Introduction
Private companies are a powerful economic and political actor. However, for decades, business power was understudied. Drawing on the notion of “structural power,” defined as the power derived from capitalists’ control over investment decisions, scholars emphasized how the business class could influence state policies without the need to organize or act collectively.Footnote 2 The implication was that the question of how business elites organize to influence policymaking was thought to be largely irrelevant.Footnote 3
However, despite their powerful position in society, business elites frequently loss their political battles with the state and other organized groups.Footnote 4 Against this backdrop, in the last few decades, scholars have become interested in forms of power beyond structural power and in the more general question of how business elites influence government policies.Footnote 5 For example, analysts have pointed out how businesses’ structural power is usually supplemented by the mobilization of “instrumental” power resources, defined as employers’ deliberate use of “political” means, such as lobbying, campaign donations, and the collective organization, through business associations.Footnote 6 More recently, others have used the concept of “discursive power”Footnote 7 to describe economic elites’ ability “to shape social values, public discourses, and public opinion… in such ways as to create a climate of acceptance and, indeed, support for their interests and agenda.”Footnote 8
Building on the notion of discursive power, scholars have pointed out the need to study legitimacy as a fundamental source of power for the business class. When discursive power is strong, businesses’ privileges and interests are more likely to be deemed legitimate,Footnote 9 and entrepreneurs are more able to justify privilege and influence by encouraging “non-coercive forms of control or domination over society as a whole.”Footnote 10 In other words, discursive power is key for the construction of hegemony in the Gramscian sense of the term, i.e., the construction of images, discourses, and narratives that legitimize business and thereby enable business elites to influence public discussion by transforming profit-making motives into society-wide concerns.Footnote 11
This emphasis on legitimacy as a source of power has allowed scholars to develop a multidimensional approach to power (i.e., an approach that analyzes not only how business groups influence decision-making or limit it to noncontroversial issues but also how they shape the preferences of the subordinated classes to make them legitimize the existing order of things).Footnote 12 However, empirical research on the factors that affect businesses’ legitimacy remains scarce. This is fundamental for understanding to what extent the rest of society accepts (or not) the discourses, narratives, and images framed by the business class to legitimize its power.
In this article, we contribute to this promising research agenda by analyzing how individual- and macro-level factors associated with class politics and power shape people’s propensity to trust private companies. We focus on trust in private companies because it is a central component of business legitimacy.Footnote 13 When citizens trust private organizations, these organizations are more respected and accepted as legitimate actors in society, and citizens are more likely to believe, feel, or expect that they will produce positive social outcomes.Footnote 14
Using data from the Latinobarómetro survey, we focus on 15 Latin American countries over a ten-year period (2005–2015). As we will show below, our data cover a period usually known as the “Pink Tide,” when several Latin American countries were ruled by leftist parties.Footnote 15 Therefore, we study trust in private companies in contexts in which, except perhaps for the cases of Colombia, Mexico, Paraguay, and Panama, left or center-left governments challenged business power and, to varying degrees, implemented redistributive, “post-neoliberal” reforms.Footnote 16
Integrating the literatures on confidence in institutions and organizationsFootnote 17 and on class and power resources,Footnote 18 we argue that businesses’ success in building discursive power, as expressed in individual trust in private companies, is limited by structural (class-based) and political (power resource) factors that operate at the micro and macro levels.
At the micro (individual) level, it is limited by structural mechanisms associated with people’s class location as well as political mechanisms associated with people’s political orientations and attitudes toward political institutions. Using logistic regression models, we find that these mechanisms explain why, despite business elites’ efforts to mobilize discursive power, trust in companies is lower among those who are in underprivileged positions in the class structure (e.g., working-class people or the informal self-employed) and among those who identify with the political left and have less confidence in political institutions.
At the macro (contextual) level, businesses’ success in building discursive power is limited by structural factors associated with countries’ levels of income inequality and by power resource factors associated with class politics and power. Using a three-level multilevel modeling strategy that allows us to distinguish between cross-sectional and longitudinal effects, we show that trust in private companies is significantly lower in countries ruled by the left or with low levels of inequality. We explain these findings by arguing that, in the period studied, the arrival of progressive governments aiming to implement redistributive reforms undermined business legitimacy. Likewise, we explain the positive cross-sectional association between income inequality and trust in business by showing how, in highly inequal societies, the business class can strengthen its discursive power at the expense of political institutions. On the other hand, our analysis of longitudinal effects shows that trust in private companies declines in countries where inequality rises over time or where citizens’ trust in political institutions increases. We argue that this is because, during the Pink Tide, business legitimacy declined in countries where inequality increased, and because the re-legitimation of political institutions was achieved in opposition to the power of business elites.
At the end of this article, we show how our analysis enables us to identify three patterns of business legitimacy in Latin America. The first pattern can be observed in the countries with the highest levels of trust in private companies (e.g., Dominican Republic, Paraguay, and Panama), where inequality was comparatively high and left parties were weak. The second pattern can be observed in the countries with the lowest levels of trust in private companies (e.g., El Salvador, Chile, and Argentina), where leftist parties were comparatively stronger, inequality was below average, and except for Chile institutional trust increased (at least between 2005 and 2010). Finally, the third pattern includes countries with intermediate levels of confidence in private companies such as Uruguay, Guatemala, Peru, Nicaragua, and Mexico. Our analysis suggests that these countries had some characteristics that may have positive effects on business legitimacy (e.g., weak left parties in Peru and declining institutional trust in Uruguay) and, at the same time, other characteristics that may have had negative effects on it (e.g., increasingly strong left parties in Peru and low inequality and strong left parties in Uruguay).
