University of Southern California Dissertations and Theses

The erosion of democracy in the post–Cold War world: understanding the phenomenon and its causes

(USC Thesis Other)

The erosion of democracy in the post–Cold War world: understanding the phenomenon and its causes

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

THE EROSION OF DEMOCRACY IN THE POST-COLD WAR WORLD:
UNDERSTANDING THE PHENOMENON AND ITS CAUSES

by

Dávid Somogyi

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(POLITICAL SCIENCE AND INTERNATIONAL RELATIONS)

August 2021

Copyright 2021 Dávid Somogyi
ii

Table of Contents

List of Tables iv
List of Figures vi
Abstract vii
Chapter 1. Introduction 1
Chapter 2. Literature Review 7
2.1. Description 7
2.2. Operationalization 11
2.3. Democratic Erosion and the Media 12
2.4. Explanation 18
2.5. Conclusion 25
Chapter 3. The Building Blocks of Democratic Erosion: An In-depth case Study 28
of Hungary
3.1. Introduction 28
3.2. Theory 31
3.2.1. Analytical Categories 31
3.2.2. Sequence of Steps 39
3.3. Data Analysis 42
3.3.1. Introduction of the Dataset 43
3.3.2. Analytical Categories 46
3.3.3. Sequence of Steps 53
3.4. Conclusion 63
Chapter 4. Media Freedom and the Erosion of Democracy: A Comparative Case 65
Study of Five Countries
4.1. Introduction 65
4.2. Theory 67
4.2.1. Analytical Categories 67
4.2.2. Sequence of Steps 76
4.3. Data Analysis 83
4.3.1. Introduction of the Dataset 84
4.3.2. Analytical Categories 88
4.3.3. Sequence of Steps 98
iii

4.4. Conclusion 108
Appendix 4.1 111
Appendix 4.2 112
Chapter 5. The Warning Signs: Explaining Vulnerability to Democratic Erosion 114
5.1. Introduction 114
5.2. Theory 115
5.3. Data Analysis 120
5.3.1. Polarization and the Quality of Democracy 120
5.3.2. Polarization and the Onset of Democratic Erosion 123
5.4. Conclusion 129
Chapter 6. Conclusion 131
References 137

iv

List of Tables

Chapter 1
Table 1.1. Summary of Empirical Chapters 6

Chapter 3
Table 3.1. Timing of the steps of democratic erosion by category 42
Table 3.2. Coding of the steps of democratic erosion: analytical categories 44
Table 3.3. Coding of the steps of democratic erosion: weight 45
Table 3.4. Descriptive statistics: analytical categories 46
Table 3.5. Descriptive statistics: weight 46
Table 3.6. Steps of erosion not classified into any of analytical groups 1-5 51
Table 3.7. Color coding of the steps of democratic erosion according to weight 53
Table 3.8. Distribution of the steps of erosion by group in the three terms of the Orbán 61
government

Chapter 4
Table 4.1. Steps of erosion: analytical groups 75
Table 4.2. Tools at the government’s disposal to effectuate steps of erosion 79
Table 4.3. Classification of the analytical groups into the phases of democratic erosion 83
Table 4.4. Hypotheses 84
Table 4.5. Coding of the weight of individual steps in the dataset 87
Table 4.6. Examples from the dataset 87
Table 4.7. Descriptive statistics: analytical categories 87
Table 4.8. Descriptive statistics: weight 88
Table 4.9. Analytical group M1: steps of dem. erosion coded as major or significant 88
Table 4.10. Analytical group M1: examples from the dataset (major or significant) 89
Table 4.11. Analytical group M2: steps of dem. erosion coded as major or significant 89
Table 4.12. Analytical group M2: examples from the dataset (major) 90
v

Table 4.13. Analytical group M3: steps of dem. erosion coded as major or significant 90
Table 4.14. Analytical group M3: examples from the dataset (major) 91
Table 4.15. Analytical group M4: steps of dem. erosion coded as major or significant 91
Table 4.16. Analytical group M4: examples from the dataset (major) 92
Table 4.17. Analytical group M5: steps of dem. erosion coded as major or significant 92
Table 4.18. Analytical group M5: examples from the dataset (major) 92
Table 4.19. Analytical group M6: steps of dem. erosion coded as major or significant 93
Table 4.20. Analytical group M6: examples from the dataset (major or significant) 93
Table 4.21. Analytical group M7: examples from the dataset 94
Table 4.22. Analytical group M8: steps of dem. erosion coded as major or significant 95
Table 4.23. Analytical group M8: examples from the dataset (major) 95
Table 4.24. Steps of erosion not classified into any of analytical groups M1-8 96
Table 4.25. Timing of the steps of erosion: theory vs. data for each analytical group 105
Table 4.26. Descriptive statistics: analytical categories (individual countries) 111
Table 4.27. Descriptive statistics: weight (individual countries) 111

Chapter 5
Table 5.1. List of variables used in the analysis (linear regression) 121
Table 5.2. Linear regression results 121
Table 5.3. Cases of democratic erosion 124
Table 5.4. Year the Eroder was elected 126
Table 5.5. List of variables used in the analysis (logistic regression) 127
Table 5.6. Logistic regression results 127
Table 5.7. Logistic regression results (Y-1 polarization) 129

vi

List of Figures

Chapter 3
Figure 3.1. The political process and the government’s points of attack 33
Figure 3.2. Timing of the steps of democratic erosion by category 42
Figure 3.3. Analytical group 1: temporal distribution of the steps of erosion 54
Figure 3.4. Analytical group 2: temporal distribution of the steps of erosion 55
Figure 3.5. Analytical group 3: temporal distribution of the steps of erosion 57
Figure 3.6. Analytical group 4: temporal distribution of the steps of erosion 58
Figure 3.7. Analytical group 5: temporal distribution of the steps of erosion 59
Figure 3.8. Temporal distribution of the steps of erosion not in analytical groups 1 to 5 60
Figure 3.9. Temporal distribution of the steps of erosion: overall 61

Chapter 4
Figure 4.1. Media ecosystem (country-level) 68
Figure 4.2. Potential obstacles in the way of political information 71
Figure 4.3. Analytical group M1: temporal distribution of the steps of erosion 99
Figure 4.4. Analytical group M2: temporal distribution of the steps of erosion 99
Figure 4.5. Analytical group M3: temporal distribution of the steps of erosion 100
Figure 4.6. Analytical group M4: temporal distribution of the steps of erosion 101
Figure 4.7. Analytical group M5: temporal distribution of the steps of erosion 102
Figure 4.8. Analytical group M6: temporal distribution of the steps of erosion 103
Figure 4.9. Analytical group M7: temporal distribution of the steps of erosion 104
Figure 4.10. Analytical group M8: temporal distribution of the steps of erosion 105
Figure 4.11. Temporal distribution of the steps of erosion (Hungary) 107
Figure 4.12. Temporal distribution of the steps of erosion (Turkey) 112
Figure 4.13. Temporal distribution of the steps of erosion (Venezuela) 112
Figure 4.14. Temporal distribution of the steps of erosion (Ecuador) 113
Figure 4.15. Temporal distribution of the steps of erosion (Poland) 113
vii

Abstract

Democratic erosion is the slow, step-by-step subversion of democracy by elected leaders. It is the
greatest threat to contemporary democracies, including – as evidenced by Donald Trump’s
presidency in the United States – established ones. Interest in this insidious phenomenon has
grown exponentially in recent years, but research is still in its early stages. For the most part,
descriptions of the process lack depth and generalizability; and very few theories have been
proposed. The aim of this dissertation is to contribute to a more in-depth description of the
phenomenon and a better understanding of its onset. The research questions are: what exactly are
the building blocks – or steps – of democratic erosion; and what are some of the variables that
increase its likelihood?
To answer the first question, two original datasets are introduced. Both are comprehensive
lists of steps of democratic erosion – one based on a case study of Hungary; and the other one on
the comparative case study of Hungary, Poland, Turkey, Venezuela, and Ecuador, which focuses
on a single aspect of democratic erosion: the changes that affect the media. The presentation of
the hundreds of steps is structured around a small number of analytical groups derived from an
assessment of the Eroder’s likely targets and available tools. The categories are then used to
theorize about the relative timing of the different types of steps. To answer the second research
question, a theoretical framework is proposed to help structure thinking about the variables that
might lead to democratic erosion. A new way to operationalize backsliding is presented to define
the set of positive cases; and large-N analysis is used to test the most promising variable
suggested by the literature: polarization.
The data support the utility of the proposed analytical groups. While some groups show clear
temporal patterns in line with prior expectations, others do not. The tentative conclusion is that
democratic erosion is not a uniform process, but larger groups of cases might exist, pointing to
the possibility of distinct “varieties of erosion.” The regressions confirm that polarization plays a
role in the onset of democratic erosion, as well as being detrimental to the quality of democracy
in general. In addition, another variable of interest – trust in the legislature, a proxy for
satisfaction with the political system – receives empirical support.
The contributions of the dissertation include two original datasets that might be used for
future research and easily replicated in other countries. Furthermore, the dissertation makes a
number of conceptual additions to the literature: the idea that individual steps should be the unit
of analysis; the utility of all-encompassing analytical categories; the notion that different types of
steps may follow each other in a given sequence; and the consideration of the tools the
government has at its disposal at various points in time, and how that affects the timing of the
steps. These ideas provide a bridge between description and explanation.

1

Chapter 1
Introduction

The end may come about quickly, as a result of one ill-fated event; or slowly, following a
gradual, almost imperceptible decay. What is true for people and empires and much more can, as
scholars have recently realized, also be applied to democracy. While previously, the collapse of
democracies was usually a sudden, violent affair orchestrated by opponents of the ruling
government, in the past few decades, another threat has emerged: the slow, insidious subversion
of the democratic regime driven by the government itself. This phenomenon is called the erosion
of democracy, or democratic backsliding.
Already in 1988, Guillermo O’Donnell makes a distinction between the quick and the slow
death of democracy. Samuel Huntington (1996) goes as far as to pinpoint erosion as the main
threat to third-wave democracies. Interest in the phenomenon has exploded in recent years, as
more and more cases have been identified, often in connection with the rising tide of populism.
Worryingly, it became apparent that even established democracies may be vulnerable to erosion,
as European Union member states and hitherto poster boys of the successful political and
economic transition from communism, Hungary and Poland, have grown increasingly autocratic,
or “illiberal.” With the election of Donald Trump in the United States, interest in the
phenomenon has transcended academia. Even though Trump failed to win reelection, democratic
erosion is still an acute problem in many other countries, and thus the phenomenon needs to be
studied and understood.
2

Research on democratic erosion falls, broadly speaking, in two categories: some works (e.g.
Bermeo 2016) provide a general overview of the phenomenon, and possibly mention a case or
two to illustrate their points; while others (e.g. Corrales 2015) center on individual (or small
groups of) affected countries. Initially, the focus has been on identifying the phenomenon (e.g.
Schedler 1998); then on describing it (e.g. Erdmann 2011); whereas, more recently, a growing
number of attempts have been made to explain some of its aspects (e.g. Svolik 2020). As a result,
we now have a good understanding of what democratic erosion is, and many of its characteristics
have been documented. However, detailed accounts of what exactly happens during the process
in individual countries are still missing from the literature; and we have not yet identified the
changes that happen across all cases, beyond the very general (e.g. attacks on the media). In
other words, current descriptions of democratic erosion are neither deep, nor broad enough.
Conceptual, explanatory work is even less conclusive. The applicability of existing theories –
e.g. on democratization – has not been established (see Lust and Waldner 2015), and, for now,
there are only a few theories that address democratic erosion itself (e.g. Haggard and Kaufman
2021). More progress has been made in identifying variables that might play a role in the
process, though most of them are still in need of strong theoretical underpinnings as well as
empirical support. The most promising candidate is polarization.
With these gaps in the literature in mind, the aim of this dissertation is twofold. First, it
intends to contribute to a more granular description of democratic erosion both directly, by
presenting and evaluating in-depth case studies, and indirectly, by offering conceptual tools that
may facilitate future research. Its second purpose is to help identify the variables that increase a
country’s vulnerability to democratic erosion. It hopes to achieve that by proposing a theoretical
framework to guide thinking, and by testing the most promising variable.
3

The aim of political science is to understand political phenomena. Since many of these
phenomena are, by definition, central to the functioning of human societies, the significance of
this work often extends beyond scholarly interest. Accordingly, gaining a more in-depth
knowledge of democratic erosion would have practical implications. The more we know about
the process, the more likely we are to be able to prevent it; or, if it has already started, to
recognize and possibly reverse it. Due to the exploratory nature of most contributions so far, the
systematic approach required to proceed to the next level of understanding is missing from the
literature. This dissertation provides a structure others may build on; and datasets that either did
not exist, or not in this form, that may be used for a more thorough description of erosion and as
an empirical basis for future conceptual work.
The dissertation is organized as follows. The next chapter (Chapter 2) gives a review of the
current literature on democratic erosion, with an emphasis on its description, operationalization
and explanation, as well as the ways in which the process affects one particular actor: the media.
Three empirical chapters follow (Chapters 3 to 5), with the aim of advancing the research goals
outlined above. The concluding chapter (Chapter 6) offers a summary of the dissertation’s main
findings and considers avenues for future research.
Chapter 3. The goal of the first empirical chapter is to shed light on the building blocks of
democratic erosion, i.e. to provide a more or less comprehensive list of the steps employed by
governments to undermine various aspects of democracy. It intends to do so in a structured way,
by defining five analytical groups that supposedly contain most steps. The classification is based
on an examination of the components of the political process that may influence a government’s
chances of being reelected, and are therefore principal targets. The analytical categories are then
used to offer insight on the relative timing of the different types of steps.
4

The empirical part of the chapter consists of an in-depth case study of a single country:
Hungary. The archives of the leading independent online news site are searched from 2010,
Prime Minister Viktor Orbán’s return to power, and each event that might be construed as a part
of democratic erosion is added to a dataset. The data confirm the hypotheses suggested by the
theory. While there might be additional steps of erosion in other countries, the chapter advances
our knowledge of democratic erosion in several ways: so far, there have been no attempts to
create all-encompassing analytical groups; no consideration of the relative timing of the various
strategies; and no case study of such detail in the literature.
Chapter 4. The aim of the second empirical chapter is, again, to identify steps of erosion in
order to contribute to a more granular description of the process. However, this chapter sacrifices
some depth for greater generalizability: it includes five countries instead of one, but focuses on a
single aspect of democratic erosion – the changes that affect the media. As in the previous
chapter, the presentation of the steps is structured around a number of analytical categories. This
time, they are derived from a consideration of the flow of political information and the potential
obstacles a government might put in its way. The relative timing of the various types of steps is
then assessed, bearing in mind the tools available to the government at different points in time
following its election to carry out these steps.
The hypotheses are tested empirically by using a dataset assembled from the in-depth case
study of five countries known to be (or have been) affected by democratic erosion: Hungary,
Poland, Turkey, Venezuela, and Ecuador. Freedom House’s yearly Freedom of the Press reports
are searched for steps of erosion, which are compiled into a single list. The data confirm the
utility of the analytical categories and reveal clear temporal patterns, though only for some of the
groups. The overall conclusion is that democratic erosion is not a uniform process, but there
5

might be larger groups of cases, pointing to the existence of distinct “varieties of erosion.” This
possibility is only hinted at in the current literature. In addition, this chapter offers some of the
same contributions as the previous one – analytical groups, observations about timing, and a
novel dataset – that might serve future research.
Chapter 5. The goal of the third empirical chapter is to move beyond description, and
contribute to answering a key question concerning democratic erosion: what are the variables
that increase a country’s vulnerability to backsliding? The theoretical contribution of the chapter
is a framework in which variables of interest might be placed, designed to help structure thinking
about the early stages of the phenomenon.
The empirical part is intended to assess the most promising variable: polarization. Two
questions are considered: does greater polarization (1) decrease the quality of democracy in
general; and (2) increase the likelihood of democratic erosion? To answer the first question, a
dataset is built from Varieties of Democracy indicators and other sources, such as the World
Bank. Values are gathered for all countries in the world from 1989. For the second question, a
way to operationalize democratic erosion is proposed, yielding a set of positive cases. The
universe of observations is limited to election years in 28 European Union member states (still
including the United Kingdom) and 18 Latin American countries since 1989. Both hypotheses
are confirmed by the data, and thus so is the role of polarization – in fact, both issue and identity
polarization – in the onset of democratic erosion. In addition, the results draw attention to
another variable: trust in the legislature, a proxy for satisfaction with the political system – also
touched upon by the literature.

6

Table 1.1. Summary of Empirical Chapters
Chapter 3 Chapter 4 Chapter 5
Objective Compile complete set of
steps of democratic
erosion; test hypotheses
derived from theory
Compile complete set of
steps (limited to media);
test hypotheses derived
from theory
Test one variable
potentially responsible
for democratic erosion:
polarization
Cases Hungary Five positive cases (1) All countries; (2) 28
EU and 18 LA countries
Time frame 2010-2019 1999-2019 1989-2019
Theory 5 analytical categories;
sequence of steps
8 analytical categories;
sequence of steps
Framework to identify
variables of interest
Method In-depth case study Comparative case studies Large-N (regression)
Source Own compilation from
online news site
Freedom House Various, e.g. Varieties of
Democracy, World Bank

7

Chapter 2
Literature Review

The aim of this chapter is to provide a comprehensive overview of the academic literature on
democratic erosion, with a special emphasis on aspects of the phenomenon that feature in later
chapters. The areas of interest are: the description of democratic erosion; its operationalization;
its effect on media freedom; and attempts to explain the phenomenon. The latter part will be
limited to variables, mechanisms, and theories that have been proposed with democratic erosion
itself in mind. While the broader literature on democratic stability and democratic breakdowns
may provide insight into the phenomenon, it is too vast to tackle here, and its applicability to
democratic erosion has not been established yet.

2.1. Description
The possibility that democracy may decay over time instead of falling overnight has been
considered for at least three decades. In his analysis of Brazil, O’Donnell (1988) makes a
distinction between the “quick death” of a conventional military coup and the “slow death” of a
gradual erosion of democratic practice (281). Schedler (1998), too, likens the erosion of basic
democratic features to slow death (94). Once the phenomenon has been identified, scholars soon
recognized the danger it poses to democracy. Huntington (1996) goes as far as to claim that
erosion, defined as “the intermittent or gradual weakening of democracy by those elected to lead
it” (8), is the main threat to third-wave democracies. Democratic erosion has remained a cause
for concern in the ensuing two decades. In her analysis of backsliding, Bermeo (2016) concludes
8

that classic, overt cases (open-ended coups d’état; executive coups, or autogolpes; and election-
day vote fraud) are becoming less common, but the more insidious forms (promissory coups, or
temporary military interventions to “restore democracy”; executive aggrandizement; and
strategic harassment and manipulation) are gaining prevalence. Similarly, Ginsburg and Huq
(2018b) assert that “slow decay” is now the “most common species of democratic recession”
(43). Overall, though, as Waldner and Lust (2018) note, mainstream academic interest in
democratic erosion is still relatively new.
While an exact definition – that is, a concise description – of democratic erosion is yet to
emerge, several defining elements of the phenomenon recur in the literature, with a few notable
differences. The key points a description has to cover are: who drives the process; what the
changes are that, together, make up democratic erosion; how they happen; and what the outcome
is. There appears to be little ambiguity with respect to the first of these points. Initially,
O’Donnell (1988) equates the slow death of democracy with the “gradual constriction of civilian
authority” by the military (287). In later articles, however, the state (Bermeo 2016) or, more
precisely, democratically elected executives (Huntington 1996; Levitsky and Way 2002;
Diamond 2015, 2021; Bermeo 2016; Przeworski 2019; Svolik 2020; Haggard and Kaufman
2021) are unequivocally identified as the driving force behind democratic erosion.
As for what the changes of interest are, the answers range from the most general, i.e. a decline
in the quality of democracy (Lust and Waldner 2015), to the very specific, explored in individual
case studies. Most authors opt for some degree of generalization, offering analytical categories,
defined by either the temporal sequence of the changes, or their objects (targets). For example,
Coppedge (2005) identifies two consecutive stages of democratic erosion in his case study of
Venezuela: (1) the removal of political actors in a position to check the president; and (2) the
9

appointment of new, loyal actors. Lust and Waldner (2015) conceptualize the process as a series
of changes that undermine three key aspects of democracy: competitive electoral procedures,
civil and political liberties, and accountability. For Ginsburg and Huq (2018b), the three areas
under threat are competitive elections, liberal rights to speech and association, and the rule of
law. According to them, democratic erosion unfolds as a combination of five mechanisms: (1)
constitutional amendments; (2) the elimination of horizontal checks; (3) the centralization and
politicization of the bureaucracy; (4) the contraction of a shared public sphere; and (5) the de
facto suppression of political competition, and thus the prospect of rotation out of elected office
(72-73).
Overall, there seems to be a consensus on the two main types of changes that make up the
process of backsliding. The first type involves attacks on actors within the formal political/legal
system. Examples include the concentration of power by the executive; the erosion of checks and
balances; and the weakening of institutions of accountability (Diamond 2015; Lust and Waldner
2015; Bermeo 2016; Coppedge 2017; Waldner and Lust 2018). The second type targets actors
outside of the political/legal system. Civil liberties, such as the freedom of the press and the
freedom of assembly, may be eroded, leading to a reduction of political pluralism. (O’Donnell
1995; Diamond 2015; Lust and Waldner 2015; Mainwaring and Pérez-Liñán 2015; Coppedge
2017; Lueders and Lust 2018).
Some authors have additional items on the list of changes. For example, the disrespect of
established norms by the executive has received growing scholarly attention in recent years. Lust
and Waldner (2015) emphasize that the “informal procedures that shape elections, rights, and
accountability” may also be negatively affected by backsliding (7). Przeworski (2019) goes as far
as to include the erosion of norms in his definition of backsliding (172). Similarly, the
10

weakening of certain long-standing norms in the politics of the United States has a central place
in Levitsky and Ziblatt’s (2018) argument. Diamond (2015) points to another feature of
democratic erosion: the accumulation of wealth by the leaders “for themselves and their families,
cronies, clients, and parties” (149).
There seems to be an agreement among scholars of erosion on some characteristics of the
process, i.e. how it happens. First, adjectives like slow, gradual, incremental, discontinuous, etc.
are used to describe how it takes places in small steps, over a relatively long period of time
(O’Donnell 1988; O’Donnell 1995; Huntington 1996; Schedler 1998; Levitsky and Way 2002;
Erdmann 2011; Diamond 2015, 2021; Lust and Waldner 2015; Bermeo 2016; Levitsky and
Ziblatt 2018; Waldner and Lust 2018; Svolik 2020; Haggard and Kaufman 2021). It is precisely
this feature of the process that led scholars to compare it to erosion. It is also what makes the
study of these small steps of interest to those who want a more in-depth understanding of the
phenomenon. Another, though less widely mentioned characteristic of backsliding is that it is
often achieved through legal channels (constitutional assemblies, legislatures, etc.) which
provides a degree of domestic and international legitimacy to even otherwise controversial
changes (O’Donnell 1995; Weyland 2013; Bermeo 2016). As Svolik (2020) points out,
vulnerabilities within the democratic process are exploited to subvert democracy. In Gandhi’s
(2019) words, executives are “following the letter of the law to effectively change the spirit of
the law” (14).
Finally, as a result of the previous two features, democratic erosion is an insidious process.
The effect of individual steps on democratic quality may be ambiguous (i.e. easy for the
executive to trivialize), and the steps are, on their own, usually not dramatic enough to elicit a
strong response (O’Donnell 1995; Schedler 1998; Erdmann 2011; Bermeo 2016). The apparent
11

legality of the process allows the executive to go as far as to claim that it is in fact acting in the
interest of democracy (Przeworski 2019). Ginsburg and Huq (2018b) emphasize that each of the
changes may be defensible in isolation; it is their cumulative effect that threatens democracy.
Similarly, Scheppele (2013) warns that “perfectly legal and reasonable constitutional
components” may be “stitched together to create” a “Frankenstate” (560).
As pointed out earlier, O’Donnell (1988) initially put the “slow death” of democracy in the
context of the struggle between civilian leaders and the military. Accordingly, the outcome of
democratic erosion was assumed to be a “highly repressive, military-based power” (287). Now,
however, scholar tend to describe the result of the process as a hybrid regime, falling somewhere
in the gray zone between functioning democracies and overt autocracies (Schedler 1998;
Levitsky and Way 2002; Bermeo 2016). Crucially, elections are still held, and they are
reasonably free (Lust and Waldner 2015; Mainwaring and Pérez-Liñán 2015; Bermeo 2016;
Waldner and Lust 2018). However, the fairness of elections is undermined. The executive tilts
the electoral playing field in its own favor by changing electoral rules, using state funds for its
campaign, etc. (Svolik 2020). According to Varol (2015), the systemic advantages the executive
enjoys lead to a situation in which “partisan alternation might not occur even in the face of
changing political preferences by the electorate.” (1684)

