The disruption of an institutionalised and polarised party system: Discontent with democracy and the rise of Nayib Bukele in El Salvador

Abstract

Studies on party system collapse or individual-party breakdowns view programmatic inconsistency or convergence as necessary for abrupt party system change. In theory, a new or fringe contender can suddenly emerge and disrupt the party system under such circumstances. We test that claim by examining Nayib Bukele’s 2019 presidential election victory in El Salvador. With data from the AmericasBarometer, we estimate probit models and predictive margins to examine the individual-level determinants of disruption in an institutionalised and ideologically polarised party system. The empirical results reveal that Bukele won amid salient ideological differences between traditional parties and that critical views towards democracy fueled his core support. Therefore, we conclude that a significant disruption in an institutionalised party system can occur notwithstanding robust ideological differences between leading contenders. Critical attitudes towards democracy can represent a driving force behind a party system’s disruption.

Keywords

El Salvador Latin America left-right ideology mavericks party system disruption

Studies on Latin American politics claim that programmatic inconsistency or convergence between contenders is necessary for abrupt party system change, including individual-party breakdowns or full-blown collapse. In theory, when voters no longer distinguish between party platforms, new or fringe contenders can suddenly emerge and disrupt a party system (Lupu, 2016; Morgan, 2011; Roberts, 2014).

Nayib Bukele’s rise to El Salvador’s presidency casts doubt on that narrative. Following the country’s transition to democracy in the mid-1990s, the leftist Farabundo Martí National Liberation Front (FMLN) and rightist Nationalist Republican Alliance (ARENA) comprised an institutionalised and ideologically polarised party system (Azpuru, 2010; de Zeeuw, 2010; España-Nájera, 2016; Jones, 2011; Lucas, 2013; Roberts, 2014). Both parties received a combined majority of votes in presidential and legislative elections. Yet, in February 2019, voters entrusted Bukele with the presidency by electing him with 53% of support in a first-round vote. A former FMLN mayor of the capital city, Bukele ran for the presidency following his expulsion from the leftist party. The populist maverick firmly criticised the party’s system’s shortcomings, including pervasive poverty, widespread corruption, and violence. Virtually overnight, Bukele’s triumphant presidential bid upended decades of two-party rule. Two years after winning the presidency, his newly formed party, New Ideas, obtained an overwhelming majority of seats in the unicameral congress – confirming the electoral realignment.

Bukele’s rise to power presents an intriguing puzzle. If the party system’s disruption occurred notwithstanding distinct ideological platforms between leading contenders, additional factors should explain its sudden change. We use the AmericasBarometer survey to examine the core electoral backing for the FMLN, ARENA, and Bukele weeks before the 2019 presidential contest. We find that party system change occurred amid robust ideological differences between the leftist FMLN and rightist ARENA. Instead, it was attitudes towards democracy that fueled Bukele’s core support. While the core voting blocs for traditional parties display no substantial differences in their views towards democracy, Bukele’s core support base was less likely to support and be satisfied with democracy.

In the next section, we review the theory on party system change, explain why views towards democracy matter, and formulate our hypotheses. In the third section, we justify our case selection. The fourth section describes the data and methodology. We then present and analyse the empirical results. In the last section, we summarise our conclusions.

Theory and hypotheses

The decay and collapse of party systems in Latin America have led to different explanations for their sudden change (Lupu, 2016; Mainwaring, 2018; Morgan, 2011; Roberts, 2014; Seawright, 2012). There is relative consensus that declining partisanship and the voters’ inability to distinguish between party platforms precede party system disruption. Morgan (2011) shows how programmatic decline results in falling partisanship, which, in turn, facilitates the collapse of party systems. Seawright (2012) views corruption scandals as contributing to a decline in party identification – one of the most proximate factors of party system collapse. Roberts (2014) asserts that it mattered who oversaw economic reform during Latin America’s dual-transitions era (from state-directed development to neoliberalism and from dictatorship to democracy). Cases in which centre-right parties adopted neoliberalism experienced less policy convergence and, eventually, displayed lower electoral volatility. Countries where centre-left parties implemented free-market reforms ultimately endured programmatic dealignment and high volatility. Lupu (2016) argues that policy inconsistency or convergence is central to brand dilution and individual-party breakdowns. Due to inconsistent or convergent positions, incumbent parties with weak party brands will likely break down under economic crises.

Whereas the theory advances our understanding of party system change, the question remains whether such specifications apply to the sudden disruption of institutionalised and ideologically polarised party systems. Since democratisation, El Salvador’s institutionalised party system aligned voters ideologically (Azpuru, 2010; España-Nájera, 2016; Jones, 2011; Lucas, 2013; Roberts, 2014; Wade, 2016). The country avoided the programmatic dealignment common in Latin America following the dual-transition era because the ruling ARENA implemented neoliberalism, and the FMLN remained in opposition (Roberts, 2014). Although there has been clientelism (Koivumaeki, 2010), the system exhibited strong party brands and programmatic differences. The two main parties adopted programmatic linkages ‘to emphasize their ideological distinctiveness’ (Lucas, 2013: 2). Furthermore, identification on the left-right ideological scale was a strong predictor of vote choice (Zechmeister, 2015), and ‘voters had party preferences that corresponded to their economic policy views’ (Baker and Greene, 2015: 183).

Bukele’s triumphant rise in the institutionalised and ideologically polarised party system questions existing theories’ ability to explain the country’s abrupt shift. We apply Downs’ (1957) logic to explain the party system’s sudden disruption. In this vein, ideologically aligned and polarised party systems provide new or fringe parties an opportunity to break with the status quo if their platforms appeal to the median voter. But if polarisation in El Salvador was a constant, why did the disruption unfold in 2019 and not sooner? While Downs’ logic might explain why voters elected Bukele, it does not account for when the disruption occurred. Thus, additional factors should explain the timing of the party system’s abrupt shift. We explore how critical attitudes towards democracy contribute to such an outcome.

