Provincial dynamics in the 2017 national elections in Ecuador

This article focuses on the provincial dimension of national politics in Ecuador. Based on an empirical analysis of the party vote shares for the different elections of February 2017, it aimed at understanding why parties dominate the elections in some provinces while remaining peripheral actors in others. This territorial heterogeneity of the vote in national elections was measured using a set of three quantitative indicators: the electoral success of provincial parties, the provincial-national incongruence of the vote and the provincialization of party systems. This article concludes that the provincial distribution of the votes remains one of the key elements in order to understand national elections in Ecuador and that it relies on a subtle combination of province-specific factors, party characteristics and party strategies.


Introduction
M ost parts of the scholarly literature on electoral processes in Ecuador -including comparative works -deal exclusively with national elections, with an even larger dominance of the presidential ones. Parties and media actors follow the same pattern as most of the campaign coverage and party dynamics focus on the presidential races. Voters also tend to give more importance for national (presidential) elections as for example indicated by the larger turnout and the lower number of invalid and blank votes for these elections. Yet, Ecuador is a decentralized country, where provincial, cantonal and parochial elections have a great importance. 1 In 2014, the results of the mayoral elections created wave-lengths that went over the borders of the cantons and had an impact at the national level (Pachano, 2015). At the interaction between the national and the sub-national level, the existence of provincial seats in the national assembly presents a particular interest for political scientists.
Given that the elections of the president, of 15 seats in the national assembly and of the Andean representatives all rely on a nationwide electoral district, the interests of analysing election results aggregated at the national level are rather obvious. But a closer look at these elections reveals another pattern: there are significant sub-national variations of the voting behaviour. This variation across different parts of the Ecuadorian territory has been observed by many analysts of the Ecuadorian elections over time. 2 The existence of differences between the national party system at the national level and the provincial systems has been a constant since the beginning of the democratic period (see for example Pachano, 2004Pachano, , 2007Sánchez, 2008;Freidenberg y Alcántara, 2001;. Together with relatively high levels of volatility and fragmentation, the weak nationalization is probably the third most important feature of the Ecuadorian party system. 1. There are also elections for the Andean parliament but, similarly to European countries, these elections remain of a second-order nature and are poorly dealt with in the literature. 2. In some cases, these sub-national entities correspond to the existing provincial level, while several authors also refer to larger geographical units (such as the sierra, costa, etc.) or to large cities and their agglomeration (mostly Guayaquil and Quito).
In this article, I investigate the importance of the provincial level 3 in the understanding of national elections in Ecuador. This 'provincialization' of the national elections can take several forms: the presence of provincial candidates and parties, a focus on provincial interests in the party manifestos or in the campaign, the formation of electoral alliances between national and provincial parties, a concentration of the votes in a specific province, etc. This article will focus on aspects related to the demand side of an election: the vote shares obtained by provincial and national parties.
Ecuador is a particularly interesting case for the analysis of the provincialization of national elections. Its national assembly broadly gathers two different types of seats: the so-called national seats, that are allocated based on a national electoral district, and the so-called provincial seats, that are allocated based on a provincial electoral district and whose magnitude depends on the size of the population. 4 When electing its representatives, the Ecuadorian voter therefore disposes of two different ballots, potentially leading to split-or straight-ticket voting behaviours. In addition, the electoral calendar implies that all elections take place in the same day, allowing a direct comparison of voting behaviours for the two types of seats.
Second, these national elections took place in a peculiar political environment. In 2017, the presidential incumbent Correa was -for the first time since 2006 -not competing in the elections. In the different electoral contests that followed, Alianza Pais, the party of Correa, gathered its votes in a rather uniform way across provinces and become the main driver of the nationalization of the Ecuadorian party system (Polga-Hecimovich, 2014). It will therefore be interesting to observe whether the 2017 elections and the dominance of Alianza Pais preserves this trends or -on the contrarywhether it leads to a return to a more traditional pattern where provinces are key to understand elections results.
