ECONOMY, POLITCAL TRUST AND
POPULISM
ABSTRACT
Various scholars have demonstrated the effect of political trust on the support for populism. In recent years scholars have also tried to link macro and micro economic factors to the support for populism. This research combines both schools of literature to assess whether there is a mediating effect of political trust between economy and populist support. Drawing on the school of economic voting theory, this research investigates two levels on which political trust can have a mediating effect: the sociotropic and egotropic level. Using the European social survey data, I’ll demonstrate that the link between economy, political trust and populism is an inconsistent relationship.
Name: Max Biermans Student Number: 11143495 Master thesis Political science
2
INDEX
Abstract ... 1
Introduction ... 3
Populism ... 4
Support for populism ... 6
Political trust ... 7
Economy and political trust ... 8
Economic voting theory and political trust ... 8
Sociotropic ... 9
Egotropic ... 10
Political Trust and populism ... 11
Culture and populism ... 13
Explanation of the data ... 16
Explanation of the variables ... 17
Explanation of the model ... 18
A multi-level random intercept model ... 18
Controlling for country fixed effects ... 19
Grouping of the data ... 19
Regression results ... 21
H1: Economic downturn decreases the level of political trust. ... 22
H2: A deteriorating personal economic situation has a negative effect on political trust... 22
H3: A decline in political trust leads to populist support. ... 23
H4.1: Political trust has a mediating effect between recession and populism ... 24
H4.2: Political trust has a mediating effect between the personal economic situation and populism ... 26
Conclusion ... 28
Sugestions for further research ... 29
Bibliography ... 30
Appendix Populist parties per country ... 35
Appendix Regression results different trust indicators ... 36
3 INTRODUCTION
That the economy plays an important role in determine how people vote is without a doubt. Dassonneville & Lewis-Beck (2014a)have shown that the state of the economy has a statistically significant effect in determining the outcome of an election. Especially after the financial crisis of 2008 scholars have tried to link the state of the economy to another well discussed subject in the political science; the support for populism.
This research will contribute by investigating whether political trust has a mediating role between the economy and voting for a populist party. By doing so, I want to answer the question: Is there a mediating role
of political trust between the support for populism and economy in Europe? Following literature on economic
voting, I’ll make a distinction between egotropic and sociotropic level. In economic voting literature this is used to assess what influences the election outcomes; are these personal economic situations (egotropic) or are these national economic factors (sociotropic). I’ll assess the impact of both “recession” and “feeling about household income” on political trust. Then, I’ll assess whether political trust has a mediating effect between these economic indicators and the support for populist parties. I’ll look at the period of 2002 till 2016 using the ESS (2016) dataset combined with economic data from Giuliani & Massari (2017). I’ll theorize that the state of the economy can influence the trust people have in a negative way, thereby fostering the rise in populism in Europe. By no means do I claim that the state of the economy is the only determine factor in affecting trust in political institutions. There are many more factors that influence the trust people have in their political system. I’ll examine the cultural backlash theory as proposed by Inglehart (2016) in this research.
I’ll test whether the following general hypotheses hold:
(H4.1.2) In times of economic downturn, people lose trust that the mainstream parties can improve this situation. This decline in trust makes it more likely to vote in favour of populist parties;
(H4.2.2) Through the mechanism of losing trust in the political elite to improve his/her personal economic situation a person will be more likely to support populist parties.
I found that there is a statically significant relation between egotropic and sociotropic factors on political trust and an independent effect of egotropic, sociotopic and political trust on populist support. However, the effects of the egotropic model on populist support go against expectations. People who are feeling that it is very difficult to live from current household income having the lowest chance of voting for a populist party, compared to people who are living comfortably on household income. I do find a strong mediating effect of political trust between the egotropic model and populist support. The mediating effect of political trust between sociotropic economic factors and populist support is negligible. I’ll conclude that the relationship between economy, trust and populism is inconsistent at best, following the conclusion from Inglehart (2016). Political trust seems to matter most at the egotropic level, especially in the category for people who find that it is very difficult to live from current household income. Yet, there seems to be no linear relation between deteriorating personal income and populist support. The mediating effect of trust between
4 GDP growth and populist support is neglectable, however there seems to be a direct effect of GDP growth on populist support. This indicates that that although political trust isn’t the mediation factor, other mediating mechanism could be in place. For instance the changing attitude towards immigrant during low economic growth(R. F. Inglehart, Norris, & School, 2016). These findings indicate that there is a complicated interaction between economy, culture, trust and populist support.
The rest of this research will continue as follows: First, I’ll define what is meant with populism. Second, I’ll assess why populism is often linked to crisis. Third, political trust will be defined and the mediating effect of political trust between economy and populist support is described. The hypotheses will follow from this theoretical framework. Fourth, the research is placed within the culture versus economy debate in
explaining the support for populism. Lastly, the hypotheses will be tested, and the findings are compared to the literature.
POPULISM
Populism is a hot topic in the academic literature. It has given rise to thousands of books and articles (Mudde, 2004). This has led to several influential definitions of what political scientists consider as populism. None of the definitions has been as influential as the one from Cas Mudde (2004). He describes populism as a “a thin ideology that considers society ultimately separated into two homogeneous and antagonistic groups,
“the pure people” versus “the corrupt elite”, and which argues that politics should be an expression of the volonté Générale (general will) of the people”. This definition can be broken down in three parts:
The thinness of the ideology
The division of society into two groups; “the pure people” and “the corrupt elite”.
Politics, and democracy itself, should be an expression of the general will of the pure people.
The “thinness” of the ideology means that populism isn’t a ‘thick ideology’ like conservatism or socialism (Kriesi & S., 2015). Rather it is a ‘thin’ ideology that can be combined with other ideologies. The thinness of the ideology also makes it a difficult concept to isolate for analysis (Mudde, 2004). Scholars agree on at least one classification of populism: the party advocates in a way that society is divided against the lines of ‘the corrupt elite’ and ‘the pure people’ and that the general will of the people is neglected. Who the “corrupt elite and the pure people” are, can vary per context. Mudde & Kaltwasser (2012) have demonstrated that, the notion of who belongs to what group can vary depending on the country or continent that is being examined. In Europe we observe that populism is often related to some form of nationalism, defining the “pure people” along the line of ethnicity. This often leads to the confusion that populism is connected to ‘right wing nationalistic’ parties. However, this is a misconception on the notion of “the pure people”. While right wing populist parties see the pure people often as a nationalistic group of people defined by country borders or ethnicity, left wing populist parties regard “the pure people” as ‘the common man’ or the ‘little guy’ (Kriesi & S., 2015; Mudde &
5 Populist parties share the notion that the people are a homogenous group and have a general will.
According to their vision, the main problem with current democracies is that this general will of the people has been neglected by the “corrupt elite”. The corrupt elite is also considered to be a homogenous group.