To conclude the article, we reflect on how our findings contribute to the recent research on trust, class, and power.
Legitimacy and trust as a source of power for business
When the business sector combines structural, instrumental, and discursive forms of power, governments are less able to pass redistributive reforms, even when the population demands it.Footnote 19 Structural, instrumental, and discursive power resources exist in a mutually dependent relationship. For instance, discursive power is stronger when structural power is strong (i.e., when firms’ economic prominence reinforces the importance attributed to business ideas)Footnote 20 or when instrumental power is strong, particularly when business associations can mobilize discourses and narratives to gain legitimacy vis-à-vis the public and the state.Footnote 21
However, discursive power is more than simply a result of structural or instrumental power. Discursive power has enhancing effects on the other forms of power, and it becomes crucial when the structural or instrumental power resources are comparatively weaker. Madariaga et al. (Reference Madariaga, Maillet and Rozas2021) demonstrate that discursive power enabled Chilean agribusiness elites to influence state policies by portraying any attempt to reform Chile’s neoliberal water code as harmful to the country, even though their position in national and international markets was subordinated and their structural power was relatively low. Along the same lines, Pérez Ahumada (Reference Pérez Ahumada2023a) shows that the mobilization of neoliberal and entrepreneurial narratives is a basic precondition for the successful construction of business associational power (which, according to the author, is a central component of instrumental power). Pérez Ahumada shows that discursive power allows business associations to condense class identities, promote collective action, and build alliances with small and medium-sized employer associations. Similarly, Crabtree et al. (Reference Crabtree, Wolff and Durand2023) show that, compared to their counterparts from Bolivia and Ecuador, Peruvian business elites have had more power to oppose government redistribution, not only because they have strong structural or instrumental power but also because they have been able to remain legitimate to society. This explains why the Peruvian middle classes have remained a reliable supporter of pro-business agendas and why, in the early 2010s, Peru remained one of the few countries in the region that kept the policy prescriptions implemented during the Washington Consensus.
Class politics, power, and business legitimacy
All in all, the abovementioned literature suggests that legitimacy, rooted in the mobilization of discursive power resources, is fundamental for business to advance its interests vis-à-vis the state and other actors in society. That said, in this article, we argue that business elites’ legitimacy is not stable or uncontested. Instead, business legitimacy is limited by individual- and macro-level factors that shape citizens’ propensity to trust private companies. Building on the literatures on social class and institutional trust as well as the research espousing the power resources approach (PRA), we argue that this is because, at the individual level, trust in private companies depends on business elites’ ability to shape people’s economic (class-based) and political interests. At the contextual level, business legitimacy also depends on the extent to which business elites can deploy discursive power to resist antibusiness discourses framed from below by social movements (e.g., organized labor) and mobilized politically by leftist governments and parties with the goal of reducing inequality via redistribution.
Individual-level determinants of trust in private companies
Trust in institutions or organizations is the result of people’s expectations, feelings, and evaluations regarding the ability of such institutions or organizations to produce positive outcomes.Footnote 22 At the individual level, these expectations are based on structural mechanisms associated with people’s position in the class structure, and political (ideational and normative) mechanisms associated with how people assess the role of institutions or organizations in society.Footnote 23
In relation to the first type of mechanism, socioeconomic position is a significant predictor of trust. Upper-income people trust more in political institutions (e.g., the government) than the poor because they perceive more benefits from them.Footnote 24 Likewise, people with higher income or who are more financially stable trust more in financial organizations, such as banks and pension fund companies, because they have more frequent interaction with these organizations, perceive more benefits from them, or have higher levels of general trust.Footnote 25
These studies’ emphasis on socioeconomic position reinforces the evidence from the literature on social class. This literature demonstrates that relative to those who are in a privileged position in the class structure (e.g., employers and high-level managers), working-class people are more likely to endorse redistributive policies,Footnote 26 have greater trust in labor unions,Footnote 27 or perceive strong conflicts between workers and managers.Footnote 28 In this literature, the concept of class refers to people’s position in relation to production or employment.Footnote 29 Therefore, in defining individuals’ subordinate (i.e., dominated or exploited) or dominant (i.e., managerial or exploiting) roles in the economic system, class position shapes not only people’s levels of job autonomy, exposure to market risks, and income levels, but also their material interests, as expressed in their sociopolitical attitudes.Footnote 30
That said, research on trust in organizations shows that this is also shaped by political and normative mechanisms. This research shows, for example, that due to the historical linkages between labor unions and leftist parties, trust in unions is significantly greater among those who identify with the left or have collectivist orientations.Footnote 31 The identification with the left is also associated with progressive socioeconomic values—e.g., the belief that government should reduce income differences or that workers need strong unions to protect their interests.Footnote 32 These investigations certainly do not measure trust in private companies. Unlike labor unions that represent the collective interests of workers, private companies are organizations mainly concerned with the private interests of their owners. However, the existing research on trust in financial organizations, such as banks, is consistent with this evidence. This research shows that trust in banks is greater among those who have right-wing (pro-market) orientations, probably because they have political values that place more importance on individualism and competition than on government redistribution.Footnote 33
Recent research also suggests that trust in organizations can also be shaped by institutional trust. For example, confidence in financial organizations is positively correlated with confidence in nonfinancial institutions.Footnote 34 In a recent article, DowFootnote 35 finds that although general trust (i.e., trust in others) does not affect confidence in financial institutions and banks, institutional trust contributes positively and significantly to it. He argues that this may be because people’s confidence in institutions transfers to financial institutions.