Discontent with democracy and party system disruption

Although emerging democracies, including those in Latin America, have made significant inroads since the third wave, the quality of democracy varies substantially (Levine and Molina, 2011). In some cases, problems in democratic governance have resulted in reversals to authoritarianism (Diamond, 2015). For decades, scholars have warned how democracy’s weaknesses might fuel support for independent, maverick, or outsider candidates (Mainwaring, 2006; Carreras, 2012) and feed the rise of authoritarian and populist forces (de la Torre, 2017). Mainwaring (2006: 18) observed, ‘Newcomers present themselves as the champions of fresh ideas and efficient and ethical governments (and in most cases, of popular causes). They claim to be more democratic than the old system’. Miller (2011: 2) states, ‘populists and outsiders are also more likely to arise where democracy is weak or is perceived not to work well’. Carreras (2012: 1462) argues, ‘political independents may provide an electoral option to citizens who have lost faith in political institutions and political parties. In many cases, political outsiders gain prominence by using an anti-establishment and anti-party discourse’. Research shows how democracy’s shortcomings, including party systems, have helped outsiders win elected office – an outcome best evidenced by Alberto Fujimori in Peru (Levitsky and Cameron, 2003) and Hugo Chávez in Venezuela (Weyland, 2003).

Still, the extent to which critical attitudes towards democracy impact a party system remains unclear. Whereas there is consensus that the rise of independent, maverick, or outsider candidates commonly reflects the regime’s problems, studies tend not to incorporate individual-level views on democracy in their empirical analyses (Lupu, 2016; Morgan, 2011; Roberts, 2014; Seawright, 2012). Elsewhere, critical opinions on democracy and system support fuel political outsiders and populists (Voogd and Dassonneville, 2020). In Latin America, critical attitudes towards democracy translate into growing support for candidates who do not run on the existing parties’ platform. The decline in support for democracy in Venezuela before the late 1990s led voters to place ‘their faith’ in Hugo Chávez (Romero, 1997: 24). Weyland reported,

popular dissatisfaction with the actual state of democracy has a highly significant impact on the vote intentions for Hugo Chávez . . . Supporters of this radical populist clearly tended to reject the way in which the political class had been running the country and to agree with the mounting criticism of the established political system. (Weyland, 2003: 833)

More recently, Rennó (2020: 13) found that Bolsonaro’s supporters in Brazil were ‘more dissatisfied with democracy’ in comparison to voters that backed other candidates.

El Salvador offers insights into how attitudes towards democracy might explain a party system’s sudden disruption. While the country was home to one of Latin America’s most stable party systems, El Salvador’s democracy suffered from notorious shortcomings. Wade (2016) argues that elites seized the country’s peace process and constrained democratisation. Indeed, ARENA and the FMLN received a combined majority backing in elections. But both parties ‘failed to address the country’s most serious problems’ (Wade, 2016: 114). Such shortcomings – including pervasive poverty, widespread corruption, and violence – resulted in a gradual decline in support for democracy (Wade, 2016). In turn, these conditions made the conditions ripe for the rise of a maverick presidential candidate.

We aim to explore two crucial features concerning Bukele’s rise and El Salvador’s party system disruption. First, we assess if party system disruption unfolded despite distinct ideological platforms between leading political party contenders. Second, we examine if critical views towards democracy served as a robust predictor of Bukele’s core support. We formulate the following hypotheses:

H1: Sudden disruptions in institutionalised and ideologically polarised party systems can occur when a new or previously fringe contender draws substantial support from voters situated away from both ends of the ideological spectrum.

H2: In highly institutionalised and ideologically polarised party systems, declining support and satisfaction with democracy fuels the core support for challengers who run against the institutionalised party system.

Case selection

In the late 1970s, prolonged authoritarianism and repression strengthened the prospect of armed insurgency in El Salvador. The FMLN, an umbrella organisation of leftist groups, and ARENA, a far right-wing party with ties to death squads, represented both extremes of the conflict that erupted into a full-blown civil war (Koivumaeki, 2010; Wood, 2000). After the 1984 semi-competitive election, ARENA shifted from the extreme right embodied by Roberto D’Aubuisson, the party’s founder, to the free-market policies championed by Alfredo Cristiani. Elected as ARENA’s first president in 1989, Cristiani signed the (Wade, 2016; Wood, 2000), ending the civil war that left 75,000 dead. In the wake of the Accords, the FMLN and ARENA embraced electoral politics. ARENA won the presidency four consecutive times in 1989, 1994, 1999, and 2004. The FMLN won the presidency in 2009 and 2015. As Figure 1 shows, until 2015, El Salvador’s traditional parties received up to a combined 90% vote share in presidential contests. Legislative results point to a similar trend.

Figure 1.

Selected political indicators for El Salvador, 1994–2021.

The bottom-left plot in Figure 1 displays the AmericasBarometer data from 2006 to 2018. Party identification with ARENA remains stable, even weeks before the 2019 critical election. In turn, partisanship with the FMLN fluctuated more, rising from 2006 to 2010, dropping in 2012, and then growing in 2014. The results suggest that presidential election cycles conditioned identification with the FMLN (as it peaked in 2010 and 2014). FMLN attachments dropped in 2016 and reached an all-time low in 2018, coinciding with Bukele’s New Ideas party rising swiftly to 12%. The bottom-right plot shows left-right ideological self-placement by partisanship and past vote in presidential elections. FMLN and ARENA partisans are on both extremes of the ideological scale – evidencing Wade’s (2016: 92) depiction of a ‘no man’s land’ in the centre. While ARENA voters display strong rightist attachments, FMLN voters turned slightly to the centre. The shift is most evident in 2010 and 2012 after the FMLN won its first term in office. Although it moved to the left before the 2015 election, the 2016 and 2018 surveys show a centrist shift. The party system remained ideologically polarised until a novel party, New Ideas, emerged as a centrist contender in 2018.

