POLITICAL CONSEQUENCES OF ECONOMIC INEQUALITY

By
Chunho Park

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Political Science – Doctor of Philosophy
2017

ABSTRACT
POLITICAL CONSEQUENCES OF ECONOMIC INEQUALITY
By
Chunho Park
This dissertation consists of three essays that investigate various political
consequences of economic inequality in democracies. While focusing on the manners in
which political actors, including political parties and voters, respond to rising economic
inequality in electoral competition, each essay provides explanations of why voter-party
linkages based on redistributive preferences weaken when economic inequality increases.
The first essay investigates why, counterintuitively, the poor do not vote for leftist
parties at the ballot box. While previous studies answer this question by focusing on potential
factors distracting the poor from their economic interests, they fail to account for the
economic and institutional contexts that may affect the poor’s voting calculus. In order to fill
this gap, this chapter theorizes that poor voters rely on changes in economic inequality to
evaluate the performance of leftist governments. Specifically, I demonstrate that the poor
support leftist parties only if the leftist government successfully advances the economic wellbeing of the poor by reducing economic inequality. Employing a hierarchical regression
analysis using survey data from 54 elections across 21 advanced democracies, I find that
income-based voting decreases when the wealth gap widens under leftist governments.
The second essay focuses on right-wing parties’ responses to changes in economic
inequality in electoral competition. This chapter argues that the varying degrees of political
constraints in advanced and emerging democracies incentivize right-wing parties to respond
in different manners to the various levels of economic inequality. Specifically, rightist parties
in advanced democracies attempt to politicize social issues in the face of high inequality. The
reason underlying this attempt is that in advanced democracies stronger political constraints

imposed on the strategic choice of party leadership curb opportunistic policy moderation of
the rightist parties. In nascent democracies, however, the right-wing parties opt for more
leftist positions within the economic dimension. I find supporting evidence for the predictions
using 1754 party platforms of 475 parties in 44 democracies.
The last essay empirically examines factors that may affect the intensity of ethnic
appeals of political parties in electoral competition. In order to investigate the determinants of
ethnic appeals, I focus on political and economic conditions that shape the incentive of
political parties to engage in ethnic appeals in their pursuit of electoral gain. Relying on
previous research studying ethnic politics, I then identify political and economic factors that
are argued to incentivize political entrepreneurs to mobilize voters around ethnic issues. I find
consistent evidence that economic inequality between (or within) ethnic groups is positively
(or negatively) correlated with the intensity of parties’ ethnic appeals using the information
on party platforms of 386 parties across 27 democracies, whereas I fail to find supporting
evidence for the effects of the other factors on ethnic appeals. The results of empirical
analysis provide important implications for policy makers to minimize the negative
consequences of ethnic politics in ethnically divided societies.
This dissertation contributes to the better understanding of the relationship between
inequality and redistribution by offering alternative mechanisms of how greater economic
disparity causes the breakdown of programmatic voter-party linkages based on economic
preferences. Each chapter demonstrates how rising economic inequality may induce political
agents - voters and parties – to respond it in a manner that de-emphasizes a redistribution
issue in their pursuit of self-interest in elections. In doing so, this dissertation highlights the
importance of dynamics between political actors in electoral politics in understanding the
relationship between economic inequality and redistributive outcomes in democracies.

Copyright by
CHUNHO PARK
2017

To My Parents, for Their Relentless Support and Love

v

ACKNOWLEDGEMENTS

As I complete this dissertation, I would like to express my appreciation to people for
their support and guidance. My utmost and sincere gratitude should go to my academic
advisor, Professor Eric C. C. Chang. He never lost faith in me even during the time I made
glacial progress due to multiple personal issues. His intellectual sharpness has always
motivated me to think harder and deeper. He has also consistently inspired me to pursue
novel ideas supported by rigorous empirical evidence while working on this dissertation. I
would have never completed this dissertation without his persistent moral support as well as
intellectual guidance.
I also owe my dissertation committee members a debt of gratitude for their
invaluable inputs. Professor Cristian Bodea was my mentor during the first year of my Ph.D.
study in East Lansing. She helped me adjust myself to the new environment and advised on
my coursework plans. Professor Jeffrey Conroy-Krutz has always been supportive of my
ideas. His class was a great opportunity for me to learn the basics of electoral politics in a
comparative perspective and his emphasis on the importance of fleshing out abstract ideas
with real world examples always impressed me. I was more than fortunate to invite Professor
Christian Houle to my dissertation committee. The fourth chapter of this dissertation would
have not been possible without his generous permission to use his dataset. I am also deeply
grateful to him for the opportunity to join his research project. It has been a great learning
experience for me to take one step further.
I am indebted to many other people at Michigan State University. I would like to
thank Professors Melinda Gann Hall, Thomas Hammond, Richard Hula, Steve Kautz, and
Charles Ostrom for their generous financial support. I am also very grateful to Karen Battin,
Rhonda Burns, and Sarah Krause for their superb and timely administrative support. My
vi

graduate life at South Kedzie Hall would have been much tougher without their excellent
assistance. I have benefited from many distinguished teachers: Professors Michael Bratton,
Michael Colaresi, Eric Juenke, Kunho Kim, Dan Lee, Carolyn Logan, Arijit Mukherjee, Ani
Sarkissian, Sandra Schneider, Peter Schmidt, Cory Smidt, Gary Solon, and Jeffrey
Wooldridge. I would like to thank them for their teaching and inspiration. I also thank my
colleagues for their friendship and time spent together: Hyun Jin Choi, Seoyoun Choi,
Kangwook Han, Sung Min Han, Petra Hendrickson, Masaaki Higashijima, Shih-Hao Huang,
Jiyoung Jang, Marty Jordan, Changkuk Jung, Brian Kennedy, Hyunwoo Kim, Jerry Lavery,
Helen Lee, Erica Maylee, Johann Park, Eva Radakovic, Jessica Schoenherr, Emma Slonina,
Nate Smith, Melanie Whitley, Wen-Chin Wu, Fangjin Ye, and Jiho Yoon. My life in East
Lansing would have been much duller without these people.
I would have never considered coming here and studying political science if not for
my teachers in South Korea. My sincere appreciation goes to my M.A. thesis adviser,
Professor Chan Wook Park. He always encouraged me to continue my study for a Ph.D.
despite that I was not a good student during my undergraduate years. I would also like to
express my gratitude to Professors Chang Jae Baik, Dal-Joong Chang, Su-Ik Hwang, and
Kyung Hoon Leem for their mentoring and guidance since the beginning of my graduate
study.
Many special thanks also go to my friends since my teenage years, Hyun-Chul Jang,
Jae-Ho Ki, Chan-Yeon Lee, Seung-Won Lee, Jung-Hwan Lim, Seung-Soo Roh, Ho-Woong
Ryu, and Inho Yang. I always appreciate their presence in my life and the joyful memories for
the last 25 years. I am also grateful to my undergraduate friends for serious as well as casual
conversations. I am truly blessed to have friends to count on whenever I am in need of moral
and intellectual support.

vii

Last but not least, this dissertation would not have been completed without the
support of my family. First, I would like to express my utmost gratitude to my wife, Seoyoon,
for her love, support, and faith in me. She has been the rock of my life. It is not a coincidence
that I began to make progress with this dissertation after we started dating, as she has been
my biggest inspiration during the last three years of my graduate school journey. As a Ph.D.
student at University of Michigan, she was willing to move to East Lansing only to stay with
me. While pursuing her own degree, she has been always supportive and remained positive
even when my progress is delayed. My sister and brother-in-law, Jinui Park and Sung-Il Han,
have always been the biggest supporters for my Ph.D. study. They always welcomed me
warmly, along with my nephew Jayden, whenever I visited South Korea and took time out of
their busy schedule to spend time with me. My parents-in-law, Ok-Hee Chong and Ji-Won
Woo, have waited for the completion of this dissertation with patience and consistently
encouraged me that my endeavor will pay off some time soon. Finally, I am deeply and
forever indebted to my parents, Moon-Ja Hong and Hong-Woo Park. They have always
believed in me and respected my decision at every “critical juncture” of my life. They have
never forced me to do anything and always supported what I chose to do since my teenage
years. If I have any virtues to be proud of, they are all legacies of my parents. This
dissertation is dedicated to my parents for their unconditional love and unwavering support.

viii

TABLE OF CONTENTS

LIST OF TABLES …………………………………………………………………...…....... xi
LIST OF FIGURES ……………………………………………………………………….. xiii
CHAPTER 1 INTRODUCTION ……………………………………………………............. 1
1.1. Puzzle and Motivation ……………………………………………………………... 1
1.2. Arguments ………………………………………………………………………..... 6
1.3. Contributions ………………………………………………………………………. 9
CHAPTER 2 WHY DO THE POOR NOT VOTE FOR THE LEFT? INCOME-BASED
VOTING AND INEQUALITY CHANGES IN ADVANCED DEMOCRACIES ……….... 12
2.1. Introduction ……………………………………………………………………... 12
2.2. Literature Review: Why Do the Poor Not Vote for the Left (or Support Redistribution)? ……………………………………………………………………………. 16
2.3. Theory ………………………………………………………………………….... 19
2.3.1. Building Blocks: Partisan Model of Economic Outcomes and Economic Voting
………………………….……………………………………….,,.................. 20
2.3.1.1. Partisan Model of Economic Outcomes ……………………………........ 20
2.3.1.2. Economic Voting ……………………………………………………........ 21
2.3.2. Inequality Changes and the Poor’s Vote Choice …………………………...... 23
2.3.2.1. Inequality Changes and Different Expectations According to Government
Partisanship ………….………………………………………………….... 23
2.3.2.2. Sticking to or Defecting from the Left? The Poor’s Choice at the Ballot Box
…………………………………………………………………................. 26
2.4. Data, Measurements, and Model Specification ………………………………...... 28
2.4.1. Cases and Data …………………………………………………………….... 28
2.4.2. Dependent Variable …………………………………………………………. 30
2.4.3. Independent Variables ………………………………………………………. 31
2.4.4. Model Specification and Estimation Strategy …………………………….... 33
2.5. Estimation Results ………………………………………………………………. 34
2.5.1. Main Analysis ………………………………………………………………. 34
2.5.2. Testing Causal Mechanism: Do the Poor Not Vote for Parties in Government
When Inequality Rises? …………………………………………………...... 40
2.5.3. Robustness Tests and Additional Analysis …………………………………. 43
2.5.3.1. Robustness Tests ……………………………………………………….... 43
2.5.3.2. Additional Analysis on the Moderation Effect of Religiosity …………... 50
2.6. Conclusion ………………………………………………………………………. 52
APPENDIX ……………………………………………………………………………. 55
CHAPTER 3 CHANGING POSITION OR PLAYING FIELD? THE DISTINCT EFFECTS
OF ECONOMIC INEQUALITY ON STRATEGIES OF RIGHT PARTIES IN OLD AND
NEW DEMOCRACIES ……………………………………………………………………. 60
3.1. Introduction ……………………………………………………………………… 60
3.2. Literature Review: Party Competition in the Policy Space …………………....... 63

ix

3.3. Theory: Party Strategies, Political Constraints, and Economic Inequality ……… 65
3.3.1. Policy Space: Definition ……………………………………………………. 65
3.3.2. Issue Positioning versus Issue Emphasis …………………………………… 66
3.3.3. Political Constraints to Strategic Choice of Parties ………………………… 67
3.3.4. Political Parties and Electoral Competitions in Old and New Democracies .. 70
3.3.5. Economic Inequality and Right Parties’ Strategies in Old and New Democracies
………………………..………………………………………………….......... 73
3.4. Data, Measurements, and Model Specification …………………………….......... 76
3.4.1. Data and Cases …………………………………………………………….... 76
3.4.2. Dependent Variables: Issue Emphasis and Issue Positioning …………......... 77
3.4.3. Independent Variables ……………………………………………................. 78
3.4.4. Control Variables ……………………………………………………………. 78
3.4.5. Model Specification and Estimation Strategy ………………………………. 80
3.5. Estimation Results ………………………………………………………….......... 82
3.5.1. Main Analysis …………………………………………………………......... 82
3.5.2. Testing Causal Mechanism: Do Political Constraints Really Matter? …….... 87
3.5.3. Robustness Tests ………………………………………………………......... 92
3.6. Conclusion …………………………………………………………………........ 100
APPENDIX ………………………………………………………………………....... 102
CHAPTER 4 WHICH FACTORS MOTIVATE POLITICAL PARTIES TO ENGAGE IN
ETHNIC APPEALS IN ELECTORAL COMPETITION? ……………………………...... 107
4.1. Introduction ……………………………………………………………………… 107
4.2. Literature Review: The Different Perspectives on Ethnic Politics and the Politicization
of Ethnicity ……………………..……………………………………………...... 109
4.2.1. Primordialism versus Constructivism ………………………………………. 109
4.2.2. Why Study Ethnic Appeals? …………………………………………........... 110
4.3. Ethnic Appeal and Its Determinants: Conditions under Which Ethnic Identities
Become Salient ……………………….………………………………………...... 114
4.4. Data, Measurements, and Model Specification …………………………………. 119
4.4.1. Data and Cases …………………………………………………………........ 119
4.4.2. Dependent Variable: Measuring Ethnic Appeals …...………………………. 120
4.4.3. Independent Variables ………………………………………………………. 122
4.4.4. Control Variables ………………………………………………………......... 124
4.4.5. Model Specification and Estimation Strategy ………………………………. 126
4.5. Estimation Results ……………………………………………………………….. 127
4.5.1. Main Analysis ………………………………………………………………. 127
4.5.2. Robustness Tests ……………………………………………………………. 132
4.6. Conclusion ………………………………………………………………………. 134
APPENDIX …………………………………………………………………………... 137
CHAPTER 5 CONCLUSION …………………………………………………………...... 140
BIBLIOGRAPHY ………………………………………………………………………… 143

x

LIST OF TABLES

Table 2.1. Estimation Results of Hierarchical Linear Models ..……………………………... 36
Table 2.2. Estimation Results of Hierarchical Probit Models .…………………………….... 41
Table 2.3. Robustness Checks 1: Alternative Measures of Government Partisanship .……... 45
Table 2.4. Robustness Checks 2: Alternative Measure of Income (Quintile) and Introducing
Country-level Heterogeneity .……………………………………………………………….. 46
Table 2.5. Robustness Checks 3: Alternative Measure of Dependent Variable (Dichotomous)
……………………………………………………………………………………………..... 47
Table 2.6. Robustness Checks 4: Fixed Effects Estimation and OLS with Cluster-robust
Standard Errors ……………………………………………………………………………... 49
Table 2.7. The Effect of Poor Conditional on Religiosity .………………………………….. 51
Table 2.8. List of Country-elections, Partisanship of Incumbent Governments, and Missing
Variables ……………………………………………………………………………….……. 56
Table 2.9. Details Regarding Control Variables .…………………………………………..... 58
Table 2.10. Summary Statistics .…………………………………………………………... 59
Table 3.1. Regression Analysis of Emphasis on Social Issues and Positions on Economic Issues
in Advanced and Emerging Democracies …………………………………………………… 84
Table 3.2. Regression Analysis of Emphasis on Social Issues and Positions on Economic Issues
Depending on Inequality Level and Degree of Political Constraints .……………………..... 89
Table 3.3. Robustness Checks 1: Alternative Classifications of Old and New Democracies
(Four or Six Elections) .…………………………………………………………………....... 93
Table 3.4. Robustness Checks 2: Alternative Classifications of Old and New Democracies
(Years of Democratization) .……………………………………………………………….. 94
Table 3.5. Robustness Checks 3: Alternative Setup of Three-level Hierarchy (Country and
Country-election Clusters) ………………………………………………………………..... 95
Table 3.6. Robustness Checks 4: Excluding Non-left and Non-right Parties .……………..... 98
Table 3.7. Robustness Checks 5: Alternative Measure of Niche Party .…………………….. 99
Table 3.8. List of Country-elections (Old and New Democracies) …...……………………. 103

xi

Table 3.9. Policy Categories for Emphasis on Social Issues …….………………..……….. 105
Table 3.10. Policy Categories for Position on Economic Issues ……………………...……. 105
Table 3.11. Details Regarding Control Variables …………………………..………………. 106
Table 3.12. List of Indicators in V-Dem Database for Political Constraints ...……………... 106
Table 3.13. Summary Statistics ...……………………………………………..…………… 106
Table 4.1. Estimation Results of Hierarchical Linear Models …....…………..……………. 129
Table 4.2. Robustness Checks: Alternative Measure of Ethnic Appeals and Excluding Outliers
…………………………………………..………………………………..…..……………. 133
Table 4.3. Policy Categories for Ethnic Appeals …………….…………………………….. 138
Table 4.4. Summary Statistics .…………………………...………………………………... 138
Table 4.5. List of Countries and Election Years (28 Countries) ..………...………………… 139

xii

LIST OF FIGURES

Figure 2.1. Effects of Poor Conditional on Inequality Change under Leftist and Non-leftist
Government ………………………………………………………….……………………... 37
Figure 2.2. Predicted Economic Ideology of Parties Voted for by the Poor and Non-poor under
Leftist and Non-leftist Government ......……………………….……………………………. 38
Figure 2.3. Predicted Probabilities of the Poor’s Voting for Incumbent Conditional on
Inequality Change under Leftist and Non-leftist Government .……….…………………… 42
Figure 2.4. Effects of Poor on Left-wing Voting Conditional on Inequality Change
(Dichotomous, Predicted Probabilities) ……………………………….……….…………… 48
Figure 3.1. Effect of Right on Emphasis on Social Issues Conditional on the Level of Inequality
in Advanced and Emerging Democracies .……………….…………….…………………… 85
Figure 3.2. Effect of Right on Positions on Economic Issues Conditional on the Level of
Inequality in Advanced and Emerging Democracies .………………….…………………… 86
Figure 3.3. Effect of Right on Emphasis on Social Issues Conditional on the Level of Inequality
under Different Degrees (Strong or Weak) of Political Constraints ..…………….…………. 90
Figure 3.4. Effects of Right on Position on Economic Issues Conditional on the Level of
Inequality under Different Degrees (Strong or Weak) of Political Constraints ……………. 91
Figure 4.1. Predicted Ethnic Appeal at the Election-level Conditional on the Levels of BGI and
WGI ……………………………………………………………………………………….. 130
Figure 4.2. Predicted Ethnic Appeal at the Party-level Conditional on the Levels of BGI and
WGI ……………………………………………………….…………………………......... 131

xiii

CHAPTER 1
INTRODUCTION

“In a democracy the poor will have more power than the rich, because there are more of
them, and the will of the majority is supreme.” – Aristotle, The Politics, Book 6

“The flaw in the pluralist heaven is that the heavenly chorus sings with a strong upper-class
accent.” – E. E. Schattschneider, The Semisovereign People: A Realist’s View of Democracy
in America

1.1. Puzzle and Motivation
This dissertation examines how economic inequality is associated with the strategic
decisions of political agents, including voters and parties, in electoral politics. I investigate
impaired programmatic linkages between voters and parties based on economic interests or
preferences by addressing the following three questions: (1) why do the poor not support the
left?; (2) how do right-wing parties respond to rising inequality in their electoral platforms?;
and (3) which factors do motivate political parties to intensify their ethnic appeals?
An election is the principal mechanism through which representative democracy
operates. As the “one man, one vote” principle is the bedrock of democracy, any set of
individuals to constitute the majority of society, in principle, can secure their demands
through their influence in government. In order to influence decision-making, people choose
as their representatives those who can best serve their interests at the ballot box. This intrinsic
feature of representative democracy gives clout to those forming the majority. This is also
why democracy is supposed to be favorable for the poor (Acemoglu and Robinson 2001;
Brown and Mobarak 2009; Bueno de Mesquita et. al. 2003; Lake and Baum 2001;
Przeworski et. al. 2000; Stasavage 2005). As Tocqueville observed, “the greatest number has
1

always been composed of those who had no property,” and thus “universal suffrage really
gives the government of society to the poor” (Tocqueville 2010: 336).
The idea of democracy as a redistribution mechanism is well-reflected in the standard
model of the political economy of elections. Meltzer and Richard (1981) provide a concise,
but powerful, model to summarize how democracy serves the poor’s interests. Their model
starts with two plausible assumptions: the preference of the median voter is crucial in
determining policy outcomes (Downs 1957) and the income distribution is right-skewed
toward the top end. When these assumptions hold, the income level of the median voter is
located at a point less than the average income, due to the right-skewness of the income
distribution, and thus the median voter, whose preference is decisive in government
decisions, always supports redistribution through taxation (Meltzer and Richard 1981). This
perspective considering democracy as a redistribution mechanism that favors the poor is
reflected in the literature on democratization, where economic inequality is a key factor in
democratization due to the inherent promise of redistribution after transition to democracy
(Acemoglu and Robinson 2006; Boix 2003). The relationship between economic inequality
and redistribution can also be predicted by the Meltzer-Richard (hereafter MR) model. If the
logic behind the MR model is correct, greater economic disparity should be associated with
more redistribution to ameliorate economic inequality. This association is because the median
voter wants more redistribution as the distance between the mean and median voters in their
positions in the income distribution increases with economic inequality.
Although the prediction of the MR model is confirmed in a few previous studies
(e.g., Kenworthy and Pontusson 2005; Milanovic 2000), the prediction does not always hold
in the real world. Not only does democracy sometimes fail to redistribute or enhance the
poor’s welfare more than non-democracies (see, e.g., Ross 2006; Timmons 2010), but
democratic countries with a more unequal income distribution also tend to redistribute less

2

than those with lower inequality (e.g., Alesina and Glaeser 2004; Benabou 1996; Perotti
1996). The negative association between inequality and redistribution is known as the Robin
Hood Paradox (Lindert 2004). In particular, recent years have also witnessed growing
income disparity in a number of advanced democracies. Gini indices of disposable income
inequality have increased in 13 OECD countries, notably the United Kingdom and the United
States, from 1978 to 2002 (Beramendi and Cusack 2009). Moreover, less economic disparity
has not been necessarily followed by democratization (Bermeo 2009). While democracies
have lower levels of economic inequality than autocracies on average, the gap is not that
great and the difference disappears if only countries below a $10,000 GDP per worker
threshold are compared (Gallagher and Hanson 2009).
Many alternative explanations have been proposed for the discrepancy between the
prediction of the MR model and the relationship between inequality and redistribution. First,
power resources theory suggests the exact opposite prediction: unequal societies redistribute
less than equal ones (Korpi 1983; Stephens 1979). The power resources theory challenges the
core assumption of the MR model that every citizen has equal power and influence over
decision-making in government. To the contrary, this theory focuses on the importance of
organizational resources, such as unions with wider coverages and high densities as bases of
strong leftist parties. It argues that policy outcomes, including the size of redistribution, are
byproducts of the power balance between classes and power asymmetry increases in favor of
the rich as inequality increases. Even though the low-income classes have numbers, they
increasingly fewer resources to pursue their interests as inequality rises. As a result, higher
inequality tends to produce less redistribution. While focusing on other factors, such as
unequal accesses to the policy-making process between the poor and rich or low turnout rates
among the low-income voters, another strand of research also proposes an explanation for the
negative association between inequality and redistribution (Benabou 1996, 2000; Bartels

3

2008; Pontusson and Rueda 2010; Stiglitz 2012). There is also another body of research that
underscores a social insurance aspect of social policies for high-income citizens (Moene and
Wallerstein 2001) or specific skills workers (Iversen and Soskice 2001). This literature argues
that the preferences over redistribution are not determined by citizens’ relative positions in an
income distribution, as the MR model assumes, but by the size of their unpredicted income
shocks (Moene and Wallerstein 2001) or the composition of their skills (Iversen and Soskice
2001). Rehm (2009) also recognizes the importance of the insurance aspect, as he finds that
the extent to which citizens are exposed to the risk of unemployment is crucial in determining
their preferences for redistribution. Lastly, several previous studies focus on electoral
institutions to explain redistributive consequences in advanced democracies. In general, they
reach the consensus that the PR electoral system redistributes more than majoritarian system.
Persson and Tabellini (2000, 2003) find that parties try to draw support from ‘swing’ electoral
districts by targeting government expenditure on public goods toward those particular
districts. Iversen and Soskice (2006) develops a model of class coalition among three groups
– the rich, the poor, and the middle class. In their model, the PR system tends to produce a
left-wing government, since the middle class voters have an incentive to ally with the poor to
tax the rich under the PR system. On the other hand, the majoritarian system tends to produce
a two-party system with two parties as class coalitions and the election promises of two
parties – the center-left and center-right –, which appeal to the middle class are not credible.
On this occasion, the middle class voters are more likely to elect the center-right, as they fear
the excessive extraction by the left than the right’s post-election deviation.
While I agree with the criticisms that the MR model oversimplifies the process
whereby citizen preferences over redistribution via taxation are constructed and fails to
account for institutional contexts and the power relation between classes, my focus is not on
these specific criticisms, but examines instead an important aspect of representative

4

democracy neglected in deriving the prediction of the MR model: the programmatic linkages
between voters and parties based on economic interests or preferences. Ideally, the idea of
representative democracy operates based on parties as agents that represent their
constituencies’ interests or preferences and implement what they promised during their
electoral campaigns if elected into office. However, the assumption of the Meltzer-Richard
model that there is a strong tie between parties and voters based on income levels does not
always hold in the real world as described in the previous section.
The breakdown of programmatic ties between parties and their core constituents is
also a significant issue in terms of representation. The concept of representation is widely
discussed in the political science literature, but it is a multi-layered term with many different
interpretations. What I mean here by representation is substantive representation, where
elected officials act as agents of their constituents, promoting constituents’ interests (Pitkin
1967). Other forms of representation, such as symbolic or descriptive representation,
emphasize the degree of the demographic correspondence or the accuracy of the resemblance
between principals and their agents (Pitkin 1967). However, modern political science
literature has largely focused on substantive representation, viewing representation as a link
between the preferences of the represented and policy measures of the representative. In other
words, representation should mean that the representative implement policies on behalf of
and for the interests of the represented. In this sense, the empirical research on substantive
representation usually focuses on the correspondence between citizens and parties as
principals and agents, respectively, using indicators measuring ideological congruence
between the median voter and the government (Budge and McDonald 2007; Powell 2000,
2006, 2009; Powell and Vanberg 2000). While the conceptualization and empirical strategy
are compelling, I believe representation can be better conceptualized and understood by
considering political reality where there are inevitable conflicts of heterogeneous interests

5

and political parties representing those interests. The heterogeneous interests among citizens
lead to different preferences over economic policies (Drazen 2000). Those conflicting
interests are sometimes resolved by an appropriate compromise, but, more often than not,
they cannot be reconciled. This conflict leads to one economic policy winning over other
alternatives. As a result, some groups are represented, but others remain underrepresented.
In addition, representative democracy is a party democracy in which multiple
representative parties compete. In representative democracies, the authority of policy
decisions usually belongs to political parties in government. Ideally, the political parties are
supposed to be an agent of the entire citizenry, but the reality is that they are perceived to
represent respective parts of the citizenry rather than representing as a whole, and have
particular preferences over alternatives according to their partisanship and the interests or
preferences of their core constituents. This idea also accords with the existing literature on
power resource theory and the traditional theory of partisan politics regarding political parties
as representatives of social classes in the “democratic class struggle” (Esping-Andersen 1990;
Korpi 1983; Stephens 1979). Therefore, this dissertation employs the substantive
representation concept, which presumes inevitable conflicts of interests among groups with
different preferences over policy outcomes represented by divergent political parties. This
conceptualization is distinct from one that sees representation as congruence between
representatives and the median of the body politic.

1.2. Arguments
In order for the prediction of the MR model to hold in representative democracies, it
follows that leftist (or rightist) parties that advocate higher (or lower) levels of redistribution
and progressive tax policies are more likely to pursue the poor’s (or the rich’s) interests when
they are governing. However, the programmatic linkages between voters and parties assumed

6

in traditional partisan politics, oftentimes, gives way. In reality, leftist parties do not pursue
the poor’s interests once elected in office or the poor do not necessarily vote for parties on the
left. Recent studies on partisan politics and the welfare state also indicated that traditional
voter-party linkages have weakened, and parties falling along the same ideological dimension
choose different goals and policies from one another across advanced democracies
(Häusermann et al. 2013; King and Rueda 2008; Kitschelt 1994). In addition, studies on
electoral competition have shown that the left can employ tax policies divergent from the
poor’s preferences (Dixit and Londregan 1996; Roemer 1998). No less significant is the fact
that the poor also do not always support leftist parties. In other words, they do not always
elect representatives that pursue (or claim to pursue) their interests. Income-based voting is
not always substantial even in advanced democracies. There is considerable variation in
income effects on vote choice across countries, with the effect of sometimes producing
results contrary to expectations in some countries (Norris 2004). Each chapter of this
dissertation attempts to provide explanations of how economic inequality is associated with
the fraying programmatic ties between voters and parties based on economic preferences. In
doing so, my aim is to illuminate the centrality of economic inequality in understanding
electoral politics in democracies.
The second chapter establishes the causal mechanism that explains the poor’s
electoral choice while emphasizing the importance of inequality changes and government
partisanship to the poor’s evaluation of government performance. More specifically, I argue
that the poor do not support the left if leftist parties in government fail to enhance their
welfare. While the disadvantaged reasonably expect that a leftist government to improve
economic inequality, the poor are not willing to vote for the left if the left in government fails
to do so. I hypothesize, therefore, that the prevalence of income-based voting depends on the
performance in improving economic inequality when the left is governing.

7

The third chapter examines how right-wing parties attempt to turn voters’ attention
away from redistribution issues when facing high economic inequality and how the response
of rightist parties can differ depending on the level of political constraints. The MR model
provides a parsimonious explanation of how democracy benefits the poor, but it also has an
important implication for the electoral strategies of rightist parties. According to the MR
model, the extent to which electoral competition is favorable for the left increases with
economic inequality, as the number of citizens who prefer redistribution via progressive
taxation also increases. Moreover, policy decisions over tax and redistribution are more likely
to emerge as central issues as inequality increases. In this chapter, I demonstrate that rightwing parties, in the face of rising inequality, try to improve their electoral fortune by luring
the middle- or low-income voters away from the left or drawing voter attention away from
economic issues, and that the right’s specific strategy differs depending on the degree of
political constraints imposed on party leadership.
Finally, the fourth chapter empirically examines conditions which induce political
parties to adopt ethnic appeals in their electoral campaign. To this end, I distinguish the
political activation of ethnic identities from the social salience of ethnicity and consider
ethnic appeals of political entrepreneurs as a primary mechanism that connects those two. In
other words, the political activation of ethnicity is a product of political entrepreneurs’ efforts
to build their support base by appealing to ethnic identities. In order to study conditions that
motivate political parties to intensify their ethnic appeals, I draw the political and economic
determinants - electoral competitiveness, electoral institutions, the information availability,
and economic inequalities between and within ethnic groups (hereafter BGI and WGI) - of
political parties’ ethnic appeals. I find strong evidence that the intensity of ethnic appeals
increases (or decreases) with the level of BGI (or WGI).

8

1.3. Contributions
Each of the three chapters answers the questions set forth at the very beginning of
this introduction, thereby contributing to the extant literature. First, this dissertation offers an
alternative explanation to the existing ones for a long-lasting puzzle: why do the poor not
vote for the left? The extant literature on this issue has explored various factors that draw
voter attention away from economic or redistributive preferences influencing the poor’s
preferences and voting (e.g., De La O and Rodden 2008; Huber and Stanig 2007). Unlike the
previous studies, the second chapter gives particular weight to the political (i.e., government
partisanship) and economic (i.e., inequality changes) contexts in which the less well-off make
their electoral choices. While many studies focus on noneconomic issues when they
investigate voting behavior and policy outcomes, the second chapter highlights the
importance of income inequality, which is more directly related to their economic welfare
than overall prosperity, in explaining the poor’s voting behavior. I provide a compelling
explanation for this long-lasting puzzle by focusing on the poor’s voting calculus based on
government performance relevant to their economic situation. While doing so, the importance
of the supply side of redistribution and resulting outcomes in the poor’s voting decision is
underscored.
My argument in the third chapter is in line with the recent literature on the
relationship between party strategies and income disparity (e.g., Tavits and Letki 2014; Tavits
and Potter 2015), I contribute to the extant literature by incorporating political constraints on
party leadership into my theoretical framework for studying responses of right-wing parties to
greater economic disparity. Indeed, the central findings of the second chapter indicate that the
varying levels of political constraints lead the right in advanced and nascent democracies to
respond to increasing economic inequality, while treating two types of party strategies –
position-taking and manipulating issue salience –as substitutes for the right. The central

9

finding of the third chapter represents a significant expansion of the extant literature, which
studies either aspect of party strategies only (Tavits and Letki 2014; Tavits and Potter 2015).
Finally, the fourth chapter is one of the very rare studies to analyze ethnic appeals of
political parties from a comparative perspective. This chapter also has important policy
implications on how to minimize the negative consequences of ethnic politics. In its
investigation of the determinants of ethnic appeals in electoral competition, this chapter
underlines the importance of BGI and WGI in preventing the ethnification of politics. The
negative impacts of ethnic divisions in a society have been well-documented. Thus, ample
political science research suggests various means, such as designing electoral institutions, to
prevent ethnic divisions from developing into politically activated cleavages (e.g., Horowitz
1985; Lijphart 1977; Reilly 2001). However, my findings in this chapter imply that it is
critical for government to distribute material benefits from economic development across, not
along, ethnic lines in order to alleviate ethnic grievances due to economic disparity between
ethnic groups. Based on the findings that high BGI (or low WGI) induces political parties to
engage in ethnic appeal for their electoral gain, the fourth chapter recommends policy-makers
to bend every effort to prevent the socio-economic conditions – high BGI and low WGI –
from occurring. This recommendation arises from the importance of controlling extreme
ethnic appeals of political entrepreneurs and their efforts for ethnic mobilization.
These dissertation chapters taken as a whole contribute to the literature on the effects
of inequality on various political outcomes, such as voting behavior and party strategy. My
dissertation also offers mechanisms of how explanations of mechanisms through which
increasing economic inequality leads to the non-politicization of redistributive issues. In light
of the prediction of the MR model for the association between inequality and redistribution,
the findings of this dissertation imply that rising inequality may undermine the programmatic
linkages between voters and parties because of the strategic responses of those political

10

agents in pursuing their interests. Each of the next three chapters derives testable implications
by developing theories of the political consequences of economic inequality while focusing
on the incentives of key political actors, voters and parties, and their strategic responses to
economic inequality. The results from empirical analysis of the theoretical hypotheses are
also presented in each chapter. The final chapter summarizes the findings and concludes the
dissertation.

11

CHAPTER 2
WHY DO THE POOR NOT VOTE FOR THE LEFT? INCOME-BASED VOTING
AND INEQUALITY CHANGES IN ADVANCED DEMOCRACIES

“A house may be larger or small; as long as the neighboring houses are likewise small, it
satisfies all social requirement for a residence. But let there arise next to the little house a
palace, and the little house shrinks to a hut.” – Karl Marx, Wage-labour and Capital

2.1. Introduction
Why do the poor not vote for the left? The association between an individual’s
income level and her vote choice has long been of interest to political scientists. The claim of
income-based voting echoes the partisan theory of economic outcomes, positing that differing
partisan compositions of government produce distinctive social and economic policies and
distributive outcomes (Alt 1985; Hibbs 1977, 1987). Given that a left-wing government
provides more redistribution via higher taxes for the poor (e.g., Boix 1998; Bradley et al.
2003), leftist parties should be favored by the poor who benefit from redistribution. For this
reason, the poor are usually assumed to be a core constituency for left parties (e.g., Iversen
and Soskice 2006). The tie between the left and the poor based on redistributive preferences
is at the core of the vast literature on the political economy of elections. For example, the
seminal Meltzer-Richard model is built on the very assumption that redistributive preferences
of citizens are determined by their relative positions in terms of income distribution and that
such preferences are decisive in their electoral choices (Meltzer and Richard 1981). Indeed,
ample empirical evidence supporting the effect of relative income on redistributive
preferences is found in a large number of micro-level studies (e.g., Brooks and Brady 1999;
Cusack et al. 2008; Finseraas 2009; Gelman 2008; Iversen 2005; McCarthy et al. 2006;

12

Schmidt and Spies 2014; Stegmueller 2013).
Yet despite this theoretical plausibility of income-based voting, in actual practice, the
poor frequently fail to vote for left parties and instead elect representatives who are not
expected to pursue their economic well-being (Alesina and Giuliano 2011). Recent empirical
studies investigating individual-level economic preferences also find that the poor do not
always prefer more redistribution than the rich (Alesina and Glaeser 2004; Finseraas 2009b;
Gaskin et al. 2013; Klor and Shayo 2010; Scheve and Stasavage 2006; Shayo 2009).
Moreover, relative income is not a dominant predictor of vote choice. The effect of an
individual’s income level on her vote choice is known to be weak, especially compared to the
effects of other factors, such as ideological position and religiosity (e.g., Dalton 2006; Norris
2004).
The goal of this chapter is to provide an explanation as to why the poor do not
choose the left at the ballot box by examining economic factors, particularly income
inequality. The effect of income on vote choice has been examined in the comparative
literature, suggesting and testing different explanations (e.g., De La O and Rodden 2008).
However, the recent literature has generally revolved around proposing and testing potential
voting cues which can be substituted for self-interests of poor voters. This branch of research
has focused on potential factors on the “second-dimension” distracting the poor, such as
religion, national identity, and ethnicity in explaining voting behavior deviating from the
expectation of the standard assumption of political economy of elections (e.g., Scheve and
Stasavage 2006; Huber and Stanig 2007; De La O and Rodden 2008). This chapter
contributes to the existing literature by suggesting instead a simple, but novel explanation
based on the poor’s rational calculus of voting decisions: other factors may play a major role
in the poor’s voting decision when their income-based voting is not expected to make
differences in their welfare.