3. Even if the comparative literature mobilizes the concept of 'regional' level and 'regional' elections when referring to his sub-national level of government, the use of the concept of provinces directly refers to the institutional reality of the decentralization process in Ecuador. 4. The vote of Ecuadorians living abroad (and therefore their six seats in the national assembly) has not been included in these analyses. The votes in the three provinces (Guayas, Pichincha, Manabí) that allocate national legislative seats based on sub-provincial districts have been aggregated at the provincial level in order to facilitate the empirical comparison across provinces.
In this article, I will analyze the electoral performance of the national and provincial parties and to observe the extent to which these elections are nationalized or provincialized. This article is structured into two complementary sections. In a first section, I will specifically look at the performance of provincial parties in the national legislative elections, not only as stand alone political but also in their alliance with national parties. The second section will deal with a specific aspect of the provincialization process, i.e. in how far national parties concentrate their votes in specific provinces for the different elections. Both sections will rely on various empirical indicators collected at the provincial level. In lieu of conclusion, I will summarize the findings of these analyses and hint towards likely explanations of these multi-level phenomena.

Provincial parties in the 2017 elections
Provincial parties 5 can be considered as the first source of observed differences across provinces. The logic is rather simple: the larger the vote shares obtained by a provincial party in a province, the more the elections in this province display a different pattern from the national level and from other provinces. Provincial parties are often defined as political parties that compete and obtain votes in only one province of a country (see for example Brancati, 2008 on regional parties). Provincial parties can participate in elections in different levels of government (local, provincial, national etc.), as long as they compete only within the provincial border as, for instance, in the case of the provincial seats for the national assembly. In the framework of this article, we use this definition to distinguish between national and provincial parties rather than the official classification of parties by the National electoral Council (even if both classifications broadly lead to a similar distinction).
Yet, provincial parties do not always compete on their own, but they may also unite their electoral destiny with another party and create an 5. In this article, "party" will be used indiscriminately for all parties and political movements. electoral alliance. The Ecuadorian Codigo de la Democracia allows parties to choose between two options: to participate to the elections on their own or to create an alliance with one or several other political parties. There are no legal restrictions on the number of parties included in the alliance or on their nature -there can be national, provincial parties or both in the alliance.
In the 2017 elections, the electoral alliance that attracted the largest attention is the alliance between two national parties: CREO and SUMA. These parties participated jointly in all electoral processes: for the presidential elections, for the legislative elections (national seats) and for the Andean elections. The elections for the provincial seats in the legislative assembly did not follow the same pattern as the nature of this alliance varied across provinces (see below). But the alliance CREO-SUMA, was not the only one gathering national parties as there was other alliances present for the Andean elections, such as the alliance APAIS-PSE or the Acuerdo por el Cambio, gathering ID, UP and MUPP. The later alliance also presented a joint candidate for the presidential elections. At the provincial level 6 , electoral alliances seem to be a frequent phenomenon for national parties. We can distinguish between two types of alliances: between two -or more -national parties; or between one -or more -national party and one provincial party. The alliance between AP and PSE in the province of Azuay is an example of the former, while the alliance between AP and the Movimiento Autonomico Regional (MAR) in the province of El Oro is an example of the later. Methodologically speaking, the task of identifying provincial parties participating in the 2017 elections is complicated by the fact that not all provincial parties mention the provincial level of government in their party label, or that an alliance between two national parties may take a provincial label, such as the Alianza Despierta Orellanence that gathers the PSP and Avanza in the province of Orellana. In addition, the same label can be used by different alliances across provinces. This is for instance the case of the 'Acuerdo por el Cambio' that gathers the CD-UP (Manabí), CD-UP-ID 6. There are even intra-provincial differences. Since the largest provinces are divided into different electoral districts, it is possible to witness different parties and alliances competing in the various districts of a province. In 2017, that was the case in the province of Manabí. In the North district, there was an alliance between the PSC and a regionalist party, while in the South district, both parties competed separately.
