Is there a demand for autocracies in Europe? Comparing the attitudes of Hungarian and Italian university students toward liberal democratic values inspired by János Kornai

In the European Union right-wing and left-wing populist parties are increasingly becoming stronger. Meanwhile in Central and Eastern Europe autocracies are emerging and becoming stabilized. Italy and Hungary are two notable examples of these processes. Italy is the only country in Western Europe where a coalition of purely populist parties won an election, while Hungary has the most mature autocracy in the European Union. By using survey methodology, we examined the preferences of Hungarian and Italian students regarding the values of liberal democracy. We seek answers to the questions whether there are any significant differences between the proportion of Hungarian students and Italian students who identify themselves with the values of liberal democracy and which of these values of liberal democracy they consider to be protected values. Based on our results, we claim that students from both countries are more likely to support liberal democratic values than to support either right-wing or left-wing populist values, even if the distributions of the two groups differ significantly. We found that Italian students adhere more strongly to liberal democratic values, while Hungarians are more open to left-wing and right-wing populism. Our results confirm that in Hungary, because of the values that many people hold, conditions are conducive for establishing a sustainable autocracy, while in Italy, the demand for such a system is much weaker.


Introduction
Across Central and Eastern Europe, a new political system is emerging and becoming stabilized. It has variously been labeled as a hybrid regime (Bogaards 2009), an illiberal democracy (Zakaria 1997), competitive authoritarianism (Levitsky and Way 2010) and autocracy (Kornai 2016). Such transformations have been observed in Hungary (Ádám 2019;Kornai 2015Kornai , 2016 and other Central and Eastern European countries (Kapidžić 2020;Sata and Karolewski 2020), and similar tendencies have gained momentum even in Western Europe (Bernhard and Kriesi 2019). Kornai (2000) suggests that to understand such social transformations, we should focus on institutions-both formal and informal-and on the incentives of social actors. By employing Kornai's approach, we aim to develop a deeper understanding of why autocracies have emerged and stabilized in Central and Eastern Europe while liberal democracies and the institutions that safeguard them have remained stable in other areas of the European Union. We contend that the incentives for individuals to act collectively, such as voting in elections or choosing a bundle of policy proposals, are strongly connected to the personal values that one accepts and defends.
In line with Kornai's (2016) work, we define such newly developed autocracies as political systems in which multiparty elections are held regularly but are not necessarily fully free and certainly not fair. In autocracies, the main liberal rights (political and civic rights, horizontal accountability, and an electoral regime) are eliminated by those in power. Autocrats attack liberal democratic values to develop a system in which they can maintain power indefinitely. 1 Populism is a prevalent feature of autocracies (Rodrik 2018;Wintrobe 2018). In our approach, we focus on one of the points from Schmitter's (2019, p. 77) composite definition, which conceives of populism as a strategy of opportunistic leaders: "The core of that strategy is 'opportunism'-the search for issues and appeals that are either 'transversal' (i.e., cut across those articulated by existing parties) or 'nascent' (i.e., are ignored or understated by existing parties)".
Populism falls into one of two general categories, right-wing or left-wing populism, based on the types of messages it sends to the electorate. Accordingly, we define right-wing populism as a strategy that cleaves society on cultural and ethnic grounds and left-wing populism as a strategy that cleaves society on economic grounds (Halmai 2019;Huber and Schimpf 2017;Mudde 2007;Rodrik 2018). The typical policy recommendations of each type of populism can be deduced from opportunistic behavior aimed at maximizing power, which is the ultimate goal of all populists.
Both types are antagonistic to one or more values of liberal democracy (Zanotti and Rama 2020), which are considered by Kornai (2018) to be protected values. 2 Protected values are those by which citizens abide even when they would benefit from abandoning them; citizens will not accept any kind of trade-off regarding such values (Baron and Spranca 1997).
Our hypothesis is the following: If the majority of Hungarian voters had adopted the same value system that Kornai does, in which liberal democratic values are a "top priority" (Kornai 2018, p. 62), then an "illiberal turn" (Rupnik 2012) could have not happened, since Hungarians would have rejected alternatives that diverge from those values. 1 Kornai (2016, p. 565) states that in an autocracy "[t]he government cannot be removed through a peaceful and civilized procedure". However, autocrats can lose local elections, like Viktor Orbán did in Budapest in 2019. 2 In a momentous speech at a conference celebrating his 90th birthday, Kornai (2018, p. 62) stated: "I don't accept a simple trade-off, saying that an X thousand-dollar increment in growth is equivalent to the imprisonment of a freedom fighter. That is not acceptable for me. I do have a ranking, or more specifically, a lexicographic ranking, where my top priority goes to freedom, liberties, liberty of speech, liberty of free expression of thoughts, liberty of the press, autonomy for the individual and for voluntary communities, democracy, state of law". Wintrobe's (2018) model is a suitable theoretical starting point for our research since he shows how an autocracy can emerge from a liberal democratic political system. 3 He identifies three ways in which liberal democracy becomes self-enforcing: (1) strong democratic norms; (2) constitutional protections, regular elections and relative income equality; and (3) widespread social commitment to liberal democratic values. Looking at the recent trends in Central and Eastern Europe, the first two mechanisms have not been sufficient safeguards for liberal democracy. If the values of the people diverge from liberal democratic principles, then a democratically elected leader may offer policy proposals that are closer to the preferences of the majority, such as proposing the political repression of a specific disliked group. Such a process goes on to maximize the leader's power. However, if the loyalty of the citizens shrinks when repression rises, then a strongman has no chance to build a full-fledged dictatorship. Nevertheless, if the power gained through such repression is greater than the loss in loyalty, then a stable autocracy can emerge. The turning point depends on society's values regarding liberal democracy.
It is a realistic assumption that a strongman who creates an autocracy is a populist leader, offering right-wing and/or left-wing values to voters. Whether his policies are characteristically left-wing or right-wing is determined by the values of the voters, which the strongman uses to indulge them opportunistically. To maximize her power, an autocratic leader formulates a mixture of policies based on left-and rightwing populist ideas (Ivaldi and Mazzoleni 2020).
Contrary to Wintrobe (2018), we believe that most leaders who build up autocracies propose both cultural and economic populist policies to maximize their power, focusing only on what resonates in the hearts and minds of voters.