13

This chapter also calls attentions to the rationality of poor voters by testing indirectly
whether the poor are rational enough to recognize their personal gain or loss based on
changes in economic conditions and, in turn, choose their representative based on it. The poor
are depicted in previous studies as irrational voters who do not understand their economic
interest and can easily be manipulated to mobilize around other issues and identities because
the poor tend to have lower education levels than the rich, and thus are less likely to be wellinformed voters (Verba et. al. 1995). This chapter challenges this argument. Scholars studying
the relationship between information and vote choices argue that well-informed voters are
more likely to make “correct” choices, but it does not necessarily mean that the poor lack the
ability to make decisions based on their self-interests (Ansolabehere et al. 2006). The poor
may sometimes understand the policy-making process through which governmental policies
are determined on the basis of limited information, but they are still rational in that they can
make their voting decision by noticing how much their personal economic situation has been
improved or declined under a leftist government. In this sense, the poor may not support the
left not because they are irrational and easily distracted, but rather because they are aware
that the left is not their best choice when considering their economic self-interests.
This chapter also contributes to the literature by incorporating different expectations
according to governmental partisanship into the existing literature on economic voting. The
literature on the political business cycles theory suggests that macroeconomic outcomes and
fiscal policies, including GDP growth, inflation, unemployment rates, and governmental
spending, change according to governmental partisanship. As much as the partisan effect on
economic outcomes is significant, citizens have expectations on what kind of policy goals are
pursued according to the partisan composition of the government. The poor understandably
expect they will benefit more from the left in government. When the left fails to serve the
interests of the poor, however, how do the poor respond to that failure in upcoming elections?

14

Will they still support the left if the party pledges to care about the interests of its core
constituency during the election period? I further the existing literature on economic voting
with new insights by incorporating governmental partisanship and voters’ position on income
distribution into economic voting theory. While party platforms and expected policy positions
work as one of the criteria which help voters decide which party they would vote for, there is
inevitable uncertainty of judgment based on party platforms which may not be credible. On
the other hand, past performance provides voters with more reliable information on what
parties would do when they are in office than party platform does. Therefore, it makes sense
to speculate that the poor also decide whether they will support the left again based on past
performances of the left during its terms. While the existing literature on economic voting
focuses on growth rate and inflation as factors influencing voters’ choice, this chapter
attempts to show that voters evaluate governmental performances based on different
indicators according to their positions on income distribution and they also have different
expectations on the government depending on its partisanship. Finally, this chapter links work
on comparative political behavior to the literature on comparative political economy by
situating individuals’ voting behavior in the context of macro-level economic conditions and
governmental partisanship (Rueda 2008).
This chapter proceeds as follows. The next section, section two, establishes the
argument of this paper based on two well-established bodies of literature on political
economy: economic voting and the partisan model of economic outcomes. The third section
describes details of the research design and presents a series of results from empirical
analyses. In that section, I will also conduct an additional analysis to make sure that the
results of empirical analysis are produced by the proposed causal mechanism. Implications of
the findings are discussed in the fourth and final section.

15

2.2. Literature Review: Why Do the Poor Not Vote for the Left (or Support
Redistribution)?
Existing literature on the political economy of elections pays substantial attention to
turnout rates of poor voters to investigate why politicians do not care about the interest of the
poor and explain variations in efforts to ameliorate economic inequality (Franzese 2002;
Mahler 2002; Pontusson and Rueda 2010). It is known that the turnout rate is lower among
the poor than the rich because they do not have sufficient information and they are not welleducated to make decisions precisely for their own interests. The explanation of
underrepresentation of the poor based on their low turnout rate is quite convincing, but this
also leaves a large part of the story of underrepresented poor citizens unexplained (Bartels
2008).1
Recent scholarly efforts to address this topic emphasize factors which may have an
effect on the poor’s voting behavior and redistributive preferences. In his seminal work,
Roemer (1998) introduces this “second-dimension” argument and demonstrates that when
parties compete in the two-dimensional policy space along economic and non-economic
dimensions (e.g., social values), the poor face a trade-off between their preferences on
economic and non-economic issues. In this case, the poor with religious beliefs vote for rightwing parties, especially if non-economic issues are sufficiently salient around election time.
This insight provides fertile ground for empirical research that investigates the discrepancy
between individuals’ income levels and their redistributive preferences. Subsequent research
attempts to test this “distraction” effect of religiosity on the poor’s redistributive preferences
and vote choices. For example, Scheve and Stasavage (2006) regard religion as a

1

While electoral democracy is built on the rule of one-man one-vote, the rule does not hold in reality
and citizens have proportional political clouts to resources available for them (Alesina and Glaeser
2004). This argument is also in line with the empirical evaluation of the Meltzer-Richard model
concluding that the median voter is not benefited from redistribution under democracy (Milanovic
2000).
16

psychological substitute for social insurance, and posit that this feature of religion leads the
religious poor to prefer lower levels of social insurance spending than the secular poor. Their
individual level analysis finds evidence that there is a significant negative correlation
between religious attendances and preferences for social spending. Similarly, Stegmuller
(2013) finds that religiosity leads citizens, regardless of their socioeconomic status, to have
more conservative preferences on both economic and moral dimensions. The effect of
religiosity on citizens’ preferences on the redistributive issue also motivates further research
to examine how religiosity moderates income-based voting. For example, De La O and
Rodden (2008) argue that preferences on the moral issues dimension matter more for the
religious in their vote choices than do their redistributive preferences. They argue that the
discrepancy is caused by the fact that the religious vote according to their positions on moral
issues rather than economic preferences or interests. They find that the poor are more easily
distracted from their economic self-interest than the rich, especially under a PR system and,
accordingly, the poor’s right-wing voting is caused by their positions on moral issues rather
than economic preferences.
There is also a growing body of evidence that various identifications with social
groups besides those based on religion affect the preferences of the poor regarding
redistribution and their vote choices. First of all, there have been several explanations of how
ethnic group loyalty and ethnic fractionalization induce the poor to withdraw support for
redistribution and the left. Alesina and Glaeser (2004) argue that the level of redistribution in
ethnically divided societies is likely to be low because individuals in one ethnic group do not
want to spend revenue on redistribution which may make other ethnic groups better off. As a
result, ethnic fractionalization tends to reduce support for redistribution. Antipathy to
immigrants also plays a role in determining preferences for redistribution and voting
behavior. Finseraas (2009b) investigates an effect of growing immigrant populations on the

17

Western European welfare system. He attempts to test anti-solidarity and distraction
mechanisms that can explain the relationship between increasing ethnic heterogeneity and
support for left parties. While he fails to find evidence for the anti-solidarity mechanism, it
turns out that even citizens favoring redistribution can be distracted by antipathy to
immigrant ethnic groups, which leads them to vote for right parties. Strong national identity
is also argued to distract the poor from their economic interests. Shayo (2009) demonstrates
that the poor are more likely to have a strong national identity than the rich because, as
inequality increases, the poor want to identify with their nation rather than with a lowerincome group perceived as a low status group. Moreover, the poor are more likely to demand
lower levels of redistribution as they have strong national identity. Klor and Shayo (2010)
design an experiment to examine how social identity affects preferences over redistribution
and voting behavior. The result of the experiment shows that a considerable number of social
identifiers deviate from their payoff maximization incentive to support their ingroup.
However, it does not mean that social identity is a stronger factor shaping voting behavior of
citizens than economic interests because they tend to support their ingroup only when doing
so is not too costly.
The goal of this chapter is to establish and empirically examine an alternative
mechanism that drives the poor’s right-wing voting. To this end, I emphasize the importance
of government policies and economic outcomes to the poor’s evaluations of their own
economic well-being and their different expectations of the poor regarding economic
outcomes conditional on government partisanship. The second-dimension arguments have
explored potential factors influencing the poor’s preferences and voting, finding convincing
evidence that the poor can vote against their self-interest. This approach, however, is not
completely satisfactory since it fails to account for the economic contexts in which
disadvantaged voters make their voting decisions. These second-dimension arguments

18

prioritize explanations from the demand side of redistribution (i.e., from citizens or voters),
but generally give little attention to the supply side of redistribution policies (i.e., from parties
or government).2 Individuals, however, do not evaluate their personal economic situation in a
vacuum. Moreover, while a government is not solely responsible for economic outcomes and,
which may have unintended consequences in some cases, government policies implemented
during the term of government should contribute to changes in economic outcomes. In
particular, various economic and institutional contexts should receive serious attention to
understand the poor’s electoral choice. In this sense, I argue that the supply side of
redistribution policies and resulting outcomes matter. This argument parallels the works of
Evans and Tilley (2012a, 2012b) which investigates the decline of class voting in the United
Kingdom since the 1980s. They attribute the decline of class voting to an ideological
convergence between the two main parties, positing that the effect of left-right values on vote
choices depends on the degree of party polarization (Evans and Tilley 2012a, 2012b). While
both the works of Evans and Tilley and the present study focus on the supply side of
redistribution policies, there is also a major difference. The works of Evans and Tilley
account for differences in perceived ideologies and policy positions between parties. In
contrast, I emphasize the importance of policy outcomes and voter’s expectation on economic
outcomes according to government partisanship in the poor’s voting calculus.

2.3. Theory
The basic intuition behind my theory is that the core support group for the left (i.e.,

2

Among studies focusing on religiosity, a notable exception is Huber and Stanig (2011), which
suggests a unique causal mechanism of how religion induces the religious poor to prefer lower tax
rates and more limited redistribution than do the secular poor. To establish the causal mechanism that
explains the poor’s voting, they focus on the financial support from the state to religious
organizations. They find that, as financial support from the state for that state’s religious organizations
increases, the religious poor support parties with similar ideological positions favored by the secular
poor.
19

the poor) withdraw their support if parties representing them (i.e., the left) fail to advance
their interests. It is reasonable for the less well-off to want a leftist government to enhance
their welfare by giving high priority to reducing economic inequality. However, if a leftist
government fails to speak to the interests of its core support group, the poor then have less or
no incentive to support the left. I expect therefore that the effect of income on voting depends
on the performance of an incumbent leftist government in improving economic inequality.
Specifically, I predict that income-based voting will prevail when the leftist government
produces economic outcomes that meet voter expectations regarding inequality changes. In
contrast, we are more likely to see the poor’s voting that contradicts their self-interest if
inequality increases under the leftist government. To further establish a causal explanation, I
mostly rely on two of the most studied ideas of political economy literature: the partisan
model of economic outcomes and economic voting.

2.3.1. Building Blocks: Partisan Model of Economic Outcomes and Economic Voting
2.3.1.1. Partisan Model of Economic Outcomes
The vast literature on the partisan model of economic outcomes finds a strong
association between government partisanship and various economic outcomes. In his
pioneering work, Hibbs (1977, 1987) investigates the effect of government partisanship on
inflation and unemployment in the United States, and he finds that leftist governments prefer
low-unemployment and high-inflation outcomes. On the other hand, right-wing governments
pursue policies that prioritize lower inflation at the cost of higher unemployment in
accordance with the preferences of its core constituency (i.e., the wealthy). While both
partisan groups prefer lower inflation and unemployment rates, they weigh on those factors
differently due to the trade-off between inflation and unemployment. In addition, power
resource theory, which was developed in the comparative political economy literature,

20

suggests that the presence and strength of left-wing parties in government is crucial for
redistributive effects of government policies (Korpi 1983; Stephens 1979). Indeed, numerous
empirical studies have affirmed a partisan effect on economic outcomes in advanced
democracies, including unemployment rates (Alt 1985), redistribution (Bradley et al. 2003),
and poverty rates (Brady 2003; Brady et al. 2009), as well as policy instruments, such as
labor market policy (Rueda 2005, 2008), fiscal policy (Cusack 1999; Mulas-Granados 2003),
tax policy (Beramendi and Rueda 2007), and welfare state spending (Allan and Scruggs
2004; Korpi and Palme 2003). Simply put, left- and right-wing parties prioritize different
economic policies congruent with their ideological principles and the economic preferences
of their support groups. The left is expected to level the economic playing field and reduce
inequality, whereas the right is expected to pursue a balanced budget and keep inflation low.
If voters are well aware of parties’ ideological principles and their pursued policy goals, a
condition which generally holds in advanced democracies, the voters should expect parties to
comply with their revealed principles and policy preferences (Tufte 1978).

2.3.1.2. Economic Voting
The other building block of my argument draws on the literature on economic voting.
The gist of economic voting theory is that records of government performance in managing
the national economy, most notably economic growth, before an election are crucial for
citizens’ decisions to support the party of the incumbent government (see Lewis-Beck and
Stegmaier 2000, 2007, for extensive reviews of the literature). While the early literature on
economic voting employs a simple mechanism of reward and punishment on the past
performance of the incumbent government (e.g., Fiorina 1978, 1981; Kernell 1978; Key
1966; Kramer 1971; Lewis-Beck 1988; Tufte 1978), there has been the development on the
economic voting literature. The recent development of economic voting theory is summarized

21

by the following three points (Lewis-Beck and Paldam 2000; Lewis-Beck and Stegmaier
2000, 2007). First of all, as many survey data sets are available, it has become possible for
political scientists to investigate whether voters choose candidates retrospectively (Norpoth
1996) or prospectively (Erikson et al. 2000; MacKuen et al. 1992). Both types of economic
voting are supported by empirical evidence, but Nadeau and Lewis-Beck (2001) find
evidence that voters in the United States cast ballots based on retrospective evaluations if the
incumbent president is running for reelection and they vote prospectively otherwise.
However, subsequent generations of studies also suggest that a retrospective evaluation
should sometimes be considered as a guide to predictions for the future performance of the
incumbent government (Keech 1995; Duch and Stevenson 2008). Secondly, there have been
debates on sociotropic-pocketbook economic voting, but empirical results generally support
the sociotropic hypothesis (Kinder and Kiewiet 1979, 1981; Kiewiet 1983). Lastly, crossnational research on economic voting has been conducted extensively during the past two
decades. Although early studies investigate economic voting in single-country including the
United States and Western European democracies (Alverez and Nagler 1995; Kiewiet 1983;
Kinder et al. 1989; Nadeau and Lewis-Beck 2000, 2001), much scholarly work has been done
on the topics involving economic voting in the cross-national context recently (Anderson
1995, 2000; Duch and Stevenson 2008; Powell and Whitten 1993; Samuels 2004; Whitten
and Palmer 1999). This growing body of literature on economic voting from the comparative
perspective also provides us with valuable insights of how various factors, such as
institutional features (Duch and Stevenson 2008; Powell and Whitten 1993) and international
economic conditions (Kayser and Peress 2012), affect citizens’ economic perceptions and
voting behavior.

22

2.3.2. Inequality Changes and the Poor’s Vote Choice
2.3.2.1. Inequality Changes and Different Expectations According to Government
Partisanship
The ideas and findings of the two theories, -- partisan effects on economic outcomes
and economic voting, -- are widely accepted in comparative studies of elections. The
argument of this chapter draws on the insights from the numerous research studies related to
these two theories. The argument here is that poor voters support leftist parties at the ballot
box only if a left-wing government successfully pursue economic well-being of the poor by
reducing income inequality. On the other hand, if the leftist incumbent fails to perform well
on that matter, then the effect of relative income on the poor’s voting decision is likely to
decrease. The left is usually perceived as representing the preferences of the lower-income
voters, since leftist parties are committed to prioritizing redistribution and equality.
Therefore, the poor are expected to support leftist parties given that their economic
preferences are commensurate with the economic goals that the left pursues. However, the
poor have an incentive to do so only when the leftist government produces outcomes that
reflect progress in advancing the economic well-being of the poor. In other words, the poor
are less incentivized to support leftist parties if their economic well-being deteriorates under
their watch. As a result, a voter’s income is no longer a relevant voting cue for the poor if
their economic condition worsens or stagnates with leftist parties at the helm.
How do poor voters evaluate the performance of a leftist government in advancing
their economic well-being? A general consensus in the economic voting literature is that a
GDP growth rate in an election year, which indicates the state of the national economy or the
incumbent’s competence in managing the national economy, is the most important source for
economic evaluations of voters. However, “it makes little sense for voters at the bottom of the
income distribution, experiencing relatively flat (or declining) income growth to reward

23

incumbents for national economic growth” (Linn and Nagler 2014: 4). Rising economic
inequality is detrimental especially to the poor (Alesina et al. 2004) and they can be
disadvantaged even in prosperous economies. In other words, the poor should base their
evaluation of the incumbent less on indicators of overall national prosperity and more on the
extent to which the incumbent enhances or undermines their own economic conditions. This
argument is particularly relevant in an era of rising economic inequality in advanced
democracies. The rise of economic inequality in advanced democracies since the 1980s is
well recognized. According to OECD (2011), the average Gini coefficient in OECD countries
increased by almost 10% from the mid-1980s (0.29) to the late 2000s (0.316). During this
period, the pace of income growth varied across income groups and real disposable
household income of the poor grew much more slowly than that of the rich (OECD 2011). In
many countries, the poor’s average income stagnates or even declines when the national
mean income increases. For example, the real disposable household income increased by
0.3 % each year from the mid-1980s to the late 2000s in Japan. Yet those in the bottom decile
had suffered income loss by 0.5 % per year during the same period (OECD 2011). Similarly,
increasing GDP per capita has not benefitted those at the bottom of the income ladder for
years in the United States (Stiglitz 2012). The average income of citizens in the top income
quintile has increased by 1.1% per year; those in the lower end of the income distribution
have experienced decreases in their income by 0.33% on average for the last three decades in
the United States (Linn and Nagler 2014). This significant disparity implies that indicators
measuring overall national prosperity, such as the annual GDP grow rate, may not be very
informative to the poor in evaluating how much the current government enhances their
welfare. If the growth of national wealth does not benefit low-income citizens, why would
they decide their vote choices based on the government’s performance in terms of the
aggregate wealth? Instead, it makes more sense that the poor would look to an indicator that

24

is more directly relevant to their economic well-being.
Previous studies have questioned the ability of a voter, especially a low-information
voter, to make electoral choices that enhance her welfare (e.g., Delli Carpini and Keeter
1996). In other words, many voters do not have sufficient knowledge about pertinent political
facts to make informed voting decisions. Even though ordinary people who are uninterested
in politics and government affairs are not very knowledgeable when they are asked about
questions on political facts in surveys, however, this does not mean that they make their
voting decisions without relevant information. Prior and Lupia (2008) demonstrate that even
uninterested people lacking knowledge in politics can make informed choices if much is at
stake in their decision and there is sufficient time to gather political information for their
decision. In other words, such poor performance of voters on survey-based pop quizzes is not
“sufficient to infer a general lack of capability at politically charged moments, such as
elections” (Prior and Lupia 2008). Another group of research on economic voting also
suggests that citizens are fairly competent to evaluate macroeconomic conditions when
making vote choice (Duch et al. 2000; Duch and Stevenson 2011; MacKuen et al. 1992).
Given that poor voters can make their decision at the ballot box in an informed way, I
highlight the importance of income inequality in the poor’s voting decision, since inequality
is more directly relevant to the poor’s welfare than other economic indicators that represent
overall prosperity. I further posit that a change in income inequality between elections, rather
than the inequality level itself, better captures performance of the government in addressing
economic disparity. As the levels of economic outcomes are generally results of long-term
structural factors, such as welfare state system, labor market institutions, and the distribution
of power among organized interest groups, it is hard to attribute the level itself to the
incumbent’s policy or competence (Becher and Donnelly 2013). In this sense, changes in
economic outcomes will provide voters with better information on the performance of the

25

current government. From the simple retrospective perspective based on the rewardspunishments mechanism, unsatisfactory performance of the leftist government in dealing
with inequality leads lower-income voters to punish the incumbent in the upcoming election.
The economic outcomes that result from the way the leftist government addresses the poor’s
interests frustrate its core support group, and, consequently, motivate the poor to express their
disappointment at the ballot box. From the prospective perspective, they are also more likely
to abandon leftist parties in government if the left-wing government fails to ameliorate
inequality, since the disappointing past performance of the left incumbent can signal the
inability or irresoluteness of the party of the incumbent government to improve the poor’s
economic well-being in the future.
The question then arises whether income-based voting is also associated with
government performance in addressing economic inequality under centrist or rightist
incumbent. I speculate that inequality changes will not have the moderation effect on voting
under non-leftist government because voters already expect that center or rightist parties
place less or little emphasis on reducing inequality. Given that the poor recognize that
improving inequality is not high on the priority list of a centrist or rightist government, they
are less likely to evaluate government performance in accord with inequality changes. I
therefore expect that the prevalence of income-based voting depends on inequality changes
only under leftist governments.

2.3.2.2. Sticking to or Defecting from the Left? The Poor’s Choice at the Ballot Box
If poor voters abandon the incumbent leftist party in government to blame the
incumbent for the subpar performance of leveling inequality, they have a couple of
alternatives at the ballot box: they may vote for another left-wing party among the opposition
parties or they may vote for a non-leftist party. For example, with respect to the first option,

26

the poor in Netherlands can decide to support the Socialist Party (Socialistische Partij) if they
are not happy with the government led by the Labor Party (PdvA). This option is available
only in multiparty democracies with proportional representation (PR) electoral system, where
there are more than one electorally viable leftist parties and at least one of them remains in
opposition. As for the second option, the poor are more likely to choose a non-leftist party
when they make choices in the presence of cross-cutting economic and non-economic
cleavages (Roemer 1998). For instance, a religious poor voter who seeks an alternative to the
left in government may rely on his or her religious orientation if the leftist incumbent ignores
the poor’s economic interests.
While the poor may choose the first option if certain conditions hold (i.e., multiparty
democracy under PR system), I argue that the poor are more likely to defect to the non-left.
The first reason is that when a left-wing government is formed, most electorally viable leftwing parties participate in government to form a majority (e.g., Social Democrats-Social
Liberal-Socialist People’s Party coalition, 2011-2014 in Denmark; SPD-Alliance ‘90/Greens
coalition, 1998-2005 in Germany; Labor-Socialist Left-Centre Party coalition 2005-2013 in
Norway; and Social Democrats-Green Party, 2014-present in Sweden). In many cases, even if
the largest left-wing party forms a minority government, the minority government is
sustained with support of minor leftist parties in the legislature (e.g., Socialist Party minority
government supported by Left Bloc, the Portuguese Communist Party, and the Greens, 2015present in Portugal). A coalition or minority government is very common in multiparty
democracies under the PR electoral system, and thus poor voters usually have a very limited
menu of left-wing parties in opposition from which to choose at the ballot box.3
More importantly, when the parties in government are evaluated poorly by voters due

3

A single-party majority government is rare in advanced democracies adopting the PR system. In
fact, there is only one example of a left-wing single-party majority government since 2000, the
Socialist Party (PS) government of Portugal formed in 2005.
27

to their performance in managing the national economy, it also undermines the reputations of
opposition parties that are ideologically close to the parties in government (Williams and
Whitten 2015). In other words, the poor performance of a left-wing government in reducing
inequality hurts the ideological brand of the left in general as well as the electoral fortunes of
the leftist parties in government. Consequently, left parties in opposition are likely to lose
votes in elections even when they are not responsible for the poor performance of the left
incumbent. Furthermore, such “spatial contagion” effects tend to be stronger when the
incumbent government is a coalition or minority government (Williams and Whitten 2015).
Therefore, poor voters are less likely to choose another left-wing challenger when the poor
voters are disappointed with an incumbent left-wing government. On the other hand, electoral
competitions in advanced democracies rarely revolve around a single issue, and parties
usually tackle multiple issues simultaneously in order to draw support from voters. This
presence of multiple cross-cutting cleavages provides voters with alternative voting cues that
the voters can rely on other than redistributive preferences. If the left fails to perform well on
the inequality issue, then a poor voter is likely to vote for an economically right-wing party,
relying on another voting cue related to a non-economic issue. These considerations yield the
main hypothesis about the moderation effect of inequality changes on income-based voting:

Hypothesis 2-1: The effect of income on voting should, all other things being equal, decrease
with rising income inequality under a left-wing government.

2.4. Data, Measurements, and Model Specification
2.4.1. Cases and Data
The scope of this study is voting behavior in national elections held in advanced
democracies. I excluded nascent democracies where voters are less likely to have clear beliefs

28

about parties’ competence and policy positions due to the short period of democratic
experience (Keefer 2007; Tavits 2006; Tucker 2006).4 It is also hard in such cases for voters
to identify which party can best represent their self-interest; hence voter-party programmatic
linkages, which are generally strong and stable in advanced democracies, are expected to be
weak in nascent democracies (Ezrow et al. 2014; McAllister and White 2007). To analyze
micro-level vote choice, I used surveys from the Comparative Study of Electoral Systems
(CSES) data set (CSES 2003, 2007, 2013). The CSES is suitable for the purpose of this study
because the three waves of multinational post-election surveys contain pertinent information
about respondents’ income levels and vote choices in national elections from 1996 to 2011.5
As the main hypothesis posits that the moderation effect of inequality changes differs
depending on government partisanship, it is also necessary to classify cases into elections
under leftist and non-leftist incumbents. I split elections into two groups according to
incumbent partisanship based on the Comparative Welfare States Data Set (CWS, Brady et al.
2014). This dataset provides information about partisan compositions of the government in
22 advanced democracies between 1960 and 2011. The coding procedure should be
straightforward if the government is composed of a single party or multiple parties in the
same partisan group. When the government comprises multiple parties from different partisan
groups, the procedure becomes more complicated. In such cases, I classified the executive as

4

Tucker (2006) also points out a lack of consensus on how parties in new democracies are classified
among scholars.
5
Most advanced democracies have adopted parliamentary system except the United States
(presidential system) and France (semi-presidential system). It is best to look at answers of
respondents to a question asking their vote choices in legislative elections for parliamentary
democracies, but this is less clear for the two countries with the other forms of government. In
principle, it is more reasonable to focus on elections determining the composition of the executive
branch. Therefore, I choose voting records for presidential elections for the United States. This choice
is also reasonable in that economic voting in the United States is generally stronger in presidential
elections than congressional elections (Erikson 1990; Marra and Ostrom 1989). For French elections,
the second wave of the CSES survey provides voting records only for the presidential election in
2002; on the other hand, voting records only for legislative elections are available in the third wave.
This data availability practically limits options for French elections.
29

leftist, centrist, or rightist, according to which group of parties (i.e., left-wing, center, or rightwing) has more than 50% of the legislative seats held by the government or has a plurality
and more than 40% of the legislative seats held by incumbent parties. Yet if both leftist and
rightist parties have more than 40% of the legislative seats held by the government, then the
government is coded as a centrist one.

2.4.2. Dependent Variable
The dependent variable is the economic ideology of a party voted by each respondent.
Following Huber and Stanig (2009, 2011), I pulled information about the positions of
political parties on the tax/redistribution policy dimension from the Party Policy in the
Modern Democracies dataset (PPMD, Benoit and Laver 2006). The PPMD provides
information on parties’ positions on the tax/redistribution issue based on evaluations of
country experts on each party.6 The value of this measure ranges from 1 (“raising taxes to
increase public services”) to 20 (“cutting public services to cut taxes”). I created the
dependent variable by matching voting records of respondents from the CSES to positions of
parties on the tax/redistribution policy issue from the PPMD. Combining these three datasets,
the CSES, PPMD, and CWS, provides the original sample covering 55 elections held from
1996 to 2011 in 21 advanced democracies.7

6

The numbers of country experts responding to the survey vary from 15 (Australia) to 166 (the
United States) and these large numbers of responses from the experts provide reliability of this
measure (Benoit and Laver 2006).
7
See the Appendix to this chapter for the list of elections. The list of countries includes Australia,
Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, the
Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and
the United States. The Belgian election in 2003 is dropped from subsequent analyses since the
Belgian survey does not ask respondents’ income levels in that year.
30

2.4.3. Independent Variables
The primary goal here is to examine how the direction and magnitude of income
effects change, depending on inequality changes and government partisanship. In order to test
the moderation effect of changes in inequality, I considered three main independent variables:
income (poor), changes in economic inequality (inequality change), and the interaction
between the two key independent variables.
I first derived respondents’ self-reported relative positions on the income distribution
from the CSES to measure the income level, and then, following Huber and Stanig (2011),
defined respondents in the two lowest quintiles as the poor. With respect to inequality
change, there are two options to measure performance of leftist governments in improving
income inequality. The first option is to measure inequality change with pre-tax/transfer
income inequality. Recognizing the extent to which pre-tax/transfer income inequality has
been ameliorated through government policy on tax and spending is not easy, however,
because pre-tax/transfer income inequality is relatively harder for ordinary citizens to observe
(Finseraas 2009a). More importantly, pre-tax/transfer income inequality already reflects the
extent to which governments implement policies (e.g., labor market policy) favorable to the
poor (Huber and Stephens 2012; Rueda 2008). For these reasons, changes in post-tax/transfer
income inequality between elections appears to be the better measure for the poor to evaluate
the performance of the leftist government.8 To measure inequality change, I drew Gini
coefficients from the Luxembourg Income Study (LIS) and the Standardized World Income
Inequality Database (SWIID, Solt 2009). The LIS dataset has been praised for its reliability
and comparability, and was therefore a primary source for measuring economic inequality.
Yet there are many missing years in the LIS database, so I utilized the SWIID as a secondary

8

Post-tax/transfer income inequality also captures inequality in disposal income among the
population which ultimately matters for citizens’ welfare (OECD 2012).
31

source to pull Gini coefficients to fill in missing observations. The SWIID is constructed to
create inequality measures ensuring comparability of inequality, and also covers the widest
range of country-year observations among existing inequality datasets. Specifically, I first
pulled Gini coefficients from the LIS wherever they are available, and filled in missing
observations using the SWIID.9 Gini coefficients in a year when incumbent assumed office
are then subtracted from Gini coefficients in the previous calendar year before the election for
election year studied to measure how well an incumbent government deals with income
inequality.10 I also constructed an interaction term between poor and inequality change in
order to examine the moderation effect of inequality change on income-based voting as the
main hypothesis suggests. Finally, I control for a set of micro-level variables, such as age,
gender, levels of education, religiosity, union membership, employment status (i.e.,
unemployed and self-employed), and self-placed ideological position, all of which are argued
to affect vote choice.11

9

I believe that measuring inequality levels for each year based on the LIS/SWIID is the best available
option in terms of comparability of inequality measures and coverage, yet some might fear that the
Gini coefficients from the SWIID is not as reliable as the LIS to capture changes in inequality. To test
the accuracy and reliability of the inequality measures constructed based on the LIS/SWIID, I
calculate the correlation coefficient between the inequality measure used and post-tax/transfer Gini
index for 21 countries from OECD Statistics (http://stats.oecd.org/index.aspx?queryid=66670#) for 20
years (1992-2011). Gini coefficients from OECD Statistics are relatively sparse as compared to those
from the LIS/SWIID, available only for 221 country-years. I find that the correlation between the
LIS/SWIID and the OECD Gini coefficients is extremely high (ρ=0.97), so it is reasonable to
conclude that the LIS/SWIID inequality measure ensures sufficient reliability enough to make valid
inferences on movements in inequality, though this measure may be imperfect.
10
I use Gini coefficients in the previous calendar year before the election, instead of figures in the
election year, to allow voters enough time to internalize the state of the national economy.
11
Details regarding the control variables are presented in the Chapter 1 Appendix. While these
controls are available in surveys for most countries, one or more question items are not available for
18 election surveys. An additional country-election case (Sweden in 2006) was dropped from the
analysis with full controls, since no respondent answered that he or she attends religious services
“once a week/more than once a week” in that survey. The percentage of religious respondents are
relatively lower in the previous surveys in Swedish elections as compared to the other countries in the
CSES, yet the fact that there is no respondent with high religiosity casts doubt on the reliability of the
survey. I therefore dropped the Swedish election in 2006, but including the case in the analysis does
not change the result in terms of substantive conclusions. As a result, 35 elections from 18 countries
are used in the main analysis with full controls.
32

2.4.4. Model Specification and Estimation Strategy
As for the estimation strategy, the data structure is hierarchical with the income level
and other factors specific to individual respondents and the change in inequality specific to
each country-election. In other words, the income level is an individual-level variable and the
change in income inequality is a country-year level variable, which is invariant across
observations within the same country-election. Ignoring this data structure and intraclass
correlation between observations in the same clusters may result in smaller standard errors of
estimates and overstated confidence over them, especially with large group sizes like national
election survey (Angrist and Pischke 2009; Moulton 1986, 1990; Steenbergen and Jones
2002).12 To address this issue, I specify a linear mixed regression model, which models a
structure of a covariance matrix directly by introducing cluster-specific unobserved
heterogeneity across elections (Steenbergen and Jones 2002; Gelman and Hill 2007). While
estimated coefficients form the statistical model should be unbiased, they can still cause
problems in drawing inferences if the variance-covariance matrix is not correctly modeled. To
account for this possibility, I also employ a post-estimation adjustment regarding the
clustering data structure by estimating more generalized forms of the Huber-White robust
standard error, which is consistent even if the correlation structure is misspecified (Huber
1967; White 1980). Specifically, the statistical model is presented as follows:

Economic	 Ideology	 of	 Party	 Voted 	
	

	
	

12

Failing to take the hierarchical data structure with large group sizes into account can lead to
significantly deflated standard errors even with small intraclass correlation (Angrist and Pischke
2009).
33

From the preceding discussion and previous studies, the signs of coefficients on the
key independent variables can be predicted. The main goal here is to estimate the coefficient
on the cross-level interaction variable ( ), which represents the moderation effect of
inequality change. The coefficient on poor (

captures the difference in the economic

ideologies of parties supported by the poor and the wealthy when there is no change in
income inequality between elections. Therefore,

should be negative. I also expect,

however, that the income effect diminishes if economic inequality deepens when the left is
governing, and thus

should be positive in the case of a leftist incumbent.

2.5. Estimation Results
2.5.1. Main Analysis
Table 2.1 reports the estimation results of the multilevel linear regression models.
First, I estimate coefficients for the three main independent variables without controls. The
first column of Table 2.1 shows that the effect of poor does not depend on inequality change
if government partisanship is not taken into account. The coefficient on the interaction term is
not only very small, but also imprecisely estimated. The insignificance of the moderation
effect is expected, since this model does not take into account the poor’s different
expectations about economic outcomes depending on government partisanship. However, a
strong interaction effect supporting the main hypothesis is detected when cases are split
according to the partisan composition of the incumbent government. The effect of poor
decreases with changes in inequality when leftist parties are in office and this effect is
statistically significant (p < .05).
The last three columns of Table 2.1 report estimation results of models with full
controls in 35 elections where all independent variables are available. The coefficient
estimates from the model with full controls are similar to the results from the basic models.

34

The coefficient for the interaction term actually becomes larger in magnitude and this is also
precisely estimated (p < .01) under a left-wing incumbent and the sign of coefficient also
accords with the main hypothesis. Substantively, an increase in Gini coefficient by 1 is
associated with a decrease in the effect of poor on the economic ideology of the party voted
for by 0.147 when leftist parties are in power. This moderation effect is fairly strong,
especially if the varying effect of poor depending on inequality change is compared to the
coefficients on the other covariates. Union membership appears to be the strongest predictor
among factors in terms of the magnitude of the effect (-0.701), leaving alone ideological
position. If the Gini coefficient decreases by 2 under a left-wing government, the point
estimate of the effect of poor is -0.763, in which the absolute value is larger than that of the
union membership effect.
The moderation effect of inequality change is also strong enough to eliminate the
effect of poor on voting. Figure 2.1 plots the different moderation effects of inequality
change under alternative partisan compositions of government. The first graph shows that the
effect of poor consistently decreases commensurate with the extent to which inequality
deepens under leftist government. Income-based voting virtually diminishes and parties voted
for by the poor are not meaningfully different, in terms of economic ideology, from those
supported by the rich when the Gini coefficient increases by 2 for the term of a leftist
government.
Figure 2.2 shows the predicted economic ideologies of parties voted for by low- and
high-income voters depending on inequality change. The two graphs on the right-hand side

35

Table 2.1. Estimation Results of Hierarchical Linear Models
DV: the
Economic
Ideology of
the Party a
Respondent
Votes for
Constant
Poor
Inequality
Change
Poor	
Inequality
Inequality Level

Model 1:
w/o
Controls
(All)

Model 2:
w/o
Controls
(Leftist
Govt.)

Model 3:
w/o
Controls
(NonLeftist
Govt.)

Model 5:
w/
Full
Controls
(Leftist
Govt.)
5.839**
(0.614)
-0.469**
(0.066)
0.105
(0.170)
0.147**
(0.021)
0.021
(0.043)
0.189
(0.259)
0.013
(0.240)
0.056
(0.313)
-0.110
(0.217)
0.013
(0.193)
0.032
(0.113)
(omitted)
0.196**
(0.058)
0.242
(0.149)
0.198
(0.266)
-0.701**
(0.099)
-0.417*
(0.172)
0.554**
(0.171)
1.007**
(0.094)
0.665

Model 6:
w/
Full
Controls
(NonLeftist
Govt.)
6.571**
(0.419)
-0.561**
(0.082)
-0.030
(0.196)
-0.339**
(0.112)
-0.024
(0.054)
0.185
(0.131)
-0.132
(0.196)
-0.059
(0.185)
-0.161
(0.185)
-0.110
(0.153)
-0.056
(0.125)
(omitted)
0.074
(0.089)
0.021
(0.094)
0.057
(0.175)
-0.751**
(0.071)
-0.348
(0.185)
0.618**
(0.116)
0.928**
(0.062)
0.492

11.042**
(0.110)
-0.572**
(0.094)
0.031
(0.127)
-0.055
(0.096)
0.039
(0.025)

11.013**
(0.169)
-0.643**
(0.073)
0.093
(0.169)
0.126*
(0.052)
0.025
(0.030)

11.059**
(0.162)
-0.452**
(0.130)
-0.078
(0.190)
-0.408**
(0.154)
0.050
(0.043)

0.655

0.435

0.756

6.228**
(0.349)
-0.550**
(0.061)
0.051
(0.140)
-0.008
(0.091)
-0.003
(0.033)
0.191
(0.121)
--0.057
(0.153)
-0.005
(0.164)
-0.142
(0.140)
-0.055
(0.117)
-0.013
(0.086)
(omitted)
0.128*
(0.060)
0.107
(0.082)
0.120
(0.150)
-0.725**
(0.057)
-0.381**
(0.128)
0.595**
(0.100)
0.962**
(0.052)
0.597

18.757

18.225

19.055

12.204

12.076

12.259

54

22

32

35

14

21

Religious
Age 18-24
Age 25-34
Age 35-44
Age 45-54
Age 55-64
Age 65 / Older
Male
Education
(Secondary)
Education
(University)
Union
Membership
Unemployed
Self-employed
Ideology

N (Countryelections)
N

Model 4:
w/
Full
Controls
(All)

65,693

24,675

41,018

30,916

13,029

17,887

* p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

36

Figure 2.1. Effects of Poor Conditional on Inequality Change under Leftist and Non-leftist
Government

-.5
-1
-2

-1.5

Effect of Poor

0

.5

Leftist Government

-2

-1

0

1

2

3

2

3

Changes in Inequality

-.5
-1
-2

-1.5

Effect of Poor

0

.5

Non-leftist Government

-2

-1

0

1

Changes in Inequality

Based on Model 5 and 6 of Table 2.1. Dashed lines are 95 percent confidence intervals.

37

Figure 2.2. Predicted Economic Ideology of Parties Voted for by the Poor and Non-poor
under Leftist and Non-leftist Government
Non-poor, Leftist Govt.