Overall, no less than 39 provincial parties participated in the legislative elections of 2017 for the so-called provincial seats. 7 This figure as relatively high, especially when compared to the previous provincial elections. In 2014, Ttwenty-eight (28) provincial candidates participated in the twentythree (23) provincial elections (Dandoy, 2014). These provincial parties not only vary according to the provinces and to their election strategies, they also have varying electoral successes (see below). There are no provincial parties in only four provinces: Cañar, Esmeraldas, Imbabura and Morona Santiago (see Table 1). In these provinces, voters only had the choice to express their vote(s) to national parties. On the contrary, five provinces witness the presence of no less than three provincial parties: Azuay, Carchi, Chimborazo, Manabí and Santa Elena. The number of provincial parties participating in the legislative elections does not depend on the size of the provinces. For example, we observe the presence of two provincial parties in large provinces such as Guayas or Pichincha, as well as in smaller provinces such as Pastaza or Zamora Chinchipe. Table 1.

Province
Provincial party Stand-alone In alliance Total Azuay As these elections rely on a nationwide electoral district, there are by definition no provincial party competing in the presidential and Andean elections, as well as for the so-called national seats in the national legislative elections.
In the 2017 legislative elections (provincial seats), a majority of provincial parties decided to create an alliance with a national party: 23 provincial parties opted for this strategy, representing on average almost one alliance containing such parties per province. Most of these alliances are created together with one national party (with parties such as AP, CREO or PSC). Only six alliances gather one provincial party and two national parties and two alliances gather one provincial party and three national parties (in Loja and Pichincha). Surprising, not a single alliance was created between two or more provincial parties. These figures correspond to the observations made for the 2014 provincial elections, i.e. (Dandoy, 2014) a majority of provincial candidates for the position of prefect created an alliance with at least one national party, while only 8 candidates decided to participate to the elections on their own and only one alliance was created between two provincial parties. The electoral performance of provincial parties Once provincial parties are identified -whether they participate in the elections on their own or in electoral alliances -the next step is the calculation of their vote shares. As mentioned above, the elections for the national assembly provides a unique setting for the analysis of national election results at the level of the provinces. As seats are clustered by province and as national parties can opt for very different campaign strategies (on their own, in alliance with a national party, in an alliance with a provincial party, etc.), the measurement of the parties votes shares requires several methodological decisions.
Concerning electoral alliances, the number of votes for each individual party, is equal to the total number of votes gathered by the alliance divided by the number of partners. For instance, the 8.586 votes obtained by the alliance CREO-SUMA in the province of Galápagos will be divided equally (4.293 votes between the two parties. In the case of provinces with different party configurations across sub-provincial electoral districts, I summed the district results of each party at the provincial level. For example in the province of Manabí, the regionalist party MACHETE presented a list on its own in the South district (28.030 votes) and formed an electoral alliance with the PSC in the North district (112.936 votes). As a consequence, MACHETE 8 was attributed 84.498 votes (=28.030+112.936/2) in the province of Manabí.
Our calculations indicate that provincial parties are rather successful in the national legislative elections (provincial seats). On average, they manage to gather no less than 13,39% of the votes. In comparison, provincial parties obtained 22,01% of the votes in the provincial elections of 2014 (Dandoy, 2014). This difference may be explained by the simultaneity of the legislative (the two types of seats) and the presidential elections. Following the secondorder model, elections that occur on the same day as the presidential ones tend to display the same election result as the presidential contest (see for example Reif, Schmitt, 1980). This electoral performance of the provincial parties has also to be understood in a political and media context that stressed 8. Acronym for 'Movimiento de Acción Cívica de Hombre y Mujeres por el Trabajo y la Equidad'. the importance of national parties, national (presidential) candidates and national political issues over provincial issues.
