seats. Absolute majority:176 seats. *The alliance between PP, Cs, PNV and CC creates a tie in Congress. The solution suggested to avoid it was that a single MP of Nueva Canarias, a party in a pre-electoral coalition with the PSOE, could abstain.

Now the combination of the PP+Ciudadanos rose to 169 seats (163 on 20D) while the PSOE+UP went down to 156 seats (161 on 20D). Nevertheless, any potential workable majority required the agreement of different parties from the left–right blocs. First, Mariano Rajoy could try a grand coalition with the PSOE or a minority government with the abstention of the PSOE and Ciudadanos. Second, Mariano Rajoy could agree to the investiture vote of Ciudadanos, also incorporating Coalición Canaria and/or the PNV, but this would also require the abstention of the PSOE (or part of it). Third, the potential options for Pedro Sánchez were reduced. Now the mere abstention of pro-independence parties was not sufficient; instead he required their active support. Finally, a three-way alliance could be tried to fulfil the PSOE’s own internal vetoes but with the pessimistic prospect created by the previous attempt.

After the elections, a number of things changed in comparison with 20D. First, Ana Pastor, a former PP minister, was appointed as president of the Congress, thanks to an agreement between the PP and Ciudadanos (Cruz & Sanz 2016). Meanwhile, the PSOE insisted that it would not try to form a government after the results of 26J, but that it would not abstain in favour of the PP or provoke a third election because of it. It was impossible to uphold all three decisions at the same time. Second, on 28 July Mariano Rajoy accepted a mandate from the King to form a government and, one month later, the PP and Ciudadanos formally signed an agreement with 150 measures, some of them replicating the pact with the PSOE, in order to attract the latter’s support (Sánchez 2016). Coalición Canaria then joined the pact, and thus Mariano Rajoy reached 170 seats in Congress. Nevertheless, he was rejected on 31 August and on 3 September (Garea 2016).

After the 25 September regional elections in Galicia and the Basque Country, Pedro Sánchez announced that he would not agree to abstain in the investiture vote and that the PSOE would hold a congress and primaries to elect a national candidate days before the deadline for new elections, 31 October. During the following week, a dispute broke out in the party; 17 members of the party executive committee resigned in order to force Pedro Sánchez’s withdrawal. He resisted, and on 1 October, in a Federal Committee, the factions fought it out while calling into question each other’s legitimacy. Finally, Sánchez was defeated on his proposal for a party congress (132 against and 107 in favour) and he handed in his resignation (Díez 2016b).

After those events, the PSOE was led by a caretaker committee that had a majority of members critical of Sánchez. On 23 October the PSOE, in a new Federal Committee, decided that it would abstain unconditionally to avoid a third general election. Days later, Mariano Rajoy was appointed candidate again by the King. He was defeated in the first round on 27 October, but elected two days later with the affirmative votes of the PP, Ciudadanos and CC (170) and the abstention of the parliamentary group of the PSOE (68 seats) except for 15 MPs who decided not to follow party discipline (El País 2016). This breakdown of party discipline was led by the Partit del Socialistes de Catalunya (PSC), the sister branch of the PSOE in Catalonia (7 seats), and other pro-Sánchez MPs. Thus, the political deadlock in Spain was broken when, finally, the PSOE, after an internal struggle, approved a minority PP government.

Conclusions

The 2016 general election was caused by the incapacity of Spanish parties to reach an agreement to form a new government. This fact was unprecedented in Spanish politics, as was the new multiparty system born of the emergence of Podemos and Ciudadanos in 2015. According to the Spanish Constitution, a failed investiture has to lead, if new attempts are not successful in two months, to a new general election. This was the case when Pedro Sánchez, the leader of the PSOE, was unable to forge a parliamentary coalition to oust the government of the PP. Why was it so difficult to forge this pact? This article identifies the role played by institutional elements (positive parliamentarianism, constructive vote of no confidence, and weak parliament) and, especially, contextual elements. Among the latter, political fragmentation and polarisation, increased complexity, and party volatility all increased uncertainty and room for electoral competition; together these factors made negotiations more complex.

