Introduction

Populism has long been viewed as a threat to liberal democracy and the recent rise of populist parties and executives has motivated new research seeking to understand the connection between populism and democracy. Populists view the procedures and institutions that characterize stable democracies as impediments to true popular sovereignty and often support their dilution or even abolition in favor of more “direct” links between the government and the citizenry (^{Rovira Kaltwasser et al., 2017}; ^{Hawkins et. al., 2017}; ^{Mudde and Rovira Kaltwasser, 2018}; ^{Urbinati, 2017}). A growing literature finds consistent evidence that populist leaders adversely impact the quality of democracy in countries that they govern (^{Houle and Kenny, 2018}; ^{Huber and Schimpf 2016}; ^{Kenny, 2020}; ^{Ruth-Lovell and Grahn, 2022}; ^{Vittori, 2022}).

It remains unclear, however, the extent to which individuals approve of this decline in democracy. Recent studies suggest that voters with populist attitudes decry existing democratic deficits in their countries, perhaps leading them to support populist politicians, but that these voters prefer democratic institutions, processes and values (^{Zaslove and Meijers, 2023}; ^{Wuttke et al., 2023}). It is thus possible that populist-led democratic backsliding is partly unwelcome, even by those who display more populist dispositions. Of course, such individuals might also welcome specific kinds of backsliding even while they claim to value democratic principles.

We seek to advance the understanding of attitudes about populism and democracy in two ways. First, we explore underlying dimensions of both populism and democracy rather than treating them as holistic concepts. Scholars define both in complex, multi-dimensional ways, but empirical examinations have often treated them as unitary concepts because measurement has lagged conceptual understandings (^{König et al., 2022}). Scholars typically point to three dimensions of populism: pro-people orientation, anti-elitism and Manicheanism. We utilize these three dimensions to produce new hypotheses about how they might differentially impact attitudes about democracy. At the same time, we also break down democracy into a multi-dimensional concept. Unlike populism, there is no established scholarly agreement about the nature of the dimensions, though other recent scholarship has focused on examining different facets of democracy (^{Claassen et al., 2023}). We identify four dimensions of support for democracy and theorize that different populist dimensions might have different impacts on them. We also identify two different ways to measure satisfaction with democracy.

Second, we apply the study of populism and democracy to an important new population: bureaucratic officials. Despite the attention that has been paid to citizens’ populist leanings, support for democracy or the intersection of the two, it is difficult for citizens to meaningfully influence the degree of democratic backsliding in the short term. Bureaucrats, on the other hand, play a central role in translating populist positions into policy, crafting and executing the details required for actual government action. A staunchly anti-populist or pro-democracy bureaucracy could delay or even block executive action deemed detrimental to democratic institutions and functioning. Scholars examining the interplay between populist governments and bureaucrats in different country case studies have found a variety of outcomes, from strong bureaucratic resistance to near-complete populist remaking of the bureaucracy (^{Bauer et al., 2021}).

Despite the importance of bureaucrats as institutional actors that potentially influence populist politicians’ impact on the quality of democracy, we know almost nothing about bureaucrats’ attitudes. Scholars who study bureaucrats rarely ask about their political attitudes, focusing instead on patronage systems, bureaucratic oversight, and external influences on bureaucrats (^{Brierley et al., 2023}). We thus bring a new institutional actor to the populism-democracy research agenda. Bureaucrats can be considered political elites because their “institutional roles afford them higher levels of influence over public policy” (^{Kertzer and Renshon, 2022, p. 535}). Studies of political elites can be quite valuable when important questions arise about whether elites may share similar patterns with the general population (^{Kertzer and Renshon, 2022, p. 536}).

Given the dearth of prior research on bureaucratic attitudes on democracy, our hypotheses assume that bureaucrats think about populism and democracy in the same way as citizens. While bureaucrats may be considered elites in terms of their potential political influence, they are not socio-economic elites that have clearly identifiable interests distinct from “ordinary” citizens on most issues. Bureaucrats are likely to have more education than typical citizens, but they still face similar vulnerabilities to economic problems or physical insecurity and are not members of an identifiable social or cultural group that would shape their views in fundamentally different ways. Of course, bureaucrats might also differ from typical citizens by having more education and being more politically aware. They also have a professional commitment to the existing, mostly democratic, political institutions in their country. Hence, our hypotheses are preliminary and merit further investigation.