2.2. Operationalization
As democratic erosion becomes an object of scholarly scrutiny, a growing demand for more in-
depth quantitative research is to be expected. Accordingly, there have been efforts to
operationalize the phenomenon; yet, no single method has emerged for now. This is hardly
surprising. Any method presupposes several choices, such as: what measure(s) of democratic
12

quality will be used; what will be the lower limit for the starting point (if any); how much the
value of the chosen measure may decrease over what period of time; etc. Most of these questions
have no ready answer in the present body of knowledge on democratic erosion, and thus
individual choices may very well differ.
For example, Burke and Leigh (2010) use Polity IV scores; Erdmann (2011), Mainwaring and
Pérez-Liñán (2015), and Svolik (2019) look to Freedom House; while Coppedge (2017)
Kaufman and Haggard (2019), Lührmann and Lindberg (2019), and Haggard and Kaufman
(2021) rely on V-Dem data to define the scope conditions of erosion. Ginsburg and Huq (2018b)
consider all three sources. Waldner and Lust (2018) imply that several measures should be used
together: one each for “competition, participation, and accountability” (95). The choice of
measure matters. As Lueders and Lust (2018) show, “the average pairwise correlation coefficient
between measures of regime type is between 0.85 and 0.86, while correlations among the
indicators of regime change are disturbingly low: between 0.33 (backsliding) and 0.46
(transition)” (738). Of course, different ways to operationalize democratic erosion lead to
different sets of cases, though notable examples – such as Hungary, Venezuela, and Turkey –
feature in most of them.

2.3. Democratic Erosion and the Media
Since the analysis of the media in eroding democracies constitutes an intersection between
political science and media/communication studies, it would not be surprising if the latter, with
its rich history, had already provided ample insight into the process. However, that is not the
case. Of course, the connection between different regimes and the freedom of the media has been
explored. A lot has been written about the media in democracies, especially the United States
13

and Western Europe, though efforts have been made to widen the scope of the research. Hallin
and Mancini (2004), for example, describe three media systems based on factors like the
structure of media markets: the polarized pluralist model, the democratic corporatist model and
the liberal model. Later (2011), they extend their analysis beyond the Western world, including
regions such as Africa. Other authors also focus on the media in democratic regimes – e.g.
Plattner (2012) discusses the impact of technological progress and the resulting “new media” on
politics.
Similarly, there is no shortage of research on the problems the media have to face in non-
democratic countries. This scholarship tends to focus on the ways in which media outlets are
suppressed, and are thus unable to function as institutions of accountability. Naturally, just like
non-democratic regimes differ from each other, so do the levels of media freedom across these
countries. For example, as Egorov et al. (2009) observe, resource-poor dictators allow freer
media so as to incentivize bureaucrats, and thereby improve the quality of government.
While work on democratic media systems has gone beyond individual cases, looking to
generalize and build theories, the analysis of the media in non-democratic regimes is usually
restricted to single countries. Unsurprisingly, the state of the media in China, the world’s leading
non-democratic country, has inspired considerable literature (e.g. Tang and Iyengar 2011). Some
of these works (e.g. Stockmann 2012; He 2008) go beyond mere description, and attempt to
establish how recent changes, such as market liberalization and technological progress, might
influence the freedom of the press. The findings are typically less optimistic than one might
expect. Other non-democratic countries and regions are also studied – for example Singapore and
Malaysia (Rodan 2004) and Central Asia (Kenny and Gross 2008).
14

These accounts do point out some changes in the media system, but they are mostly static, and
no connection is proposed between the evolution of the media and regime changes. In fact, there
is little general knowledge about what happens to the media during transitions, either to or from
democracy. The expectation, of course, is that democratization will lead to – or is at least
associated with – relatively free media, while a breakdown will have the opposite effect; yet it is
unclear how the transition itself takes place. Similarly, it can only be assumed that democratic
erosion is characterized by a gradual decrease in the level of media freedom, but not much is
known about the process. In the absence of theories that can be applied to democratic erosion, a
possible strategy is, again, to assess what relevant case studies have to say about the media.
No account of democratic erosion can be complete without at least a brief reference to the
measures that target the media. Accordingly, most case studies listed in the following section
include, in one way or another, the media. Moreover, there is an increasing number of studies
that focus on the media themselves in the affected countries: in Russia (e.g. Arutunyan 2009;
Duffy 2015; Lipman and McFaul 2001; Mickiewicz 2008; Pallin 2017; Robertson and Greene
2017; Shevtsova 2015; Tolz and Teper 2018; Vartanova 2011); Turkey (e.g. Akser 2018; Corke
et al. 2014; Esen and Gumuscu 2016; Öniş 2016; Över 2021; Yesil 2016; Yıldırım, Baruh, and
Çarkoğlu 2020); Venezuela (e.g. Corrales 2015; Corrales and Penfold-Becerra 2007; Hidalgo
2009); Ecuador (e.g. Bermeo 2016; Conaghan 2008; Conaghan 2016; de la Torre 2013); Bolivia
(e.g. Anria 2016; Lehoucq 2008); Hungary (Bajomi-Lázár 2013, 2017; Bátorfy and Urbán 2020;
Griffen 2020; Kerpel 2015; Surowiec and Štětka 2020); Poland (e.g. Kerpel 2015; Surowiec and
Štětka 2020 Surowiec, Kania-Lundholm, and Winiarska-Brodowska 2020); etc.
Even though these works typically scrutinize individual (or a small group of) countries, and
vary considerably in their focus and depth, taken together, they point to a number of common
15

patterns that characterize democratic erosion. Examples include: complete control over the
public media, reduced to little more than a mouthpiece for the government (e.g. Bajomi-Lázár
2013); distortion of the media market by channeling advertising money to friendly outlets while
withholding state funds from critical media (e.g. Bátorfy and Urbán 2020); acquisition of media
outlets by business circles close to the government (e.g. Yesil 2016); etc. Like democratic
erosion itself, the changes that affect the media tend to be subtle and gradual, though not without
notable exceptions (e.g. imprisonment of journalists in Turkey; see Corke et al. 2014).
While most of these case studies do not attempt to go beyond description, some of them offer
a more structured approach. For example, Bajomi-Lázár (2013) introduces the concept of “party
colonization” (69) as a framework to explain the measures the Orbán government has deployed
against the public media and domestic regulatory authorities. The crux of his argument is that the
reason for taking control of the media is not just to silence criticism: it also serves to redistribute
valuable resources – airtime, frequencies, positions and money – to party loyalists. Other authors
(e.g. Akser 2018; de la Torre and Ortiz Lemos 2016; Dragomir 2018; Griffen 2020) group
individual episodes of democratic erosion into broader categories, which is a necessary first step
toward a more general understanding of the process. Dragomir (2018), for instance, identifies
four types of financial strategies that authorities employ to dominate the media sector: (1) public
funding for state-administered media; (2) state advertising; (3) state subsidies; and (4) market-
disruption measures (1132). These strategies are especially effective due to financial constraints
many media outlets grapple with.
In addition to works on democratic erosion itself, one might conceivably gain insight into the
process – and its impact on the media – from research that centers on populism (e.g. Aalberg et
al. 2016; Alvares and Dahlgren 2016; Vachudova 2020). In fact, most of the leaders at the helm
16

of eroding democracies (Chávez, Erdoğan, President Jair Bolsonaro of Brazil, etc.) have been
called populists. According to Aalberg et al. (2016), populism is closely connected to democratic
illiberalism, since it considers “the people” to be always right, and thus any restrictions on
majority rule unjustified; and it rejects intermediaries and institutions (18). Vachudova (2020)
goes as far as to claim that ethnopopulism is a successful strategy “for winning votes and
concentrating power” (318). Yet, there are limits to the utility of this strand of literature. The
connection between populism and democratic erosion has no conceptual underpinnings. Leaders
considered populist have either not attempted to undermine democracy (e.g. British Prime
Minister Boris Johnson), or were not successful in their (arguable) efforts (e.g. U.S. President
Donald Trump). More importantly, most populists have never been in a position to do real harm
to democracy, and thus research on them cannot extend to the process of democratic erosion.
All things considered, there is a growing literature on the media in eroding democracies, and
key parts of the phenomenon have been described in detail. However, the results so far are
neither comprehensive nor systematic enough to provide a complete picture of the process that
cuts across boundaries. As such, significant gaps remain, which need to be addressed.
Social media. An overview of the literature would not be complete without a discussion of the
role the social media might play in, and the ways it might be affected by, democratic erosion.
Scholars quickly recognized the positive impact that information technology – and social media
in particular – might have on democracy. However, initial optimism was soon tempered by the
realization that enemies of democracy might also make good use of the new tools. Diamond
(2010) emphasizes this duality: on the one hand, “liberation technology” (70) allows for greater
access to information and more meaningful participation in politics; on the other hand,
authoritarian states can more efficiently control the Internet and punish dissenters. Shirky (2011)
17

also warns that the use of social media “does not have a single preordained outcome,” its impact
being dependent on the strength of the civil society. Worryingly, it is often populist actors who
benefit from social media (e.g. Engesser et al. 2017; Gerbaudo 2018). Zhuravskaya, Petrova, and
Enikolopov (2020) conclude that “the political roles of the Internet and social media are not yet
fully understood” (433).
When it comes to democratic erosion, two questions seem especially relevant. First, can social
media keep functioning and remain free from manipulation once traditional media have been
taken over by the Eroder? Second, can social media assume, at least in part, the role of traditional
media in such a situation? Based on the available literature, neither answer is particularly
reassuring. Malevolent actors have a growing arsenal of tools they can deploy to disrupt social
media: “bots” that generate automated messages (e.g. Forelle et al. 2015); “trolls” who derail
online discussions (e.g. Aro 2016; Saka 2018); misinformation (e.g. Weeks and Gil de Zúñiga
2021); etc. Even censorship has become more sophisticated, making use of targeted viruses,
strategically timed denial-of-service attacks, etc. (Deibert et al. 2010). According to Gunitsky
(2015), social media can contribute to autocratic resilience in four ways: through counter-
mobilization; discourse framing; preference divulgence; and elite coordination (42). He warns
that autocratic adaptation to the new technologies is only likely to accelerate, as regimes learn
from their own mistakes and each other (50). Most of the aforementioned tools are available not
just to autocrats, but also to Eroders, too.
Furthermore, social media is unlikely to be able to replace traditional media if the latter’s
independence has been undermined. As Walker and Orttung (2014) note, social media are in the
“insurgency phase” of their development (73), and thus television still has no real competitors.
Chadwick (2017) places the emphasis on the “interactions among older and newer media logics”
18

(4), calling the media system in the United States and Britain a “hybrid.” Similarly, Postill
(2018) cautions against treating social media as a “realm apart from the rest of the media
environment” (761).
It also needs to be noted that some of the success stories associated with social media (e.g.
Said-Hung and Segado-Boj 2018) have more to do with their potential to mobilize, and not to
transmit political information the way traditional media do. Interestingly, even the former is open
to some debate: in their case study of the Arab Spring, Wolfsfeld, Segev, and Sheafer (2013) find
that the use of the new media tended to follow, and not precede, protest activity.
It seems, therefore, that while social media have their uses, they alone are unlikely to save the
freedom of the press once traditional media have been subjugated by the government.

2.4. Explanation
Several individual variables have been proposed to play some role in the onset of democratic
erosion, or in its eventual success. In his quantitative analysis of incumbent takeovers, Svolik
(2015) concludes that a presidential system and fuel exports increase the risk of takeovers, while
greater GDP per capita, GDP growth and, surprisingly, a military past decrease it. Maeda (2010)
also points to the relative vulnerability of presidential systems. Gibler and Randazzo (2011) find
that an independent judiciary can prevent regime change toward authoritarianism, though it
needs to be well established before its beneficial effects can manifest. The issue with these
results is that the cases on which they are based are not limited to democratic erosion. For
example, Svolik includes Hitler’s power grab in 1933; whereas Maeda looks, among others, at
the suspension of the democratic process in Peru by Fujimori.
19

Other works focus on democratic erosion itself, but are less conclusive. Erdmann (2011) links
democracy’s proneness to erode to its age and economic well-being, but concludes that
democratic quality can decline in young and old democracies as well as poor and rich ones.
Ginsburg and Huq (2018b) distinguish two ways in which democratic erosion can take place, and
note that one of them (partisan degradation) has a simple prerequisite: one side needs to win
convincingly. Kaufman and Haggard (2019) also emphasize the importance of majority and
supermajority control of the parliament by the executive’s party, which in effect eliminates the
legislature as a major source of oversight. Similarly, Svolik (2019) calls the legislature’s
“complicity” an essential part of the process.
Diffusion – the idea that the appetite for and the “best practices” of democratic erosion spread
internationally – has been considered as a potentially significant variable. However, while de la
Torre (2017) acknowledges that the Venezuelan example might have played a role in Bolivia and
Ecuador, he claims that diffusion was not the main cause of democratic erosion in the latter
countries. Buzogány (2017) reaches the same conclusion when assessing whether Hungary was
affected by authoritarian diffusion from Russia. Akyuz and Hess (2018) – building on the work
of Levitsky and Way (2010) – identify another way in which international factors might play a
role. They note that in Turkey, the onset of democratic erosion coincided with the rising
influence of nondemocratic states, such as China, which reduced the – real and perceived – costs
of backsliding for Erdoğan.
Others point to ideological factors, such as “counterrevolutions against liberalism” (Krastev
and Holmes 2018) or “the confusion or the collusion of political liberalism with economic
liberalism” (Rupnik 2018). In a similar vein, Vachudova (2020) argues that “ethnopopulist”
parties – such as Viktor Orbán’s Fidesz – took advantage of the unpopularity of neoliberal
20

economic policies by purposefully conflating neoliberalism with liberal democracy (326).
However, these theories are built on limited case studies of Hungary and Poland, and fail to
address why the erosion of democracy is less of an issue in other – typically less successful –
post-communist countries of the region. Also, the applicability of these theories to cases beyond
Central and Eastern Europe is questionable.
While our understanding of democratic erosion is still lacking, there are two potential
explanatory variables that show considerable promise. The first is the crisis of the political
system. De la Torre (2017) notes that Hugo Chávez’s rise to power was preceded by years of
partisan decay and political instability in Venezuela. Bolivia and Ecuador also experienced
“crises of political parties and the institutional framework of democracy” (1271) before the
Venezuelan model could be emulated. Foa and Mounk (2017) use survey data – including
evidence of consistently low levels of confidence in politicians and political institutions – to
reach the same conclusion for both Venezuela and Poland. Based on their study of Latin
American countries, Pérez-Liñán, Schmidt, and Vairo (2019) conclude that one institutional
factor in particular proved to be detrimental to democratic stability: the “concentration of
institutional power by the executive” (611), or presidential hegemony. Coppedge (2005) shows
that political decline in Venezuela is consistent with economic explanations, but the rise and
decay of “partyarchy” is also an important part of the story. Gauna (2017) takes Coppedge’s
model one step further, and lays the blame for the eventual challenge to democracy on economic
presidentialism – the “highly autonomous executive control of a state-controlled economy” (33)
– which deprived the political system of its ability to adapt.
The latter contributions suggest that the crisis of the political system is connected to issues
with the economy. The causal relationship seems to run both ways. Ginsburg and Huq (2018b)
21

observe the synergy of the two types of crises, noting that the economic difficulties due to a
decade-long decline in the Venezuelan oil industry coincided with a collapse in the country’s
party system in the 1990s. Hunter and Power (2019) go as far as to identify a “perfect storm” of
four simultaneous crises in Brazil, including a political crisis of rising polarization and falling
trust in established parties, and a serious downturn in the economy after 2013. The subsequent
disillusionment with establishment parties, blamed for the crises, led to the rise of Jair Bolsonaro
in 2018. Erdmann (2011) concludes that scholars should consider how an economic crisis can
turn into a political crisis, which will then endanger democracy. While the interaction of these
two is more complex, they almost certainly play some role in the onset of democratic erosion.
The other promising explanatory variable is polarization. Its adverse effect on democracy has
been recognized decades ago, with scholars like Dahl (1971) observing that polarization leads to
crisis in competitive (i.e. democratic) systems. As Mason (2015) explains, (social) polarization is
not necessarily rooted in conflicting issue positions; it is instead driven by “partisan identity and
political identity alignment” (128). There is an emotional and social connection to the party,
whose supporters behave “more like a sports fan than like a banker choosing an investment”
(129). They remain fiercely loyal and protective even if they disagree with some of the party’s
issue positions.
Polarization has, unsurprisingly, resurfaced in the context of democratic erosion, though the
mechanism through which it leads to a decline in the quality of democracy is still debated.
According to Kaufman and Haggard (2019), polarization affects the elite (weakens norms of
tolerance and self-restraint) as well as the electorate (generates support for illiberal actors). In a
later work (Haggard and Kaufman 2021), they identify three interrelated ways in which
polarization sets the stage for democratic erosion: it reduces appetite for policy compromises
22

within the government, resulting in disaffection with democracy; it magnifies the appeal of anti-
system political actors; and it increases tolerance for derogations from democracy (14).
Levitsky and Ziblatt (2018) attribute the weakening of democratic norms to “extreme partisan
polarization” – a sort of existential conflict. They warn that a respect for norms is necessary for
constitutional checks and balances to effectively defend democracy. Svolik (2019, 2020) argues
that a prerequisite of democratic erosion is people’s willingness to trade off democratic
principles for partisan interests, which increases with polarization. Similarly, Somer and McCoy
(2018) point out that polarization can undermine democracy in a number of ways, but they – like
Diamond (2021) – add that causality may flow in both directions. They are the only ones to note
that polarization may have potential benefits to democracy, such as greater engagement in
politics or the increased cohesion of pro-democracy actors.
McCoy, Rahman, and Somer (2018) go one step further, and outline a causal chain linking
polarization to democratic erosion (26). First, a leader or a movement mobilizes a new or
previously disunited group by exploiting grievances and employing divisive rhetoric (i.e. “us
versus them”). This leads to mass negative partisanship and affective polarization, enhancing
zero-sum perceptions. Due to a reduced willingness to cooperate with a group increasingly seen
as an existential threat, conflict becomes more attractive than cooperation. This, in turn, results in
a greater tolerance of illiberalism and democratic erosion.
Another strand of research focuses on countries instead of individual variables. There is a
growing number of case studies of varying depth that look at states that are, or have recently
been, affected by democratic erosion: Russia (e.g. Lanskoy and Myles-Primakoff 2018; Lanskoy
and Suthers 2013; Lust and Waldner 2015; Robertson and Greene 2017; Sestanovich 2007;
Shevtsova 2015; Smyth and Soboleva 2014); Turkey (e.g. Akyuz and Hess 2018; Bermeo 2016;
23