9

9

E c o n o m ic Id e o lo g y o f P a rtie s V o te d F o r
10
11

E c o n o m ic Id e o lo g y o f P a rtie s V o te d F o r
10
11

12

12

Poor, Leftist Govt.

0

1

2

3

-2

-1

0

1

Changes in Inequality

Changes in Inequality

Poor, Non-leftist Govt.

Non-poor, Non-leftist Govt.

2

3

2

3

E c o n o m ic Id e o lo g y o f P a rtie s V o te d F o r
10
11
9

9

E c o n o m ic Id e o lo g y o f P a rtie s V o te d F o r
10
11

12

-1

12

-2

-2

-1

0

1

2

3

-2

Changes in Inequality

-1

0

1

Changes in Inequality

Based on Model 5 and 6 of Table 2.1. Dashed lines are 90 percent confidence intervals.

show that high-income voters are not responsive to inequality changes regardless of
government partisanship. The poor’s voting, on the other hand, is not only strongly affected
by inequality change, but also the direction of such effects also clearly depends on the
partisan composition of government. Overall, the coefficient estimates reported in Table 2.1
provide strong evidence that support the main hypothesis. Indeed, the poor are less likely to
38

vote for the left if a leftist government fails to reduce economic inequality. These results
indicate that increasing inequality depresses income-based voting under leftist governments.
Additional noteworthy, and unexpected, findings of Model 3 and 6 are that the effect
of poor on left-wing voting strongly depends on inequality change and that the effect of poor
consistently increases with inequality change under a non-leftist government. The absolute
value of the coefficient on the interaction term is even larger in the opposite direction, and
political parties chosen by the poor at the ballot box are more economically conservative as
inequality decreases under a non-leftist government. Why does the performance of centrist or
right-wing government in addressing income inequality also affect the poor’s evaluation of
the incumbent government and their vote choice? One possible explanation is that rising
inequality under a non-leftist incumbent motivates the poor to want parties that may advance
their economic welfare (i.e., the left) to be in power; on the flip side, they may have less
incentive to support left-wing parties if the non-leftist incumbent performs well in reducing
inequality.
The results of the empirical analysis show that the prevalence of income-based
voting depends on changes in inequality, yet it is not clear whether the results are produced
through the suggested causal mechanism. The reason is that this empirical analysis does not
test the poor’s voting for or against incumbents, which is the test more directly linked to the
theoretical intuition behind the hypothesis, but rather the poor’s voting for more or less proredistribution parties. For this reason, it is necessary to conduct an additional analysis to more
directly test the causal mechanism, which is whether the poor actually are more likely to vote
for opposition parties when inequality rises. To this end, I examine whether the suggested
causal mechanism – the poor’s economic voting and different expectations on economic
outcomes conditional on government partisanship – drives the moderation effects of
inequality change in the estimation results.

39

2.5.2. Testing Causal Mechanism: Do the Poor Not Vote for Parties in Government
When Inequality Rises?
I construct an additional statistical model to examine the poor’s voting for incumbents
following the common practice of the economic voting literature. To measure the poor’s
voting for incumbents, parties in government are identified from the Database of Political
Institutions (DPI) by Beck et al. (2001). The statistical model includes all the control
variables used in the previous section and also controls for annual GDP growth rates in the
election years.13 I estimate coefficients of the following equation to examine whether the
probability of voting for incumbent parties by the poor decreases with rising inequality:

Pr	 Voting	 for	 Incumbent

1 	

	Φ

	

The causal mechanism driving the moderation effect of inequality change suggests
that the poor are less likely to vote for incumbents as inequality rises if they evaluate the
government based on its performance in improving inequality, and thus

of the equation

should be negative.
The estimation results presented in the first column of Table 2.2 provide support for
the intuition behind the main hypothesis. The coefficient estimate on inequality change under
left-wing governments reach a conventional level of statistical significance (p < 0.05) and the
direction of the coefficient is negative, which indicates that the poor are less likely to vote for
incumbent parties when the left is in power as inequality rises. Figure 2.3 plots the
probability of the poor’s voting for incumbent parties at different levels of inequality change.

13

Annual GDP growth rates are pulled from OECD Statistics
(https://stats.oecd.org/index.aspx?queryid=350#).
40

Table 2.2. Estimation Results of Hierarchical Probit Models
DV: Voting for
Incumbent
Constant
Inequality Change
Inequality Level
GDP Growth
Religious
Age 18-24
Age 25-34
Age 35-44
Age 45-54
Age 55-64
Age 65 / Older
Male
Education
(Secondary)
Education
(University)
Union Membership
Unemployed
Self-employed
Ideology
Intraclass Correlation
(ρ)
N (Country-elections)
N

Model 7: Poor, Leftist Govt.

Model 8: Poor, Non-leftist Govt.

0.703**
(0.176)
-0.117*
(0.058)
0.015
(0.012)
0.034
(0.033)
-0.330**
(0.121)
-0.095
(0.095)
0.013
(0.049)
-0.063
(0.061)
-0.116
(0.072)
-0.081
(0.063)
(omitted)
-0.065
(0.036)
-0.102
(0.060)
-0.133
(0.136)
0.162**
(0.047)
0.073
(0.097)
-0.249**
(0.085)
-0.145**
(0.031)
0.029

-0.799*
(0.401)
-0.027
(0.236)
-0.026
(0.045)
-0.033
(0.075)
0.110
(0.082)
-0.284**
(0.107)
-0.255*
(0.115)
-0.255*
(0.123)
-0.115
(0.090)
-0.166**
(0.050)
(omitted)
-0.105*
(0.052)
0.003
(0.049)
-0.048
(0.103)
-0.134**
(0.048)
-0.210**
(0.079)
0.168
(0.104)
0.208**
(0.035)
0.284

14

21

4,681

6,326

* p<0.05, ** p<0.01 (two-tailed); Cluster bootstrap standard errors are given in parentheses.

41

Figure 2.3. Predicted Probabilities of the Poor’s Voting for Incumbent Conditional on
Inequality Change under Leftist and Non-leftist Government

.5
.4
.2

.3

Predicted Prob.

.6

.7

Poor, Leftist Govt.

-2

-1

0

1

2

3

2

3

Changes in Inequality

.5
.4
.2

.3

Predicted Prob.

.6

.7

Poor, Non-leftist Govt.

-2

-1

0

1

Changes in Inequality

Based on Model 7 and 8 of Table 2.2. Dashed lines are 95 percent confidence intervals.

42

When a left-wing government reduces income inequality by 2, the estimated probability that
a low-income voter supports an incumbent party is 0.54 approximately, yet the probability of
the poor’s voting for an incumbent decreases up to 0.31 as inequality change approaches its
highest level. On the other hand, the poor are also less likely to support incumbent parties
when inequality grows under non-leftist governments, yet the estimation results show that the
coefficient on inequality change is not precisely estimated and the coefficient is also
negligible in terms of its magnitude. These results indicate that while the moderation effects
of inequality change exist under non-leftist governments, such effects are not produced by the
poor’s economic voting, but rather by another mechanism, or simply due to chance. However,
the results with the poor under leftist governments support the logic behind the moderation
effect of inequality change in the main analysis.

2.5.3. Robustness Tests and Additional Analysis
2.5.3.1. Robustness Tests
While the estimation results of the main analysis presented in Table 2.1 render
supporting evidence for the main hypothesis, it is also possible that these estimates may be
products of particular choices of measures for each variable and model specification. In fact,
there are many alternative ways to test the main hypothesis, and thus I conduct a host of
robustness checks to test whether the estimation results are sensitive to alternative measures
and model specifications.
I first utilize alternative sources for partisan composition of government, since the
results may not support my argument if elections are categorized based on different sources
of government partisanship. To entertain this possibility, I collect information about the
partisan composition of government from the Comparative Political Parties Data Set (CPPD,
Swank 2013) and the Comparative Political Data Set (CPDS, Armingeon et al. 2013) to

43

examine robustness of the original results. Table 2.3 reports the estimation results based on
those alternative classifications of government partisanship. While the absolute magnitude of
the moderation effect under left-wing government slightly decreases, the coefficient estimates
are still statistically significant (p < .01) and substantively meaningful in expected directions.
The moderation effect also remains statistically significant under non-leftist government as
the main analysis suggests.
I then redo the main analysis with an alternative measure of income (quintile). The
first three columns of Table 2.4 report estimation results of Model 15-17, and the results
remain unchanged. In addition, I introduce unobserved heterogeneity at the country level and
re-estimate the three-level hierarchical models. The last three columns of Table 2.4 present
the estimation results, which are almost identical to the results of the main analysis.
In the next step, I test the main hypothesis with a dichotomous measure of the
dependent variable (i.e., left-wing voting). I adopt a continuous measure in the original
analysis as the dependent variable, given that classifying parties into two groups is
intrinsically arbitrary. While this claim is valid, I also construct a binary variable to capture
respondents’ left-wing voting and examine whether using the alternative measure changes the
original results. Specifically, respondents voting for parties that are placed at less than 10.5
on the tax/spending dimension are regarded to vote for left-wing parties following Huber and
Stanig (2009, 2011). The estimation results presented in Table 2.5 still support the main
hypothesis. The coefficient estimate on the interaction term for the model under a leftist
incumbent reach a conventional level of statistical significance (p < .01), and the sign of the
coefficient also accords with the theoretical expectation. To facilitate substantive
interpretation of the mediating effects, I estimated the effect of poor on predicted
probabilities of left-wing voting depending on inequality change under alternative
government partisanship based on coefficients from the estimation results. Figure 2.4 plots

44

Table 2.3. Robustness Checks 1: Alternative Measures of Government Partisanship
DV: the Economic
Ideology of the
Party a
Respondent Votes
for
Constant
Poor
Inequality Change
Poor	
Inequality
Inequality Level
Religious
Age 18-24
Age 25-34
Age 35-44
Age 45-54
Age 55-64
Age 65 / Older
Male
Education
(Secondary)
Education
(University)
Union
Membership
Unemployed
Self-employed
Ideology

N (Countryelections)
N

Model 9:
CPPD
(Leftist)

Model 10:
CPPD
(Non-Leftist)

Model 11:
CPDS
(Leftist)

Model 12:
CPDS
(Non-Leftist)

5.960**
(0.703)
-0.426**
(0.070)
0.115
(0.187)
0.141**
(0.019)
0.028
(0.051)
0.340
(0.252)
0.207
(0.209)
0.160
(0.321)
-0.012
(0.219)
0.134
(0.164)
0.038
(0.112)
(omitted)
0.139*
(0.061)
0.199
(0.168)
0.285
(0.298)
-0.723**
(0.113)
-0.406*
(0.187)
0.564**
(0.174)
0.960**
(0.097)
0.737

6.415**
(0.387)
-0.596**
(0.071)
-0.019
(0.185)
-0.280**
(0.106)
-0.025
(0.041)
0.127
(0.131)
-0.244
(0.197)
-0.095
(0.175)
-0.213
(0.172)
-0.159
(0.150)
-0.044
(0.114)
(omitted)
0.118
(0.086)
0.050
(0.088)
0.010
(0.166)
-0.731**
(0.066)
-0.380*
(0.183)
0.603**
(0.121)
0.962**
(0.061)
0.465

6.091**
(0.769)
-0.435**
(0.072)
0.140
(0.161)
0.141**
(0.020)
-0.006
(0.050)
0.485*
(0.217)
0.192
(0.228)
0.184
(0.333)
0.050
(0.212)
0.187
(0.157)
0.085
(0.100)
(omitted)
0.139*
(0.065)
0.178
(0.174)
0.252
(0.310)
-0.797**
(0.089)
-0.468*
(0.183)
0.579**
(0.179)
0.964**
(0.104)
0.555

6.337**
(0.378)
-0.574**
(0.077)
-0.093
(0.190)
-0.272*
(0.107)
0.004
(0.044)
0.092
(0.130)
-0.198
(0.193)
-0.093
(0.169)
-0.232
(0.169)
-0.171
(0.147)
-0.062
(0.113)
(omitted)
0.119
(0.083)
0.072
(0.086)
0.038
(0.164)
-0.700**
(0.071)
-0.331
(0.176)
0.590**
(0.117)
0.960**
(0.060)
0.541

12.213
12

12.173
23

12.570
11

12.001
24

10,899

20,017

10,140

20,776

* p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

45

Table 2.4. Robustness Checks 2: Alternative Measure of Income (Quintile) and Introducing
Country-level Heterogeneity
DV: the
Economic
Ideology of
the Party a
Respondent
Votes for
Constant

Model 13:
Alternativ
e Measure
of Income
(Quintile,
All)
5.356**
(0.345)
0.272**
(0.028)
-0.027
(0.279)
0.022
(0.044)
-0.003
(0.033)
0.206**
(0.122)
-0.095
(0.149)
-0.053
(0.161)
-0.229
(0.138)
-0.159
(0.116)
-0.078
(0.084)
(omitted)
0.108
(0.061)
0.057
(0.083)
-0.018
(0.149)
-0.747**
(0.057)
-0.293*
(0.127)
0.578**
(0.100)
0.954**
(0.052)

Model 14:
Alternativ
e Measure
of Income
(Quintile,
Leftist)
5.055**
(0.595)
0.240**
(0.033)
0.294
(0.193)
-0.045**
(0.017)
0.022
(0.044)
0.207
(0.257)
0.004
(0.238)
0.020
(0.310)
-0.183
(0.214)
-0.075
(0.193)
-0.029
(0.113)
(omitted)
0.178**
(0.062)
0.206
(0.151)
0.096
(0.269)
-0.709**
(0.098)
-0.338
(0.180)
0.543**
(0.171)
1.000**
(0.094)

Model 15:
Alternativ
e Measure
of Income
(Quintile,
Non-leftist)
5.734**
(0.404)
0.267**
(0.034)
-0.678*
(0.288)
0.168**
(0.047)
-0.026
(0.054)
0.197
(0.133)
-0.199
(0.182)
-0.118
(0.181)
-0.256
(0.182)
-0.229
(0.149)
-0.122
(0.123)
(omitted)
0.050
(0.088)
-0.038
(0.093)
-0.103
(0.169)
-0.790**
(0.074)
-0.252
(0.175)
0.592**
(0.111)
0.919**
(0.062)

Model 18:
ThreeLevel
Model
(Nonleftist)
6.219**
5.666**
6.585**
(0.434)
(0.805)
(0.437)
Poor
-0.550**
-0.470**
-0.561**
(0.066)
(0.064)
(0.082)
Inequality
0.060
0.155
0.044
Change
(0.100)
(0.120)
(0.187)
-0.008
0.147**
-0.338**
Poor	
(0.092)
(0.017)
(0.114)
Inequality
Inequality
-0.036
-0.012
-0.039
Level
(0.044)
(0.052)
(0.066)
Religious
0.191
0.191
0.184
(0.158)
(0.289)
(0.129)
Age 18-24
-0.061
0.015
-0.133
(0.184)
(0.322)
(0.209)
Age 25-34
-0.009
0.057
-0.061
(0.222)
(0.395)
(0.187)
Age 35-44
-0.146
-0.111
-0.165
(0.198)
(0.284)
(0.194)
Age 45-54
-0.059
0.012
-0.113
(0.151)
(0.205)
(0.158)
Age 55-64
-0.016
0.032
-0.057
(0.112)
(0.129)
(0.127)
Age 65 / Older
(omitted)
(omitted)
(omitted)
Male
0.128
0.196**
0.074
(0.067)
(0.063)
(0.109)
Education
0.110
0.224
0.024
(Secondary)
(0.089)
(0.179)
(0.096)
Education
0.122
0.190
0.060
(University)
(0.177)
(0.301)
(0.189)
Union
-0.725**
-0.699**
-0.751**
Membership
(0.075)
(0.112)
(0.085)
Unemployed
-0.380*
-0.419**
-0.345
(0.165)
(0.159)
(0.199)
Self0.595**
0.552**
0.619**
employed
(0.122)
(0.183)
(0.122)
Ideology
0.963**
1.007**
0.929**
(0.060)
(0.125)
(0.063)
0.420
0.702
0.231
0.592
0.670
0.483
0.141
0.061
0.245
12.148
12.038
12.180
12.204
12.077
12.259
N (Countries)
18
10
15
N (Elections)
35
14
21
35
14
21
N
30,916
13,029
17,887
30,916
13,029
17,887
* p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.
46

Model 16:
ThreeLevel
Model
(All)

Model 17:
ThreeLevel
Model
(Leftist)

Table 2.5. Robustness Checks 3: Alternative Measure of Dependent Variable (Dichotomous)
DV: : the Economic
Ideology of the
Party a Respondent
Votes for

Model A19:
Binary
Dependent
Variable (Leftwing Voting,
All)

Constant

1.736**
(0.152)
0.181**
(0.028)
0.027
(0.062)
0.009
(0.050)
-0.003
(0.018)
-0.106
(0.070)
0.025
(0.063)
0.026
(0.066)
0.070
(0.064)
0.037
(0.046)
0.020
(0.042)
(omitted)
-0.010
(0.026)
-0.079*
(0.032)
-0.118*
(0.054)
0.290**
(0.026)
0.176**
(0.042)
-0.300**
(0.036)
-0.352**
(0.024)
0.220

Poor
Inequality Change
Poor	

Inequality

Inequality Level
Religious
Age 18-24
Age 25-34
Age 35-44
Age 45-54
Age 55-64
Age 65 / Older
Male
Education
(Secondary)
Education
(University)
Union Membership
Unemployed
Self-employed
Ideology
Intraclass
Correlation (ρ)
N (Countryelections)
N

35

Model 20:
Binary
Dependent
Variable
(Left-wing
Voting,
Leftist)
1.970**
(0.279)
0.147**
(0.020)
0.005
(0.063)
-0.058**
(0.014)
-0.005
(0.021)
-0.172
(0.155)
-0.014
(0.111)
-0.033
(0.121)
0.021
(0.103)
-0.010
(0.087)
-0.042
(0.071)
(omitted)
-0.051
(0.034)
-0.145**
(0.051)
-0.162
(0.088)
0.290**
(0.043)
0.177*
(0.070)
-0.298**
(0.057)
-0.350**
(0.051)
0.118
14

30,916

13,029

Model 21:
Binary
Dependent
Variable
(Left-wing
Voting, Nonleftist)
1.555**
(0.212)
0.181**
(0.041)
0.042
(0.169)
0.168**
(0.065)
0.009
(0.047)
-0.067
(0.070)
0.062
(0.074)
0.071
(0.073)
0.106
(0.079)
0.074
(0.050)
0.065
(0.050)
(omitted)
0.024
(0.033)
-0.037
(0.036)
-0.084
(0.064)
0.296**
(0.032)
0.178**
(0.059)
-0.298**
(0.044)
-0.354**
(0.026)
0.268
21
17,887

* p<0.05, ** p<0.01 (two-tailed); cluster bootstrap standard errors are given in parentheses.

47

Figure 2.4. Effects of Poor on Left-wing Voting Conditional on Inequality Change
(Dichotomous, Predicted Probabilities)

.15
.1
.05
-.05

0

Effect of Poor

.2

.25

Left Govt.

-2

-1

0

1

2

3

2

3

Changes in Inequality

.15
.1
.05
-.05

0

Effect of Poor

.2

.25

Non-left Govt.

-2

-1

0

1

Changes in Inequality

Based on Model 20 and 21 of Table 2.5. Dashed lines are 95 percent confidence intervals.

48

Table 2.6. Robustness Checks 4: Fixed Effects Estimation and OLS with Cluster-robust
Standard Errors
DV: : the
Economic
Ideology of the
Party a
Respondent
Votes for
Constant
Poor
Inequality
Change
Poor	
Inequality
Inequality
Level
Religious
Age 18-24
Age 25-34
Age 35-44
Age 45-54
Age 55-64
Age 65 / Older
Male
Education
(Secondary)
Education
(University)
Union
Membership
Unemployed
Self-employed
Ideology
Intraclass
Correlation (ρ)
N (CountryElections)
N

Model 22:
FixedEffects
(All)

Model 23:
FixedEffects
(Leftist)

Model 24:
FixedEffects
(Nonleftist)

6.304**
(0.311)
-0.548**
(0.061)

6.081**
(0.478)
-0.470**
(0.066)

6.489**
(0.421)
-0.557**
(0.084)

-0.008
(0.092)

0.148**
(0.021)

-0.341**
(0.114)

0.191
(0.122)
-0.058
(0.154)
-0.006
(0.164)
-0.143
(0.141)
-0.056
(0.118)
-0.013
(0.086)
(omitted)
0.128*
(0.060)
0.114
(0.081)
0.126
(0.151)
-0.719**
(0.057)
-0.381**
(0.128)
0.594**
(0.100)
0.963**
(0.052)
0.049

0.189
(0.259)
0.014
(0.242)
0.057
(0.314)
-0.111
(0.218)
0.012
(0.194)
0.032
(0.114)
(omitted)
0.196**
(0.058)
0.241
(0.146)
0.196
(0.264)
-0.693**
(0.100)
-0.416*
(0.174)
0.553**
(0.171)
1.008**
(0.094)
0.059

0.184
(0.132)
-0.133
(0.196)
-0.060
(0.185)
-0.163
(0.187)
-0.110
(0.153)
-0.055
(0.126)
(omitted)
0.073
(0.089)
0.032
(0.095)
0.068
(0.177)
-0.746**
(0.072)
-0.348
(0.186)
0.618**
(0.117)
0.930**
(0.063)
0.042

35

14

21

30,916

13,029

17,887

Model 25:
OLS with
Clusterrobust
Standard
Errors
(All)
7.546**
(0.389)
-0.550**
(0.061)
0.137*
(0.056)
-0.008
(0.087)
-0.203**
(0.069)
0.170
(0.123)
-0.084
(0.156)
-0.048
(0.164)
-0.193
(0.140)
-0.100
(0.119)
-0.048
(0.087)
(omitted)
0.128*
(0.061)
0.088
(0.084)
0.125
(0.146)
-0.735**
(0.055)
-0.355*
(0.134)
0.600**
(0.103)
0.962**
(0.052)

Model 26:
OLS with
Clusterrobust
Standard
Errors
(Leftist)
7.415**
(0.529)
-0.465**
(0.065)
0.087
(0.070)
0.142**
(0.021)
-0.267**
(0.048)
0.199
(0.260)
0.021
(0.244)
0.057
(0.315)
-0.110
(0.218)
0.009
(0.194)
0.035
(0.114)
(omitted)
0.196**
(0.059)
0.180
(0.148)
0.176
(0.263)
-0.696**
(0.100)
-0.418*
(0.174)
0.547**
(0.170)
1.006**
(0.094)

35

14

30,916

13,029

Model 27:
OLS with
Clusterrobust
Standard
Errors
(Non-leftist)
7.197**
(0.681)
-0.573**
(0.081)
0.328*
(0.157)
-0.303**
(0.101)
-0.049
(0.190)
0.162
(0.129)
-0.148
(0.198)
-0.112
(0.189)
-0.226
(0.190)
-0.169
(0.157)
-0.092
(0.129)
(omitted)
0.073
(0.089)
0.025
(0.096)
0.076
(0.177)
-0.760**
(0.073)
-0.279
(0.193)
0.632**
(0.121)
0.930**
(0.063)
21
17,887

* p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses; Country
dummies are included in the models estimated via OLS and coefficients on them are not reported.
49

the moderation effect of inequality change on left-wing voting. While the point where the
effect of poor on left-wing voting disappears is slightly different, substantive interpretation of
the moderation effect is almost identical regardless of the type of the dependent variable.
Finally, I estimate the models with alternative estimation methods. First, I estimate
fixed-effects models by relaxing the assumption of uncorrelatedness between unobserved
heterogeneity at the country-election level and independent variables. Then, I re-estimate the
models including country dummy variables with OLS. The estimation results in Table 2.6
indicate that the results from the main analysis are robust to these alternative estimation
methods. In summary, the moderation effect of inequality change on the poor’s voting is
robust to all the alternative measures and model specifications, and thus it is safe to conclude
that the evidence supporting the main hypothesis is very strong and stable.

2.5.3.2. Additional Analysis on the Moderation Effect of Religiosity
While examining the moderation effect of inequality change is the primary interest, I
also test whether the effect of religiosity on voting is disproportionately stronger for the poor
as previous studies suggest. The estimation results for Model 28 show that voters with
religious beliefs support more economically conservative parties, yet this effect is not
statistically significant. Also, I construct an interaction term between religious and poor in
order to examine the disproportionate effect of religiosity. Previous studies argue that
income-based voting is more prevalent among the secular poor than the religious poor (De La
O and Rodden 2008; Huber and Stanig 2007), thus the coefficient on the interaction term
should be positive. The coefficient estimates in the first column of Table 2.7 show that there
is no evidence supporting this argument. The coefficient estimate on the interaction variable
between poor and religious is not only estimated with large error, but also its sign is opposite
to the expected direction. This result renders little support for the “distraction” hypothesis

50

Table 2.7. The Effect of Poor Conditional on Religiosity
DV: the Economic
Ideology of the Party
a Respondent Votes
for
Constant
Poor (Income)
Inequality Change
Inequality Level
Religious
Poor

Religious

Age 18-24
Age 25-34
Age 35-44
Age 45-54
Age 55-64
Age 65 / Older
Male
Education
(Secondary)
Education
(University)
Union Membership
Unemployed
Self-employed
Ideology

N (Country-elections)
N

Model 28:
Interaction b/w Poor &
Religiosity
(All)
6.228**
(0.348)
-0.547**
(0.065)
0.048
(0.161)
-0.003
(0.033)
0.207
(0.156)
-0.037
(0.161)
-0.059
(0.154)
-0.005
(0.164)
-0.142
(0.140)
-0.056
(0.118)
-0.014
(0.086)
(omitted)
0.128*
(0.060)
0.107
(0.082)
0.119
(0.150)
-0.725**
(0.057)
-0.382**
(0.128)
0.595**
(0.100)
0.962**
(0.052)
0.597
12.204
35
30,916

Model 29:
Interaction b/w Poor &
Religiosity (All, the
modest religious)
6.169**
(0.339)
-0.561**
(0.071)
0.052
(0.164)
-0.004
(0.033)
0.301*
(0.151)
0.028
(0.168)
-0.021
(0.153)
0.023
(0.165)
-0.120
(0.140)
-0.035
(0.118)
0.002
(0.085)
(omitted)
0.138*
(0.060)
0.105
(0.082)
0.117
(0.149)
-0.727**
(0.057)
-0.377**
(0.127)
0.596**
(0.100)
0.959**
(0.052)
0.625
12.194
35
30,916

* p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

regarding the poor’s voting behavior. The moderation effect of religious is not detected in the
estimation result of Model 28, but this result can be also a product of the measure used in the
51

original analysis. Following Huber and Stanig (2011), respondents declaring that they attend
at religious services “once a week/more than once a week” or identifying themselves as “very
religious” are classified as the religious in the original analysis, but I extend this category by
including respondents answering “two or more times a month” or “somewhat religious.” The
estimation result for Model 29 in Table 2.7 shows that the previous result is not mere a
product of a particular measure of religiosity. While the result indicates that religiosity per se
causes voters to support more conservative parties, but I cannot find evidence that this effect
is particularly stronger for the poor. The effect of religious on voting turns out to be stronger
with the new measure of religiosity, there is still no evidence that the poor are more likely to
be distracted from their self-interest than the rich.

2.6. Conclusion
This paper examines the poor’s voting behavior with a focus on their economic
interests, which interests have been underappreciated in previous studies. The declining
relevance of income, economic interests, and class to citizens’ voting has been a popular
subject of political science research. In particular, numerous political science research studies
offer explanations as to why the poor sometimes vote against their own interests. The recent
debate on this puzzle has mainly revolved around factors which can be substituted for
economic preferences of poor voters. There is also a journalistic argument that cultural and
moral issues virtually “outweigh” economic issues in the United States (Frank 2004). While
previous research has produced fruitful contributions to the better understanding of the poor’s
voting behavior, this study is motivated by the observation that very little attention has been
paid to the poor’s economic interests per se. Instead of examining other voting cues, I attempt
to offer an explanation for the seeming contradiction by focusing on the poor’s economic
interests, which should be affected by government policies. Specifically, I demonstrate that

52

the poor’s evaluations of the left-wing government’s performance in improving economic
inequality are of critical importance to the poor’s voting calculus. The findings of this chapter
provide strong evidence to support the hypothesis that income-based voting becomes
prevalent when the leftist government successfully ameliorates economic inequality. This
paper therefore contributes to the existing literature on the poor’s right-wing voting by
suggesting a simple, but novel explanation for this long-lasting puzzle: the poor do not
support leftist parties because of policy outcomes against their self-interest delivered by their
representative.
This paper also highlights the importance of contextual factors in studying microlevel political behavior. Political scientists studying political economy and political behavior
have argued that macro-level contextual factors, including economic conditions and political
institutions, should be taken seriously in order to advance understanding of political behavior
at the individual level (e.g., Anderson and Singer 2008; Rueda 2008). This paper echoes this
argument and emphasizes the importance of parties in government as suppliers of economic
policies that influences citizens’ political behavior. The voluminous literature on the partisan
model of economic outcomes suggests that economic policies and macroeconomic outcomes
should differ according to government partisanship. If this argument is correct, citizens
should expect that different policy goals will be pursued depending on the partisan
composition of the government (Duch and Stevenson 2008). It follows that the poor expect
they will benefit more from the left in government. However, if a leftist government falls
short of these expectations by failing to produce expected economic outcomes, the outcome
will disappoint its core constituency – the poor. In this case, the less well-off may explore
different criteria for their voting in the upcoming election. The findings of this chapter
contribute to the literature by properly locating micro-level behavior in the context of the
macro-level economic condition (i.e., inequality changes) and the institutional features (i.e.,

53

government partisanship).
Finally, this paper calls attention to the rationality of the poor by indirectly testing
whether the poor are able to understand their personal loss during a time of growing
economic inequality. Previous research assumes that the poor are readily mobilized around
other politicized issues. That argument mainly relies on findings that the poor tend to be less
educated and ill-informed (e.g., Verba et al. 1995). The extant literature on the poor’s rightwing voting also tends to depict the poor as somewhat irrational agents who tend to be easily
distracted from their material interests (e.g., De La O and Rodden 2008) or receive greater
psychological benefits from religious beliefs than the wealthy (e.g., Scheve and Stasavage
2006). My findings in this chapter present a different story. While it is argued that wellinformed voters are more likely to make “correct” choices, it does not necessarily mean that
the poor lack the ability to consider their voting choices based on their self-interest (e.g.,
Ansolabehere et al. 2006). My findings suggest that the poor decide to support non-leftist
parties not because they are irrational, but rather because they recognize that the left is no
longer a valid or competent advocate of their interests. In other words, they have a good
reason for not voting for the left. These findings illuminate the rationality in the political
behavior of the poor. The assumption that the poor have ill-informed opinions on politics is
widely accepted, yet the findings in this chapter calls this assumption into question and
invites further scholarly investigation of the rationality of the poor’s political behavior in
future research.

54

APPENDIX

55

Table 2.8. List of Country-elections, Partisanship of Incumbent Governments, and Missing
Variables
Country/Election
Australia 1996
Australia 2004
Australia 2007
Austria 2008
Belgium 1999
Belgium 2003
Canada 1997
Canada 2004
Canada 2008
Denmark 1998
Denmark 2001
Denmark 2007
Finland 2003
Finland 2007
Finland 2011
France 2002
France 2007
Germany 1998
Germany 2002
Germany 2005
Germany 2009
Greece 2009
Ireland 2002
Ireland 2007
Italy 2006
Japan 1996

CWS
Left
Right
Right
Center
Center
Center
Center
Center
Right
Left
Left
Right
Left
Center
Center
Left
Right
Right
Left
Left
Center
Right
Center
Center
Right
Right

CPPD
Left
Right
Right
Center
Left
Left
Center
Center
Right
Left
Left
Right
Left
Center
Center
Left
Right
Right
Left
Left
Center
Right
Right
Right
Right
Right

CPDS
Left
Right
Right
Center
Center
Center
Right
Right
Right
Left
Left
Right
Center
Center
Right
Left
Right
Right
Left
Left
Center
Right
Right
Right
Right
Right

The Netherlands 1998
The Netherlands 2002
The Netherlands 2006
The Netherlands 2010
New Zealand 1996
New Zealand 2002
New Zealand 2008
Norway 1997
Norway 2001
Norway 2005
Norway 2009
Portugal 2002
Portugal 2005
Portugal 2009
Spain 1996
Spain 2000
Spain 2004
Spain 2008

Left
Left
Center
Center
Right
Left
Left
Left
Left
Right
Left
Left
Right
Left
Left
Right
Right
Left

Center
Center
Center
Center
Right
Left
Left
Left
Left
Right
Left
Left
Center
Left
Left
Right
Right
Left

Center
Center
Right
Center
Right
Left
Left
Left
Left
Right
Left
Left
Right
Left
Left
Right
Right
Left

Sweden 1998
Sweden 2002
Sweden 2006
Switzerland 1999

Left
Left
Left
Right

Left
Left
Left
Right

Left
Left
Left
Right

56

Missing Variables
Self-employed
Self-employed
Income
Self-employed
Religiosity
Unemployed
Religiosity, Self-employed

Self-employed

Ideology, Religiosity,
Unemployed, Self-employed

Religiosity
Religiosity
Religiosity
Self-employed
Self-employed
Religiosity
Union Membership, Selfemployed
Religiosity

Table 2.8 (cont’d)
Switzerland 2003
Right
Right
Right
Self-employed
Switzerland 2007
Right
Right
Right
Self-employed
United Kingdom 1997
Right
Right
Right
United Kingdom 2005
Left
Left
Left
United States 1996
Center
Center
Right
United States 2004
Right
Right
Right
United States 2008
Right
Right
Right
Self-employed
* Comparative Welfare State Dataset (CWS, Brady et al. 2014): cases are coded as leftist (or rightist)
government if leftist (or rightist) parties as a group have more than 50% of the legislative seats held
by the government, or have more than 40% of the legislative seats held by incumbent parties and a
plurality of the legislative seats; if both left-wing and right-wing groups have more than 40% of the
legislative seats held by the government, then it is considered as centrist government even though one
of the two have more than 50% of the seats, or have more than 40% of the seats and a plurality of the
seats; if centrist parties hold more than 50% of the legislative seats held by incumbent parties, or have
more than 40% of the legislative seats held by the government and a plurality of the seats, then it is
coded as centrist government.
* Comparative Political Parties Dataset (CPPD, Swank 2013): cases are coded as leftist (or rightist)
government if leftist (or rightist) parties as a group have more than 50% of the seats, or have more
than 40% of the seats and a plurality of the seats in the cabinet; if both left-wing and right-wing
groups have more than 40% of the seats in the cabinet, then it is considered as centrist government
even though one of the two have more than 50% of the seats, or have more than 40% of the seats and
a plurality of the seats in the cabinet; if centrist parties hold more than 50% of the seats in the cabinet,
or have more than 40% of the seats and a plurality of the seats in the cabinet, then it is coded as
centrist government.
* CPDS (Armingeon et al. 2013): Governmental Partisanship coded from 1 to 5

57

Table 2.9. Details Regarding Control Variables
Variable
Religiosity

Age

Male
Education
(Secondary)
Education
(University)
Union Membership
Unemployment

Description
1 if a respondent attends church weekly, 0
otherwise; if this question is not asked
(Canada, Finland, and Spain), another
question asking about respondents’
religiosity is used. I assigned 1 if a
respondent self-describes as “very
religious,” 0 otherwise
Six dichotomous variables are created for
“age.” For each dichotomous variable, if a
respondent belongs to the range for the
variable, 1 is assigned to her (or him), 0
otherwise.
1 if a respondent is male, 0 otherwise.
1 if a respondent has completed secondary
school or has an experience of higher
levels of education, but fails to complete
university degree, 0 otherwise.
1 if a respondent has completed a
university degree or has an experience of
higher levels of education, 0 otherwise.
1 if a respondent or someone in household
other than respondent is a member of
union, 0 otherwise.
1 if a respondent is unemployed, 0
otherwise.

Self-employed

1 if a respondent is self-employed, 0
otherwise.

Ideology

Self-placement of a respondent on a scale
from 0 (left) to 10 (right).

58

Survey Items
A2015/A2016 (the 1st wave),
B2023/B2024 (the 2nd wave), and
C2023/C2024 (the 3rd wave)

A2001 (the 1st wave), B2001 (the
2nd wave), and C2001 (the 3rd
wave)
A2002 (the 1st wave), B2002 (the
2nd wave), and C2002 (the 3rd
wave)
A2003 (the 1st wave), B2003 (the
2nd wave), and C2003 (the 3rd
wave)
A2003 (the 1st wave), B2003 (the
2nd wave), and C2003 (the 3rd
wave)
A2005/A2006 (the 1st wave),
B2005/B2006 (the 2nd wave), and
C2005/C2006 (the 3rd wave)
A2007 (the 1st wave), B2010 (the
2nd wave), and C2010 (the 3rd
wave)
A2009 (the 1st wave), B2012 (the
2nd wave), and C2012 (the 3rd
wave)
A3031 (the 1st wave), B3045 (the
2nd wave), and C3013 (the 3rd
wave)

Table 2.10. Summary Statistics
Variable
Economic Ideology of Voted Party
(Continuous)
Left-wing Voting (Dichotomous)
Poor
Income (Quintile)
Inequality Change
Inequality Level
GDP Growth Rate
Religiosity
Religiosity (Moderate)
Age 18-24
Age 25-34
Age 35-44
Age 45-54
Age 55-64
Age 65 or Older
Male
Education (Secondary)
Education (University)
Union Membership
Unemployed
Self-employed
Ideology

Obs
75,771

Mean
10.797

Std. Dev.
4.411

Min
2.1

Max
18.7

75,771
86,276
86,276
102,147
102,147
102,147
81,598
81,598
101,397
101,397
101,397
101,397
101,397
101,397
101,950
99,215
99,215
93,054
98,512
67,812
86,271

0.510
0.403
2.967
0.160
-0.027
2.187
0.164
0.274
0.091
0.161
0.196
0.188
0.167
0.197
0.482
0.455
0.182
0.317
0.048
0.098
5.176

0.500
0.490
1.387
0.940
4.116
2.250
0.370
0.446
0.287
0.368
0.397
0.391
0.373
0.398
0.500
0.498
0.386
0.465
0.215
0.297
2.279

0
0
1
-2.017
-6.931
-5.6
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

1
1
5
3.368
8.000
5.9
1
1
1
1
1
1
1
1
1
1
1
1
1
1
10

59

CHAPTER 3
CHANGING POSITION OR PLAYING FIELD?
THE DISTINCT EFFECTS OF ECONOMIC INEQUALITY ON STRATEGIES OF
RIGHT PARTIES IN OLD AND NEW DEMOCRACIES

“The new conflict can become dominant only if the old one is subordinated, or obscured, or
forgotten, or loses its capacity to excite the contestants, or becomes irrelevant.” – E. E.
Schattschneider, The Semisovereign People: A Realist’s View of Democracy in America.

