I Don’t Like It Unless It’s for Me: Voters’ Perceptions of Pork-Barrel Politics in Central and Eastern Europe

ABSTRACT

Public resources are often allocated neither effectively nor equitably, and they serve as a tool for securing re-election. Despite compelling evidence for pork-barrel politics, little is known about voters’ perceptions of this practice. Moreover, limited findings from Western countries are also contradictory – voters perceive such a strategy both positively and negatively. To contribute with findings from Central and Eastern Europe (CEE), we conducted two survey experiments – in Czechia and Slovakia – which randomized people’s exposure to fairness of the distribution and profit for the respondent’s country. Our results suggest that once voters realize their profit from pork-barrel politics, they are less critical of unfair distributions of resources and the responsible decision-maker.

Introduction

Effective and fair distribution of public resources is seen as one of the key elements of a well-functioning democracy. In the real world, however, it turns out that resources are often allocated neither effectively nor equitably, and that the allocation of public finances is used by incumbents as a tool for securing votes and reelection (Denemark 2000). The existing scholarly literature shows that politicians strive to achieve favorable political outcomes, and that they intentionally bias money distribution in accordance with their political aims in order to affect local electoral behavior (Costa-i-Font, Rodriguez-Oreggia, and Lunapla 2003; Denemark 2000). To this end, political parties typically adopt one of two strategies: they either target safe seats where a party member can win with a clear margin or, on the contrary, they focus on channeling resources to swing constituencies to avert the risk of losing. Existing research finds evidence for both core and swing strategies without giving either of them the upper hand (Dahlberg and Johansson 2002; Denemark 2000; Dixit and Londregan 1996).

The literature provides compelling evidence for pork-barrel patterns being present in distributive politics (Milligan and Smart 2005). Slovakia and the Czech Republic, the focus of this study, are not exceptional cases breaking this pattern (Spáč 2016a, 2016b). However, there is only limited empirical evidence to reveal voters’ perceptions of this inherently unfair attempt to secure their votes. Moreover, existing findings are contradictory: it appears that voters may perceive a pork-barrel strategy both positively (Braidwood 2015) and negatively (Bøggild 2016). Nevertheless, little effort has been dedicated to explaining this ambiguity.

Our research fills in the gap among studies exploring the patterns in voters’ perceptions of pork-barrel politics. Findings have so far been made solely in regions with long democratic traditions, such as Western Europe, the Nordic countries, and Australia (e.g., Bøggild 2016; Braidwood 2015). However, the perceptions of voters in countries where a higher degree of clientelism and corruption of political elites are characteristic have not been properly studied by researchers. Therefore, we focus on two Central and Eastern European (CEE) countries that meet these criteria; namely, Slovakia and the Czech Republic (“Special Eurobarometer 470: Corruption” 2017; Transparency International 2019). Our survey-embedded experiment was designed to test the effect of manipulating the two main attributes of distribution politics: (1) the fairness of such a distribution and (2) the benefits for the respondents. The responses were collected for a representative sample of the populations of the two countries using an Internet panel facilitated at our request by the respected and well-known FOCUS survey research agency.

We addressed the question: How do fairness and possible benefits influence voters’ perceptions of responsible decision-makers? Our research provides empirical evidence for a trend among voters such that, once they realize that they are profiting from pork-barrel politics, they are less critical of an unfair distribution of resources.

The paper continues as follows. First, we review the literature dealing with pork-barrel politics, with special attention given to people’s perceptions of these practices. Then we describe our data and applied methods. Lastly, we present our results, followed by a concluding discussion of the findings.

The Emergence of Clientelism

If there is any widespread agreement in the field of party politics and clientelism studies, it is that political parties and politicians in office are increasingly taking advantage of public resources. Political actors intentionally direct money from state budgets to recipients in a way that is intended to improve their electoral position. This strategy has emerged gradually as a response to the transformation of party organizations. Partisan de-alignment radically decreased the traditional type of available resources (Bartolini and Mair 1990; Dalton and Wattenberg 2000). As political parties became less and less dependent on their membership, they grew closer to the state and started to use its resources to sustain themselves (Katz and Mair 1995, 2009). The developing relationships between parties and the state gained complexity after the introduction of direct public funding of political parties (Van Biezen and Kopecky 2007, 2014). However, the general intention of parties to use public resources to support their own survival and improve their electoral performance remained the same.

Research has convincingly proven that the intended strategy of parties and politicians when allocating public resources is to influence local electoral behavior in their favor. This often happens through investments made at the lower levels of state administration (Costa-i-Font, Rodriguez-Oreggia, and Lunapla 2003; Denemark 2000, 2014; Milligan and Smart 2005). The aim is to make improvements to infrastructure and services so that citizens can immediately feel the positive change. In a specific context, the money might flow in a more direct manner to voters who are willing to have their votes bought by political machines without paying attention to the content of the political competition (Birch 1997; Hasen 2000; Kitschelt 2000; Stokes 2005; Stokes et al. 2013).

Another branch of research into clientelism identified a link between parties and private donors who were willing to make campaign contributions in exchange for public procurement contracts and influence over policy development (Gherghina and Volintiru 2017). Various research papers found institutional causes for this relationship (Hoare 1992; Khemani 2003) as well as additional background factors that influenced the emergence of clientelist practices conducted by parties (Ansolabehere, Snyder, Jr., and Ueda 2004; Stratmann 1995).