Diamond 2015; Esen and Gumuscu 2016; Müftüler-Baç and Keyman 2012; Öniş 2016; Turam
2012); Venezuela (e.g. Corrales 2015; Corrales and Penfold-Becerra 2007; Hidalgo 2009;
Seawright 2012); Bolivia (e.g. Anria 2016; Lehoucq 2008); Ecuador (e.g. Bermeo 2016;
Conaghan 2008, 2016, 2017; de la Torre 2013; de la Torre and Ortiz Lemos 2016); Hungary (e.g.
Bakke and Sitter 2020; Bánkuti, Halmai, and Scheppele 2012; Grzymala-Busse 2019; Przybylski
2018); and Poland (e.g. Bakke and Sitter 2020; Grzymala-Busse 2019; Przybylski 2018). More
recently, (potential) cases like India (e.g. Heller 2020; Mukherji 2020); Brazil (e.g. Daly 2020;
Heller 2020); Israel (Rogenhofer and Panievsky 2020); Croatia (Čepo 2020); Serbia (Castaldo
2020); Bangladesh (Lorch 2021; Riaz 2021); Thailand and the Philippines (Lorch 2021); etc.
have also received their share of attention.
However, in spite of growing scholarly interest in democratic erosion, there are virtually no
theories that address the phenomenon in its entirety. Even middle-level generalizations, like the
ones Linz (1978) provides for classic breakdowns, are missing. This is the conclusion that Lust
and Waldner (2015) reach in their thorough assessment of the current state of knowledge. They
consider the applicability of a number of hypotheses generated by scholarship on regime change
to the study of backsliding (e.g. Acemoglu and Robinson 2006; Boix 2003; Fish 2002; Levitsky
and Way 2002, 2010; Linz 1990; Lust and Waldner 2013; Mainwaring 1993; Mazzuca 2013;
Powell 1982; Putnam 1994; Reynolds 2011; Seawright 2012; van de Walle 2003). The caveat is
that these hypotheses were not conceived with erosion in mind. While some of them are
undoubtedly relevant to the issue of democratic erosion, and point to a number of promising
avenues for future research, none of them intends to tell the whole story, and none of them has
been conclusively proven.
24

There are, of course, a number of studies that attempt to draw more general conclusions. It is
perhaps Mazzuca’s (2013) analysis that falls closest to a theory of democratic erosion. He
attributes the Latin American cases of backsliding to “rentier-populist coalitions” (110), or
alliances between the government and the informal sector. The former redistributes income
derived from mineral or agricultural sources to the latter in exchange for political support. There
are two keys to resisting rentier populism: the state of the political parties, and that of the capital
markets. Would-be autocrats only succeed where these institutions are in a crisis.
Haggard and Kaufman (2021) also offer a more holistic explanation of democratic erosion.
Drawing on in-depth case studies, they assert that three interrelated causal mechanisms are at
play: “the pernicious effects of polarization; realignments of party systems that enable elected
autocrats to gain legislative power; and the incremental nature of derogations, which divides
oppositions and keeps them off balance.” (iii) Democratic erosion is “the outcome of a complex
causal chain” (2): anti-system parties gain a significant vote share as a result of polarization,
which gives them control of the executive, and then the legislature. The latter ultimately leads to
the collapse of the separation of powers (5).
A potential new direction of research is offered by Ginsburg and Huq (2018a), who point out
that work on democratic erosion tends to be characterized by selection bias. Instead, they
propose to study “near misses,” or “cases in which a democracy is exposed to social, political, or
economic forces that could catalyze backsliding, yet somehow overcomes those forces and
regains its footing.” (17) Examples include Finland between the two world wars; Colombia in
2010; and Sri Lanka in 2015.
Overall, though, as Waldner and Lust again conclude in 2018, “no crystallized, coherent
literature evaluates rival hypotheses about a relatively fixed set of cases” (97) and therefore
25

“there is no readily available set of theories that we as a community can uncontroversially adopt,
adapt, and apply to the problem of backsliding.” (106) In the same vein, Luo and Przeworski
(2019) concede that “as of now we know little empirically about the vulnerability of particular
institutional arrangements to backsliding.” (29)

2.5. Conclusion
Literature on democratic erosion has seen dynamic growth in recent years. Budding interest has
created what can be described as a positive feedback loop: the more we know about the
phenomenon, the more cases have been proposed, which in turn has attracted even greater
scholarly attention, as well as a broader recognition of this new threat to democracy.
Description. While democratic erosion has no universally accepted definition, there is general
agreement on what the fundamental features of the phenomenon are. It is driven by
democratically elected executives, who subvert institutions of accountability, concentrate power,
and undermine civil liberties. Their ultimate aim is to tilt the electoral playing field to such an
extent that their reelection is all but assured. The process is slow and gradual (hence “erosion”),
consisting of individual steps that are often legal and seemingly harmless on their own, and thus
difficult to identify, criticize and reverse. Some disagreements remain, mostly on how
democratic a country has to be at the start and at the end of the process to be considered a
positive case.
Operationalization. Accordingly, there is no consensus yet on how to operationalize the
phenomenon. Different authors use different measures of democratic quality and define different
algorithms to tell apart positive and negative cases. As a result, there is no broadly accepted set
of cases. Since there is little debate about the features of democratic erosion (see above), the
26

various sets overlap significantly, typically including well-known cases such as Venezuela,
Turkey, and Hungary. However, the differences matter when it comes to borderline cases like
Israel or, more recently, the United States.
Media. Even though media/communication studies offer few insights into democratic erosion,
and research on populism has limited applicability to the process itself, there is no shortage of
works that touch upon the media in eroding democracies. Most country studies devote at least
some space to the changes that affect the media, and there is also an increasing number of works
that focus exclusively on the plight of the fourth estate. Consequently, much is known about the
strategies employed to undermine the freedom of the press. Governments take control of the
public media and use them as tools of propaganda; distort the media market by selectively
disbursing state funds and advertising money, while applying pressure on private advertisers to
follow suit; acquire valuable media outlets through friendly business groups and turn them into
tools of propaganda; etc.
The existing literature has two shortcomings. First, very few attempts have been made to
provide a comprehensive overview of all the measures that target the media in eroding
democracies. Some case studies are more detailed than others, but these works never go beyond
individual countries, and are thus unable to offer general conclusions. Second, the changes that
affect the media are rarely presented in a systematic way, in relation to a theoretical framework
and each other. For now, conceptual work is still in its early stages.
Explanation. Being a complex phenomenon, several aspects of democratic erosion warrant
attention. Understanding what variables increase a country’s vulnerability to backsliding is one
of the key questions, with both theoretical and practical significance. Since research on
democratic erosion has only recently gathered pace, most of it is descriptive; yet a number of
27

variables have already been proposed to play a role in the onset of the process. The most notable
examples are: majority control of the legislature; international factors, such as diffusion and the
relative leverage of democratic and non-democratic powers; ideological variables, like backlash
against neoliberal economic policies, etc.
Two variables recur in the literature more often than others. One of them is the – broadly
defined – crisis of the political system, which supposedly leads to democratic erosion either own
its own or in tandem with another type of crisis – mostly economic. The causal mechanism at
play is not yet clear, in spite of a number of contributions. The most promising variable is
polarization, the subject of several works of varying depth and methodological approach. Indeed,
available evidence seems to suggest that polarization might be a part of the story, even though
many questions about its role remain unanswered. For example, does it increase vulnerability to
erosion, or act as a catalyst for the ongoing process (or both)? What is the mechanism behind its
corrosive effect? Does polarization matter on the elite level, or within the whole population?
Even the type of polarization at play needs to be determined.
Considerable progress notwithstanding, there is still no consensus on what variables might be
important either before or during democratic erosion. More complex explanatory theories have
not been proposed yet, with a few notable exceptions.

28

Chapter 3
The Building Blocks of Democratic Erosion: An In-depth case Study of
Hungary

3.1. Introduction
Democratic backsliding, or erosion, is a gradual, step-by-step decline in democratic quality
driven by governments that originally rose to power through free and fair elections. While
institutions of accountability – and elections themselves – remain in place, they are subverted by
the executive, undermining their ability to effectively check its power. Democratic erosion is a
global phenomenon, with diverse cases such as Venezuela (from 1999), Turkey (from 2003), or
Hungary (from 2010). The realization that backsliding is a real threat and may affect even
countries assumed to be safely democratic, like EU members Hungary and Poland, resulted in
growing academic interest.
However, even though the phenomenon has been recognized decades ago, research is still in
its early phase. Erosion has no commonly accepted definition, and scholars are yet to reach a
consensus on how to operationalize it. Part of the issue is that some measures of democracy may
not be able to capture the small, insidious changes that characterize the process. As long as this
poses a problem, efforts to establish the population of cases may also be inconclusive.
Unsurprisingly, there are no credible theories that would explain the causes of erosion, i.e.
whether a country’s vulnerability is increased by structural factors.
It is safe to conclude that we have very limited knowledge about democratic erosion. A root
cause of this problem is that we do not even know what erosion entails, i.e. what exactly happens
29

in an eroding democracy. If backsliding is indeed a step-by-step decline in democratic quality,
identifying and making sense of these steps is paramount. It is not to say that no such steps are
mentioned in the literature. Key differences between democratic and non-democratic regimes are
well known, thus some changes can be expected for a country in transition. As such, authors
focusing on specific cases may mention how the independence of the courts or media freedom is
undermined. However, these are usually no more than examples meant to illustrate the author’s
general point. No effort has been made so far to provide a comprehensive list of the steps that
make up democratic erosion, and the steps themselves have never come under academic scrutiny.
It is not hard to see how having a clearer picture of these steps would further research in all
three areas mentioned above. A definition of democratic erosion needs to include a description of
the set of steps that make up the process (leaving room for individual variations). Clearly, the
closer we are to knowing the full set, the more accurate the description, and thus the definition, is
likely to be. As for operationalization, the main questions are what measure(s) to use, and what
sort of change to look for. If we find that, for example, the courts and the media are affected by
the process, we need a measure or a set of measures that capture both. Similarly, attempts at
operationalization may work with the assumption that the changes occur within a certain time
frame, which can only be established by knowing how long individual steps take, how much
time elapses between them, etc. For potential theories focusing on the process of democratic
erosion, it is of obvious importance to find the building blocks of this process. Structural theories
would also benefit from having a more detailed picture of the dependent variable – besides
identifying what exactly it is. For example, if research uncovers that subverting the
Constitutional/Supreme Court is a prerequisite of further backsliding, it may be worth
30

ascertaining what structural variables made the Constitutional Court vulnerable to subversion in
the first place.
Finally, a familiarity with the steps that make up erosion may have value beyond academic
research: it may also help to recognize the phenomenon in time. This is especially important
considering the insidious nature of backsliding, with individual steps rarely being egregious or
particularly difficult to justify, and executives often claiming that they in fact follow popular
will, thereby strengthening, not weakening, democracy.
With all that in mind, the intention of this chapter is to shed light on the building blocks of
democratic erosion. In Part 2, I propose a general theory that attempts to group these steps into
analytical categories that are all relevant, and that, together, contain most of the steps. The
second part of the theory uses these categories to offer insight into the timing and sequence of
the different steps. In Part 3, I ground this theory by conducting an in-depth case study of
Hungary, where Viktor Orbán’s reelection in 2010 marked the beginning of a still ongoing spell
of democratic erosion. I take into account each and every event that might be construed as a step
of backsliding, and evaluate them in light of the theory. Part 4 offers a summary of the main
findings of the chapter.
Hungary, of course, is but one case. As a small member state of the European Union with a
long communist past, it is in many ways different from other countries affected by erosion. In
order to identify general patterns, research will need to include other eroding democracies. This
chapter intends to establish a point of reference, to which later cases can be compared.

31

3.2. The Building Blocks of Democratic Erosion: a Theory
The aim of this section is to provide a basic theoretical framework for the study of the steps of
democratic erosion. At this stage, the theory is meant to facilitate the description of the process,
and thus makes no attempt to explain it. The erosion of democracy consists of many – possibly
hundreds of – different steps. For a better overview of the process, as well as for greater
convenience in later research, it makes sense to group these steps into a few key analytical
categories. The first part of this section establishes these groups, or, in other words, highlights
the types of steps that take place during democratic erosion. Once the categories are in place, a
logical follow-up is to analyze how they relate to each other. In the second part of this section, an
argument is made about the sequential nature of democratic erosion, and, accordingly, the
relative timing of the groups is explored.
3.2.1. Analytical Categories
Knowing the components of a machine gives a competent engineer a good idea of where the
machine might in theory be broken. I use this approach to determine the key categories of the
steps of democratic erosion. The “machine,” in this case, has to be a part of the political process.
I work with the assumption that in democratic politics, the government’s main objective is to
keep its power, or, in other words, to be re-elected. This is also true in eroding democracies,
where, by definition, free elections still take place. The government, or some of its members,
may well have other goals in mind (such as accumulating wealth, making history, etc.), but none
of these is likely to be achieved without a substantial amount of time in power. It is therefore the
“components” of the political process that have the potential to influence a government’s
chances of being re-elected that need to be examined, for the steps of democratic erosion are
expected to be closely connected to them.
32

The government’s path to re-election. The following figure shows the main elements of the
political process that may affect electoral results, and their relationship to each other. (An arrow
pointing from A to B means that A has an impact on B.)
A basic assumption of democratic politics is that governments are re-elected or driven out of
office based on their performance. The performance of a politician is an aggregate of how they
behave in public (i.e. how honest, loyal, etc. they are), and the result of the policies they
champion. The relative importance of integrity and efficiency may vary. If engulfed by a
corruption scandal, the politician is unlikely to be saved by sound policies. In case of failed
reforms, few voters will care that the politician responsible for the failure is otherwise a model
citizen. It also bears mention that usually, a political actor’s performance is judged relative to
that of rival politicians or parties; one with a low performance may prevail at the polls if the
other options are even worse. Of course, integrity and efficiency may affect each other.
Corruption, for example, tends to undermine efficiency, while it is more difficult to break well-
designed laws. Finally, both integrity and efficiency may be influenced by external factors, such
as a global economic crisis or a foreign power intent on causing trouble.
While performance should be the main determinant of one’s electoral success, in reality, most
voters have little first-hand knowledge about the political actors’ integrity or efficiency. What
truly matters, therefore, is the voters’ perception of the leaders’ performance. This is an
aggregate of their personal experience (e.g. how a new tax policy affects them) and the second-
hand information they get from various sources. Obviously, experience is the more reliable of the
two; however, it is also much more limited. Information can increase a politician’s chances of
being re-elected, if it is favorable to them (pro) or unfavorable to their rivals (con). In the
opposite case, the likelihood of success will decrease. The amount of positive and negative
33

information on a political actor available to voters depends on two factors: sources that provide it
(such as political parties, civil society organizations, etc.) and the resources these actors possess
(money, volunteers, etc.), which determine how effectively their messages can be transmitted.
Again, the two factors may affect each other. For example, civil society organizations may help
raise money for various causes, while money can buy the political support of certain sources.
Similarly to performance itself, the perception of one’s performance is judged in comparison to
that of their opponents.

Figure 3.1. The political process and the government’s points of attack

ELECTORAL RESULTS
ELECTORAL DECISION
PERCEPTION
IRREGULARITIES
EXPERIENCE INFORMATION
PRO
CON
RESOURCES
SOURCES
EFFICIENCY INTEGRITY
PERFORMANCE
EXTERNAL FACTORS
GOVERNMENT CHECKS
1
2 3
4
5
34

Voters rely on their perception of the politicians’ performance to guide their decision on
Election Day. (How exactly they process that input is the subject of an extended literature.) The
electoral decision is, first, the choice of going to the polls or not, and, in the former case, the
expression of one’s political preferences. However, the exact same voting behavior by the
electorate may lead to different electoral results, depending on the electoral system in place.
(Contrasting proportional representation and first-past-the-post voting is common place in
political science, and that is just one of the many possible variations in the electoral system.) Of
course, there can also be irregularities that affect the electoral results, like administrative
problems, malfunctioning machines, etc. In a democracy, outright cheating is not possible.
Unsurprisingly, the government has a key role in the process shown by the above graph. Its
integrity and efficiency determine the performance rivals are compared to, which is then, to some
extent, experienced by the voters. It is itself a source of information with considerable resources,
and can have a direct impact on the circumstances in which other sources function, including the
amount of resources available to them. Governments can also influence the electoral decision,
e.g. by making it harder to vote, and thereby increasing the number of absentees. Given a large
enough majority in the legislature, they are in control of the electoral system, and thus may
choose the way votes add up to the electoral results.
In liberal democracies, there is another crucial group of political actors. They are the state
institutions with the power to oversee the actions of the executive, and possibly reverse them.
Examples include the courts, electoral commissions, etc. These checks also affect how other
actors – like the sources of information – function, protecting them from abuse and, at the same
time, making sure they themselves respect the laws. Some checks may also be involved in
preventing or remedying electoral irregularities.
35

Corrupting the System. All of the above stands true for perfectly well-functioning as well as
eroding democracies. As mentioned previously, the ultimate purpose of governments in both
types of countries is to keep power. The main difference is that in the latter, leaders are willing to
take steps that decrease democratic quality to that end. The question, then, is at what points and
how the process depicted by the graph can be corrupted.
Political actors make choices that determine their integrity and efficiency. Making these
choices with electoral victory in mind is an integral part of democratic politics. Moreover, once
they are made, objective reality cannot be changed: breaches of integrity cannot be undone, and
chosen policies will run their course. Consequently, the leaders’ performance is not an area in
which eroding democracies are expected to differ from the norm. Perception, on the other hand,
is much more malleable. The voters’ first-hand experience is hard to manipulate. Again, policies
that favor certain groups are fair game in a democracy. A possible way to corrupt the system is to
directly transfer resources (money, food, etc.) to voters, though whether such a step is a sign of
erosion or not depends on the context.
Information, in turn, is highly vulnerable to manipulation by a government willing to sacrifice
democratic quality. Logically, this can be done in two ways: the voice of pro-government
sources can be amplified, typically by increasing their number or the resources available to them
(both ultimately dependent on state resources); or that of anti-government sources can be
reduced. Since political actors cannot be banned or resources taken away arbitrarily if the
pretense of democracy is to be maintained, the latter is expected to happen chiefly through
restrictive legislation, or through political or economic pressure.
In a democracy, even an eroding one, elections need to be free. Fairness, however, is more
complex and less tangible, and thus leaves a lot of room for the government to change the rules
36

of the game. The various rules governing elections may very well be tailored to the strengths of
the government or the weaknesses of its opponents. Of course, the electoral system is
occasionally modified in well-functioning democracies, too, and no government will want to go
against its own interests. However, if the changes have no justification other than to serve those
in power, they can probably be considered steps of democratic erosion. The electoral decision
may be affected, by discouraging voters to take part in the elections, or even by changing the
electorate; and so may the way in which electoral results are calculated. Other than engineering
the electoral system, the government may also be behind some of the irregularities that occur
around the elections. Even though outright cheating is incompatible with democracy, there may
indeed be more irregularities in case there is ongoing erosion.
No account of a decline in democratic quality can be complete without touching upon the
“immune system” of democracy, or the institutions that serve as checks. As long as they are
intact, the government will have trouble eroding democracy, since its more controversial
decisions can be overturned, and even if they are not, opposition by the checks may turn public
opinion against those in power. Consequently, dealing with these checks is not just a possible
way to corrupt the system; it is necessary for the erosion of democracy to proceed. Furthermore,
as mentioned previously, some of the checks have power over other political actors, too – or even
society as a whole – thus as a result of the government taming them, not only will they cease to
use their powers as intended, but they may go as far as to abuse them.
Having described the “machine” that determines a government’s re-election and identified the
“components” that might be broken, this knowledge can now be used to compile a list of
analytical categories that, together, contain most steps of democratic erosion.
37

1. Changing the rules that regulate elections to the advantage of the government. For a step to
be classified into this group, the change needs to build on the government’s strengths or its
opponents’ weaknesses. Moreover, it cannot have a reasonable justification beyond the
government’s interest. Possible examples include changes in the electoral procedures; changes in
the way electoral campaigns are regulated, including their financial aspect; gerrymandering; etc.
Laws that increase the electoral threshold or make it harder to cooperate may work especially
well against a fragmented opposition.
2. Using state resources to advance the government’s interests. The government typically
controls significantly more resources (primarily money, but also data, expertise, labor, etc.) than
other political actors. Most of these do not belong to the people in power, but that does not mean
they cannot use them to their advantage, in case they are willing to disregard certain norms (and
possibly laws). As mentioned above, resources help amplify the voice of pro-government
sources, both by increasing their number and loyalty, and by multiplying the resources they can
then spend on spreading their message. Resources can also be devoted to directly influencing
voter experience about the government.
3. Weakening potential sources of criticism. This group contains the steps meant to reduce the
voice of anti-government sources, thereby decreasing their capacity to consistently reach a large
number of voters, or, in other words, to shape the information that influences electoral decisions.
Steps in this group may affect the circumstances in which these sources operate (e.g. the
restriction of certain rights, such as the freedom of assembly); or the amount of resources
available to them (e.g. opposition parties losing part of their state funding, or media outlets being
burdened with higher taxes).
38

4. Weakening or taking over institutions of accountability. There are two possible ways of
dealing with the checks that might derail a government’s anti-democratic agenda. Some of their
powers may be taken away, so that they lose their ability to effectively oppose the government.
Alternatively, the executive may contrive to “take over” an institution of accountability, making
sure the people in positions of power are loyal to the government. Packing the
Constitutional/Supreme Court is a typical example of this strategy. When there are no empty
seats to fill, a possibility is to create some, e.g. by lowering the retirement age. Another way of
forcing out the leadership of an institution is by reorganizing it, e.g. by merging it with another
institution. Also included in this group are steps that make it easier for the new leadership of
subverted checks to serve the executive’s interests – for example laws that increase the
institution’s power, or grant greater discretion to its leaders.
5. Actual instances of abuse of power. The government’s influence over checks does not bode
well for democratic quality. However, for the hypothetical threat to materialize, actual abuses of
power need to happen. That is exactly the case when institutions of accountability fail to keep an
eye on the government and reverse transgressions, even when mandated by law. Another type of
abuse in this group is when checks turn against the government’s rivals, increasing the political
or economic pressure these groups have to face. Abuses of power by other state institutions and
state-owned enterprises have also been classified here.
From an analytical standpoint, these five categories seem adequate to cover the most
important steps of democratic erosion, and thus tell an accurate story of the process. The
categorization may of course be refined in light of a growing body of empirical data.