Our contribution focuses on breaking populism and support for democracy into constituent parts and investigating relationships between them. We utilize existing studies of citizens to help formulate baseline expectations about the sub-population of bureaucrats but do not seek to explicitly compare bureaucrats and citizens. Because we offer innovations in both disaggregating measures of populism and support for democracy, explicit comparisons are impossible in any case. We discuss the implications of our findings about bureaucrats in the results section by relating them to existing findings about citizens. Differences are suggestive of the need for further research into the political attitudes of this understudied population.

In 2021-22, we surveyed bureaucrats in eight Latin American countries asking about their attitudes on populism and democracy, among other topics. Recognizing that both these phenomena are complex and multi-dimensional, we explored attitudes in some detail. We broke down populism into three well-known underlying dimensions: anti-elitism, pro-people sentiments, and a Manichean (good vs. evil) worldview. Working with measures that have been validated in cross-national surveys of citizens, we asked two questions on each dimension (^{Castanho Silva et al., 2020}).

With respect to democracy, we are interested in the two different (yet related) issues of satisfaction with democracy and support for democracy, which ^{Norris (2011)} describes as beliefs about the supply of and demand for democracy. Scholars often treat these issues separately and considerable scholarship looks at only one of these two possible outcomes (^{Curini et al., 2011}; ^{Ezrow and Xezonakis, 2011}; ^{Claassen, 2020}; ^{Gidengil et al., 2022}).² The essential difference between satisfaction and support can be illustrated by conceptualizing variation on the two dimensions. Individuals might think that their system fails to correspond well with democratic norms (low satisfaction, or supply) and favor more democratic practices (strong support for democracy, or demand). People in this category are “dissatisfied democrats.” Other combinations are of course possible; for example, individuals might be dissatisfied with the current democratic system and favorable to undemocratic values (dissatisfied non-democrats).

We find that the populism-democracy nexus is more nuanced than the received wisdom, or our own expectations, would suggest. First, we find that the three dimensions of populism commonly utilized in studies of voters do not yield a consistent underlying index in our sample of bureaucrats and this is reflected in dramatically different distributions of respondents for each dimension. We, thus, analyze each of the three dimensions separately.

We find that bureaucrats are relatively satisfied with their political systems, especially compared to voters, but that they have widely varying opinions on whether their country is a full democracy. We find no association between underlying dimensions of populism and satisfaction with democracy, but we do find that bureaucrats with higher scores on each dimension of populism are less likely to think their country is a full democracy. Turning to support for democracy, respondents who express more pro-people sentiments more strongly support the dimensions of democracy that we test. In contrast, we find that those with a Manichean outlook are less supportive of all tested dimensions of democracy. Associations between those with anti-elite attitudes and support for democracy are mostly non-significant.

Combining these results, we find little evidence that bureaucrats are dissatisfied democrats, in contrast to some studies of populist voters that find that voters fall into that category (^{Zaslove and Meijers, 2023}; ^{Rovira Kaltwasser and Van Hauwaert, 2020}). Pro-people bureaucrats are largely satisfied democrats and Manichean bureaucrats are non-democrats while anti-elite bureaucrats are no more or less favorable toward democracy than other bureaucrats. These complexities suggest the need for more research on disaggregated populist attitudes and fine-grained measures of support for democracy. Our results suggest that populist leaders’ threats to democracy may be moderated by the pro-democracy orientations of some bureaucrats. At least among Latin American bureaucrats, pro-people and anti-elite bureaucrats are both less antagonistic to democracy than conventional wisdom about populism suggests and less antagonistic than those who hold Manichean views.