Regarding Europe, among others, mainstream political parties; bankers; academics and Brussels belong to this group (Ziller & Schübel, 2015). As the notion of the corrupt elite is not limited to these groups, there is within Europe quite some variation. It is not the aim of this research to look at the variation of who is considered to be the corrupt elite across Europe. I will classify each party that claims to speak against á corrupt elite to be populist. A list of parties is presented in appendix 1. This is in line with the minimal definition of Mudde (2004) and used in several other researches (R. F. Inglehart, Norris, & School, 2016;Kriesi & S., 2015; Mudde, 2016).
Last, the notion that politics (or democracy itself) should be an expression of the general will of the people, is a characteristic of populist parties. Populist parties claim that democracy is failing because the general will of the people is being neglected by the ruling elite. If democracy is truly to function properly, the will of the people should be placed central in democracy1 (Kriesi & S., 2015;Abts & Rummens, 2007; Kaltwasser, 2014; Mudde & Kaltwasser, 2012). This leaves the question if populism is a threat to democracy. Though it isn’t the main scope of this research, the way it is presented in today’s media it is worth clearing my view on this matter: I agree with the conclusion from Mudde & Kaltwasser(2012) that populism is both a threat and a corrective measure to democracy depending on context in which it is presented2.
Though the definition used by Mudde (2004) is one of the most influential, the debate on the definition hasn’t been settled . Some have defined populism as a communication style (Gidron & Bonikowski, 2013; Rooduijn & Pauwels, 2011). Or as a political strategy ‘through which a personalistic leader seeks or exercises government power based on direct, unmediated, noninstitutionalized support from a large numbers of mostly unorganized followers’(Gidron & Bonikowski, 2013; Weyland, 2001). Both definitions are complementary to the minimal definition (Mudde, 2014).
Many scholars have argued that populism is inherent to some form of crisis. This can be a cultural crisis(R. F. Inglehart et al., 2016;Kessel, 2015; Van Hauwaert & Van Kessel, 2018;Malone, 2014), economic
crisis(Hernández & Krieis, 2016; Kriesi & S., 2015) and/or political crisis(Beasley-Murray, 2006; Moffitt, 2015, 2017b; Moffitt & Tormey, 2014). Whether populism is inherent to some form of crisis is examined in more detail in the next chapter.
1This means that if the general will of the people goes against the law/ constitution, the latter should be changed.
2 Not a very satisfactory answer. For the readers interested in this debate I would advise to read (Abts & Rummens, 2007; Kaltwasser, 2014; Mudde & Kaltwasser, 2012) to get a clear overview of this debate.
6 SUPPORT FOR POPULISM
To link crisis and populism together seems intuitive. As Kriesi (2015) puts it: “Given its essentially anti-elitist orientation, populism can be expected to thrive on popular dissatisfaction with the elites. Such dissatisfaction can have different origins, depending on the national context, but it is certainly expected to increase in crisis situations”. However, how (and if) a crisis influences populism remains debated.
Some have argued that populism needs a crisis to surface (Laclau, 2005; Stavrakakis ,2005). Laclau (2005) argues that ‘some degree of crisis ...is a necessary precondition for populism’. While Stavrakakis (2005) argues that “the emergence of new discourses and new identities is always related to the dislocation or crisis of previously hegemonic discursive orders . . . this is also the case with populist discourses”.
Others aren’t convinced that there is a causal link between crisis and populism. As Mudde (2007) notes
‘the key problem in this literature is the relationship between these variables and the overarching concept of crisis’– that is, these variables do not automatically equal crisis.
Moffitt (2015) argues that the causal relation between crisis and populism is debatable. Populism tends to foster crisis through rhetoric and mediatisation. One should thus be careful to argue that a crisis is needed to foster populism, as populism could also foster (a perceived) crisis.
I agree with Mudde (2007). There is no formal definition of crisis. Speaking about “a crisis” depends on your benchmark of “normal”. As summarized by Moffitt (2015) “… crisis being a contested phenomenon that lacks clear and distinct boundaries; the fact that crisis itself is a product of complex causality; and the aforementioned difference between crisis and the variables we associate with crisis. The second problem is that we cannot truly conceptualize a ‘neutral’ or ‘objective’ crisis that populism can be measured against.”
However, there does seem to be some significant correlation between variables we might associate with crisis – for example, economic instability, unemployment and political dissatisfaction. Though we should be careful in linking crisis to populism it is interesting to see how the variables normally associated with a crisis influence the support for populism in a country.
Especially interesting is the research done by Kriesi (2015). He assesses two different types of crises: political and economic crisis. For an economic crisis indicator, Kriesi takes the macro economic indicators unemployment, GDP growth and public debt. For political crisis he takes the indicators of electoral volatility and trust in parliament. Their time span is focussed on the financial crisis (2008) and to assess the impact of this crisis on populism. Kriesi (2015) hypothesizes that an economic crisis and/or a political crisis has a positive effect on the rise of populist support in a country. He finds a notable but uneven surge of populism in Europe.
As explained in the chapter on the support for populism, they should refrain from using a standardized notion of crisis for each country. This is one of the reasons why Kriesi (2015) finds scattered and mixed
evidence of a rise in populism in countries where there seems to have been a modest or ‘no crisis’. GDP growth remained above zero in most countries but there was still a drop in GDP growth compared to the historical
7 average. In explaining their support for populist parties, they should have looked at the change within a
country and not necessarily between countries. It is to be expected that people judge the performance of the economy and the political arena to historical figures, rather than using a set benchmark.
Next to that, I expect a different mechanism to be in place. While they distinguish between economic and political crisis, I think that this distinction is arbitrary and that macroeconomic indicators have a large impact on the trust people have in their government. Economic- and political crises should not be seen as separate but as a causal mechanism3. This has been demonstrated (Van Erkel & Van Der Meer, 2016;Kroknes, Jakobsen, & Grønning, 2015). Kriesi (2015) recognizes this problem to a certain extent, as he hypothesizes that if both an economic crisis and a political crisis occur populist parties are expected to do even better. The findings in his research support this hypothesis partially, but he doesn’t explain how this mechanism works.
The examination of evaluating crisis and populism together should have provided the reader with a word of caution. This is important to understand when examining the period of 2002 - 2016. The linking of crisis and populism was problematic for two reasons. First, because the definition of crisis itself is debatable and second because the variables associated with crisis can influence a mediating variable: political trust. In the rest of this research I’ll theorize how political trust is influenced by economy and link political trust to populism. Here I’ll combine literature on political trust and a partly modified theory of economic voting. Then, I’ll demonstrate how political trust can be linked to populism. This will be partly based on the research done by Van Erkel & Van Der Meer (2016) and Kroknes (2015).
POLITICAL TRUST
I support the definition of political trust as used by Weber & Saris (2016): “political trust as a belief that
those in authorities and power will not harm the citizens but look after their interest. In this sense, similar to the relationship of social trust with society, citizens have trust in political institutions when these operate according to the principles of justice and impartiality”.