However, this success of the provincial parties is not homogeneous across the Ecuadorian territory. We can observe huge variations of their performance across provinces. Graph 1 represents the vote shares for all provincial parties at the provincial level for the legislative elections (provincial seats). In provinces such as Los Ríos, Sucumbíos, Pichincha, Napo or Zamora Chinchipe, provincial parties do not perform well. They obtain less than five per cent of the votes, even in the case of Pichincha where two provincial parties participate in the elections. Probably the least successful provincial party is the Alianza Bolivariana Alfarista Los Ríos (ABA) in Los Ríos that managed to only obtain 0,9% of the votes. On the other side of the axis, we observe very successful provincial parties.
In the provinces of Carchi and Santa Elena, they even obtain more than 40% of the votes. In that sense, they became major players on the electoral scene and managed to obtain several seats in the national assembly. In Carchi, no less than three parties provincial participated in the legislative elections. The Movimiento Lidera-Carchi (MLC) obtained 13,07% and the Movimiento Social Conservador Del Carchi (MSCDC) gathered 20,09% of the votes. The alliance between the national party ID and the provincial Integración Democrática was similarly successful (17,14%). Only MPAIS was more successful than the three provincial parties with 23%. In Santa Elena, there were similarly two provincial parties competing on their own and one in an alliance with a national party. The Movimiento Peninsular Creyendo en Nuestra Gente (MPCNG) and the Movimiento Provincial Salinas Independiente (MPSI) respectively obtained 21,4% and 2,7% while the alliance between MPAIS and the Movimiento Provincial Frente de Lucha Ciudadana (MPFDLC) gathered no less than 37,9% of the votes.
In Carchi, the candidate that received the largest amount of votes (Javier Cadena with 26.709 votes) belongs to the MSCDC. The third most popular candidate (René Yandún with 22.262 votes) was part of the alliance ID-IDC. Based on their electoral successes with a provincial party, both candidates obtained a seat in the national assembly. In Santa Elena, the situation is even more favourable to the provincial parties as all three provincial seats were obtained by candidates elected thanks to provincial parties. Carlos Cambala (73.157 votes) and Noralma Zambrano Castro (61.554 votes) were elected based on the electoral alliance between MPAIS and MPFDLC, while Jimmy Candell (52.078 votes) represents the MPCNG in the national assembly.

The provincialization of elections
Besides the electoral success of province-based parties, the observed differences across provinces and between the national and provincial levels of government can emerge from another (larger) process at the level of the provincial party system. In the political science literature, nationalization is a well-studied phenomenon. Conceptualized in Western Europe -among others by the seminal works of Daniele Caramani (2004) -the idea of nationalization relies on the long term evolution of party systems towards an increasingly dominance of national parties and, simultaneously, a gradual decline of the important of regional and peripheral parties. Even if research on party systems in Latin America led to diverging results, the concept of nationalization is often used as a classic indicator in order to measure change and compare party systems across countries.
The provincialization (or de-nationalization if we refer to Caramani's concept) of an election in a concept can be defined in two different ways. First, elections are provincialized when the parties that compete in elections are different across levels government (i.e. national and provincial levels). The presence and strength of provincial parties is de facto taken into consideration in this definition, but it also includes national parties that do not present candidates in specific provinces. Second, elections are said to be provincialized when the vote shares obtained by national parties vary across provinces. In this definition, provincial parties are not considered and only the varying electoral performance of national parties (including the alliances between national parties) is taken into account.

Incongruence of the vote
This first definition can be operationalized via the measurement of the incongruence. Incongruence refers to the overall differences in election results between two elections -be it two elections over time, two elections in different districts, or two elections at different levels of government. This dissimilarity index, sometimes referred to as the Lee Index, is identical to the Pedersen Index of electoral volatility but, instead of comparing an election with another earlier election, the elections for two different types of seats is compared in the same province. The dissimilarity index is calculated by taking the sum of absolute differences between regional and national vote shares for each party and dividing the sum by two (for more information about the different types of congruence measures and their application to national and regional elections, see Schakel, 2013; Dandoy, Schakel, 2013).