Theoretical Context

Latin America has a long history of well-known populist leaders, stretching from Juan and Eva Perón in Argentina to Andres Manuel Lopez Obrador in Mexico, leading one scholar to declare, “Latin America is the land of populism” (^{De la Torre 2017, p. 195}). Populism’s rise in various countries has been attributed to sources as diverse as party fragmentation and unstable interparty alliances (^{Borges, 2021}) and the commodity boom (^{Grigera, 2017}). Scholars have found populism can lead to democratic backsliding in a variety of ways, including “undermining the autonomy” of key political and judicial institutions (^{Arteaga Botello, 2021, p. 97}) and the centralization of power, suppression of dissent and creation of affective polarization (^{McKay and Colque, 2021}). The negative effects of populist leaders on democratic quality that are demonstrated by these case studies are bolstered by cross-national quantitative analyses showing strong correlations between populist leaders and the decline of democracy at the country level (^{Hawkins et al., 2023}).

Populist executives, however, do not necessarily have a completely free hand to simply impose their will in dismantling democratic institutions. Indeed, while the level of democracy has dropped in Latin America over the past decade, the political systems have not declined into unfettered authoritarianism in most countries. Perhaps populist leaders prefer a middling outcome and do not wish to push further down the path away from democracy. An alternative possibility is that they have met resistance in their attempts. One plausible source of resistance is public servants, a large enough obstacle for many political leaders in Latin America and elsewhere to spend significant time and energy trying to bend them to their will, sometimes successfully (^{Muno and Briceño, 2021}). Scholars examining bureaucrats often perceive “interactions between politicians and the bureaucracy as multifaceted and complex; regularly, they are more a matter of negotiation or collaboration than of top-down command and control” (^{Bauer et al., 2021, p. 5}). Relatively few attempts have been made, however, to examine these dynamic government-bureaucrat relations in the context of democratic quality. One experimental approach in Brazil found that bureaucrats are quite willing to “take democratic norms into account when making decisions” through some form of non-compliance with high-level anti-democratic directives (^{Guedes-Neto and Peters 2021, 242}).

The study of populism at a sub-national level has evolved from an early focus on the emergence of populist parties, particularly in Europe, to an emphasis on voters. Considerable scholarship has focused on how to measure populist attitudes and on both the causes and consequences of populism in the electorate (^{Castanho Silva, 2020}). Although three attitudinal tendencies have come to be considered indicative of populism-anti-elitism, pro-people orientation and Manicheism (^{Hawkins and Rovira Kaltwasser, 2019})-there is no consensus on how to measure them or even whether the three dimensions cohere well if measured separately. One systematic study comparing seven populist attitude scales found only three of those scales cohere well but suggested that the lack of coherence derived mostly from poor questions rather than problematic conceptualization (^{Castanho Silva et al., 2020, p. 420}). At the same time, some find evidence that the Manichean dimension operates differently from the other two (^{Akkerman et al., 2014}), and others suggest modifications of a good/evil frame to tap the moral dimension of populism (Elchardus and Spruyt, 2014; ^{Spruyt et al., 2016}). A group of authors (^{Schulz et al., 2018, p. 322}) reviewing different data sets recently concluded that a model with “three dimensions of populism is superior to a one-dimensional conceptualization in a number of ways.”

Recent scholarship examines the relationship between citizens’ populist attitudes and views about democracy. Initially, populist attitudes were theorized to threaten democracy due to the preference for direct unmediated ties to an unconstrained leader. Indeed, a robust literature finds a strong correlation at the national level between populist political leaders and a decline in democracy (^{Hawkins et al., 2023}). Surprisingly, however, populist citizens have been found to favor democracy over other political systems and have also been found to be favorable toward democracy as a system (^{Zaslove and Meijers, 2023}; ^{Rovira Kaltwasser and Van Hauwaert, 2020}; Hawkins et al., 2018), suggesting a potential disconnect between populist leaders’ aims and their followers’ preferences.