The two-party establishment had some limited progress in reducing poverty. However, in the years leading to Bukele’s triumph, the country’s economic situation did not improve much (see Appendix 1). Moreover, corruption and violence emerged as El Salvador’s two leading problems. Political elites proved hopelessly corrupt. Francisco Flores (1999–2004) and Antonio Saca (2004–2009), two ARENA presidents, sacked state funds. Flores died under house arrest in 2016 for embezzling foreign aid. After pleading guilty, a local court ordered Saca to 10 years in prison for misappropriating US$300 million. President Mauricio Funes (2009–2014), from the FMLN, embezzled more than US$350 million and fled to Nicaragua. The top-left plot in Figure 2 shows the Varieties of Democracy (V-DEM) political corruption index, with an interval that ranges from low to high values (V-DEM, 2021). The results reveal how corruption was high following democratisation, presenting a U-shaped growth in the years leading to Bukele’s election. Meanwhile, violence escalated. The top-right plot in Figure 2 displays data from the World Bank for intentional homicides. Although the rate peaked in the immediate years following regime transition, reaching 141 homicides per 100 thousand inhabitants in 1995, it gradually fell to 48 in 2002. Afterwards, escalating violence between the state’s security forces and maras (gangs) brought the rate back up to 105 homicides per 100 thousand inhabitants, turning El Salvador into the world’s homicide capital (Watts, 2015).

Figure 2.

Selected indicators for El Salvador, 1994–2018.

The extent to which corruption and violence impacted public opinion varied. According to the AmericasBarometer, an average of 68% of adults surveyed viewed politicians as corrupt from 2008 to 2018. But less than 1% considered corruption the country’s most pressing issue. Violence perceptions, however, told a different story. In 2014, 54% thought that crime was the most severe problem affecting the country. In 2016 and 2018, the figure declined somewhat, to 42% and 36%, respectively (see Appendix 2).

The survey evidence points to a decline in favourable attitudes towards democracy. The bottom-left and bottom-right plots in Figure 2 use the AmericasBarometer to portray support (7-point scale) and satisfaction (4-point scale) with democracy in El Salvador from 2006 to 2018. Those who supported democracy the most (values 6 and 7 in the 7-point scale) fluctuated from 37% in 2006 to 48% in 2008 to 36% in 2010. After growing to 42% in 2012, they drop to 38% (2014), 30% (2016), and reach 31% in the weeks leading to Bukele’s election. Second, those who supported democracy the least (values 1 and 2 in the 7-point scale) reached 15% in 2006 and subsequently fell below 10 (except 2012). Those who weakly supported democracy (the middle values of 3, 4, and 5) continuously grew from 48% in 2012 to 59% in 2018. Hence, Salvadorans only half-heartedly supported democracy in the years leading to the party system’s disruption.

Satisfaction with democracy reveals a similar trend. In 2006 and 2008, Salvadorans were more dissatisfied than satisfied with democracy. In 2010, the year following the FMLN’s rise to the presidency, satisfaction with democracy reached 64%. It remained in the high 50s in 2012 and 2014. However, in 2016, dissatisfaction with democracy surpassed satisfaction (59% vs 41%, respectively). And in 2018, satisfaction with democracy reached a low of 37%, as close to two-thirds of Salvadorans remained dissatisfied with the country’s political regime.

Figure 3 presents a scatterplot that allows viewing El Salvador comparatively with other Central American countries. The data reveal that satisfaction and support for democracy in El Salvador were at one of their lowest points before Bukele’s election. Although views concerning democracy were more favourable from 2008 to 2014, coinciding with government alternations from ARENA to the FMLN, overall attitudes worsened in 2016 and 2018. The data also hint at similar trends in the region. After all, the inhabitants of the surveyed Central American countries displayed their most critical views towards democracy in the 2018 AmericasBarometer wave – even in Costa Rica, the regional beacon of democracy. This outcome suggests that views on democracy might be a relevant factor fueling demand for change elsewhere in the Americas.

Figure 3.

Attitudes towards democracy in selected Central American countries, 2006–2018.

Thus, El Salvador’s two-party establishment proved excessively corrupt and incapable of delivering concrete solutions to the country’s pressing problems. The problems originating from the two-party system coincided with growing popular discontent towards democracy. Bukele (2018) tapped into that discontent by persistently criticising the two-party system. He regularly made statements questioning traditional parties, like ‘they do not care about any limits, nor the truth, nor the due process, nor the right of the people to choose freely in elections. They want power, at any cost’.

One of the fascinating aspects of Bukele concerns his shifting ideological platform. Bukele broke onto the political scene as the leftist FMLN mayor of Nueva Cuscatalán (2012–2015) and San Salvador (2015–2018). In a 2012 interview, Bukele said, ‘I’m from the radical left’ (Dalton, 2012). Nevertheless, Bukele’s ideological address changed as he ran for the presidency. After his 2017 expulsion from the FMLN, Bukele sought to create his New Ideas political party. But he was unable to register the party before the legal deadline. As a result, Bukele sought the presidential nomination of Democratic Change (CD), a small progressive party founded in 2005. Yet, the Supreme Electoral Tribunal cancelled the party’s legal status – a move Bukele criticised as being politically motivated. Undeterred, Bukele won the Grand Alliance for National Unity (GANA) nomination, a conservative party founded by former ARENA President Antonio Saca.

Above all, Bukele is a populist who campaigned against the two-party establishment by appealing to centrist voters despite his shifts. Meléndez-Sánchez argues that

Bukele’s personal brand has also appealed to moderate voters who favor change but fear extremism. Although Salvadorans are generally unhappy with their political system, most identify as ideological centrists . . . Bukele has mitigated the risk of alienating these moderate voters with his authoritarian and populist appeals in part by balancing them with his sleek and polished personal brand – a far cry from the more brazen style that is often associated with political extremism. (Meléndez-Sánchez, 2021: 23)

Whereas Bukele flirted with centre-left and centre-right ideas during the campaign, as president, he has become more closely aligned with conservative values – evidenced by his iron-fist approach to dealing with crime.

Data and methodology

We rely on the AmericasBarometer (n.d.) by the Latin American Public Opinion Project (LAPOP) to present probit maximum likelihood estimators and predictive margins. The 2018 survey’s fieldwork, representative of the country’s adult population, was conducted between 13 November and 6 December. Hence, the data allow us to examine attitudes weeks before the presidential contest of 3 February 2019.