3.1. Introduction
The aim of this chapter is to explain the manner of response by right-wing parties to
the growing economic inequality by right-wing parties in advanced and emerging
democracies. The central finding is that when faced with a high level of economic inequality,
the political right in new democracies opt to profess more leftist positions on the economic
policy dimension. On the other hand, right-wing parties in advanced democracies are more
inclined to emphasize social issues compared to left-wing parties even though taking issue
positions closer to the median voter is a more attractive option than politicizing social issues.
The reason is that in advanced democracies stronger political constraints on strategic choices
of party leadership curb opportunistic policy moderation of the right-wing parties. Thus
instead, right-wing parties in advanced democracies attempt to politicize non-economic
issues, including moral values, religion, and national identity, in order to draw voter attention
away from economic issues.
Optimal choice of policy positioning has been one of the central issues in political
science research. Ample research has been conducted to investigate how political parties
adjust their positions in response to various environmental factors, such as public opinion
60

(Adams et al. 2004, 2006; Ezrow et al. 2011), past election records (Somer-Topcu 2009), and
economic conditions (Adams et al. 2009; Barth et al. 2015; Haupt 2010; Ward et al. 2011). In
addition, several recent empirical studies also recognize that party competition usually takes
place in the multidimensional policy space and analyze how parties manipulate the salience
of economic and non-economic issues in their platforms in response to changing economic
conditions (e.g., Tavits and Letki 2014; Tavits and Potter 2015; Ward et al. 2015). While I
develop my theoretical arguments building on the extant literature, I also seek to contribute to
the current understanding of party competition in the three following ways.
First, previous studies generally focus on either the party strategies on position or
salience in response to various exogenous factors, but they fall short of incorporating those
options available for parties into a single theoretical framework. For instance, several recent
studies explored the effect of economic inequality on party positions on the welfare state
policy (Barth et al. 2015), or on the amount of emphasis parties place on the “values”
dimension (Tavits and Potter 2015); meanwhile, those studies focus only on a single aspect of
party strategies – issue emphasis or issue positioning – and fail to account for the other option
available to the parties (cf. Tavits and Letki 2014). However, it is reasonable to believe that
parties are able to employ both strategies in the pursuit of their electoral success (Meguid
2005, 2008; Wagner 2012; Ward et al. 2015). Given this understanding that both strategies are
considered, I regard the two strategic options as substitutes for each other in the parties’
efforts in pursuing their electoral gains.
Second, while there are notable attempts to explain how economic inequality affects
party competition (e.g., Barth et al. 2015; Tavits and Letki 2014; Tavits and Potter 2015), they
fail to take into account different political environments that confront parties in advanced or
nascent democracies. To address this void, I analyze how the contextual differences between
the old and new democracies lead their right-wing parties to choose distinct strategies. In
61

doing so, this chapter complements the literature by accounting for the varying degrees of
political constraints on strategic party decisions that bear on the contextual differences
between advanced and emerging democracies.
Finally, this chapter offers an important implication for understanding the class
politics in emerging democracies. While growing economic disparity entailed by economic
liberalization and globalization promotes the formation of interest-based preferences over
economic policies among citizens (Kitschelt 1992), these preferences sometimes fail to
translate into citizens’ vote choices (Gijsberts and Niewbeerta 2000). My findings in this
chapter suggest a theoretical possibility for the discrepancy. Specifically, rising inequality
creates incentives for right parties to move leftwards, thus making it difficult for voters to
identify representatives who best represent their interests. Furthermore, the party convergence
on the economic dimension reduces the salience of economic issues, thereby discouraging
voters to base their electoral choices on their economic preferences (Evans and Tilley 2012a).
In this regard, this chapter contributes to providing a better understanding of why the
development of certain social cleavages are not reflected in the political realms in new
democracies.
The next section provides a review of the literature on party strategies on position
and salience, and exploration of various factors that may affect parties’ electoral strategies.
Then, in the third section, I draw a set of hypotheses based on a discussion of party strategies
and the political constraints imposed on strategic choices of party leadership in both
advanced and emerging democracies. The fourth section presents a series of results from the
empirical analyses. I also conduct additional analyses to test whether the results that confirm
the hypotheses are driven by the proposed causal mechanism. Concluding remarks and a
discussion of the implications of the findings are presented in the last section.

62

3.2. Literature Review: Party Competition in the Policy Space
A political party can be defined as an institution that aims to seek positions in
government as well as shared particular policy goals (e.g., Strøm and Müller 1999; Ware
1996). For this reason, parties’ electoral platforms reflect their broad ideological principles,
but they are also products of strategic considerations of electoral and political environments
that may affect their fortunes in the elections. In this sense, the extant literature on party
competition in democratic elections generally regards the drafting of party platforms as an
attempt to speak to citizens and maximize the party vote shares.
This sizable literature portrays electoral competition as efforts of political parties, as
unitary agents, to position themselves properly on the unidimensional policy space in order to
draw support from a larger segment of citizens (Downs 1957; Enelow and Hinich 1984; cf.
Budge et al. 2010; Schumacher et al. 2013). Based on this theoretical framework, ample
empirical research provides explanations of how parties decide their positions within the
policy space in response to environmental incentives. First, parties shift their positions in
response to a change in the distribution of voter preferences on the policy dimension (Adams
et al. 2004, 2006). Yet parties also respond to different groups of citizens depending on the
parties’ programmatic profiles (i.e., mainstream or niche). Mainstream parties respond to
opinion shifts in the mean voter position in the whole of the electorate, whereas niche parties
are more responsive to their core supporters (Ezrow et al. 2011). Election results in the past
also motivate parties to shift their positions. Somer-Topcu (2009) finds that only those parties
that lost votes in the past election change their policy positions in the current election, but this
effect fades between current and past elections (cf. Adams et al. 2004; Adams et al. 2011).
Lastly, economic conditions, such as economic globalization and inequality, are also
proposed as factors that are taken into consideration in understanding parties’ position-taking.
Adams et al. (2009) find evidence that non-leftist (i.e., centrist and rightist) parties change
63

their ideological positions in response to globalization, whereas leftist parties are generally
unresponsive to economic conditions related to globalization. However, Haupt (2010) finds
that rightist and leftist parties behave similarly in response to economic openness. In addition,
Ward et al. (2011) argue that the effect of globalization on party positions depends on the
position of the median voter. In particular, the influence of globalization on policy positions
of social democratic parties takes effect only when the median voter is leftist, since the
parties face pressure to move rightward, created only by the advent of globalization and
otherwise nonexistent. There are also notable attempts to uncover the mechanisms behind
party response to domestic economic conditions in electoral competition. For instance,
Pontusson and Rueda (2010) find that increase in inequality induces left-wing parties to move
leftward only when the poor are politically mobilized. On the other hand, Barth et al. (2015)
find that leftist parties in OECD democracies advocate less generous welfare state policy
when faced with higher inequality.
While numerous studies focus on parties’ policy positions in the unidimensional
policy space, electoral competitions rarely center around one single issue (e.g., De Sio and
Weber 2014; Miller and Schofield 2003; Roemer 1998; Schofield and Sened 2006; cf. Hix et
al. 2006). Therefore, a substantial body of literature on party competition assumes a
multidimensional policy space that comprises the two main policy domains: economic
(“interests”) and social (“values”) issues dimensions (Miller and Schofield 2003; Roemer
1998; Schofield and Sened 2006; Tavits and Potter 2015). Such multidimensionality of the
policy space encourages political scientists to study issue emphasis as well as position-taking
(e.g., Green-Pedersen 2007; de Vries and Hobolt 2012; Hobolt and de Vries 2015; Meguid
2005, 2008). First, one branch of research focuses on the fact that parties are inclined to
politicize issues on which those parties are recognized as competent or credible (Budge and
Farlie 1983). For example, U.S. voters believe that the Republican Party is more capable of
64

fighting inflation, whereas the Democratic Party is better at dealing with unemployment
(Petrocik 1996). On the other hand, recent studies find that parties also emphasize issues that
are central to the public debate (Klüver and Sagarzazu 2016; Spoon and Klüver 2014). In
particular, the strategy to emphasize issues of priority may be successful only if those issues
are salient enough (Bélanger and Meguid 2008). Vote-seeking parties with greater resources
are also more likely to engage in “riding the wave” of public concern by addressing issues
that are central in the public debate (Wagner and Meyer 2014).
Several recent studies have direct relevance to the research question here. Tavits and
Letki (2014) investigate how economic inequality affects the polarization of party systems in
post-communist democracies, basing their theory on the literature regarding the relationship
between inequality and party polarization witnessed in advanced democracies. Their
argument, specifically, is that the rise in inequality leads to an increased prevalence of
economic interests among the citizens and, accordingly, party-system polarization within the
sphere of economic issues. Moreover, Tavits and Potter (2015) demonstrate that right-wing
parties attempt to politicize issues pertaining to the social values in order to draw voter
attention away from economic issues as inequality grows, while leftist parties, in constrast,
are incentivized to emphasize economic issues in the face of growing economic disparity.

3.3. Theory: Party Strategies, Political Constraints, and Economic Inequality
3.3.1. Policy Space: Definition
In accordance with the previous studies, I define the two-dimensional policy space as
one that is constituted by the dimensions of economic and social issues (Miller and Schofield
2003; Tavits and Letki 2014; Tavits and Potter 2015). Issues within the economic dimension
include the classical conflict of economic interests, such as redistribution, progressive
income/property taxes, social welfare, and state regulation/deregulation. On the other hand,
65

issues related to national identity and patriotism, immigration, moral values, religiosity, and
the degree of punishment for criminals to preserve public order are categorized as social
issues.

3.3.2. Issue Positioning versus Issue Emphasis
When competing in a two-dimensional policy space, political parties have to decide,
first, where they position themselves along each of the policy dimensions, and second, which
issues they want to emphasize. This also implies that when encountered with adverse
circumstances during electoral competitions (i.e., being positioned farther away from the
median voter than their competitors), parties have two options to turn the situation around to
their favor. First, parties can take an issue position that is closer to the median voter rather
than the positions they prefer in the policy dimension. This idea is very much in line with the
classical proximity model of party competition in which voters choose parties close to their
ideal points in a one-dimensional policy space (Down 1957). In addition, the
multidimensional policy space allows parties to manipulate the salience of each policy
dimension. In other words, parties are also able to politicize issues on the other dimension to
divert voter attention to the policy dimension in which they are more likely to win. This
strategy echoes Riker’s concept of heresthetics, which is defined as the strategic manipulation
of the political structure (e.g., dimensional salience) to construct a winning majority (Riker
1986).14
The process of formulating party strategy profiles consequentially leads to the
question of determining which option is better in terms of a costs and benefits analysis. Both

14

According to Riker, among three varieties of heresthetical maneuvers – agenda control, strategic
voting, and manipulation of issue dimensions –, the latter is “just about the most frequently attempted
heresthetical device” (Riker 1986: 150).
66

strategies – position-taking and manipulating salience – can be used as substitutes for each
other in pursuing electoral gains (Meguid 2005, 2008; Wagner 2012), yet they are not
completely interchangeable in that they vary in effectiveness and availability. I argue that,
given that both are successful, position-taking is more effective than emphasizing issues on
the other dimension for achieving electoral gains. The prime expected benefit of positiontaking is a proximal gain by attracting voters who would have voted for the rival parties. In
addition, another concomitant benefit of position-taking is drawing voter attention away from
the issue dimension disadvantageous to the party. This mechanism can be explained by the
fact that convergence of political parties on an issue reduces the importance of that particular
issue at the ballot box (Green and Hobolt 2008; Meguid 2008; Miller and Schofield 2003;
Wagner 2012). Emphasizing issues on the other policy dimension, however, may not be as
effective for electoral gains compared to the strategy of positon-taking, since the outcome of
this option constitutes only one aspect of expected effects by position-taking – drawing voter
attention away from a particular policy dimension. Moreover, if the issue that the party
desires to downplay is already established as a prominent issue for the voters, manipulating
issue salience by politicizing alternative issues is less likely to have an effect (Bélanger and
Meguid 2008).

3.3.3. Political Constraints to Strategic Choice of Parties
Position-taking may be a more attractive option in the pursuit of electoral advantages
than manipulating issue salience. Nevertheless, it should also be noted that due to various
political constraints, this strategy may not always be feasible. In order to analyze what
constitute such political constraints, I borrow two concepts from Meguid (2008):
organizational and reputational constraints. First, I define organizational constraints as agents
or institutions that restrict the autonomy of party leadership in the decision-making process.
67

Organizational constraints usually take effect within the party organization in the forms of
factions and party activists; party strategies to seek popular support are sometimes
constrained by the dynamics of interactions between various agents within the party (Bäck
2008; Budge et al. 2010; Pedersen 2010; Strøm and Müller 1999).15 Especially in
comparison to party leaders, party activists are generally more committed to a certain
ideological identity and less willing to abandon policy principles for votes or office (Aldrich
1995; Miller and Schofield 2003; Garand 2010; Panebianco 1988; Schumacher et al. 2013;
Strøm 1990). Therefore, intraparty decision-making in selecting a party position over a
certain issue sometimes involves a considerable amount of time, and, in some cases, it is even
possible to be unable to reach an agreement (Meguid 2008). Consequently, it is challenging
for parties that have decision-making power dispersed among various internal units to change
their issue positions in accordance to the median voter position or in response to officeseeking incentives (Pedersen 2012; Schumacher et al. 2013; Strøm 1990). In addition, parties’
strategic choices may be also constrained by citizens with partisan attachments to each party.
Parties that have participated in electoral competitions throughout a long period have core
supporters and usually enjoy stable and strong long-term partisan ties with such supporters. In
order to achieve electoral gains, parties may attempt to make opportunistic moves at times,
but changing issue positions always involves considerable uncertainty in terms of
effectiveness (Budge 1994). As Petrocik posits, “a Democrat’s promise to attack crime by
hiring more police, building more prisons, and punishing longer sentences would (…)
provoke (decisive) distress among Democratic constituents” (Petrocik 1996: 829). Therefore,
it is likely that such parties will be reluctant to change their political stance for uncertain

15

To take into account organizational constraints, I relax the popular assumption of a party as a
unitary agent in the literature on party competition (see Budge et al. 2010; Miller and Schofield 2003;
Schumacher et al. 2013).
68

short-term gains while risking the loss of assured support from their core constituencies, and,
for this reason, such long-standing ties between parties and their core supporters act as
constraints to parties’ attempts to pursue opportunistic position-taking.
Yet the organizational aspect is not the only factor that imposes constraint on the
strategic decisions of parties. Policy reputation can also deter parties from carrying out
opportunistic positioning. First, the intertemporal policy inconsistency can cause a party to
lose its programmatic credibility (Kitschelt and Rehm 2015). Voters tend to “discount” a
party’s policy that contradicts its previous stance, given that voters have a priori knowledge
regarding the party’s ideological principle and policy position in the past elections (Kitschelt
and Rehm 2015; Tomz and van Houweling, 2014). Furthermore, as Meguid posits, “a party
that pursues consistent policy objectives will be considered more responsible and trustworthy
than one that vacillates between opposing policy positions” (Meguid 2008: 106). This implies
that parties lacking intertemporal policy consistency are more likely to suffer reputational
cost in terms of valence evaluation (DeBacker 2015; Lupu 2014; Tomz and van Houweling
2014). The intertemporal inconsistency of policy positions can be detrimental to the general
reputation of a political party, especially if a rightist party take a leftist position or, inversely,
if a leftist party takes a rightist one (Meyer and Müller 2014; Tomz and van Houweling
2014). Therefore, the reputational constraint from past issue positions also discourages
parties from implementing strategic position-taking.16
The decision to enforce the issue salience strategy is relatively free, however, from
these constraints mentioned above. As Wagner states, “positional decisions precede salience
choices in party strategies” (Wagner 2012: 66). His statement implies that party activists, who
16

To be clear, I do not argue that parties are always forced to take policy positions that are identical
to their previous positions, and that there is no intertemporal variation in party position-taking under
strong political constraints. My argument is that stronger organizational and reputational constraints
should allow considerably smaller sets of points within a policy space for party position-taking.
Opportunistic position-taking is also much more costly when political constraints are strong.
69

possess stronger preferences in regard to their desired policy positions, are generally less
adamant about parties’ choice of issue salience than about policy positions (Wagner 2012).
Moreover, voters usually identify and support parties based on which stance the parties take
on specific issues rather than on how much importance the parties give to those issues. For
this reason, congruence in issue position between voter and party, in most cases, is a prime
condition for forging a programmatic tie between the two. However, the programmatic tie
formulated through issue salience should be subject to a certain degree of issue congruence.
“If an individual does not share a party’s issue stance, then it is irrelevant that she finds that
particular party to be the owner of the policy position” (Bélanger and Meguid 2008: 483).
This is the case even for niche parties, whose issue salience is a core element of party identity
(Wagner 2012). For example, French voters who prioritize multiculturalism and
assimilationist policies are least likely to vote for the far-right Front National (FN), even if
the FN places its anti-immigrant stance on its highest priority in its election campaign.
To summarize, issue positioning can be a much more effective option for political
parties in the pursuit of electoral gains. Nevertheless, party leaders are reluctant to employ
this strategy when they are constrained by well-institutionalized party organization or policy
reputations developed throughout history.17 In such cases, party leadership will opt to
manipulate issue salience in its electoral platform, which enables party leadership to protect
its reputation and easily secure the approval of party activists and core supporters.

3.3.4. Political Parties and Electoral Competitions in Old and New Democracies
Both issue emphasis and issue positioning are tools that are utilized by parties in

17

In fact, the organizational and reputational constraints are to some extent interrelated. For example,
vote-seeking party activists may try to hinder strategic position-taking to maintain programmatic
credibility or the general reputation of their party (Kitschelt and Rehm 2015).
70

order to improve their electoral fortunes, but the effectiveness of each tool depends on the
strength of organizational and reputational constraints being imposed on the strategic choice
of parties. I emphasize the importance of environmental differences in advanced and nascent
democracies in understanding the degree of constraints imposed on parties’ strategic choice
during electoral competition. Three key differences between advanced and nascent
democracies that are relevant to the varying degree of political constraints are summarized in
the following discussion.
First, party organization and decision-making process in emerging democracies are
not as well-institutionalized as those in advanced democracies. Literature on party
organizations and party systems in new democracies in Southern and East-Central Europe
shows that weak party organizations allow party leaders to exercise stronger authority in the
decision-making process (van Biezen 2003, 2005). This low level of party institutionalization
enables parties to make timely programmatic response to changing environments around
elections (Lupu and Riedl 2013). Moreover, political parties in nascent democracies are
sometimes characterized as personalistic or flexible, meaning that they operate according to
the individual decisions made by party leaders rather than by an institutionalized decisionmaking bodies (Bader 2008; Choi et al. 2014).18 In short, party leaders in nascent
democracies are usually less constrained by party activists or strong organizational units, and
as a consequence, possess greater autonomy in intraparty decision-making process.
Second, new democracies generally show weak and unstable voter-party linkages
(Ezrow et al. 2014; Lupu and Riedl 2013; McAllister and White 2007; Rose 1995). Citizens

18

Lupu and Riedl (2013) emphasize the effects of uncertainty in new democracies on the low degree
of party institutionalization. Specifically, party elites in emerging democracies intentionally maintain
a flexible (or uninstitutionalized) party organization and restrain intraparty contestation in order to
cope with uncertainties inherent to new democracies. Ware also notes that “(parties) would want an
organization that can be modified to be effective in new circumstances (Ware 1996: 140)” in regimes
with chronic instability.
71

of new democracies are less likely to be attached to particular parties due to their lack of past
experience of democratic elections. The linkages between parties and voters are usually
established and maintained by voters’ support for prominent figures associated with specific
parties rather than by programmatic appeals or ideological commitments (Mainwaring and
Tocal 2006). Such lack of party roots in the society is closely related to weakly
institutionalized party systems in new democracies (Mainwaring 1999). In contrast, voterparty linkages based on voters’ organizational and psychological attachments to parties are
more durable and stronger in advanced democracies. The stability of voter-party linkages in
advanced democracies is evident, especially compared to the linkages in nascent democracies
where citizens are less likely to have inherited partisan attachments to particular parties due
to the experience of shorter democratic episodes (Dalton and Klingerman 2007; Dalton and
Weldon 2007; Mainwaring 1999). Moreover, frequent splits and mergers among parties and
changes in party labels also prevent citizens from building long-lasting ties to particular
parties in emerging democracies (Rose and Munro 2003).
Finally, political parties in emerging democracies are more likely to lack policy
reputations among voters. Voters in nascent democracies generally find it hard to associate
certain parties with their positions on the policy spectrum due to their lack of experience in
observing how parties behave once they are in office (Tucker 2006). Keefer (2007)
demonstrates that underdeveloped policy reputations of parties and subsequent lowcredibility in their electoral promises encourages political parties to invest in patron-client
relationships rather than in programmatic appeals aimed at a broad group of citizens. He also
presents evidence that politicians in emerging democracies prefer clientelistic networks with
targeted good provision because of such low credibility and the lack of policy reputations.
The correlation between the level of political credibility and the age of democracy may not be
deterministic. In addition, parties in nascent democracies sometimes inherit their reputations
72

from the period prior to democratization during which such reputations have developed.
Nevertheless, policy reputations of parties in advanced democracies are more likely to
develop and are in better form, since policy reputations take time to build (Keefer 2007;
Kitschelt 1995) and party systems tend to stabilize as a democracy matures (Lupu and Stokes
2010). These differences in political environment between old and new democracies indicate
that political constraints on opportunistic position-taking of parties are stronger in advanced
democracies.

3.3.5. Economic Inequality and Right Parties’ Strategies in Old and New Democracies
With the increase in economic inequality, economic interests gain greater
significance in electoral choice in both advanced democracies (McCarthy et al. 2006;
Leighley and Nagler 2013) and nascent ones (Tavits and Letki 2014). For instance, growing
inequality creates fertile ground for the ensuing formation of class cleavage and its
subsequent increase in importance to the electoral competition within Central and Eastern
European countries (Evans 1997; Gijsberts and Niewbeerta 2000). Rising inequality also
increases the right-skewedness of the income distribution, and therefore the proportion of the
population below the mean income also increases (Finseraas 2009a). This implies that
electoral competition that focuses on economic issues will be detrimental to the rightist
parties; while left-wing parties are willing to take their preferred positions in the economic
dimension, right-wing parties are faced with the necessity of overcoming the challenge. The
previous discussion about party strategies implies that the efforts by right-wing parties to
achieve electoral gain can take two forms. First, right-wing parties can try to highlight social
issues to draw voter attention away from economic issues. The other option is to position
themselves more towards the left than their preferred positions in the economic dimension as
a means to reduce the distance between themselves and the median voter (and also the left73

wing parties).19
Of the two options, which is a more effective strategy? The former should be a more
attractive option to the right, as discussed above. Shifting to the left helps the right secure
partial support from the middle-income group, but the convergence in the economic
dimension also mitigates the importance of economic issues in the electoral competition. As a
result, the importance of income as a voting cue will be reduced in the minds of voters (Evans
and Tilley 2012a). Importantly, it is risky for parties to downplay economic issues in their
policy platforms in the face of high inequality given that this entails a high salience of
economic issues in voters’ perception and polarized preferences over redistribution (Spoon
and Klüver 2014).
However, political constraints can discourage rightist parties from moving to the left
on the economic policy dimension. Party elites or activists usually have heterogeneous policy
preferences, and by using some degree of their possessed power they may exert influence on
important decision making of the party. In general, the intraparty decision-making process in
advanced democracies are well-institutionalized through prescribed party regulations. This

19

Tavits and Letki (2014) predict that increasing inequality is positively associated with party-system
polarization in post-communist democracies, but it is not clear from their study why right-wing parties
move further rightward on the economic dimension as inequality increases. The reason is that
increasing party polarization within the economic dimension is likely to create an attraction of voters’
attention toward economic issues, and as a consequence, increase their income-based voting (e.g.,
Evans and Tilley 2012; Green and Hobolt 2008; Hobolt and de Vries 2015; Miller and Schofield
2003; Wagner 2012). In a similar vein, Meguid (2005, 2008) also points out that mainstream party’s
decision to take a position on the opposite side of the issue taken by a niche party increases the
perceived relevance of that issue, in turn, enhancing the niche party’s fortune in electoral competition.
The argument that inequality and party polarization are positively associated with each other is
developed in the context of established democracies where citizens have relatively clear information
on the policy positions of individual parties, and where links between voters and well-institutionalized
parties are established based on programmatic ties (e.g., Garand 2010; Pontusson and Rueda 2008).
However, this condition is less likely to hold in most nascent democracies, especially if citizens have
not had any experience in electoral competition and if most parties of the regime are less
institutionalized or newly founded after democratization. In this sense, it is more reasonable to
speculate that as inequality increases, right-wing parties want to move toward the position of the
median voter in order to appeal to a larger segment of citizens, while downplaying the significance of
economic cleavage.
74

constraint makes it difficult for party leaders to make opportunistic decisions in regard to
party positioning within the economic dimension in response to contextual factors. On the
other hand, party leadership in emerging democracies tends to exert much stronger control
over the party’s important decisions. In particular, strong autonomy of party leadership is a
feature that is more prevalent in center-rightist parties in younger democracies (Enyedi and
Linek 2008). In addition, weak voter-party linkages in new democracies allow parties more
freedom in changing their policy positions.20
Parties in advanced democracies must also endure the risk of losing their reputation
when they attempt opportunistic position-taking. While right-wing parties should calculate
the cost (i.e., losing policy credibility and general reputation) and benefit (i.e., proximal
gains) expected from their policy shifts toward the left, opportunistic position-taking will not
produce an effective outcome once the voters become aware of the low credibility of the
election promise (Kitshelt and Rehm 2015). In advanced democracies, in fact, parties that
moderate their policy positions between elections are likely to suffer falls in their vote shares
(Somer-Topcu 2015). Therefore, it is more reasonable for the rightist parties in advanced
democracies to seek alternative options to enhance their electoral fortunes. Rightist parties in
emerging democracies, in contrast, face relatively weaker reputational constraints, thanks to
meager experience in democratic elections and lack of policy reputation (Dalton and
Klingemann 2007). Parties in emerging democracies are not as strongly constrained by their
past issue positions as their counterparts in advanced democracies, and citizens in emerging
democracies are also less likely to have solid beliefs or information on the parties’ policy
positions. This weak reputational constraint imposed in the emerging democracies allows
rightist parties to utilize a more effective option in the electoral competition.

20

Dalton and McAllister (2015) find that parties in advanced democracies are less likely to change
their policy positions between elections than those in new democracies.
75

In sum, I predict that the varying degrees of the two political constraints –
organizational and reputational – in advanced and emerging democracies incentivize rightist
parties to behave differently in response to greater economic disparity. Specifically, the
difference between rightist and non-rightist parties in the degree in which they emphasize
social issues will grow in tandem with inequality in advanced democracies. In nascent
democracies, on the other hand, high levels of inequality that challenge the rightist parties
incentivizes them to take more leftist policy positions in the economic dimension. An
integrative understanding of these relations leads to the formulation of the main hypotheses
set forth below regarding the different effects that inequality have on the strategies of rightist
parties in old and new democracies.

Hypothesis 3-1 (Changing Playing Field in Old Democracies): In comparison with nonrightist parties, rightist parties in old democracies should, all other things being equal, put
greater emphasis on issues pertaining to the social dimension when economic inequality is
high.
Hypothesis 3-2 (Changing Position in New Democracies): In comparison with non-rightist
parties, rightist parties in new democracies should, all other things being equal, position
themselves to the left in the economic dimension when economic inequality is high.

3.4. Data, Measurements, and Model Specification
3.4.1. Data and Cases
To test the main hypotheses, I take the information on parties’ platforms from the
Comparative Manifesto Project (CMP) dataset (Volkens et al. 2015). The CMP data, which
covers more than 50 countries, contains quantified information regarding the relative
attention parties pay to individual issue categories. Since the CMP data also includes the
76

information on parties in non-democracies, elections that were held in non-democratic years
are excluded according to classifications by the Boix-Miller-Rosato (Boix et al. 2013) and the
Cheibub-Gandhi-Vreeland (Cheibub et al. 2010) datasets. Specifically, elections held in years
that are classified as non-democracy in either dataset are dropped from the analysis.
Furthermore, small parties are unlikely to draft their policy platforms based on office-seeking
motivation when faced with environmental changes (Pedersen 2012), and thus parties with
less than three percent of total votes are also excluded.
The main hypotheses expect that the effect of inequality will be different in old and
new democracies. Therefore, all observations in this analysis are split into two groups.
Following Tavits and Potter (2015), five consecutive country-elections after democratization
are coded as new democracies. In cases where countries had experienced authoritarian
interruption between democratic episodes, five consecutive democratic elections after three
or more years of authoritarian rule are classified as elections in new democracies.21

3.4.2. Dependent Variables: Issue Emphasis and Issue Positioning
Based on the frequency of rhetoric in electoral platforms, the CMP provides the
information that enables measuring of a party’s emphasis on individual issue areas. In order
to test the main hypotheses, two dependent variables are constructed: (1) the extent to which
each party gives importance to social issues, and (2) party positions within the economic
dimension. I measure party emphasis on social issues using the sum of scores which measure
the parties’ relative attention to 16 policy categories related to social issues (Tavits and Potter
2015). I also identify 25 policy categories related to economic issues in both old and new

21

Termination of democracy due to foreign occupation during the Second World War is not take into
account here. List of country-elections in the sample and details on how they are classified into old
and new democracies are presented in Table 3.8 in the Appendix to this chapter.
77

democracies.22 To measure left-right positions in the economic dimension, the sum of scores
given to 15 leftist issues are subtracted from the total scores from the 10 rightist policy
categories.23 For this measure, smaller values indicate positions that are more left-leaning.

3.4.3. Independent Variables
I construct three key independent variables: right, inequality, and the interaction term
to test the moderation effect of inequality on both issue positioning and issue emphasis by
right-wing parties. Based on the CMP, liberal, Christian democratic, conservative, or
nationalist parties are classified as rightist parties consistent with previous studies (Burgoon
2013; Tavits and Potter 2015). In order to measure the level of economic inequality during
each election, the gross (pre-tax/transfer) Gini coefficients are taken from the Standardized
World Income Inequality Database (SWIID; Solt 2009). The Gini coefficients from the years
prior to the election years are used to for the elections held during the first half of the election
year.24

3.4.4. Control Variables
I also include a set of control variables which may potentially affect the estimation of
the moderation effect suggested by the main hypotheses. First of all, ethno-linguistic and
religious fractionalizations are argued to have an effect on both party decisions – on issue
positioning and issue emphasis. In societies with high levels of ethnic fractionalization,
citizens are more likely to support less redistribution since individuals with affiliation to
specific ethnic groups do not wish to spend revenue on other ethnic groups (Alesina and
22

Valence issues (Stokes 1963; e.g., economic growth) are not used in constructing the dependent
variable that measures party positions on economic issues.
23
Policy categories used to construct the dependent variables are presented in Table 3.9 and 3.10 in
the Appendix to this chapter.
24
As for the elections held after June 30th, the Gini coefficients in election years are used.
78

Glaeser 2004). Support for leftist policies also tends to decrease when ethnic fractionalization
increases because even those who favor redistribution are influenced by their antipathy to
other ethnic groups (Finseraas 2009b). As for the strategy of emphasis on social issues,
parties are more likely to appeal to social issues, including ethno-linguistic identity and
religiosity, when such social cleavages are present. I therefore include ethnic, linguistic, and
religious fractionalization measures in the statistical models to control for their rightward (or
positive) effects on position-taking in regard to economic issues (or emphasis on social
issues). Fractionalization indexes are taken from Alesina et al. (2003).
Economic globalization is another factor that is argued to have an effect on the
election platforms. Previous studies have also examined the effect of economic globalization
on policy positioning, but the resulting evidence is mixed in terms of the direction of effect.
Various economic indicators that measure economic globalization tend to have significant
effects on party’s policy position, but the direction of effect varies depending on which
indicator is used (Adam et al. 2009; Haupt 2010). While economic globalization is claimed to
push parties to converge on neoliberal economic policies by creating competitive pressure
(Steiner and Martin 2012; Ward et al. 2015), it should also be noted that the debate on the
neoliberal convergence is far from settled (e.g., Garrett and Mitchell 2001; Plümper et al.
2009; Rodrik 1998). For this reason, I remain open to the direction of effect that economic
globalization has on positions within the economic policy dimension. At the same time, Ward
et al. (2015) demonstrate that economic globalization is positively associated with the party
implementation of emphasis on non-economic issues during elections. Following this
argument, I predict that economic globalization intensifies parties’ emphasis on social issues
in electoral platforms. The measure for economic globalization is taken from the KOF Index
of Globalization (Dreher 2006; Dreher et al. 2008).
The effect of unemployment rates on party positioning is unclear. Intuitively, as
79

unemployment rate rises, parties would propose more aggressive labor market policies and
promise greater job protection. However, as Petrocik (1996) posits, a right-wing party may
have its own way to deal with unemployment. For example, a rightist party may address the
problem of a growing unemployment rate by stressing “the importance of stimulating
business opportunities through investment credits and less regulations” (Petrocik 1996: 829).
To measure the unemployment rates, information on harmonized unemployment rates for
advanced democracies are drawn from the Comparative Welfare States data set (Brady et al.
2014). I utilize the information from OECD and the World Bank’s World Development
Indicators (WDI) database to fill in the missing observations. I also control for the three
party-election level variables (new party, niche party, and party size) and the three additional
election-level variables (GDP growth, turnout rates, and the effective number of electoral
parties) following Tavits and Potter (2015).25

3.4.5. Model Specification and Estimation Strategy
I construct models with the two dependent variables to test the main hypotheses. The
data structure is hierarchical, as party-election level observations are clustered at the party
level as well as the country-election level. Both clusters are nested within each country.
While the dependent variables are measured at the party-election level, the key independent
variables, right and inequality, are measured at the party level and country-election level,
respectively. Ignoring this data structure can lead to overconfidence in the accuracy of the
estimates (Angrist and Pischke 2009; Steenbergen and Jones 2002), so I estimate multilevel
linear models while allowing random intercepts both at the country and party levels with

25

All economic variables, including unemployment rates, economic globalization index, and annual
GDP growth rates, are lagged for elections held in the first half of the election years. Details regarding
the coding schemes of these control variables are presented in Table 3.11 in the Appendix to this
chapter.
80

cluster-robust standard errors at the country level as was done in previous studies (Tavits and
Potter 2015; Ward et al. 2015). Observations with at least one missing value for any
independent variables are dropped out, and therefore the main hypotheses are examined using
1754 observations from 475 parties in 44 democracies. Specifically, the statistical models are
presented as follows:

(1) Emphasis	 on	 Social	 Issues

	

(2) Position	 on	 Economic	 Issues

where
variances,

and

	

are country- and party-level random intercepts of which estimates of
and

, will be derived. The main goal is to estimate

, which represents

how discernable issue emphasis or issue positioning of right-wing parties is from that of nonrightist counterparts given the level of economic inequality. As for the first model that uses
issue emphasis as the dependent variable, it is expected that as inequality increases right-wing
parties in old democracies will lay more emphasis on social issues than other parties.
However, the rightist parties in younger democracies are expected to take more leftist
positions when challenged by increasing inequality. Therefore, while the sign of
first model should be positive for advanced democracies,
have a negative sign only for emerging democracies.

81

in the

in the next model is expected to

3.5. Estimation Results
3.5.1. Main Analysis
The estimation results for the two main models are presented in Table 3.1. The results
of the first two columns confirm my hypotheses about issue emphasis of right-wing parties in
response to the different levels of inequality. The coefficient for the interaction term between
right and inequality statistically significant (p < .01). This implies that in advanced
democracies rightist parties are inclined to emphasize social issues more than non-rightist
parties when economic inequality rises. Substantively, an increase in the Gini coefficient by 1
is associated with an increase in the difference in emphasis on social issues between rightist
and non-rightist parties by 0.301. This result is in line with the first hypothesis. On the
contrary, the effect of economic inequality on party’s emphasis on social issues is not
statistically significant in new democracies. In terms of the magnitude, the coefficient for the
interaction term is not much smaller than that in advanced democracies, but the coefficient is
not precisely estimated. On the other hand, the estimation results in the rightmost column
show the rightist parties takes a closer position to the left as inequality rises. The interaction
term between right and inequality is statistically significant (p < .05), and as a result, the
difference between rightist and non-rightist parties in new democracies regarding their
positions on economic issues decreases with increasing economic inequality.
Even though the directions of coefficients on multiplicative interaction terms are
consistent with the theoretical predictions and the coefficients are precisely estimated, it is
also necessary to calculate the effects of right and their standard errors in order to enhance
the validity of inference and substantive interpretations (Brambor et al. 2006). To this end, I
create graphs to display the effect of right on each dependent variable depending on the
different levels of inequality. Based on the estimation results, Figure 3.1 and Figure 3.2
graphically present the effect of right over all possible ranges of inequality. The solid line in
82

each graph represents the point estimate of the effect and the dashed lines indicate the 95%
confidence interval. The above graph in Figure 1 shows that the difference between rightist
and non-rightist parties in their emphasis on social issues is not statistically significant when
the Gini coefficient is less than 30. However, as inequality grows, the gap between the two
groups consistently increases throughout the ranges of inequality level. The bottom graph in
Figure 3.1 plots the effect of right on the emphasis on social issues in new democracies. At
all possible values of inequality, the effect of right holds values that are not significantly
different from zero. More importantly, the solid line shows an upward inclination overall, but
we cannot conclude that the effect of right consistently increases or decreases over the range
of inequality levels as there exists great uncertainty in the estimates. On the other hand, the
bottom graph in Figure 3.2 shows that the difference between rightist and non-rightist parties
within the economic dimension in new democracies consistently decreases with inequality.
While the right-wing’s position within the economic dimension is clearly distinct from that of
other parties when inequality is low, the difference slopes downward with the rise of
inequality and effectively diminishes as inequality approaches its highest level. This
moderation effect is non-existent in advanced democracies. The above graph in Figure 3.2
shows that right-wing parties take more conservative positions in the economic dimension
throughout the entire range of inequality levels.
The findings strongly support the main hypotheses. Higher economic inequality
encourages right-wing parties to make attempts to turn electoral competition to their
advantage, albeit in different manners. The results confirm that specific strategies of the rightwing groups diverge according to different environmental incentives in advanced and
emerging democracies.