Research questions
Based on the discussion above, we set out to study whether liberal democratic values are protected values in Hungary and Italy and, if not, whether voters are open to right-wing or left-wing populist policy proposals. Thus, following Baron's (1999Baron's ( , 2002 work on protected values, we decided to carry out survey-based research. The survey was designed to gather data to ascertain the sets of values held by Hungarian and Italian university students and to test our hypotheses statistically. We chose Hungary and Italy as examples because both countries are unique: Hungary is the leading example of autocracy in Central and Eastern Europe (Bustikova and Guasti 2017; Győrffy 2020), while Italy is a "showcase of populist parties" (Verbeek and Zaslove 2016, p. 304) and is the only Western European county that has been led by a government comprised entirely of populist parties.
Our research questions and hypotheses are as follows: • RQ1: Do any significant differences exist between the proportion of Hungarian students and the proportion of Italian students who identify with liberal democratic values?
• H0: The proportions of Hungarian and Italian students who identify with liberal democratic values do not differ significantly.
• RQ2: Do Hungarian and Italian students differ in terms of which liberal democratic values they consider to be protected?
• H0: Hungarian and Italian students do not differ in terms of which values of liberal democracy they consider to be protected values.
• RQ3: In which country would it be easier to establish a sustainable autocracy, Hungary or Italy?
• H0: It would be easier to establish a sustainable autocracy in Italy than in Hungary.

Methodology
We conducted an online survey 4 consisting of three parts. The first part was designed to gather demographic data from the respondents. Each of the other two parts contained a set of ten questions, and each question in the first set had a parallel question in the second set. The question pairs were designed to measure a given value of liberal democracy.
In the first set of questions, we asked whether the student considers liberal democratic values to be protected. The questions corresponded to ten liberal democracy values that were deemed to be the most characteristic of such political systems based on the prior literature. Our main goal was to determine if respondents considered a given value to be a protected value and then to compare the answers from both sets of questions to see if their answers were consistent.
The second set of questions was designed to give a more precise picture of the given value by providing a specific proposal clearly violating the given democratic value. In that set, every question had four possible answers. One answer indicated approval of the proposed policy by evoking a reason associated with right-wing populism. The second possible answer indicated approval based on left-wing populist reasoning. The third possible answer offered the respondent a trade-off-giving up some (not so strongly held) protected values in exchange for a policy enhancing the utility of the respondent directly. The fourth possible answer indicated that the respondent rejected the policy proposal.
Within both sets of questions, we targeted ten values of liberal democracy in order to measure the respondents' attitudes towards specific values of liberal democracy. They include the following: I. political rights: (1) the right to assemble, associate and demonstrate and (2) freedom of speech, opinion and the press; II. civil rights: (3) equal opportunity and protection against discrimination, (4) the protection of minority rights, (5) human rights (such as the protection of life), (6) rule of law, (7) freedom of religion, and (8) protection of property rights; III. electoral regime: (9) universal and equal voting rights in regular, free and fair elections; and IV. horizontal accountability: (10) the separation of powers.