39

3.2.2. Sequence of Steps
It seems reasonable to assume that the steps described above are not used by the government at
random; the erosion of democracy is, to a degree, a systematic process. A logical way to
structure steps is to establish how the five analytical categories follow each other in time. There
is reason to believe sequencing the steps of erosion makes theoretical sense. First, some steps
may be needed to enable others, which should affect the order in which the different methods are
employed by the government. Second, some opponents may represent more of a direct threat to
the government’s chances of being re-elected than others, meaning that they need to be dealt
with first. Third, it is improbable that governments aiming for democratic erosion would come to
power with a complete plan. A more likely scenario is that their appetite grows with each
successful step, meaning that easier ones will come before those that require more complicated,
possibly unlawful methods.
In what follows, I discuss three consecutive phases of democratic erosion: Phases 0, A, and B.
The main goal here is to connect the analytical categories presented in the previous section to the
different phases. In case the empirical data show that the categories are indeed characterized by a
separation in time, the sequential nature of erosion would be confirmed. Below, the five
categories are referred to as Groups 1 to 5.
Phase 0: the Eroder comes to power. Democratic erosion is by definition a process driven by
the executive, meaning that any political actor intent on undermining democracy in this way first
needs to come to power. This phase is of primary importance for research on the prerequisites
and causes of democratic erosion. However, since most of the steps assume that the political
actor behind erosion – the Eroder – has control over the legislature, access to state resources,
etc., this period of time has less relevance to the description of the process.
40

Phase A: changing the institutional setup. The erosion of democracy truly begins once the
Eroder has assumed power. The process is by definition gradual and characterized by the
appearance of legality, meaning that the new government cannot do whatever it pleases. The
selection of steps at this stage is expected to be determined by two considerations. First, they
need to be feasible with the tools the Eroder has at his disposal: a legislative majority, the right to
appoint new leaders to certain institutions, etc. More informal – and possibly illegal – steps will
usually have to wait until the government has consolidated its power, and other relevant actors
have adapted to the new situation. Second, as mentioned above, some steps are necessary to
enable others. Obviously, these steps have to take precedence.
In light of these observations, the group that is certain to feature in this phase of erosion is
Group 4 (taking over state institutions). Dealing with institutions of accountability is crucial, lest
they attempt to thwart the Eroder’s agenda. On the other hand, when subverted, they can be used
to assist in later steps. Also, weakening and taking over these checks is usually straightforward,
provided that the government has sufficient legislative support. As such, these steps may be
undertaken right after the electoral victory, and completed relatively soon thereafter. Similarly,
many steps in Groups 1 (electoral rules) and 3 (weakening sources of criticism) only require
legislation, meaning that they may very well be parts of Phase A. Obviously, Group 1 steps are a
priority, given that the main objective for the Eroder is re-election. Logically, the “deadline” for
tweaking the electoral rules is sometime before the following elections. While it is not in theory
impossible that Phase A is significantly shorter than an electoral cycle, one would expect Group
1 steps to take place in this initial phase of democratic erosion. As for Group 3, since there can
be various sources of criticism, and targeting them may require tools other than legislation, the
41

corresponding steps probably occur in both Phases A and B – whenever the government
identifies a new opponent.
Phase B: exploiting the new institutional setup. Eventually, the first steps taken by the Eroder
bear fruit, enabling others. By this stage, the government’s power – formal and informal – is
such that it may resort to tools that would be considered controversial in well-functioning
democracies (misuse of state resources; political and economic pressure; etc.). This is a self-
reinforcing process: the more power the government has and the less control it needs to
accommodate, the broader the selection of available steps will become.
Since the abuse of power by state institutions (including checks) is made possible by the
appointment of strong pro-government leaders, Group 5 (abuse of power) is expected to follow
Group 4 (taking over state institutions), and thus feature mainly in Phase B. The same is true for
Group 2 steps (misusing state resources): the more entrenched a government is in power, the
more freely it may use – and thus misuse – state resources in order to serve its own interests,
electoral and otherwise.
Of course, even if the existence of these 2 (+1) separate phases is supported by empirical
evidence, there might be some overlap between them. Group 1 and Group 4 steps may be present
in Phase B; for example, the government may feel the need to modify the electoral system after
its re-election so as to implement the lessons learned during the recent campaign. Similarly,
some Group 2 and Group 5 steps may take place in Phase A. The sooner a state institution is
taken over, the earlier it may start to abuse its power. Overall, though, there should be
significantly more Group 1 and Group 4 steps in Phase A than later on, while the number of
Group 2 and Group 5 steps is likely to increase over time. As noted before, Group 3 steps may
occur in spurts throughout the process of erosion.
42

The above observations are summed up in the following table and figure:

Table 3.1. Timing of the steps of democratic erosion by category
Group Phase A Phase B
1 Changing electoral rules X
2 Misusing state resources X
3 Weakening sources of criticism X X
4 Taking over state institutions X
5 Abuse of power X

Figure 3.2. Timing of the steps of democratic erosion by category

3.3. Data Analysis
In this section, a new dataset containing the various steps of democratic erosion that have
occurred in Hungary since 2010 will be used to assess the theoretical points made above. This is
not meant to be a statistically rigorous testing of formal hypotheses. As argued previously, the
state of knowledge on the erosion of democracy leaves plenty of room for exploratory work that,
in turn, may eventually contribute to such testing. First, the dataset will be introduced, with an
emphasis on how it was compiled and processed. Second, it will be considered whether the data
PHASE 0
PHASE A
PHASE B
THE ERODER COMES TO POWER
GROUP 1
t
GROUP 2
GROUP 4
GROUP 5
GROUP 3
43

support the analytical categories established in the previous section. Finally, observations will be
made on the timing of the steps of democratic erosion.
3.3.1. Introduction of the Dataset
The data used for the subsequent analysis in this section was gathered in the following way. I
used the online archives of Index.hu, a Hungarian-language news site. I applied two filters to
narrow down the search: I only considered the articles that appeared in (1) the domestic section
(“Belföld”) (2) between January 1, 2010 (the year of Viktor Orbán’s return to power) and
October 31, 2018 (the time of the data gathering). I skimmed through close to a hundred
thousand articles and made note of events that may be construed as steps of the erosion of
democracy. As a result, I obtained a dataset of possible steps of erosion in chronological order.
Justification of the data-gathering strategy. My goal was to create a list of steps of democratic
erosion in Hungary that is as comprehensive as possible. Some steps may be covered up as a
result of editorial bias in outlets overly friendly to the government; blown out of proportion in
outlets aligned with the opposition; or overlooked due to a dearth of resources in outlets that lack
the financial background for complete coverage. The choice of Index.hu, Hungary’s leading,
independent news site, should make sure that no steps of erosion are left out of the list because of
these reasons. Another advantage of Index.hu is that its archive covers the whole period in focus.
Since the erosion of democracy is a domestic phenomenon, narrowing the search to the domestic
section seems reasonable, even though some steps may be presented in other sections. In fact, as
detailed in the following chapter, many steps related to the media have appeared in the cultural
section of Index.hu. Ideally, data gathering should have extended to all articles, but due to
limitations in capacity, the line had to be drawn somewhere. Still, there is no reason to assume
44

that developments with a major impact on Hungarian politics would not be addressed in the
domestic section.
Processing the data. Once compiled, the raw list was processed in preparation for further
analysis. First, the timing of the various steps was established. April 2010, the month Viktor
Orbán was elected, was coded as 1 – the first month of the new government. May 2010 was
coded as 2, June 2010 as 3, etc., until October 2018, or month 103 of the Orbán regime. (No
events of importance were recorded for the first three months of 2010.) Second, individual items
were assigned to the analytical groups described previously. The coding was as follows:

Table 3.2. Coding of the steps of democratic erosion: analytical categories
Code Meaning
0 Not a step
1 Changing the rules that regulate elections to the advantage of the government
2 Using state resources to advance the government’s interests
3 Weakening potential sources of criticism
4 Weakening or taking over institutions of accountability
5 Actual instances of abuse of power
6 Other
p Propaganda (not a step)
c Context (to a step)

Items coded 1 to 5 could be linked to the analytical groups defined in the theory section. The
number 6 was assigned to items that constituted steps of erosion, but did not fit into any of these
five categories. Items that provided context to a step of erosion (but were not steps themselves)
were assigned a number (depending on the nature of the step) and the letter c – e.g. 3c for a news
article that shed light on the background of an abuse of power by a state institution. The rest of
the items may have been news articles on controversial measures or actions by the government,
but, upon closer inspection, were deemed not to be steps of erosion. For example, nominating
leaders to state institutions is not a sign of erosion in itself. If, however, laws were changed to
45

force out the previous leadership ahead of time, that action does qualify as a step of erosion. The
difference between items labeled with 0 and those with a p is that the latter focused on key
elements of the government’s messaging, and thus, though not steps of erosion, may be of
interest for different research questions.
Finally, items on the list (excluding those that were determined not to be steps of erosion)
were classified according to the “weight” of the steps:

Table 3.3. Coding of the steps of democratic erosion: weight
Code Meaning
1 Noteworthy, but not grave/has no impact on the quality of democracy
2 Clearly a step of erosion, but its impact/gravity is small
3 Significant impact/gravity
4 Major impact/gravity

The weights were assigned in light of the context, i.e. taking into account the political
situation in Hungary at the time the decisions were made. Thus changing the electoral system
may not necessarily have an impact in itself on the quality of democracy, but reducing the
number of rounds from two to one when the opposition is fragmented certainly results in a major
increase in the government’s chances of re-election. Another caveat to note is that the
classification is somewhat arbitrary, since there is no scientific way of establishing the exact
impact or gravity of a step. However, we can still have a reasonable estimate of its magnitude,
which is all we need to distinguish, for example, “significant” from “major.” This is exploratory
work; later iterations will need to aim for increased precision.
Description of the data. The dataset consists of 483 items, out of which 238 were categorized
as true steps of erosion. The number of items in each category is as below:

46

Table 3.4. Descriptive statistics: analytical categories
Category No. of Items Percentage
1 Changing electoral rules 17 7%
2 Misusing state resources 50 21%
3 Weakening sources of criticism 26 11%
4 Taking over state institutions 22 9%
5 Abuse of power 108 45%
6 (Not assigned to any of the groups) 15 6%

As for the classification according to weight:

Table 3.5. Descriptive statistics: weight
Weight Number of Items Percentage
1 30 13%
2 87 37%
3 99 42%
4 22 9%

3.3.2. Analytical Categories
The analytical classification proposed in the theory section will be supported by the data if the
categories are (1) relevant, meaning that there are major steps of erosion in all of them; and (2)
complete, meaning that there are no major steps that do not fit into any of them. The task, then, is
to analyze all six groups of items defined above. The first five represent the analytical categories,
and thus the expectation is that items of importance will be found in them; whereas the opposite
is true for the sixth, which is a collection of the remaining, non-classified items. Examples of
steps will be provided for all groups.
1. Changing electoral rules. Out of the 17 items in this category, 1 was classified as major
and a further 5 as significant, showing that changes in the electoral system were indeed an
important part of the erosion of democracy in Hungary. The new electoral law of December 2011
contained several elements that were designed to decrease the opposition’s chances in future
47

elections. It instituted a single-round system, which made it easier for government candidates to
win their districts even without an absolute majority when facing a fragmented opposition.
Districts were redrawn, often resulting in gerrymandering; it became simpler to field candidates,
further fragmenting the opposition; etc. The law was clearly tailored to the strengths of Viktor
Orbán’s Fidesz party, and to the weaknesses of its opposition. Another step with a significant
impact was the constitutional amendment passed in October 2012 which allowed Hungarians
living permanently abroad to vote by mail, while withholding this right from Hungarian residents
who were only temporarily abroad. The first group consists mostly of Hungarian minorities in
neighboring countries – strong supporters of Fidesz; whereas the latter is made up of people
working in Western countries, who are more likely to vote for the opposition. Another
constitutional amendment (March 2013) decreed that media outlets are not allowed to charge
money for political advertisements, effectively undermining the incentive of private outlets to
broadcast such advertisements – and thereby increasing the importance of alternative channels,
often controlled by the government.
2. Misusing state resources. This second largest group of items in the dataset contains 3 major
and 32 significant steps of erosion. Most of the latter are large-scale public communication
campaigns orchestrated by the government and paid for with taxpayer money. This practice
sheds further light on the insidious nature of democratic erosion. Governments may very well
decide to launch communication campaigns in fully functioning democracies, and it is not
unreasonable to assume that the messages should be somewhat similar to those endorsed by the
party that forms the government. However, the scale of these campaigns in Hungary and the fact
that their sole tangible purpose is to reinforce Orbán’s messaging marks them clearly as steps of
erosion. The opposition (and other political actors, like the civil society) is unable to compete
48

with the vast state resources. As a result, the government’s messages dominate in key issues. The
less access a community has to alternative sources of information (like in Fidesz’s rural
heartlands), the more this is true. Examples for such campaigns include the one titled “Hungary
Does Better” (March to November 2010) for USD 2.2 million and another one in July 2015 for
USD 2.1 million claiming that recent reforms were successful.
The most egregious steps blur the line between public and party interest even further. For
example, the rights to a USD 532,000 campaign were given to Fidesz by the government for
USD 700 (in February 2014 – two months before the parliamentary elections). Similarly, in
March 2016, the Cabinet Office of the PM gave the rights for the slogan “the Hungarian reforms
are working” to Fidesz for free. Some state-owned enterprises were also used to divert public
resources. The electricity company MVM spent more than USD 7.7 million between January
2016 and May 2017 to advertise in pro-government media outlets. Since most outlets depend
heavily on income from advertising, such an intervention has the potential to seriously disrupt
the media market. MVM’s decisions were clearly not based on business considerations, since it
is not operating in a competitive environment, and some (not government-friendly) outlets were
overlooked in spite of their broader reach. While advertising can be justified to a certain extent,
MVM also supported CÖF, the “civil society” organization sponsored by the government, with
USD 1.8 million in 2017. While the state resource thus hijacked is mostly money, it can also be
expertise. For example, pro-government think tank Századvég was given USD 5.9 million in
2017 for its analyses – which, crucially, are not available to the public, or even to the opposition
parties.
3. Weakening sources of criticism. Among the 26 items in this group, 4 were coded as major
and 11 more as significant, which proves that the existence of this category is well justified when
49

discussing the erosion of democracy. In November 2010, a change in the constitution narrowed
the set of issues that the Constitutional Court (CC) may address, significantly weakening the
check on executive power. To make the matter even more egregious, the move came as a direct
answer to the CC striking down a piece of legislation – one that was thus removed from its
competence. Since then, legislation concerning for example taxes cannot be brought to the CC.
In a similar move, another constitutional amendment (November 2013) was used to add a series
of laws that were previously struck down by the CC directly to the constitution. At the same
time, the set of issues the CC could address was further restricted. In another example of a step
with major implications (June 2017), civil society organizations were obliged by law to report
any financial support from abroad, and to indicate on every publication that they were recipients
of such support. These new rules came at a time when some civil society organizations were
branded as foreign agents by the government.
4. Taking over state institutions. Overall, 22 items were placed in this group, 7 of which were
classified as major and 12 as significant. Again, these numbers imply that this is indeed a useful
category when describing the erosion of democracy. One of the major steps of democratic
erosion was the adoption of a new constitution in April 2011. A new constitution may be
completely harmless to the quality of democracy, but it may also fit into more than one of the
analytical categories discussed here. In the Hungarian case, the number of justices on the
Constitutional Court was increased from 11 to 15, paving the way for a government-friendly
majority. Another part of the new constitution decreased the retirement age for judges from 70 to
62 years, meaning that many of them were forced into retirement – and replaced by the new,
Fidesz-nominated leadership of the National Judicial Council. (In September 2012, after the law
was struck down by the CC, the retirement age was only decreased to 65 years, but judges were
50

still barred from holding leadership positions above 62.) Another step in a different area that
proved to be major in the Hungarian context was the creation of National Media and
Infocommunications Authority (July 2010) to supervise the media, with its powerful leader
appointed directly by the prime minister.
5. Abuse of power. This is by far the biggest group, covering 108 individual items, which
contain 6 major and 33 significant steps of democratic erosion. It should come as no surprise that
the government using nominally independent institutions for its own ends is a central aspect of
erosion. These institutions – which include the public media, state-owned enterprises, etc. – have
a great variety of functions and associated powers, meaning that the abuses they may commit are
also diverse. For example, a step coded as major was the suspension of the tax number of several
civil society organizations in September 2014 as a part of a concerted attack by the government
against them. (In Hungary, as in most other countries, a valid tax number is needed for an
organization to function.) Later, it turned out that neither the attack, nor the move by the
National Tax and Customs Administration was in any way justified. Another egregious abuse of
power was the State Audit Office’s decision to impose a crippling fine on Jobbik, the strongest
opposition party at the time, less than six months before the parliamentary elections (December
2017). Some smaller parties were also fined. To make matters even worse for them, there was no
legal route to appeal these decisions. Other steps in this group included the “occupied”
Constitutional Court’s reluctance to decide in controversial issues (November 2017 and June
2018); the National Media and Infocommunications Authority’s decision to strip opposition
radio Klubrádió of its frequency (December 2011); the National Election Commission’s rejection
of referendum initiatives by the opposition; the Hungarian News Agency’s repeated failure to
report on the suspicious enrichment of Fidesz politicians and their family members; etc.
51

(This last point raises a methodological problem: when the abuse is not a single event, but
becomes part of the “normal” functioning of the institution – and thus loses its immediate
newsworthiness – it cannot be captured by processing news articles.)
Other steps of erosion. As shown above, the analytical categories defined in the theory section
are all necessary to describe the erosion of democracy. The remaining question is whether they
are sufficient, too, or, in other words, whether they cover most of the steps in the dataset. The
number of items left out of the five groups is 15, out of which 1 was coded as major and another
6 as significant. The following table summarizes these steps:

Table 3.6. Steps of erosion not classified into any of analytical groups 1-5
Date Month Weight Description
10/08/16 79 4 Publication of biggest opposition newspaper Népszabadság
suspended; workers claim a “coup” has taken place
02/08/13 35 3 Electricity and gas companies will have to highlight in
orange (color of Fidesz) on every bill how much money the
consumer is saving
06/03/14 51 3 The editor-in-chief of Origo.hu leaves with immediate
effect; he does not deny being under political pressure
07/04/14 52 3 Bill on advertisement tax passed: media companies taxed
after advertisement income – greatest burden on RTL Klub
12/01/16 81 3 Changes in the leadership of 8 regional newspapers
08/15/18 101 3 Critical journalists at conservative periodical Kommentár
fired; the periodical to become even more conservative
09/24/18 102 3 The conservative periodical Századvég was withdrawn
because of an article critical to the government; the
editorial board resigned

There is no denying the relevance of these events to the erosion of democracy in Hungary.
The closure of Népszabadság, the biggest left-leaning newspaper (October 2016) was a clear
blow to media freedom. When the editor-in-chief of Origo.hu, a hitherto independent, widely
read news site resigned (June 2014), he did little to quell rumors of political pressure – namely,
the prospect of lucrative government contracts to Origo.hu’s owner, German telecommunications
52

company Deutsche Telekom. When entrepreneurs affiliated with the government bought a large
portfolio of regional newspapers, the leadership of these outlets was immediately removed
(December 2016). A less conspicuous, but – in the Hungarian context – still significant step was
the law taxing the advertising income of media companies in a way that disproportionately
affected RTL Klub, the country’s biggest TV channel, putting it at a disadvantage compared to
its main, pro-government rival (TV2).
What makes the classification of these items especially difficult is that most of them were not
initiated by the government or state institutions, but by formally apolitical actors. Népszabadság
was closed by an owner with only informal ties to Fidesz, who claimed that the newspaper was
not profitable anymore. Similarly, the regional newspapers mentioned above were not purged by
the government. Of course, the government shares indirect responsibility for these events, as it
distorted the media market though legislation (e.g. the tax on advertising income) and strategic
advertising in friendly outlets.
Do these items undermine the analytical categories defined above? Not really, for two
reasons. First, their number is relatively small: only 15 out of 238, or 6 percent (including 5
percent of the major steps and 6 percent of the significant steps). However, these numbers were
even smaller for the first analytical group. The big difference is that those items could be
gathered in a single group (changing the electoral system), while these have little in common
when it comes to what exactly happens. Thus, even though it has to be noted that the analytical
categories do not cover every single step of democratic erosion, a closer look at the left-out items
confirms that no additional groups are needed. Furthermore, 6 of the 7 items on the above list are
related to the media, so an in-depth analysis of what happens to the media during the erosion of
democracy should properly address them (see Chapter 4).
53

3.3.3. Sequence of Steps
As mentioned above, the months starting with April 2010 have been numbered from 1 to 103
(October 2018). In order to visualize the timing of the steps of erosion, 103 rows in an Excel
document have been assigned numbers from 1 to 103, and a cell was colored in these rows for
each step that took place in that specific month. Four colors have been used, in line with the
different weights defined previously:

Table 3.7. Color coding of the steps of democratic erosion according to weight
Code Meaning Color
1 Noteworthy, but not grave/has no impact on the quality of democracy