83

Table 3.1. Regression Analysis of Emphasis on Social Issues and Positions on Economic
Issues in Advanced and Emerging Democracies

Constant
Right
Economic
Inequality
Right
Inequality
GDP Growth
Unemployment
Economic
Globalization
Effective Number
of Parties
Turnout Rates
Niche
Size
New Party
Ethnic
Fractionalization
Religious
Fractionalization
Linguistic
Fractionalization

N (Country)
N (Party)
N

Model 1:
Emphasis on
Social Issues
(Old)
7.629
(7.095)
-7.101*
(3.256)
-0.186
(0.100)
0.301**
(0.077)
0.017
(0.050)
-0.036
(0.093)
0.169**
(0.052)
-0.512
(0.342)
-0.000
(0.036)
3.449**
(1.282)
-1.497
(1.855)
0.077
(0.509)
-3.004
(7.063)
0.926
(3.791)
11.236
(10.808)
13.678
32.231
40.257
36
310
1359

Model 2:
Emphasis on
Social Issues
(New)
9.915
(7.719)
-6.441
(5.428)
-0.073
(0.116)
0.189
(0.143)
0.036
(0.099)
-0.150
(0.084)
0.015
(0.055)
0.177
(0.268)
0.023
(0.047)
5.063**
(1.301)
-1.689
(1.800)
0.042
(0.872)
5.478
(4.635)
5.403
(4.088)
2.502
(4.506)
2.404
7.116
45.976
23
229
395

Model 3:
Positions on
Economic
Issues (Old)
-18.028*
(8.708)
3.699
(6.693)
-0.345**
(0.129)
0.211
(0.159)
-0.332*
(0.164)
0.363*
(0.160)
0.012
(0.079)
0.468
(0.546)
0.073
(0.063)
-0.810
(2.008)
4.947
(3.120)
2.664*
(1.232)
-6.706
(8.439)
2.054
(3.335)
12.558
(7.487)
12.901
82.580
105.505
36
310
1359

Model 4:
Positions on
Economic
Issues (New)
-4.496
(15.998)
18.973**
(5.127)
0.108
(0.175)
-0.265*
(0.133)
-0.023
(0.147)
0.298*
(0.146)
-0.179*
(0.080)
0.352
(0.319)
-0.086
(0.101)
-0.138
(1.482)
2.574
(3.190)
-0.031
(1.090)
-4.587
(7.121)
-3.192
(7.508)
-1.720
(5.251)
13.703
37.763
46.305
23
229
395

*p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

84

Figure 3.1. Effect of Right on Emphasis on Social Issues Conditional on the Level of Inequality
in Advanced and Emerging Democracies

3

5

2

0

|

|||| | || ||| | ||||||||||||||||||||||||||||||||||
|||||||||||||||||||
|||||||||
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||| ||||||||||||||||||| ||||||||| ||| ||

25

30

35

40

45

50

| | | || |

||

|

0

|

||

|

0

-10

1

-5

Effect of Right

4

Percent of Observations

10

5

6

15

Advanced Democracies

55

Level of Inequality

3

5

2

0
-10

1

-5

Effect of Right

4

Percent of Observations

10

5

6

15

Emerging Democracies

|

|

|||| | || ||| | ||||||||||||||||||||||||||||||||||
|||||||||||||||||||
|||||||||
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||| ||||||||||||||||||| ||||||||| ||| ||

25

30

35

40

45

50

| | | || |

55

Level of Inequality

Based on Model 1 and Model 2 of Table 1. Dashed lines are 95 percent confidence intervals.

85

Figure 3.2. Effect of Right on Positions on Economic Issues Conditional on the Level of
Inequality in Advanced and Emerging Democracies

3

10

2

5

|

|||| | || ||| | ||||||||||||||||||||||||||||||||||
|||||||||||||||||||
|||||||||
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||| ||||||||||||||||||| ||||||||| ||| ||

25

30

35

40

45

50

| | | || |

||

|

0

|

||

|

0

-5

1

0

Effect of Right

4

Percent of Observations

15

5

6

20

Advanced Democracies

55

Level of Inequality

3

10

2

5
-5

1

0

Effect of Right

4

Percent of Observations

15

5

6

20

Emerging Democracies

|

|

|||| | || ||| | ||||||||||||||||||||||||||||||||||
|||||||||||||||||||
|||||||||
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||| ||||||||||||||||||| ||||||||| ||| ||

25

30

35

40

45

50

| | | || |

55

Level of Inequality

Based on Model 3 and Model 4 of Table 1. Dashed lines are 95 percent confidence intervals.

86

3.5.2. Testing the Causal Mechanism: Do Political Constraints Really Matter?
Although the key findings in the previous section are consistent with the main
hypotheses regarding different responses of right-wing parties to economic inequality, the
results should still be interpreted with caution because it is possible these results were
produced by mechanisms other than the varying degrees of political constraints. It is therefore
important to conduct an additional investigation that tests the moderation effect of the
political constraints in order to substantiate the suggested causal mechanism. To capture the
degree of political constraints, I draw four continuous measures that are related to the extent
to which the autonomy of party leadership is constrained from the Varieties of Democracy (VDem) database (Coppedge et al. 2016). Next, I create an additional independent variable
(constraint) to measure the degrees of political constraints by adding up the values of those
four variables.26 To test whether the moderation effect of inequality is conditioned on the
degree of political constraints, I estimate the following two equations with the three-way
interaction term between right, inequality, and constraint. As Braumoeller (2004) and
Brambor et al. (2006) prescribe, all constitutive terms of the three-way interaction term are
included in the equations as follows:

(3) Emphasis	 on	 Social	 Issues
	

	

26

The mean and standard deviation of constraint are 6.997 and 2.641 respectively, and this variable
ranges from -1.589 to 11.988. Higher values indicate stronger constraints. The prime and unique
advantage of employing measures from the V-Dem is that the measures change over time, thus
reflecting the development of party institutions and programmatic ties between voters and parties as a
democracy matures. Table 3.12 in the Appendix to this chapter lists the indicators used to construct
constraint.
87

(4) Position	 on	 Economic	 Issues
	

	

The estimation results of the models are presented in Table 3.2, but the estimated
coefficients and their statistical significances tell little about the moderation effects of
political constraints. Instead, I estimated the effect of right on each dependent variable
conditional on the level of inequality at different degrees (i.e., strong or weak) of political
constraints based on the coefficient estimates in the two equations.27
Figure 3.3 provides a depicted explanation of the estimated effect of right on the
emphasis on social issues throughout all possible ranges of inequality under strong and weak
political constraints. The first graph plots the conditional effect of inequality under strong
constraints, which shows an upward slope that is quite similar to that observed from the effect
of right in advanced democracies in Figure 3.1. As inequality grows, right-wing parties place
increasingly greater emphasis on social issues compared to non-rightist parties under strong
political constraints. The solid line representing the effect of right is almost flat in the bottom
graph of Figure 3.3, in contrast, and the difference in social issue emphasis between rightist
and non-rightist parties is not statistically significant over the entire range of inequality.
Figure 3.4 plots the effect of right on positions within the economic dimension at
different levels of inequality under strong and weak political constraints. The bottom graph

27

The values for strong and weak constraints are set to the integers closest to the value of one
standard deviation away from the maximum and minimum of constraint (i.e., strong=9 and weak=1).
88

Table 3.2. Regression Analysis of Emphasis on Social Issues and Positions on Economic
Issues Depending on Inequality Level and Degree of Political Constraints

Constant
Right
Economic Inequality
Political Constraints
Right

Inequality

Right

Constraints

Inequality

Constraints

Right
Inequality
Constraints
GDP Growth
Unemployment
Economic Globalization
Effective Number of Parties
Turnout Rates
Niche
Size
New Party
Ethnic Fractionalization
Religious Fractionalization
Linguistic Fractionalization

N (Country)
N (Party)
N

Model 5:
Emphasis on Social Issues
10.368
(7.626)
-1.313
(6.089)
-0.131
(0.138)
-0.247
(0.998)
-0.029
(0.161)
-0.412
(0.843)
-0.003
(0.025)
0.034
(0.021)
0.019
(0.050)
-0.016
(0.079)
0.115**
(0.042)
-0.268
(0.212)
0.013
(0.026)
4.480**
(1.033)
-1.201
(1.421)
0.551
(0.469)
2.283
(4.741)
-1.215
(3.405)
9.695
(7.950)
9.249
23.333
42.964
43
466
1726

Model 6:
Position on Economic Issues
-7.111
(14.224)
26.229**
(9.563)
-0.328
(0.280)
-0.640
(1.754)
-0.475*
(0.237)
-2.538
(1.524)
0.009
(0.038)
0.079*
(0.036)
-0.238*
(0.121)
0.336**
(0.115)
-0.019
(0.059)
0.288
(0.380)
0.013
(0.053)
-0.395
(1.463)
4.746*
(2.310)
1.557
(0.874)
-4.125
(6.687)
-0.905
(3.433)
8.134
(5.853)
13.078
66.017
97.378
43
466
1726

*p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

89

Figure 3.3. Effect of Right on Emphasis on Social Issues Conditional on the Level of
Inequality under Different Degrees (Strong or Weak) of Political Constraints

3

5

2

0

|

|||| | || ||| | ||||||||||||||||||||||||||||||||||
|||||||||||||||||||
|||||||||
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||| ||||||||||||||||||| ||||||||| ||| ||

25

30

35

40

45

50

| | | || |

||

|

0

|

||

|

0

-10

1

-5

Effect of Right

4

Percent of Observations

10

5

6

15

Strong Constraints

55

Level of Inequality

3

5

2

0
-10

1

-5

Effect of Right

4

Percent of Observations

10

5

6

15

Weak Constraints

|

|

|||| | || ||| | ||||||||||||||||||||||||||||||||||
|||||||||||||||||||
|||||||||
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||| ||||||||||||||||||| ||||||||| ||| ||

25

30

35

40

45

50

| | | || |

55

Level of Inequality

Based on Model 5 of Table 3.2. Dashed lines are 95 percent confidence intervals.

90

Figure 3.4. Effects of Right on Position on Economic Issues Conditional on the Level of
Inequality under Different Degrees (Strong or Weak) of Political Constraints

3

10

2

5

|

|||| | || ||| | ||||||||||||||||||||||||||||||||||
|||||||||||||||||||
|||||||||
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||| ||||||||||||||||||| ||||||||| ||| ||

25

30

35

40

45

50

| | | || |

||

|

0

|

||

|

0

-5

1

0

Effect of Right

4

Percent of Observations

15

5

6

20

Strong Constraints

55

Level of Inequality

3

10

2

5
-5

1

0

Effect of Right

4

Percent of Observations

15

5

6

20

Weak Constraints

|

|

|||| | || ||| | ||||||||||||||||||||||||||||||||||
|||||||||||||||||||
|||||||||
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||| ||||||||||||||||||| ||||||||| ||| ||

25

30

35

40

45

50

| | | || |

55

Level of Inequality

Based on Model 6 of Table 3.2. Dashed lines are 95 percent confidence intervals.

91

shows that the effect of right consistently decreases as inequality increases under weak
political constraints and the effect effectively diminishes when Gini coefficient is above 48.
On the other hand, when strong political constraints restrict the autonomy of party leadership
in the decision-making process, the effect of right slightly increases with the rise of
inequality. In this case, however, the conditional effect of inequality fails to reach the
conventional levels of statistical significance, and the estimated position of right-wing parties
is more conservative than non-rightist parties regardless of inequality levels. The results are
in line with the estimation results from the main analysis. When faced with strong political
constraints, right-wing parties attempt to politicize social issues as inequality rises. Rightwing parties take more leftist positions within the economic dimension as inequality rises
under weak political constraints; however, there is no discernable effect of right on the
strategy of social issues emphasis. These findings validate the suggested causal mechanism,
which links the degree of political constraints with the different responses of right-wing
parties to economic inequality in advanced and nascent democracies.

3.5.3. Robustness Tests
The results that support my theoretical arguments may be driven by particular
measurements or model specifications. In order to ensure that the results are not products of
those particular choices, I conduct a series of robustness checks. First, I adopt different
criteria to classify old and new democracies. I split elections into two groups with a criterion
of five consecutive democratic elections. As this process is intrinsically arbitrary, I estimate
the same models with both expanded (six consecutive elections) and restricted (four
consecutive elections) categories of new democracies in order to ensure that this arbitrary
procedure does not affect the results. The estimation results are presented in Table 3.3. While

92

Table 3.3. Robustness Checks 1: Alternative Classifications of Old and New Democracies
(Four or Six Elections)
Model 7:
Emphasis
on Social
(Old, 4
Elections)

Model 8:
Emphasis
on Social
(New, 4
Elections)

Model 9:
Position
on Econ.
(Old, 4
Elections)

Model 10:
Position
on Econ.
(New, 4
Elections)

Model 11:
Emphasis
on Social
(Old, 6
Elections)

Model 12:
Emphasis
on Social
(New, 6
Elections)

Model 13:
Position
on Econ.
(Old, 6
Elections)

Model 14:
Position
on Econ.
(New, 6
Elections)

Constant

6.442
(6.507)

13.603
(8.584)

2.419
(17.799)

7.586
(7.889)

11.534
(6.477)

-14.275
(10.292)

-8.023
(12.837)

Right

-7.213
(2.869)

-5.991
(5.806)

16.107*
(8.016)
3.941
(6.165)

-7.239*
(3.295)

-4.996
(4.680)

4.012
(7.130)

Economic
Inequality

-0.170
(0.100)

-0.124
(0.148)

-0.201
(0.103)

-0.073
(0.091)

-0.374*
(0.145)

19.591*
*
(4.396)
0.129
(0.135)

Right
Inequality
GDP
Growth
Unemplo
yment
Economic
Global.

0.293**
(0.070)
0.075
(0.053)
-0.002
(0.091)
0.159**
(0.049)

0.168
(0.150)
0.034
(0.106)
-0.133
(0.090)
0.021
(0.062)

0.360**
(0.126)
0.197
(0.148)
-0.291*
(0.144)
0.360*
(0.146)
0.004
(0.074)

21.427*
*
(5.549)
0.029
(0.222)
-0.338*
(0.149)
-0.035
(0.175)
0.177
(0.185)
-0.212*
(0.088)

0.306**
(0.079)
-0.003
(0.053)
-0.025
(0.102)
0.174**
(0.055)

0.166
(0.122)
0.030
(0.094)
-0.123
(0.079)
0.013
(0.046)

0.217
(0.167)
-0.332
(0.191)
0.460*
(0.182)
0.008
(0.082)

Effective
N of
Parties
Turnout
Rates
Niche

-0.414
(0.318)

0.107
(0.273)

0.624
(0.460)

-0.036
(0.412)

-0.545
(0.393)

0.104
(0.258)

0.556
(0.558)

-0.285*
(0.111)
0.050
(0.109)
0.278*
(0.116)
0.205**
(0.071)
0.446
(0.310)

0.006
(0.031)
3.628**
(1.260)
-1.096
(1.727)
-0.031
(0.491)
-3.133
(6.612)
1.799
(3.356)
10.291
(10.130)
11.942
29.323
39.871
40
347
1453

0.008
(0.052)
5.372**
(1.290)
-3.686
(2.016)
0.035
(1.118)
7.551
(4.662)
4.019
(3.327)
2.011
(5.212)
1.011
5.523
52.519
22
189
301

0.066
(0.058)
-0.442
(1.827)
5.101
(2.792)
2.422*
(1.043)
-4.254
(8.015)
1.247
(3.347)
8.163
(6.907)
13.680
76.671
102.076
40
347
1453

-0.093
(0.112)
-0.305
(1.888)
2.959
(3.241)
-0.344
(1.315)
-1.907
(5.952)
-0.468
(7.299)
-2.692
(5.103)
9.701
50.180
40.747
22
189
301

0.006
(0.042)
3.561**
(1.322)
-2.488
(1.884)
-0.081
(0.575)
-2.322
(8.391)
1.305
(4.083)
10.883
(12.985)
15.681
32.958
40.826
31
274
1287

0.006
(0.037)
5.181**
(1.357)
-1.209
(1.833)
0.007
(0.855)
4.359
(4.923)
3.667
(3.989)
3.925
(4.509)
2.203
9.389
44.607
23
240
467

0.036
(0.079)
-0.686
(2.260)
4.744
(3.357)
2.349
(1.275)
-8.137
(9.300)
1.909
(3.715)
12.271
(8.335)
15.911
86.797
106.665
31
274
1287

-0.038
(0.089)
0.003
(1.098)
1.479
(2.899)
0.210
(0.957)
-0.250
(6.137)
-4.420
(7.111)
-0.216
(4.992)
12.303
31.449
53.262
23
240
467

Size
New
Party
Ethnic
Frac.
Religious
Frac.
Linguistic
Frac.

N (Country)

N (Party)
N

*p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

93

Table 3.4. Robustness Checks 2: Alternative Classifications of Old and New Democracies
(Years of Democratization)

Constant
Right
Economic
Inequality
Right 	
Inequality
GDP Growth
Unemployment
Economic
Globalization
Effective N of
Parties
Turnout Rates
Niche
Size
New Party
Ethnic
Fractionalization
Religious
Fractionalization
Linguistic
Fractionalization

N (Country)
N (Party)
N

Model 15:
Emphasis on
Social Issues
(Old,
Democracy
b/a 1980)
8.009
(7.315)
-7.622*
(3.021)
-0.219*
(0.091)
0.312**
(0.071)
-0.081
(0.071)
-0.078
(0.112)
0.169**
(0.052)
-0.522
(0.405)
0.013
(0.041)
3.284**
(1.179)
-0.611
(1.759)
0.455
(0.569)
-0.285
(8.374)
3.064
(4.368)
9.132
(13.561)
20.184
30.523
40.545
25
252
1320

Model 16:
Emphasis on
Social Issues
(New,
Democracy
b/a 1980)
9.927
(6.704)
-2.159
(4.593)
-0.008
(0.084)
0.092
(0.122)
0.136*
(0.058)
-0.059
(0.084)
0.027
(0.042)
0.021
(0.218)
0.012
(0.043)
4.758**
(1.823)
-5.034**
(1.761)
-0.192
(0.871)
1.493
(4.043)
0.719
(3.221)
5.436
(3.662)
0.014
8.954
50.477
19
223
434

Model 17:
Position on
Economic Issues
(Old,
Democracy
b/a 1980)
-12.956
(9.449)
4.905
(6.512)
-0.399**
(0.135)
0.201
(0.153)
-0.271
(0.174)
0.441*
(0.202)
0.003
(0.082)
0.615
(0.545)
0.030
(0.073)
-0.502
(2.397)
4.994
(3.062)
1.598
(1.158)
-10.702
(8.961)
5.726
(3.339)
10.416
(8.395)
13.136
88.019
104.728
25
252
1320

Model 18:
Position on
Economic Issues
(New,
Democracy
b/a 1980)
-12.984
(13.696)
22.775**
(4.252)
0.232
(0.127)
-0.359**
(0.105)
-0.031
(0.134)
0.143
(0.139)
-0.159*
(0.078)
0.168
(0.321)
-0.073
(0.112)
-0.069
(1.287)
2.652
(3.194)
1.232
(1.298)
2.970
(8.802)
-2.071
(7.971)
-0.658
(6.731)
17.519
28.711
62.487
19
223
434

* p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

94

Table 3.5. Robustness Checks 3: Alternative Setup of Three-level Hierarchy (Country and
Country-election Clusters)

Constant
Right
Economic Inequality
Right

Inequality

GDP Growth
Unemployment
Economic Globalization
Effective N of Parties
Turnout Rates
Niche
Size
New Party
Ethnic Fractionalization
Religious
Fractionalization
Linguistic
Fractionalization

N (Country)
N (Country-Election)
N

Model 19:
Emphasis on
Social Issues
(Old)

Model 20:
Emphasis on
Social Issues
(New)

5.701
(5.916)
-6.081*
(2.413)
-0.110
(0.103)
0.256**
(0.062)
0.029
(0.043)
-0.005
(0.090)
0.135**
(0.050)
-0.234
(0.290)
-0.011
(0.029)
3.101*
(1.250)
-0.036
(1.689)
0.321
(0.731)
-1.584
(7.079)
1.673
(3.076)
10.789
(10.318)
15.611
5.320
56.579
36
260
1359

11.500
(7.720)
-6.736
(4.825)
-0.080
(0.110)
0.196
(0.129)
0.033
(0.091)
-0.146
(0.081)
0.012
(0.050)
0.202
(0.269)
0.007
(0.047)
5.191**
(1.275)
-1.398
(1.938)
-0.211
(0.851)
5.390
(4.829)
5.063
(3.969)
2.539
(4.766)
2.229
4.393
49.112
23
79
395

Model 21:
Position on
Economic
Issues
(Old)
-13.398**
(8.281)
-0.610
(5.412)
-0.351**
(0.116)
0.366**
(0.124)
-0.375*
(0.164)
0.285*
(0.131)
-0.058
(0.071)
0.053
(0.593)
0.102
(0.056)
1.639
(2.174)
3.814
(3.610)
2.647
(1.408)
-8.600
(8.712)
3.142
(3.638)
15.446*
(7.724)
15.580
19.613
170.010
36
260
1359

Model 22:
Position on
Economic
Issues
(New)
1.135
(18.248)
17.751**
(4.059)
0.081
(0.176)
-0.234*
(0.103)
-0.000
(0.139)
0.288*
(0.139)
-0.186*
(0.077)
0.292
(0.282)
-0.115
(0.119)
0.115
(1.243)
-1.349
(2.439)
-0.868
(1.219)
-4.605
(7.779)
-4.027
(8.009)
-2.587
(5.361)
16.671
0.177
77.971
23
79
395

*p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

there are slight changes in the magnitudes of coefficients for the interaction terms, all
estimation results are not substantively different from the results of the main analysis. In

95

addition, I classify old and new democracies with an alternative criterion based on the year of
democratization. Specifically, I classify countries whose current democratic episode began in
1979 or afterward as new democracies (Scarrow 2010). A list of old democracies includes
Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland,
Ireland, Israel, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal,
Romania, Spain, Sweden, Switzerland, the United Kingdom, and the United States. Albania,
Bulgaria, Croatia, Cyprus, Czech Republic, Estonia, Georgia, Hungary, Latvia, Lithuania,
Macedonia, Mexico, Moldova, Poland, Serbia, Slovakia, Slovenia, South Korea, Turkey, and
Ukraine are coded as new democracies. Table 3.4 presents the estimation results, and the
main results are robust.
Next, I introduce an alternative random intercept into the statistical model. The data
structure can be understood as hierarchical, as party-election level observations are clustered
at the country-election level, instead of the party level, and those country-election clusters are
nested within each country. I also estimate the same model with an alternative setting of a
hierarchical data structure (i.e., the country and country-election levels) in order to entertain
the possibility that this particular setup of data structure affects the results. The estimation
results of the multilevel linear models are presented in Table 3.5 and all the key results
remain robust.
I also exclude non-partisan parties (i.e., parties which are neither left-wing nor rightwing) and re-estimate the statistical models following Tavits and Potter (2015). Therefore, the
coefficient estimate on (right

inequality) represents the extent to which the difference in

the outcome variables between rightist and leftist (instead of non-rightist) parties in
conditional on economic inequality. The estimation results are presented in Table 3.6.
The estimation results of Model 3 in Table 3.1 indicate that non-rightist parties are more
likely to take leftist positions when inequality increases. This was not an expected result of
96

my theoretical discussion, and, therefore, I conducted an additional analysis to determine
which segment of non-rightist parties - left-wing (social democratic, communist, and
ecologist) parties or non-partisan (agrarian, ethnic-regional, special issues, and
coalition/alliance) parties – drives this effect. I excluded those non-partisan parties which do
not fall into the left-right dichotomy from the sample, and examined whether the effect of
inequality on leftist parties’ positions on economic issues can still be detected. The result
showed that the effect of inequality on the leftist parties is only half the magnitude of the
result yielded by an analysis on non-partisan parties and, moreover, the effect itself was not
statistically significant. Additionally, the difference in positions within the economic
dimension between leftist and rightist parties did not display meaningful changes over the
various ranges of inequality. Indeed, it can be noted that as inequality increases, those nonpartisan parties take more leftist positions. Thus, it is reasonable to conclude that a
statistically significant coefficient for inequality is a product of opportunistic position-taking
by parties that are neither leftist nor rightist.
Lastly, I estimate the two main equations using an alternative measure of niche.
While I believe that my measure of niche, which focuses on the salience theory, is more
appropriate to control for a “nicheness,” other empirical studies also use an alternative
definition and measure of niche based on the ideological extremity of parties (Adams et al.
2006; Tavits and Potter 2015). In particular, introducing this alternative measure into
estimation may affect the results for parties’ position-taking. Therefore, I redo the main
analysis using this alternative measure of niche. Specifically, I classify ecologist, communist,
and nationalist parties as niche parties based on the information on party family from the
CMP. The estimation results with this variable are presented in Table 3.7. The results still
render as strong support as the original results, so it is safe to conclude that the main results
are robust to the alternative measure of nicheness. The series of robustness tests of the main
97

Table 3.6. Robustness Checks 4: Excluding Non-left and Non-right Parties
Model 23:
Emphasis on
Social Issues
(Old)

Constant

7.040
(7.483)
Right
-6.367
(3.434)
Economic Inequality -0.184
(0.107)
0.290**
Right
Inequality
(0.080)
GDP Growth
0.032
(0.052)
Unemployment
-0.050
(0.090)
Economic
0.183**
Globalization
(0.055)
Effective N of Parties -0.541
(0.357)
Turnout Rates
-0.008
(0.036)
Niche
3.036*
(1.419)
Size
-1.636
(1.901)
New Party
-0.212
(0.522)
Ethnic
-3.421
Fractionalization
(7.485)
Religious
1.709
Fractionalization
(4.077)
Linguistic
10.164
Fractionalization
(11.531)
16.162
33.000
40.325
N (Country)
36
N (Party)
272
N
1197

Model 24:
Emphasis on
Social Issues
(New)

Model 25:
Position on
Economic
Issues
(Old)

Model 26:
Position on
Economic
Issues
(New)

3.058
(8.156)
-2.728
(5.534)
0.011
(0.127)
0.106
(0.144)
0.043
(0.102)
-0.167
(0.103)
0.029
(0.057)
0.371
(0.252)
0.045
(0.054)
3.633*
(1.442)
-1.283
(2.003)
-0.311
(1.093)
5.975
(5.465)
6.015
(4.318)
1.840
(4.467)
1.991
5.556
44.688
23
195
338

-26.354**
(8.059)
10.513
(7.273)
-0.241
(0.145)
0.134
(0.178)
-0.330*
(0.156)
0.269
(0.149)
-0.000
(0.067)
0.488
(0.573)
0.100
(0.056)
-0.618
(1.786)
7.200*
(3.012)
1.750
(0.937)
-4.298
(7.162)
2.545
(3.267)
9.658
(5.915)
11.084
49.488
101.842
36
272
1197

-18.375
(12.084)
26.020**
(5.837)
0.240
(0.172)
-0.406**
(0.146)
-0.008
(0.151)
0.310*
(0.137)
-0.113
(0.071)
0.369
(0.310)
-0.035
(0.073)
-1.868
(1.341)
3.937
(2.745)
0.625
(0.985)
-5.233
(7.496)
-5.921
(6.964)
0.961
(5.622)
11.497
22.122
46.832
23
195
338

Model 27:
Position on
Economic
Issues
(Old, Neither
Left nor
Right Only)
27.096
(38.847)
-0.574*
(0.259)
-0.633
(0.452)
0.843
(0.429)
0.026
(0.201)
-2.510*
(0.955)
-0.119
(0.256)
0.080
(5.014)
-12.564
(29.915)
10.156
(6.305)
-86.587**
(15.686)
2.834
(13.527)
89.237**
(24.397)
100.502
90.277
119.736
19
38
162

*p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

98

Table 3.7. Robustness Checks 5: Alternative Measure of Niche Party

Constant
Right
Economic Inequality
Right

Inequality

GDP Growth
Unemployment
Economic Globalization
Effective N of Parties
Turnout Rates
Niche
Size
New Party
Ethnic Fractionalization
Religious
Fractionalization
Linguistic
Fractionalization

N (Country)
N (Party)
N

Model 28:
Emphasis on
Social Issues
(Old, Niche)

Model 29:
Emphasis on
Social Issues
(New, Niche)

7.552
(7.208)
-7.825
(3.369)
-0.198*
(0.100)
0.319**
(0.078)
0.015
(0.051)
-0.035
(0.093)
0.172**
(0.052)
-0.497
(0.339)
0.002
(0.036)
1.450
(1.194)
-1.884
(1.962)
0.122
(0.507)
-2.413
(6.982)
1.227
(3.792)
11.461
(10.641)
13.012
33.703
40.252
36
310
1359

8.870
(8.273)
-5.328
(6.275)
-0.065
(0.134)
0.169
(0.159)
0.030
(0.100)
-0.132
(0.081)
0.015
(0.056)
0.206
(0.274)
0.038
(0.045)
0.889
(0.927)
-3.263
(2.051)
-0.245
(0.878)
6.998
(4.921)
5.048
(4.107)
2.230
(4.468)
2.453
10.636
45.435
23
229
395

Model 30:
Position on
Economic
Issues
(Old, Niche)
-15.095**
(8.999)
3.152
(6.422)
-0.324*
(0.133)
0.181
(0.147)
-0.352*
(0.163)
0.363*
(0.165)
0.020
(0.078)
0.391
(0.559)
0.075
(0.064)
-8.776**
(1.661)
0.844
(3.478)
2.727*
(1.204)
-8.143
(8.193)
1.239
(3.360)
12.979
(6.979)
14.544
71.129
105.167
36
310
1359

Model 31:
Position on
Economic
Issues
(New, Niche)
-3.650
(16.138)
18.863**
(4.863)
0.104
(0.171)
-0.265*
(0.127)
-0.024
(0.150)
0.295*
(0.148)
-0.186*
(0.080)
0.377
(0.323)
-0.080
(0.107)
-2.962*
(1.459)
2.135
(3.091)
-0.233
(1.047)
-3.890
(7.506)
-3.492
(7.680)
-2.107
(5.363)
14.567
35.414
46.751
23
229
395

*p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

results confirm that my findings are not products of particular choice of cases, measures or
estimation methods. Therefore, it is safe to conclude that the main arguments of this chapter
are strongly supported by empirical evidence.

99

3.6. Conclusion
This chapter examines different responses of rightist parties to the increase in
inequality in advanced and nascent democracies. Relevant literature has already pointed out
how rising inequality creates adverse conditions for rightist parties in electoral competition
and identified how economic disparity affects the electoral strategy and election platform of
each party (e.g., Tavits and Letki 2014; Tavits and Potter 2015). This chapter shares the
argument that the level of inequality plays an important role in parties’ strategic decisions on
which position they wish to take within the economic dimension and on how much emphasis
each party places (or is willing to place) on social issues. This chapter contributes to the
extant literature concerning the effect of inequality on party strategy in the three following
ways.
First, the literature regarding the effect of inequality on party strategies has
overlooked political constraints which may potentially act as limitations to the strategic
options of party leaders. However, the contextual differences between advanced and nascent
democracies create differences in the degrees of political constraints imposed on the strategic
choices of parties. I demonstrate that those constraints lead the right-wing parties in advanced
and emerging democracies to choose different options despite the fact that position-taking is a
more attractive option. This is a major improvement on previous studies, which focus
exclusively on nascent democracies (Tavits and Letki 2014) or treat the age of democracy as
one of the control variables (Tavits and Potter 2015).
Second, this chapter attempts to offer a better understanding of the two party
strategies – issue positioning and issue emphasis – by treating them as substitutable responses
to changes in the conditions surrounding electoral competitions. Specifically, I investigate
how right-wing parties respond to rising inequality by considering both issue positioning and
issue emphasis together as the parties’ efforts to achieve electoral gains. Using the dataset
100

covering party platforms in 44 electoral democracies, I find evidence that supports my
theoretical predictions.
Finally, my findings in this chapter also have an important implication for emerging
democracies with underdeveloped class politics and an absence of credibility: the findings
imply that nascent democracies with a highly unequal income distribution are more likely to
experience policy-shifting and convergence of political parties in the economic dimension.
Such frequent policy-shifting and convergence prevent voters from identifying parties that
would best represent their interests, and thus will likely lead to unstable voter-party linkages
and a delay in party system institutionalization (Lupu 2013, 2014). Even though parties tend
to develop policy reputations as parties and voters accumulate electoral experiences, parties
will inevitably suffer through hardship in building long-term policy reputations if their
economic policies are not discernable from those of their rivals. If establishing stable policy
reputations for political parties is a prerequisite to ensuring the credibility of electoral
promises in new democracies (Keefer 2007), high inequality can be a major obstacle that
prevents nascent democracies from evolving into well-performing democracies with stable
voter-party linkages based on programmatic appeals. While I do not tackle this issue in-depth,
my findings call for further scholarly investigation on various mechanisms of how economic
inequality hinders the development of stable partisan politics grounded on programmatic
linkages in emerging democracies.

101

APPENDIX

102

Table 3.8. List of Country-elections (Old and New Democracies)
Country

Election-year
Old Democracies

Albania
Australia
Austria
Belgium
Bulgaria
Canada
Croatia
Cyprus
Czech
Denmark
Estonia
Finland
France
Georgia
Germany
Greece
Hungary
Iceland
Ireland
Israel
Italy
Japan
Korea, South
Latvia
Lithuania
Luxembourg
Macedonia
Mexico
Moldova
Netherlands
New Zealand
Norway

New Democracies
1997, 2001

1961, 1963, 1966, 1969, 1972, 1974, 1975, 1977,
1980, 1983, 1984, 1987, 1990, 1993, 1996, 1998,
2001, 2004, 2007, 2010
1986, 1990, 1994, 1995, 1999, 2002, 2006, 2008
1981, 1985, 1987, 1991, 1995, 1999, 2003, 2007,
2010
2005, 2009
1962, 1963, 1965, 1968, 1972, 1974, 1979, 1980,
1984, 1988, 1993, 1997, 2000, 2004, 2006, 2008,
2011
2011
2006, 2010
1964, 1966, 1968, 1971, 1973, 1975, 1977, 1979,
1981, 1984, 1987, 1988, 1990, 1994, 1998, 2001,
2005, 2007, 2011
2011
1962, 1966, 1970, 1972, 1975, 1979, 1983, 1987,
1991, 1995, 1999, 2003, 2007, 2011
1962, 1967, 1968, 1973, 1978, 1981, 1986, 1988,
1993, 1997, 2002, 2007, 2012

1991, 1994, 1997, 2001

2000, 2003, 2007, 2011
1996, 2001, 2006
1992, 1996, 1998, 2002

1999, 2003, 2007

2004, 2008
1976, 1980, 1983, 1987, 1990, 1994, 1998, 2002,
2005, 2009
1989 (November), 1993, 1996, 2000, 2004,
2007, 2009, 2012 (May), 2012 (June)
2010
1995, 1999, 2003, 2007, 2009
1977, 1981, 1982 (February), 1982 (November),
1987, 1989, 1992, 1997, 2002, 2007, 2011
1977, 1981, 1984, 1988, 1992, 1996, 1999
1968, 1972, 1976, 1979, 1983, 1987, 1992, 1994,
1996, 2001, 2006, 2008
1963, 1967, 1969, 1972, 1976, 1979, 1980, 1983,
1986, 1990, 1993, 1996, 2000, 2003, 2005
2008, 2012
1989, 1994, 1999, 2004, 2009
2008, 2011
2010
1963, 1967, 1971, 1977, 1981, 1982, 1986, 1989,
1994, 1998, 2002, 2003, 2006, 2010
1978, 1981, 1984, 1987, 1990, 1993, 1996, 1999,
2002, 2005, 2008, 2011
1961, 1965, 1969, 1973, 1977, 1981, 1985, 1989,
1993, 1997, 2001, 2005, 2009
103

1974, 1981, 1985, 1989 (June)
1994, 1998, 2002, 2006

1961

1992, 1996, 2000, 2004
1998, 2002
2008
1994, 1998, 2002, 2006
2000, 2003, 2006, 2009
1994, 1998, 2001, 2005, 2009

Table 3.8 (cont’d)
Poland
Portugal
Romania
Serbia
Slovakia
Slovenia
Spain
Sweden
Switzerland
Turkey
Ukraine
United
Kingdom
United States

2005, 2007, 2011
1987, 1991, 1995, 1999, 2002, 2005, 2009, 2011
2006, 2010, 2012
2011
1993, 1996, 2000, 2004, 2008, 2011
1964, 1968, 1970, 1973, 1976, 1979, 1982, 1985,
1988, 1991, 1994, 1998, 2002, 2006, 2010
1983, 1987, 1991, 1995, 1999, 2003, 2007, 2011
2002, 2007, 2011
1964, 1966, 1970, 1974 (February), 1974
(October), 1979, 1983, 1987, 1992, 1997, 2001,
2005, 2010
1964, 1968, 1972, 1976, 1980, 1984, 1988, 1992,
1996, 2000, 2004, 2008

104

1991, 1993, 1997, 2001
1979, 1980, 1983, 1985
1992, 1996, 2000, 2004, 2008
2007, 2008
1994, 1998, 2002
1996, 2000, 2004, 2008
1982, 1986, 1989

1987, 1991, 1995, 1999
1998, 2002, 2006, 2007

Table 3.9. Policy Categories for Emphasis on Social Issues
Variable
per601
per602
per603
per604
per605
per606
per607
per608
per705
per706
per2022
per2023
per6072
per6081
per7051
per7052

Content
National Way of Life: Positive
National Way of Life: Negative
Traditional Morality: Positive
Traditional Morality: Negative
Law and Order: Positive
Civic Mindedness: Positive
Multiculturalism: Positive
Multiculturalism: Negative
Underprivileged Minority Groups
Non-economic Demographic Groups
Restrictive Citizenship: Positive
Lax Citizenship: Positive
Multiculturalism pro Roma: Positive
Multiculturalism pro Roma: Negative
Minorities Inland: Positive
Minorities Inland: Negative

Table 3.10. Policy Categories for Position on Economic Issues
Variable
per403
per404
per406
per409
per412
per413
per503
per504
per701
per4121
per4122
per4123
per4124
per4131
per4132

Left
Content
Market Regulation
Economic Planning
Protectionism: Positive
Keynesian Demand Management
Controlled Economy
Nationalization
Equality: Positive
Welfare State Expansion
Labor Groups: Positive
Social Ownership: Positive
Mixed Economy: Positive
Publicly-Owned Industry: Positive
Socialist Property: Positive
Property-Reinstitution: Negative
Privatization: Negative

Variable
per401
per402
per407
per414
per505
per702
per4011
per4012
per4013
per4014

105

Right
Content
Free Market Economy
Incentive: Positive
Protectionism: Negative
Economic Orthodoxy
Welfare State Limitation
Labor Group: Negative
Privatization: Positive
Control of Economy: Negative
Property-Reinstitution: Positive
Privatization Voucher: Positive

Table 3.11. Details Regarding Control Variables
Variable
GDP Growth
Effective Number of Parties
Niche Party (Megiud 2005,
2008)
Niche Party (Adams et al.
2006)
Party Size
New Party

Description
Annual GDP growth taken from World Bank’s World
Development Indicators (WDI) database.
The effective number of electoral parties taken from Gallagher
(2015); as for missing observations, figures are taken from
Bormann and Golder (2013) wherever they are available.
1 for ecologist, ethnic-regionalist, and nationalist parties, 0
otherwise.
1 for ecologist, communist, and nationalist parties, 0 otherwise.
Each party’s seat share in the legislature ((N of seats)/(Total N of
Seats)) based on information from the CMP
1 for parties competing in a democratic election for the first time,
0 otherwise

Table 3.12. List of Indicators in V-Dem Database for Political Constraints
Indicator
Party Organization (v2psorgs)
Party Branches (v2psprbrch)
Candidate Selection
(v2pscnslnl)
Party Linkages (v2psprlnks)

Question
How many political parties for national-level office have
permanent organizations?
How many political parties have permanent local party
branches?
How centralized is legislative candidate selection within the
parties?
Among the major parties, what is the main or most common
form of linkage to their constituent?