3
The Italian students received an English version, while the Hungarians received a Hungarian version of the survey. The translation was carried out with a double-blind and backtranslation procedure (Brislin 1970). By implementing such procedures, survey questions consciously were formulated in a way that eases translation and understanding. First, a pilot survey was administered to 10 non-Hungarian 5 and 10 Hungarian students to test the primary survey instrument. We also interviewed the students to ensure that both language versions were understood as intended and no meaning was lost or misinterpreted. Based on that test, minor changes to the questionnaire were made, and then the primary survey was administered.

Respondents
The two sample groups had the following characteristics: 103 people (48 women, 55 men) from Universitá di Trento in Italy and 169 people (113 women, 56 men) from Corvinus University of Budapest in Hungary. All respondents were majoring in economics and business administration at either the bachelor's or master's level; their ages ranged from 18 to 25 years old. 6 We followed a convenience sampling method to choose the groups of students; therefore, the results of the analysis should be carefully handled with this in mind. The reader should also be cautious when drawing conclusions from the results because the convenience sampling method is non-probabilistic, and the sample was not representative.

Target variables
Based on the survey results, we constructed seven target variables, where each variable had a numeric range between 0 and 10 computed from the answers given to the survey questions. Table 1 summarizes the purpose of each variable: Our target variables are measured on an ordinal scale, which means that we had to conduct tests that allow such target variables. Owing to that feature, we tested our data using a combination of methods, namely, ordered logit tests, nonparametric Mann-Whitney U-tests and chi-square independence tests. We specified ordered logit models for the seven computed target variables regarding gender, age and citizenship.
In addition, we performed a nonparametric test for the seven computed target variables regarding gender groups and citizenship groups. The main idea behind that test is that if a statistically significant difference exists between the two groups of respondents, then the mean rank helps to decide which group is more likely to show the behavior measured by the target variable's score.

Results
We present the results of the survey in the order of the research questions.