Overall, the literature captures a variety of schemes employed by political actors to exploit clientelism. It deals with straightforward vote buying, which has evolved into more complex forms in which the interaction between parties (patrons) and voters (clients) is surrounded by additional actors. Such frameworks are characterized by their application to different time intervals, since parties are not always oriented toward short-term goals (such as increasing their gains in the next elections), but may be seeking consolidation of a long-lasting relationship with the electorate (Gherghina 2013). However, even though political parties and politicians are together a central driver of clientelism, the success of their efforts to thereby support their own sustainability is in the hands of voters. Therefore, it is equally relevant to study the voters’ perceptions of clientelism in order to understand the wider societal consequences of parties’ and politicians’ engagement in clientelism.

Voters’ Perceptions of Pork-Barrel Politics

The use of fair procedures should be a cornerstone of the legitimacy of any democratic political system. As Esaiasson and Öhberg (2020) note, fair procedural arrangements evoke the “good loser” norm. This standard assumes that people living in a democracy must be able to accept unfavorable political decisions, if these decisions result from a fair process. However, this somewhat idealized view of the functioning of democracy clashes with the incentives of politicians who try to influence the rules of the game in their favor (Bowler, Donovan, and Karp 2006). In reality, political support is largely inspired by the ability of politicians to deliver favorable outcomes (Magalhães 2016).

The notion that politicians need to make favorable decisions to please their constituents and secure reelection is recognized in classic works of political science (Downs 1957; Schumpeter 1943). However, this theoretical position stands in stark contrast to the field of social psychology and its “procedural fairness theory.” The theory formulates a set of rules characteristic of a “fair process” that need to be followed by decision-makers in order to maintain the trust and support of constituents as a prerequisite for being reelected. The main principle of this theory lies in the need for the impartiality of the decision-maker and the absence of any personal motives influencing the outcomes of his decision-making (Leventhal 1980; Thibaut and Walker 1975; Tyler 1990). As Robert J. MacCoun (2005, 171–172) put it, citizens “care deeply about the process by which conflicts are resolved and decisions are made, even when outcomes are unfavorable or the process they desire is slow and costly.”

These theoretical assumptions are confirmed, for example, by studies in the field of behavioral economics that suggest individuals are very strict in evaluating responsible actors who base their decisions on their own personal interests (e.g., Falk, Fehr, and Fischbacher 2008). In addition, Hibbing and Alford (2004) found in their series of economic experiments that people’s responses to a particular output are strongly influenced by the procedure behind that output. Respondents in their study evaluated actors responsible for the redistribution of resources less positively when the allocators kept more resources for themselves.

The theory of procedural fairness is not new in the research on political support. Classic scholarly works discovered that decision-makers are evaluated based on their ability to deliver favorable outcomes, but also on the basis of perceived fairness of the process—including the sources, decisions, and distribution of these outcomes to selected societal groups (Lind and Tyler 1988; Thibaut and Walker 1975). Research has also confirmed that political representatives who allow citizens to speak during decision-making processes are perceived in a more positive light (Tyler 1994; Ulbig 2008). If we focus on politicians, taking part in illegal activities or making various personal mistakes can have serious negative consequences for their integrity and the public’s trust (Ulbig and Miller 2012). In addition, there is quite a lot of convincing evidence that political scandals of various kinds can have a significantly negative impact on voter decision-making (e.g., Bhatti, Hansen, and Olsen 2013; Bowler and Karp 2004).

Some authors have tried to focus on the effects of the impartiality of decision-makers on people’s attitudes. Hibbing and Theiss-Morse (2002) dealt with the question of what American citizens actually expect from their democratically elected government. Their findings suggest that people prefer impartial, empathetic decisions made by non-self-interested decision-makers. Stacey G. Ulbig (2002) came to a similar conclusion. Based on the procedural fairness theory, he demonstrated that judgments about policy output and the processes by which policies are developed are key sources of trust in government. Using an experimental design, Mark D. Ramirez (2008) examined the impact of process fairness on support for public institutions. As expected, the subjects in the experiment showed a higher level of support for the Court in those cases where the rules of impartiality and fairness were met. On the other hand, Anderson and Tverdova (2003) showed that for those who voted for the winning side in previous elections, the win itself was more important than fair and impartial procedures. Hence, it turns out that the extent to which the decision-making process is considered fair is also conditioned by the favorability of the outcome. In their experimental research, Esaiasson and Öhberg (2020) discovered that if persons benefit from the output of a political process, they increasingly tend to consider the process to be more fair. Hence, favorable outcome seems to increase the perceived procedural fairness.

All these findings suggest that the fairness of these processes has a major impact on the general public’s evaluation of decision-makers. But political struggle is inevitably linked to the subjective ambitions of politicians, which are fully reflected in their efforts to succeed in elections. As a result, unfair processes not only affect voter decision-making in elections, but also affect society’s overall satisfaction with democracy and political legitimacy (Gilley 2006). Dahlberg and Holmberg (2014) even argue that impartial bureaucratic procedures matter more for satisfaction with democracy than electoral outcomes.