Based on this evidence, we hypothesize that business groups’ success in building discursive power, as expressed in people’s trust in private companies, varies depending on people’s class position and political orientations. In other words, we hypothesize that
H1a: Trust in private companies is lower among people in underprivileged classes (e.g., working-class people and the informal self-employed) than among the members of privileged classes (e.g., employers and high-level managers).
H1b: Trust in private companies is lower among those who identify with the political left than among those who identify with the center or the right.
H1c: Trust in private companies is greater among those who have greater institutional trust.
Cross-national variation in trust in private companies
In the last few decades, empirical studies have been devoted to studying cross-national variations in attitudes toward institutions, actors such as labor unions, and to a much lesser extent, business groups. Similar to research on the micro-level determinants of trust, these studies show that cross-national variations in trust are explained by structural factors usually associated with economic processes (e.g., economic development and income inequality), and by power resource factors associated with class politics and power (e.g., the strength of organized labor and of left parties).
Empirical studies show that citizens’ confidence in institutions tends to be lower in more unequal societiesFootnote 36 and in countries suffering economic downturns or crises.Footnote 37 Economic crises also affect trust in financial organizations because people’s perception of their financial situation worsens, thereby undermining these organizations’ reputations.Footnote 38
In line with this evidence, research suggests that higher levels of economic inequality increase not only the perception of class conflictFootnote 39 but also citizen demand for government intervention.Footnote 40 Therefore, we hypothesize that
H2a: The higher the level of inequality, the lower the trust in companies.
The abovementioned discussion indicates that contextual-level factors such as inequality and economic decline affect individual confidence in institutions. This, in turn, might undermine citizens’ confidence in organizations such as private enterprises.Footnote 41 However, studies focusing on class-based political conflict suggest that, in certain political contexts, the macro-level relationship between institutional trust and trust in private companies might not follow this pattern. As our discussion on business discursive power suggested, business elites’ legitimacy depends on their ability to convince society that they can successfully improve people’s lives and that, therefore, they are necessary for social prosperity.Footnote 42 In Latin America, where inequality is high and institutional trust is low,Footnote 43 economic elites usually mobilize discursive power against governments and political actors such as left parties. Crabtree et al. (Reference Crabtree, Wolff and Durand2023) argue that this is what happened during Latin America’s left-wing administrations of the early 2000s. They show, for instance, that the accusations of corruption leveled against former Ecuadorian left-wing president Rafael Correa enhanced business legitimacy because they helped Ecuadorian entrepreneurs mobilize neoliberal, anti-state discourses that had lost popular support during the Correa administration.Footnote 44
This evidence implies that at the country level, the relationship between institutional trust and trust in private companies might be negative. Therefore, our second macro-level hypothesis is that in Latin America,
H2b: The higher the country’s level of institutional trust, the lower the trust in companies.
Trust in companies can also be affected by other class-based dynamics, expressed in the conflict between labor unions and business associations, and between the latter and left parties.Footnote 45 As is well known, scholars espousing the PRA argue that political conflict reflects antagonism between classes (e.g., between workers and employers) and between the political parties representing their interests—most notably between left and right parties.Footnote 46 Therefore, when linkages between left-wing parties and organized labor are strong, redistributive social policy is more likely to be implemented and class inequality is less pronounced.Footnote 47 Conversely, when right-wing parties and/or the business class are more powerful, redistributive policies are less developed, and class disparities are larger.Footnote 48
The argument developed by power resource scholars is key for understanding how, in the last few decades, sociopolitical conflict in Latin America has affected business legitimacy. In the 1990s, countries such as Argentina, Bolivia, Brazil, Ecuador, Venezuela, and Uruguay witnessed intense anti-neoliberal mass protests that culminated, between the late 1990s and early 2000s, in the election of leftist presidents.Footnote 49 This inaugurated a period usually known as the Pink Tide, in which left-wing parties reconstructed their historical linkages with the popular sectors, and these sectors were reincorporated into the political system.Footnote 50
These developments affected business elites’ capacity to monopolize political influence because business groups found themselves forced to contend with competing social actors, such as labor unions and new collective actors formed by indigenous people’s movements, urban squatter movements, and environmental movements.Footnote 51 Business structural power was also undermined during the Pink Tide. Taking advantage of the favorable conditions produced by the commodity boom, progressive governments achieved greater autonomy vis-à-vis the capitalist class and implemented redistributive reforms despite business opposition.Footnote 52 Furthermore, left-wing governments and social movements alike called into question the legitimacy of the discourses and narratives framed by business elites during the transition to neoliberalism to justify their power and the “natural” association between private initiative and development.Footnote 53 In this new context, progressive governments successfully increased taxes on the rich and implemented redistributive reforms to benefit the poor.Footnote 54 In countries such as Argentina and Uruguay, governments also overcame business opposition and passed reforms to empower unions vis-à-vis employers.Footnote 55
The governments’ commitment to dismantling neoliberal institutions differed greatly by country. So did the level of confrontation between governments and business elites. However, differences aside, during the Pink Tide, the balance of power became more favorable the the popular classes, especially compared with the neoliberal period.Footnote 56 Even in Chile, where the center-of-left governments were much less antagonistic to the business class, business elites lost credibility after anti-neoliberal protests erupted in the early 2010s.Footnote 57
Quantitative empirical research suggests that these sociopolitical developments reflecting the balance of power between classes might affect individual attitudes toward private enterprises. In countries where labor unions and leftist parties are stronger, class attitudes toward redistribution are more polarized because the debate on inequality is politicized, and workers are induced to interpret their experiences in class terms.Footnote 58 Adding weight to this argument, Kelley and EvansFootnote 59 find that in these countries, which they refer to as “strong welfare states,” citizens are significantly more likely to support policies to “level down” the rich via wealth redistribution.