Table 3.13. Summary Statistics
Variable
Emphasis on Social Issues
Left-Right on Economic Issues
Right
Economic Inequality
New Party
Niche Party (Meguid 2008)
Niche Party (Adams et al. 2006)
Party Size
GDP Growth Rate
Effect Number of Party
Turnout Rates
Ethnic Fractionalization
Religious Fractionalization
Linguistic Fractionalization
Unemployment Rate
Economic Globalization
Political Constraints

Obs
2255
2255
2263
2128
2263
2263
2263
2263
2115
2257
2241
2263
2244
2244
2200
1999
2208

Mean
14.444
-10.992
0.490
39.823
0.186
0.137
0.200
0.190
2.487
4.822
72.699
0.254
0.415
0.257
7.120
67.788
6.994

106

Std. Dev.
9.632
14.924
0.500
6.154
0.389
0.343
0.400
0.162
3.620
1.841
13.713
0.199
0.221
0.198
5.392
15.649
2.637

Min
0
-68.182
0
21.221
0
0
0
0
-11.403
1.91
34.94
0.002
0.005
0.002
0
25.75
-1.589

Max
88.889
66
1
59.740
1
1
1
0.871
18.621
13.82
95.43
0.712
0.824
0.644
36
98.2
11.988

CHAPTER 4
WHICH FACTORS MOTIVATE POLITICAL PARTIES TO ENGAGE IN
ETHNIC APPEALS IN ELECTORAL COMPETITION?

“Ethnicity offers political leaders the promise of secure support. Politicians who can count
have something they can count on.” – Donald L. Horowitz, Ethnic Groups in Conflict.

4.1. Introduction
When and why do political parties attempt to appeal to ethnic identities of voters?
Under what conditions are parties more likely to emphasize ethnic issues in their electoral
campaign? This study discusses these questions and attempts to answer them.
In order to answer the questions, I focus on the incentive of political parties and
leaders to mobilize voters around ethnic issues in their pursuit of electoral success and
various political and economic conditions that shape these electoral incentives of political
entrepreneurs. To this end, this chapter investigates the effect of political and economic
conditions, including electoral competitiveness, electoral institutions, information shortage,
and economic inequalities between and within ethnic groups. The central findings are that
political elites and parties are incentivized to appeal to ethnic identities when BGI is high and
WGI is low. The reason is that the social salience of ethnicity increases under those
conditions. More specifically, a large wealth gap between ethnic groups leads to the
accumulation of ethnic grievances, and, in turn, creates distributional conflicts along ethnic
lines. In addition, the low economic disparity within an ethnic group increases homogeneity
in economic interests among members of the ethnic group, which leads to high internal
cohesion of the ethnic group. Political elites are then incentivized to exploit the high social
salience of ethnicity for their electoral benefits by emphasizing ethnicity in their pursuit for
107

electoral gain. On the other hand, I fail to find supporting evidence for the other factors,
including electoral competitiveness, electoral institutions, and information shortage.
The various aspects of the ethnicization of politics, or the degree to which ethnicity is
salient in politics, including ethnic voting, ethnic conflicts, and party system ethnicization,
have been popular subjects of political science research for decades. The effects of ethnic
divisions in society on various political phenomena and economic outcomes and the political
significance of ethnic diversity have been also well documented in previous studies (e.g.,
Alesina et al. 2003; Alesina and La Ferrara 2005; Bates 1974; Cederman et al. 2010;
Habyarimana et al. 2007; Horowitz 1985; Ordeshook and Shvetsova 1994; Petersen 2002;
Rabushka and Shepsle 1972). Another strand of political science research also shows that the
mere existence of multiple ethnic groups does not necessarily lead to the political salience of
ethnic cleavages (e.g., Dunning and Harrison 2010; Posner 2004, 2005). This implies that it is
important to understand the conditions under which ethnic cleavages are more likely to gain
political significance. Moreover, if ethnic identities are socially constructed and the salience
of ethnic cleavages are endogenous to the efforts of political mobilization, as the
constructivist view of ethnicity argues (Chandra 2004; Posner 2005), it is necessary to study
the role of political entrepreneurs in the politicization of ethnicity and the logic behind ethnic
appeals of political leaders and parties in the pursuit of electoral gain (Gadjanova 2013).
However, ethnic appeals of political parties during electoral campaigns have remained
underexplored, despite their ubiquity and their potential significance in understanding the
politicization of ethnic identities (Gadjanova 2015).
The remainder of this chapter will proceed as follows. The next section briefly
reviews the comparative politics literature on different perspectives on ethnic politics and
politicization of ethnicity. In the third section, I draw a series of hypotheses based on a
discussion of the conditions under which ethnic identities are more likely to be activated. A
108

series of estimation results of the statistical model are presented in the fourth chapter, in
which additional robustness tests, conducted to check that the results that confirm the
hypotheses are not produced by the particular choices of model specifications or
measurements, are also discussed. The last section offers concluding remarks and discusses
the implications of the findings.

4.2. Literature Review: The Different Perspectives on Ethnic Politics and the
Politicization of Ethnicity
4.2.1. Primordialism versus Constructivism
Scholars commonly define “ethnic identity” as “a subset of categories in which
descent-based attributes are necessary to determine eligibility for membership” (Chandra
2012: 58). Importantly, ethnic divisions have been long argued to be negatively associated
with various political and economic outcomes, including the stability of democracy
(Horowitz 1985; Rabushka and Shepsle 1972), economic growth (Easterly and Levine 1997;
Norris 2012), inter-group violence (Fearon and Laitin 1996; Posen 1993) and public goods
provision (Alesina et al. 1999). This family of research assumes, either implicitly or
explicitly, that ethnic categories are fixed and exogenously given (Chandra 2012). This view
of ethnic politics, commonly known as the primordialist view (e.g., Geertz 1973; Gellner
1983; Rabushka and Shepsle 1972), considers ethnic divisions or heterogeneity to be
exogenously determined (Ordershook and Shvetsova 1994), and ethno-cultural identities to
originate from “the givens of social existence” (Geertz 1973). According to this perspective,
inter-group animosities and conflicts between ethnic groups are prone to exist due to the
inherent differences in kinship, race, religions, or cultures.
While the primordialist perspective reflects the popular notions that describe ethnic
politics and conflicts, the constructivist approach rejects the primordialist assumptions about

109

ethnic identities (Chandra 2012). First, constructivists argue that ethnic identities can change
as opposed to what primordialists assume. Moreover, while primordialists argue that each
person is characterized by a singular ethnic identity, constructivists demonstrate that
individuals may have multiple ethnic identities. Therefore, not only do ethnic identities
changes in terms of their strength, but they can also change within “repertoires” of multiple
ethnic identities which are ascribed to individuals. Which identity is recognized by
individuals and politically activated at a certain time and place also depends on various
political and economic contexts. The findings of constructivists imply that the political
salience of ethnic divisions does not stem from innate cultural differences, and the
relationship between ethnic groups is in large part determined by political and socioeconomic contexts.28 In other words, constructivists have demonstrated that ethnic identities
are not exogenous, but instead endogenous to political and economic conditions.

4.2.2. Why Study Ethnic Appeals?
The constructive perspective, which has been the dominant view on ethnic politics,
not only considers the political salience of ethnicity to be determined endogenously from
political and economic conditions, but also underscores the role of political entrepreneurs in
driving voters to identify with their in-group members in the process through which ethnicity
is politically activated. Here, I define “political activation” of ethnicity as an individual’s

28

For example, in his natural experimental study, Posner (2004) demonstrates that the relationship
between the same pair of ethnic groups can be dramatically different depending on the political
conditions they face. Specifically, he designs a natural experiment that examines the relationship
between Chewas and Tumbukas, who reside in both Zambia and Malawi, and finds that the two
groups have maintained an amicable relation in Zambia, but they are in bitter political competition in
Malawi. The reason is that the political salience of the ethnic cleavage is determined by the relative
sizes of those ethnic groups to the political arena of competition in each country: Chewas and
Tumbukas are too as small groups for political elites to mobilize for their political gains in Zambia,
but the relative sizes of those two ethnic groups are large enough to be attractive bases of political
support in Malawi.
110

political choice, including participating in violence or conflicts, choosing a party affiliation,
and voting, relying on her professed or assigned membership in an ethnic category.29 While
the existence of ethnic divisions within a society and self-recognition by individuals of
themselves as members of a particular ethnic group are prerequisites for the political
activation of ethnic identities, those necessary conditions do not automatically lead to high
political salience of ethnic identities. In other words, the political salience of ethnicity is not a
necessary consequence of its social salience, especially when there exist other social
cleavages that cross-cut ethnic identities (Dunning and Harrison 2010), and the political
activation of ethnicity usually requires a certain level of efforts on the side of political leaders
to seek support on an ethnic basis.
A subset of constructivists also assumes that political elites strategically activate a
certain nominal ethnicity category and target particular ethnic groups (e.g., co-ethnics) to
mobilize them to build their political support (e.g., Bates 1974; Brass 1974, 1991; Chandra
2004, 2005; Huber 2014; Posner 2004, 2005). As Posner suggests, “viewing one’s group as a
unified cultural entity may be a prerequisite for the development of a political salient
cleavage between one’s own group and one’s neighbor. But it in no way guarantees that the
cleavage between the two groups will become salient” (Posner 2004: 537). Political parties
and leaders are assumed to be “instrumentally rational” in the sense that they want to
maximize popular support, so they weigh costs and benefits of ethnic mobilization. If this
logic of ethnic mobilization holds true, the strategic consideration of politicians to draw
electoral support at the ballot box should also provide the micro-foundation of their efforts to
mobilize voters around ethnic cleavages for electoral gain.
29

Here, I use the term “political activation” of ethnicity based on Chandra’s (2012) definition. She
defines the activation of ethnicity as “the act of choosing membership in some category or being
assigned membership in that category” (Chandra 2012: 115). Activated ethnic identity is distinguished
from “nominal” ethnic identity, which is an inactivated identity category or decent-based attributes by
which an individual is characterized.
111

One of the major tools for politicians’ ethnic mobilization efforts is to appeal to
members of a particular ethnic category in their electoral platforms. Previous studies find that
ethnic identities become more salient as an election date approaches. The ethnic mobilization
effort of political leaders to garner support at the ballot box is suggested as the core
mechanism for the increasing ethnic salience around elections (Eifert et al. 2010; Higashijima
and Nakai 2016). In other words, ethnic appeals of political entrepreneurs around election
time significantly promotes the political activation of ethnic identities and they also operate
as catalysts for the political activation of ethnic categories. While the repertoire of ethnic
appeals is not limited to engaging in ethnic favoritism or specifying policies that benefits a
particular ethnic group in election manifestos,30 these are all important components of
appeals to the targeted ethnic group, as parties’ explicit ethnic messages and policy
statements regarding ethnic issues in their official platforms can be interpreted as strong
signals of what they want to propose to voters (Gadjanova 2015).31
In order to fully understand the process of the political activation of ethnicity, it is
also necessary to investigate various conditions that incentivize political entrepreneurs to
engage in expressive ethnic appeals and promise policy favors to particular ethnic groups.
While the political activation of ethnicity is subject to manipulation of political elites seeking
electoral benefits, it is important to recognize that the attempt of politicians to capture
electoral support by appealing to voters’ ethnic identities are also constrained by ongoing
socio-economic conditions related to the structure of ethnic divisions.32 In other words, elites

30

The list of ethnic appeals can include recruiting and advertising co-ethnic candidates, holding
large-scale rallies for the targeted ethnic group around election time, and organizing intra-ethnic
gatherings to promote a party’s position at the grossroot level (Higashijima and Nakai 2016).
31
In this sense, ethnic appeals have both programmatic components (support for a set of policies and
issue positions) as well as expressive ones (symbols and emotions). The former component can be
more important, especially because blatant discrimination or overly exclusive ethnic appeals against
certain ethnic groups are sometimes neither acceptable nor effective (Gadjanova 2015).
32
Chandra defines “ethnic structure” as “any concept that describes nominal descent-based attributes
that characterize individuals or populations or the nominal categories generated from these attributes”
112

can take advantage of existing ethnic cleavages for their electoral benefits, but they cannot
create ethnic identities which are not already in place. Even though the extent to which
individuals identify with a certain ethnic category is subject to change when exposed to
ethnic mobilization by political leaders, the repertoire of descent-based attributes for their
recognized ethnicity is generally bounded in the short term. It also takes long for certain
ethnic identities to change due to their stickiness (Horowitz 1985; Chandra 2012).33 For
example, the repertoire of ethnic identities created during the British colonial period in
Zambia has persisted even after Zambia’s independence and democratization (Posner 2005).
Therefore, as a strategic agent, each political entrepreneur should take into account existing
ethnic demography and related socio-economic conditions in a society to make her ethnic
appeals effective in seeking electoral support. This reasoning is in line with Posner (2004),
who underlines the importance of politicians’ efforts to understand ethno-cultural
demography to be successful in electoral competition.
In sum, the role of ethnic appeals around elections is critical for understanding the
political salience of ethnicity, since they are the key mechanisms that develop nominal or
socially activated ethic categories into politically activated one. By using ethnic appeals in
their electoral campaigns, political entrepreneurs manipulate ethnic categories or the degree
to which voters identify with a certain ethnic group. However, the effort of political
entrepreneurs for ethnic mobilization are also constrained by existing distribution of decentbased attributes and related political and economic conditions. To explore this issue, the next

(Chandra 2012: 11). Here, I use this term as “the distribution of attribute and category repertoires in a
population” (Chandra 2012: 11).
33
It is also quite rare for new ethnic categories to be created through recombination of existing ethnic
groups (Ferree 2012). These boundedness and stickiness are why political support based on ethnic
categories is stable and more rigid in loyalties (Geertz 1973; Gellner 1983; Horowitz 1985) and less
vulnerable to short-term factors than class- or income-based support (Alonso 2008).
113

section will investigate the determinants of political entrepreneurs’ ethnic appeals involving
conditions in ethnically divided societies.

4.3. Ethnic Appeal and Its Determinants: Conditions under Which Ethnic Identities
Become Salient
The main constructivist criticisms of primordialism is that it fails to explain
variations in the political salience of ethnic cleavages across countries and over time.
Therefore, constructivists pay close attention to factors, including political institutions,
information shortage, and economic inequality, that are conducive or unfavorable to the
politicization of ethnic identities.
First, competitive elections crucially determine the salience of ethnic identities. For
example, Eifert et al. (2010) argue that citizens in ethnically heterogenous societies are more
likely to politically identify themselves as members of their own ethnic group during the
periods around closely fought national elections.34 In a similar vein, Wilkinson (2004)
suggests that political leaders sometimes utilize anti-minority events to mobilize their coethnics when they appeal to voters in highly competitive elections. He also finds that
politicians, when they do not need electoral support from minority groups, precipitate ethnic
violence against ethnic minority groups by allowing violence against minorities to happen,
using data on Hindu-Muslim riots in India and other case studies of Ireland, Malaysia,
Romania, Bulgaria, and the United States.35 Therefore, I speculate that political parties are
motivated to appeal to ethnic identities in closely fought elections:

34

Technically, Eifert et al. (2010)’s argument is that citizens identify with their ethnic group more
strongly as an election date approaches and the effect of electoral proximity is conditional on electoral
competitiveness. However, election manifestos are generally issued around elections, so I assume that
an election date and a time when election manifestos are issued are in close proximity.
35
Competitive elections at the national level also shape incentives for political elites to decide which
ethnic cleavage they want to mobilize voters around to enhance their electoral fortunes. Posner (2005)
demonstrates that salient ethnic cleavages changed due to democratization and the subsequent
114

Hypothesis 4-1: Political parties are more likely to strongly appeal to ethnicity when parties
intensely compete in the election.

Configurations of electoral institutions, majoritarian or proportional representation
(PR) electoral systems, are also argued to affect the degree to which ethnicity is politicized.
While the consociationalist model of democracy recommends ethnically divided societies to
adopt PR electoral systems for equal representation of minorities (e.g., Lijphart 1977, 1999),
PR has been also criticized on the grounds that it can reinforce ethnic identities and aggravate
ethnic divides within the society (e.g., Horowitz 1985, 1991; Reilly 2001). For example, the
political salience of ethnic divisions and ethnic tensions between dominant “whites” and
minorities, including Maori and Pacific Islanders, has increased after New Zealand adopted
mixed-member proportional system (Norris 2008). The latter argument favors majoritarian
systems that incentivize political parties to engage in inter-group compromise and vote
pooling to form electoral coalitions (Horowitz 1985; Reilly 2001). On the other hand, Huber
(2012) finds that PR is actually negatively associated with the ethnicization of politics using
his newly designed measures for the extent to which politics is ethnicized. While there are
many compelling explanations for the effects of electoral institutions on the political
activation of ethnicity, there is no clear consensus on the direction of the effects. As there are
two competing arguments on the possible relationship between the ethnic salience and
electoral system, I derive the following two hypotheses:

introduction of competitive elections in Zambia. Specifically, ethnic entrepreneurs try to mobilize
voters by appealing to voters’ tribal identities under single-party rule, but they attempt to organize
electoral support along linguistic lines in multi-party system where electoral competition takes place
at the national level. As a result, citizens were more likely to identify themselves as members of one
of the 73 tribes before democratization, and the language categories, consisting of the four main
language groups, became politically salient under multi-party rule where party labels at the national
level are significant for political entrepreneurs to be successful in electoral competition at the district
level.
115

Hypothesis 4-2-1: Political parties are more likely to strongly appeal to ethnicity under
majoritarian electoral system.
Hypothesis 4-2-2: Political parties are more likely to strongly appeal to ethnicity under PR
electoral system.

The extent to which information on parties or candidates are available for voters
affects the political salience of ethnicity. In democracies where voters are linked to political
parties or elites based on the patron-client network rather than on programmatic appeals or
ideological commitments, electoral competition is more likely to revolve around ethnic
issues. In her seminal work, Chandra (2004) explains the logic of “ethnic head counts”
behind party politics and electoral competition in India. She argues that an individual’s ethnic
identities are much more “costless” information than other non-ethnic identities in a limited
information situation, and thus people are more likely to refer to the information on ethnic
category when they distinguish between candidates. This mechanism of ethnic politics is selfenforcing, as voters expect political leaders to favor co-ethnics in the distribution of state
benefits and elites expect co-ethnic voters to support them at the ballot box. In a similar vein,
Carey (2015) demonstrates that the increasing information about candidates decreases voting
along ethnic lines in Sierra Leone. Ferree (2006) tests the three mechanisms of ethnic voting,
including expressing identities, policy and performance evaluations, and racial heuristics, and
finds that voters use race as an information shortcut for their electoral choice in South Africa.
Along these lines, Conroy-Krutz (2013), based on his survey experiment in Uganda, supports
the argument for the role of information shortage in explaining ethnic voting.
These findings can be generalized for other nascent democracies where it is difficult
for voters to collect relevant programmatic information to their vote choice and clientelistic

116

networks are prominent forms of voter-party linkages (e.g., Posner 2005). The prevalence of
voter-party linkages based on clientelistic networks is closely associated with voters’
difficulty in obtaining information on parties’ programmatic positions (Keefer 2007).
Chandra (2004) focuses on limited information settings in patronage-democracies to explain
ethnic politics in India, and argues that political entrepreneurs have an incentive to mobilize
voters along ethnic lines in order to effectively secure electoral support in those settings.
Birnir (2007) finds that voters tend to cast their votes relying on ethnic cues, since ethnic
identities, such as race, languages, religions, or traditional clothes, are much more identifiable
or visible than any other social categories like income and social status in information-poor
nascent democracies. As a result, ethnic cleavages are more likely to become salient in new
democracies, but ethnic salience also decreases as democracy matures. Thus, I establish the
two hypotheses on the relationship between the information availability and ethnic appeals:

Hypothesis 4-3-1: Political parties are more likely to strongly appeal to ethnicity in new
democracies.
Hypothesis 4-3-2: Political parties are more likely to strongly appeal to ethnicity when a
clientelist network is a prevalent form of a voter-party linkage.

The final condition is economic inequalities between and within ethnic groups. There
has been a general consensus that high BGI is the main source for members of disadvantaged
ethnic groups to feel a sense of deprivation (Gurr 1970; Hechter 1975), and the resulting
grievances sometimes lead to ethnic conflicts (Bates 1974; Horowitz 1985). Recent studies in
ethnic politics has underscored economic disparities between and within ethnic groups as
factors that play major roles in various aspects of ethnic politics, including ethnic conflict and

117

civil war (Cederman et al. 2011; Cederman et al. 2013; Esteban and Ray 2011; Gubler and
Selway 2012; Kuhn and Weidmann 2015; Østby 2008a; Stewart 2000).
Building on the previous studies, I suggest that ethnic identities become easier to be
politically activated by ethnic appeals of political entrepreneurs as BGI increases and WGI
decreases. First, ethnic grievances among members of poor ethnic groups are expected to
become prevalent when BGI is high. As group identity is an essential part of individual’s
identity, they usually evaluate their status by comparing their group to others (Horowitz
1985). Therefore, as BGI increases, members of the disadvantaged ethnic groups are more
likely to feel frustration and, at the same time, be antagonistic toward outgroup members,
especially members of rich ethnic groups. In other words, collective grievances of
disadvantaged ethnic groups against relatively rich ethnic groups increases with the wealth
gap between ethnic groups. Higher BGI also creates distributional conflicts along ethnic
lines, as class conflicts over redistributive policies are likely to develop into strife between
ethnic groups. As a result, tension between members of the rich and poor ethnic group over
material benefits increases with BGI, and ethnic cleavages are likely to be more socially
salient under this condition (Cedermann et al. 2011; Houle 2015; Østby 2008a, 2008b).
On the other hand, Houle (2015) argues that low WGI leads members of the ethnic
groups to share similar living conditions and policy preferences. As a result, individuals of
the ethnic group are more likely to identify themselves with their ethnic groups and in-group
loyalties increases under low WGI.36 Low WGI also increases internal cohesion of the ethnic

36

Actually, the effect of WGI on political phenomena involved in ethnicity can differ depending on a
dependent variable of interest. For example, high WGI “can lead to intra-group resentments which
group leaders buy off by directing animosity against other groups” (Stewart 2000: 253). One strand of
research on the effect of WGI argues that WGI is positively associated with the likelihood of ethnic
conflicts or civil war onset, since political elites find it easy to recruit in-group rebels, thanks to high
WGI, as the opportunity costs of participating in violent conflicts for regular group members decrease
and the elites have more resources to use for mobilization (e.g., Esteban and Ray 2011; Kuhn and
Weidmann 2015). Huber and Suryanarayan (2016) also test the effect of WGI on party system
ethnification, but the effect of WGI is neither significant nor consistent in terms of its direction.
118

group, as the group members are more likely to have homogenous economic interests and
preferences (Houle 2015). Political entrepreneurs, then, find it more attractive to appeal to the
ethnic group with high internal cohesion.
As BGI increases or WGI decreases, economic and ethnic cleavages reinforce rather
than cross-cut each other (Gubler and Selway 2012; Østby 2008a, 2008b; Selway 2015).
Consequently, an ethnic group is more likely to resemble, in the words of Rabushka and
Shepsle, “a consensual corporate group in conflict with similar corporate entities” (Rabushka
and Shepsle 1972: 68) under high BGI and low WGI.37 Under those conditions, ethnic
appeals of political entrepreneurs will be most effective, as members of ethnic groups
targeted by the appeals have strong prior attachments to the ethnic group (Valenzuela and
Michelson 2016). Therefore, political parties are induced to target certain ethnic groups using
ethnic appeals in their electoral campaigns when BGI is high and WGI is low:

Hypothesis 4-4-1: Political parties are more likely to strongly appeal to ethnicity as BGI
increases.
Hypothesis 4-4-2: Political parties are more likely to strongly appeal to ethnicity as WGI
decreases.

4.4. Data, Measurements, and Model Specification
4.4.1. Data and Cases
To test the main hypotheses, I pull the information on parties’ election platforms
from the Comparative Manifesto Project (CMP) data set (Volkens et al. 2015). The CMP
37

Rabushka and Shepsle defines “intracommunal consensus” as “the members of an ethnic
community perceive and express preferences about political alternatives identically.” “Intercommunal
conflict” is also a situation where “communities are in disagreement on all issues that face the
collectivity” (Ragushka and Shepsle 1972: 67). Under high BGI (“intercommunal conflict”) and low
WGI (“intracommunal consensus”), ethnic groups and ethnic relations are more likely to achieve, at
least on redistribution issues, those two conditions.
119

data, which covers more than 50 countries, provides quantified information on party
platforms by disaggregating documents into quasi-sentences, which are assigned to one of the
policy categories. One potential problem with using the CMP dataset is that it contains very
little information on party platforms in African countries where ethnicity plays a critical role
in electoral politics. However, the CMP dataset is still the best option in terms of its coverage
and the logic of ethnic appeals suggested in the previous sections can be applied to many
democracies in other regions, including most European democracies. Moreover, the fact that
in advanced democracies party platforms generally better reflect the actual policy positions of
political parties also justifies studying ethnic appeals using the CMP.
In order to exclude the information on electoral platforms in non-democracies,
elections that were held in non-democratic years are ruled out based on the Boix-MillerRosato (Boix et al. 2013) and the Cheibub-Gandhi-Vreeland (Cheibub et al. 2010) datasets.
Specifically, I drop the information on platforms in elections held in years that are coded as
non-democracy in either dataset.

4.4.2. Dependent Variable: Measuring Ethnic Appeals
The dependent variable is the intensity of ethnic appeals. In spite of the importance
of ethnic appeals for understanding ethnic politics, there are only very few attempts to
rigorously measure ethnic appeals (e.g., Gadjanova 2013, 2015; Protsyk and Garaz 2013) and
there is no consensus on how to measure ethnic appeals. The CMP data set also contains
policy categories which can be used to measure parties’ ethnic appeals in their election
manifestos (Gadjanova 2015), though they may not be exhaustive and are imperfect (Protsyk
and Garaz 2013). The problem making the measurement of ethnic appeals based on the CMP
dataset more complicated is that the interpretations of policy statement that belongs to the
same policy category can also vary depending on where or when the statement is proposed.
120

This implies that it would be “conceptual stretching” to include policy categories, such as
decentralization and industry protection, which may or may not be linked to the welfare of
certain ethnic groups, in the construction of measurements for ethnic appeals without
specifying ethno-cultural contexts where the policy statement is suggested (Gadjanova 2015;
Satori 1970).
To avoid this problem, I attempt to include only policy categories with explicit and
direct relevance to ethnic issues. For this purpose, I refer to Protsyk and Garaz (2013) and
identify nine manifesto categories for five “multicultural” appeals (i.e., positive attitudes
towards the preservation of ethno-cultural diversity) and two categories for “integrationist”
appeals (i.e., negative attitudes against multiculturalism).38 Then, I add up all scores on the
seven categories relevant to ethnic appeals to come up with the measure of ethnic appeals at
the party-election level.39 In addition, the average of parties’ ethnic appeals is also calculated
in each election to see if the hypotheses are supported at the election level.

38

Details on the policy categories for each group are presented in Table 4.3. Protsyk and Garaz
(2013) design their own coding scheme with new policy categories as an alternative to the CMP
policy categories. As they make clear, measuring ethnic appeals based on the CMP policy categories
is never perfect and this may cause the problem of “undercounting” ethnic appeals (Protsyk and Garaz
2013). However, their measures cover only four East-European countries (Bulgaria, Moldova,
Romania, and Ukraine), so while their coding efforts are very sensible, it is also hard to use their
measures for a cross-national analysis with a much larger number of countries. Their study finds that
the other CMP policy categories, such as “National Way of Life: Positive,” are also relevant to
measuring ethnic appeals depending on context, but “definitions of such relevant categories are, as a
rule, too inclusive for them to be considered valid measures of party position on issues related to
titular group identity” (Protsyk and Garaz 2013: 300). For these reasons, I take a conservative
approach to avoid the problem of “overcounting,” rather than undercounting, and conceptual
stretching.
39
I measure the dependent variable by simply adding all relevant policy categories to ethnic appeals,
unlike Gadjanova (2015) who use principal components factoring to come up with the two measures
of ethnic appeals. The reason why I do not follow her approach is that the scores on the seven policy
categories are not strongly enough correlated to conduct principal components analysis (PCA). In
general, the correlation between the variables should be stronger than 0.3 to conduct factor analysis or
principal components analysis (Tabachnik and Fidell 2001; Yong and Pearce 2013), but the pairwise
correlation coefficients between the nine variables are pretty weak. The largest correlation coefficient
is 0.31, which is between “Cultural Autonomy: Positive” and “Minorities Abroad: Positive,” but most
of the correlation coefficients are smaller than 0.1, with the next largest one as 0.12 (between
“Multiculturalism: Positive” and “Cultural Autonomy: Positive”). For this reason, I adopt the simplest
scheme to measure the strength of ethnic appeals, as the CMP data is also uniquely appropriate for
121

4.4.3. Independent Variables
The first hypothesis suggests that parties are more likely to engage in ethnic appeals
as elections are more closely fought, so I measure electoral competitiveness using vote shares
of parties in each election. Specifically, I compute the gap in vote shares between the two
parties with most votes, the winner and the runner-up, and multiply the gap by minus one
following Eifert et al. (2010). Therefore, this measure is larger, the more competitive the
election is. The information on vote shares of each party is obtained from the CMP dataset.
Previous studies also underscore the role of electoral institutions in the politicization
of ethnic identities, but they are still divided on which allocation rule - PR or majoritarian
system - induces politicians to appeal to ethnicity. To test the effect of electoral institutions on
ethnic appeals, I construct a dichotomous measure, PR, based on the information on electoral
system from Bormann and Golder (2013). They classify electoral systems into three
categories, PR, majoritarian, and mixed systems. As for the last category, I classify mixedmember proportional (MMP) systems as PR and mixed parallel systems as majoritarian
systems. I also pull the information on the mean district magnitude of the lowest electoral tier
from Bormann and Golder (2013). As the marginal effect of district magnitude is expected to
decrease as its value increases, I take the log of the mean district magnitude to account for
this logic.
Ethnic appeals are more likely to be effective when voters lack the relevant
information on parties’ policy positions or ideological dispositions. While it is hard to
measure the information availability, I adopt two proxies for the amount and features of
information on parties for voters. First, I draw the information on the features of voter-party
linkages from the Varieties of Democracy (V-Dem) database (Coppedge et al. 2016). The V-

measuring the intensity of emphasis on policy issues (Klingemann et al. 2006).
122

Dem data set provides the information (“Party Linkage”) on whether parties are linked to
their constituents based on programmatic/policy preferences or clientelistic networks. When
the common feature of voter-party linkages is clientelistic, parties are more likely to further
develop their support bases relying on ethnic categories, rather than invest in building ties to
voters based on programmatic preferences. Therefore, I use this measure, linkage, as the first
proxy for the information availability. Larger values indicate the higher prevalence of
programmatic linkages. Second, voters in new democracies do not have as much information
for electoral choices as those in advanced democracies due to the lack of past electoral
experiences. Policy reputations of political parties are also underdeveloped in new
democracies which leads parties to rely on the patron-client networks because their
programmatic promises are less likely to be credible (Keefer 2007). Therefore, I construct a
dichotomous variable for new democracy. Specifically, I classify five consecutive countryelections after democratization as nascent democracies (Tavits and Potter 2015). As for
countries who experienced democratic breakdown and reinstatement, I code five consecutive
democratic elections after three or more years of authoritarian interruption as elections in new
democracies.
In order to measure BGI and WGI, I draw BGI and WGI measures from Houle
(2015).40 He uses a series of survey datasets, including Demographic and Health Surveys,
World Values Survey, Latinobarometer, International Social Survey Program, and
Comparative Study of the Electoral Systems, to construct BGI and WGI measures. The
information on monetized income (WVS, Latinobarometer, ISSP, and CSES) or asset-based
wealth (ABW, DHS) is utilized to measure BGI and WGI for each ethnic group and country.

40

I am very grateful to Christian Houle for his generous permission to use his dataset.
123

Only ethnic groups listed in the Ethnic Power Relations (EPR) dataset are included for his
calculation.41 The formula for BGI for a given ethnic group is as follows:

BGI


 g
  log 
G






2

where g refers to the average income (or ABW) of members of the ethnic group, and G refers
to the average income (or ABW) of the country. This formula is designed to capture
deviations of the average income (or ABW) of a given ethnic group from the country average
income (or ABW). BGI should be 0 if the average income (or ABW) of an ethnic group is
equal to the country average (Cederman et al. 2011). WGI is measured by the Gini coefficient
in the income (or ABW) among all members of that group. Houle (2015) also calculated
country-level indicators for BGI and WGI by calculating the averages of group-level BGI and
WGI weighted by the size of the ethnic groups for each country. Therefore, the country-level
BGI indicates the average level of income disparity between ethnic groups of each country.
Likewise, the country-level WGI captures the weighted average of the Gini coefficients for
all groups of a country. The resulting measure captures the average level of inequality among
members of ethnic groups in that country.

4.4.4. Control Variables
I also include a set of control variables which may potentially affect the estimation of
the effects of the independent variables. First, I include three party level control variables,
including ethnic party, leftist party, and rightist party, and one party-election level control

41

Wimmer et al. include an ethnic group in EPR dataset only “if at least one significant political actor
claims to represent the interests of that group in the national political arena, or if members of an ethnic
category are systematically and intentionally discriminated against in the domain of public politics”
(Wimmer et al. 2009: 325).
124

variable, which is party size. Based on the party family variable in the CMP data, I classify
ethno-regional parties as ethnic party. Following previous studies, ecologist, communist, or
social democratic parties are coded as leftist party, and liberal, Christian democratic,
conservative, or nationalist parties are classified as rightist party (Burgoon 2013; Tavits and
Potter 2015). I also control for party size, measured by the proportion of legislative seats to
the total number of seats, as larger parties are more likely to try to appeal to broader
constituencies. Therefore, party size is expected to have a negative impact on ethnic appeal.
Two country-level variables, ethno-linguistic fractionalization and Western
democracies, are also controlled for. The high level of ethno-linguistic fractionalization
provides multiple bases for electoral support which political parties can appeal to, and thus
ethnic appeals are expected to be prevalent as the ethnic fractionalization index increases.
The ethnic fractionalization index is taken from Alesina et al. (2003). A dummy variable for
Western democracies is also included, since the frequency of extreme or violent ethnic
conflicts in Western democracies is not as large as that in other regions.
In addition, a battery of country-election level controls is also controlled for. The
number of ethnic parties can affect the strength of ethnic appeals at both country-election and
party-election levels and the increasing number of ethnic parties should have a positive effect
on ethnic appeals, but the direction of effect is not clear for the party-level analysis. If other
parties attempt to “outbid” ethnic parties by taking more extreme parties (Rabushka and
Shepsle 1972), then ethnic appeals should increase with the number of ethnic parties. On the
other hand, mainstream parties may ignore ethnic appeals proposed by ethnic parties in order
not to increase the political salience of ethnicity in electoral competition (Meguid 2005,
2008). The direction of the effect of economic development is also indecisive. According to
modernization theories, economic development and ensuing societal modernization creates
various social cleavages cross-cutting each other (Deutsch 1961; Lipset 1959, 1960).
125

However, Bates (1974) also argues that economic development may also create distributional
conflicts between ethnic groups if economic cleavages develop along ethnic lines. Therefore,
while the number of ethnic parties and log of GDP per capita are controlled for in estimation,
I do not specify any clear expectations regarding the directions of the effects here. Other than
these two factors, country-election level controls that may affect the political salience of
ethnicity, including turnout rates and the effective number of parties, are also included.

4.4.5. Model Specification and Estimation Strategy
In order to test the main hypotheses, I construct statistical models with the two
dependent variables, which are ethnic appeal at the party-election level and the average of
ethnic appeal of parties at the country-election level. The data structure is hierarchical, as
party-election level observations are clustered at both the party level and country-election
level, and the party level clusters are also nested within each country. Failure to take into
account this hierarchical data structure can lead to underestimation of the uncertainty in the
estimates (Angrist and Pischke 2009; Steenbergen and Jones 2002). For this reason, I
estimate multilevel linear models while allowing random intercepts at the higher levels, the
country and party levels, and also measure the uncertainty of the estimates with cluster-robust
standard errors at the country level following previous studies (Tavits and Potter 2015; Ward
et al. 2015). Observations with at least one missing value for any independent variables are
dropped out, and as a result, the full models are estimated using 215 and 1,279 observations
at the country-election and party-election level, respectively, from 27 democracies.
Specifically, the statistical models are presented as follows:

	

(1) Election Level:

	

	

	
126

, where

;

	

(2) Party-election Level:
	

	
, where

where

and

variances,

are country- and party-level random intercepts of which estimates of
and

, will be derived. Other than

and

, the signs of the coefficients

can be predicted based on the discussion on the relationship between ethnic appeal and each
independent variable. As elections become more competitive, parties are expected to appeal
to ethnicity more strongly, and thus

should be positive. A higher linkage indicates that

parties are tied to voters based on programmatic preferences, so
negative. On the other hand,

is expected to be

should be positive as ethnic appeals are expected be

prevalent in new democracies. As for the coefficients for BGI and WGI, the sign of
should be positive as ethnic appeal is expected to increase with BGI, whereas the hypothesis
4-2 predicts that

has a negative sign.