The contrast between political science and social psychology in their approaches to politicians seeking reelection led Troels Bøggild (2016, 901) to a complex theoretical expectation that “politicians must present voters with favorable policy outcomes but without appearing as if they pursue these policies based on a personal, vote-maximizing motive for reelection.” In his experiments conducted in Denmark, Bøggild (2016) observed that subjects respond to the reelection efforts of politicians who use the impartial distribution of public finances with a lower level of trust and reduced willingness to vote for these politicians even when they can benefit from distribution. However, this conclusion is not fully in line with the outcomes of experiments organized in the United States by Travis Braidwood (2015). In these experiments, “[r]ecipients are only moved when they are made aware of projects in policy arenas of individual importance. Thus, pork is a nuanced policy tool that must be wielded strategically to gain electoral reward from specific subsets of a constituency” (Braidwood 2015, 1). Therefore, it appears that voters can perceive pork-barrel strategies both positively and negatively, depending on the context and the information available.

Hypotheses: Does Unfair Distribution Damage a Voter’s Perception of the Decision-Maker and the Policy?

The most reasonable explanation for these different findings is that the perception of unfair processes is a contextually conditioned effect. Since Bøggild’s study (2016) was conducted in Denmark, which is the least corrupt country according to the 2019 Corruption Perceptions Index (CPI) released by Transparency International, it may be that the Danish people are less inclined to tolerate corruption and clientelism because it rarely occurs in their daily life. A similar context applies to Braidwood’s United States, which ranks 23rd among the countries in the CPI. To test this hypothesized effect of the political context with respect to corruption, we decided to replicate Bøggild’s Study 1 in the CEE context. We selected Slovakia and the Czech Republic, where the partisan-motivated distribution of public resources has been convincingly proven in analyses of public budgetary data (Spáč 2016a, 2016b). Moreover, Slovakia occupies the 59th and the Czech Republic the 44th spots among the countries included in the CPI, significantly lower than Denmark and the United States.

There is another reason for the selection of Czech Republic and Slovakia for our experiment, which concerned a mock-up scenario of a grant scheme at the level of the European Union (EU). The EU operates as a political space in which member states regularly compete in various decision-making processes, be they negotiations on institutional matters, EU budget decisions, or foreign policy arrangements (Bayer 2020; De Wilde 2012; Varju 2019). For political representatives and inhabitants of the Czech Republic and Slovakia, given their status as net receivers, the EU might be primarily perceived as an arena in which their national and financial interests clash. The analysis of party manifestos in Slovakia confirms such a claim (Haughton and Rybář 2009), and other literature shows that CEE countries intentionally defend their regional interests in EU financial matters (Walsch 2018). Such a viewpoint might lead CEE countries to show a higher acceptance of materially beneficial policies, even in cases of low fairness of allocation procedures.

We formulate a set of hypotheses to be experimentally tested in Slovakia and the Czech Republic based on the theoretical expectations of the procedural fairness theory and contributions from the empirical literature. Our main goal is to test the hypothesis about the impact of the fairness of decision-making and self-benefit on voters’ evaluation of politicians. While our hypotheses reflect the same theoretical basis as Bøggild’s (2016), we expect that in the context of Central Europe the causal mechanism works slightly differently, and that voters will put stronger emphasis on self-benefit than on the fairness of the allocation of resources. Hence, we examine the effects of two relevant attributes related to the distribution of public resources—(1) fairness of the distribution and (2) benefits for voters resulting from the distribution—on evaluation of the decision-maker in three important aspects: (a) trust in the decision-maker, (b) willingness to vote for the decision-maker, and (c) support for the policy solution. By a combination of these five items, we reach the following set of three hypotheses based on the reasoning explained below:

Trust hypothesis (1): People are more inclined to trust a decision-maker distributing resources in an unfair manner if they benefit from the distribution.

Willingness to vote hypothesis (2): People are more inclined to vote for a politician distributing resources in an unfair manner if they benefit from the distribution.

Policy support hypothesis (3): People are more inclined to support a policy solution that has been adopted unfairly if they benefit from the policy.

While trust, willingness to vote, and policy support each express a slightly different aspect of the relationship between voter and politician, all three hypotheses respond to the same theoretical and empirical argument. Each of our hypotheses addresses the fundamental principle of procedural fairness theory, which is based on the necessity of impartiality on the part of decision-makers. This theory works with the assumption that voters place great emphasis on the motives behind political decisions and that they refuse to support decisions that are motivated by the decision-maker’s desire for personal gain (Leventhal 1980; Tyler 1990). In other words, politicians should act neutrally and impartially in order to gain the support of voters. Moreover, fair and unbiased decision-making should strengthen voter trust in the decision-maker (De Cremer 2004) and increase electoral support. These tendencies were confirmed by Bøggild (2016) in his study.

However, as some other studies show, this mechanism may not apply under all circumstances; sometimes voters can tolerate or overlook unfair decision-making, especially in situations in which they themselves are benefiting from the outcomes of these policies (Braidwood 2015; Tyler 1990). As Brockner and Wiesenfeld (1996, 191) noted, “when procedural justice is relatively low, outcome favorability is more apt to be positively correlated with individuals’ reactions.” Hence, we presuppose that people in post-communist countries like Slovakia and the Czech Republic have similarly learned to be less sensitive to the unfair behavior of politicians, a supposition that led us to formulate the hypotheses above. If we manage to find similar patterns in both countries under the scope of our study, we can reasonably expect that a similar trend could apply to a wider range of countries with similar historical and political trajectories, particularly within the CEE region.