First research question: do any significant differences exist between the proportion of Hungarian students and the proportion of Italian students who identify with liberal democratic values?
The results reported in Table 2, show that the Hungarian students marked answers corresponding to liberal democratic values 6.78 times, on average, while the Italian students chose such answers 7.75 times on average. Hungarian students chose both right-wing and left-wing populist answers more frequently than their Italian counterparts did. The frequency of answers indicating self-interested motives was minimal in both countries. In both countries, left-wing populist replies were more frequent than right-wing populist replies. Median ( Although Hungary already is seen as an autocracy 7 (Lührmann et al. 2020;Kornai 2016;Repucci 2020), many respondents chose statements corresponding to liberal democratic values.
We estimated ordered logit models to determine whether or not the distributions of the two groups (Hungarian and Italian students) regarding the target variables were identical (Table 3).
We found that, compared with Hungarian students, Italian students are more likely to identify with liberal democratic values (1% significance level). We found no statistically significant difference between age groups and genders when controlling for the other variables.
Looking at Table 4, we can conclude that, compared with Hungarian students, Italian students are less likely to be open to right-wing populist values. The result is statistically significant at the 5% level. No statistically significant differences can be seen between age groups or genders (controlling for all other variables).
If we look at the target variable regarding left-wing populism, we can state that, compared with Hungarian students, Italian students are less likely to be open to left-wing populist values ( Table 5). The result is statistically significant at the 1% level. No statistically significant differences can be seen between age groups or genders (controlling for all other variables).
Moreover, compared with Hungarian students, Italian students are less likely to be consistent in rejecting liberal democratic values (statistically significant at the 1% level). 8 We also performed the Mann-Whitney U-test to reinforce the robustness of our findings. The result was exactly the same as with ordered logit models, with a small size effect. 9 Based on those results, we can state that regarding the target variables with statistical significance, the Italian and Hungarian groups do not have similar distributions. To observe the directions of the between-group differences, we looked at the mean ranks of the test output. Looking at the target variable of identifying with liberal democratic values, the mean rank of Italian students dominates the Hungarian students' rankings stochastically, which means that they identified with liberal values more widely than Hungarian students did. For the target variables of openness towards right-wing and left-wing populist values, the mean ranks of Hungarian students is higher, which means that they identified with these values more widely than Italian students did. 10 Based on the estimates, we reject the null hypothesis of the first research question. The Italian and Hungarian students' attitudes towards liberal democratic values are not identical. Italian students adhered more to liberal democracy, while Hungarians were more open to left-wing and right-wing populism. According to our results, 14.6% of the Italian students and 11.8% of the Hungarian students chose answers indicating that they considered liberal democratic values to be protected. No students in our sample gave absolutely zero answers corresponding to liberal democratic values. A total of 65.1% of the Italian students and 44.9% of the Hungarian students chose eight or more answers indicating that liberal values are protected values. 11 Those subjects are voters who would never vote for parties that jeopardize liberal democracy. Populist, autocratic parties would not benefit from catering to that segment of the electorate; they would instead identify them as enemies.
Only 2.9% of the Italian students and 5.9% of the Hungarian students gave answers corresponding to liberal democratic values only one or two times. Those individuals would support right-wing or left-wing populist parties and would become the core supporters of a strongman seeking to establish an autocracy.
32% of the Italian students and 49.1% of the Hungarian students chose answers corresponding to liberal democratic values three to seven times. They are voters for whom parties would fight to win over, and they ultimately would decide whether a strongman could stabilize his power or not.
In addition to conducting an analysis based on our questionnaire, which is a snapshot of the current situation, we compared our results with the European Social Survey's (ESS) dataset to check the dynamics of the change in liberal democratic values. Italy participated in the ESS in 2002ESS in , 2004ESS in , 2012ESS in , 2016ESS in and 2018. The Hungarian dataset was collected during all nine survey The boundary between the group of students we assume never would vote for a party that rejects liberal democracy and the group we assume could be convinced to vote for a party with populist policies that erode liberal democracy is arbitrary. However, the result is robust; modification would affect neither the results nor our answer to the first research question. If the boundary were between 6 and 7 instead of 7 and 8, then the results would be the following: 59.8% of the Hungarians chose 7 or more values of liberal democracy as protected, 11.8% of them chose 3 or less, and 28.4% chose between 4 and 6. While 82.5% of the Italian students chose 7 or more such statements corresponding to the values of liberal democracy, 3.9% chose 3 or less and 13.6% chose between 4 and 6 such statements.
rounds between 2002 and 2018. The ESS did not measure all of the values we included in our survey. 12 However, some relevant questions were asked in 2002, 2012 and 2018. One of them was a statement related to whether "gays and lesbians should be free to live life as they wish", with which we can measure civil rights values (minority rights, equal opportunity) (Table 6).
If we look at the differences between the two countries, it can be seen that the gap between Hungary and Italy has widened consistently. The Italian data have been more or less stable; however, the tolerance or acceptance of exclusion has been growing rapidly in Hungary since 2012. 13 The picture seen here corresponds with the results of our survey.
Another statement with which we measured liberal democratic values is related to immigration. 14 Table 7 measures attitudes towards human rights.
The table shows that the difference between the numbers of Hungarian and Italian citizens who would accept letting many or some immigrants enter the country has declined, while the difference between Hungarians and Italians who would not let any immigrants 12 In 2012, the ESS focused partly on Europeans' understanding of democracy (Ferrin and Kriesi 2014). 13 Looking at the results of the ESS survey in 2010, it can be seen that the acceptance rate in Hungary fallen from 48.3% (2010) to 34.4% (2018). The source of data from 2010 is: ESS5-2010, ed.3.4. 14 The question we asked was as follows: "Now, using this card, to what extent do you think [country] should allow people of a different race or ethnic group from most [country] people to come and live here?" Source: European Social Survey datasets, ESS1-2002, ed.6.6. We should allow a few 25.9 23 34.9 We should allow none 9.9 12.5 15.8 enter the country has increased. The difference between the countries is substantial: 2.87 times more Hungarians than Italians would prefer to stop all immigrants from entering their country. Those numbers also are in line with our results and make the latter more robust. Based on our results, we claim that students from both countries are more likely to support liberal democratic values than to support nationalistic, ethno-cultural populist values or economic populist values, even if the distributions of the two groups differ significantly. However, we cannot say that each of the liberal democratic values had the same support in both countries. That is our second research question.