4.5. Estimation Results
4.5.1. Main Analysis
The estimation results for the two main models are reported in Table 4.1. The first
two columns of Table 4.1 present the estimation results of the first model with the dependent
variable as average ethnic appeal. The results confirm my hypotheses about the effects of
BGI and WGI on ethnic appeal at both election and party levels. The coefficient for BGI
equals 0.068 when control variables are not included, and this magnitude even increases up to
0.076 with the controls included in the model. The uncertainty in the estimate also decreases
after the effects of control variables are controlled for, and the coefficient is precisely enough
127

estimated to be confident about the positive effect of BGI on ethnic appeals (p < .01). These
results show that political parties, on average, emphasize ethnic issues more to appeal to
certain ethnic groups as BGI increases. In addition, WGI turns out to have a negative effect
on average ethnic appeal, which is consistent with the other hypothesis. Even though the
magnitude of the coefficient on WGI also increases when the controls are included in the
model, and the coefficient is significantly different from zero (p < .01). Therefore, we can
conclude that, on average, high WGI tends to discourage political parties from appealing to
ethnicity during campaigns.
While the estimation results confirm the effects of BGI and WGI in the expected
directions, I fail to find consistently supporting evidence for the other hypotheses. While the
signs of coefficients for variables measuring electoral systems are consistent across the
models, the coefficients are not estimated with sufficient precision to conclude that they are
statistically different from zero. As for variables representing the information availability for
voters, the coefficients for those variables not only fail to reach statistical significance, but
are also inconsistent in terms of their signs. The effect of competitiveness is most consistent
in terms of its direction among the independent variables besides BGI and WGI. The positive
coefficients for competitiveness implies that political parties tend to appeal more to ethnicity
as the gap in the vote shares between the winner and runner-up decreases, and this result is
consistent with the hypothesis regarding electoral competitiveness. However, the coefficient
reaches conventional statistical significance only in one estimation result, and it is hard to
safely conclude that the coefficients for competitiveness are statistically different from zero
due to large uncertainty in the estimates.
The effects of BGI and WGI are not only statistically significant, but also
substantively meaningful. Substantively, an increase in BGI by one standard deviation at the
election level (24.44) changes the predicted value of the average ethnic appeal by 1.853,
128

Table 4.1. Estimation Results of Hierarchical Linear Models
Model 1:
Countryelection Level
(Basic)
Constant
3.469*
(1.635)
0.004
Electoral
(0.015)
Competitiveness
Electoral Institutions
PR System
0.294
(0.338)
District
-0.103
Magnitude (Log)
(0.127)
Information
New Democracy
0.155
(0.391)
Voter-party
-0.046
Linkages
(0.192)
Inequality
BGI
0.068**
(0.016)
WGI
-0.182*
(0.076)
Ethno-linguistic
Fractionalization
GDP per capita
(Log)
Western
Democracies
N of Ethnic
Parties
Effective N of
Parties
Turnout Rates

Model 2:
Countryelection Level
(Full Controls)
5.339
(4.286)
0.017
(0.016)

Model 3:
Party-election
Level
(Basic)
3.246
(1.234)**
0.014
(0.014)

0.428
(0.293)
-0.059
(0.112)

0.237
(0.299)
-0.111
(0.134)

0.429
(0.363)
-0.108
(0.145)

-0.141
(0.437)
0.396
(0.277)

0.187
(0.412)
0.172
(0.219)

-0.070
(0.553)
0.503
(0.258)

0.076
(0.012)**
-0.208
(0.061)**
3.046
(1.385)*
-0.319
(0.508)
-0.814
(0.505)
0.233
(0.136)
-0.168
(0.073)*
0.010
(0.014)

0.051
(0.015)**
-0.147
(0.064)*

0.059
(0.012)**
-0.183
(0.059)**
1.219
(1.147)
0.066
(0.489)
-1.228
(0.664)
-0.133
(0.123)
-0.259
(0.095)**
0.006
(0.013)
4.282
(1.267)**
-0.432
(0.470)
-0.336
(0.479)
-0.737
(0.352)*
0.075

Ethnic Party
Leftist Party
Rightist Party
Party Size
0.512
1.705

N (Country)
N (Party)
N

Model 4:
Party-election
Level
(Full Controls)
3.841
(4.045)
0.028
(0.014)*

0.229
1.640

27

27

214

212

0.000
6.844
4.351
27
386
1,286

4.609
4.307
27
386
1,279

*p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.
129

Figure 4.1. Predicted Ethnic Appeal at the Election-level Conditional on the Levels of BGI
and WGI

4
3
2
0

1

Predicted Ethnic Appeals

5

6

Linear Prediction of Ethnic Appeals
Depending on BGI (Election-Level)

10

30

50

70

Between-group Inequality

4
3
2
1
0

Predicted Ethnic Appeals

5

6

Linear Prediction of Ethnic Appeals
Depending on WGI (Election-Level)

15

20

25

30

Within-group Inequality

Based on Model 2 of Table 1. Dashed lines are 95 percent confidence intervals.

130

Figure 4.2. Predicted Ethnic Appeal at the Party-level Conditional on the Levels of BGI and
WGI

4
3
2
0

1

Predicted Ethnic Appeals

5

6

Linear Prediction of Ethnic Appeals
Depending on BGI (Party-Level)

10

30

50

70

Between-group Inequality

4
3
2
1
0

Predicted Ethnic Appeals

5

6

Linear Prediction of Ethnic Appeals
Depending on WGI (Party-Level)

15

20

25

30

Within-group Inequality

Based on Model 4 of Table 1. Dashed lines are 95 percent confidence intervals.

131

which is larger than the standard deviation of the average ethnic appeal, when other variables
are set at the mean or the median. The change in WGI by one standard deviation (4.217) also
decreases average ethnic appeal by 0.877. The estimation results at the party-election level
presented in the right columns of Table 1 also consistent with the results of analysis at the
country-election level, in terms of both magnitude and statistical significance of the
coefficients.
The supporting evidence for the hypotheses on BGI and WGI are graphically
presented in Figure 1 and 2. Based on the estimation results of the models with full controls,
Figure 1 and Figure 2 graphically present the linear prediction of average ethnic appeal and
ethnic appeal at the party-election level depending on the level of BGI and WGI. The solid
line in each graph represents the point estimate of the linear prediction and the dashed lines
are the 95% confidence interval of the point estimate. While the point estimates of the linear
prediction of ethnic appeal in Figure 1 are more precisely estimated, the two pairs of plots are
similar in terms of substantive interpretation. The linear predictions of ethnic appeal in the
two figures consistently increase (or decrease) with BGI (or WGI). These results suggest that
higher BGI or low WGI encourages political parties and entrepreneurs to strongly appeal to
ethnicity as they expect that ethnic appeals are more likely to be electorally rewarded under
those conditions, since they increase the social salience of ethnicity. High BGI and low WGI
together create an attractive environment for political entrepreneurs to mobilize voters by
appealing to their ethnic identities as the social salience of ethnicity increase under those
conditions.

4.5.2. Robustness Tests
I also conducted additional robustness tests to check the sensitivity of the main
results. First, I estimated the models with an alternative measure of ethnic appeal. While I
132

Table 4.2. Robustness Checks: Alternative Measure of Ethnic Appeals and Excluding
Outliers
Model 5:
Countryelection Level
(Ethnic
Appeals)
Constant
5.526
(6.259)
0.019
Electoral
(0.016)
Competitiveness
Electoral Institutions
PR System
2.331**
(0.548)
District
-0.298
Magnitude (Log)
(0.174)
Information
New Democracy
0.757
(0.577)
Voter-party
0.449
Linkages
(0.354)
Inequality
BGI
0.084**
(0.020)
WGI
-0.282**
(0.085)
Ethno-linguistic
3.201*
Fractionalization
(1.496)
GDP per capita
0.381
(Log)
(0.714)
Western
-1.849*
Democracies
(0.890)
N of Ethnic
0.164
Parties
(0.164)
Effective N of
-0.448**
Parties
(0.135)
Turnout Rates
-0.021
(0.021)
Ethnic Party
Leftist Party
Rightist Party
Party Size
0.864
3.437

N (Country)

27

Model 6:
Party-election
Level
(Ethnic
Appeals)
1.664
(5.492)
0.022
(0.016)

Model 7:
Countryelection Level
(Outliers)

Model 8:
Party-election
Level
(Outliers)

9.421
(5.212)
0.030
(0.016)

7.831
(5.656)
0.034*
(0.015)

1.641*
(0.640)
-0.197
(0.187)

0.279
(0.319)
-0.028
(0.117)

0.150
(0.414)
-0.056
(0.155)

0.847
(0.715)
0.343
(0.391)

0.094
(0.514)
0.411
(0.278)

0.100
(0.596)
0.483
(0.270)

0.066**
(0.018)
-0.272**
(0.091)
2.744*
(1.149)
1.052
(0.668)
-2.187*
(1.088)
-0.365**
(0.114)
-0.568**
(0.119)
-0.003
(0.017)
3.750**
(1.244)
-2.081**
(0.644)
-0.188
(0.731)
-3.484**
(0.948)
0.000

0.070**
(0.018)
-0.324**
(0.112)
2.887*
(1.373)
-0.302
(0.500)
-0.759
(0.613)
0.249
(0.133)
-0.158*
(0.077)
0.002
(0.015)

0.052**
(0.014)
-0.296*
(0.114)
0.837
(1.225)
0.114
(0.497)
-1.162
(0.736)
-0.127
(0.127)
-0.255**
(0.096)
-0.002
(0.013)
4.482**
(1.355)
-0.476
(0.513)
-0.294
(0.518)
-0.731
(0.410)
0.020

13.743
11.340
27

133

0.153
1.705

24

4.908
4.443
24

Table 4.2 (cont’d)
N (Party)
N

212

386
1,279

196

354
1,199

*p<0.05, ** p<0.01 (two-tailed); Cluster-robust standard errors are given in parentheses.

take a conservative approach to measure ethnic appeals, the two policy categories regarding
“National Way of Life” may include statements that are directly related to ethnic groups
(Protsyk and Garaz 2013). To entertain this possibility, I include the two policy categories
and calculate the sum of scores again. Next, I redo the main analysis by excluding four
countries, Albania, Moldova, Sri Lanka, and Turkey, that are the biggest outliers in terms of
BGI and WGI. Both BGI and WGI for Sri Lanka are extremely large and those for the other
three countries are also outliers in the opposite direction. To see whether the results are
products of those outliers, I exclude the four countries from the analysis and re-estimate the
models. The estimation results are presented in Table 4.2 and all the main results regarding
the effects of BGI and WGI are robust to these changes. It is noteworthy that PR has a
positive effect on ethnic appeals with the alternative measure of ethnic appeal and the effect
is statistically significant. In addition, competitiveness also has an expected positive impact
on ethnic appeal when outliers are excluded as it does in the main analysis.

4.6. Conclusion
This chapter has investigated conditions under which political entrepreneurs are
induced to appeal to ethnicity. A relatively recent development in the literature regards
ethnicity as socially constructed and the mere existence of multiple ethnic categories in a
society does not necessarily result in the politicization of ethnic identities. Rather, the
political activation of ethnic identities results from political entrepreneurs’ strategic
considerations of political and economic conditions that shape politicians’ incentive to build

134

their political support based on ethnic groups. In this regard, ethnic appeals of political parties
during their electoral campaign deserve more attention from political science research as
ethnic appeals are major tools for parties to draw support from certain ethnic groups at the
ballot box. While underscoring the crucial role of political entrepreneurs and their ethnic
appeals in the process through which ethnic identities are politically activated, I investigated
the political and economic determinants - electoral competitiveness, electoral institutions, the
information availability, and BGI and WGI - of political parties’ ethnic appeals. The central
findings are that ethnic appeals during electoral campaigns are positively (or negatively)
associated with the level of BGI (or WGI). These findings are in line with recent comparative
studies that have focused on distributional conflicts and asymmetries between and within
ethnic groups in studying ethnic politics (e.g., Alesina et al. 2016; Baldwin and Huber 2010;
Cederman et al. 2011; Cederman et al. 2013; Houle 2015; Huber and Suryanarayan 2016;
Kuhn and Weidermann 2015). These conditions, high BGI and low WGI, raise the social
salience of ethnic cleavage, and thus motivate political entrepreneurs to appeal to ethnic
identities of voters in electoral competition. On the other hand, I fail to find empirical
evidence for the effects of other political correlates on ethnic appeals, and if there is any
supporting evidence, it is at most mixed. This is the first cross-national study, to my best
knowledge, to investigate the determinants of political parties’ ethnic appeals, which have
been unexplored so far in spite of their importance for understanding ethnic politics.
The implications of this study are clear. The vast literature on ethnic politics has
found plenty of evidence that ethnic divisions have negative impacts on various political and
economic outcomes, including the stability of democracy, public goods provision, economic
development, and civil war onset. While the existence of multiple ethnic categories is a fact
of life in most countries, political scientists have long studied how to minimize the negative
consequences of ethnic divisions in ethnically divided societies. The first suggested tool is to
135

design electoral institutions that are able to deal with ethnic divisions by allowing minorities
to be represented in the legislature (e.g., Liphart 1977) or motivating political parties to draw
support across ethnic groups (e.g., Horowitz 1985). Or, the politicization of ethnicity is
argued to be prevalent in new democracies where the ethnicity of candidates is the most
accessible information for voters, but ethnic politics is also expected to fade away as a
democracy matures (Birnir 2007). My findings, however, suggest that designing institutional
mechanisms may not be the best way to prevent the political activation of ethnic identities. I
also fail to find any compelling evidence of the relationship between the maturity of
democracy and the effort of ethnic mobilization on the politicians’ side.
Rather, the central findings imply that government should find a way to alleviate
economic disparity between ethnic groups and make every effort, especially in developing
countries, to distribute the fruit of economic development across ethnic groups in order to
keep ethnic divisions from developing into politically activated social cleavages. If certain
groups are favored in the distribution of government benefits, then the wealth gap between
rich and poor ethnic groups increases and ethnic and economic cleavages reinforce each
other. This condition will invite political entrepreneurs, regardless of whether they are ethnicminded or not, to appeal to voters’ ethnic identities to exploit the conditions for their electoral
gain. Regulations of ethnic politics on the side of the elite by outlawing ethnic parties or
banning particular forms of electoral campaign are very common in Africa (Moroff 2010),
but these measures have not been actually effective as violent ethnic conflicts remain
prevalent in many African countries and ethnic politics is still the most prominent aspect of
African politics. In this sense, policy recommendations for deterring politicians from
engaging in exclusive and extreme ethnic appeals should focus on the socio-economic
conditions that may cause political entrepreneurs to rely on ethnic mobilization.

136

APPENDIX

137

Table 4.3. Policy Categories for Ethnic Appeals
Multicultural
Per 607 Multiculturalism: Positive
Favorable mentions of cultural diversity and
cultural plurality within domestic societies. May
include the preservation of autonomy of
religious, linguistic heritages within the country
including special educational provisions.
Per 6071 Cultural Autonomy: Positive
Favorable mentions of cultural autonomy.

Integrationist
Per 608 Multiculturalism: Negative
The enforcement or encouragement of cultural
integration. Appeals for cultural homogeneity in
society.
Per 6081 Multiculturalism pro-Roma:
Negative
Negative mentions of cultural autonomy of
Roma.

Per 6072 Multiculturalism pro Roma:
Positive
Favorable mentions of cultural autonomy of
Roma.
Per 7051 Minorities Inland: Positive
References to manifesto country minorities in
foreign countries; positive references to
manifesto country minorities.
Per 7052 Minorities Abroad: Positive
References to ethnic minorities living in the
manifesto country such as Latvians living in
Estonia.

Table 4.4. Summary Statistics
Variable
Ethnic Appeals (Election)
Ethnic Appeals (Party-Election)
BGI (Election)
BGI (Party-Election)
WGI (Election)
WGI (Party-Election)
Electoral Competitiveness
PR System
Linkage
New Democracy
Number of Ethnic Parties
Effective Number of Parties
Average District Magnitude
(Log)
Turnout Rates
Ethnic Fractionalization
GDP per Capita (Log)
Western Democracy
Ethnic Party
Leftist Party
Rightist Party
Party Size

Obs
218
1296
218
1299
218
1299
1293
1295
1299
1299
1299
1299
1295

Mean
1.593
1.703
49.760
48.291
28.090
28.132
0.918
0.726
1.853
0.382
1.079
4.976
1.795

Std. Dev.
1.822
3.227
24.436
17.743
4.217
3.807
0.081
0.446
1.018
0.486
1.611
1.955
1.534

Min
0
0
0.252
0.252
13.603
13.603
0.6213
0
-1.372
0
0
1.97
0

Max
9.675
35.29
220.555
220.555
43.817
43.817
1.000
1
3.130
1
8
10.28
6.109

1292
1299
1299
1299
1299
1299
1299
1293

71.369
0.356
9.235
0.646
0.120
0.333
0.451
0.162

14.855
0.187
0.630
0.478
0.325
0.471
0.498
0.169

34.94
0.093
7.158
0
0
0
0
0

94.31
0.712
10.278
1
1
1
1
0.871

138

Table 4.5. List of Countries and Election Years (28 Countries)
Country
Albania
Australia
Belgium
Bulgaria
Canada
Croatia
Czechoslovakia
Estonia
Finland
Greece
Hungary
Latvia
Lithuania
Macedonia
Mexico
Moldova
Netherlands
New Zealand
Romania
Slovakia
Slovenia
Spain
Sri Lanka
Switzerland
Turkey
Ukraine
United Kingdom
United States

Election Years
1992, 1996, 1997, 2001
1961, 1963, 1966, 1969, 1972, 1974, 1975, 1977, 1980, 1983, 1984, 1987,
1990, 1993, 1996, 1998, 2001, 2004, 2007
1961, 1965, 1968, 1971, 1974, 1977, 1978, 1981, 1985, 1987, 1991, 1995,
1999, 2003, 2007
1991, 1994, 1997, 2001, 2005
1962, 1963, 1965, 1968, 1972, 1974, 1979, 1980, 1984, 1988, 1993, 1997,
2000, 2004, 2006
2000, 2003, 2007
1990, 1992
1992, 1995, 1999, 2003, 2007
1962, 1966, 1970, 1972, 1975, 1979, 1983, 1987, 1991, 1995, 1999, 2003,
2007
1977, 1981, 1985, 1989, 1990, 1993, 1996, 2000, 2004, 2007
1994, 1998, 2002, 2006
1993, 1995, 1998, 2002
1992, 1996, 2000, 2004
1994, 1998, 2002, 2006
2003, 2006
1994, 1998, 2001, 2005
1963, 1967, 1971, 1972, 1977, 1981, 1982, 1986, 1989, 1994, 1998, 2002,
2003, 2006
1960, 1963, 1966, 1969, 1972, 1975, 1978, 1981, 1984, 1987, 1990, 1993,
1996, 1999, 2002, 2005
1992, 1996, 2000, 2004
1994, 1998, 2002, 2006
1992, 1996, 2000, 2004
1979, 1982, 1986, 1989, 1993, 1996, 2004
1960, 1965, 1970
1963, 1967, 1971, 1975, 1979, 1983, 1987, 1991, 1995, 1999, 2003, 2007
1965, 1969, 1973, 1977, 1987, 1991, 1995, 1999, 2002, 2007
1994, 1998, 2002, 2006, 2007
1964, 1966, 1970, 1974, 1979, 1983, 1987, 1992, 1997, 2001, 2005
1960, 1964, 1968, 1972, 1976, 1980, 1984, 1988, 1992, 1996, 2000, 2004

139

CHAPTER 5
CONCLUSION

This dissertation project has explored the various political consequences of economic
inequality in electoral democracies. According to the MR model, greater income disparity
creates favorable conditions for left parties in electoral competition, which is likely to result
in more redistribution in democracies. While the voluminous literature has addressed the
discrepancy between the predictions of the MR model and electoral and redistributive
outcomes in the real world, this project approaches this puzzle by focusing on the overlooked
aspect of representative democracy in the MR model: the programmatic party-voter linkages
based on economic preferences. The findings of each chapter in this dissertation have
important implications on how programmatic ties between parties and citizens are severed
when economic inequality rises. Specifically: (1) rising inequality under a leftist government
induces poor voters to choose the right at the ballot box; (2) right-wing parties attempt to
politicize non-economic issues or make economic issues irrelevant to vote choice in the face
of rising inequality; and (3) inequality along ethnic lines incentivizes political entrepreneurs
to intensify ethnic appeals.
The second chapter examines why, counterintuitively, the poor do not vote for leftist
parties at the ballot box. While previous studies answer this question by focusing on potential
factors distracting the poor from their economic interests, they fail to account for the
economic and institutional contexts that may affect the poor’s voting calculus. Relying on the
extant literature on economic voting and the partisan model of economic outcomes, I theorize
that poor voters rely on changes in economic inequality to evaluate the performance of leftist
governments. More specifically, I demonstrate that the poor support leftist parties only if the
leftist government successfully advances the economic well-being of the poor by reducing

140

economic inequality. Employing a hierarchical regression analysis using survey data from 54
elections across 21 advanced democracies, I find that income-based voting decreases when
the wealth gap widens under leftist governments. This result is driven by the poor’s voting for
opposition parties in the face of rising inequality when the left is in office.
The third chapter investigates how economic inequality affects the electoral
platforms of right-wing parties in democratic elections. I argue that the different degrees of
political constraints in advanced and emerging democracies induce right-wing parties to
respond in different manners to rising economic inequality. Rising inequality increases the
right-skewedness of the income distribution and this creates the adverse condition for rightwing parties in electoral competition; the more right-skewed the income distribution is, the
more voters are located below the mean income. Rightist parties have two options to
overcome this situation: a leftward shift within the economic policy dimension and emphasis
on non-economic issues in their platform. While the former option is more effective to
enhance electoral fortune, rightist parties in advanced democracies choose to politicize social
issues in the face of high inequality. The reason is that in advanced democracies stronger
political constraints imposed on the strategic choice of party leadership curb opportunistic
policy moderation of the rightist parties. In nascent democracies, however, the right-wing
parties are able to opt for more leftist positions within the economic dimension due to weak
political constraints. I find supporting evidence for the arguments using 1754 party platforms
of 475 parties in 44 democracies from the CMP dataset.
The fourth chapter focuses on whether rising economic inequality creates mutually
cross-cutting or reinforcing ethnic cleavages of economic interests and ethnic identities.
Specifically, this chapter empirically examines political and economic conditions under
which political entrepreneurs, who seek to build electoral support bases, are incentivized to
appeal to ethnic identities, relying on previous studies on ethnic politics. I fail to find
141

consistent support for hypotheses on the effects of other determinants, whereas evidence
indicates that high BGI (or low WGI) induces political parties to intensify ethnic appeals in
their electoral platforms. The findings of this chapter imply that rising inequality does not
necessarily result in the salience of a redistribution issue. Rather, if economic inequality
increases in a way that aggravates distributional conflicts between ethnic groups, this leads
electoral competition to revolve around ethnic issues, instead of redistribution.
All the conclusions reached in this dissertation indicate that higher economic
inequality does not result in more redistribution. Although a voluminous body of literature
addresses this topic, each chapter of this dissertation contributes to the extant literature by
offering alternative mechanisms of how greater economic disparity causes the breakdown of
programmatic voter-party linkages based on economic preferences. In each chapter, I
demonstrate how rising economic inequality may induce political agents - voters and parties
– to respond it in a manner that de-emphasizes a redistribution issue in their pursuit of selfinterest in elections. If elections do not revolve around redistribution and economic issues,
unlike the assumption made in the MR model, then representative democracy fails to operate
as the redistribution mechanism that many theories of democracy predict. In this sense, this
dissertation highlights the importance of dynamics between political actors in electoral
politics in understanding the relationship between economic inequality and redistributive
outcomes in representative democracy.

142

BIBLIOGRAPHY

143

BIBLIOGRAPHY

Acemoglu, Daron, and James A. Robinson. 2001. “A Theory of Political Transition.”
American Economic Review 91(4): 938-963.
Acemoglu, Daron, and James A. Robinson. 2006. Economic Origins of Dictatorship and
Democracy. New York, NY: Cambridge University Press.
Adams, James, Michael Clark, Lawrence Ezrow, and Garret Glasgow. 2004. “Understanding
Change and Stability in Party Ideologies: Do Parties Respond to Public Opinion or to
Past Election Results?” British Journal of Political Science 34(4): 589-610.
Adams, James, Michael Clark, Lawrence Ezrow, and Garret Glasgow. 2006. “Are Niche
Parties Are Fundamentally Different from Mainstream Parties? The Causes and the
Electoral Consequences of Western European Parties’ Policy Shifts, 1976-1988.”
American Journal of Political Science 50(3): 513-529.
Adams, James, Lawrence Ezrow, and Zeynep Somer-Topcu. 2011. “Is Anybody Listening?
Evidence That Voters Do Not Respond to European Parties’ Policy Statements During
Elections.” American Journal of Political Science 55(2): 370-382.
Adams, James, Andrea B. Haupt, and Heather Stoll. 2009. “What Moves Parties? The Role of
Public Opinion and Global Economic Condition in Western Europe.” Comparative
Political Studies 42(5): 611-639.
Aldrich, John H. 1995. Why Parties? The Origin and Transformation of Political Parties in
America. Chicago, IL: University of Chicago Press.
Alesina, Alberto, Reza Baqir, and William Easterly. 1999. “Public Goods and Ethnic
Divisions.” Quarterly Journal of Economics 114(4): 1243-1284.
Alesina, Alberto, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat, and Romain
Wacziarg. 2003. “Fractionalization.” Journal of Economic Growth 8(2): 155-194.
Alesina, Alberto, RaFael Di Tella, and Robert MacCulloch. 2004. “Inequality and Happiness:
Are Europeans and Americans Different?” Journal of Public Economics 88(9-10): 20092042.
Alesina, Alberto, and Paola Giuliano. 2011. “Preferences for Redistribution.” In Handbook of
Social Economics. Jess Benhabib, Matthew O. Jackson, and Alberto Bisin, eds. San
Diego, CA: North-Holland, 93-131.
Alesina, Alberto, and Edward L. Glaeser. 2004. Fighting Poverty in the US and Europe: A
World of Difference. New York, NY: Oxford University Press.
Alesina, Alberto, and Eliana La Ferrara. 2005. “Ethnic Diversity and Economic
144

Performance.” Journal of Economic Literature 43(3): 762-800.
Alesina, Alberto, Stelios Michalopoulos, and Elias Papaioannou. 2016. “Ethnic Inequality.”
Journal of Political Economy 124(2): 428-488.
Allan, James P., and Lyle Scruggs. 2004. “Political Partisanship and Welfare State Reform in
Advanced Industrial Societies.” American Journal of Political Science 48(3): 496-512.
Alonso, Sonia. 2008. “Enduring Ethnicity: The Political Survival of Incumbent Ethnic Parties
in Western Democracies.” In Controlling Governments: Voters, Institutions, and
Accountability. Jose Maria Maravall and Ignacio Sanchez-Cuenca, eds. Cambridge, UK:
Cambridge University Press, 82-104.
Alt, James E. 1985. “Political Parties, World Demand, and Unemployment: Domestic and
International Sources of Economic Activity.” American Political Science Review 79(4):
1016-1040.
Alvarez, Michael R., and Jonathan Nagler. 1995. “Economics, Issues, and the Perot
Candidacy: Voter Choice in the 1992 Election.” American Journal of Political Science
39(3): 714-744.
Anderson, Christopher J. 1995. Blaming Government: Citizens and the Economy in Five
European Democracies. Armonk, NY: M. E. Sharpe.
Anderson, Christopher J. 2000. “Economic Voting and Political Context: A Comparative
Perspective.” Electoral Studies 19(2-3): 151-170.
Anderson, Christopher J., Matthew M. Singer. 2008. “The Sensitive Left and the Impervious
Right: Multilevel Models and the Politics of Inequality, Ideology, and Legitimacy in
Europe.” Comparative Political Studies 41(4-5): 564-599.
Angrist, Joshua D., and Jörn-Steffen Pischke. 2009. Mostly Harmless Econometrics: An
Empiricist's Companion. Princeton, NJ: Princeton University Press.
Ansolabehere, Stephen, Jonathan A. Rodden, and James M. Snyder. 2006. “Purple America.”
Journal of Economic Perspective 20(2): 97-118.
Aristotle. 1988. The Politics. Cambridge, UK: Cambridge University Press.
Armingeon, Klaus, Romana Careja, David Weisstanner, Sarah Engler, Panajotis Potolidis,
and Malene Gerber. 2013. Comparative Political Data Set III, 1990–2011. Bern:
Institute of Political Science, University of Berne.
Bäck, Hanna. 2008. “Intra-Party Politics and Coalition Formation Evidence from Swedish
Local Government.” Party Politics 14(1): 71-89.
Bader, Max. 2008. “Fluid Party Politics and the Challenge for Democracy Assistance in
Georgia.” Caucasian Review of International Affair 2(2): 1-10.

145

Bartels, Larry M. 2008. Unequal Democracy: The Political Economy of the New Gilded Age.
Princeton, NJ: Princeton University Press.
Barth, Erling, Henning Finseraas, and Karl O. Moene. 2015. “Political Reinforcement: How
Rising Inequality Curbs Manifested Welfare Generosity.” American Journal of Political
Science 59(3): 565-577.
Barth, Fredrik. 1969. Ethnic Groups and Boundaries. Prospect Heights, NY: Waveland Press.
Bates, Robert H. 1974. “Ethnic Competition and Mobilization in Contemporary Africa.”
Comparative Political Studies 6(4): 457-484.
Baldwin, Kate, and John D. Huber. 2010. “Economic versus Cultural Differences: Forms of
Ethnic Diversity and Public Goods Provision.” American Political Science Review
104(4): 644-662.
Becher, Michael, and Michael Donnelly. 2013. “Economic Performance, Individual
Evaluations, and the Vote: Investigating the Causal Mechanism.” Journal of
Politics 75(4): 968-979.
Beck, Thorsten, George Clarke, Alberto Groff, Philip Keefer, and Patrick Walsh. 2001. “New
Tools and New Tests in Comparative Political Economy: The Database of Political
Institutions.” World Bank Economic Review 15(1): 165-176.
Bélanger, Éric and Bonnie M. Meguid. 2008. “Issue Salience, Issue Ownership, and Issuebased Vote Choice.” Electoral Studies 27(3): 477-491.
Benabou, Roland. 1996. “Inequality and Growth.” NBER Macroeconomics Annual: 11-73.
Benabou, Roland. 2000. “Unequal Societies: Income Distribution and the Social Contract.”
American Economic Review 90(1): 96-129.
Benoit, Kenneth, and Michael Laver. 2006. Party Policy in Modern Democracies. London,
UK: Routledge.
Beramendi, Pablo, and Thomas R. Cusack. 2009. “Diverse Disparities: The Politics and
Economics of Wage, Market, and Disposable Income Inequlities.” Political Research
Quarterly 62(2): 257-275.
Beramendi, Pablo, and David Rueda. 2007. “Social Democracy Constrained: Indirect
Taxation in Industrialized Democracies.” British Journal of Political Science 37(4): 619641.
Bermeo, Nancy. 2009. “Does Electoral Democracy Boost Economic Equality?” Journal of
Democracy 20(4): 21-35.
Birnir, Johanna K. 2007. Ethnicity and Ethnic Politics. New York, NY: Cambridge University
Press.

146

Boix, Carles. 1998. Political Parties, Growth and Equality. New York, NY: Cambridge
University Press.
Boix, Carles. 2003. Democracy and Redistribution. New York, NY: Cambridge University
Press.
Boix, Carles, Michael K. Miller, and Sebastian Rosato. 2013. "A Complete Data Set of
Political Regimes, 1800–2007." Comparative Political Studies 46(12): 1523-1554.
Bormann, Nils-Christian, and Matt Golder. 2013. “Democratic Electoral Systems around the
World, 1946-2011.” Electoral Studies 32(2): 360-369.
Bradley, David, Evelyne Huber, Stephanie Moller, François Nielsen, and John D. Stephens.
2003. “Distribution and Redistribution in Postindustrial Democracies.” World Politics
55(2): 193-228.
Brady, David. 2003. “The Politics of Poverty: Left Political Institutions, the Welfare State,
and Poverty.” Social Forces 82(2): 557-588.
Brady, David, Andrew S. Fullerton, and Jennifer M. Cross. 2009. “Putting Poverty in Political
Context: A Multi-Level Analysis of Adult Poverty across 18 Affluent
Democracies.” Social Forces 88(1): 271-299.
Brady, David, Evelyne Huber, and John D. Stephens. 2014. Comparative Welfare States Data
Set. University of North Carolina and WZB Berlin Social Science Center.
Brambor, Thomas, William R. Clark, and Matt Golder. 2006. “Understanding Interaction
Models: Improving Empirical Analyses.” Political Analysis 14(1): 63-82.
Brass, Paul R. 1974. Language, Religioin and Politics in North India. Cambridge, UK:
Cambridge University Press.
Brass, Paul R. 1991. Ethnicity and Nationalism: Theory and Comparison. Newbury Park,
CA: Sage Publications Ltd.
Braumoeller, Bear F. 2004. “Hypothesis Testing and Multiplicative Interaction
Terms.” International Organization 58(4): 807-820.
Brooks, Clem, and David Brady. 1999. “Income, Economic Voting, and Long-Term Political
Change in the U.S., 1952-1996.” Social Forces 77(4): 1339-1374.
Brown, David S., and Ahmed Mushfiq Mobarak. 2009. “The Transforming Power of
Democracy: Regime Type and the Distribution of Electricity.” American Political
Science Review 103(2): 193-213.
Budge, Ian. 1994. “A New Spatial Theory of Party Competition: Uncertainty, Ideology and
Policy Equilibria Viewed Comparatively and Temporally.” British Journal of Political
Science 24(4): 443-467.

147

Budge, Ian, Lawrence Ezrow, and Michael D. McDonald. 2010. “Ideology, Party
Factionalism and Policy Change.” British Journal of Political Science 40(4): 781-804.
Budge, Ian, and Dennis J. Farlie. 1983. Explaining and Predicting Elections: Issue Effects
and Party Strategies in Twenty-Three Democracies. London, UK: George Allen &
Unwin.
Budge, Ian and Michael D. McDonald. 2007. “Election and Party System Effects on Policy
Representation: Bringing Time into Comparative Perspective.” Electoral Studies 26(1):
168–179.
Burgoon, Brian. 2013. “Inequality and Anti-Globalization Backlash by Political Parties.”
European Union Politics 14(3): 408-435.
Casey, Katherine. 2015. “Cross Party Lines: The Effects of Information on Redistributive
Politics.” American Economic Review 105(8): 2410-2448.
Cederman, Lars-Erik, Kristian S. Gleditsch, and Halvard Buhaug. 2013. Inequality,
Grievance, and Civil War. New York, NY: Cambridge University Press.
Cedarman, Lars-Erik, Nils B. Weidmann, and Kristian Skrede Gleditsch. 2011. “Horizontal
Inequalities and Ethno-Nationalist Civil War: A Global Comparison.” American
Political Science Review 105(3): 478-495.
Cederman, Lars-Erik, Andreas Wimmer, and Brian Min. 2010. “Why Do Ethnic Groups
Rebel? New Data and Analysis.” World Politics 62(1): 87-119.
Chandra, Kanchan. 2004. Why Ethnic Parties Succeed: Patronage and Ethnic Head Counts in
India. New York, NY: Cambridge University Press.
Chandra, Kanchan. 2005. “Ethnic Parties and Democratic Stability.” Perspectives on Politics
3(2): 235-252.
Chandra, Kanchan. ed. 2012. Constructivist Theories of Ethnic Politics. New York, NY:
Oxford University Press.
Cheibub, Jose. A., Jennifer Gandhi, and James R. Vreeland. 2010. “Democracy and
Dictatorship Revisited.” Public Choice 143(1-2): 67-101.
Choi, Joon Nak, Myung-koo Kang, and Gi-Wook Shin. 2014. “Democratic Transition and
Intraparty Politics: The Distribution of Key Party Positions in Democratizing South
Korea.” Journal of Korean Studies 19(1): 7-36.
Comparative Study of Electoral Systems (www.cses.org). 2003. CSES Module 1. Full Release
[dataset]. August 4, 2003.
Comparative Study of Electoral Systems (www.cses.org). 2007. CSES Module 2. Full Release
[dataset]. June 27, 2007.