It seems plausible that populist citizens do not oppose democracy as a regime, but believe it has been corrupted, and are therefore dissatisfied with its functioning and favor remedies believed to better empower the voice of “the people.” Satisfaction with democracy is typically understood as “an expression of short-term evaluation of system outputs, and as a useful measure of the discrepancy between democratic norms and actual processes” (^{Curini et al., 2011}). Support for democracy, on the other hand, is conceptualized not as evaluations of the current political system but rather as aspirations, values and goals with respect to the organization of political life (^{Norris, 2011, pp. 31-33}). New research has explicitly tested the relationship between populist attitudes, satisfaction with democracy, and support for key institutions of executive oversight as well as mediating institutions such as political parties. A few studies have found that individuals with stronger populist attitudes are in fact dissatisfied democrats: more supportive of democracy but more dissatisfied with its functioning (^{Zaslove et al., 2020}; ^{Rovira Kaltwasser and Van Hauwaert, 2020}). Such populist voters are less likely to protest, are more supportive of referendums, and are more supportive of deliberative forms of political participation compared to individuals with weaker populist attitudes (^{Zaslove et. Al. 2020}). Moreover, populist voters do not necessarily reject liberal institutions in general but do reject mediated representation through political parties and judicial oversight (^{Zaslove and Meijers, 2023}).

Of course, these findings represent the relationship between populism and satisfaction and support for democracy among citizens. As discussed above, bureaucrats may be similar to ordinary citizens in their exposure to uncertainty and economic and social pressures. However, we also recognize that they differ from an average citizen in terms of being more highly educated, highly politically aware, and involved in policy making. We do not anticipate strong differences between citizens and bureaucrats but do posit that due to their personal involvement in policymaking they may be less dissatisfied with the way democracy works than average citizens. Even pro-people and anti-elite bureaucrats are likely to better understand and agree with the nuanced reasons why policy deviates from the will of the people.

Based on the lack of consensus in the literature on what dimensions are key to the concept of populism and on some previous studies where the dimensions do not cohere well, we proceed to develop our argument by analyzing each of the dimensions separately. While we expect the three dimensions to sometimes affect democratic attitudes in similar ways, we also develop some arguments about why the Manichean dimension may differ from the pro-people and anti-elite dimensions, depending on the outcome in question.

Research Design

Given the plausible importance of bureaucratic actors in influencing how democracies function in the face of backsliding executives and the paucity of knowledge about public servants, we implemented a survey of them from December 2021 to March 2022 in eight important countries: Argentina, Brazil, Chile, Colombia, Mexico, Panama, Paraguay, and Peru. We assembled the subject pool by using Python scripts and human labor to scrape official government websites in each country and identify possible respondents. We included only national-level government employee names and email addresses for all available government agencies, excluding only security services. Where emails were unavailable, we created email addresses based on the email conventions we observed from other addresses on the website. Emails were then double-checked for duplicates and cycled through a program to check for the likelihood of success. We excluded those found to have a nonexistent domain or syntax errors, those that were disabled or those that failed to be delivered due to a dead server, a rejected attempt, a relay error or some unknown reason. We then randomly selected a sub-set of addresses and utilized the emails to send a link to the survey directly to bureaucrats. There was no incentive or reward for respondents to take the survey.

We chose the countries based on availability of names and email addresses from government websites. Many countries did not have easily accessible public-servant contact information. Our goal was to gather at least 500 responses from each of these countries; we succeeded (or nearly so) in Brazil, Chile, Colombia, Mexico and Panama but received fewer responses in the other three countries (Table A1). Expecting a number of our emails to bounce and many others to find their way to spam filters, we sent out email requests to 50,000 randomly selected bureaucrats in each country, followed by a reminder a few days later. For the three countries with lower response rates, we judged it better to accept the lower response rate rather than sending out repeated reminders or expanding our subject pool. We worried that the passage of time could affect answers, making it undesirable to put all the data into the same analysis. In the end, 3,704 bureaucrats answered at least one question in our survey. About 31 percent of our emails bounced, giving an average response rate of 1 percent among non-bounced emails, though this varied by country. We cannot of course claim that our results are generalizable to all bureaucrats in these countries, but a recent meta-analysis of elite sampling found that random samples of elites typically generate “consistent and unbiased estimators of population parameters” (^{López 2023, p. 1303}). Our approach is consistent with methodological choices recommended by ^{Kertzer and Renshon (2022)}, such as reasonable sample sizes, the importance of anonymity and a focus on first contact.