Second research question: do Hungarian and Italian students differ in terms of which liberal democratic values they consider to be protected?
Pearson's chi-squared test helps to answer that question. That test asks whether a statistically significant difference exists between the expected frequencies and the observed frequencies in one or more categories of a contingency table. To determine the size of the effect of the chi-square independence test, we adopted Cramér's V. 15 We find a statistically significant difference between Italian and Hungarian students: • At a significance level of 1% for the right to assemble and associate, human rights, universal and equal voting rights; • at a significance level of 5% for equality of opportunity and protection against discrimination, freedom of religion; and • at a significance level of 10% for the protection of minority rights.
No statistically significant differences are found regarding freedom of speech and opinion, property rights, rule of law, or separation of power. 16 When we looked at the strength of the association of values and citizenship, we found strong associations (Cramér's V ≥ 0.15) for all values of liberal democracy, which were statistically significant at the 1%, 5% or 10% level. Nevertheless, we found very strong associations for human rights (Cramér's V was 0.354) and for universal and equal voting rights (Cramér's V was 0.364). The results for universal and equal voting rights are particularly interesting because they are not only one of the core elements of liberal democracy but also essential for procedural democracy.
A smaller proportion of the Hungarian respondents than Italian respondents considered liberal democratic values to be protected. The Hungarians were proportionally more open to both left-wing and right-wing populism regarding all the values except for the protection of property rights and the rule of law. The differences are largest for the value of human rights, for which 27.81% of the Hungarian respondents gave a right-wing populist answer and only 0.97% of the Italian respondents did so, and for universal and equal voting rights, for which 72.19% of the Hungarian students chose a left-wing populist answer and only 37.86% of the Italian students did so. It is worth noting that the majority of the Hungarian students also identified with liberal democratic values except in relation to universal and equal voting rights (Table 8). 17 Pearson's chi-squared test and our descriptive statistical analyses 18 allow us to conclude that Italian students are more likely to consider the following values to be protected: the right to assemble and associate, equality of opportunity and protection against discrimination, human rights, the protection of minority rights, freedom of religion, and universal and equal voting rights. We found no significant differences between the Italian and Hungarian respondents regarding the values of freedom of speech and opinion, the rule of law, the separation of power, and property rights. 19 We therefore reject the null hypothesis of the second research question: a significant difference between Hungarian and Italian students exists with respect to the liberal democratic values they consider to be protected. We found that Italian students are more consistent in considering liberal democratic values to be protected values, while Hungarian students are 17 In Part 2-aside from universal and equal voting rights-less than 50% of the Hungarians identified with the values of equality of opportunity, protection against discrimination and human rights. 18 All the descriptive statistics can be found on the Public Choice homepage in a Power BI format. 19 We conducted consistency analysis as well. The results can be downloaded from the Public Choice website.