The patterns of voters’ attitude reactions that we anticipate may be mainly due to the general population’s negative experience of the post-socialist transition, which continues to be characterized by strong clientelism (Wolchik and Curry 2018). Moreover, the CEE countries are less economically efficient, which is closely related to the levels of post-materialist values of their inhabitants (Pavlović 2015). It can, therefore, be assumed that, due to the combination of all these factors, people in CEE (unlike Western Europeans) are more interested in the output than the “throughput” (Schmidt 2013) and they give more priority to personal profit at the expense of fairness of the political process. To the best of our knowledge, there is as yet no research testing these hypotheses empirically, and our work aims to fill this gap.

Experimental Research Design and Data

To test the above-formulated hypotheses, we conducted two survey-embedded population-based experiments—the first in Slovakia, the second in the Czech Republic. Our experimental designs replicated and slightly adjusted Bøggild’s Study 1 (2016), which examined how fairness in the distribution of public resources and the benefits from such distribution influenced people’s evaluation of the decision-makers responsible for implementing these practices. Bøggild successfully confirmed the tendencies founded in the procedural fairness theory, meaning that his respondents cared about how the decision was made; however, his results were based mainly on convenience samples (students), and hence the external validity of his conclusions is questionable. Since we wanted to test and replicate findings from a specific “low corruption” West European context, we intentionally made our experiment materials as similar as possible to Bøggild’s Study 1 (2016) in order to minimize any systematic bias and to facilitate accurate comparison. The same applied to the wording of the questions measuring the dependent variables described below.

Data were collected by the Slovak and Czech branches of the FOCUS survey research agency during September 2018 in Slovakia and February 2019 in the Czech Republic. Respondents were recruited for an Internet-based questionnaire (CAWI) which was completed by 700 participants in Slovakia and 1,025 participants in the Czech Republic. In both instances, the resulting sample per country constitutes a representative sample of the population, controlling for the distribution of age, gender, nationality, region, and settlement/municipality size.

The participants were randomly assigned to one of the four experimental groups in each country. The numbers of participants assigned to the groups are given in Table 1. We checked the overall characteristics of participants across the groups (age, gender, nationality, region, and settlement/municipality size) to ensure that random assignment of participants to each experimental condition worked as expected. There were no significant differences.

Table 1. Distribution of Participants into Experimental Groups

	Slovakia		Czech Republic
	Fair distribution	Unfair distribution	Fair distribution	Unfair distribution
Benefit	Decision-maker: Slovak	Decision-maker: Slovak	Decision-maker: Czech	Decision-maker: Czech
	N = 186	N = 174	N = 269	N = 256
No benefit	Decision-maker: Hungarian	Decision-maker: Hungarian	Decision-maker: German	Decision-maker: German
	N = 171	N = 169	N = 241	N = 259
Total	700 participants		1,025 participants

Experimental Setup, Manipulation, and Measures

Before approaching the questionnaire, participants were presented with one of four short texts describing a mock-up situation: the European Union (EU) was about to launch a new grant scheme supporting health care; hence, it was necessary to select a few EU member states to take part in the pilot round.¹ Situating the story at the EU level (above national politics) decreased the possible bias resulting from the fact that respondents could bring their own partisanship/party affiliation into the evaluation of decision-makers. No contextual factors or political events at the time of the experiment should have affected the results in any specific direction. We are aware that attitudes toward the EU can factor into the process of respondents’ making evaluations, but thanks to the use of the experimental design and random assignment, this could not systematically affect the observed relationship.

Experimental manipulation was based on a 2 × 2 factorial matrix. The first dimension was fairness of the distribution and the respondents were presented with a description in which the responsible decision-maker (i.e., a fictional chairman of the European Parliamentary Committee) distributed the resources either fairly (i.e., the participating countries were selected randomly) or unfairly (i.e., the chairman intentionally included his own country in the pilot). The second dimension was benefit and its implementation varied between the countries. In Slovakia, the responsible decision-maker was either a Slovak politician who decided to distribute the money to Slovakia (included in a small group of EU countries), or a Hungarian politician distributing the money to Hungary (included in a small group of EU countries). In the experimental design implemented in the Czech Republic, the nationality of the decision-makers was Czech or German, respectively.² In both instances, the goal was to simulate that benefits were allocated to either the respondents’ home country (i.e., benefit) or their neighbors (i.e., no benefit). It must be emphasized that the home country of all respondents participating in the Slovak experiment was Slovakia, while the Czech Republic was the home country of all respondents in the Czech experiment. Therefore, if European resources were to be distributed to Hungary/Germany, none of the respondents should expect any direct benefit.

Three dependent variables—(1) trust, (2) willingness to vote for, and (3) support for the distribution policy—were operationalized by three questions presented to respondents after reading the text. Trust in the decision-maker was measured by the question: “How much trust do you have in a politician like [decision-maker’s name]?” Willingness to vote for a decision-maker was measured by the question: “To what extent would you be willing to vote in the elections for a politician like [decision-maker’s name]?” Lastly, support for the policy was measured by the question: “To what extent do you agree with the EU politicians’ decision about a health care trial scheme?” This set of questions allowed us to compare the evaluation of the decision-maker as a responsible actor (first and second questions) with the citizen’s perception of the pork-barrel practice as policy (third question).