Unfortunately, there is no systematic information on the population of Latin American bureaucrats, so we cannot precisely measure the ways in which our sample differs from the population. However, our sample seems consistent with common-sense expectations about demographic features of bureaucrats, as detailed in Table A1. About 55 percent of respondents report having a post-graduate degree, and less than 1 percent of respondents report having less than a secondary education degree. We observe a relatively normal distribution of ideology with a slight skew to the left. Respondents in our survey report working in a wide range of agencies, from the environment to public finance, with Public Management being the largest group, at about 12 percent of the sample, followed by Justice at nearly 10 percent. The mean amount of time respondents report working for the government is approximately 14 years, with a standard deviation of 10 years. A little more than 50 percent of respondents identify as male. Respondents were overwhelmingly hired (95 percent) rather than appointed by elected officials. Nearly two-thirds of them spend at least a little of their time supervising others, suggesting that we are not simply sampling teachers or custodians or other low-level employees. About 45 percent report taking a civil service exam. Overall, we feel confident that our sample aligns with reasonable expectations for this population.

We completed a systematic review of the recent survey work done on populist attitudes and selected a set of six questions proposed by ^{Castanho Silva et al. (2018)}. We came to this conclusion in large part because of the work of ^{Castanho Silva and others (2020)} in comparing seven frequently used populism question batteries. Due to the multinational nature of our project, we prioritized the scale that tested as having the highest level of cross-national validity. In previous tests, the scale also performed well on coherence, conceptual breadth and external validity. The populism questions follow, with respondents asked to rank their views on a five-point scale from strongly disagree to strongly agree. These have been re-coded as necessary so that higher values indicate more populist attitudes. Our independent variable for each dimension of populism is the mean of the two questions listed.

Findings and Discussion

In contrast to prior results using measures of citizens’ populist attitudes, our own sample produced responses that indicate that, among bureaucrats, different attitudinal dimensions of populism do not reliably correlate with each other. Cronbach’s Alpha of the six questions is a mere 0.06 (far below the rule of thumb acceptable threshold of 0.60 for an index). One possibility for this poor result for a previously validated survey block is that the questions do not capture the way that bureaucrats think about the world. One of the anti-elitism questions, for example, asks whether government officials use their power to try to improve people’s lives. Bureaucrats may interpret that question as referring to themselves and answer it differently than citizens would. Likewise, another question asks whether politicians have to spend time among ordinary people to do a good job at helping those people. Again, it is possible that bureaucrats would think of themselves and their role in helping people when answering this question and that most citizens would not understand the question the same way. Bureaucrats would not likely consider themselves politicians, but they may see themselves as making it possible for politicians to not spend time among the people in order to do a good job. The results suggest that scholars exercise much caution when attempting to measure populist attitudes in different populations and particularly among public officials.

A second possibility, which is not mutually exclusive with the first problem, is that bureaucrats do not think about populism in the cohesive way that scholars expect. In fact, reasonable evidence exists to suggest that the underlying dimensions of populism also often do not cohere in surveys of many citizen populations (^{Castanho Silva et al., 2020}). One reasonable response to the coherence problem is to break the concept of populism into its underlying dimensions: pro-people, anti-elitism and Manicheanism. In our survey, we had two questions to measure each of these dimensions. Within each dimension, however, responses to those paired questions still do not correlate highly with each other. At the same time, the conceptual argument is reasonably strong that each dimension cannot be further sub-divided into more dimensions. Hence, we adopt the approach of creating three separate indices and run our models with all three indices. As a robustness check, we also run our base models with each populism-oriented attitude measure individually. The results, discussed below, show that models using two separate measures for each dimension yield qualitatively similar results as putting questions together in an index for each dimension.

Figure 1 is a violin plot mapping the distribution of opinion for each of the three different dimensions of populism. As discussed above, each dimension of populism is measured as the average index of two questions on a scale from one to five. The anti-elite measure most resembles a normal distribution, though it is somewhat skewed toward higher values. The distributions of the pro-people and Manichean dimensions are nearly mirror images of each other, with the bulk of respondents scoring high on the pro-people dimension, but low on the Manichean dimension. The bureaucrats in our sample do not demonstrate a combination of attitudes that reflect populism as a unified concept in the way that many scholars have discussed. Rather, they adopt very different views about the different dimensions. Many tend to be quite pro-people and anti-Manichean while there is a range of opinion about anti-elitism.