Third research question: in which country would it be easier to establish a sustainable autocracy, Hungary or Italy?
To answer our third research question, we checked whether the demand from Hungarian and Italian students is sufficient to construct a stable autocracy in Hungary or Italy. 20 The members of society who are fully committed to liberal democratic values are not the focus of a strongman, since they would never vote for him. The group of voters who fully reject liberal democracy likewise does not attract too much attention, since they would never turn to parties that advocate liberal democracy. The autocrat therefore needs to focus on voters whose preferences for liberal democratic values are unstable. If that uncommitted group of voters comprises a large fraction of the electorate, then the autocrat has a good chance of obtaining power and holding onto it. Examining our survey results, we can conclude that in both countries, a demand for an autocratic, populist strongman is evident. However, in Italy, the number of likely autocrat supporters plus those who have unstable value systems regarding liberal democracy make up a rather small portion of the students surveyed, while in Hungary, the same group represents a majority or at least a significant proportion of the students surveyed. 21 It also is clear from the results of the survey that the Hungarian students would not support a dictatorship in Hungary, since most of them at least partially adhere to liberal democratic values. The system primarily is internally constrained; external constraints are collateral (Bozóki and Hegedűs 2018).
The literature also suggests that right-wing populism is more fertile ground than left-wing populism for autocratic regimes (Halmai 2019). 22 As indicated, to answer our research question, we looked in depth at the composition of the group of students from our sample who did not indicate a clear preference for liberal democratic values. 23 We found that a larger proportion of Hungarian students systematically chose right-wing populist statements than did their Italian counterparts (Table 9).
The most striking difference is the case of the value attached to human rights; 42.2% of the Hungarians agreed with the right-wing populist statement, whereas none of the Italians did so. Similarly, regarding the protection of minority rights, 30.1% of the Hungarian 20 Please note that our research is rather exploratory, and we shed light on some social patterns that should be investigated further. 21 Based on Inglehart and Norris (2016), we can assume that more university students than the general population consider the values of liberal democracy to be protected values, which means that the ratio might even be higher in the given society. Moreover, people also consider other aspects of their value systems when they vote, which means that even in Italy, populist leaders can win elections and gain power. However, based on our results, we state that they cannot establish a stable autocracy, since the demand side in Italy does not reach the level needed to sustain an autocracy and demolish liberal democracy. Unlike Italy, the demand for right-wing and left-wing populist groups is stable in Hungary. 22 While right-wing populists see national states as non-pluralistic entities, left-wing populists do not exclude any group from society (Huber and Schimpf 2017). 23 Representing the students who have chosen 3-7 such statements, which correspond with the values of liberal democracy. The samples from both countries are rather small, so our results should be interpreted with caution. respondents selected the right-wing populist statement, while only 9.1% of the Italian students marked the same. It also is noteworthy that in both countries, strong reservations about giving voting rights to undereducated people are observed.
Based on our analyses, we argue that the value systems that people hold in Hungary create conditions that are conducive to establishing a sustainable autocracy, given that (1) a large proportion of the Hungarian respondents did not consider liberal democratic values to be protected and (2) a considerable proportion of the Hungarian respondents who did not demonstrate a stable preference for liberal democratic values also demonstrated a tendency towards right-wing populism, especially regarding human rights and the protection of minority rights. We likewise can assert that the value systems held by Italian people do not create conditions conducive to establishing a sustainable autocracy.
Therefore, based on our survey, we reject the third research question's null hypothesis and state that it would be easier to establish a sustainable autocracy in Hungary than in Italy.

Conclusion
Autocracy is becoming a prevalent political system across the world, including within the European Union. Populism is flourishing in both Eastern and Western Europe. In this paper, we investigated the demand for establishing sustainable autocracies. We examined the value systems of Hungarian and Italian students regarding liberal democracy to look for any differences in their preferences for each of the ten liberal democratic values. We applied survey methods to map out and understand the respondents' preferences. Although our research is merely explorative owing to the many limitations of our convenience samples, the results are robust and straightforward. First, Italian and Hungarian students do not have identical attitudes towards liberal democratic values. Italian students adhere more strongly to those values, while Hungarians are more open to left-wing and right-wing populism. Second, based on the statistical analyses, we can conclude that Italian students are more likely to consider the following values protected: the right to assemble and associate, human rights, and universal and equal voting rights (at a significance level of 1%); equality of opportunity and protection against discrimination, freedom of Table 9 The ratio of respondents' choices between liberal democratic values (LD), right-wing populism (RWP), left-wing populism (LWP) and self-interest seeking (SI) in the group of students who selected three to seven statements corresponding to liberal democratic values. Source: Authors' own calculation We examine only the values that are significantly different between the Hungarian and Italian students Based on those findings, we can conclude that a demand exists in both countries for an autocratic, populist strongman. However, in Italy, the number of unambiguous supporters of autocracy and those who have nonstable value systems regarding liberal democracy make up a much smaller proportion of the student body, whereas in Hungary, they represent the majority, or at least a significant proportion of it. Our results confirm that in Hungary, because of the values that many people hold, conditions are conducive for establishing a sustainable autocracy, while in Italy, the demand for such a system is much weaker.