With all three questions, respondents were offered an 11-point scale ranging from 0 to 10. The questionnaire form included an anchor at each extreme: Zero represented “complete lack of trust,” “definitely would not vote for,” or “completely disagree,” respectively. Ten meant the opposite, namely “definitely trust,” “definitely would vote for,” and “completely agree.”

Estimation Strategy

In order to examine the effects of benefit and fairness on every dependent variable, we built five models for each factor; that is, (1) trust in the decision-maker, (2) willingness to vote for the decision-maker, and (3) support for the policy. The first regression model in each part of the analysis included only the two independent variables—benefit and fairness. The second model added a country dummy variable to control for the impact of country-specific characteristics on statistical results. The third model included a two-way interaction term between benefit and fair distribution to examine whether their effects are mutually dependent. The fourth regression implemented three-way interactions between the two main independent variables and a country dummy for the Czech Republic (coded as 1 if the survey interview took place in the Czech Republic and coded as 0 if the participant took part in the experiment carried out in Slovakia). This dummy enables us to examine the effects of the two main independent variables—benefit and fairness—separately for Slovakia and the Czech Republic. If the findings are similar for both countries, we can conclude that the findings are not influenced by the country-specific context or other unobserved factors typical for one of these countries. Instead, we will be able to propose that the findings are robust and can be generalized to other countries similar to the Czech Republic and Slovakia, such as those in other parts of Central and Eastern Europe.

Given the random assignment and equally distributed sociodemographic characteristics, any observable and unobservable confounders were expected to cancel out their effects across the groups. Despite these expectations, we estimate also the fifth models in each group (i.e., models 5, 10, and 15 in Tables A1–Tables A3, respectively), which include a set of individual-level control variables to demonstrate that none of the personal background characteristics constitute a potential confounder that would bias the findings of the statistical models. Various socio-demographic characteristics could be claimed to impact individuals’ motivations with respect to their evaluation of politicians or policies (which is the goal of this experiment as well). Following the logic of statistical modeling, including these potential confounders in the model(s) holds their impact at zero, which enables us to examine the effects of the main independent variables—benefit and fairness—without being worried that the results are biased by some sociodemographic factor.

The set of control variables includes gender (male/female); age (measured in five categories: 18–29, 30–39, 40–49, 50–59, and 60+); education (measured in four categories: Primary school, Secondary school without graduation examination, Secondary school with graduation examination, College/University); and settlement/municipality size (measured in four categories: Less than 4,999, 5,000–19,999, 20,000–99,999, More than 100,000).

The results of regressions are presented in Tables A1–Tables A3 included in Appendix A (for the sake of smoother presentation of the findings). Trust in the decision-maker is tested in Models 1–5, results for Willingness to vote for the decision-maker are presented in Models 6–10, and finally, results for Support for the policy are shown in Models 11–15.

Despite estimating the models with individual-level controls, as well to check whether our randomization worked as intended, we followed the standards for experimental studies; thus, the visualized marginal effects of the relevant variables—benefit and fairness—are based on the model specifications interacting these two factors and controlling for the country where the survey experiment took place. These models—namely models 3, 8, and 13 in Tables A1–Tables A3, respectively—do not control for the individual-level control variables. Nevertheless, as can be seen in Tables A1–Tables A3, the results are similar across various model specifications controlling for potential confounders. Therefore, the decision to visualize the effects has no impact on the results and the findings would be similar if any other model specification was chosen.

Analysis of Benefiting Voters’ Perceptions of (Un)fair Decision-Makers

Our survey experiment examines the effect of two factors—(1) benefiting from the distribution of public resources and (2) the fairness of such a distribution in three respects reflected in our hypotheses. We focus on how these factors influence the individual’s (1) trust in the decision-maker, (2) willingness to vote for the decision-maker, and (3) support for the policy determining the above-mentioned aspects of the funding. In general, politicians are incentivized to direct finances to the regions where they can expect electoral gains (Costa-i-Font, Rodriguez-Oreggia, and Lunapla 2003; Denemark 2000, 2014; Milligan and Smart 2005); however, as we hypothesized in the theoretical section reflecting procedural fairness theory (Leventhal 1980; Thibaut and Walker 1975; Tyler 1990), we may ask whether such unfair behavior aggravates the public’s perception of the decision-maker and the policy.

In the specific context of Slovakia and the Czech Republic, where a relatively high occurrence of corruption is typical (Transparency International 2019), our respondents were always inclined to consider the decision-maker more trustworthy by roughly 1.6–1.8 points on an 11-point scale (at p < .01) as long as they benefited from the resource distribution. We estimated five various model specifications (including one controlling for potential individual-level confounders), all of which yield similar results (see Table A1 in Appendix A).

When it comes to fairness, respondents tend to reveal a higher trust in the decision-maker. However, it is higher by 0.6–0.7 points on an 11-point scale (at p < .01). This amounts to only one-third of the effect size of benefiting from an unfair distribution. In other words, the effect of benefiting from a resource distribution is three times larger than perceiving the decision-maker’s behavior as fair or impartial. Moreover, these findings are roughly consistent in Slovakia and the Czech Republic (see Figure A1 in Appendix A).

Figure 1 showcases the effect based on Model 3 (from Table A1, included in Appendix A), which interacts benefit and fairness while controlling for the country where the survey experiment took place. The same visualization strategy, following the standards for experimental studies, is used in the subsequent figures, as well.

Figure 1. Trust in the decision-maker: marginal effects of interaction terms based on Model 3 in Table A1, included in Appendix A.