2 Clearly a step of erosion, but its impact/gravity is small

3 Significant impact/gravity

4 Major impact/gravity

For clarity’s sake, cells representing steps with a greater weight were grouped to the left. The
resulting figure is an exact visualization of the dataset. Its disadvantage is that the great number
of rows and the relatively small number of cells in each one (especially for individual categories)
makes it hard to draw conclusions about the timing of the steps. For a more in-depth analysis,
months were organized in groups of six (months 1 to 6, 7 to 12, etc.) and then in groups of 12
(months 1 to 12, 13 to 24, etc.), one for each year following Viktor Orbán’s return to power. The
cells were colored in the same way as before.
Before analyzing the 5+1 analytical categories and their relative timing, a strong caveat is in
order. As noted above, even though the methodology employed to build the dataset seems
adequate, some steps may be missing; and assigning weights to them is somewhat subjective. In
addition to these concerns, it is not always straightforward to determine the exact timing of a
step. For example, in case of a public communication campaign that lasts for several months,
54

which of them should be used for the analysis? Should we instead consider the month in which
the decision was made to launch the campaign? Also, reports about a step may emerge months
after the fact, without being able to pinpoint the exact time the abuse took place. Thus it is
possible that some of the cells have been assigned the wrong month. However, since we look at
semesters and years, and not individual months, most of these problems should not have a
significant impact on the big picture.
1. Changing electoral rules
Expected pattern: most steps early on; some steps later

Figure 3.3. Analytical group 1: temporal distribution of the steps of erosion

A few important observations can be made based on this figure. First, it is apparent that, even
though some steps happen right after the election of the Fidesz government, the key piece of
legislation (the new electoral law) only takes place late in the second year of Orbán’s
1 - 6 1 - 12
7 - 12 13 - 24
13 - 18 25 - 36
19 - 24 37 - 48
25 - 30 49 - 60
31 - 36 61 - 72
37 - 42 73 - 84
43 - 48 85 - 96
49 - 54 97 - 108
55 - 60
61 - 66
67 - 72
73 - 78
79 - 84
85 - 90
91 - 96
97 - 102
103 - 108
Month Month
55

premiership. After another hiatus, the majority of the steps occur in the 21 months leading up to
the parliamentary and municipal elections of 2014 (months 49 and 55, respectively). Clearly,
there was no need to hurry these decisions, as long as the new system was in place on time. After
the 2014 elections, there is a conspicuous absence of further steps for more than three years. It is
safe to assume that the government was satisfied with the changes, having secured 67 percent of
the parliamentary seats with only 45 percent of the votes. There is some fine-tuning of the
system before the 2018 elections, but these latter steps pale in comparison to previous ones, in
terms of both number and significance.
Conclusion: steps somewhat later than expected, but still relatively early on
2. Misusing state resources
Expected pattern: some steps early on; most steps later

Figure 3.4. Analytical group 2: temporal distribution of the steps of erosion

1 - 6 1 - 12
7 - 12 13 - 24
13 - 18 25 - 36
19 - 24 37 - 48
25 - 30 49 - 60
31 - 36 61 - 72
37 - 42 73 - 84
43 - 48 85 - 96
49 - 54 97 - 108
55 - 60
61 - 66
67 - 72
73 - 78
79 - 84
85 - 90
91 - 96
97 - 102
103 - 108
Month Month
56

According to the figure, there have been considerably more – almost twice as many – steps in
the second term of the government as in the first one, similarly to the previous analytical group
(17 against 32 out of 50). In fact, the number of steps has remained consistently high in the
second term, with at least 4 steps every year, out of which at least 2 were significant.
Discounting the first year of the term, these numbers are even more impressive: there have been
at least six major or significant steps in each of these three years. Since these steps were in large
part public communication campaigns, it can be concluded that by the second term, constant
propaganda has become the norm for the Orbán government. It is also apparent that there is a
spike in the number of steps before the 2014 elections (months 37 to 48), as should be expected
from the connection with political communication. This spike is only matched by the year
leading up to the 2018 elections (months 85 to 96). Tellingly, there is a notable decrease in the
number of steps after both elections: there are no steps in months 49 to 54, and only one in
months 97 to 102. Clearly, there is less need for communication after elections.
Conclusion: pattern as expected
3. Weakening sources of criticism
Expected pattern: steps both early on and later
The figure suggests that steps in this group have been happening right from the beginning of
the period under scrutiny. There have only been 5 semesters out of 18 without any changes of
this kind (9 if we only consider major and significant steps). Even so, the first two years stand
out, implying that it was a priority for the government to curb the power of various institutions. It
is also apparent that the intensity of such steps decreases between months 43 and 84 – for more
than three years. Finally, there is another spike in the number of steps between months 85 and
100. Overall, months 1 to 24 and 85 to 100 contain 17 out of the 26 steps in this group (65%).
Interestingly, in the first period, the steps typically target state institutions, like the Constitutional
57

Court, while in the later period, it is actors outside of politics that come into focus, from the civil
society to the Hungarian Academy of Sciences.

Figure 3.5. Analytical group 3: temporal distribution of the steps of erosion

Conclusion: more steps early on, but pattern mostly as expected
4. Taking over state institutions
Expected pattern: most steps early on; some steps later
As shown by the above figure, the timing of the steps in this group follows a clear pattern –
the clearest for all 5+1 analytical categories. The great majority of the steps – 17 out of 22, or
77% – take place in the first 30 months of the Orbán government. The ratio is even higher for the
most significant steps (weight = 4): 6 out of 7, or 86%. Accordingly, there are only a few
scattered steps after the fifth semester. The notable exception is a constitutional amendment that
made it possible to establish a system of administrative courts, taking away politically sensitive
issues from existing courts and the Supreme Court (June 2018). This move was a reaction to the
1 - 6 1 - 12
7 - 12 13 - 24
13 - 18 25 - 36
19 - 24 37 - 48
25 - 30 49 - 60
31 - 36 61 - 72
37 - 42 73 - 84
43 - 48 85 - 96
49 - 54 97 - 108
55 - 60
61 - 66
67 - 72
73 - 78
79 - 84
85 - 90
91 - 96
97 - 102
103 - 108
Month Month
58

fact that the previous attempt to occupy courts was only partly successful: there proved to be
considerable institutional resistance to the government-friendly head of National Judicial
Council. This suggests that some steps of erosion do not have the intended effect, but also that
occasional failures do not discourage the government.

Figure 3.6. Analytical group 4: temporal distribution of the steps of erosion

Conclusion: pattern as expected
5. Abuse of power
Expected pattern: some steps early on; most steps later
Since this group contains the greatest number of steps, it is no surprise that there is only a
single semester – the very first one – without any kind of abuse. In fact, from month 13, there are
at least five steps of erosion every year, with two or more of them significant. However, that
does not mean the distribution of steps is homogeneous. Out of the 108 steps in this group, 25
1 - 6 1 - 12
7 - 12 13 - 24
13 - 18 25 - 36
19 - 24 37 - 48
25 - 30 49 - 60
31 - 36 61 - 72
37 - 42 73 - 84
43 - 48 85 - 96
49 - 54 97 - 108
55 - 60
61 - 66
67 - 72
73 - 78
79 - 84
85 - 90
91 - 96
97 - 102
103 - 108
Month Month
59

take place in the government’s first term (2010-2014), 73 in the second (2014-2018), and 11
more in the third, ongoing term. (The corresponding values for the major steps: 0, 4 and 2 out of
6). Clearly, there were significantly more instances of abuse in the second term than the first one.
It seems safe to assume that time was needed to create an environment supportive of abuse,
through legislation and the occupation of institutions. Also, it is telling that, even though the
dataset only covers 7 months of the third term (months 97 to 103), there have already been 11
steps of erosion, including 2 major and 3 significant. At this rate, the third term will see even
more cases of abuse than the second.

Figure 3.7. Analytical group 5: temporal distribution of the steps of erosion

The greatest spike in the number of steps comes the semester (and the year) after the
government’s (first) reelection (months 49-54). These steps include a concerted (and, as it later
turned out, completely baseless) attack against a group of civil society organizations and the
refusal by competent authorities to investigate alleged irregularities in connection with the
elections. Interestingly, there is another spike the semester (and the year) before the 2018
elections. These observations imply that elections play a key role in the timing of at least some of
the steps of erosion. This is true in part because the ultimate goal of all steps is to increase the
1 - 6 1 - 12
7 - 12 13 - 24
13 - 18 25 - 36
19 - 24 37 - 48
25 - 30 49 - 60
31 - 36 61 - 72
37 - 42 73 - 84
43 - 48 85 - 96
49 - 54 97 - 108
55 - 60
61 - 66
67 - 72
73 - 78
79 - 84
85 - 90
91 - 96
97 - 102
103 - 108
Month Month
60

government’s chances on Election Day, but also because they act as constraints: those in power
are less likely to make controversial moves right before elections, when they have no time to
course correct, if needed.
Conclusion: pattern as expected
Other steps of erosion. Since steps of erosion in this group are few and different in nature, no
patterns are expected. This is mostly confirmed by the tables, though it is notable that there are
markedly more steps in the second term of the Orbán government (9 out of 15) than in the first
(3). Also, the seven months of the third term have already brought 3 more steps, which suggests
that the final tally for this term may very well end up surpassing that of the pervious one; though,
because of the diverse nature of these steps, one should refrain from extrapolating.

Figure 3.8. Temporal distribution of the steps of erosion not in analytical groups 1 to 5

1 - 6 1 - 12
7 - 12 13 - 24
13 - 18 25 - 36
19 - 24 37 - 48
25 - 30 49 - 60
31 - 36 61 - 72
37 - 42 73 - 84
43 - 48 85 - 96
49 - 54 97 - 108
55 - 60
61 - 66
67 - 72
73 - 78
79 - 84
85 - 90
91 - 96
97 - 102
103 - 108
Month Month
61

Relative timing of the groups. The figure and the table below serve to summarize what we
know about the timing of the steps in each group.

Figure 3.9. Temporal distribution of the steps of erosion: overall

Table 3.7. Distribution of the steps of erosion by group in the three terms of the Orbán government

The figure is a collection of the six figures already seen separately when discussing individual
groups. The table contains the number of steps for each group in the two terms of the Orbán
government since 2010 (with data for the first seven months of the third). Weighted totals take
into account that not all steps have the same weight, and thus consider the total weight of the
steps instead of their raw numbers. (E.g. two steps, weighing 2 and 3 respectively, would count
as 5.) Percentages have been added for greater clarity.
1 - 6
7 - 12
13 - 18
19 - 24
25 - 30
31 - 36
37 - 42
43 - 48
49 - 54
55 - 60
61 - 66
67 - 72
73 - 78
79 - 84
85 - 90
91 - 96
97 - 102
103 - 108 1 2 3 4 5 6
Month
No. of steps % No. of steps % No. of steps % No. of steps % No. of steps % No. of steps %
1 11 65 17 34 15 58 19 86 25 23 3 20
2 5 29 32 64 8 31 2 9 72 67 9 60
3 1 6 1 2 3 12 1 5 11 10 3 20
1 24 67 44 32 37 55 60 86 50 21 7 18
2 11 31 90 66 21 31 6 9 161 68 23 61
3 1 3 3 2 9 13 4 6 27 11 8 21
Raw
Weighted
Group 1
Term
Group 2 Group 3 Group 4 Group 5 Group 6
62

Before the analysis, let us remember the predictions made in the theory section. The
assumption was that erosion has three phases: one before the election of the new government
(Phase 0), and two afterwards (Phases A and B). The steps in Groups 1 (electoral rules) and 4
(taking over state institutions) were assumed to take place mostly (though not exclusively) in
Phase A, whereas Phase B would contain most of the steps in Groups 2 (misusing state
resources) and 5 (abuse of power). Group 3 steps (weakening sources of criticism) would be
found in both periods of time.
If we view the first term of the Orbán government as Phase A and the second (and early third)
as Phase B, the figure and the table confirm the predictions. The intuitive conclusions one may
draw from the former are supported by the numbers in the latter: 65 percent of Group 1 steps
have happened in the first term, and only 35 after that. The weighted numbers tell the same story:
67 percent against 33. The results are even more convincing for Group 4 steps, with 86 percent
taking place in the first term no matter how we calculate the numbers. The second (and third)
term dominates when it comes to Group 2 and Group 5 steps. For the latter, only 23 (raw) or 21
(weighted) percent of the steps occur in the first term. For the former, the corresponding values
are 34 and 32, which are somewhat bigger, but still below one third. Finally, 58/55 percent of
Group 3 steps happen in the first term and 42/45 later. Overall, these numbers are well in line
with the predictions in the theory section. One may argue that Phase B starts even earlier than the
second term, or month 49. In fact, setting the starting point at month 31 would make sense for 4
of these 5 groups, the exception being Group 1, which would mostly slip into Phase B as a result.
This is a question that will in all likelihood be answered by analyzing other cases of democratic
erosion. If, at the end, Phase A is deemed to be at least as long as the government’s first term,
that would suggest that a single term is not enough to completely undermine democracy.
63

3.4. Conclusion
In the theory section, two sets of predictions were made about the steps of democratic erosion.
The first one centered on a list of five analytical categories theorized to be necessary and
sufficient to contain all the steps. The second one dealt with the relative timing of the steps in
these groups, or, in other words, the possibility that steps might occur in distinct phases. The
dataset based on the actions of Viktor Orbán’s government since 2010 mostly confirmed these
predictions. The five aforementioned groups are indeed useful when describing the steps of
erosion, and even though some steps do not fit into any of them, no additional groups can be
defined, so the classification is more or less complete. Similarly, the data showed that steps in
certain groups typically happen earlier than others, thus asserting that erosion is sequential,
consisting of separate phases, makes theoretical sense.
The story that emerges from the Hungarian case is that the Eroder starts its tenure by making
it possible to occupy and subvert institutions, thereby creating the necessary background for later
steps. Relatively early, but certainly before the following elections, the electoral system is also
changed to the advantage of the government. At the same time, the power of actors that may
check or criticize the government is curbed; a process that continues in the next phase of erosion.
An assumption that needs to be supported with further data is that state institutions are targeted
in the initial phase and non-state actors later on. Once the checks and balances have been
weakened, there is less opposition to the steps in the second phase of erosion: abusing the power
of state institutions (including state-owned enterprises) and diverting public resources to advance
the government’s interests.
64

Even though these first, exploratory results are encouraging, further research is needed in both
areas. The classification can be fine-tuned; differentiating between the targets of Group 2 steps is
just one possibility. Of special interest are steps that are not carried out by the government or
state institutions, but by allies of the ruling party, such as friendly oligarchs. The chapter on the
media (Chapter 4) will have more on that. Going more in-depth with the analytical categories
will obviously have an impact on our understanding of the different phases of erosion. However,
the most important task is to extend the data analysis to countries other than Hungary, to see to
what extent the above results are generalizable.

65

Chapter 4
Media Freedom and the Erosion of Democracy: A Comparative Case Study of
Five Countries

4.1. Introduction
There are several reasons to assume that the media play a key role in democratic erosion. The
Dahlian criteria of a true democracy, or polyarchy, include the freedom of expression and the
availability of alternative sources of information (Dahl 1971), both organically connected to the
state of the media. Since erosion proceeds by undermining these criteria, it would be surprising if
the media were not affected. Let us also keep in mind that the government’s main goal is re-
election, which only indirectly depends on its performance. What truly matters is the perception
of its performance, and since voters have limited personal experience about the impact of
government policies, it is the information they receive about them – its substance as well as its
framing – that will determine their electoral decision. Information is provided by certain sources,
some of which are in direct contact with citizens, such as political parties and civil society
organizations. This is especially true during electoral campaigns. However, in modern mass
democracies, much of the information is relayed indirectly. That is where the media come into
the picture.
To put it simply: representative democracy is based on the accountability of those in power;
accountability is only possible if the voters know about government policies and their
consequences; and the media are the main channel of information between the government and
the governed. In an ideal world, political actors would only be able to influence information
66

through their actual performance. However, since it plays such a crucial role in determining
one’s electoral fortunes, politicians routinely make an effort to shape it even in the highest-
quality democracies. As a consequence, information may become distorted, dissociated from the
political actors’ true performance.
This is made possible by a variety of problems with the media, from extreme partisanship to
owners with a political agenda. There is indeed a whole strand of literature on media bias (e.g.
Bennett 2016; D’Alessio and Allen 2000; Lee and Solomon 1991; Lippmann 1922; or more
recently Eberl, Boomgaarden, and Wagner 2017; Puglisi and Snyder 2015). One would expect a
government orchestrating democratic erosion to go further: to attempt to control the flow of
information as tightly as possible, so as to amplify positive coverage and filter out any piece of
news that might reduce its chances of being re-elected. In the process, political information will
become less and less reflective of actual performance, with regards to both the government and
its opponents. This makes the media a natural target, and thus an important part of the
description of erosion.
As concluded previously (see Literature Review), there is no shortage of works that focus on
the media in eroding democracies. Consequently, much is known about the strategies employed
to undermine the freedom of the press. However, two shortcomings were identified: the list of
the various measures that target the media is neither comprehensive (if not for individual
countries), nor systematic enough to provide the basis for a more theoretical approach. The goal
of this chapter is to address these shortcomings.
The chapter is organized as follows. In the next part, I formulate a theory of how the media
are affected by erosion. My goal is twofold: to provide a useful framework for thinking about
erosion by establishing analytical categories that encompass most steps; and to make
67

observations on the timing and sequence of the different steps. Part 3 presents a case study of
actual steps of democratic erosion that relate to the media. Steps in Hungary (2010-), Turkey
(2003-), Venezuela (1999-2012), Ecuador (2007-2016) and Poland (2015-) are considered. The
number of countries in focus, as well as linguistic limitations, makes it impossible to go as in-
depth as a single subject would allow. On the other hand, this setup is more conducive to a
comparative approach, and therefore the resulting observations are more likely to be generally
true for countries suffering from democratic erosion.

4.2. Media Freedom and the Erosion of Democracy: a Theory
In the following section, hundreds of different steps of erosion will be presented. In order to
make sense and to be a useful basis for further research, the presentation needs to be systematic.
To this end, analytical categories need to be defined to capture the types of steps that characterize
the process. The first part of this section establishes these groups by building a theory around the
main players and the consideration of how the flow of information may be distorted. At this
point, the theory is not meant to provide causal hypotheses, but to facilitate description, which
might in turn lead to the study of causality. In the second part, I explore the relative timing of the
different steps, i.e. which analytical groups come first, and which ones tend to occur in later
stages of the process of erosion.
4.2.1. Analytical Categories
The main players of a country’s media ecosystem are presented in the figure below.

68

Figure 4.1. Media ecosystem (country-level)

The government can have a major impact on the media in a variety of ways – most obviously
through legislation. The assumption here is that the executive branch either has control over the
legislature (i.e. the parties that make up the majority are its allies), or has the right to issue
decrees to the same effect. Democratic erosion can only gather steam if this assumption is true
(see Chapter 5). Public media are owned by the state, and thus led by government appointees.
They are set apart from private media outlets by their raison d’être: while those exist primarily to
make money, public media are meant to provide public services – information, culture, sports,
etc. – for free. The final player is the media authority: a sort of intermediary between legislation
and private media outlets, which has the right to allocate state resources (such as radio
frequencies), punish outlets that breach regulations, etc. (While international authorities might
GOVERNMENT
MEDIA AUTHORITY
PUBLIC MEDIA
PRIVATE MEDIA
APPOINTMENT OF
PARTY-AFFILIATED
LEADERSHIP
GREATER POWERS
GREATER DISCRETION
APPOINTMENT OF
PARTY-AFFILIATED
LEADERSHIP
GREATER RESOURCES
GREATER REACH
(See below)
1
2
69

play a role in shaping the media landscape, e.g. by setting standards, the focus here is on the
domestic, state-run regulatory agency.)
Even in a well-functioning democracy, there may be connections between the government and
the other three players. For example, the government usually has the right to appoint the
leadership of the public media and the domestic media authority. The editorial line of certain
private media outlets may fall close to government ideology; in fact, there are outlets, typically
newspapers, founded by political parties. However, such connections do not mean that these
players blindly serve government interests, or, in other words, that the government’s wishes
supersede legal and professional considerations. The erosion of democracy is expected to change
that: the other three players represent an obvious target for a government intent on controlling
the flow of information. It seems reasonable to assume that most steps of erosion involving the
media will be directed against these players, thus it makes sense to have them serve as a basis for
the classification of said steps. The public media are limited in number, have one, well-defined
owner (the state) and often a centralized leadership, meaning that the steps targeting it will
probably fit into a single analytical group. The same can be said about the media authority. This
leads us to the first two categories.
M1. Taking control of the public media. The end goal for a government more interested in
cementing its own power than preserving democracy is public media that function as loyal and
effective channels for transmitting its political messages. Loyalty depends chiefly on the media’s
leadership, thus in the course of democratic erosion, we should see a change in that area, with the
new leaders emerging from the government’s political orbit. They will then make sure that
critical voices within the organization are silenced. The public media can only be effective
conveyors of government propaganda if they reach a wide audience, thus the number and
70

thematic diversity of public media outlets is expected to grow. This, of course, can only happen
if they are not constrained by a lack of resources, so another logical step is to significantly
increase the budget available to the public media sector as a whole.
M2. Taking control of the media authority. While the media authority is supposedly a
politically neutral player whose role is to ensure compliance with relevant legislation and
represent the state in certain issues, it can be a powerful tool against media freedom. Since it has
a degree of control over market access, it can be used to filter out – or force into compromise –
outlets that are critical of the government. Pressure can also be applied by routinely punishing
anti-government media, either by fining them, or, in more extreme cases, by (temporarily)
closing them down. A consequence of its usefulness is that the media authority is likely to be
targeted by the government in the process of erosion. The observed steps should serve the same
purpose as in the case of the public media: to increase loyalty and effectiveness. Again, the
former can be guaranteed by appointing a new, loyal leadership. The media authority is not
expected to reach masses of voters, thus greater effectiveness is brought about by greater power
(greater potential impact in more issues) and greater discretion (greater leeway in interpreting the
laws and regulations). The latter may lead to subjective criteria when deciding about new
entrants to the market, fines of arbitrary magnitude, etc., creating the possibility for the
discrimination of critical outlets.
Other institutions may also have an – albeit smaller – impact on the flow of information. An
example is the data protection authority, a type of institution that exists in many countries (e.g.
the Office of the Privacy Commissioner of Canada, or the Federal Commissioner for Data
Protection and Freedom of Information in Germany), tasked with protecting information privacy.
Steps targeting these institutions will also be included in this group.
71

The private media constitute a more diverse group in terms of their ideological alignment,
financial background, etc., thus bending them to the government’s will is likely to be a more
complex process, leading to more analytical categories. In order to help with classification, the
figure below presents the flow of information, with an eye on how it can be distorted.