We use the LAPOP data to study the core supporters of the FMLN, ARENA, and Bukele. We made this decision because the survey does not explicitly ask for vote intention for individual candidates or parties. Instead, respondents are asked whether they would ‘not vote’, ‘vote for the incumbent’, ‘vote for the opposition’, or ‘cast a blank or null vote’. Since this question makes it challenging to distinguish between those who voted for Bukele or ARENA (both opposed to the FMLN government), we rely on partisanship to distinguish between voting blocs. Employing the question, Which party do you currently identify with? we coded dummy variables for respondents who mentioned the FMLN or ARENA. For Bukele’s core supporters, we coded a dummy that groups respondents who sympathised with New Ideas and GANA.

Therefore, the core electoral support for the FMLN equals one for respondents who vote for the incumbent and identify with that leftist party. ARENA’s core supporters are respondents who vote for the opposition and identify with that rightist party. Bukele’s core supporters are those who vote for the opposition and identify with New Ideas or GANA. We include respondents who do not indicate a voting preference or do not identify with a political party in the reference category. We code individuals who did not answer both questions as missing values (N = 176).

The robustness checks examine core support for traditional parties over time. We code the same dependent variable (combining vote intention and partisanship) for the FMLN and ARENA, employing the 2008, 2010, 2014, and 2016 LAPOP surveys. We excluded the 2012 survey because that year’s wave only asked satisfaction with democracy to less than half of respondents (N = 695 out of a 1497 sample), making it challenging to conduct meaningful statistical analysis.

The probit maximum likelihood estimators include three primary independent variables: ideological self-placement, attitudes towards democracy (support and satisfaction), and perceiving crime as the country’s most pressing issue. We utilise left-right ideological self-placement to determine differences in the core support for the FMLN, ARENA, and Bukele. We use the following question: ‘According to the meaning that the terms “left” and “right” have for you, and thinking of your own political leanings, where would you place yourself on this scale?’ We normalise the 10-point scale (1–10) to range from zero to one for leftist and rightist values.

In addition, the models incorporate attitudes towards democracy. For support, we rely on this question: Democracy may have its problems, but it is still better than any other form of government. To what extent do you agree or disagree with this statement? Respondents use a 7-point scale, where 1 is ‘strongly disagree’, and 7 is ‘strongly agree’. For satisfaction with democracy, we use, Would you say that you are very satisfied, satisfied, dissatisfied, or very dissatisfied with the way democracy works in El Salvador? We normalise all responses from zero to one for negative and positive views, respectively.

We turn to the following question to assess the effect that crime, as a salient issue, had on political support: In your opinion, what is the most serious problem faced by the country? We coded a dummy for those who mentioned crime. Building on previous research identifying corruption as a determinant of party system change (Seawright, 2012), we sought to include its measurement in our models. However, we faced two problems. First, corruption had a low response in the question focusing on the country’s most serious problem (see Appendix 2). Second, LAPOP asks a second question on corruption, as follows: Taking into account your own experience or what you have heard, corruption among public officials is ‘very common’, ‘common’, ‘uncommon’, or ‘very uncommon’? However, the 2018 wave for El Salvador only asked that question to half of its respondents (N = 720). As a result, the statistical models focus primarily on crime and not corruption. Appendix 3 includes both variables to test their impact, revealing no substantial effect on Bukele’s backing.

The models include controls for the economic outlook, political engagement, and socio-demographic variables. We use the following question for national economic outlook views: Do you think that the country’s current economic situation is better than, the same as or worse than it was 12 months ago? For pocketbook economic outlook, we use, Do you think that your economic situation is better than, the same as, or worse than it was 12 months ago? We normalise all responses to range from zero to one for negative and positive views. We also introduce a control for political engagement, which normalises a 4-point scale ranging from ‘none’ to ‘a lot’. All estimates include socio-demographic controls for religion (Catholic and Evangelical), ethnicity (White and Mestizo), years of education, urban residence, age, and gender. The dummies for Catholic and Evangelical religions contrast those preferences with each other and all other choices (i.e. Protestant, non-Christian eastern religions, traditional religions, none, agnostic, or atheist). The dummies for White and Mestizo also contrast each other and all other options (i.e. Indigenous, Black, mulatto, and other). We code dummies for religion, race, urban residence, and gender and normalise values for education and age. Table 1 summarises the survey data (Appendix 4 summarises the data used for the robustness checks).

Table 1.

Summary of survey data.

Variable	N	Mean	Std. Dev.	Min.	Max.
Core FMLN supporters	1063	0.04	0.19	0	1
Core ARENA supporters	1063	0.09	0.29	0	1
Core New Ideas / GANA supporters	1063	0.17	0.38	0	1
Core New Ideas supporters	1063	0.14	0.34	0	1
Left-right ideology	1072	0.52	0.30	0	1
Support for democracy	1151	0.62	0.26	0	1
Satisfaction with democracy	1151	0.42	0.24	0	1
Crime as national priority	1175	0.36	0.48	0	1
National economic outlook	1161	0.26	0.31	0	1
Pocketbook economic outlook	1168	0.36	0.33	0	1
Interest in politics	1171	0.40	0.35	0	1
Catholic	1175	0.45	0.50	0	1
Evangelical	1175	0.21	0.41	0	1
White	1175	0.26	0.44	0	1
Mestizo	1175	0.50	0.50	0	1
Education	1175	0.52	0.26	0	1
Urban	1175	0.66	0.47	0	1
Age	1175	0.40	0.16	0.18	0.99
Women	1175	0.47	0.50	0	1

Source: Authors, based on the AmericasBarometer.

FMLN: Farabundo Martí National Liberation Front; ARENA: Nationalist Republican Alliance; GANA: Grand Alliance for National Unity.

Results and analysis

In the first section, we discuss the individual-level determinants of core support for the FMLN, ARENA, and Bukele weeks before the 2019 contest. In the second section, we present the robustness checks that focus on the core support for the FMLN and ARENA from 2008 to 2016. This research strategy allows us to distinguish between the why and when of El Salvador’s party system disruption.

Tables 2 and 3 show the results of the probit estimates that focus on core support for the FMLN and ARENA in late 2018. There are four models for each dependent variable. The first model centres on the effect of left-right ideology. The second model omits left-right ideology and adds attitudes towards democracy. The third model includes crime as the country’s most pressing problem. In the last model, we include all predictors and controls. Table 4 replicates the models for Bukele’s core supporters. In models we do not report here, we ran the same predictors and controls using partisanship as a dependent variable. Those models revealed no substantial differences from the output we report below.