Why is trust so important in politics? As stated by Kroknes (2015): “Trust links ordinary citizens to the institutions that are intended to represent them, and thereby enhances both the legitimacy and the
effectiveness of democratic government”. The concept of political trust can be described as multidimensional. Different terms are often used to describe the same phenomenon. The basic idea is that “trust is a—probably
the—main component of social capital. Social capital is a necessary condition of social integration, economic
efficiency, and democratic stability(Newton, 2001). Trust can be divided into three forms: thick, interpersonal, and systemic or institutional trust (Newton, 2001). In this research I am concerned with institutional trust as we want to assess the effect of the economy on how people trust the political institutions4 and the effect
institutional trust has on populism. The casual relationship between trust and economic efficiency is still
3 I’ll extend this argument in the next chapter
4The trust in political institutions will be named as political trust throughout this research. Concentrating not on the trust people have in
8 debated(Van Erkel & Van Der Meer, 2016). Trust can foster economic efficiency and economic efficiency can foster trust. In this research I’m only concerned with the relationship between economy and trust5.
Economy influences political trust. People trust governments more if they have shown a capacity to generate economic growth, create jobs, provide access to social services and perform in a transparent
manner(Kroknes et al., 2015). When there is doubt if the authorities can manage the national economy, it may create mistrust. High political trust signals that the institutions function properly and effectively and helps to keep democracy alive (Ringdal, Kristen ; Listhaug, 2007). In the next chapter, the link between economy and trust will be assessed in greater detail.
ECONOMY AND POLITICAL TRUST
The effect of economy on political trust has been extensively theorized (Van Erkel & Van Der Meer, 2016;Funke, Schularick, & Trebesch, 2016; Funke & Trebesch, 2017; Kroknes et al., 2015; Mishler & Rose, 2001; R. F. Inglehart et al., 2016; Kinder & Kiewiet, 1979). Some scholars find that there is a significant effect of economy on political trust (Van Erkel & Van Der Meer, 2016; Funke et al., 2016; Funke & Trebesch, 2017; Kroknes et al., 2015). Others find the relationship to be insignificant and emphasize more on the cultural factors (R. F. Inglehart et al., 2016; Mishler & Rose, 2001). In this chapter I’ll theorize that the effect of economy on political trust can happen on two levels. Borrowing from economic voting theory6, I’ll name them egotropic (for the individual level) and sociotropic (for the macro level).
ECONOMIC VOTING THEO RY AND POLITICAL TRUST
The basic mechanism7 in economic voting is that “citizens reward the government with their vote under a good economy and punish it by voting for the opposition under a bad economy. They thus look back at the economic performance before an election. This is called retrospective voting. This idea is old, and
intuitive”(Lewis-beck & Lobo, 2017). There is discussion among economic voting theorist whether the national economy matters, or the personal economic situation is determining the outcome of elections.
The economy can have a sociotropic effect on the voting behaviour in which people take into account the general state of the economy, indicated by the macroeconomic indicators like unemployment, GDP growth or inflation. It can also have an egotropic effect, in which people care less about the general state of the economy but more about their own economic situation like household income or employment status(Lewis-Beck & Paldam, 2000). One would expect that egotropic and sociotropic show almost identical patterns in explaining voting behaviour. Yet, multiple studies have shown that there is an ecological fallacy: the observed
5As the estimation results show, the coefficient of political trust hardly changes when adding economic variables indicating that this is also
the stronger causal relationship.
6 . I’m aware that economic voting theory normally limits itself to the punish reward system of incumbent parties, but I think it can be
extended to assess the effect on an entire mainstream political system through the mechanism of political trust.
7This is just a part of the economic voting theorem. The theory is much broader and extends to a prospective vs retrospective
debate(Lewis-Beck & Stegmaier, 2009). For my paper I’ll use the basic principle of retrospective voting to explain how the economy could matter to explain an increase/decrease of political trust.
9 links between the macro-economy and national elections were actually spurious at the level of the individual (Dassonneville & Lewis-Beck, 2014b; Lewis-beck & Lobo, 2017). Both the marco- and individual economic levels have an independent effect on the election outcome.
I want to use the same punishment-reward system, albeit not explained directly in number of votes, but its effect on political trust. By doing so, I stretch the theory to go beyond focussing at the punishment of incumbent parties and extend it to punishment of the entire mainstream political spectrum. The difference between the individual and macro level is interesting to investigate outside the spectrum of election outcomes. Do people lose political trust because the overall economy is performing bad and do they lose political trust if their personal economic situation is deteriorating?
The general mechanism would be that in a bad economic situation (either at the country level or personal level) people lose trust in the mainstream political parties in dealing with this situation. This decline in political trust then leads to the search for parties that are outside the mainstream political spectrum anti-establishment /anti-elite parties. This could explain why we observe indicators like negative GDP growth and drop in household income to be correlated with support for populism . Next, I’ll examine these indicators in more detail by explaining the sociotropic and egotropic models.
SOCIOTROPIC
As described by Van Erkel & Van Der Meer (2016) : “The regulation of the macroeconomy is one of the most salient issues for citizens, one that citizens hold their government strongly responsible for”. Hence, economic performance is likely to have direct repercussions for the level of trust in government. If economic performance deteriorates, this will drive down confidence that governments are sufficiently competent to solve problems. It turns out that not only incumbent parties decline in trust. Also, opposition parties show a decline in trust when the economy performs badly. People thus lose faith in the entire current political system at the point when the economy is deteriorating. This would be in line with the findings of Giuliani & Massari (2017) that “new parties” are likely to perform better if the economy is performing bad. These new parties can claim that they are unrelated to the current situation. Arguably, voters choose for populists because they hold protest attitudes which coincide with the discursive elitist critique populist parties offer, with the anti-elitist message being the attacks on the ‘established political order’ (Schumacher & Rooduijn, 2013). If they blame the ‘established political order’ for the bad economic situation, in this case measured as the trust they have in parliament and politicians, they are more likely to vote for a populist party. The overall hypothesis for the sociotropic model is:
H4.1.2: In times of economic downturn, people lose trust that the mainstream parties can improve this situation. This decline in trust makes it more likely to vote in favour of populist parties.
This hypothesis can be broken down in three preliminary hypothesis that need to be tested before we can assess whether political trust has a mediating effect:
10
H1: Economic downturn decreases the level of political trust. H3: A decline in political trust leads to a rise in populist support. H4.1.1: An economic downturn leads to a rise in populist support.