Figure 4.2. Potential obstacles in the way of political information

Let us consider the performance of political actors a part of objective reality: a politician is
either corrupt or not; the rate of unemployment either decreases or not; carbon dioxide emissions
have an actual value; etc. Political information, defined as a simple description of reality, cannot
in itself be distorted. That changes, however, as performance is perceived, i.e. information makes
its way to the voter. The focus here is on indirect transmission, independent of the voters’
personal experiences (which, of course, can also be subjective). Information may be distorted for
a number of reasons in well-functioning democracies. The sheer number of political actors and
actions, some of which are highly complex, makes it necessary to filter and interpret pieces of
information, which opens the way for subjectivity. The media’s own biases are well documented
(see above). Of interest here are the distortions engineered by the government in an eroding
democracy. That means the voters themselves do not need to be considered. While they may
distort information in a variety of ways, some of which may be amplified in the process of
INFORMATION JOURNALIST/OUTLET VOTERS
NO ACCESS 3
NO J/O TO TRANSMIT 4
DO NOT TRANSMIT
RESTRICTIONS
PUNISHMENT
DROWN OUT
5
7 6
8
72

erosion (e.g. lack of interest in politics, excessive partisanship), these are not directly related to
the media.
The information’s journey has two stages: first, it needs to reach the media, which will then
pass it on to the voters. The flow of information may be distorted at both points: it is possible
that the media will not acquire certain pieces of information; and, even if they do, they may
decide not to share them with the public. Both can happen organically, but also as a result of the
government’s interference. Another way in which the government may attempt to keep
information away from voters is to manufacture more content, hoping – based on the average
voter’s limited time and interest in politics – that critical information will be drowned out. A
third point of interest is the transmitter itself: even if information is available and ready to be
transmitted, the government may force the outlets willing to do the job out of business. With
these considerations in mind, let us define the remaining analytical groups.
M3. Limiting access to information. The government may attempt to keep information it
deems to be detrimental to its odds of being re-elected away from the voters. Cases of corruption
involving leading politicians are an obvious example, but this practice may extend to any facet of
decision-making or reporting about outcomes. Access may be limited physically: for example, a
party may ban reporters (from certain outlets) from its events, or boycott them in some other
way. It may also be difficult to enter state institutions to request information. Another set of
obstacles is administrative. The government may classify information beyond what is justifiable;
deny press credentials; or charge prohibitive sums of money for access to institutions or
documents.
M4. Eliminating critical outlets. Another way in which the government may distort the flow
of information is by reducing the number of sources able and willing to report on controversial
73

issues. Closing down such outlets, or even attempting to do so, are steps of erosion that belong in
this category. The outlet may be eliminated completely or in part. For example, a radio station
that uses several local frequencies to broadcast may lose some, but not all, of them. The
government does not necessarily have to destroy critical information; it may be enough to limit
the number of people it reaches. Similarly, the outlet may survive, but a journalist or a certain
program may need to be taken off the air. Closing down an outlet for a limited amount of time
also counts as eliminating critical reporting. It is worth noting that in eroding democracies, the
façade of the rule of law is mostly kept intact, thus the government rarely makes direct moves
against the media. Instead, the media authority and government-friendly owners lead the charge.
M5. Regulating content. In a well-functioning democracy, editorial decisions are made within
media outlets. As democratic erosion progresses, editorial boards may face more and more
restrictions. Two types of information may influence electoral fortunes: pieces that strengthen
the government’s position; and pieces that weaken it. Accordingly, there are two types of
restrictions: the obligation to publish content (or feature certain political actors) and the
prohibition to do so. The latter, in its extreme form, is censorship; though one would not expect
to see much of that in countries still aspiring to be called democracies.
M6. Influencing content through financial pressure. The bottom line of media outlets (and
other companies sharing the same owners) has the potential to be significantly affected by
government policies. In an eroding democracy, one would expect the government to use this
leverage as a bargaining chip in order to obtain more favorable coverage. Taxes and fees may
serve this purpose, though it is difficult to tailor them to specific outlets, depending on how
amenable they are to cooperation. The discriminative allocation of state advertising is a more
versatile tool, especially at a time when many outlets – including most traditional broadsheets –
74

struggle financially. Crucially, many owners have business interests outside of media, which
increases the number of sticks and carrots available to the government. Steps that distort the
media market to the point that they may lead to changes in ownership (and thus content) are also
included in this category.
M7. Influencing content through the threat of punishment. Withholding advertising money,
mentioned in the previous paragraph, is a sort of punishment. Steps in this group are
conceptualized as being even more direct. Some of them are legal in nature: journalists may have
to go through the hassle of a long and costly trial, with the potential of a prison sentence or a
hefty fine waiting at the end. (Fines for outlets are classified into the previous category.)
Administrative punishments may include increased attention from state institutions like the tax
authority. In some cases, punishment may even be physical, though that is more likely to happen
in dictatorships than in eroding democracies. The point of punishing critical outlets is of course
to make sure they become less confrontational in the future.
M8. Building a pro-government media empire. Critical coverage cannot be completely
suppressed in even a superficially democratic setting. Ultimately, what matters to the
government is the relative amount of supportive and critical coverage. It is therefore crucial to
have outlets that can produce the former, so as to drown out the latter. The pro-government
media portfolio may be expanded in two ways. First, new outlets may be created; second,
existing critical – or not sufficiently supportive – outlets may be acquired and subverted. The
second solution has the benefit of killing two birds with the same stone. Involved in the process
are usually pro-government business groups and a lot of advertising money.
The above six groups are, of course, not completely independent from each other. For
example, the restrictions presented in Group M5 (regulating content) often serve as a basis for
75

the legal/administrative punishments that make up Group M7. Financial pressure (Group M6)
may in certain cases force an outlet out of the market (Group M4); in other cases, it may lead to a
change in ownership (Group M8). New owners, in turn, may get rid of the most critical
journalists and programs (Group M4). Yet, in spite of these connections – and the fact that as
many as eight groups have been defined above – the analytical categories (see table below) seem
distinct and theoretically relevant enough to stand on their own. Since the classification is based
on a theoretical framework that was designed to cover all eventualities, the number of steps that
do not fit into any of these eight groups should be limited. Whether that is truly the case is
assessed in Part 3.

Table 4.1. Steps of erosion: analytical groups
M1 Taking control of the public media
M2 Taking control of the media authority
M3 Limiting access to information
M4 Eliminating critical outlets
M5 Regulating content
M6 Influencing content through financial pressure
M7 Influencing content through the threat of punishment
M8 Building a pro-government media empire
Social media. The analytical categories were defined with traditional media in mind. Since
social media play an increasing role in everyday life, it is possible that their prominence has
made certain categories obsolete, while necessitating the creation of new ones. In order to decide
whether that is the case, let us consider how social media affect the journey of political
information from its source to the voter (see Figure 2.2). There are three notable changes: (1)
information may reach the voters directly, without going through traditional media; (2)
individual journalists may transmit information efficiently without the backing of an
organization; and (3) it is easier for voters to share information with each other.
76

These changes notwithstanding, there is no reason to rethink the classification. First, as
suggested by the literature (see Literature Review), social media are still far from supplanting
traditional media as conveyors of political information, so any steps that undermine the latter’s
freedom remain relevant for would-be Eroders, and thus for scholars. Second, the potential
obstacles in the way of political information (see Figure 2.2) are mostly the same for social
media. Access to information (Group M3) may be easier, but only in certain cases. Classified
information, or data that public institutions are unwilling to divulge, are still out of reach.
Similarly, content on social media may be regulated by governments (Group M5), and the threat
of being punished (Group M7) may affect online journalists and other users. On the other hand,
financial pressure on media outlets (Group M6) will probably play less of a role, considering that
journalists may choose to share information as private individuals.
The one obstacle that social media may help to circumvent is the lack of critical outlets
(Group M4), given their decentralized nature and the great number of users. Of course, whole
applications may be suspended or banned. Also, not everyone has access to social media,
especially in poorer countries and outside of urban centers. Finally, it is precisely these two
features that make it easier for governments to drown out critical information (Group M8) by
deploying online “trolls,” misinformation, etc. (see Literature Review).
In sum, while the growing impact of social media on the flow of information needs to be
acknowledged, the obstacles remain the same, and therefore so should the analytical groups.
4.2.2. Sequence of Steps
The classification of the various steps of democratic erosion provides a useful theoretical
framework for further descriptive work. In addition, it may also make it easier to identify
patterns in the way the different types of steps follow each other in time, thereby contributing to
77

a deeper understanding of the phenomenon. Having greater knowledge of the sequence of steps
would have practical implications, too: it would help to assess how far along the process a
country is; anticipate steps, and thus resist, or even reverse them before they can do lasting
damage; etc. Of course, such universal patterns may not exist at all. It is entirely possible that
democratic erosion affects the media in different ways across countries or regions. There are
reasons why this would not be surprising: media systems (Hallin and Mancini 2004, 2011) might
differ; the initial level of media freedom might be lower in certain countries, leaving the Eroder
with less to do; there might be considerable variations in the way the media are regulated, as well
as the ownership structure; etc. Moreover, other factors might also play a role, such as civic
culture, Internet penetration, etc. Still, if we consider democratic erosion to be a distinct
phenomenon, at least some similarities should be expected across cases. Part 3 is designed to
bring greater clarity to this issue.
Let us assume for now that the steps of democratic erosion do follow a certain temporal
pattern, and consider what that pattern might be. In order to do that, it seems expedient to first
establish what general tools the government has at its disposal to effectuate these steps.
Certain tools are available to any government, whether democracy is eroding or not. These –
let us call them direct – tools include legislation; drafting the country’s budget; the right to
appoint functionaries to key positions (or refuse to do so); etc. Similarly, the decision to respect
norms – e.g. to hold regular press conferences – or not is entirely up to the government. The
extent of some of these powers depends on the degree of control the government has over the
legislature. A minority government will have to make compromises with other actors, and is thus
unlikely to be able to deploy these tools against democracy. On the other hand, overwhelming
legislative support will leave the government with considerable room for maneuver; it may even
78

rewrite some of the rules, if that is necessary to make the most of its tools. For example,
restructuring an institution may make it possible to appoint new leaders earlier than otherwise.
Crucially, direct tools are available from the moment the new government assumes power; they
have no prerequisites.
State institutions (including state-owned enterprises) have certain regulatory, financial, etc.
powers. In high-quality democracies, these institutions function independently of the
government’s political interests. However, once the erosion of democracy has started, and the
government has imposed its will on these institutions, their powers become indirect tools it may
use to further the process. These tools include regulatory oversight: the power to decide whether
an organization works in accordance with relevant laws, and impose a punishment if it does not;
law enforcement itself; and, in the case of state-owned enterprises, the right to spend money on
advertising. Unlike the previous group, indirect tools are not immediately available to the
government, for they require the subversion of state institutions. Some of them may fall quickly;
for others, like the courts, the process may take considerably longer.
The government gains access to the third group of – informal – tools through its allies. Again,
there is nothing inherently antidemocratic in some sort of limited cooperation between political
actors and business leaders. However, the relationship becomes problematic if businesspeople
and enterprises owe their success to their place in the government’s political orbit; or if they are
forced to serve the government’s interests. Once that happens, the government may use their
resources as tools to entrench its power. These tools include disposal over huge sums of
advertising money; and the ability to acquire other companies – sometimes reshaping the market
in the process. Of the three groups, this one should become available to the government last.
Increasing its allies’ economic power often requires help from various state institutions, which
79

need to be subverted first; and time for the changes to take place, provided that the semblance of
the rule of law is to be preserved. Similarly, non-aligned enterprises will not capitulate to the
government overnight: democratic erosion needs to be in a sufficiently advanced stage for some
of the threats and promises to be credible.

Table 4.2. Tools at the government’s disposal to effectuate steps of erosion
Direct Indirect Informal
Legislation Regulatory oversight Private advertising
State budget Law enforcement Acquisitions
Appointment Public advertising
Let us now determine which of these tools are required for the government to carry out the
steps of erosion in our eight analytical categories. While the availability of a step is only a part of
the picture, since the government may refrain from resorting to it for various reasons, it is
nonetheless a good basis for establishing expectations on relative timing.
Taking control of the public media (Group M1) consists chiefly of replacing their leadership
and increasing their budget. Other changes – such as letting go of critical journalists and
expanding the public media’s reach – follow from these initial steps. The required tools are
exclusively direct ones, thus one would expect the brunt of these steps to happen soon after the
government takes power – though finding the most efficient way to structure and regulate these
outlets may necessitate more than one round of legislation.
The same can be said about taking control of the media authority (Group M2), which requires
the appointment of new leaders and legislation to increase their power and discretion to make
more subjective decisions. Since a compliant media authority will itself provide powerful tools
for further steps of erosion, subverting it is expected to be a priority for the new government.
Again, later steps may happen so as to fine-tune the system.
80

Physically banning reporters from certain events with the aim of limiting access to
information (Group M3) does not require any tools, and therefore may happen at any time,
depending on how confrontational the government is. However, administrative obstacles
presuppose specific legislation and firmly pro-government state institutions. The former is
possible right after the elections; the latter once institutions have been taken over. Yet, while
some of these tools are immediately available, one has to wonder whether limiting access to
information is truly a priority for a government with a transformative agenda that has just gained
power, and thus has barely generated any information worth hiding. Instead, it is more likely that
this area will only come into focus once other priorities have been taken care of.
As noted above, eliminating critical outlets (Group M4) falls to the media authority (and
possibly other state institutions), and later to government-friendly owners. The media authority
may use its regulatory powers to set conditions for continued operation that certain outlets cannot
meet; revoke their license; impose prohibitive punishments; etc. In addition to traditional media
organizations, social media platforms and online communities may be targeted in a similar
fashion. This tool is available to the government as soon as the media authority’s loyalty has
been assured. However, major private media are unlikely to be forced out of the market in this
way in a supposed democracy, with possibly a few exceptions. Government-friendly owners
represent an informal tool, so they are not expected to play a significant role until the later stages
of democratic erosion.
Regulating content (Group M5) is possible through either legislation or the regulatory power
of state institutions – mainly, again, the media authority. Legislation is available to the
government early on. Yet, one needs to bear in mind that extreme restrictions on media content
are incompatible with even an eroding democracy, while minor ones are unlikely to have a
81

significant impact, and thus should not be a priority for the government. The best guess with
regards to the timing of these steps is that a few of them may occur early on, but we should see
most of them once democracy has sufficiently eroded.
While influencing content through financial pressure (Group M6) may involve legislation, the
main tool at the disposal of the government is the strategic use of advertising money. This is a
combination of an indirect and an informal tool (public and private advertising). The informal
use of a normally direct tool (state funds) to woo or threaten major media owners may also play a
role. Overall, steps in this group are expected to feature relatively early on, but not before the
government has established its hold over state institutions. Once its informal power has
sufficiently grown, more steps are likely to follow.
Influencing content through the threat of punishment (Group M7) tends to happen within the
framework of regulatory oversight and law enforcement, both indirect tools. That alone suggests
that some time needs to pass after the government has assumed power for steps in this group to
become prevalent. The same conclusion follows from an observation made previously, namely
that steps in Group M5 (regulating content) tend to serve as a basis for those in Group M7.
However, one type of “punishment” is available to the government and its allies early on: critical
media outlets and journalists may be named and shamed, and the barrage of personal attacks may
eventually lead to self-censorship.
Building a pro-government media empire (Group M8) is undertaken by pro-government
business groups, who either create new outlets or acquire existing ones. Creation is possible any
time (in fact, some of it may even predate the elections); though, as erosion progresses, these
business groups will control exponentially more resources, and thus will have greater room for
maneuver. Similarly, time is needed to weaken critical outlets, precipitating their sale.
82

Acquisitions are, of course, an informal tool. All things considered, steps in this group are
expected to feature more prominently once democratic erosion has already gathered steam, i.e. a
couple of years after the election of the Eroder, following other steps.
Bearing in mind the three types of tools available to the government to further democratic
erosion, the analytical categories can tentatively be assigned to three corresponding phases.
Phase M
D
: contains steps that only require direct tools, and thus happen immediately after the
government comes to power. Groups M1 (public media control) and M2 (media authority
control) are expected to be a part of this phase, with possibly a few steps added later on.
Phase M
ID
: comprises steps that require mostly indirect tools, and therefore occur after state
institutions have been subverted by the government. Groups M3 (limiting access to information),
M4 (eliminating critical outlets), M5 (regulating content) and M6 (financial pressure) should be
in this phase, though they are all expected to feature in the next one, too.
Phase M
IF
: contains steps that mostly require informal tools (even though they may also
involve direct and indirect ones). This phase begins once the government’s non-state allies have
been sufficiently strengthened, and it has built up the credibility that is necessary to pursue
effective “carrot and stick” tactics. Group M8 (pro-government media empire) is an obvious part
of this phase. Group M7 (threat of punishment) is also classified here, for it presupposes a
relatively advanced stage of erosion.
It needs to be noted that no group is restricted to a single phase; in fact, there are some that
will feature to a similar degree in successive phases. Also, it is difficult to determine when
exactly Phase M
ID
and Phase M
IF
begin. State institutions are not all subverted at the same time,
and even when they have been taken over, some time is needed to turn them into effective
enforcers of the government’s will. As for the government’s informal power, there is no
83

conceptual basis for measuring it, and much of it is hidden anyway from the public. Cultural
differences may also affect the length of these phases in various countries. However, precision is
not the goal here. If the data (see Part 3) do confirm the existence of reasonably distinct phases,
establishing the sequence of steps will become significantly easier: analytical groups will be
measured against each other by comparing the phases they take place in, and not the dozens of
individual steps they contain.

Table 4.3. Classification of the analytical groups into the phases of democratic erosion
Group Phase M
D
Phase M
ID
Phase M
IF

M1 Taking control of the public media X .
M2 Taking control of the media authority X .
M3 Limiting access to information . X X
M4 Eliminating critical outlets X X
M5 Regulating content X X
M6 Influencing content through financial pressure X X
M7 Influencing content through the threat of punishment . X
M8 Building a pro-government media empire . . X

4.3. Data Analysis
The purpose of this section is to test whether the arguments made in Part 2 conform to the reality
of democratic erosion. To this end, I use a new dataset compiled from Freedom House’s yearly
Freedom of the Press reports. The theoretical framework yielded eight analytical groups. The
classification is empirically supported if (1) all eight groups are relevant (i.e. contain steps of
democratic erosion); and (2) there are only a few steps that do not belong in any of the groups,
and they not constitute a separate, ninth group. When considering the sequence of steps, the
underlying assumption was that there are patterns in the timing of the different types of steps
across all affected countries. If that is indeed the case, we should observe that (3) at least some of
the analytical groups occur in specific stages of democratic erosion; and (4) they may possibly be
grouped into distinct phases. The theory led to concrete predictions about the approximate timing
84

of the analytical groups and their place in three phases; however, any clear pattern would be a
positive result. If any of these hypotheses is confirmed by the data, our understanding of the
process of democratic erosion would increase considerably.