Table 2.

Determinants of core support for the FMLN, 2018.

Variables	(1)	(2)	(3)	(4)
Left-right ideology	−1.280*** (0.315)			−1.274*** (0.321)
Support for democracy		0.441(0.393)		0.268(0.410)
Satisfied with democracy		0.385(0.361)		0.520(0.395)
Crime as a national priority			−0.151(0.192)	−0.140(0.204)
National economic outlook	1.233*** (0.323)	1.325*** (0.312)	1.405*** (0.305)	1.138*** (0.334)
Pocketbook economic outlook	0.187(0.337)	0.257(0.322)	0.260(0.317)	0.140(0.346)
Interest in politics	1.200*** (0.281)	1.303*** (0.275)	1.296*** (0.270)	1.207*** (0.287)
Catholic	0.471** (0.233)	0.439** (0.223)	0.478** (0.220)	0.466* (0.238)
Evangelical	0.241(0.298)	0.287(0.286)	0.234(0.285)	0.298(0.305)
White	0.0779(0.282)	0.0299(0.272)	0.135(0.267)	−0.00410(0.289)
Mestizo	0.249(0.254)	0.156(0.240)	0.205(0.241)	0.233(0.256)
Education	−0.194(0.412)	0.225(0.398)	0.119(0.389)	−0.114(0.419)
Urban	0.0143(0.235)	0.0831(0.228)	0.0180(0.223)	0.0257(0.241)
Age	2.692*** (0.599)	2.734*** (0.583)	2.751*** (0.570)	2.630*** (0.612)
Women	−0.0107(0.198)	−0.114(0.190)	−0.0861(0.186)	−0.0436(0.202)
Constant	−3.924*** (0.572)	−5.253*** (0.610)	−4.753*** (0.545)	−4.287*** (0.644)
McFadden’s R²	0.355	0.309	0.304	0.363
Nagelkerke R²	0.396	0.346	0.341	0.404
N	972	1032	1049	964

Source: Authors, based on the AmericasBarometer.

The dependent variable equals one if a respondent states they would vote for the incumbent and identify with the FMLN. Entries are unstandardised probit estimates, with standard errors in parenthesis.

***

p < 0.01, **p < 0.05, *p < 0.1 (two-tailed tests).

Table 3.

Determinants of core support for ARENA, 2018.

Variables	(5)	(6)	(7)	(8)
Left-right ideology	2.453*** (0.271)			2.490*** (0.277)
Support for democracy		−0.136(0.250)		−0.410(0.293)
Satisfied with democracy		0.554** (0.234)		0.333(0.276)
Crime as a national priority			−0.182(0.123)	−0.212(0.143)
National economic outlook	−0.457* (0.260)	−0.791*** (0.230)	−0.817*** (0.229)	−0.431(0.266)
Pocketbook economic outlook	0.207(0.224)	0.306(0.196)	0.315(0.194)	0.205(0.227)
Interest in politics	0.805*** (0.200)	0.717*** (0.175)	0.762*** (0.172)	0.777*** (0.204)
Catholic	0.125(0.161)	0.0829(0.141)	0.0943(0.139)	0.118(0.163)
Evangelical	0.339* (0.178)	0.344** (0.157)	0.311** (0.155)	0.350* (0.181)
White	−0.0564(0.179)	−0.0112(0.158)	0.00817(0.156)	−0.0277(0.182)
Mestizo	−0.309* (0.171)	−0.214(0.149)	−0.171(0.148)	−0.309* (0.174)
Education	0.611** (0.309)	0.0127(0.274)	0.0637(0.268)	0.525* (0.316)
Urban	−0.101(0.150)	−0.172(0.135)	−0.178(0.134)	−0.129(0.153)
Age	1.386*** (0.456)	1.283*** (0.392)	1.369*** (0.387)	1.299*** (0.461)
Women	−0.0235(0.135)	0.00873(0.119)	0.00311(0.118)	−0.0153(0.136)
Constant	−3.998*** (0.411)	−2.328*** (0.339)	−2.064*** (0.290)	−3.900*** (0.454)
McFadden’s R²	0.264	0.092	0.085	0.275
Nagelkerke R²	0.327	0.119	0.11	0.34
N	972	1032	1049	964

Source: Authors, based on the AmericasBarometer.

The dependent variable equals one if a respondent states they would vote for the opposition and identify with ARENA. Entries are unstandardised probit estimates, with standard errors in parenthesis.

***

p < 0.01, **p < 0.05, *p < 0.1 (two-tailed tests).

Table 4.

Determinants of core support for Bukele (2018).

Variables	(9)	(10)	(11)	(12)
Left-right ideology	−0.820*** (0.187)			−0.785*** (0.192)
Support for democracy		−0.596*** (0.224)		−0.576** (0.232)
Satisfied with democracy		−0.474** (0.191)		−0.511** (0.204)
Crime as a national priority			0.136(0.100)	0.134(0.107)
National economic outlook	−0.193(0.193)	−0.0526(0.185)	−0.0876(0.182)	−0.129(0.197)
Pocketbook economic outlook	−0.0599(0.181)	−0.00823(0.176)	−0.0520(0.173)	−0.00535(0.184)
Interest in politics	1.069*** (0.152)	1.149*** (0.147)	1.108*** (0.143)	1.114*** (0.155)
Catholic	−0.0314(0.117)	−0.00324(0.113)	−0.0380(0.112)	−0.0206(0.119)
Evangelical	−0.000960(0.145)	−0.0559(0.141)	−0.0614(0.138)	−0.0182(0.148)
White	0.0440(0.161)	−0.0132(0.155)	−0.0454(0.152)	0.0491(0.164)
Mestizo	0.247* (0.141)	0.169(0.133)	0.162(0.131)	0.223(0.142)
Education	0.806*** (0.255)	0.774*** (0.249)	0.856*** (0.243)	0.716*** (0.262)
Urban	0.104(0.128)	0.0955(0.124)	0.145(0.122)	0.0695(0.131)
Age	0.524(0.354)	0.431(0.340)	0.299(0.335)	0.615* (0.362)
Women	−0.188* (0.107)	−0.191* (0.103)	−0.193* (0.101)	−0.179* (0.108)
Constant	−1.778*** (0.286)	−1.622*** (0.285)	−2.153*** (0.255)	−1.316*** (0.320)
McFadden’s R²	0.134	0.134	0.12	0.149
Nagelkerke R²	0.193	0.193	0.174	0.214
N	972	1032	1049	964

Source: Authors, based on the AmericasBarometer.