Research by Kroknes (2015) has shown that, political trust is affected by times in which unemployment is high and GDP growth is low. Their study contributes to the understanding of the relationship between economic performance and political trust, as well as how citizen’s trust is affected by an economic crisis. A severe economic crisis must be taken seriously by policy makers, as it can contribute to reducing citizen’s confidence in and support of political institution, especially during the financial crisis of 2008. Yet, their results also show that there are some exceptional cases in which political trust doesn’t decline with deteriorating economic performance. We do see an increase in populist support in these countries. These countries include: Netherlands, Sweden, Switzerland, Estonia, Czech Republic8. How can this be the case? The vast array of possible indicators complicates the economy–trust story. According to some analyses, trust declined in advanced industrial democracies at the same time as some objective macro measures of economic
performance (GDP, inflation, and unemployment, for example) were improving or visa versa. That objective macro trends do not, at first glance, appear to track with political trust suggests the two are not causally related(Wroe, 2016).
However, both the research of Van Erkel & Van Der Meer (2016) & Kroknes (2015) show that there is a relation. It is therefore worthwhile to test if macro-economic performance has a significant effect on the level of political trust, controlled for country specifics. Besides the debate whether macro-economic performance evaluation has a significant effect there is a conventional trend that personal economic experiences and evaluations do not play much of a role (Wroe, 2016). However, in line with the reasoning of Wroe (2016), “the
conventional wisdom is wrong”. This brings me to the egotropic effect economy has on political trust. This
might be the missing link that can explain why in general we can observe an increase / decrease in macro-economic performance which goes against the trend in political trust.
EGOTROPIC
As argued by Algan, Guriev, Papaioannou, & Passari,(2017) The differential impact of crisis on economy is key to understand the rise of anti-establishment, often populist parties, and the accompanying sizable drop of trust towards political parties, the European Union, and the domestic legal system. They argue that the economic growth of the past 20 years has not been distributed equally. Ergo, it can be captured in the “losers of globalization theorem”. Algan (2017) argues that: “Globalization in general and the EU project in particular, have been successful in promoting growth but have not done as well in terms of assuring that the gains from this growth have been shared broadly. Large parts of the society feel left behind and have risen against the establishment, national and European institutions”. It can be the case that the general
11 economic figures are good by country standards but there is a group that is still experiencing wage
drops(Kessel, 2015).
When looking at the literature “the losers of globalization” are typically described as “the blue-collar workers” or “the lower educated angry white man” (especially popular in media9). However, when scholars try to find a link between social economic status from that group and the support for populist parties the results are inconsistent at best(R. F. Inglehart et al., 2016). Perhaps the mechanism of losing political trust can clarify the inconsistency. Wroe (2016) has shown that the effect of personal economic insecurity has a significant effect on political trust in the United States. He argues that the increase of economic insecurity has led to the decrease in political trust. Political trust’s importance is also evident at the egotropic level. “A lack of trust, increases voters’ support for non-incumbent and third party candidates and elite-challenging initiatives”(Wroe, 2016). Besides the sociotropic effect, the egotropic effect should be tested as having a separate effect on political trust. This is in line with recent research done by Gangl & Giustozzi (2018) who concluded that; both macro and micro economic situations work though a different mechanism and should be estimated separately. If they indeed influence political trust through a separate mechanism we would observe an independent effect of both macro-economic performance (sociotropic) and individual economic situations (egotropic).
An overall hypothesis for the egotropic level can be formulated:
H4.2.2: Through the mechanism of losing trust in the political elite to improve his/her personal economic situation a person will be more likely to turn to parties that are populist parties.
This hypothesis can be separated in three testable hypotheses that need to be true before the overall hypothesis can be tested:
H2: A deteriorating personal economic situation has a negative effect on political trust H3: A decline in political trust leads to a rise in populist support.
H4.2.1 deteriorating personal economic situation leads to a rise in populist support.
H3 is the same in the sociotropic model & egotropic model, as people can lose trust due to sociotropic factors or egotropic factors. Either way, they lose trust in the political institutions which are the same in both models. It is worth explaining why I expect trust to influence support for populism.
POLITICAL TRUST AND POPULISM
In Guiso's (2017) Populism: Supply and Demand he makes a compelling case on the influence of political trust and the support for populist parties. He distinguishes a demand and supply side for a surge of
12 populism: “The demand side involves people’s fears and enthusiasms, to which politicians pander. The supply side lies in the politicians’ claim to be on the side of the people and against the elite”.
He argues that: “Previous crises, which basically resulted in the failure of only one side simply generated political cycles and did not leave space for the emergence of populist platforms, which requires substantial disappointment and falling electoral turnout across the political spectrum”. Populist parties are more likely to foster when countries are faced with a crisis of economic security that incumbent parties (whether left-wing or right-wing) find hard to address. Disappointing voters will lose faith in them (Guiso, 2017).
As people lose faith in either left-wing or right-wing parties this gives way for “new parties” to arise. This was also observed by Giuliani & Massari (2017). This is further supported by Algan (2017), who shows that in elections after 2008, the sharpest decline of trust in institutions and establishment politics was in regions with the highest increase in unemployment. Otten(2017) reaches similar results showing that in the aftermath of the financial crisis, mistrust of European institutions and national government to deal with the negative consequences led to the rise of anti-establishment parties. He describes it as: “if one were to pick one word to describe prevailing attitudes toward the political establishment in Europe – to describe attitudes toward national parties and politicians and the institutions of the European Union – that word would be dissatisfaction. If one were to pick two words, they would be widespread dissatisfaction”(Otten, 2017).
So why is trust in politics important when explaining populism? The general mechanism is that when people lose faith in the government they will look at the opposition parties. This is especially true during times of economic hardship. Dassonneville & Lewis-Beck (2014a) have shown that during economic hard times the incumbent parties are likely to lose votes and opposition parties are likely to gain votes. Yet, opposition parties can’t capture the entire amount of votes as some people move to alternative options of not voting or new parties, as demonstrated by Giuliani & Massari (2017). This can give rise to new populist parties and bigger, already established populist parties. As the notion of being anti-elite becomes more attractive towards voters. This leads to the following hypothesis:
H3: A decline in political trust leads to a rise in populist support.
The difference made in economic voting theory between sociotropic and egotropic effects, and how these factors influence political trust, can very well be the key in explaining the spurious outcomes we observe in the levels of political trust in some countries10 in correlation with the macro economic performance of those countries. Yet, it is unrealistic to expect that political trust is only influenced by sociotropic and egotropic economic indicators. Next to that, it is unrealistic to expect that a decline in political trust is the only
explanatory variable for populist support. It is important to view this thesis in context with this economy versus culture debate concerning populist support. In the next part I’ll describe the theory of Inglehart (2016) how
13 culture can affect the support for populist parties. Also, I provide the reader with a word of caution when looking at cultural factors in determining an increase in support for populism.