Figures for individual countries can be found in Appendix A as Figure A1.

Hence, the results provide strong empirical support for the Trust hypothesis (1), confirming that if people benefit from the distribution of public resources, they are inclined to trust the decision-maker even if the resources are distributed in an unfair manner.

When it comes to the willingness to vote for a decision-maker distributing resources to the respondent’s benefit, the results are roughly comparable to those for trust. Individuals taking part in our experiment were more likely to indicate higher willingness to cast a ballot for the decision-maker whose actions benefit them by approximately 1.8–2.0 points on an 11-point scale (at p < .01).

Also in this scenario, fairness slightly increased the intention of individuals to vote for the decision-maker (at p < .01); however, the effect was roughly three times smaller, with coefficients ranging from 0.6–0.7 (see Table A2 in Appendix A). This means that benefiting from resource distribution increases the motivation of Slovaks and Czechs to vote for the responsible decision-maker to larger degree than whether the distribution is carried out in an impartial manner. Moreover, the findings are again consistent across the Czech Republic and Slovakia (see Figure A2 in Appendix A).

The effects are visualized in Figure 2, which is based on Model 8 (see Table A2 in Appendix A), which interacts benefit and fairness and controls for the country of the experiment. However, the results would be similar if estimates of any other model specification were chosen. Hence, it is possible to conclude that the empirical evidence strongly supports the Willingness to vote hypothesis (2) and that people are, indeed, more inclined to vote for a decision-maker distributing resources in an unfair manner if they benefit from the distribution.

Figure 2. Willingness to vote for the decision-maker: marginal effects of interaction terms based on Model 8 in Table A2 in Appendix A.

Figures for individual countries can be found in Figure A2 in Appendix A.

In the case of support for the policy, if the public resources were directed to the home country of the respondent, the participants tended to evaluate the policy in a more positive light by about 2.6 points on an 11-point scale (at p < .01). This means that benefiting from the resource distribution provided a positive boost to the evaluation of the policy related to the distribution of resources. Again, the effect of fairness was much smaller (coefficients are around 0.3 for the dependent variable measured on 11-point scale) and even failed to reach the threshold for statistical significance in the most complex models examining the country variations and controlling for potential individual-level confounders (see Models 14 and 15 in Table A3 in Appendix A). Therefore, the gap between the effects of benefit and fairness is even wider when it comes to the evaluation of the policy, compared to the perception of the decision-maker.

The results for policy support are displayed in Figure 3. They are based on Model 13 (see Table A3 in Appendix A) interacting the two main variables—benefit and fairness—and controlling for the country where the experiment took place. However, similar results are generated by all five model specifications (see Table A3). Thus, the findings consistently suggest that benefiting from the resource distribution has a positive impact on evaluations of the distribution policy. Moreover, the results are again roughly comparable between Slovakia and the Czech Republic (see Figure A3 in Appendix A). These observed trends thus provide strong empirical support for the Policy support hypothesis (3).

Figure 3. Support for the policy: Marginal effects of interaction terms based on Model 13 in Table A3 in Appendix A.

Figures for individual countries can be found in Figure A3 in Appendix A.

Overall, it is possible to conclude that the fairness (or impartiality) of the distribution of public resources does matter for the Slovak as well as the Czech participants in our experiments, as is expected in procedural fairness theory (Leventhal 1980; Thibaut and Walker 1975; Tyler 1990). However, the effect is relatively small in its magnitude. In the scenarios with partiality in the resource distributions, if the home country of the respondent benefits from the funds, this provided a comparatively much larger boost to the trustworthiness and willingness to vote for the decision-maker, as well as support for such a distribution policy. This means that benefiting from the resource distribution has a larger positive impact on evaluations of the actors responsible for the policy as well as for the policy itself, especially whe compared to the effect of fair and impartial policymaking. Moreover, these trends are consistent across two countries, Slovakia and the Czech Republic, and our sample was representative of both populations.

These results become even more interesting if we compare them to Bøggild’s original Study 1, the findings of which we tried to replicate. Bøggild’s results (2016, Tables 1, 907) are in stark contrast to ours. In his study, the benefit to the experimental subjects (or outcome favorability, as he refers to it) had no effect. The coefficients were small and not statistically significant. However, the effect of fairness (or “impartial decision-maker,” as he refers to it) was statistically significant and positive, indicating that Bøggild’s subjects were inclined to positively evaluate the decision-maker who distributes resources in a fair manner. This difference between Denmark, on the one hand, and the Czech Republic and Slovakia, on the other, suggests that the evaluation of pork-barrel practices is contextually conditioned; however, more comparative research needs to be done to capture the particular mechanism and causes of this trend.

Conclusion

This study examined the evaluation by voters of pork-barrel practices by politicians in the distribution of public financial resources. Political science literature is rich in explanations of why it is essential for parties and politicians to employ pork-barrel practices to secure reelection; however, little is known about the effects of these inherently unfair practices on voters’ evaluations of the decision-makers. In Denmark, Bøggild (2016) reached a conclusion that participants in his experiments were less inclined to trust and vote for politicians and to support their policies when they were adopted in an unfair manner. However, Bøggild’s conclusion could be strongly driven by the fact that Denmark is the least corrupt country in the world (Transparency International 2019) and, thus, its people may be more sensitive to the partial distribution of resources.