Based on this evidence and the discussion on Latin America’s Pink Tide, we hypothesize that
H2c: The stronger the power of left-wing parties, the lower the trust in companies.
H2d: The stronger the power of trade unions, the lower the trust in companies.
Data and methods
Data
This study uses data from the Latinobarómetro survey. Due to data availability, we restricted our analysis to three years (2005, 2010, and 2015) and to the following 15 countries: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, El Salvador, Guatemala, Mexico, Nicaragua, Panama, Paraguay, Peru, and Uruguay. After listwise deletion, our final sample contained 23,407 individuals. Considering the objectives of this work, our sample included only the respondents who were employed and who could therefore be classified in a class position.
Methods
In this article, we followed a two-step strategy. First, we fit binary logistic regression models with country and year-fixed effects. This allowed us to analyze the association between individual-level predictors and trust in companies, controlling for other relevant individual attributes as well as for unobserved country and year characteristics. The country and year-fixed effects also allowed us to estimate whether trust in companies vary cross-nationally and over time.
Second, to analyze the contextual factors that account for cross-national differences in trust in companies, we used multilevel logistic regression models to estimate several random-intercept models with individual and country-level variables.Footnote 60 Considering the objectives of this research, in preliminary analyses, we also fit models with random slopes to test whether the relationship between the independent variables of interest (e.g., social class) and trust in private companies varied across countries. These analyses suggested that adding random slopes for the independent variables did not improve the model fit. In theoretical terms, this implies that the relationship between these variables and trust in companies does not differ across countries. Therefore, in the next section, we only report the results of the random-intercept models.
Multilevel modeling requires adequate country samples to produce reliable estimates of “country effects” and accurate estimates for the fixed effects and variance components.Footnote 61 This is a challenge because, for this article, we used data from only 15 countries. Therefore, to increase the size of the country sample, we followed prior researchFootnote 62 and used three rounds of the Latinobarómetro survey: 2005, 2010, and 2015.
This design provided us with two advantages. First, it allowed us to fit multilevel models on 23,407 individuals (level 1) nested in 45 country-years (level 2) and 15 countries (level 3). Second, this 3-level design allowed us to distinguish between cross-sectional relationships and longitudinal relationships.Footnote 63 To do so, we followed Fairbrother’sFootnote 64 suggestions and input the contextual variables into the models in two ways. First, to capture cross-sectional relationships, we input the means of the country-level (level 3) variables for each country. These variables, which we labeled “Mean” in the regression tables, indicate the average score of each contextual attribute spanning the three survey years. Second, to capture longitudinal relationships, we input the country-year variables (level 2). These variables, which we labeled “Difference” in the tables, were calculated as the difference between a country’s score for each of the three years and the country’s average value for the three years.Footnote 65
These three-level multilevel models are highly complex. They required heavy computation and some models (e.g., those estimating the effect of country-level institutional trust and trade union power) had convergence issues. To address these issues, we rescaled the country-level variables (see below), and, for the two models with convergence issues, we optimized the estimation of the coefficients by setting to 0 the number of quadrature points for the adaptive Gauss-Hermite approximation to the log-likelihood (i.e., we set nAGQ = 0 in the function glmer of R’s lme4 package).Footnote 66 This might have affected the precision of parameter estimations.Footnote 67 To check the robustness of our results, we ran all the multilevel logistic models presented in this article with nAGQ=0 and nAGQ=1 (the default setting in glmer). We found little to no difference in the results. This suggests that, for our sample, optimizing the estimation of the coefficients did not affect their precision. In further robustness checks, we also fit multilevel linear probability models, which produced nearly identical results as the models reported here (see Table A1, Appendix).
All data and codes used in this article are available in the Open Science Framework (OSF) repository: https://osf.io/cehgm/
Dependent variable
The dependent variable is “Trust in private companies.” The Latinobarómetro survey measures this variable by asking the following question: “How much trust do you have in each of the following groups, people, or institutions? Would you say you have a lot (1), some (2), a little (3), or no trust (4)?” In this article, we focus on the “Private companies” option. Following similar investigations,Footnote 68 we recoded this variable into a dichotomous outcome: 0 = a little or no trust; 1 = some or a lot of trust in private companies.
Individual-level independent variables
The first level-1 independent variable is social class. Drawing upon the class distinctions proposed by WrightFootnote 69 we created an 8-class scheme based on four variables included in the Latinobarómetro survey. Using the respondents’ current employment situation, we distinguished between salaried workers, employers, and the self-employed. Additionally, using the respondents’ type of employment, we divided salaried employees into managers and nonmanagerial workers. Following prior research,Footnote 70 we also used this variable to divide the self-employed into those who belonged to the “formal petite bourgeoisie” (i.e., the professional self-employed, such as doctors, lawyers, and accountants) and those who belonged to the “informal self-employed” class (that is to say, nonprofessional and less-educated independent workers).