CULTURE AND POPULISM
Research of Van Erkel & Van Der Meer (2016) and Kroknes (2015) show that there is a significant relation between macroeconomic indicators and the level of political trust. Research of Wroe (2016) show that
individual economic situations matter as well. Yet, not everything can be seen from an economic perspective. Scholars like Inglehart (2016) find that the cultural factors provides the most reasonable explanation for the surge of populism in Europe. They compare the theory of economic insecurity and the cultural backlash theory. Economic insecurity theory is defined as “rising economic insecurity and social deprivation among the left-behinds has fuelled popular resentment of the political classes.” The left-behinds are the low-waged unskilled workers, the long-term unemployed, household’s dependent on shrinking social benefits, residents of public housing, single-parent families, and poorer white populations living in inner-city areas with
concentrations of immigrants.
The cultural backlash thesis suggests that the surge in votes for populist parties cannot be explained as a purely economic phenomenon that leads to a decline in political trust but is in large part as a reaction against progressive cultural change. This argument builds on the ‘silent revolution’ theory of value change. The high levels of insecurity experienced by the people of Western societies brought a shift toward post-materialist values, such as cosmopolitanism and multiculturalism. This generated a rising support for left-libertarian parties and the accompanied tolerated view on immigration and the European union. However, this rising support in turn led to a counter reaction of certain parts of society who feel uncomfortable with this change.(R. Inglehart & Norris, 2016).
The distinction between culture and economics factors might not be so evident. Interactive processes may link these factors. For example, changes in the workforce and social trends in globalized markets increase economic insecurity, this might stimulate a counterreaction among traditionalists towards cultural shifts. It is not an either/or question, but rather a mechanism with interactive effects (R. Inglehart & Norris, 2016). Overall, Inglehart (2016) concludes that cultural values, combined with several social and demographic factors, provide the most consistent explanation for support for populist parties. Arguing that “the evidence in this study suggests that it would be a mistake to attribute the rise of populism directly to economic inequality alone” (R. Inglehart & Norris, 2016).It is important to take the cultural backlash theory into account. First, I’ll argue that in large part I agree with Inglehart that both economy and culture are important. Second, I’ll provide a word of caution in linking the cultural backlash theory to populism.
As I’m interested in the relationship between economy, trust and populism, I will only look at the effect of economic indicators on the level of political trust. By doing so, I do not claim that culture isn’t an important factor in determining political trust. It can very well be the case that changes in perceived culture and the perceived failure to adequately deal with these changes causes distrust towards the political system (as argued
14 by Inglehart). It would also be a mistake to determine causality only based on economic indicators. Although economic indicators are important, the interplay between economics and culture might be the key in determining how political trust changes over time and how this influences populism in Europe.
This research aims to provide a suggestion that economic indicators affect political trust which then affects a rise in populism. I do not claim that economic indicators like household income and GDP growth are
the only cause of distrust and therefore populism. It would be very interesting to see a model which accounts
for more factors that affect political trust. But due to scope and time limitations, I’ll only test the economic side of the debate. However, not mentioning cultural effects would be a mistake, as the interplay between
economy and culture could be the key in fully understanding the support for populism. I’ll briefly describe the argument that culture is a determining factor for populism. I’ll suggest that the cultural argument should be used with caution, as it can mistakenly claim that right-wing characteristics like Eurosceptic and
anti-immigration are part of the populist ideology. When examining the results, the cultural argumentation will be considered.
Across Europe, we see that most right-wing populist parties hold similar views on issues related to the cultural backlash theorem; governments should decrease immigration as much as possible (an anti-immigration stance) and view the EU as either a money depleting project or as an infringement of their national sovereignty (Arzheimer, 2016). Many scholars also take these two indicators as a measurement of support for the cultural backlash theorem (Funke et al., 2016). One can easily mistake this as a characteristic of populism, especially considering that a lot of research has been about the rise of right-wing populism (Mudde, 2014; Moffitt, 2017a; Muis & Immerzeel, 2017; Lucassen & Lubbers, 2012). Scholars have more and more focussed their attention also on left wing populism and started to look at the populist ideology on both sides of the left-right spectrum (Schumacher & Rooduijn, 2013; Abts & Rummens, 2007; Gidron & Bonikowski, 2013; Kessel, 2015; Kriesi & S., 2015; Rooduijn & Pauwels, 2011; Rooduijn, Van der Brug, & de Lange, 2016). The early focus on right wing populism might have left the reader with the idea that anti-immigration and anti-EU are characteristics of populism as such and related to the cultural backlash argument. This would be a mistake.
Anti-immigration and Anti-EU aren’t a characteristic of populism. It might very well be a characteristic of right wing parties, but it is more related to the right of the political spectrum in which the party places itself (so to say its “thick ideology”) then it has to do with populism as such. Populism should be explained by a view of distrust towards “the ruling elite”. A thought example might be best to explain the difference: it can very well be the case that someone has strong ideological beliefs against the EU and Immigration. Therefore, he votes for a party that expresses those concerns (a right-wing party). Another person has the same strong ideological beliefs, but he feels that the current government has done so little to counter the EU or Immigration that he lost trust in the party system. When he decides to vote, he votes for a party that is outside the system and expresses his ideological beliefs (a right-wing populist party). The mechanism at work is a subtle difference, yet very important to understand if one wants to explain populism. Only the second example shows a voter who holds an anti-establishment believe, something which is at the heart of populist ideology. To explain why
15 people vote for a populist party one should take into account the anti-elite sentiment and not necessary the attitude towards the EU or immigration. Considering the factors immigration and Euroscepticism as
explanation for populism is therefore not entirely true. It could very well be the explanation for the rise of
rightwing parties. Populism, however, should be explained by looking at the anti-elite sentiment. I
acknowledge that it is difficult to assess whether someone votes for a party because of the “thick ideology”, or because of the “thin ideology”, losing trust in the ruling elite.
This chapter placed the hypotheses in the larger context of the economy versus culture debate. To find whether political trust has a mediating effect between economy and populist support two models will be tested in the next chapter; “the sociotropic model” and “the egotropic model”.
16 EXPLANATION OF THE DATA
As mentioned in the theoretical framework the following hypotheses can be tested:
SOCIOTROPIC MODEL
H1: Economic downturn decreases the level of political trust. H3: A decline in political trust leads to populist support.
H4.1.1: An economic downturn leads to a rise in populist support.
H4.1.2: In times of economic downturn, people lose trust that the mainstream parties can improve this situation. This decline in trust makes it more likely to vote in favour of populist parties.
EGOTROPIC MODEL
H2: A deteriorating personal economic situation will have a negative effect on political trust. H3: A decline in political trust leads to populist support.
H4.2.1 deteriorating personal economic situation leads to a rise in populist support.
H4.2.2: Through the mechanism of losing trust in the political elite to improve his/her personal economic situation a person will be more likely to turn to parties that are populist parties.