148

Comparative Study of Electoral Systems (www.cses.org). 2013. CSES Module 3. Full Release
[dataset]. March 27, 2013.
Conroy-Krutz, Jefferey. 2013. “Information and Ethnic Politics in Africa.” British Journal of
Political Science 43(2): 345-373.
Coppedge, Michael, John Gerring, Staffan I. Lindberg, Svend-Erik Skaaning, Jan Teorell,
with David Altman, Michael Bernhard, M. Steven Fish, Adam Glynn, Allen Hicken,
Carl Henrik Knutsen, Kelly McMann, Pamela Paxton, Daniel Pemstein, Jeffrey Staton,
Brigitte Zimmerman, Rachel Sigman, Frida Andersson, Valeriya Mechkova, and Farhad
Miri. 2016. “Varieties of Democracy: Dataset v6.” Varieties of Democracy (V-Dem)
Project.
Cusack, Thomas R. 1999. “Partisan Politics and Fiscal Policy.” Comparative Political
Studies 32(4): 464-486.
Cusack, Thomas R., Torben Iversen, and Philipp Rehm. 2008. “Economic Shocks, Inequality,
and Popular Support for Redistribution.” In Democracy, Inequality, and Representation:
A Comparative Perspective. Pablo Beramendi and Christopher J. Anderson, eds. New
York, NY: Russell Sage Foundation, 203-231.
Dalton, Russell J. 2006. Citizen Politics: Public Opinion and Political Parties in Advanced
Industrial Democracies. Washington, DC: Congressional Quarterly Press.
Dalton, Russell J., and Hans-Dieter Klingemann. 2007. “Citizens and Political Behavior.” In
The Oxford Handbook of Political Behavior. Russell J. Dalton, and Hans-Dieter
Klingemann. eds. New York, NY: Oxford University Press.
Dalton, Russell J., and Steven Weldon. 2007. “Partisanship and Party System
Institutionalization.” Party Politics 13(2): 179-196.
Dalton, Russell J., and Ian McAllister. 2015. “Random Walk or Planned Excursion?
Continuity and Change in the Left-Right Positions of Political Parties.” Comparative
Political Studies 48(6): 759-787.
DeBacker, Jason M. 2015. “Flip-Flopping: Ideological Adjustment Costs in the United States
Senate.” Economic Inquiry 53(1): 108-128.
De La O, Ana L., and Jonathan A. Rodden. 2008. “Does Religion Distract the Poor? Income
and Issue Voting around the World.” Comparative Political Studies 41(4-5): 437-476.
Delli Carpini, Michael X., and Scott Keeter. 1996. What Americans Know about Politics and
Why It Matters. New Haven, CT: Yale University Press.
De Sio, Lorenzo, and Till Weber. 2014. “Issue Yield: A Model of Party Strategy in
Multidimensional Space.” American Political Science Review 108(4): 870-885.
Deutsch, Karl W. 1961. “Social Mobilization and Political Development.” American Political
Science Review 55(3): 493-514.
149

de Vries, Catherine E., and Sara B. Hobolt. 2012. “When Dimensions Collide: The Electoral
Success of Issue Entrepreneurs.” European Union Politics 13(2): 246-268.
Dixit, Avinash, and John Londregan. 1996. “The Determinants of Success of Special Interests
in Redistributive Politics.” Journal of Politics 58: 1132-1155.
Downs, Anthony. 1957. An Economic Theory of Democracy. New York, NY: Harper and
Row.
Drazen, Allan, 2000. Political Economy in Macroeconomics. Princeton, NJ: Princeton
University Press.
Dreher, Axel. 2006. “Does Globalization Affect Growth? Evidence from a New Index of
Globalization.” Applied Economics 38(10): 1091-1110.
Dreher, Axel, Noel Gaston, and Pim Martens. 2008. Measuring Globalization: Gauging Its
Consequences. New York, NY: Springer.
Duch, Raymond M., Harvey D. Palmer, and Christopher J. Anderson. 2000. “Heterogeneity
in Perceptions of National Economic Conditions.” American Journal of Political
Science 44(4): 635-652.
Duch, Raymond M., and Randolph T. Stevenson. 2008. The Economic Vote: How Political
and Economic Institutions Condition Election Results. New York, NY: Cambridge
University Press.
Duch, Raymond M., and Randolph T. Stevenson. 2011. Context and Economic Expectations:
When Do Voters Get It Right? British Journal of Political Science 41(1): 1-31.
Dunning, Thad, and Lauren Harrison. 2010. “Cross-cutting Cleavages and Ethnic Voting: An
Experimental Study of Cousinage in Mali.” American Political Science Review 104(1):
21-39.
Easterly, William, and Ross Levine. 1997. “Africa’s Growth Tragedy: Policies and Ethnic
Divisions.” Quarterly Journal of Economics 112(4): 1203-1250.
Eifert, Benn, Edward Miguel, and Daniel N. Posner. 2010. “Political Competition and Ethnic
Identification in Africa.” American Journal of Political Science 54(2): 494-510.
Enelow, James M., and Melvin J. Hinich. 1984. Spatial Theory of Voting: An Introduction.
New York, NY: Cambridge University Press.
Enyedi, Zsolt, and Lukásˇ Linek. 2008. “Searching for the Right Organization: Ideology and
Party Structure in East-Central Europe.” Party Politics 14(4): 455-477.
Erikson, Robert S. 1990. “Economic Conditions and the Congressional Vote: A Review of the
Macrolevel Evidence.” American Journal of Political Science 34(2): 373-399.

150

Erikson, Robert S., Michael B. MacKuen, and James A. Stimson. 2000. “Bankers or Peasants
Revisited: Economic Expectations and Presidential Approval.” Electoral Studies 20(23): 295-312.
Esping-Andersen, Gøsta. 1990. The Three Worlds of Welfare Capitalism. Princeton, NJ:
Princeton University Press.
Esteban, Joan, and Debraj Ray. 2011. “A Model of Ethnic Conflict.” Journal of European
Economic Association 9(3): 496-521.
Evans, Geoffrey. 1997. “Class Inequality and the Formation of Political Interests in Eastern
Europe.” European Journal of Sociology 38(2): 207-234.
Evans, Geoffrey, and James Tilley. 2012a. “The Depoliticization of Inequality and
Redistribution: Explaining the Decline of Class Voting.” Journal of Politics 74(4): 963976.
Evans, Geoffrey, and James Tilley. 2012b. “How Parties Shape Class Politics: Explaining the
Decline of the Class Basis of Party Support.” British Journal of Political Science 42(1):
137-161.
Ezrow, Lawrence, Margit Tavits, and Jonathan Homola. 2014. Voter Polarization, Strength of
Partisanship, and Support for Extremist Parties. Comparative Political Studies 47(11):
1558-1583.
Ezrow, Lawrence, Catherine E. De Vries, Marco Steenbergen, and Erica Edwards. 2011.
“Mean Voter Representation and Partisan Constituency Representation: Do Parties
Respond to the Mean Voter Position or to Their Supporters?” Party Politics 17(3): 275301.
Fearon, James, and David D. Laitin. 1996. “Explaining Interethnic Cooperation.” American
Political Science Review 90(4): 715-735.
Ferree, Karen E. 2006. “Explaining South Africa’s Racial Census.” Journal of Politics 68(4):
803-851.
Ferree, Karen E. 2012. “How Fluid Is Fluid? The Mutability of Ethnic Identities and
Electoral Volatility in Africa.” In Constructive Theories of Ethnic Politics. Kanchan
Chandra, ed. New York, NY: Oxford University Press, 312-340.
Finseraas, Henning. 2009a. “Income Inequality and Demand for Redistribution.”
Scandinavian Political Studies 32(1): 94-119.
Finseraas, Henning. 2009b.” Anti-Immigration Attitudes, Support for Redistribution, and
Vote Choice in Europe.” Norwegian Institute for Social Research Working Paper: 1-25.
Fiorina, Morris P. 1978. “Economic Retrospective Voting in American National Elections: A
Micro-Analysis.” American Journal of Political Science 22(2): 426-443.

151

Fiorina, Morris P. 1981. Retrospective Voting in American National Elections. New Haven,
CT: Yale University Press.
Frank, Thomas. 2004. What’s the Matter with Kansas? How Conservatives Won the Heart of
America. New York, NY: Henry Holt and Company.
Franzese Jr., Robert F. 2002. Macroeconomic Policies of Developed Democracies.
Cambridge, UK: Cambridge University Press.
Gadjanova, Elena. 2013. “What is an Ethnic Appeal? Policies as Metonymies for Ethnicity in
the Political Rhetoric of Group Identity.” Ethnopolitics 12(3): 307-330.
Gadjanova, Elena. 2015. “Measuring Parties’ Ethnic Appeals in democracies.” Party Politics
21(2): 309-327.
Gallagher, Mary, and Jonathan K. Hanson. 2009. “Coalitions, Carrots, and Sticks: Economic
Inequality and Authoritarian States.” Perspectives on Politics 42(4): 667-672.
Gallagher, Michael. 2015. Election Indices
(http://www.tcd.ie/Political_Science/staff/michael_gallagher/ElSystems/index.php).
Garand, James C. 2010. “Income Inequality, Party Polarization, and Roll-Call Voting in the
U.S. Senate.” Journal of Politics 72(4): 1109-1128.
Garrett, Geoffrey, and Deborah Mitchell. 2001. “Globalization, Government Spending and
Taxation in the OECD.” European Journal of Political Research 39(2): 145-177.
Gaskin, Ben, Matt Golder, and David A. Siegel. 2013. “Religious Participation and Economic
Conservatism.” American Journal of Political Science 57(4): 823-840.
Geertz, Clifford. 1973. The Interpretation of Cultures. New York, NY: Basic Books.
Gellner, Ernest. 1983. Nations and Nationalism. Ithaca, NY: Cornell University Press.
Gelman, Andrew. 2008. Red State, Blue State, Rich State, Poor State: Why Americans Vote
the Way They Do. Princeton, NJ: Princeton University Press.
Gelman, Andrew, and Jennifer Hill. 2007. Data Analysis Using Regression and
Multilevel/hierarchical Models. New York, NY: Cambridge University Press.
Gijsberts, Merove, and Paul Niewbeerta. 2000. “Class Cleavages in Party Preferences in the
New Democracies in Eastern Europe: A Comparison with Western Democracies.”
European Societies 2(4): 397-430.
Green, Jane, and Sara B. Hobolt. 2008. “Owning the Issue Agenda: Party Strategies and Vote
Choices in British Elections.” Electoral Studies 27(3): 460-476.
Green-Pedersen, Christoffer. 2007. “The Growing Importance of Issue Competition: The
Changing Nature of Party Competition in Western Europe.” Political Studies 55(3): 607152

628.
Gubler, Joshua, and Joel S. Selway. 2012. “Horizontal Equality, Cross-cutting Cleavages and
Civil War.” Journal of Conflict Resolution 56(2): 206-232.
Gurr, Ted. 1970. Why Men Rebel. Princeton, NJ: Princeton University Press.
Habyarimana, James, Macartan Humphreys, Daniel N. Posner, and Jeremy M. Weinstein.
2007. “Why Does Ethnic Diversity Undermine Public Goods Provision?” American
Political Science Review 101(4): 709-725.
Haupt, Andrea B. 2010. “Parties’ Responses to Economic Globalization: What Is Left for the
Left and Right for the Right?” Party Politics 16(1): 5-27.
Häusermann, Silja, Georg Picot, Dominik Geering. 2013. “Rethinking Party Politics and the
Welfare State: Recent Advances in the Literature.” British Journal of Political Science
43(1): 221-240.
Hechter, Michael. 1975. Internal Colonialism: The Celtic Fringe in British National
Development. London, UK: Routledge and Kegan Paul.
Hibbs, Douglas A. 1977. “Political Parties and Macroeconomic Policy.” American Political
Science Review 71(4): 1467-1487.
Hibbs, Douglas A. 1987. The American Political Economy. Cambridge, MA: Harvard
University Press.
Higashijima, Masaaki, and Ryo Nakai. 2016. “Elections, Ethnic Parties, and Ethnic
Identification in New Democracies: Evidence from the Baltic States.” Studies in
Comparative International Development 51(2): 124-146.
Hix, Simon, Abdul Noury, and Gérard Roland. 2006. “Dimensions of Politics in the European
Parliament.” American Journal of Political Science 50(2): 494-511.
Hobolt, Sarah B., and Catherine E. de Vries. 2015. “Issue Entrepreneurship and Multiparty
Competition.” Comparative Political Studies 48(9): 1159-1185.
Horowitz, Donald L. 1985. Ethnic Groups in Conflict. Berkeley, CA: University of California
Press.
Horowitz, Donald L. 1991. A Democratic South Africa? Constitutional Engineering in a
Divided Society. Berkeley, CA: University of California Press.
Houle, Christian. 2015. “Ethnic Inequality and the Dismantling of Democracy: A Global
Analysis.” World Politics 67 (3): 469-505.
Huber, Evelyne, and John D. Stephens. 2012. Democracy and the Left: Social Policy and
Inequality in Latin America. Chicago, IL: University of Chicago Press.

153

Huber, John D. 2012. “Measuring Ethnic Voting: Do Proportional Electoral Laws Politicize
Ethnicity?” American Journal of Political Science 56(4): 986-1001.
Huber, John D., and Piero Stanig. 2007. “Why Do the Poor Support Right-Wing Parties? A
Cross-National Analysis.” Unpublished paper, Columbia University.
Huber, John D., and Piero Stanig. 2009. “Individual Income and Voting for Redistribution.”
Working Paper, September 9.
Huber, John D., and Piero Stanig. 2011. “Church-State Separation and Redistribution.”
Journal of Public Economics 95(7): 828-836.
Huber, John D, and Pavithra Suryanarayan. 2016. “Ethnic Inequality and the Ethnification of
Political Parties: Evidence from India.” World Politics 68(1): 149-188.
Huber, Peter J. 1967. “The Behavior of Maximum Likelihood Estimates under Nonstandard
Conditions.” In Proceedings of the Fifth Berkeley Symposium in Mathematical Statistics
and Probability, Vol.1: Statistics. Berkeley, CA: University of California Press, 221-233.
Iversen, Torben. 2005. Capitalism, Democracy, and Welfare. New York, NY: Cambridge
University Press.
Iversen, Torben, and David Soskice. 2001. “An Asset Theory of Social Policy Preferences.”
American Political Science Review 95(4): 875-893.
Iversen, Torben, and David Soskice. 2006. “Electoral Institutions and the Politics of
Coalition: Why Some Democracies Redistribute More Than Others.” American Political
Science Review 100(2): 165-181.
Kayser, Mark A., and Michael Peress. 2012. “Benchmarking across Borders: Electoral
Accountability and the Necessity of Comparison.” American Political Science Review
106(3): 661-684.
Keech, William R. 1995. Economic Politics: The Costs of Democracy. New York, NY:
Cambridge University Press.
Keefer, Philip. 2007. “Clientelism, Credibility, and the Policy Choices of Young
Democracies.” American Journal of Political Science 51(4): 804-821.
Kenworthy, Lane, and Jonas Pontusson. 2005. “Rising Inequality and the Politics of
Redistribution in Affluent Countries.” Perspectives on Politics 3(3): 449-471.
Kernell, Samuel. 1978. “Explaining Presidential Popularity.” American Political Science
Review 72(2): 506-522.
Key, V. O. Jr. 1966. The Responsible Electorate. New York, NY: Vintage Books.
Kiewiet, Roderick. 1983. Macroeconomics and Micropolitics: The Electoral Effects of
Economic Issues. Chicago, IL: University of Chicago Press.
154

Kinder, Donald R., and Roderick Kiewiet. 1979. “Economic Discontents and Political
Behavior: The Role of Personal Grievances and Collective Economic Judgments in
Congressional Voting.” American Journal of Political Science 23(3): 495-527.
Kinder, Donald R., and Roderick Kiewiet. 1981. “Sociotropic Politics: The American Case.”
British Journal of Political Science 11(2): 129-161.
Kinder, Donald R., Gordon Adams, and Paul Gronke. 1989. “Economics and Politics in the
1984 American Presidential Election.” American Journal of Political Science 33(2):
491-515.
King, Desmond, and David Rueda. 2008. “Cheap Labor: The New Politics of “Bread and
Roses” in Industrial Democracies.” Perspectives on Politics 6(2): 279-297.
Kitschelt, Herbert. 1992. “The Formation of Party Systems in East Central Europe.” Politics
and Society 20(1): 7-50.
Kitschelt, Herbert. 1994. The Transformation of European Social Democracy. New York,
NY: Cambridge University Press.
Kitschelt, Herbert. 1995. “Formation of Party Cleavages in Post-Communist Democracies:
Theoretical Propositions.” Party Politics 1(4): 447-472.
Kitschelt, Herbert, and Philipp Rehm. 2015. “Party Alignments: Change and Continuity.” In
The Politics of Advanced Capitalism. Pablo Beramendi, Silja Häusermann, Herbert
Kitschelt, and Hanspeter Kriesi, eds. New York, NY: Cambridge University Press, 179201.
Klingemann, Hans-Dieter, Andrea Volkens, Judith Bara, Ian Budge, and Michael McDonald.
2006. Mapping Policy Preferences II: Estimates for Parties, Electors, and Governments
in Eastern Europe. New York, NY: Oxford University Press.
Klor, Esteban F., and Moses Shayo. 2010. “Social Identity and Preferences over
Redistribution.” Journal of Public Economics 94(3): 269-278.
Klüver, Heike, and Iñaki Sagarzazu. 2016. “Setting the Agenda or Responding to Voters?
Political Parties, Voters and Issue Attention.” West European Politics 39(2): 380-398.
Korpi, Walter. 1983. The Democratic Class Struggle. London, UK: Routledge and Kegan
Paul.
Korpi, Walter, and Joakim Palme. 2003. “New Politics and Class Politics in the Context of
Austerity and Globalization: Welfare State Regress in 18 Countries, 1975–
95.” American Political Science Review 97(3): 425-446.
Kramer, Gerald H. 1971. “Short-term Fluctuation in U.S. Voting Behavior, 1896-1964.”
American Political Science Review 65(1): 131-145.

155

Kuhn, Patrick M., and Nils B. Weidmann. 2015. “Unequal We Fight: Between- and WithinGroup Inequality and Ethnic Civil War.” Political Science Research and Methods 3(3):
543-568.
Lake, David, and Matthew Baum. 2001. “The Invisible Hand of Democracy: Political Control
and the Provision of Public Services.” Comparative Political Studies 34(6): 587-621.
Leighley, Jan E., and Jonathan Nagler. 2013. Who Vote Now? Demographics, Issues,
Inequality, and Turnout in the United States. Princeton, NJ: Princeton University Press.
Lewis-Beck, Michael S. 1988. Economics and Elections: The Major Western Democracies.
Ann Arbor, MI: University of Michigan Press.
Lewis-Beck, Michael S., and Martin Paldam. 2000. “Economic Voting: An Introduction.”
Electoral Studies 19(1): 13-21.
Lewis-Beck, Michael S., and Mary Stegmaier. 2000. “Economic Determinants of Electoral
Outcomes.” Annual Review of Political Science 3(1): 183-219.
Lewis-Beck, Michael S., and Mary Stegmaier. 2007. “Economic Models of Voting.” In The
Oxford Handbook of Political Behavior. Russell Dalton and Hans-Dieter Klingemann,
eds. New York, NY: Oxford University Press, 518-537.
Lijphart, Arend. 1977. Democracy in Plural Societies. New Haven, CT: Yale University
Press.
Lijphart, Arend. 1999. Patterns of Democracies: Government Forms and Performances in 35
Countries. New Haven, CT: Yale University Press.
Linn, Suzanna, and Jonathan Nagler. 2014. “Economic Voting and Economic Inequality: US
Presidential Elections 1952-2012.” In MPSA Conference, April 3-6.
Lipset, Seymour M. 1959. “Some Social Requisites of Democracy: Economic Development
and Political Legitimacy.” American Political Science Review 53(1): 69-105.
Lipset, Seymour M. 1960. Political Man: The Social Basis of Politics. New York, NY:
Doubleday.
Lupu, Noam. 2013. “Party Brands and Partisanship: Theory with Evidence from a Survey
Experiment in Argentina.” American Journal of Political Science 57(1): 49-64.
Lupu, Noam. 2014. “Brand Dilution and the Breakdown of Political Parties in Latin
America.” World Politics 66(4): 561-602.
Lupu, Noam, and Jonas Pontusson. 2011. “The Structure of Inequality and the Politics of
Redistribution.” American Political Science Review 105(2): 316-336.
Lupu, Noam, and Rachel B. Riedl. 2013. “Political Parties and Uncertainty in Developing
Democracies.” Comparative Political Studies 46(11): 1339-1365.
156

Lupu, Noam, and Susan Stokes. 2010. “Democracy, Interrupted: Regime Change and
Partisanship in Twentieth-century Argentina.” Electoral Studies 29(1): 91-104.
MacKuen, Michael B., Robert S. Erikson, and James A. Stimson. 1992. “Peasant or Bankers?
The American Electorate and the U.S. Economy.” American Political Science Review
86(3): 597-611.
Mahler, Vincent A. 2002. “Exploring the Subnational Dimension of Income Inequality: An
Analysis of the Relationship between Inequality and Electoral Turnout in the Developed
Countries.” International Studies Quarterly 46(1): 117-142.
Mainwaring, Scott. 1999. Rethinking Party Systems Theory in the Third Wave of
Democratization. Stanford, CA: Stanford University Press.
Mainwaring, Scott, and Mariano Torcal. 2006. “Party System Institutionalization and Party
System Theory after the Third Wave of Democratization.” In Handbook of Party
Politics. Richard S. Katz and William Crotty, eds. London, UK: SAGE Publications Ltd,
204-227.
Marra, Robin F., and Charles W. Ostrom. 1989. “Explaining Seat Change in the US House of
Representative, 1950-1986.” American Journal of Political Science 33((3): 541-569.
Marx, Karl. 1971. Wage-labour and Capital. New York, NY: International Publishers.
McAllister, Ian, and Stephen White. 2007. “Political Parties and Democratic Consolidation in
Post-Communist Societies.” Party Politics 13(2): 197-216.
McCarthy, Nolan, Keith T. Poole, and Howard Rosenthal. 2006. Polarized America: The
Dance of Ideology and Unequal Riches. Cambridge, MA: MIT Press.
Meguid, Bonnie M. 2005. Competition between Unequals: The Role of Mainstream Party
Strategy in Niche Party Success. American Political Science Review 99(3): 347-359.
Meguid, Bonnie M. 2008. Party Competition between Unequals: Strategies and Electoral
Fortunes in Western Europe. New York, NY: Cambridge University Press.
Meltzer, Allan H., and Scott F. Richard. 1981. “A Rational Theory of the Size of
Government.” Journal of Political Economy 89(5): 914-927.
Meyer, Thomas M., and Wolfgang C. Mueller. 2014. “Testing Theories of Party Competition:
The Austrian Case.” Party Politics 20(5): 802-813.
Milanovic, Branko. 2000. “The Median-Voter Hypothesis, Income Inequality, and Income
Redistribution: An Empirical Test with the Required Data.” European Journal of
Political Economy 16(3): 367-410.
Miller, Gary, Norman Schofield. 2003. “Activists and Partisan Realignment in the United
States.” American Political Science Review 97(2): 245-260.
157

Moene, Karl Ove, and Michael Wallerstein. 2001. “Inequality, Social Insurance, and
Redistribution.” American Political Science Review 95: 859-874.
Moroff, Anika. 2010. “Party Bans in Africa: An Empirical Overview.” Democratization
17(4): 618-641.
Moulton, Brent R. 1986. “Random Group Effects and the Precision of Regression Estimates.”
Journal of Econometrics 32(3): 385-397.
Moulton, Brent R. 1990. “An Illustration of a Pitfall in Estimating the Effects of Aggregate
Variables on Micro Units.” Review of Economics and Statistics 72(2): 334-338.
Mulas-Granados, Carlos. 2003. “The Political and Economic Determinants of Budgetary
Consolidation in Europe.” European Political Economy Review 1(1): 15-39.
Nadeau, Richard, and Michael S. Lewis-Beck. 2000. “French Electoral Institutions and the
Economic Vote.” Electoral Studies 19(2-3): 171-182.
Nadeau, Richard, and Michael S. Lewis-Beck. 2001. “National Economic Voting in U.S.
Presidential Elections.” Journal of Politics 63(1): 159-181.
Norpoth, Helmut. 1996. “Presidents and Prospective Voter.” Journal of Politics 58(3): 776792.
Norris, Pippa. 2004. Electoral Engineering: Voting Rules and Political Behavior. New York,
NY: Cambridge University Press.
Norris, Pippa. 2008. Driving Democracy: Do Power-sharing Institutions Work? New York,
NY: Cambridge University Press.
Norris, Pippa. 2012. Making Democratic Governance Work: How Regime Shape Prosperity,
Welfare, and Peace. New York, NY: Cambridge University Press.
OECD. 2011. An Overview of Growing Income Inequalities in OECD Countries.
(http://www.oecd.org/els/soc/49499779.pdf).
OECD. 2012. “Income Inequality and Growth: The Role of Taxes and Transfers.” OECD
Economics Department Policy Notes 9. (http://www.oecd.org/eco/publicfinance/49417295.pdf).
Ordeshook, Peter C., and Olga Shvetsova. 1994. “Ethnic Heterogeneity, District Magnitude,
and the Number of Parties.” American Journal of Political Science 38(1): 100-123.
Østby, Gudrun. 2008a. “Polarization, Horizontal Inequalities and Violent Civil Conflict.”
Journal of Peace Research 45(2): 143-162.
Østby, Gudrun. 2008b. ‘‘Horizontal Inequalities, Political Environment and Civil Conflict:
Evidence from 55 Developing Countries.’’ In Horizontal Inequalities and Conflict:
158

Understanding Group Violence in Multiethnic Societies. Frances Stewart, ed.
Basingstoke, UK: Palgrave Macmillan, 136-157.
Panebianco, Angelo. 1988. Political Parties: Organization and Power. Cambridge, UK:
Cambridge University Press.
Pedersen, Helene H. 2010. “How Intra-Party Power Relations Affect the Coalition Behaviour
of Political Parties.” Party Politics 16(6): 737-754.
Pedersen, Helene H. 2012. “What Do Parties Want? Policy Versus Office.” West European
Politics 35(4): 896-910.
Perotti, Roberto. 1996. “Growth, Income Distribution, and Democracy: What the Data Say.”
Journal of Economic Growth 1(2): 149-187.
Persson, Torsten, and Guido Tabellini. 2000. Political Economics: Explaining Economic
Policy. Cambridge, MA: MIT Press.
Persson, Torsten, and Guido Tabellini. 2003. The Economic Effects of Constitutions.
Cambridge, MA: MIT Press.
Petersen, Roger D. 2002. Understanding Ethnic Violence: Fear, Hatred, and Resentment in
Twentieth-century Eastern Europe. New York, NY: Cambridge University Press.
Petrocik, John R. 1996. “Issue Ownership in Presidential Elections, with a 1980 Case Study.”
American Journal of Political Science 40(3): 825-850.
Pitkin, Hanna F. 1967. The Concept of Representation. Berkeley, CA: University of
California Press.
Plümper, Thomas, Vera E. Troeger, and Hannes Winner. 2009. “Why Is There No Race to the
Bottom in Capital Taxation?” International Studies Quarterly 53(3): 761-786.
Pontusson, Jonas, and David Rueda. 2008. “Inequality as a Source of Political Polarization: A
Comparative Analysis of Twelve OECD Countries.” In Democracy, Inequality, and
Representation: A Comparative Perspective. Pablo Beramendi and Christopher J.
Anderson, eds. New York, NY: Russell Sage Foundation, 312-353.
Pontusson, Jonas, and David Rueda. 2010. “The Politics of Inequality: Voter Mobilization
and Left Parties in Advanced Industrial States.” Comparative Political Studies 43(6):
675-705.
Posen, Barry R. 1993. “The Security Dilemma and Ethnic Conflict.” In Ethnic Conflict and
International Security. Michael E. Brown. ed. Princeton, NJ: Princeton University Press,
103-125.
Posner, Daniel N. 2004. “The Political Salience of Cultural Difference: Why Chewas and
Tumbukus Are Allies in Zambia and Adversaries in Malawi?” American Political
Science Review 98(4): 529-545.
159

Posner, Daniel N. 2005. Institutions and Ethnic Politics in Africa. New York, NY: Cambridge
University Press.
Powell, Bingham G. 2000. Elections as Instruments of Democracy. New Haven, CT: Yale
University Press.
Powell, Bingham G. 2006. “Election Laws and Representative Government: Beyond Votes
and Seats.” British Journal of Political Science 36(2): 291-315.
Powell, Bingham G. 2009. “The Ideological Congruence Controversy: The Impact of
Alternative Measures, Data, and Time Periods on the Effects of Election Rules.”
Comparative Political Studies 42(12):1475–1497.
Powell, Bingham G., and Georg S. Vanberg. 2000. “Election Laws, Disproportionality, and
Median Correspondence: Implications for Two Visions of Democracy.” British Journal
of Political Science 30(3): 383-411.
Powell, Bingham G., and Guy D. Whitten. 1993. “A Cross-National Analysis of Economic
Voting: Taking into Account Political Context.” American Journal of Political Science
21(2): 501-522.
Prior, Markus, and Arthur Lupia. 2008. “Money, Time, and Political Knowledge:
Distinguishing Quick Recall and Political Learning Skills.” American Journal of
Political Science 52(1): 169-183.
Protsyk, Oleh, and Stela Garaz. 2013. “Politicization of Ethnicity in Party Manifestos.” Party
Politics 19(2): 296-318.
Przeworski, Adam, Michael Alvarez, Jose Antonio Cheibub, and Fernando Limongi. 2000.
Democracy and Development: Political Institutions and Well-Being in the World, 19501990. New York, NY: Cambridge University Press.
Rabushka, Alvin, and Kenneth Shepsle. 1971. “Political Entrepreneurship and Patterns of
Democratic Instability in Plural Societies.” Race and Class 12(4): 461-476.
Rabushka, Alvin, and Kenneth Shepsle. 1972. Politics in Plural Societies: A Theory of
Democratic Instability. Columbus, OH: Charles E. Merrill.
Raymond, Christopher, and Moises Arce. 2013. “The Politicization of Indigenous Identities
in Peru.” Party Politics 19(4): 555-576.
Rehm, Philipp. 2009. “Risks and Redistribution: An Individual-level Analysis.” Comparative
Political Studies 42(7): 855-81.
Riker, William H. 1986. The Art of Political Manipulation. New Haven, CT: Yale University
Press.
Reilly, Benjamin. 2001. Democracy in Divided Societies: Electoral Engineering for Conflict
160

Management. New York, NY: Cambridge University Press.
Rodrik, Dani. 1998. “Why Do More Open Economies Have Bigger Government?” Journal of
Political Economy 106(5): 997-1032.
Roemer, John E. 1998. “Why the Poor Do Not Expropriate the Rich: An Old Argument in
New Garb.” Journal of Public Economics 70(3): 399-424.
Rose, Richard. 1995. “Mobilizing Demobilized Voters in Post-Communist Societies.” Party
Politics 1(4): 549-563.
Rose, Richard, and Neil Munro. 2003. Elections and Parties in New European Democracies.
Washington D.C.: CQ Press.
Ross, Michael. 2006. “Is Democracy Good for the Poor?” American Journal of Political
Science 50(4): 860-874.
Rueda, David. 2005. “Insider-Outsider Politics in Industrialized Democracies: The Challenge
to Social Democratic Parties.” American Political Science Review 99(1): 61-74.
Rueda, David. 2008. Social Democracy Inside Out: Partisanship and Labor Market Policy in
Industrialized Democracies. New York, NY: Oxford University Press.
Samuels, David. 2004. “Presidentialism and Accountability for the Economy in Comparative
Perspective.” American Political Science Review 98(3):425-436.
Satori, Giovanni. 1970. “Concept Misformation in Comparative Politics.” American Political
Science Review 64(4): 1033-1053.
Scarrow, Susan E. 2010. “Political Parties and Party Systems.” In Comparing democracies:
Elections and voting in the 21st century. Lawrence LeDuc, Richard G. Niemi, and Pippa
Norris, eds. London, UK: Sage Publications Ltd, 45-64.
Schattschneider, E. E. 1960. The Semisovereign People: A Realist’s View of Democracy in
America. New York, NY: Holt, Rinehart and Winston.
Scheve, Kenneth, and David Stasavage. 2006. “Religion and Preferences for Social
Insurance.” Quarterly Journal of Political Science 1(3): 255-286.
Schmidt, Alexander W., and Dennis C. Spies. 2014. “Do Parties “Playing the Race Card”
Undermine Natives’ Support for Redistribution? Evidence from Europe.” Comparative
Political Studies 47(4): 519-549.
Schofield, Norman, and Itai Sened. 2006. Multiparty Democracy: Elections and Legislative
Politics. New York, NY: Cambridge University Press.
Schumacher, Gijs, Catherine E. de Vries, and Barbara Vis. 2013. “Why Do Parties Change
Position? Party Organization and Environmental Incentives.” Journal of Politics 75(2):
464-477.
161

Selway, Joel S. 2015. Coalitions of the Well-being: How Electoral Rules and Ethnic Politics
Shape Health Policy in Developing Countries. New York, NY: Cambridge University
Press.
Shayo, Moses. 2009. “A Model of Social Identity with an Application to Political Economy:
Nation, Class, and Redistribution.” American Political Science Review 103(2): 147-174.
Solt, Frederick. 2009. “Standardizing the World Income Inequality Database.” Social Science
Quarterly 90(2): 231-242. SWIID Version 4.0, Septermber 2013.
Somer-Topcu, Zeynep. 2009. “Timely Decisions: The Effects of Past National Elections on
Party Policy Change.” Journal of Politics 71(1): 238-248.
Somer-Topcu, Zeynep. 2015. “Everything to Everyone: The Electoral Consequences of the
Broad-Appeal Strategy in Europe.” American Journal of Political Science 59(4): 841854.
Spoon, Jae-Jae, and Heike Klüver. 2014. “Do Parties Respond? How Electoral Context
Influences Party Responsiveness.” Electoral Studies 35: 48-60.
Stasavage, David. 2005. “Democracy and Education Spending in Africa.” American Journal
of Political Science 49(2): 343-358.
Steenbergen, Marco R., and Bradford Jones. 2002. “Modeling Multilevel Data Structures.”
American Journal of Political Science 46(1): 218-237.
Stegmueller, Daniel. 2013. “Religion and Redistributive Voting in Western Europe.” Journal
of Politics 75(4): 1064-1076.
Steiner, Nils D., and Christian W. Martin. 2012. “Economic Integration, Party Polarisation
and Electoral Turnout.” West European Politics 35(2): 238-265.
Stephens, John D. 1979. The Transition from Capitalism to Socialism. Urbana, IL: University
of Illinois Press.
Stewart, Frances, 2000. ‘‘Crisis Prevention: Tackling Horizontal Inequalities.’’ Oxford
Development Studies 28(3): 245-262.
Stiglitz, Joseph E. 2012. The Price of Inequality: How Today’s Divided Society Endangers
Our Future. New York, NY: W. W. Norton & Company.
Stokes, Donald E. 1963. “Spatial Models of Party Competition.” American Political Science
Review 57(2): 368-377.
Strøm, Kaare. 1990. “A Behavioral Theory of Competitive Political Parties.” American
Journal of Political Science 34(2): 565-598.
Strøm, Kaare, and Wolfgang C. Müller. 1999. “Political Parties and Hard Choices. In Policy,
162

office, or votes? How political parties in Western Europe make hard decisions.
Wolfgang C. Müller and Kaare Strøm, eds. Cambridge, UK: Cambridge University
Press, 1-35.
Swank, Duane. 2013. Comparative Political Parties Dataset: Electoral, Legislative, and
Government Strength of Political Parties by Ideological Group in 21 Capitalist
Democracies, 1950–2011. Electronic database. Department of Political Science, Marquette
University (http://www.marquette.edu/polisci/documents/part19502011.xls).
Tabachnick, Barbara G., and Linda S. Fidell. 2001. Using Multivariate Statistics (4th ed.).
Boston, MA: Allyn and Bacon.
Tavits, Margit. 2006. “Party System Change Testing a Model of New Party Entry.” Party
Politics 12(1): 99-119.
Tavits, Margit, and Natalia Letki. 2014. “From Values to Interests? The Evolution of Party
Competition in New Democracies.” Journal of Politics 76(1): 246-258.
Tavits, Margit, and Joshua D. Potter. 2015. “The Effect of Inequality and Social Identity on
Party Strategies.” American Journal of Political Science 59(3): 744-758.
Timmons, Jeffrey F. 2010. “Does Democracy Reduce Economic Inequality?” British Journal
of Political Science 40(4): 741-757.
Tocqueville, Alexis de. 2010 [1835]. Democracy in America: Historical-Critical Edition.
Eduardo Nolla. ed. translated from the French by James T. Schleifer. Vol. 2.
Indianapolis, IN: Liberty Fund.
Tomz, Michael, and Robert van Houweling. 2014. Political Repositioning: A Conjoint
Analysis. Unpublished Manuscript, Stanford University
(https://web.stanford.edu/~tomz/working/TomzVanHouweling-2014-08-15.pdf).
Tucker, Joshua A. 2006. Regional Economic Voting: Russia, Poland, Hungary, Slovakia, and
the Czech Republic. New York, NY: Cambridge University Press.
Tufte, Edward R. 1978. Political Control of the Economy. Princeton, NJ: Princeton
University Press.
Van Biezen, Ingrid. 2003. Political Parties in New Democracies: Party Organization in
Southern and East-Central Europe. New York, NY: Palgrave MacMillan.
Van Biezen, Ingrid. 2005. “On the Theory and Practice of Party Formation and Adaptation in
New Democracies.” European Journal of Political Research 44(1): 147-174.
Verba, Sidney, Kay L. Schlozman, and Henry Brady. 1995. Voice and Equality: Civic
Voluntarism in American Politics. Cambridge, MA: Harvard University Press.
Volkens, Andrea, Pola Lehmann, Theres Matthieß, Nicolas Merz, Regel Sven, and Annika
Werner. 2015. The Manifesto Data Collection: Manifesto Project
163

(MRG/CMP/MARPOR), Version 2015a. Berlin: Wissenschaftszentrum Berlin für
Sozialforschung (WZB).
Wagner, Markus. 2012. “When Do Parties Emphasise Extreme Positions? How Strategic
Incentives for Policy Differentiation Influence Issue Importance.” European Journal of
Political Research 51(1): 64-88.
Wagner, Markus, and Thomas M. Meyer. 2014. “Which Issues Do Parties Emphasise?
Salience Strategies and Party Organization in Multiparty Systems.” Western European
Politics 37(5): 1019-1045.
Ward, Hugh, Lawrence Ezrow, and Han Dorussen. 2011. “Globalization, Party Positions, and
the Median Voter.” World Politics 63(3): 509-547.
Ward, Dalston, Jeong Hyun Kim, Matthew Graham, and Margit Tavits. 2015. “How
Economic Integration Affects Party Issue Emphases.” Comparative Political Studies
48(10): 1227-1259.
Ware, Alan. 1996. Political Parties and Party Systems. New York, NY: Oxford University
Press.
White, Halbert. 1980. “A Heteroskedasticity-consistent Covariance Matrix Estimator and a
Direct Test for Heteroskedasticity.” Econometrica 48: 817-838.
Whitten, Guy D., and Harvey D. Palmer. 1999. “Cross-National Analyses of Economic
Voting.” Electoral Studies 18(1): 49-67.
Wilkinson, Steven. 2004. Votes and Violence: Electoral Competition and Ethnic Riots in
India. New York, NY: Cambridge University Press.
Williams, Laron K, and Guy D. Whitten. 2015. “Don’t Stand So Close to Me: Spatial
Contagion Effects and Party Competition.” American Journal of Political Science 59(2):
309-325.
Wimmer, Andreas, Lars-Erik Cederman, and Brian Min. 2009. “Ethnic Politics and Armed
Conflict: A Configurational Analysis of a New Global Data Set.” American Sociological
Review 74(2): 316-337.
Yong, An Gie, and Sean Pearce. 2013. “A Beginner’s Guide to Factor Analysis: Focusing on
Explanatory Factor Analysis.” Tutorials in Quantitative Methods for Psychology 9(2):
79-94.

164