Table 4.4. Hypotheses
Analytical Categories H1: Groups M1 to M8 are all relevant, i.e. they contain steps of erosion.
H2: Steps outside of Groups M1 to M8 do not constitute a separate group.
Sequence of Steps H3: The analytical groups occur in specific stages of democratic erosion.
H4: Democratic erosion has distinct phases.
4.3.1. Introduction of the Dataset
The case study is built on the yearly Freedom of the Press reports by Freedom House (Freedom
House 2019). These reports chronicle the state of the media in a given country, with a special
emphasis on relevant changes. Reports are available up to 2016. The case study covers five
countries that have experienced democratic erosion: Hungary (from 2010, Prime Minister Viktor
Orbán’s return to power); Turkey (from 2003, Prime Minister/President Recep Tayyip Erdoğan’s
rise to power); Venezuela (from 1999, President Hugo Chávez Frías’ first election, to 2012, the
last full year of his presidency before his death); Ecuador (from 2007 to 2017, the full tenure of
President Rafael Correa Delgado); and Poland (from 2015, the return to power of Jarosław
Kaczyński’s Law and Justice Party).
The five countries under scrutiny were chosen for a number of reasons. First, differences in
operationalization notwithstanding (see Literature Review), the literature on democratic erosion
has identified all five as clear cases of erosion. Second, taken together, they provide the kind of
diversity needed to test the hypotheses: some Eroders are from the political right, others from the
left; the countries are located in three different regions; display significant differences in terms of
size, population, GDP/capita, etc. Moreover, the variety extends to some of the factors
mentioned above: the initial level of media freedom; media regulation; ownership structure; civic
85

culture; Internet penetration; etc. (e.g. Turkey’s Freedom of the Press score was 55 on a scale of
0 to 100 in 2003; Hungary’s was 23 in 2010). Finally, the inclusion of more countries from the
same region might help to identify regional patterns even if universal ones are found not to exist.
An argument could be made to add a negative case so as to ensure that the potential patterns are
in fact due to democratic erosion, and not to general changes in the media system (e.g. the
emergence of new technologies). However, it is unlikely that many steps of erosion would occur
in stable democracies. Fortunately, the process has not started at the same time in all five
countries, but over the course of a relatively long sixteen-year period (from 1999 in Venezuela to
2015 in Poland). This should increase the robustness of the eventual conclusions.
Freedom of the Press reports were used to identify individual decisions made by the
government, by state institutions under its control or by allied business groups that either
represent a direct blow to media freedom or contribute to an environment in which such blows
are more likely to happen. In this way, a list of the steps of democratic erosion involving the
media was created for all five countries.
Assessment of the data-gathering strategy. Freedom of the Press reports have only been
published since 2001, meaning that they were not available for the first two years of the Chávez
presidency (1999 and 2000). In their stead, Freedom House’s Freedom in the World reports
(Freedom House 2019) were used for this short period. These do take into account changes
concerning the freedom of expression, though with less emphasis than the dedicated reports. It is
unfortunate that data may be scarcer for the crucial first years of the Chávez presidency, but the
main developments are likely to be captured. In Poland’s case, democratic erosion has only
started in late 2015, thus there are little more than a year’s worth of steps on the list. They alone
86

are unlikely to reveal much about the process of erosion; however, they may help shed light on
its first phase (M
D
), which might be over by the second year.
There is, of course, a limit to the level of detail Freedom of the Press reports are able to
provide. An empirical strategy that goes more in-depth would have been preferred; but the
number of countries in focus, as well as linguistic limitations, called for a compromise. Still, it
seems reasonable to assume that all significant developments were included in the reports, using
methodology that is in all likelihood consistent across time and countries. Clearly, later iterations
of this research have plenty of room (and need) for improvement. Overall, though, the Freedom
of the Press reports seem like a valid starting point.
Processing the data. A few additions were made to the raw data in order to facilitate its
analysis. First, the timing of the various steps was established. The calendar year in which the
government rose to power is marked as 1; the one after as 2; etc. When the elections took place
in one year, but the leader was only inaugurated in the next one, the latter is considered Year 1.
Naturally, these numbers denote different actual years in different countries. For example, 2010
for Hungary and 2017 for Ecuador were both coded as 1. This number was then used to
aggregate all the steps into a single list encompassing all five countries.
Next, steps were assigned to the analytical categories defined in the theory section (coded as 1
for Group M1, 2 for Group M2, etc.). When a step did not fit into any of the groups, it was coded
as 0. Finally, an attempt was made to assess their severity, or “weight.” Since it is difficult to
measure the exact impact of individual steps, which might also depend on the context, the
classification was based on personal judgment, informed by a prior overview of all the items.
The aim here was not precision, but rather a basic differentiation of the various episodes. A four-
87

point scale seemed suitable for that. The following color codes were used, with one reserved for
steps that were not classified in this way (more on that later).

Table 4.5. Coding of the weight of individual steps in the dataset
Code Meaning

Not classified

Noteworthy, but not a concrete step/no precise timing

Clearly a step of erosion, but its impact/gravity is small

Significant impact/gravity

Major impact/gravity
A few select items of the dataset are shown below:

Table 4.6. Examples from the dataset
Year Group Step Country
6 4
In May 2012, CONATEL canceled the licence of Telesangay TV, owned by
a member of the opposition.
E
7 4
Hungary’s largest independent daily, Népszabadság, which had uncovered a
string of scandals involving the ruling party, was unexpectedly suspended in
October. Soon afterward, its parent company was sold to a firm linked to an
ally of Prime Minister Viktor Orbán.
H
7 4
The Organization for Security and Cooperation in Europe reported that
approximately 3,700 websites were blocked between 2007 and 2009. The law
allows prosecutors to block sites if their content “incites suicide, pedophilia,
drug abuse, obscenity or prostitution,” or attacks Ataturk.
T
Description of the data. The aggregated dataset consists of 533 items: 63 for Hungary, 223 for
Turkey, 116 for Venezuela, 119 for Ecuador, and 12 for Poland. The number of items in each
category is as follows (for data on individual countries, see Appendix 2.1):
Table 4.7. Descriptive statistics: analytical categories
Category No. of Items Percentage
1 Taking control of the public media 16 3%
2 Taking control of the media authority 19 4%
3 Limiting access to information 26 5%
4 Eliminating critical outlets 95 18%
5 Regulating content 64 12%
6 Influencing content through financial pressure 54 10%
7 Influencing content through the threat of punishment 225 42%
8 Building a pro-government media empire 28 5%
0 (Not assigned to any of the analytical groups) 6 1%
88

As for the classification according to weight:

Table 4.8. Descriptive statistics: weight
Weight Number of Items Percentage

225 42%

36 7%

107 20%

100 19%

59 11%
4.3.2. Analytical Categories
The purpose of this section is to test whether the two hypotheses concerning the analytical
categories are supported by actual data. The first of them (relevance of all eight categories) is
true if all groups contain enough steps of erosion. While there is no formal definition of
“enough,” numbers as well as weights needs to be considered. Steps with major or significant
impact/gravity will he highlighted, with examples from several countries, when available. The
second hypothesis (no other relevant categories) is true if the (relatively few) steps that do not fit
into any of the eight categories do not constitute a group of their own. Accordingly, steps in all
8+1 categories will be analyzed below.
M1. Taking control of the public media.

Table 4.9. Analytical group M1: steps of democratic erosion coded as major or significant
Number of Items Percentage
Total 16 100%

6 38%

1 6%
Examples:

89

Table 4.10. Analytical group M1: examples from the dataset (major or significant)
Year Group Step Country
1 1
The public media, including Hungarian Television, the Hungarian News
Agency (MTI), Hungarian Radio, and Duna TV, were unified under the
jurisdiction of the Media Council, threatening their political and financial
independence. The law gives the head of the NMHH the right to nominate
the executive directors of all public media. The public broadcasters have lost
their right to produce their own news; they are obliged to use MTI’s news
coverage.
H
1 1
The December 2015 law on the public media allows the government to hire
and fire the management of the public television and radio. In response to its
passage, the heads of all public media, including public broadcaster TVP,
immediately resigned in protest.
P
2 1
In February, the head of the MTI said all news programs broadcast by public
service television and radio stations would be produced and edited by MTI
staff within the year.
H
With 16 items in this category, including 1 major and 6 significant ones, it certainly cannot be
dismissed as irrelevant. Typical steps include: cementing control through centralization; firing
critical journalists; and increasing the budget. One note of caution is warranted, though: 12 of the
16 steps occurred in just two countries: Hungary and Poland, including all major and significant
ones. Thus, while taking control of the public media was an important part of democratic erosion
in these countries, it was less so in the other three. That does not mean that the public media did
not matter for those governments. A reason for the lack of concrete steps might be that the public
media were never really free from political influence in Turkey; consequently, less effort was
needed to control them. As for Venezuela and Ecuador, the public media were not a major source
of information before the period of erosion, so, for the government, it was less a matter of taking
control, and more of expanding their reach.
M2. Taking control of the media authority.

Table 4.11. Analytical group M2: steps of democratic erosion coded as major or significant
Number of Items Percentage
Total 19 100%

11 58%

8 42%
90

Examples:

Table 4.12. Analytical group M2: examples from the dataset (major)
Year Group Step Country
1 2
In August, newly elected President Pál Schmitt […] approved legislation
which restructured supervision over Hungary’s public and private media [...]
The NMHH president also chairs the five-member autonomous Media
Council, which has the power to fine television and radio stations for
unbalanced coverage, as well as banning public subsidies to media outlets
that are found guilty […]
H
7 2
After years of failed attempts, in June the National Assembly approved a new
Communications Law. The legislation contained many controversial
provisions, including the creation of two powerful telecommunications and
media regulatory bodies - the Superintendency of Information and
Communication and the Council for the Regulation and Development of
Information and Communication - of questionable independence [...]
E
12 2
The Reform to the Organic Law for Telecommunications, which took effect
in late December, establishes telecommunications as a “public interest
service,” granting CONATEL additional powers to regulate the industry and
suspend or revoke licenses when it deems certain content to be antithetical to
the national interest or security.
V
All 19 steps of erosion in this group are either major (8) or significant (11). Moreover, as
opposed to the previous category, this one encompasses at least two steps from all five countries
included in the case study. Overall, there can be no question about the relevance of this analytical
group. Typical steps include: creating new media authorities loyal to the government; and
increasing their power.
M3. Limiting access to information.

Table 4.13. Analytical group M3: steps of democratic erosion coded as major or significant
Number of Items Percentage
Total 26 100%

9 35%

4 15%
Examples:

91

Table 4.14. Analytical group M3: examples from the dataset (major)
Year Group Step Country
6 3
Amendments to Hungary’s Freedom of Information Act approved in July
2015 allow public bodies to charge fees for vaguely defined labor costs,
potentially making the fulfilment of an information request very expensive.
The changes also allow state bodies to reject claims if the requested data is
“preparatory,” meaning that it could be used in future government decisions;
or if it is copyrighted by a third party; or if the petition is a repeat request –
even if the initial request went unanswered.
H
8 3
A new monetary and financial code passed in July 2014 granted an oversight
board the authority to designate various types of information as confidential
and established penalties for disclosure […]
E
13 3
A discriminatory accreditation system [...] is used to screen out critical
journalists, restricting access to the offices of the president and cabinet
ministers. Amendments published in the official gazette in August 2015
changed the structure and procedures surrounding accreditation. The new
regulations granted the deputy prime minister overseeing the BYEGM the
power to issue permanent press cards, and altered the composition of the
BYEGM’s Press Card Commission, increasing its membership from 13 to 15
but decreasing the number of seats for media representatives from 8 to 5.
T
Half of the 26 steps of erosion in this category are either major (4) or significant (9). With the
exception of Poland, all countries have experienced a minimum of four steps of this kind,
including at least one that is major or significant. Again, the relevance of this group cannot be
disputed. Typical steps include: barring physical access to people, locations or events; and
making certain types of information inaccessible by either legal or financial means.
M4. Eliminating critical outlets.

Table 4.15. Analytical group M4: steps of democratic erosion coded as major or significant
Number of Items Percentage
Total 95 100%

28 29%

15 16%
Examples:

92

Table 4.16. Analytical group M4: examples from the dataset (major)
Year Group Step Country
7 4
Hungary’s largest independent daily, Népszabadság, which had uncovered a
string of scandals involving the ruling party, was unexpectedly suspended in
October. Soon afterward, its parent company was sold to a firm linked to an
ally of Prime Minister Viktor Orbán.
H
8 4
The year’s most dramatic media event occurred on December 28, when the
authorities announced that the license of RCTV, a prominent pro-opposition
network, would not be renewed and the channel would go off the air by May
2007.
V
14 4
Within days of the July coup attempt, RTÜK withdrew the operating licenses
of two dozen television and radio outlets over their alleged links the Gülen
movement, and many more were later closed by executive decree.
T
The sheer number of steps (95) classified into this group proves its relevance beyond doubt.
Moreover, 43 – or almost half – of these steps were coded as major (15) or significant (28). The
numbers are especially high for Turkey (44), Venezuela (24) and Ecuador (19); not so much for
Hungary (6) and Poland (1), though, as mentioned earlier, there is barely more than a year’s
worth of data for the latter. Typical steps include: closing down outlets, either temporarily or
permanently; canceling programs; firing journalists; and, more recently, using cyberattacks to
silence critical voices.
M5. Regulating content.

Table 4.17. Analytical group M5: steps of democratic erosion coded as major or significant
Number of Items Percentage
Total 64 100%

21 33%

14 22%
Examples:

Table 4.18. Analytical group M5: examples from the dataset (major)
Year Group Step Country
4 5
Causing further alarm, the Parliament approved amendments to the Antiterror
Law in June that allow for imprisoning journalists for up to three years for
the dissemination of statements and propaganda by terrorist organizations
[…]
T
93

6 5
The penal code was also amended in December to include a broader category
of government officials covered by desacato (disrespect) provisions and to
increase dramatically the criminal penalties for slander and libel, charges
often employed by government officials to intimidate private media
journalists.
V
7 5
In December the National Assembly approved a new criminal code, which
includes provisions that could affect freedom of expression and become tools
for the persecution of journalists and government rivals. Examples include
limits on social protests that cause discord; broad privacy protections that
proscribe the dissemination of unauthorized personal information, with no
exception for issues of public interest; and the criminalization of creating
“economic panic” by disseminating false information about the economy [...]
E
Again, the dataset justifies the existence of this analytical category, with 14 of the 64 steps of
erosion listed here classified as major and 21 more as significant (adding up to more than half of
the cases). Similarly to the previous group, the number of steps in Turkey (20), Venezuela (18)
and Ecuador (17) was significantly higher than that of their Hungarian (7) and Polish (2)
counterparts. Typical steps include: passing legislation to ban or impose content; and banning or
imposing actual content, building on previous legislation.
M6. Influencing content through financial pressure.

Table 4.19. Analytical group M6: steps of democratic erosion coded as major or significant
Number of Items Percentage
Total 54 100%

19 35%

8 15%
Examples:

Table 4.20. Analytical group M6: examples from the dataset (major or significant)
Year Group Step Country
2 6
Meanwhile, government agencies ended subscriptions to opposition-friendly
outlets, while state-owned companies canceled advertising contracts with
them.
P
7 6
A tax authority controlled by the Finance Ministry fined one of the country’s
major media companies, the Dogan Group, 826 million lira (US$537 million)
in February and 3.7 billion lira (US$2.4 billion) in September for purported
tax evasion. The Dogan Group has consistently reported on the ruling party’s
shortcomings and involvement in an Islamic charity scandal in 2008, and the
tax case was widely viewed as politicized.
T
94

13 6
CONATEL imposed a $2 million fine on Globovisión in October 2011 for its
coverage of riots in El Rodeo prison. If the fine is actually imposed, the
television station would likely go bankrupt.
V
The relevance of this analytical category is again supported by the numbers. It contains a total
of 54 steps of erosion, including 8 that were coded as major and 19 more classified as significant.
Together, they make up exactly half of the cases. There were at least ten steps in all countries
included in the case study, with the exception of Poland (1). Typical steps include: advertising
selectively, while pressuring private enterprises to follow suit; imposing heavy fines; and
distorting the media market in other ways.
M7. Influencing content through the threat of punishment.
Examples:

Table 4.21. Analytical group M7: examples from the dataset
Year Group Step Country
4 7
Erdogan himself continued to launch defamation suits against members of
the media, however, filing a total of 59 cases in 2006. Rights groups report
that the total number of defamation cases increased from 2005, along with
the fines issued as punishments. Convictions against journalists are made
much less frequently than are prosecutions, but trials are time-consuming and
expensive [...]
T
4 7
Meanwhile, journalist Freddy Aponte was sentenced to pay more than
$54,000 on top of another sentence of six months in prison from a different
criminal defamation proceeding.
E
10 7
In July, reporter Dayana Fernandez and photographer Luis Torres of the
newspaper La Verdad were assaulted by Maracaibo city officials while
covering waste disposal practices. The journalists’ equipment was
confiscated and they were allegedly beaten and held for more than two hours.
V
Consisting of a total of 225 steps of erosion, the relevance of this analytical category cannot
be in doubt. In fact, one may even question whether it makes theoretical sense to leave all these
steps in a single group. Clearly, a fine is not the same as incarceration, and both are different
from physical violence. This realization is the main reason steps in this category were not coded
according to weight. Yet, there is a rationale for keeping the group together: ultimately, all the
95

steps here serve the same purpose. They are meant to create an environment in which outlets and
journalists think twice before publishing content that is critical of the government, leading to
self-censorship.
The difference between the two Central European countries and the other three has been a
recurring theme in this section. It is even more noticeable in the context of this category, which
includes 127 steps from Turkey, 46 from Venezuela and 44 from Ecuador, but only 6 steps from
Hungary and 2 from Poland.
M8. Building a pro-government media empire.

Table 4.22. Analytical group M8: steps of democratic erosion coded as major or significant
Number of Items Percentage
Total 28 100%

6 21%

9 32%
Examples:

Table 4.23. Analytical group M8: examples from the dataset (major)
Year Group Step Country
4 8
In another instance of editorial pressure, the government ordered the
transformation of the El Telégrafo newspaper from a public to a
progovernment publication, over the strong protests of the editorial staff. El
Telégrafo had been a privately owned paper until three years before, when
debts and legal troubles led the government to take it over. After protesting
Correa’s plans for the newspaper, the editor-in-chief was forced to tender his
resignation in March. Later 20 editors, columnists, and reporters resigned [...]
E
7 8
Meanwhile, progovernment media proliferated both through acquisitions and
the establishment of new outlets. In October 2016, Opimus Press bought the
publisher Mediaworks, which had recently merged with Pannon Lapok
Társasága, which controlled numerous regional newspapers. The business
weekly Figyelő, was acquired by Mária Schmidt, a government ally. [...]
Opimus Press, though its 2016 acquisitions, at year’s end controlled about
half of newspaper printing capacities in Hungary.
H
11 8
In one of the most flagrant examples of the use of economic leverage to
shape media ownership, wiretap recordings leaked in December 2013
indicated that the government dictated which holding companies would
purchase the Sabah-ATV media group in exchange for a contract to build
Istanbul’s multibillion-dollar third airport.
T
96

This analytical category features 28 steps of erosion, including 9 major and 6 significant ones.
Accordingly, it would be difficult to argue that the existence of the group is not justified. The
individual numbers for the countries under consideration are similar (between 6 and 9), with
again the exception of Poland. Typical steps include buying or creating outlets with government
backing; and seizing them, though the latter only happened in Turkey.
Other steps of erosion. In order to assess the second hypothesis (there are no other relevant
categories), the steps that do not fit into any of the above groups need to be considered. The
question is whether they constitute a separate group of their own. It would also be problematic if
there were too many such steps, since that would mean that the classification introduced in the
theory section has little descriptive value. The following table summarizes all the steps coded as
0:

Table 4.24. Steps of erosion not classified into any of analytical groups M1-8
Year Group Step Country
1 0
In July, Fidesz removed the article from Hungary’s constitution which had
previously banned information monopolies.
H
4 0
Another set of controversial amendments, passed in May, included a ban on
political advertising in independent media outlets during campaign seasons,
restricting advertising to state-owned outlets. Although the Constitutional
Court rejected the legislation, Fidesz used its parliamentary supermajority to
bypass the court and insert the amendment directly into the constitution in
March.
H
7 0
According to an investigation published by the 444.hu news outlet in
September 2016, officials from an internal security agency the previous
December had sought to compel a journalist to cooperate with them; they
confronted the with journalist information about his private life, claiming that
an unidentified but harmful actor had disclosed it and they would assist him
only if he agreed to report to them.
H
7 0
There are no reported restrictions on internet access, but the government
monitors social-media content for alleged insults against Correa and other
government officials. Under the Communications Law, media outlets are
required to set up mechanisms for commenters to register their personal data
or create systems to delete offensive comments; otherwise, outlets can be
held legally responsible for insults or other perceived offenses contained in
reader comments. The law also prohibits outlets from using information
disseminated by anonymous sources via social media.
E
97

8 0
A June 2013 presidential decree granted the government new powers to
intervene in the operations of civil society organizations, including the
authority to forcibly dissolve them; observers saw the decree as a potential
threat to media watchdog groups, among others. In January 2014,
SUPERCOM used the decree to assume administrative supervision of
Fundamedios and request extensive financial records from the group.
E
10 0
In October, public television and radio outlets provided live coverage of a
campaign launch event for Moreno, the presidential candidate of Correa’s
ruling Alianza PAIS party, at which Correa had given an address. Public
outlets failed to offer similar coverage for other candidates.
E
The six steps listed above clearly have little in common: they cover areas as diverse as the
legal approach to information monopolies and the biased coverage of elections. In fact, one
would be hard pressed to find meaningful similarities between any two of them. As a result, there
is no thematic core that could serve as a basis for a new analytical category. Moreover, as 6 out
of 533 is barely more than one percent, there is no theoretical need to add more groups: the
existing eight are sufficient to describe the erosion of democracy.
Conclusion. Both hypotheses derived from the theory were confirmed by the dataset,
providing support for the theory itself. Furthermore, while analyzing the various groups,
observations were made beyond the scope of the two hypotheses. For four groups – M1 (public
media control), M4 (eliminating critical outlets), M5 (regulating content), and M7 (threat of
punishment) – the analysis uncovered a marked difference between the behavior of the Central
European governments and that of the others. For two of them – M4 and M7 – there was also a
significant difference between the two Latin American countries and Turkey. Some of the
possible reasons for these differences have been mentioned previously. Further research is
needed to provide a more comprehensive explanation. In any case, it seems reasonable to
conclude that the erosion of democracy might happen in more than one way. However, that does
not mean there cannot be general patterns. In fact, the other four analytical categories registered
no variations in government behavior.
98

4.3.3. Sequence of Steps
In this section, I analyze the timing of the steps of erosion in the eight analytical categories. The
purpose of this exercise is to test hypotheses H3 and H4, i.e. to decide whether analytical groups
tend to occur in specific stages of democratic erosion, and whether the process has more or less
distinct phases; or, in other words, whether there are patterns in the timing of the steps across
cases. To this end, steps in each category were visualized in the following way: every step was
represented by a square colored according to its weight, placed in the column corresponding to
the (relative) year in which it took place. For clarity’s sake, squares standing for steps with a
greater weight were grouped to the bottom. Two figures are provided for each analytical
category: one in which steps are displayed by country; and an aggregate.
Some caveats are in order. While Freedom House is a highly regarded source, some steps of
erosion may be missing from the dataset. Moreover, steps are only available on a yearly basis.
2016 is the last year covered by the Freedom of the Press Reports, meaning that for Hungary,
there is only data up to Year 7. For Turkey and Venezuela, Year 14 is the end point; for Ecuador,
it is Year 10; and for Poland, it is Year 2. Going beyond the Freedom of the Press reports in later
iterations of this work will hopefully help to fine-tune the analysis.
M1. Taking control of the public media.
Expected pattern: steps immediately after Eroder comes to power
The aggregate graph makes it clear that steps in this category tend to take place in the very
early stages of democratic erosion. Indeed, all major and significant steps happen in the first two
years of the new government. Out of the 16 steps, only 7 take place after Year 2, and most of
them have little or no direct impact on media freedom. As mentioned previously, these
observations are limited to Hungary and Poland, since there are only two steps in this group from
Turkey and Venezuela (they all occur in later years) and none from Ecuador.
99

Figure 4.3. Analytical group M1: temporal distribution of the steps of erosion

Conclusion: pattern as expected, but steps mostly for Hungary and Poland
M2. Taking control of the media authority.
Expected pattern: steps immediately after Eroder comes to power
While theoretical considerations suggested that steps in this category should take place in the
early stages of erosion, the aggregate graph paints a more nuanced picture. It has in fact two
main clusters: one in the first two years of the new government, as predicted; and a slightly
smaller one much later, in Years 12 to 14. To make matters even less clear-cut, there are a couple
of major and significant steps between these two periods. These three features of the graph are
the result of three very different patterns of behavior. The first cluster is made up almost
exclusively of steps from Hungary and Poland, while those in the second one occur in Turkey
and Venezuela. Ecuador is responsible for the steps in-between.