Therefore, we designed an experiment replicating Bøggild’s study and positioned it in the context of CEE (specifically the Czech Republic and Slovakia), where corruption is much more common (Transparency International 2019). We conducted two survey-embedded experiments with 700 and 1,025 respondents, consisting of a representative sample of the Slovak and the Czech populations, respectively. Our analysis indicated that fairness in the distribution of public resources has some positive effect on respondents’ evaluations of the decision-maker and the policy at stake, but the effect was at least three times smaller compared to that when the policy resulted in benefits to the participants. If the decision brought financial gains to their country, respondents revealed a significant and positive boost in their evaluations of the decision-maker as well as the policy itself. These empirical results from the Czech Republic and Slovakia allow us to conclude that people’s perceptions of pork-barrel politics are strongly conditioned by the socio-political context. Therefore, we suggest that future studies build on these findings and explore other contextual determinants that drive people’s perceptions of clientelist practices.

Our research shows no substantial differences between the national results. In both the Czech Republic and in Slovakia we found similar patterns in behavior: people’s evaluations of both the policies and the political representatives are primarily affected by the obtained benefits. Given the political, economic, and cultural features of the two countries, we are convinced that these findings could be generalized to the wider CEE region, which includes other countries with similar political, economic, and historical trajectories.

These results, however, contradict those of Bøggild (2016) that people are less likely to trust and vote for politicians who implement policies in a partial manner. This discrepancy indicates variation among European countries concerning the roles of fairness and benefits in people’s evaluations of the decision-makers responsible for distribution policies. We offer some thoughts to explain these differences: First, countries differ with respect to the level of corruption. That may impact people’s sensitivity to unfair behavior in politics. Thus, in Western Europe, where a lower level of corruption is characteristic compared to the CEE countries (Transparency International 2019), a higher demand from society for fairness in political decision-making is likely to occur. Second, the standard of living might affect the evaluation of policy outputs: people in more financially secure societies such as those in Western Europe do not need to rely on material benefits to the same extent as those in the economically worse-off CEE countries.

Finally, our findings, compared to those obtained in Denmark, may result from a methodological choice. Our study employed population-based survey experiments. In contrast, Bøggild (2016) conducted laboratory experiments with a convenient sample—students. Students represent a specific societal segment that differs from the general population in various ways. What is more, the literature shows that using student samples in experiments has its limits and may lead to different results compared to population-based experiments (Hooghe et al. 2010; Mjelde et al. 2016; Peterson and Merunka 2014). Hence, the selection of samples may explain the observed differences to a certain degree. Given that population survey experiments do not suffer from low external validity, we suggest that other researchers who focus on experimental testing of voters’ perceptions of pork-barrel politics construct their samples accordingly.

Supplementary Files

Supplementary files (experimental design, data, and code) are available at https://doi.org/10.7910/DVN/GYHP2Q.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was funded by the Czech Science Foundation (Grant no. GA18-16928S).

Notes

1. Full wording of the experimental design and replications files are available in supplementary files at 10.7910/DVN/GYHP2Q.

2. Competing countries were selected in a way that allowed the expected effect to be verified under different conditions. Hungary shares a post-communist experience with Slovakia and is economically and politically similar. Germany represents an economically more efficient country and differs from the Czech Republic not only in its historical experience but also in its size, population composition, and so on.

Appendix A

Table A1. Trust in the Decision-Maker: Results of Regression

	Dependent variable: Trust in decision-maker
	(1)	(2)	(3)	(4)	(5)
Constant	2.442*** (0.094)	2.227*** (0.114)	2.173*** (0.127)	2.172*** (0.172)	1.844*** (0.256)
No benefit	(Reference category)
Benefit	1.551*** (0.108)	1.551*** (0.108)	1.658*** (0.153)	1.794*** (0.242)	1.811*** (0.240)
Unfair distribution	(Reference category)
Fair distribution	0.597*** (0.108)	0.601*** (0.108)	0.709*** (0.154)	0.659*** (0.243)	0.679*** (0.241)
Slovakia	(Reference category)
Czechia		0.357*** (0.110)	0.358*** (0.110)	0.361 (0.221)	0.421* (0.221)
Interactions:
Benefit*Fair distr.			−0.211 (0.215)	−0.372 (0.338)	−0.422 (0.336)
Benefit*Czechia				−0.229 (0.312)	−0.248 (0.310)
Fair distr.*Czechia				0.086 (0.315)	0.036 (0.313)
Benefit*Fair distr.*Czechia				0.269 (0.439)	0.317 (0.437)
Individual controls:
Male	(Reference category)
Female					0.431*** (0.109)
Age:
18–29	(Reference category)
30–39					−0.033 (0.183)
40–49					−0.358* (0.191)
50–59					−0.185 (0.196)
60+					0.019 (0.175)
Settlement size:
Less than 4 999	(Reference category)
5 000– 19 999					−0.115 (0.154)
20 000– 99 999					0.055 (0.140)
More than 100 000					−0.083 (0.157)
Education:
Primary school	(Reference category)
Secondary school without graduation examination					−0.010 (0.193)
Secondary school with graduation examination					0.248 (0.191)
College/University					0.536** (0.212)
Observations	1 725	1 725	1 725	1 725	1 725
R2	0.123	0.128	0.128	0.129	0.147
Adjusted R2	0.122	0.126	0.126	0.126	0.138

*p < 0.1; **p < 0.05; ***p < 0.01; OLS regression with standard errors in parentheses.