The dependent variable equals one if a respondent states they would vote for the opposition and identify with New Ideas or GANA. Entries are unstandardised probit estimates, with standard errors in parenthesis.

***

p < 0.01, **p < 0.05, *p < 0.1 (two-tailed tests).

As seen in Tables 2 and 3, left-right ideological self-placement is a strong predictor of support for the FMLN and ARENA. Although few people identified with both traditional parties, identifying with the ideological left and right increased the likelihood of backing the FMLN and ARENA, respectively. In turn, attitudes towards democracy have little to no impact as a determinant of support for both parties. The exception is ARENA’s second model (see Table 3, column 6), which shows how satisfaction with democracy increased the party’s backing; nonetheless, the effect is limited and loses significance in the full model (column 8). The results also reveal how viewing crime as the country’s most serious problem fails to gain statistically significant explanatory power as a determinant of core support for the FMLN and ARENA.

The economic outlook has a different impact on support for the ruling FMLN and the opposition ARENA. A positive view of the country’s economy increased the FMLN’s support. Conversely, individuals with a negative outlook were more inclined to vote for ARENA (although the variable’s statistical significance fails to reach substantial explanatory power in the full model). Pocketbook economic outlook has no significant effect on either party’s support. The remaining controls are consistent across models. Individuals with a greater interest in politics and older age cohorts were more likely to support traditional parties. The FMLN is also stronger among Catholic voters than individuals with other religious beliefs.

Table 4 presents the individual-level determinants of the core support for Bukele. For that group, ideology plays a central role in explaining the core support for the maverick politician. Bukele’s core support stems from the centre-left. Even more revealing is how attitudes towards democracy shed light on the critical nature of Bukele’s core voters. The individual and full models unveil how less support and satisfaction with democracy increase support for Bukele. As hypothesised, the maverick politician’s niche base was discontent with democracy in El Salvador. Viewing crime as the country’s most serious problem is a weak predictor of Bukele’s core support, as it fails to achieve statistical significance. The controls are consistent across models. There is no substantial effect deriving from the economic outlook. Individuals with attachments to Bukele were more interested in politics, as was also the case for those with higher education levels. Thus, Bukele did not just build his support among those who were uninterested in politics. People who cared about politics were also more likely to support Bukele. There are no significant differences for the remaining controls, pointing to Bukele’s transversal support.

We use the full models (columns 4, 8, and 12) to plot predictive margins with 95% confidence intervals for ideological self-placement and attitudes towards democracy. Figure 4 shows those results. The first plot illustrates the profoundly ideological nature of the party system weeks before its critical election. Core support for the FMLN and ARENA peaks at leftist and rightist values, respectively. Concurrently, identifying with the left also increases support for Bukele. The middle plot portrays the results for supporting democracy. While FMLN and ARENA voters are comparatively more sympathetic towards democracy, the negative slope seen for Bukele’s core supporters underscores their critical views. The third and last plot shows the output for satisfaction with democracy. The results point in the same direction: Bukele’s core supporters were notoriously less likely to be satisfied with democracy. Despite their critical view, no substantial differences exist for ARENA supporters, while the flattened slope for FMLN voters indicates no significant variation.

Figure 4.

Predictive margins for left-right ideology and attitudes towards democracy.

In sum, the individual-level determinants of support for traditional parties were thoroughly ideological, with FMLN attracting left-wing voters and ARENA attracting those on the right. Bukele’s core supporters also had strong centre-left ideological attachments. Yet, the fundamental difference derives from attitudes towards democracy. While the groups that supported traditional parties did not display substantial differences, Bukele’s core voters supported democracy in fewer numbers and were considerably more dissatisfied with democracy. Hence, the empirical evidence shows how the disruption of El Salvador’s party system occurred amid ideological differences between established contenders. Simultaneously, ideology and attitudes towards democracy fueled the support for a maverick candidate who disrupted the party system.

Robustness checks

The empirical results reveal how ideology and critical attitudes towards democracy increased support for Bukele’s presidential bid within an ideologically polarised party system. While we shed light on the determinants of support, the timing of the party system’s disruption remains unclear. The following question remains unanswered: Did attitudes towards democracy influence vote choice before the 2019 election? If critical views on democracy did not drive voter behaviour in the past, the rise of critical views in 2018 helps piece together the puzzle of El Salvador’s party system disruption.

We use four LAPOP survey waves (2008, 2010, 2014, and 2016) to explore core support for the FMLN and ARENA during the height of the two-party system. Figure 5 presents four coefficient plots with 95% confidence intervals for each survey wave. For simplicity purposes, we only show the results for our main variables of interest, left-right ideological self-placement, attitudes towards democracy, and viewing crime as a country priority.

Figure 5.

The determinants of support for the FMLN and ARENA, 2008–2016.

We draw three conclusions. First, the left-right ideological scale is significant for both parties across survey waves. The outcome confirms the party system’s polarised nature in the years before the 2019 election. Second, support for democracy repetitively fails to achieve statistical significance in explaining the backing for either party. Still, satisfaction with democracy points in a different direction. On one hand, it increased ARENA’s backing in 2008 and decreased it in 2014 and 2016 (p < 0.05 level). On the other hand, satisfaction with democracy enhances the likelihood of supporting the FMLN in 2014, coinciding with Salvador Sánchez’s election. In other words, the results suggest that incumbency status likely conditioned the variable’s effect. Third, despite its salience in public opinion polls, viewing crime as a national priority fails to display statistically significant explanatory power.