Democracias
Vol.5 • octubre-diciembre 2017 • pp. 3-24 Using the dissimilarity index, I measured the incongruence between the two different types of votes (for the national and provincial seats) in the elections for the Ecuadorian national assembly for each individual province. It is expected that, in comparison with the allocation of the national seats in the national assembly, voters can potentially express a more provincialized vote for the allocation of provincial seats. 9 If the two types of votes for the same national election lead to exactly similar results, the incongruence measure will equal to zero. If voters decide to elect completely different candidates for the two types of seats in the same assembly, the incongruence measure will equal to hundred. 10 Graph 2 represents the incongruence between the two different votes for each province. O r e l l a n a C h i m b o r a z o E l O r o M a n a b í C a r c h i S a n t a E l e n a Source: CNE Prepared by: Author 9. Incongruence can also be measured between the votes for the provincial seats in the national assembly and the presidential elections (or even the Andean elections) at the level of each province. Yet, other factors may intervene and the observed incongruence may be the results of other processes than provincialization (for instance, the personalization of the presidential elections or the fact that Andean elections are second-order elections). 10. Traditionally, incongruence measures vary between zero and one, but for the sake of comparison with the indicators of success of provincial party, the dissimilarity index has been scaled to vary between zero and hundred.

Incongruence between national and provincial seats in the national legislative election (per province)
The graph indicates that the 2017 legislative elections in Ecuador are characterized by a rather large incongruence of the vote (on average, provinces reach an score of 25,31), indicating a split-ticket voting behaviour between the votes for the national seats and the votes for the provincial seats. Yet, we also observe very interesting patterns as this incongruence is not homogenous across provinces, but rather reveals large differences across provinces. In the province of Los Ríos, incongruence is very low (4,53) indicating that voters in this province emitted a similar vote for the elections of the national assembly. This similarity (or absence of incongruence) may be explained by the absence of electoral alliances (with the exception of CREO-SUMA) and the weakness of the provincial party (ABA). The same conclusions apply to provinces such as Pichincha (7,61) or Esmeraldas (9,24). Altogether, incongruence between the two types of votes is weak when similar parties are present on the two ballots and when they decide to compete in the elections on their own (i.e. not forming an electoral alliance with another party).
On the other side of the axis, we observe high incongruence for the provinces of Carchi (45,61) and Santa Elena (49,5). Such high score indicates that almost half of the voters in these provinces decided to express a different vote on the two different ballots for the same election. Obviously, these high scores of incongruence originate in the electoral strength of the provincial parties present in both provinces (see above). Yet, a large part of this score is also explained by the difference of the results of the national parties. In Santa Elena for instance, FUERZA.EC obtained 6,66 % of the votes for the national seats while not even half of it (3,12 %) for the provincial seats. Similarly, AVANZA obtained only 1,89% of the votes for the national seats while obtaining three times more votes (6,24%) for the provincial seats. In this province, the incongruence is therefore not the result of party strategies, but rather a combination of successful provincial parties and split-ticket voting for national parties.

Party provincialization
This analysis of the provincialization of national elections in Ecuador relies

Democracias
Vol.5 • octubre-diciembre 2017 • pp. 3-24 on a second definition: elections are provincialized when the vote shares obtained by a party varies across provinces. This conceptualization refers to the classic measurement of the nationalization of party systems (see for instance Rose, Urwin, 1975;Jones, Mainwaring, 2003). 11 The literature in political science refers to a variety of measurements of party systems nationalization but, in the framework of this article, I will use Bochsler's party nationalization score with weighted units (2010). Based on the Gini coefficient of inequality, this measure has the advantage to taking into account sub-national heterogeneity, i.e. the size of territorial units (for more information about the formula, see Bochsler, 2010).