To examine the relationship between economy, trust and populism I used the ESS (2016) dataset and added a recession indicator based on 12 months preceding the election. This was largely based on the data set used by Giuliani & Massari (2017) but supplemented where necessary with data from the CEIC (2018) where data on GDP growth was missing. The data set contains N = 273,952 observations of 22 countries, examining 8 rounds of ESS data. I coded all the populist parties partly based on the research of Kessel (2015) & Kriesi (2015). Where necessary I checked additional literature to check which parties where considered populist. This was the case in the Baltic states (Oliver, 2016) and in Bulgaria (Cholova, 2010).
The main dependent variable is populist_vote. This is a dummy variable and considers whether someone has voted for a populist party (1 if yes 0 if no). The independent variables are the impact of GDP growth, measured as a recession indicator. Trust in politicians and trust in parliament are combined as a measure for political
trust. Lastly, Feeling about household income as a measure for the personal economic situation of the
17 EXPLANATION OF THE VARIABLES
Populist_vote is a dummy variable which categorizes every vote of being either for a populist party (=1) or
another party (=0). I didn’t include a variable for right and left wing populist parties as is seen in a research done by (Funke et al., 2016; Kitschelt, 2007; March & Rommerskirchen, 2015; Mudde, 2014; Muis & Immerzeel, 2017) because I’m interested in the ´thin ideological’ part of populism11. The ‘thin ideological part’ of populism doesn’t differ depending on the left-right spectrum.
Recessionindicator is composed of the average GDP growth 12 months before the election date. This means
that if an election was at the first of June 2003, the average GDP growth of the first of June 2002 till the first of June 2003 is used. If there was an average growth the recession indicator would be negative. This way of measuring GDP growth was used by Giuliani & Massari (2017). I used the data set of Giuliani & Massari (2017) to refer to the recession indicators where possible. If the election wasn’t included in their research (elections before 2003 and after 2015), I supplemented the data using the CEIC (2018) database to compose the average GDP growth of the 12 months preceding the election. This might create a small difference between the numbers used in their research and the added numbers as they composed their recession indicator based on a different dataset. However, the error would be minor and not really influencing the results as GDP growth often differs only in very small margins between intuitions.
Political trust is composed of two variables; trust in politicians and trust in parliament. This are categorical
variables, ranging on 0-10 scale. With 0 indicating no trust in, politicians and parliament and 10 indicating complete trust in politicians and parliament. To prevent the scattering of the results I dropped all the
observations which answered don’t know (88) or failed to answer the question (99). I’ll treat this variable as a linear variable. This might influence the results somewhat as the difference in loss in trust between 1 and 0 might be bigger than between 2 and 1.
I created a new variable as trust in parliament and trust in politicians are highly correlated (>0.8941
correlation). Political trust is the sum of trust in parliament and trust in politicians. Trust in political party hasn’t been included as it would generate 48.000 missing values due to only a limited number of ESS data is available on this subject. Therefore, it would scatter the results too much. I expect that leaving out trust in political parties would not make a large difference, as they are all strongly correlated. The method of adding these three variables together was also done by (Kroknes et al., 2015). But to be sure, I ran all the variables in separate regression models for hypothesis 1 and 2. The results are included in the appendix. The variables show similar direction of the coefficients.
Feeling about household income is a categorical variable ranging from 1 to 4. This is a subjective variable. I follow argue that the subjective economic evaluations are more important individual-level determinants of political trust than are objective criteria(Wroe, 2016).
18
CATEGORY DESCRIPTION
1 Living comfortably of present household income
2 Coping on present household income
3 Difficult to live of present household income
4 Very difficult to live of present household income
This is an ordinal variable but in line with the reasoning of Long and Freese (2016)12 I performed a Walt- test to see whether the variables can be treated as a continues variable or as an ordinal variable. The results differ per model used. When testing the effect of feeling about household income on political trust the Wald test show that the test is insignificant at 0.1169, implying it can be used as a linear variable without losing too much of its explanatory value. Yet, when testing the effect of feeling about household income directly on populism the test shows a significant result 0.0000. Therefore, it should be treated as an ordinal variable. I included each
category (1-4) as a separate variable in the analysis. Category 1 will serve as the reference category.
EXPLANATION OF THE M ODEL
To test the 4 hypotheses, I’ll use a multi-level model. Depending on the depended variable this will either be a multi-level random intercept model (for Hypotheses 1 &2) or a log multi-level random intercept model (for hypotheses 3 & 4). To test H4.1.2 & H4.2.2 I check the mediation effect of political trust on recession and feeling about household income. I do so by looking at the change in coefficient by adding the expected mediator to the model. If the coefficient and significance change, there is a significant mediating effect. This follows the approach suggested by Baron & Kenny (1986)“A mediation analysis is comprised of three sets of regression: X → Y, X → M, and X + M → Y.”
A MULTI-LEVEL RANDOM INTERCEPT MODEL
Multi-level models allow simultaneous estimation of both level 1 and level two characteristics while accounting for any clustering at level 2 that might be occurring. This allows to consider certain country characteristics. This is especially helpful when considering the impact of GDP growth on populist support, as everyone is exposed to the same economic situation in a country. Yet, the consequences of the macro-economic situation don’t have to be distributed evenly across every group (the sociotropic vs egotropic debate). There is still some clustering to be expected when assessing the individual level. Someone in Greece
12The advantage of this approach is that interpretation is simpler, but to take advantage of this simplicity you must make the strong
assumption that successive categories of the ordinal independent variable are equally spaced. For example, it implies that an increase from no publications by the mentor to a few publications involves an increase of the same amount of productivity as an increase from a few to some, from some to many, and from many to lots of publications. Accordingly, before treating an ordinal independent variable as if it were interval, you should test whether this leads to a loss of formation about the association between the independent and dependent variable.
19 will be more strongly related to another Greek then for instance to someone in Sweden. It is useful to consider multi-level modelling when looking at the individual economic situation.
CONTROLLING FOR COUNTRY FIXED EFFECTS
I.countryname is included in the regression analysis to account for country specific conditions. By including this
variable I’m interested in the variance within countries. This goes against research done by for instance (Kriesi & S., 2015) which compares GDP growth against an set benchmark (e.g. average of the EU) to conclude whether a country is performing good or bad. It makes more sense to look against the historical average within a country as this is more likely to be considered when assessing the performance of the political arena. This connects well to retrospective voting mechanism described in the literature review on economic voting theorem. I.countryname assumes that the conditions within a country have been stable enough from 2002 till 2016 to be kept constant except for the other variables included in the model (recession, political trust and
feeling about household income). This is a (very) strong assumption and can be heavily criticized. It is to be
expected that not all other conditions besides the three variables have been constant in affecting whether or not someone votes for a populist party. (Bovens & Wille, 2010; R. F. Inglehart et al., 2016; Van Hauwaert & Van Kessel, 2018) and many more scholars have shown that other factors have a significant effect on populist votes as well. Yet, by assuming these to be constant I hope to find an indication that the effect of recession indicator,
political trust and feeling about household income on populist support are significant.