Finally, we used two other variables to differentiate the class categories more clearly. First, we used the respondents’ subjective income to distinguish between large and small employers. The employers who reported that their income was sufficient and allowed them to save money were labeled as “Large employers.” The rest of the respondents in the employer class were simply defined as “Small employers.” We did so because the Latinobarómetro survey does not have a question that identifies the size of the company owned by those classified as “employers.” Second, we used educational levels to separate the salaried classes into the categories of “Expert managers” (i.e., managers with complete tertiary education or more) and “Supervisors” (i.e., managers without professional degrees). We also used this education-level variable to reclassify some inconsistent cases among the self-employed, e.g., cases that had been defined as part of the “Formal (professional) petite bourgeoisie” but whose educational level was low. Taking all these criteria into account, our class scheme contains a total of eight class positions (see Table 1).
Descriptive statistics (N = 23,407)
The second individual-level independent variable is political position. This was measured through four categories indicating the respondents’ self-identification: “Left” (scores 0–3 on the original interval scale), “Center” (scores 4–6), “Right” (scores 7–10), and “No political position” (DK/NA).
Finally, the third individual-level independent variable is trust in institutions. This was measured through a 0–15 interval scale constructed through the summation of five variables indicating the respondents’ trust in (a) the judiciary, (b) the police, (c) the congress, (d) the government, and (e) political parties (0. no trust to 3. a lot of trust). Higher scores indicate more trust. This scale is internally consistent and unidimensional (Cronbach’s alpha = 0.79; Factor Analysis [Principal Axis Method]: Eigenvalue = 2.23; Explained variance = 45 percent).
Following prior research,Footnote 71 our regression models included two sociodemographic controls: gender (0 = male; 1 = female) and age, measured in five ordinal categories (16–24, 25–34, 35–44, 45–54, and 55 years or more). Additionally, they included perceived economic situation, measured through a 1–5 interval scale indicating the respondents’ perception of the country’s present economic situation (1. very good to 5. very bad).
Country-level independent variables
The first contextual variable of interest is leftist government. This was measured by a dichotomous variable that indicates the party orientation of the government’s chief executive officer (1 = yes, that is, parties defined as socialist, communist, social democratic, or left-wing; 0 = all others). The data for this variable were obtained from the DPI 2020 Database of Political Institutions. The second variable is income inequality, and it was measured by the Gini index.Footnote 72 The third variable is trade union power and was measured by means of the net union density rate.Footnote 73 Finally, the fourth contextual variable is country-level institutional trust. We constructed this variable by calculating the mean of the 0–15 scale of individual institutional trust (see above) for each country in each year.
Due to the limited number of level 2 units, the contextual variables were analyzed in separate models. In these models, we included GDP per capita constant at 2015 US$ as controlFootnote 74 to analyze whether the effect of the contextual variables remains after controlling for differences in economic development/affluence.
All the country-level variables were transformed into Z-scores before being input into the models. In addition to facilitating model convergence, the transformation into Z-scores simplified comparisons across models. Finally, all models included year dummies—2005, 2010, and 2015 (reference category = 2005)—to control for unobserved time trends.
Results
Table 2 reports the results of four logistic regression models with country and year-fixed effects. The coefficients are presented in log odds and odds ratios. The coefficients for social class suggest that people in subordinated classes—particularly, nonmanagerial workers and the informal self-employed—are less likely to trust private companies than the members of the reference category (employers). For instance, Model 4 shows that, ceteris paribus, the odds that a person belonging to the informal self-employed class trusts private business are 34 percent lower ([1 – 0.66] *100) than the odds of someone in the employer class (p-value < 0.001). Likewise, the odds that someone in the working-class trusts private companies are 19 percent lower than someone in the employer class ([1 – 0.81] *100) (p-value < 0.1). These results and the results from the other models in the table are consistent with H1a. That said, Table 2 suggests that low levels of trust are also prevalent among some segments of the Latin American middle class—particularly, the small employers. Model 4 indicates, for example, that the odds that a small employer trusts private companies are around 20 percent lower than the odds for an employer.
Logistic Regression Models predicting trust in private companies in Latin America, 2005–2015 (standard errors in parentheses)
Note: The following variables were included in the models but not reported in this table: gender, age, and perceived economic situation (see main text for definitions).
***p < 0.001;
**p < 0.01;
*p < 0.05;
†p < 0.1.
As for individual political position, the models suggest that the odds that someone who identifies with the Center or the Right trusts private companies are, respectively, 25 and 30 percent higher than the odds of someone self-identifying as leftist (Model 4). In all models, these differences are statistically significant, which means that H1b should be accepted.
Finally, the models that include institutional trust indicate that it is positively correlated with trust in private companies. In line with H1c, Model 4 shows that an increase of 1 point on the 0–15 scale of institutional trust is associated with a rise of nearly 24 percentage points in the odds of trusting private companies (p-value < 0.001).
To explore whether trust in companies varies cross-nationally and over time, Model 4 includes interaction effects between year and country (not reported in the table). To represent graphically these interaction effects, we used the ggeffects package in RFootnote 75 and calculated the predicted probability of trusting private companies. The probabilities are presented in Figure 1. The figure shows that there are significant cross-national variations in trust in companies and that, in some countries, trust in companies has also changed over time. More specifically, in countries such as Costa Rica, Panama, Paraguay, and Dominican Republic the probability of trusting private companies is the highest in the region. In this group, Costa Rica and Panama stand out for the decline in the probability of trusting companies. In Costa Rica, for example, these probabilities decreased from 65 percent in 2005 to 50 percent in 2015. In contrast, in countries such as El Salvador, Chile, and Argentina, the probabilities of trusting companies are the lowest among all countries examined. In El Salvador trust in companies seemed to have risen somewhat, especially between 2010 and 2015, whereas in Chile it seemed to have decreased consistently over the three time points. As a result, in 2015 Chile was the country with the lowest levels of trust in private companies.