In order to calculate the score of party system provincialization, I subtracted the nationalization score from one. This provincialization measure therefore varies between zero (the party obtains exactly the vote shares in all provinces) to one (the party obtains all its votes in one single province). In Table 2, are displayed the provincialization scores per party and for all elections (legislative elections are sub-divided for the votes for national and provincial seats, respectively). Provincial parties are not included in the table as, by definition, their vote shares are concentrated into one province. As the electoral alliances between national parties for presidential, Andean and legislative (national seats) are homogenous across provinces, their joint electoral score has been included in the calculation. With the exception of the alliance CREO-SUMA, all other electoral alliances (including the ones created with a provincial parties) have been sub-divided between the different partners (see above for the calculation method, including for sub-provincial districts). Even it the analysis of provincialization across policy levels, it possesses the advantage of reducing potential biases. 12 11. For an application and a measure of the party system nationalization in the Ecuadorian case, see for instance Batlle (2009), Polga-Hecimovich (2014) and Mustillo (2017). 12. The other strategy would be to sum the score of the national parties for the legislative elections (provincial seats) in order to match their score obtained in alliances for the other types of elections. This calculation strategy would lead to a loss of variance at the provincial level.

Presidential
Legislative  We observe a large variation of provincialization across national parties. Some parties are poorly provincialized, i.e. their vote shares are fairly similar across provinces. This is mostly the case for AP that gathers about the same results in all provinces and -more surprisingly -in all four types of elections, confirming observations by  for the 2013 elections.
There is no visible incumbency effect for this party in the presidential elections, maybe because its alliance with the PSE compensates its electoral weaknesses in some provinces. The provincialization score for AP for the provincial seats in the legislative elections is slightly higher (0,16), indicating that the party obtains larger scores in some provinces than in others. 13 Several other parties also obtain similar low scores of party provincialization, such as Adelante, Compromiso and the alliance between CREO and SUMA, again with the exception of the provincial seats in the legislative elections.
Most of the national parties (such as Avanza, Concertación, Fuerza. EC, Izquierda Democrática, PSC, PSP, Unión Ecuatoriana or Unidad Popular) present medium scores of provincialization. For instance in the legislative elections (provincial seats), the PSC obtained relatively high vote shares in the province of El Oro (28,25%) while obtaining about ten times less votes in the province of Orellana (2,44%). In some cases, the observed differences across provinces are due to party campaign strategies (presenting a stand alone list or participating in an electoral alliance) leading to different electoral outcomes. Obviously, several parties obtain higher provincialization scores as they decided not to present candidates in some provinces as, for instance, Fuerza.EC in Pastaza, PSE in Carchi or Avanza in Chimborazo. 14 In this provincialization landscape, Pachakutik appears to be an outlier. 15 It score of provincialization is medium when in alliance with Izquierda Democratica and Unidad Popular in the presidential and Andean elections but crosses the symbolic threshold of the 0,5 score of provincialization for the legislative elections. In the case of the provincial seats for the legislative elections this score even reaches 0,73, indicating that the vote shares for this party are almost exclusively concentrated in a few number of provinces. Indeed, we observe that this party did not present candidates in several provinces (including in Guayas, the country's most populated province) or obtained very weak results in others (for instance 0,84% of the vote shares in Manabí). In others provinces, Pachakutik managed to obtain large vote shares, as in the province of Orellana where it became the largest party with 34,01% of the votes.

Conclusion
This article aimed at understanding the observed sub-national differences for the 2017 Ecuadorian elections. Not only turnout strongly varies across provinces, but there are also important variations in terms of party vote shares: some parties dominate the elections in some provinces while remaining peripheral actors in others. Most analyses of national elections in Ecuador focus on the national actors, national parties and national issues while acknowledging that sub-national dynamics may be at play. This article stresses the importance of the provincial level for the analysis of national election results, by attempting to measure the so-called 'provincialization' of national elections in Ecuador.