I prefer a random intercept multi-level model over, for instance, a random slope model because I expect the effects of economy to be (roughly) the same is each country. Especially when controlled for all other country characteristics (as is done by i.countryname). This is however an assumption that can be relaxed. I’ll not go into dept in this research on this, but it could be expected that economy has a different impact in Bulgaria then it has in the United Kingdom. For this research I’ll assume that the relation is (roughly) similar.
GROUPING OF THE DATA
The data will be grouped based on country and ESS round or election date. There is however an error in the model that limits the interpretation of the results. As trust in political parties is measured at the time of the ESS round (Tess) and the recession indicator is measured before the actual election data (Telection). The results will therefore be an underestimation of the true results as there is the lagged effect of the time passed. However, the results are still relevant as trust in political parties does fluctuate across time but not so much that someone who first would answer to have absolutely no confidence in political parties during elections (0) and would answer to have complete political trust (20) during the ESS round. This is also demonstrated by the results in research by (Kroknes et al., 2015) who show the largest change in political trust is 2.6 points on a scale of 30 in Hungary. Although this is considerable and does influence the results, I consider it to be within margin. However, the results should be viewed with caution(!)
To check if the different time frames scatter the results I estimated a model grouped on the election date and country (Telection), and one grouped on the ESS date and country(Tess). Though the coefficients changed a bit in
20 both models it didn’t matter for the direction of the coefficients, nor the significance level except when
estimating the effect of recession on populist votes13.
21 REGRESSION RESULTS
The first two models will be estimated with a random intercept multi-level model.
In formula the three estimations look like this:
(1): Politicaltrustij = β0 + β1recessionindicator j + β2 i.countrynamej + Uoj + eij
(2): Politicaltrustij = β0 + β1householdord4 + β2householdord3 + β3householdord2 + β4householdord1 + β5
i.countrynamej + Uoj + eij
Besides the two hypotheses tested I’ll also assess whether recession and personal economic situation has a independent effect on political trust.
(3): Politicaltrustij = β0 + + β1householdord4 + β2householdord3 + β3householdord2 + β4householdord1 +
β5recessionindicator j + β6 i.countrynamej + Uoj + eij
(1) (2) (3) Political trust recession -0.0586* -0.0515* (0.0236) (0.0224) Household -0.969*** -0.969*** Coping (0.0199) (0.0199) Household -1.815*** -1.815*** Difficult (0.0256) (0.0256) Household -2.642*** -2.642*** Very difficult (0.0359) (0.0359) constant 8.268*** 9.151*** 9.076*** (0.357) (0.344) (0.339) N 273924 268225 268225
Standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001
Graph 1: Margins plot effect recession on political trust
Table 2 Regression output effect of household income and recession on political trust
22 H1: ECONOMIC DOWNTURN DECREASES THE LEVE L OF POLITICAL TRUST.
A multi-level random intercept regression model was used to predict political trust based on the recession controlled for country fixed effects. If recession increases by 1-point political trust decreases by 0.0586462. With a 95 % confidence interval, the margins plot looks like graph 1. Based on these estimations we find evidence that there is a significant effect of recession on political trust. The results indicate that there is an effect on the sociotropic level. Political trust is declining depending on the level of recession there is in a country. This is in line with the findings of Kroknes (2015) , Van Erkel & Van Der Meer (2016) that GDP growth has a significant effect on political trust. The estimations of the first model indicate that hypothesis 1 can be confirmed, we find a significant effect of recession on political trust.
H2: A DETERIORATING PERSONAL ECONOMIC SITUATION HAS A NEGATIVE E FFECT ON POLITICAL TRUST.
I estimated the 4-cathegories of satisfaction with household income separately, using the category living comfortably on household income (1) as a reference category. As someone moves from living
comfortably on current household income (1), to coping with current household income (2) he loses -.9686636 points political trust compared to the reference category. When comparing category 1 to difficult living on current household income (3) a person loses 1.815095 points of political trust. And when comparing very difficult living on household income (4) to category 1 a person loses 2.642036 points of political trust. These findings hint on a strong significant effect between a deteriorating personal income situation and losing political trust. I described the outcomes with a margins plot in graph 2. To create the margins plot I treated
feeling with household income as a continuous variable. The Wald-test showed Prob >Chi = 0.1169 indicating
that the feeling with household income could be treated as a continuous variable in this model. I ran that separate regression and found that when someone increase with 1-point dissatisfaction with household income, political trust decreases with 0.8897936 (P< 0.001).This finding is in line with the research done by Wroe (2016) that personal economic situations influence the level of political trust. Confirming hypotheses 2.
To test if recession and personal economic situation have indeed a separate effect on political trust I included both in the final estimation model (3). As the results show the adding of both independent variables hardly changes the coefficient or the significance. For this estimation it can thus be concluded that they have an independent effect on political trust. As is in line with the expectation flowing from the debate on the significance of including either sociotrpic or egotropic measurements. Both sociotropic and egotropic economic factors have a separate, significant effect political trust. Whether this leads to supporting populist parties is examined after testing if a decline in political trust has a significant effect on populist support.
23 H3: A DECLINE IN POLITICAL TRUST LEADS T O POPULIST SUPPORT.
In formula the estimation looks like:
(3) Populist_voteij = = β0 + β1 Politicaltrustij + β2 i.countrynamej + Uoj + eij
When running this regression model Spain is dropped from the observations because before 2016 there were no populist parties with more than 1% of the votes. After 2016, Podemos entered the political arena. Yet, as I’ll only examine the elections included in the ESS rounds 2002 till 2016. There is no prediction on whether political trust affects the change of voting for a populist party in Spain. This explains the drop off observations
compared to the first three estimations.
Controlled for country fixed effects the regression model shows that if political trust increases the chance of voting for a populist party decrease with 0.0277691(P<0.001, N =237710). To illustrate this, I constructed a margins plot (graph 3). The margins plot shows that if political trust increases the chance of voting for a populist party go down. This is in line with the findings of Guiso (2017).