Figure 1 Distribution of Populism Index containing Three Dimensions 

Turning to our outcomes of interest, Latin American bureaucrats are, on balance, relatively satisfied with their country’s democracy even though many are reluctant to say their country is a full democracy as illustrated by Figure 2. Overall, about 62% of bureaucrats are satisfied or very satisfied with their country’s democracy. This stands in contrast to the views of Latin American citizens reported in other studies. Only 43 percent of citizens said they were either satisfied or very satisfied with democracy in their country (^{Lupu at al., 2021}). For the question of whether their country is a full democracy, the distribution is quite flat with all of the possible ten scores receiving some responses (as illustrated in Figure 2). In short, bureaucrats tend to be skeptical their country is a full democracy but, perhaps realistically, also incline toward satisfaction with what they have.

Figure 2 Assessment of Country’s Democracy and Satisfaction with Democracy 

Latin American bureaucrats are also rather supportive of democracy, as seen in Figure 3. Large majorities agree or strongly agree that democracy is the best form of government and that particular features of liberal democracy are desirable: political party autonomy, legislative autonomy and the ability of anti-system demonstrations to take place. Support for political party autonomy is especially strong.

Figure 3 Distribution of Democratic Support Indices 

Turning to our analysis, we find mixed evidence for the first hypothesis (Figure 4). Bureaucrats who identify with any of the three dimensions of populism are less likely to agree that their country is a full democracy. These results are statistically significant for Anti-Elite and Manichean dimensions of populism. At the same time, they are not likely to be less satisfied with democracy, a finding that differs from studies of citizens. In a sample of 11 countries in Europe and Latin America, ^{Rovira Kaltwasser and Van Hauwaert (2020)} found in every one of them that populist attitudes are associated with lower satisfaction with democracy. Others have found similar results (^{Zaslove and Meijers, 2023}). While we cannot say for certain what accounts for the absence of the widely observed relationship between populism and democratic satisfaction, bureaucrats are more intimately acquainted with the process of governance than citizens and have also attached their personal livelihoods to those processes. Bureaucrats may thus be more willing than citizens to overlook or tolerate problems with the political system, or to have more realistic expectations of what the system can deliver and how it operates. These lower expectations can result in more reluctance to express dissatisfaction. The complete results for Figure 4 can be found in Appendix Table A3, with information on the control variables in Table A2.

Figure 4 Populism Measures and Satisfaction with Democracy 

These results are robust to models with only one populist dimension in each model (Table A5) and to models with more extensive controls regarding workplace attributes, including ministry fixed effects, years in government employment, supervisory level, and dummy variables for political appointment and whether they took a civil service exam (A6). If we use our six individual measures of the populism dimensions in the same model, results are similar to the three-dimension models, suggesting that aggregation is not altering or obscuring individual results (Table A9). Among control variables, Support for President is worth a brief discussion, as partisanship is an important question in many studies of attitudes toward democracy. The question asks how respondents rate the job performance of the sitting president in their country, from very bad to very good. Those who support the president are more likely to think their country is a full democracy, perhaps because democracy produced an outcome they welcome. At the same time, they are less likely to be satisfied, a puzzling outcome that deserves further research because presidential supporters are often more satisfied with democracy (^{Anderson & Guillory, 1997}).³

With respect to support for democracy, we find that the relationship between populist-oriented attitudes and support for democracy depends on the dimension of populism being discussed and on the dimension of democracy in question (Figure 5; Table A4). Our hypotheses as a whole favor breaking populism and democracy into different dimensions and we find evidence that outcomes do indeed vary by dimension. At the same time, the particular relationships we found were often but not always consistent with our hypotheses. We measure support for democracy on four dimensions, as discussed above: conceptual, legislative autonomy, political party autonomy, and allowing anti-system demonstrations.

Figure 5 Populism Dimensions and Support for Democracy 

Those with pro-people attitudes are associated with more support for democracy across all four outcomes and three of those results are strongly significant. In sharp contrast, those with Manichean attitudes are associated with lower levels of support for democracy across all four outcomes. Anti-elitist attitudes are generally insignificant, but interestingly, they favor popular demonstrations, which is consistent with a view that democracy has been captured by elites. Tables A7 and A8 show that the results are robust to models that break out the measures of populism into different equations and to models that use more expansive control variables. Using the individual measures for the populism dimensions shows similar patterns to the three-dimensions model (Table A10). Support for the President is consistently negative and significant, which is consistent with existing literature showing that those who support the president are less likely to support democratic principles (^{Singer, 2018}), probably because democratic institutions like legislatures and parties restrict the ability of their preferred leader to enact policy.