Three different sets of quantitative indicators were mobilized in order to apprehend the relevance of the provincial phenomena in national elections: the electoral success of provincial parties, the provincial-national incongruence of the vote and the provincialization (or de-nationalization) of party systems. Vote shares for provincial seats in legislative elections were specifically investigated but some comparisons could also be made with other electoral processes. These indicators helped us understand the nature of the cross-provincial variance in national elections by identifying three factors. First, some provinces display specific voting patterns while others have a more homogeneous profile and/or tend to resemble the vote at the national level. Second, the provincial parties managed to obtain important vote shares (and even seats in some cases) in some provinces while remaining marginal Democracias Vol.5 • octubre-diciembre 2017 • pp. 3-24 or absent actors in others. Third, and with few exceptions, national parties obtained more vote shares in some provinces than in others. All together, these factors contribute to render the national elections in Ecuador a multilevel political phenomenon.
Probably the most expected outcome of the 2017 elections was the electoral performance of incumbent AP, as the presidential party also benefited from a legislative absolute majority in the previous national elections. But my observations indicate that the provincialization does not seem to affect much the vote shares of AP. The party obtained a very uniform result across the national territory for the presidential and legislative elections (national seats) and did not concentrate its vote shares in specific highly populated provinces. Its alliance with the PSE for the Andean elections did not lead to a larger or weaker provincial profile, while its different provincial alliances with the PSE or with several provincial parties did not affect its overall distribution of vote shares in the legislative elections. The post-Correa era therefore opens with a continued behaviour of AP as the largest party and the main driver of nationalization of the Ecuadorian party system.
Despite this impact of AP on the overall party system, provincial variations and provincial actors remain at the centre of the electoral dynamics. Several socio-demographic explanations can be found for this continued provincialization and have often been included in comprehensive analyses of voting behaviour in Ecuador. Yet, I would like to stress several institutional and political elements that potentially explain these observed provincial dynamics and that could be investigated in future research. First, the simultaneity between different types of elections undoubtedly has an impact on election results, particularly when the legislative elections rely on two different types of votes (the provincial and national seats). The concept of split-ticket voting signifies that, when given the chance to express two votes on the same day, voters tend to vote for different political parties. If the second-order Andean and legislative (national seats) elections are expected to lead to similar votes as for the presidential elections, the legislative elections (provincial seats) display a distinct pattern. Undoubtedly, a de-synchronisation of the electoral calendar (for instance if the legislative Democracias Vol.5 • octubre-diciembre 2017 • pp. 3-24 elections are held at mid-term to the presidential ones) would have an impact on the voting behaviour and lead to even larger vote shares for provincial parties.
Second, the rules concerning electoral alliances allow a large diversity of party strategies. For the 2017 elections parties were permitted to create alliances with one or several parties, with national parties or provincial parties, for one type of elections and not for others and, more importantly, in some provinces and not in others. This large liberty given to parties to choose their optimal strategy implies that they can opt for largely different strategies. For instance, AP is in an alliance electoral with the PSE in three provinces, with Pachakutik in one province and with different provincial parties in six provinces. Altogether, the provincialization of the national elections benefits from this possibility that parties have to adapt to each individual provincial situation.
Third, the constant importance of the distribution of the votes across provinces in national elections since the return of democracy the country also indicates that -independently on the electoral and territorial reforms and on the ideology of the incumbent party -socio-political cleavages at still at play. The most visible cleavage is probably the ethnic one (see for example the provincialization of Pachakutik's results) but the socio-economic cleavage remains the most important in today's Ecuadorian politics and there are no coherent concentration of left-wing or right-wing votes in some provinces and not in others. Interestingly, populist parties do not display different provincial patterns than other types of parties, reinforcing the idea of a continuous provincial dimension of Ecuadorian politics.