Predicted probabilities of trusting in private companies for the interaction between year and country.
To analyze the factors that account for these cross-national divergences, we estimated several multilevel logistic regression models. The results are presented in Table 3. Following the standard practice, we first fit an empty model (Model 0) and calculated the intra-class correlation.Footnote 76 The empty model suggests that around 4 percent of the total variability in the chance of trusting companies is between countries, whereas only 1 percent of that variation can be explained by changes over time. This implies that between-country variation in trust is greater than the variation over time.
Multilevel logistic models predicting trust in private companies in Latin America, 2005–2015 (standard errors in parentheses)
Note: The following variables were included in the models but not reported in this table: social class, gender, age, political position, perceived economic situation, individual-level trust in institutions, survey year, and GDP per capita (see main text for definitions).
***p < 0.001;
**p < 0.01;
*p < 0.05;
†p < 0.1.
Model 1 presents the coefficients for country-level inequality. Contrary to H2a, the “mean” coefficient indicates that there is a positive cross-sectional relationship between inequality and trust in business. In countries where inequality is greater, the odds of trusting companies are higher by nearly 19 percent. Nevertheless, consistent with the hypothesis, in this model the “difference” coefficient is significant and negative (p-value < 0.05), suggesting that in countries where inequality increases over time, the odds of trusting companies decline by about 10 percent. This means that H2a can only be partially accepted.
The effect of country-level institutional trust is tested in Model 2. The “mean” coefficient is positive but insignificant. By contrast, the “difference” coefficient is statistically significant and negative, suggesting that in countries where institutional trust rises, the odds of trusting companies decrease by around 8 percent. The implication is that the longitudinal relationship between institutional trust and business legitimacy is negative, which is in line with H2b.
Model 3 presents the coefficients for leftist government. Consistent with H2c, the “mean” coefficient indicates that when leftist parties are in power, the odds of trusting companies decrease by 37 percent (p-value < 0.05). On the other hand, the lack of significance of the “difference” coefficient suggests that, in Latin America, the relationship between leftist party power and trust in private companies does not hold longitudinally. In other words, the variation in the trust in companies over time might not be explained by a change in the relative power of left-wing parties.
Finally, Model 4 shows the “mean” and “difference” coefficients for trade union density. Although the sign of these coefficients is as hypothesized (negative), neither is statistically significant. Furthermore, we performed a deviance test that showed that this model did not significantly improve the fit to the data, compared to a model with only individual-level variables (p-value = 0.817). This implies that H2d cannot be accepted.
Discussion
Trust in private companies is a central component of discursive power. When private companies are trusted, businesses’ privileges and interests are more likely to be deemed as legitimate and citizens are more likely to believe or expect that private initiative will produce positive social outcomes.Footnote 77 Discursive power enables entrepreneurs to build hegemony (i.e., shape social values, public discourses, and public opinion to create favorable political and economic conditions).Footnote 78 However, despite the strong power of Latin American business elites, in this region, the construction of business hegemony has been far from stable and uncontested. During the Pink Tide, the business class found itself forced to contend with competing social actors, such as labor unions, social movements, and progressive governments, that to varying degrees challenged neoliberalism and businesses’ privileges.Footnote 79
In this article, we have argued that in such contexts, businesses’ attempts to gain legitimacy are restricted by people’s individual class-based and political interests and by businesses’ ability to face antibusiness discourses that usually emerge in periods of popular mobilization. This occurs because business elites’ success in building discursive power, as expressed in people’s trust in private companies, is limited at the individual level by structural mechanisms associated with individual class position and by political mechanisms reflected in people’s political orientations and in their attitudes toward political institutions. Businesses’ efforts to build discursive power are also limited at the macro level by structural factors associated with countries’ levels of income inequality and power resource associated with class politics and power, such as the countries’ levels of institutional trust and the strength of leftist parties.
Our findings largely support our argument. As for the individual-level determinants of trust, we find that those who are subordinated in production and employment relations (e.g., who are working-class or informal self-employed) are less likely to trust private companies than those who have a privileged position in those relations (e.g., employers and experts). Building on the literature on social class, we argue that this occurs because class position defines people’s exposure to market risks and access to economic welfare and economic power (i.e., control over surplus product).Footnote 80 In doing so, class location shapes people’s material interests and their attitudes to inequality and class conflict.Footnote 81 This explains why those who are in subordinated class positions (i.e., those who are excluded from production profits and are more exposed to market risks) are significantly less likely to believe that private companies are legitimate actors in society.
That said, our statistical analyses indicate that in Latin America, low levels of trust are also prevalent among some segments of the middle class, particularly small employers. Although our dependent variable does not necessarily measure political attitudes, this result is surprising considering that in industrialized nations the self-employed are argued to be a “conservative” class due to their higher levels of job autonomy.Footnote 82 Our data imply that this view should be nuanced. In Latin America, an important portion of the independent middle class is precarious because of its vulnerability to poverty and its lack of social protections.Footnote 83 Furthermore, in some countries, small employers have strong working-class identities.Footnote 84 Although more research is needed in this area, our data indicate that in contexts of high economic disparity, the attitudes of the independent middle class can be more similar to those of the working class than those of, say, employers.