Overall, these findings suggest that the relationships between attitudes about different dimensions of populism and democracy are complex. Respondents with high scores on any of the three populism dimensions are more likely to be skeptical of how well their country fits a democratic ideal, which certainly fits the conventional narrative about populism. At the same time, they are not likely to be more dissatisfied with that state of affairs, which does not fit the narrative. As discussed, it is possible that these results are different for bureaucrats than they would be for ordinary citizens.

When considering support for democracy, only those with Manichean beliefs express less support for liberal democracy across a variety of dimensions. That particular outcome is consistent with a narrative about a causal relationship between the rise of populism and the decline of democracy: populist leaders play on popular fears about the indifference and corruption of evil elites to gain support for their anti-democratic efforts. Those who see the world divided into good and evil are likely particularly susceptible to these allegations.

However, the story about populist attitudes among bureaucrats turns out once again to be more complex than that. Those with pro-people attitudes, which arguably lie at the heart of populist preferences, are more supportive of democracy across a variety of democratic dimensions. Abstracting away from everyday populist rhetoric, such an observation is logically consistent in itself: for those who value people, they prefer to make the people’s voices heard through legislatures, political parties, anti-system demonstrations and, in general, democracy. Still, these views are somewhat surprising given that populist political leaders tend to reject intermediary institutions like parties and legislatures because they believe they are susceptible to elite capture. Bureaucrats most attracted to anti-elitist views are prone to neither accept nor reject the concept of democracy or mediating institutions like legislatures and parties. Perhaps they see some hope for those institutions to serve the needs of the people, a hope consistent with their choice to become a public servant. Unsurprisingly, those with anti-elite orientations are more likely to favor anti-system demonstrations.

Conclusions

This initial investigation of the relationship between bureaucrats’ populist attitudes and democracy, as with much of the literature on populism, yielded mixed findings. While those with Manichean views fit the well-known populist narrative-distrust of liberal democracy-quite well, those with pro-people and anti-elite attitudes do not fit conventional wisdom at all. No underlying attitude is associated with less satisfaction with democracy although all three are associated with viewing their country as less than a full democracy. Populist bureaucrats may understand that building a full democracy is difficult and may view their democracy as a work in progress and thus are less likely to fault their existing system.

Manichean bureaucrats are less accepting of democracy as a concept and of particular checks on executive power provided by legislatures and political parties, thus favoring quintessential populist remedies that empower executives. Their association with support for anti-system demonstrators is neither positive nor negative, perhaps because they might reserve the right to engage in such demonstrations even if they do not fully endorse them in all cases.

Pro-people bureaucrats, in contrast, are more likely to endorse specific democratic institutions that check executive power. Of the three dimensions of populism, those holding pro-people views have underlying beliefs most consistent with democratic rule. While Manicheans are skeptical of anyone in power save their leaders, and anti-elite populists may view democracy as inevitably prone to elite capture, those with pro-people views are perhaps more likely to see democracy as amenable to improvement, and thus see value in institutions that have the potential to reflect the will of the people. Anti-elite bureaucrats have no particular association with most dimensions of democracy, but they do more strongly support anti-system demonstrators, a finding consistent with their views.

These results suggest two future avenues of fruitful research. First, investigation of populist attitudes and not just support for democracy, but also support for liberal institutions of democracy among voters and elites, is key to refining our understanding of where democracy’s potential vulnerabilities lie. Second, additional study of elite attitudes and how populism shapes their views of different forms of backsliding could shed new light on possible ballasts for democracy. Indeed, we found that bureaucrats in general are quite supportive of democracy, and pro-people bureaucrats exhibit attitudes that suggest they may be pro-active allies. Bureaucrats are elites with influence over implementation of executive directives, and the fact that some with pro-people attitudes are more supportive of liberal democracy suggests that at least some elites may seek to check executive overreach.