The robustness checks confirm two findings. First, El Salvador’s party system displayed distinct ideological platforms in the decade preceding its disruption. Second, despite minor exceptions, critical attitudes towards democracy were not significant determinants before 2018. Thus, a combination of ideology and critical attitudes towards democracy fueled Bukele’s rise to power.

Conclusion

Since democratisation in the mid-1990s, stability marked El Salvador’s party system. However, the 2019 presidential victory by Nayib Bukele represented a significant disruption, which the 2021 legislative election confirmed. While studies emphasise that inconsistent or converging party platforms facilitate individual-party breakdown and party system collapse, our findings take a step in a new direction – one where people’s critical attitudes towards democracy play a fundamental role.

Weeks before El Salvador’s 2019 presidential election, the established contenders – the FMLN and ARENA – were ideologically distinct and polarised from each other. Nonetheless, those ideological differences were insufficient to prevent the disruption of one of Latin America’s most stable party systems. Our empirical findings show that Bukele’s core supporters had ideological roots, as the maverick political drew support from centre-left voters. Although Bukele attracted support from the growing population that does not identify with parties, his support among those who continue to identify with them came from the centre-left.

Still, there was a more dominant propeller of party system change. Three decades of two-party rule resulted in critical attitudes towards democracy. Salvadorans who were less supportive and more dissatisfied with democracy voiced their discontent by supporting Bukele. The robustness checks confirm the absence of statistically significant effects stemming from both variables in the decade preceding the 2019 election. As a result, the empirical evidence corroborates our hypotheses.

El Salvador’s case sheds light on how abrupt change can occur within institutionalised and ideologically polarised party systems. While other studies on party system change stress the need to take financial distress (Lupu, 2016; Morgan, 2011), corruption scandals (Seawright, 2012), and the political alignments surrounding neoliberal reform (Roberts, 2014), we put forward the idea of incorporating individual-level views on democracy. In democracies worldwide – both affluent and developing – voters are turning their backs on established parties. Bukele’s rise to power suggests that critical views towards democracy represent a crucial factor to consider when examining party system disruption.

Footnotes

Appendix

Appendix 4.

Summary of survey data for robustness checks.

Variable	Survey wave: 2008			Survey wave: 2010			Min.	Max.
	N	Mean	Std. Dev.	N	Mean	Std. Dev.	Min.	Max.
Core FMLN supporters	895	0.38	0.49	1093	0.30	0.46	0	1
Core ARENA supporters	895	0.18	0.38	1093	0.07	0.25	0	1
Left-right ideology	1310	0.47	0.33	1372	0.47	0.28	0	1
Support for democracy	1402	0.68	0.29	1426	0.65	0.26	0	1
Satisfaction with democracy	1407	0.45	0.24	1430	0.55	0.22	0	1
Crime as national priority	1431	0.30	0.46	1445	0.55	0.50	0	1
National economic outlook	1424	0.17	0.29	1442	0.32	0.35	0	1
Pocketbook economic outlook	1429	0.28	0.34	1445	0.38	0.34	0	1
Interest in politics	1430	0.39	0.34	1444	0.38	0.34	0	1
Catholic	1431	0.54	0.50	1445	0.52	0.50	0	1
Evangelical	1431	0.34	0.47	1445	0.37	0.48	0	1
White	1431	0.19	0.40	1445	0.23	0.42	0	1
Mestizo	1431	0.67	0.47	1445	0.53	0.50	0	1
Education	1431	0.48	0.30	1445	0.49	0.28	0	1
Urban	1431	0.63	0.48	1445	0.63	0.48	0	1
Age	1431	0.38	0.16	1445	0.38	0.16	0.18	0.88
Women	1431	0.51	0.50	1445	0.51	0.50	0	1
Variable	Survey wave: 2014			Survey wave: 2016			Min.	Max.
	N	Mean	Std. Dev.	N	Mean	Std. Dev.	Min.	Max.
Core FMLN supporters	963	0.35	0.48	900	0.14	0.35	0	1
Core ARENA supporters	963	0.17	0.37	900	0.12	0.32	0	1
Left-right ideology	1257	0.47	0.32	1237	0.51	0.30	0	1
Support for democracy	1354	0.66	0.26	1343	0.60	0.27	0	1
Satisfaction with democracy	1369	0.53	0.22	1354	0.45	0.26	0	1
Crime as national priority	1380	0.53	0.50	1371	0.42	0.49	0	1
National economic outlook	1376	0.33	0.34	1367	0.17	0.28	0	1
Pocketbook economic outlook	1378	0.38	0.34	1371	0.33	0.34	0	1
Interest in politics	1378	0.39	0.34	1366	0.35	0.34	0	1
Catholic	1380	0.45	0.50	1371	0.49	0.50	0	1
Evangelical	1380	0.43	0.49	1371	0.35	0.48	0	1
White	1380	0.22	0.42	1371	0.24	0.43	0	1
Mestizo	1380	0.63	0.48	1371	0.45	0.50	0	1
Education	1380	0.48	0.26	1371	0.50	0.26	0	1
Urban	1380	0.64	0.48	1371	0.66	0.48	0	1
Age	1380	0.41	0.17	1371	0.40	0.17	0.18	0.93
Women	1380	0.55	0.50	1371	0.49	0.50	0	1

Source: Authors, based on the AmericasBarometer.

FMLN: Farabundo Martí National Liberation Front; ARENA: Nationalist Republican Alliance.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Navia acknowledges support from ANID-Millennium Science Initiative Program (NCS2021_063) and National Fund for Scientific and Technological Development (#1200317).

ORCID iD

Patricio Navia

Author biographies

Lucas Perelló is a visiting instructor of international studies at Marymount Manhattan College, and a PhD candidate in Politics at The New School. His research focuses on elections, public opinion, and party systems in Latin America.

Patricio Navia is a full professor of political science at Universidad Diego Portales in Santiago, Chile, and a clinical professor of liberal studies at New York University, NY, USA. He has published on elections, public opinion, party systems, and legislative affairs in different Latin American countries. He previously published with Perelló an article in Politics on democratic backsliding in Nicaragua.

References

AmericasBarometer (n.d.) AmericasBarometer by the Latin American Public Opinion Project (LAPOP) datasets home. Available at: http://datasets.americasbarometer.org/database/index.php?freeUser=true (accessed 26 March 2021).