The paper also explains the different stages of the bargaining process and has highlighted two interesting elements. First, parties are not always cohesive actors and they face internal constraints when forming governments. The divide within the formateur party, the PSOE, and the territorial alliances within Podemos underscored how internal veto players can perform a key role. Second, in a political system where there is a high probability of calling a new election, political actors will behave more strategically, making it more difficult to reach a compromise between negotiation and electoral interests. Traditionally in Spain, minority governments have made agreements with regional parties, which are not in direct electoral competition with – or at least in the same arena as – the PSOE or the PP. This makes it easier to reach a compromise. However, the demands of Catalan parties for independence and the fragmentation of the Congress drew attention to the two new parties, Podemos and Ciudadanos, which were in direct competition with the traditional parties.

Following the failed negotiations, the main novelty of the 26J general election was the formation of a pre-electoral coalition between Podemos and IU. The electoral campaign was mainly centred on the possibility of this party becoming the second biggest and forming a potential coalition after the election. However, the final results increased the marginal vote of the PP, which won seats and votes, while UP came third, after the PSOE. According to the polls, the main vote transfers were from Ciudadanos to the PP and from former voters of IU who did not go on to vote for UP. This new scenario has reduced party system fragmentation and electoral volatility. Moreover, the right-wing PP/Ciudadanos bloc achieved 169 seats, making the government formation process – at least in theory – more feasible. A new government was finally elected with the abstention of the PSOE, but it was proved that internal veto players could play a crucial role, one usually neglected by the literature.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Pablo Simón is a visiting professor at Universidad Carlos III de Madrid. PhD in political science (Universitat Pompeu Fabra), he has been a visiting fellow at Université de Montreal and a post-doc at Université Libre de Bruxelles. He specialises in electoral systems, party systems, fiscal federalism and political participation. His most recent publications have appeared in West European Politics, Political Studies, Publius, Social Sciences Quarterly and Acta Politica.

Notes

1. The ENEP is calculated as follows: where p is the proportion of votes obtained by party i in the election.

2. Podemos itself includes 42 seats of 69 seats. The specific territorial coalitions are En-Comú-Podem–Iniciativa per Catalunya Verds (ICV)–Esquerra Unida i Alternativa (EUiA) (Catalonia, 12 seats), Podemos–Compromís–Esquerra Unida del País Valencià (EUPV) (Valencia Autonomous Community, nine seats) and Podemos–En-Marea–ANOVA - Esquerda Unida (ANOVA-EU) (Galicia, six seats).

3. The calculation is based on a discrete index of polarisation, which accounts for the size of parties and their dispersion on the left–right scale (Alvarez & Nagler 2004).

4. This is one of the most common measures of aggregate stability and variation in party systems (Bartolini & Mair 1990) and it is quite straightforward: all the votes (p) received by each party (w) at election t (pw_t) are subtracted from all the votes received by each party at election t ‒ 1 (pw_{t – 1}).

5. Total volatility is the sum of traditional parties’ volatility and that of new parties’.

6. The agreement can be viewed on the following website: http://www.psoe.es/media-content/2016/02/acuerdo-gobierno-reformista-y-de-progreso-2016.pdf

7. The programme can be viewed at: http://podemos.info/wp-content/uploads/2016/05/acuerdo26 J_final.pdf

8. The parties are: Podemos, IU, Equo, Més Per Balears/Més per les Illes, Batzarre/Asamblea de Izquierdas, Unidad Popular en Común, Izquierda Asturiana, Construyendo la izquierda/ Alternativa Socialista, Segoviemos, Izquierda Castellana and Democracia Participativa.

9. The technical details of the survey can be consulted here: http://elpais.com/elpais/2016/07/10/media/1468169139_869059.html

10. The negative correlation is statistically at a one per cent level. In district magnitude below 5 is –0.47, in districts between 5 and 9 –0.16 and in districts above 9 is –0.34.

11. The negative correlation is statistically at a one per cent level. Where IU had less than three per cent the correlation is –0.15, between three and six per cent –0.13, between six and nine per cent –0.30 and more than nine per cent –0.41.

Appendix

2016 general election vote share correlation.

Table

	Vote share PP	Vote share PSOE	Vote share UP	Vote share Ciudadanos	Abstention
Vote share PP	1
Vote share PSOE	−0.01	1
Vote share UP	−0.52**	−0.12**	1
Vote share Ciudadanos	0.16**	0.02**	0.01	1
Abstention	−0.40**	−0.13**	0.16**	−0.10**	1

Note: N: 8,114 municipalities; data obtained from the Ministry of the Interior database.

*Statistically significant at five per cent level;

**Statistically significant at one per cent level.