Our findings also show that trust in companies is lower among those who identify with the political left. Along with reinforcing the argument that trust in organizations is shaped by people’s political views,Footnote 85 this finding emphasizes how political identities are closely associated with normative orientations related to wealth inequality and redistribution.Footnote 86 This is key for explaining why those who identify themselves with the left endorse collectivist and pro-redistributive views and, according to our data, are less likely to trust private companies. This also helps explain why those who identify with the right have stronger individualist and pro-market orientations and therefore stronger probusiness attitudes. This finding is worth noting because during Latin America’s neoliberal era, probusiness orientations permeated center-of-left and these parties abandoned class politics.Footnote 87 Against this backdrop, our data suggest that, at least during the Pink Tide, the left-right cleavage remained important for explaining individual differences in attitudes toward private companies. This suggests, in turn, that the reconstruction of alliances between parties and social movements during the Pink Tide may have had significant effects on people’s political identity. Further research relying on data covering the period before and during the Pink Tide can show whether this conclusion holds true.
As for the third individual-level determinant analyzed in this article, institutional trust, we found that it is positively correlated with trust in private companies. To our knowledge, researchers have not systematically analyzed the relationship between trust in political institutions and in private companies. Researchers have studied whether generalized trust (i.e., trust in others) or institutional trust is correlated with confidence in financial organizations, such as banks.Footnote 88 The positive coefficients found in our regression models contribute to this research by showing that at the individual level, there may be a transference of confidence from political institutions to private companies. Our data certainly do not allow us to rule out the possibility of reverse causality. Therefore, our results may also indicate a process of mutual reinforcement between confidence in political institutions and confidence in private companies. Further research relying on longitudinal data can clarify this issue.
Focusing on the macro-level determinants of trust in private companies, our multilevel logistic regression models showed a positive cross-sectional relationship between inequality and trust in private companies. This result can be explained by how, in highly unequal societies, business groups can strengthen their discursive power at the expense of political institutions. In Latin America, inequality is commonly associated with a decline in citizens’ trust in governments and the parliament.Footnote 89 Furthermore, perceived corruption is high and governments’ capacity to promote sustainable prosperity is frequently called into question.Footnote 90 In these contexts, discursive power can help business groups present themselves as the only social actor capable of ensuring economic affluence and prosperity. Consistent with this explanation, Franetovic and Castillo (Reference Franetovic and Castillo2022) find that in Latin America, individual support for redistribution is lower in more unequal countries. This suggests, again, that in a context of high inequality, little confidence in public institutions, and weak institutions, business elites may have more opportunities to exercise their hegemonic power.
In spite of this, our multilevel models showed a negative longitudinal association between inequality and trust in private companies. In other words, trust in business declines in countries where inequality increases over time. The implication is that although business discursive power can be stronger in more unequal countries, business legitimacy declines over time when inequality does not diminish. This is probably a result of how the public opinion perceives the persistence of economic disparities as an effect of businesses’ unwillingness to contribute to equality. During the Pink Tide, this occurred because in contrast to what has been shown in Europe, the existence of more organized business classes restricted rather than facilitated governments’ power to implement redistributive policies.Footnote 91 We believe that this finding should not be overlooked. In theoretical terms, it suggests that the abovementioned hegemonic power of Latin American business elites is challenged from below when citizens come to realize over time that business power does not translate into more equal societies. In methodological terms, it implies that distinguishing between cross-sectional and longitudinal effects is crucial for identifying the mechanisms behind the relationship between inequality and confidence in institutions or groups.Footnote 92
This conclusion is further reinforced if we consider the negative longitudinal effect of country-level institutional trust. This negative effect suggests that in countries where institutional trust increases over time, trust in companies declines. In line with the PRA, the political conflicts observed during the Pink Tide are crucial for explaining this finding. During the period analyzed in this article, progressive governments not only implemented redistributive policies to benefit the poor but also made efforts to integrate the popular classes into the political system.Footnote 93 These efforts increased the tension between governments and business elites.Footnote 94 The level of business-government conflict varied significantly by country (e.g., compared to countries such as Bolivia and Ecuador, such conflict was less intense in Brazil and especially in Chile, where redistributive agendas were much less radical). However, with the partial exception of Chile, in these countries, the implementation of redistributive policies increased citizens’ trust in political institutions throughout the 2000s and part of the 2010s.Footnote 95 Our findings contribute to this argument by suggesting that during the Pink Tide, increases in institutional trust produced a decline in the confidence in private companies, probably because the re-legitimation of political institutions was achieved in opposition to the power of business elites, which frequently mobilized against redistribution.
This finding emphasizes the need to distinguish between individual and contextual levels of analysis. Doing so allows us to explain why, at the individual level, institutional trust is positively correlated with trust in private companies whereas at the contextual level, the (longitudinal) relationship is negative.
As for the effect of leftist government partisanship, we found a negative cross-sectional effect of leftist government partisanship, which means that in countries ruled by left-wing governments, trust in companies is lower. From a power resource perspective, this suggests that, at least during the period studied, the arrival of progressive governments aiming to implement redistributive reforms undermined business legitimacy. As stated above, this is probably because in these contexts, business power is contested and as business groups oppose governments’ redistributive agenda, citizens come to realize that business interests are at odds with people’s desires for social change and equality.
Consistent with the PRA, we also found that the sign of the coefficients for trade union density was negative. However, these coefficients were not statistically significant. Therefore, we could not draw strong conclusions regarding the effect of union power on business legitimacy.