Dependent variable: “How much trust do you have in a politician like [decision-maker’s name]?” [0 = complete lack of trust … 10 = definitely trust].

Figure A1. Trust in the decision-maker: Comparison between Slovakia and the Czech Republic. Marginal effects of interaction terms based on Model 4 in Table A1.

Table A2. Willingness to Vote for the Decision-Maker: Results of Regression

	Dependent variable: Willingness to vote for the decision-maker
	(6)	(7)	(8)	(9)	(10)
Constant	2.016*** (0.098)	1.818*** (0.119)	1.789*** (0.132)	1.722*** (0.179)	1.353*** (0.267)
No benefit	(Reference category)
Benefit	1.817*** (0.112)	1.818*** (0.112)	1.875*** (0.159)	2.060*** (0.252)	2.084*** (0.250)
Unfair distribution	(Reference category)
Fair distribution	0.565*** (0.112)	0.568*** (0.112)	0.628*** (0.161)	0.693*** (0.253)	0.720*** (0.251)
Slovakia	(Reference category)
Czechia		0.329*** (0.114)	0.329*** (0.114)	0.440* (0.231)	0.517** (0.230)
Interactions:	(Reference category)
Benefit*Fair distr.			−0.115 (0.225)	−0.346 (0.353)	−0.405 (0.351)
Benefit*Czechia				−0.308 (0.325)	−0.347 (0.323)
Fair distr.*Czechia				−0.109 (0.328)	−0.170 (0.326)
Benefit*Fair distr.*Czechia				0.386 (0.458)	0.448 (0.455)
Individual controls:
Male	(Reference category)
Female					0.480*** (0.114)
Age:
18–29	(Reference category)
30–39					0.090 (0.191)
40–49					−0.219 (0.199)
50–59					−0.338* (0.204)
60+					−0.189 (0.182)
Settlement size:
Less than 4 999	(Reference category)
5 000– 19 999					0.009 (0.161)
20 000– 99 999					0.140 (0.146)
More than 100 000					0.043 (0.163)
Education:
Primary school	(Reference category)
Secondary school without graduation examination					−0.036 (0.201)
Secondary school with graduation examination					0.221 (0.199)
College/University					0.521** (0.221)
Observations	1 725	1 725	1 725	1 725	1 725
R2	0.144	0.148	0.148	0.149	0.167
Adjusted R2	0.143	0.147	0.146	0.146	0.159

*p < 0.1; **p < 0.05; ***p < 0.01; OLS regression with standard errors in parentheses.

Dependent variable: “To what extent would you be willing to vote in the elections for a politician like [decision-maker’s name]?” [0 = definitely would not vote for … 10 = definitely would vote for].

Figure A2. Willingness to vote for the decision-maker: comparison between Slovakia and the Czech Republic. Marginal effects of interaction terms based on Model 9 in Table A2.

Table A3. Support for the Policy: Results of Regression

	Dependent variable: Support for policy
	(11)	(12)	(13)	(14)	(15)
Constant	2.477*** (0.110)	2.433*** (0.134)	2.392*** (0.149)	2.550*** (0.202)	2.215*** (0.301)
No benefit	(Reference category)
Benefit	2.605*** (0.126)	2.605*** (0.126)	2.685*** (0.179)	2.599*** (0.283)	2.623*** (0.282)
Unfair distribution	(Reference category)
Fair distribution	0.305** (0.126)	0.306** (0.126)	0.387** (0.181)	0.268 (0.284)	0.284 (0.284)
Slovakia	(Reference category)
Czechia		0.073 (0.129)	0.074 (0.129)	−0.187 (0.259)	−0.141 (0.260)
Interactions:	(Reference category)
Benefit*Fair distr.			−0.159 (0.253)	−0.359 (0.396)	−0.404 (0.395)
Benefit*Czechia				0.139 (0.365)	0.121 (0.364)
Fair distr.*Czechia				0.194 (0.369)	0.163 (0.368)
Benefit*Fair distr.*Czechia				0.345 (0.514)	0.370 (0.513)
Individual controls:
Male	(Reference category)
Female					0.348*** (0.129)
Age:
18–29	(Reference category)
30–39					0.160 (0.215)
40–49					−0.373* (0.224)
50–59					−0.188 (0.230)
60+					−0.113 (0.205)
Settlement size:
Less than 4 999	(Reference category)
5 000– 19 999					0.022 (0.181)
20 000– 99 999					0.169 (0.165)
More than 100 000					0.117 (0.184)
Education:
Primary school	(Reference category)
Secondary school without graduation examination					−0.025 (0.227)
Secondary school with graduation examination					0.201 (0.225)
College/University					0.492** (0.250)
Observations	1 725	1 725	1 725	1 725	1 725
R2	0.201	0.201	0.202	0.204	0.215
Adjusted R2	0.200	0.200	0.200	0.200	0.207

*p < 0.1; **p < 0.05; ***p < 0.01; OLS regression with standard errors in parentheses

Dependent variable: “To what extent do you agree with the EU politicians’ decision about a health care trial scheme?” [0 = completely disagree … 10 = completely agree]

Figure A3. Support for the policy: comparison between Slovakia and the Czech Republic. Marginal effects of interaction terms based on Model 14 in Table A3.