Azpuru

(2010) The salience of ideology: Fifteen years of presidential elections in El Salvador. Latin American Politics and Society 52(2): 103–138.

Baker

Greene

(2015) Positional issue voting in Latin America. In: Carlin

Singer

Zechmeister

(eds) The Latin American Voter. Ann Arbor, MI: The University of Michigan Press, pp.173–194.

Bukele

(2018) No respetan la vida, ni las leyes, ni la Democracia, ni la Constitución, ni el Estado de Derecho. No les importa ningún límite, ni la verdad, ni el debido proceso, ni el derecho del pueblo de elegir libremente en elecciones. Quieren el poder, a cualquier costo [twitter post], 3 June. Available at: https://twitter.com/nayibbukele/status/1014197680388628481 (accessed 24 March 2021).

Carreras

(2012) The rise of outsiders in Latin America, 1980–2010: An institutional perspective. Comparative Political Studies 45(12): 1451–1482.

Dalton

(2012) Nayib Bukele: soy de izquierda radical. Huffington Post [Online]. Available at: https://www.huffpost.com/entry/nayib-bukele-izquierda_n_1882374 (accessed 20 December 2021).

de la Torre

(2017) Populism and democracy: Lessons from Latin America. The Journal of Diplomacy and International Relations 18(1): 33–46.

de Zeeuw

(2010) ‘Sons of war’: Parties and party systems in post-war El Salvador and Cambodia. Democratization 17(6): 1176–1201.

Diamond

(2015) Facing up to the democratic recession. Journal of Democracy 26(1): 141–155.

10.

Downs

(1957) An economic theory of political action in a democracy. Journal of Political Economy 65(2): 135–150.

11.

España-Nájera

(2016) The legislative elections in El Salvador, 2012 and 2015. Electoral Studies 42: 308–310.

12.

Jones

(2011) Weakly institutionalized party systems and presidential democracy: Evidence from Guatemala. International Area Studies Review 14(4): 3–30.

13.

Koivumaeki

(2010) Business, economic experts, and conservative party building in Latin America: The case of El Salvador. Journal of Politics in Latin America 2(1): 79–106.

14.

Levine

Molina

(ed.) (2011) The Quality of Democracy in Latin America. Boulder, CO: Lynne Rienner Publishers.

15.

Levitsky

Cameron

(2003) Democracy without parties? Political parties and regime change in Fujimori’s Peru. Latin American Politics and Society 45(3): 1–33.

16.

Lucas

(2013) Programmatic political competition where we least expect it: Party system development in El Salvador. Paper presented at the annual meeting of the Western Political Science Association, March, pp.28–30. Available at: http://www.wpsanet.org/papers/docs/lucas13.pdf

17.

Lupu

(2016) Party Brands in Crisis: Partisanship, Brand Dilution, and the Breakdown of Political Parties in Latin America. Cambridge: Cambridge University Press.

18.

Mainwaring

(2006) The crisis of representation in the Andes. Journal of Democracy 17(3): 13–27.

19.

Mainwaring

(ed.) (2018) Party Systems in Latin America: Institutionalization, Decay, and Collapse. New York: Cambridge University Press.

20.

Meléndez-Sánchez

(2021) Latin America erupts: Millennial authoritarianism in El Salvador. Journal of Democracy 32(3): 19–32.

21.

Miller

(2011) Why do populist outsiders get elected? A model of strategic populists [online]. IDB working paper no. IDB-WP-248. Available at: https://ssrn.com/abstract=1858063 (accessed 20 March 2021).

22.

Morgan

(2011) Bankrupt Representation and Party System Collapse. University Park, PA: Penn University Press.

23.

Rennó

(2020) The Bolsonaro voter: Issue positions and vote choice in the 2018 Brazilian presidential elections. Latin American Politics and Society 62(4): 1–23.

24.

Roberts

(2014) Changing Course in Latin America: Party Systems in Latin America’s Neoliberal Era. Cambridge: Cambridge University Press.

25.

Romero

(1997) Rearranging the deck chairs on the Titanic: The agony of democracy in Venezuela. Latin American Research Review 32(1): 7–36.

26.

Seawright

(2012) Party-System Collapse: The Roots of Crisis in Peru and Venezuela. Stanford, CA: Stanford University Press.

27.

Supreme Electoral Tribunal (2019) Elecciones. Available at: https://www.v-dem.net/data_analysis/VariableGraph/ (accessed 26 March 2021).

28.

Varieties of Democracy (2021) Variable graph: Political corruption. Available at: https://www.v-dem.net/en/analysis/VariableGraph/ (accessed 26 March 2021).

29.

Voogd

Dassonneville

(2020) Are the supporters of populist parties loyal voters? Dissatisfaction and stable voting for populist parties. Government & Opposition 55(3): 394–370.

30.

Wade

(2016) Captured Peace: Elites and Peacebuilding in El Salvador. Athens, OH: Ohio University Press.

31.

Watts

(2015) One murder every hour: How El Salvador became the homicide capital of the world [online]. The Guardian, 22 August. Available at: https://www.theguardian.com/world/2015/aug/22/el-salvador-worlds-most-homicidal-place (accessed 20 December 2021).

32.

Weyland

(2003) Economic voting reconsidered: Crisis and charisma in the election of Hugo Chávez. Comparative Political Studies 36(7): 822–848.

33.

Wood

(2000) Forging Democracy from Below: Insurgent Transitions in South Africa and El Salvador. Cambridge: Cambridge University Press.

34.

World Bank (2021a) Data: Intentional homicides (per 100,000 people). Available at: https://data.worldbank.org/indicator/VC.IHR.PSRC.P5?locations=SV (accessed 26 March 2021).

35.

World Bank (2021b) Data: GDP per capita growth (annual %). Available at: https://data.worldbank.org/indicator/NY.GDP.PCAP.KD.ZG (accessed 20 December 2021).

36.

Zechmeister

(2015) Left-right identifications and the Latin American voter. In: Carlin

Singer

Zechmeister

(eds) The Latin American Voter. Ann Arbor, MI: The University of Michigan Press, pp.195–225.