The fourth hypotheses (H4.1 & H4.2) will be tested with a logistical random intercept multi-level model. Mediating can be tested by looking at the change in coefficient of the independent variable when another independent variable is added to the equation. For political trust to have a mediating effect between economy and the chance of voting for a populist party we expect that the coefficient of both recession and feeling about household income should change when adding the independent variable of political trust to the model. The estimated models are:
(4.1.1) Populist_voteij = = β0 + β1 recessionindicator j + β2 i.countrynamej + Uoj + eij
(4.1.2) Populist_voteij = = β0 + β1 recessionindicator j + β2 Politicaltrustij + β3 i.countrynamej + Uoj + eij
(4.2.1) Populist_voteij = = β0 + β1 householdord4 + β2householdord3 + β3householdord2 + β4householdord1+ β5
i.countrynamej + Uoj + eij
(4.2.2) Populist_voteij = = β0 + β1 householdord4 + β2householdord3 + β3householdord2 + β4householdord1 + β5
Politicaltrustij + β6 i.countrynamej + Uoj + eij
24 Populist Vote (4.1.1) (4.1.2) (4.2.1) (4.2.2) (4.1.1) (4.1.2) (4.2.1) (4.2.2) recession 0.147** 0.145** 0.0571 0.0553 (0.0501) (0.0503) (0.0561) (0.0561) Political -0.0275*** -0.0256*** -0.0278*** -0.0258*** trust (0.00191) (0.00194) (0.00192) (0.00196) Household 0.207*** 0.183*** 0.205*** 0.180*** Coping (0.0233) (0.0234) (0.0234) (0.0234) Household 0.294*** 0.249*** 0.289*** 0.244*** Difficult (0.0269) (0.0271) (0.0271) (0.0273) Household 0.157*** 0.0916** 0.160*** 0.0950** Very difficult (0.0351) (0.0354) (0.0355) (0.0358) N 237710 237710 232124 232124 237710 237710 232124 232124
Standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001
H4.1: POLITICAL TRUST HAS A MEDIATING EFFECT BETWEEN RECESSION AND POPULISM For testing hypothesis 4.1.1 and 4.1.2 I estimated two different models, one grouped on the election date and country, the other grouped on ESS round and country.
Grouping on election date means that the respondents are grouped based on when the last election was. If 2002 and 2004 both say something about the party voted for in 2001 they are grouped together. This means that the number of observations increases per group (from average 1949 to 3231) and the number of groups go down (from 126 to 85). Interesting to see is that when estimating the effect of recession on populist vote this seems to matter quite substantially. The estimation based on ESS round, country is not significant while grouped on election date there is a significant effect, but adding the variable political trust seems to have little impact. Three implications are made from testing the two models:
First, how can it be the case that grouping on different categories changes the significance of recession on populist voting? This is a very odd finding and could well be an error in the model. As ESS round country & recession indicator should not differ much from a model estimated based on the question what did you vote in
Model estimated grouped on ESS round Model estimated grouped on election date
25 the last election? Both examine whether recession had an influence on the voting behaviour on the same election. The only difference is that the number of observations per group goes up and the number of groups go down. I found it a remarkable finding and it has probably something to do with how the model is specified. Yet, for the interpretation of the mediating effect I’ll focus on the model grouped on when the election date.
Second, testing if recession has a significant effect on populist support I find recession to have a significant effect of .1472604(P<0.05). Meaning that if recession increases by 1-point, people are more likely to support a populist party with roughly 14.7 %. This is partly in line with the findings of
Hernández & Krieis, (2016) & Kriesi (2015) that sociotropic indicators matter when considering voting for populist parties. This is demonstrated in the marginsplot (graph 4).
Interesting is that when the recession indicator approaches a positive number, the standard deviation goes up quite a lot. Indicating that the support for populism rises as GDP growth drops, but as soon as GDP growth turns negative the effect becomes ambiguous. This is also found by Lewis-beck & Lobo (2017). In their research they are concerned with the election outcomes in economic extraordinary times. The find that different countries react in different manners to bad economic situations. With southern countries punishing governments far more then northern countries in times of economic hardship. The link could be made that countries who punish their government more fiercely in times of crisis, are more receptible for parties outside the mainstream political spectrum. This could be an explanation why I observe such a large standard deviation when GDP growth drops below zero (recession indicator becomes positive).
Third, trust seems to have no mediating effect of political trust between recession and populist votes. The coefficient of recession in 4.1.1 = .1472604 (P<0.01) when adding the variable of political trust, it changes to .1449043(P<0.01).
How is it that trust has only a small mediating effect? One explanation might be that the time frame is incorrectly specified. As mentioned at the start of this chapter, there is an error in this research. People are asked what they voted in the last election (telection) and then how much they trust politicians, parliament and parties. When people answer this question, it can be expected that they answer the trust question at the time the survey is done (say tess). This can be a maximum of 4 years after the election. Even accounting for the fact that trust in politics doesn’t change very much per year, four years is too long for finding a significant change. The other explanation is that trust has no mediating effect between recession and populist vote. It could be that one needs to pay more attention to cultural arguments when assessing the impact of recession on populist
26 voting. One possible mediating effect could be that during a recession attitude towards immigrants change, affecting more the “thick ideology” many (right wing) populist parties. This would be more in line with the findings of (R. F. Inglehart et al., 2016).
So, in conclusion: we find indications that hypothesis 4.1.1 can be confirmed. Recession seem to matter when considering voting for a populist party or not. However, one needs to consider that as soon as the recession indicator becomes positive we see a large spreading of the results. Next to that, the grouping of the observations makes a large difference in the significance of recession on populist support. Hypothesis 4.1.2 can be rejected. There seems to be a small effect, yet the difference in significance of recession on populist vote in the two models and the incorrect time frame between trust (Tess) and recession (Telection) make it unlikely to confirm this hypothesis.
H4.2: POLITICAL TRUST HAS A MEDIATING EFFECT BETWEEN THE PERSONAL ECONOMIC SITUATION AND POPULISM
The effects of feeling about household income are tested as separate independent variables as the Wald test showed to be significant (Prob >Chi = 0.0000). Unlike the effect of recession on populist vote, the different grouping made little difference for the significance or the coefficient of each variable. For my
interpretation I’ll use the grouping based on ESS round, country14. Also, the time component plays less a role in interpreting these coefficient as feeling about household income and trust are measured both at the same time (tess). Ideally, these variables would have been measured right before the election date. The results thus show some delay and should also be viewed with some caution.
Taking “living comfortably with household income” (1) as the reference category, each step away from it increases the chance of voting for a populist party. Someone who is feeling that it is very difficult to live from household income has a .1598262 point higher chance of voting for a populist party, compared to reference category. Someone who is feeling that it is difficult .2890576 compared to reference category. And someone who is coping .2045493 compared to the reference category. Remarkable from these results is that (in contrast to political trust) personal economic situation doesn’t have a linear relation to populist support. Category 4 scores remarkable lower than category 2 and 3, with category 3 having the highest chance of voting for a populist party compared to the reference category (1). This goes against hypothesis H4.2.1: A deteriorating
personal economic situation leads to populist support.
I found these results to be remarkable: a decrease in personal economic situations isn’t linear connected to support for populism. As the results show, people who have are in the third category are the most likely to vote for a populist party. Category 4 is the least likely to vote for a populist party compared to the reference category. One would expect those who have the lowest political trust (the most disillusioned voters) to be the strongest supporters of populist parties (Guiso, 2017; Ivarsflaten, 2005; Muis & Immerzeel, 2017; Fieschi & Heywood, 2004; Norris, 2017). The relation between low political trust and populist support