In spite of this, these results allow us to explain cross-national variations in business legitimacy from a class-based, power resource perspective. Based on these results, we can identify three patterns of business legitimacy in Latin America. The first pattern includes countries where business legitimacy is consistently high, such as the Dominican Republic, Paraguay, Panama, Costa Rica, and to a lesser extent Brazil and Colombia. In the period studied, these countries had high levels of inequality and lacked electorally successful left parties, except for Brazil and to some extent Costa Rica. In this group, Costa Rica and Panama stand out because although business legitimacy remained high, it declined over time (see Figure 1). According to our data, this may have occurred because, in both countries, institutional trust increased at at least one point and because, in Costa Rica, inequality rose.
The second pattern corresponds to countries where business legitimacy was the lowest among all countries (El Salvador, Chile, and Argentina). Throughout the period studied, these countries were ruled mostly by leftist parties, inequality was below average, and except for Chile, institutional trust increased between 2005 and 2010. In this group, Chile also stands out because trust in companies seemed to have declined consistently over time (Figure 1). Our data suggest that in Chile, this trajectory is explained not because of an increase in citizens’ trust in political institutions but because of the resurgence of social movements and leftist parties in the early 2010s. This resurgence resulted in the election of a center-left coalition in 2013, which implemented redistributive reforms that were strongly resisted by the business class.Footnote 96
Finally, the third pattern includes the countries with intermediate levels of business legitimacy (Uruguay, Guatemala, Peru, Nicaragua, and Mexico). According to our data, all these countries had some characteristics that may have positive effects on business legitimacy and, at the same time, other characteristics that may have had negative effects on it. For instance, Peru’s leftist parties were weak, and throughout the whole period under study, inequality declined. This might have reinforced trust in private companies. However, by the mid-2010s, leftist parties became more influential and trust in political institutions rose, which in turn might have deteriorated business legitimacy. Likewise, in Uruguay, the left remained powerful and inequality low, which might have been detrimental to business legitimacy. However, between 2010 and 2015, institutional trust declined, which may have increased trust in private companies.
Conclusion
In this article, we studied trust in private companies in Latin America between 2005 and 2015 with the goal of analyzing how the business class builds discursive power in contexts of high inequality and political conflict. The period studied allowed us to analyze how business hegemony is contested by actors such as social movements and leftist governments that, to varying degrees, tried to challenge neoliberalism and businesses’ privileges.
Using multivariate regression techniques, we argued that business groups’ success in building discursive power, as expressed in individual trust in private companies, is limited by structural, class-based mechanisms, and by political mechanisms. These mechanisms explain why despite businesses’ ideological work, trust in companies is significantly lower among those who are in subordinated classes (e.g., in the working class or in the informal self-employed class), who identify with the political left and have less confidence in political institutions. Likewise, using multilevel modeling strategies, we showed that business groups’ success in increasing their legitimacy is restricted by structural factors associated with countries’ levels of inequality and power resource factors associated with class politics and power. In the period studied, these factors explained why trust in private companies was significantly lower in countries ruled by the left, where inequality rose or where citizens’ trust in political institutions increased.
These findings have important implications for the study of business power in Latin America. They suggest that, compared to other forms of power (particularly, structural power), business’s discursive power is more susceptible to contestation by left governments or mobilization from below. This certainly does not mean that discursive power is irrelevant or that scholars should abandon it. It means that discursive power should be seen as a secondary form of powerFootnote 97. The political implication is that progressive governments could face business opposition to redistribution by mobilizing narratives to counteract business discursive power (e.g., by emphasizing how such opposition goes against the material interests of the working class).
All in all, our findings show that business legitimacy depends on both structural (class-based) factors and political (power resource) factors that operate at the micro and macro levels. In this article we did not analyze which of these factors is more prevalent in our explanation, nor did we test how these factors interact. Further studies should address these issues. We believe that it is worthwhile to explore two-level interactions between some of these factors. For example, if government partisanship is measured by a variable that captures the degree to which left governments rely on popular mobilization, there might be significant interaction effects between government partisanship and individual class location. We speculate that when governments mobilize the working classes instead of emphasizing technocratic decision-making, the political polarization between classes might increase.
Although largely consistent with our hypotheses, this study has potential limitations that should not be ignored. As stated above, our data covered a period when business power was challenged by social movements and progressive governments aiming to implement redistributive reforms. The implication is that our findings might not be generalizable to other sociopolitical contexts (e.g., where business-government relations are less strained). We also focused on a highly inequal region, Latin America, where institutional trust is not always correlated with institutional strength. This might also affect the generalization of our results. Finally, we believe that data and measurement limitations can explain the lack of statistical significance of theoretically relevant variables, such as trade union power. In this article, we analyzed a sample of 45 country-year observations taken from a sample of only 15 countries. Additionally, we used a rough measurement of union power based on the trade union density rate. Our small sample of countries may have affected the statistical power of some estimates, and our measurement of union power might not capture other aspects involved in the concept of labor power (e.g., the degree of centralization of industrial relations or the institutional strength derived from labor policies that protect workers’ bargaining power).Footnote 98 Further research should clarify whether statistical associations are found when more refined measurements of labor power are used.
Appendix
Linear probability models predicting trust in private companies in Latin America, 2005–2015 (standard errors in parentheses)
Note: The following variables were included in the models but not reported in this table: social class, gender, age, political position, perceived economic situation, individual-level trust in institutions, survey year, and GDP per capita (see main text for definitions).
***p < 0.001;
**p < 0.01;
*p < 0.05;
†p < 0.1.