Figure 4.4. Analytical group M2: temporal distribution of the steps of erosion

H
P
T
V
E
Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14
H
P
T
V
E
Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14
100

Conclusion: pattern as expected for Hungary and Poland; not for Turkey and Venezuela
M3. Limiting access to information.
Expected pattern: possibly some steps early on, but most of them later

Figure 4.5. Analytical group M3: temporal distribution of the steps of erosion

Even though it would be incorrect to suggest that there are no steps of erosion in Years 1 to 5,
they only amount to 5 out of 26, or 19 percent, of all cases (including 3 out of 13, or 23 percent,
of major and significant steps). There is a peak in the number of cases in Years 6 and 7, after
which their number stays relatively constant over the years. The Hungarian pattern is again
different from the Turkish and Venezuelan examples. After a few early cases, Hungary peaks in
Years 6 to 7. For both Turkey and Venezuela, an early step in Year 6 is followed by a couple of
cases in Years 10 to 14. Ecuador seems to resemble Hungary, though, with only four cases, it is
impossible to draw a firm conclusion.
Conclusion: pattern as expected; timing different in Hungary/Ecuador and Turkey/Venezuela
M4. Eliminating critical outlets.
Expected pattern: no steps right after Eroder comes to power; upward trend in later years

H
P
T
V
E
Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14
101

Figure 4.6. Analytical group M4: temporal distribution of the steps of erosion

The graph shows barely any steps in Year 1, but there is a peak in Year 2, which is still early
in the process of democratic erosion. In Hungary, 3 out of 8 steps (38 percent) take place in the
first three years of the second Orbán government, including half of the more significant steps. On
the other hand, there are a consistently high number of cases from Year 5, with a clear upward
trend, driven chiefly by Turkey and Venezuela. As mentioned previously, there are fewer cases
in Hungary than in the Latin American countries (though the proportion of major and significant
steps is higher in Hungary, suggesting that the Orbán government might be more strategic in this
area), and Turkey constitutes a tier of its own.
Conclusion: pattern as expected for Turkey and Venezuela; not for Hungary
M5. Regulating content.
Expected pattern: possibly some steps early on, but most of them later

H
P
T
V
E
Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14
102

Figure 4.7. Analytical group M5: temporal distribution of the steps of erosion

The aggregate figure does not show any distinctive patterns, but, again, the picture is different
for individual countries. In Venezuela, only 2 cases out of 18 (11 percent) occur in the first five
years of democratic erosion. That includes 18 percent of significant and major steps. The
corresponding numbers are 25 percent and 20 percent for Turkey; and 24 percent and 29 percent
for Ecuador (Years 1 to 4). Consequently, most steps do take place in later stages of the process.
In contrast, steps in Hungary cover Years 1 to 5, though there is only one major step among
them. Interestingly, both Turkey and Venezuela have two main clusters of cases, though their
relative timing and size differs.
Conclusion: pattern as expected for Turkey, Venezuela and Ecuador; not for Hungary
M6. Influencing content through financial pressure.
Expected pattern: no steps right after Eroder comes to power; upward trend in later years
The aggregate graph seems to confirm prior expectations: there is only one step in Year 1, and
a large majority of the cases occur after Year 3 of democratic erosion. However, Hungary is
H
P
T
V
E
Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14
103

again an outlier: it records 4 out of its 10 cases (40 percent) in the first two years of the Orbán
government, and, notably, the single Polish case takes place as early as Year 2. In Turkey and
Venezuela, there are indeed no steps of erosion in the early years (in fact, Turkey has a single –
albeit major – case until Year 11), but no upward trend can be observed later in the process. Only
Ecuador seems to display the anticipated pattern.

Figure 4.8. Analytical group M6: temporal distribution of the steps of erosion

Conclusion: as expected for Ecuador; partly so for Turkey and Venezuela; not for Hungary
M7. Influencing content through the threat of punishment.
Expected pattern: most steps in later stages of democratic erosion
As evidenced by the aggregate graph, there are relatively few steps of democratic erosion in
Years 1 and 2, and Year 3 is also well below the average of later years (11 vs. 18). In Venezuela,
there are no steps at all in the first two years, which is especially notable in light of later numbers
(46 in total). In Turkey, Years 1 and 2 only account for 7 out of 127 (6 percent) of the cases;
whereas in Ecuador, the corresponding numbers are 3 out of 44 (7 percent). In Hungary, there
are practically no such steps, but of those few, only 1 out 8 (13 percent) takes place in the first
two years. Moreover, there is a clear upward trend in Turkey, with a sharp increase in Year 11.
H
P
T
V
E
Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14
104

(The mean for Years 1-5 is 5.2; for Years 6-10, it is 7.4; and for Years 11-14, it is 16.) The same
trend is less conspicuous, but present in Venezuela. (The mean for Years 1-5 is 1.8; for Years 6-
10, it is 3.6; and for Years 11-14, it is 4.75.)

Figure 4.9. Analytical group M7: temporal distribution of the steps of erosion

Conclusion: pattern as expected in all countries
H
P
T
V
E
Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14
105

M8. Building a pro-government media empire.
Expected pattern: most steps in later stages of democratic erosion

Figure 4.10. Analytical group M8: temporal distribution of the steps of erosion

As shown by the aggregate graph, there are no steps of note before Year 4. In fact, 21 of the
28 cases (75 percent) occur in or after Year 6, including 13 of the 15 major or significant steps
(87 percent). In Turkey, there are no steps at all before Year 11, and in Venezuela, only 2 out of
6 cases (33 percent) take place earlier. However, it is worth noting that there are relatively few
major or significant steps in this category from the two Latin American countries (two each),
suggesting that Presidents Chávez and Correa preferred other strategies.
Conclusion: pattern as expected in all countries
Assessment of the data. The following table summarizes the findings of this section:
Table 4.25. Timing of the steps of erosion: theory vs. data for each analytical group
Group Expected pattern Other pattern
M1 Taking control of the public media H, P T, V, E
M2 Taking control of the media authority H, P T, V, E
M3 Limiting access to information H, T, V, E
M4 Eliminating critical outlets T, V, (E) H
M5 Regulating content T, V, E H
M6 Infl. content through financial pressure (T), (V), E H
M7 Infl. content through the threat of punish. H, T, V, E
M8 Building a pro-government media empire H, T, V, E
According to the data, democratic erosion broadly follows the patterns anticipated in Part 2 in
all four countries for analytical categories M3, M7 and M8. When it comes to the rest of the
H
P
T
V
E
Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Year 1 2 3 4 5 6 7 8 9 10 11 12 13 14
106

groups, it is typically Hungary that stands apart from the other cases (together with Poland, when
there is enough data to draw conclusions about the latter), both in adhering to the pattern (Groups
M1 and M2) and in deviating from it (Groups M4, M5 and M6). Taken together, these results
confirm hypothesis H3, i.e. show that steps in at least some of the analytical groups tend to occur
in specific stages of democratic erosion. The process therefore does display patterns in the timing
of the different types of steps across all affected countries.
However, two caveats are in order. Most importantly, some of the similarities are only true to
a certain extent. For example, while steps that limit access to information (M3) and contribute to
the creation of a pro-government media empire (M8) did happen later in the process of
democratic erosion in all four countries, they happened, relatively speaking, much later in
Turkey and Venezuela than in Hungary and Ecuador. Also, while Hungary did follow the
anticipated pattern for Group M7 steps (threat of punishment), there were significantly fewer
such steps than in the other countries. Second, democratic erosion is still ongoing in several of
these countries, and new steps may change some of the conclusions of this chapter. For instance,
in the unlikely event that the Orbán government decides to restructure the public media once
again, the timing of Group M1 steps will need to be reexamined.
Overall, while democratic erosion is clearly not a uniform process, the data analysis has
confirmed the existence of broad general patterns. Considering all the differences between
affected countries (see above), this is a noteworthy result. Furthermore, the data reveal striking
similarities between the timing of certain types of steps in Turkey and Venezuela. This is
especially true for Group M1 (public media control), M2 (media authority control), M3 (limiting
access to information) and M4 (eliminating critical outlets) steps; and to a lesser degree for
Group M5 (regulating content) and M7 (threat of punishment) steps. Further research is needed
107

to make sense of this observation; yet it suggests that, even if democratic erosion is not a uniform
process, there may be groups of more or less similar cases.
Since steps in most analytical categories do not seem to follow a general temporal pattern
across all affected countries, it is not possible to identify distinct phases of democratic erosion. In
other words, the data analysis did not find confirmation for hypothesis H4. There is one country
in which the process so far had three more or less distinct phases: Hungary.

Figure 4.11. Temporal distribution of the steps of erosion (Hungary)

Years 1 and 2 constitute the first phase; Years 3 to 5 the second one; and Years 6 and 7 make
up the third phase. Group M1 (public media control) and M2 (media authority control) steps take
place almost exclusively in the first phase; Group M5 (regulating content) and M6 (financial
pressure) steps in the first two phases; while Group M3 (limiting access to information), M7
(threat of punishment) and M8 (building a pro-government media empire) steps predominantly in
the third phase. Only Group 4 (eliminating critical outlets) steps follow no distinct pattern. These
results reflect in many ways – though not without notable differences – the predictions made in
Part 2 (see Table 2.3). While identifying phases for a single country has little practical value, if
more cases are similar to Hungary (Poland is a promising candidate), hypothesis H4 will need to
be reconsidered.
Y 1 2 3 4 5 6 7 8
1
2
3
4
5
6
7
108

Similar figures are provided in Appendix 2.2 for the other four countries. The only
observation worthy of note is that in both Turkey and Venezuela, democratic erosion seems to
intensify significantly after Year 10, considering the total number of steps as well as that of
major and significant cases. No phases are apparent for Ecuador; whereas for Poland, there is
only two years’ worth of data, so few conclusions can be drawn at this point.

4.4. Conclusion
Even though literature on democratic erosion has been steadily growing in recent years, and
some of the scholarly interest has centered on the role that the media play in the process, the
results have two important shortcomings. First, the list of strategies that governments employ to
undermine media freedom is not comprehensive. Some of the works do endeavor to provide a
more in-depth description of the process, but they do not go beyond individual cases. Second, the
presentation of these strategies has not been systematic enough, i.e. most works merely list them
with at best a rudimentary attempt at a more organized approach. As a consequence of these
shortcomings, research so far has not been able to offer more general conclusions about the
changes that affect the media in eroding democracies. The goal of this chapter was to address
these two issues.
Based on the understanding that governments undermine media freedom by obstructing the
free flow of political information, I assessed where obstacles might be put, and used that insight
to identify eight analytical categories that should, together, contain most steps of democratic
erosion. Next, I considered how soon after their rise to power governments may employ certain
tools, and derived a series of assumptions about the timing of the different types of steps. I noted
that, in case of clear temporal patterns, democratic erosion may be found to happen in distinct
109

phases, which would be a very consequential result. I used an in-depth case study of five
countries built on Freedom House data to test these hypotheses.
The dataset confirmed that the eight analytical groups are all necessary and, together, do
contain the overwhelming majority of steps. Clear temporal patterns were also identified, though
only steps in three of the groups followed the anticipated pattern in all countries. The existence
of distinct phases of erosion was not supported by data, except for Hungary.
The findings of this chapter deepen our understanding of democratic erosion, and have the
potential to shape future research in several ways. The process consists of hundreds of steps,
which may hinder conceptual work. The existence of analytical groups – grounded in theory and
small in number, yet containing most steps – changes that. Also, the existing framework may
direct scholarly attention to certain areas, and possibly help to discover new steps. Overall, the
analytical categories contribute to a more conscious study of democratic erosion.
Democratic erosion is thought of a single phenomenon, and yet it affects countries that differ
in many ways (some of which have been touched upon in this chapter). The question is, should
we expect to observe more or less the same process across all cases, or not? The tentative
conclusion drawn in this chapter is that democratic erosion is certainly not a uniform process, but
the notable similarities between certain countries (mainly Hungary and Poland; and Turkey and
Venezuela) suggest that there may be larger groups of cases. The notion that there might be
varieties of democratic erosion is entirely missing from the literature; and, if recognized, might
reframe research. If groups of cases are indeed identified, the possibility of the process having
distinct phases may need to be reexamined.
Finally, literature so far has not considered the various tools governments may employ to
carry out steps of democratic erosion, and how their availability (in time and otherwise) might
110

shape the process. Doing so could open up further avenues for theory generation, and facilitate
the comparison of various strategies, as well as cases.

111

Appendix 4.1. Descriptive Statistics (Individual Countries)

Table 4.26. Descriptive statistics: analytical categories (individual countries)
Hungary Poland Turkey Venezuela Ecuador Total
Cat. No. % No. % No. % No. % No. % No. %
1 9 14 3 25 2 1 2 2 0 0 16 3
2 5 8 2 17 4 2 3 3 5 4 19 4
3 9 14 1 8 7 3 5 4 4 3 26 5
4 7 11 1 8 44 20 24 21 19 16 95 18
5 7 11 2 17 20 9 18 16 17 14 64 12
6 10 16 1 8 13 6 12 10 18 15 54 10
7 6 10 2 17 127 57 46 40 44 37 225 42
8 7 11 0 0 6 3 6 5 9 8 28 5
0 3 5 0 0 0 0 0 0 3 3 6 1
Total 63 100 12 100 223 100 116 100 119 100 533 100

Table 4.27. Descriptive statistics: weight (individual countries)
Hungary Poland Turkey Venezuela Ecuador Total
Wt. No. % No. % No. % No. % No. % No. %

6 10 2 17 127 57 46 40 44 37 225 42

5 8 0 0 4 2 18 16 12 10 36 7

20 31 2 17 43 19 17 15 27 23 107 20

18 29 7 58 25 11 22 19 29 24 100 19

14 22 1 8 24 11 13 11 7 6 59 11
Total 63 100 12 100 223 100 116 100 119 100 533 100

112

Appendix 4.2. Temporal Distribution of the Steps of Erosion (Individual
Countries)

Figure 4.12. Temporal distribution of the steps of erosion (Turkey)

Figure 4.13. Temporal distribution of the steps of erosion (Venezuela)

Y 1 2 3 4 5 6 7 8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Y 1 2 3 4 5 6 7 8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
113

Figure 4.14. Temporal distribution of the steps of erosion (Ecuador)

Figure 4.15. Temporal distribution of the steps of erosion (Poland)

Y 1 2 3 4 5 6 7 8
1
2
3
4
5
6
7
8
9
10
Y 1 2 3 4 5 6 7 8
1
2
114

Chapter 5
The Warning Signs: Explaining Vulnerability to Democratic Erosion

5.1. Introduction
The aim of the previous chapters was to present a descriptive account of democratic erosion,
with a special focus on one case (Hungary in Chapter 3) and one area affected by erosion (the
media in Chapter 4). While description is a necessary first step, the ultimate aim of academic
research is to provide a thorough understanding of the phenomenon. Accordingly, a range of
theories will be needed to explain various aspects of the process, from its inception to its
successful conclusion and aftermath. An important question is why democratic erosion starts in
certain countries, but not in others. In other words, what are the variables that increase a
country’s vulnerability to erosion? The purpose of this chapter is to address this question.
The importance of understanding democratic erosion, and in particular the circumstances that
lead to it, goes well beyond the world of academia. If the variables that play a key role at the start
of the process are recognized, vulnerable countries may be identified and erosion prevented or
reversed before lasting damage has been done. Due to the nature of the phenomenon, the more
time the autocratic leadership had to implement its plans, the harder it becomes to stop the
process. In fact, we know of very few countries in which erosion has been reversed, though the
relative novelty of the phenomenon may in part account for that.
This chapter is organized as follows. In Part 2, I outline a framework in which future theories
of democratic erosion may be placed. While not a concrete theory in itself, such a framework
may help to structure thinking about the early stages of the phenomenon. In addition, I identify
115

the variable that most consistently comes up in the academic literature on the origins of
democratic erosion. In Part 3, I introduce a new dataset based on a variety of high-quality
sources, and use it to test the hypothesis outlined in the theory section. The results are
summarized and assessed in the concluding part.

5.2. Vulnerability to Democratic Erosion: a Theoretical Framework
Since democratic erosion is a highly complex phenomenon, pinpointing all the variables that
increase a country’s vulnerability to backsliding goes beyond the scope of this chapter. Instead, I
present a theoretical framework that yields a number of key questions a comprehensive theory
focusing on the initial stages of the process will have to answer.
Considering that democratic erosion is a way for an autocratic leadership (the Eroder) to
monopolize power in a country, a logical starting point is to consider what happens to the main
depositories of power: the three branches of government. Obviously, democratic erosion will not
happen unless a political actor willing to orchestrate it comes to power. In other words, the
Eroder taking control of the executive branch is a necessary condition for democratic erosion to
begin. The same can be said about the legislative branch, albeit with some reservations. Since
specific legislation and loyal appointees are needed for democratic erosion to proceed, and in
democracies, parliaments are usually responsible for both, the Eroder will have a hard time
carrying out its plans in the face of a hostile legislature. Of course, executives may rule by decree
and have the power to appoint certain public servants, and legislatures may be circumvented
altogether. The 1999 Constitutional National Assembly in Venezuela served this purpose. All
things considered, though, it is difficult to imagine erosion to progress beyond a few initial steps
unless the Eroder controls the legislature.
116

The importance of dominating the judicial branch is more dependent on the exact powers the
judiciary has in a country, from the range of issues it can take up to the practical impact of its
rulings. However, in democracies, the judiciary is supposed to act as a check and a balance on
the other branches, thus it is likely to pose an obstacle to democratic erosion at least in certain
areas. Moreover, there are countries – like the United States – where the judiciary is strong
enough to altogether stop the process. In such places, neutering it is of paramount importance to
the Eroder. In practice, one would expect the Eroder to gain control of the legislative branch
either at the same time as or soon after being elected, and undermine the effectiveness of the
judiciary some time later, but still early in the process.
Are there any other (more or less) necessary conditions for democratic erosion to be self-
sustaining? Once the Eroder dominates all three branches of the government, there are two
potential sources of pressure that may prevent it from pursuing its agenda. The first is public
opinion. As mentioned in previous chapters, the goal of the Eroder is to create circumstances in
which its electoral victory is practically a foregone conclusion. Angering the public goes directly
against this goal. For this reason, political and societal actors that credibly express or amplify
criticism – such as the civil society and the media – need to be reined in. This, however, is less of
a priority for the Eroder, especially in the early years, when the following elections are still far
away in time. Furthermore, once in control of the three branches of government, the Eroder
usually has very effective tools against such actors. In other words, the presence of factors that
reduce the initial strength of these political and societal actors may help, but are not necessary
for the successful start of democratic erosion.
The other potential source of pressure is external. In theory, a foreign power may obstruct
parts the Eroder’s agenda. Indeed, one would assume that an organization like the European
117

Union would stand in the way of attempted democratic erosion, at least in its member states. Yet,
it is precisely the failure of the EU to deal with offenders such as Hungary and Poland that shows
how difficult it is to slow down the process from abroad. Again, variables that decrease foreign
actors’ leverage may help the Eroder, but are certainly not necessary.
Overall, a comprehensive theory of the initial stages of democratic erosion will have to
answer the following questions:
1. What are the variables that increase the likelihood that the Eroder will gain control of the
executive branch? Since elections almost always have a victor, the focus of this question needs
to be on the Eroder itself: a political actor ready to undermine democracy. Such actors are rarely
part of the established elite; instead, they position themselves against the old status quo, and
typically campaign on a platform of radical change. Ironically, they often claim to be the true
democrats – champions of the people, standing in opposition to a corrupted, elitist system. There
have always been populist outsiders in established democracies. They usually have no hope of
gaining power. Why, then, is there a surge in demand for their radical ideas? Disillusionment
with the elite is surely a part of the answer.
2. What are the variables that increase the likelihood that the Eroder will gain control of the
legislative branch? In parliamentary systems, executive power is held by the parties that control
the legislature, thus the Eroder can automatically count on parliamentary support, at least
initially. Similarly, in countries where executive and legislative elections take place
simultaneously, the factors that help the Eroder to power may also deliver it the legislature.
Variables that may matter include the type of legislature, since a unicameral parliament is, in
theory, easier to control than one with a strong second chamber. Whether the Eroder commands
118

a majority, or even a supermajority, in the legislature may also depend on how proportional or
majoritarian the electoral system is.
3. What are the variables that increase the likelihood that the Eroder will gain control of the
judicial branch? The judiciary may be neutered in two ways. First, the range of issues it can take
up may be narrowed, or the impact of its rulings limited. In this case, the judiciary may still wish
to act as a check on the Eroder, but loses its ability to do so effectively. Second, the judicial
branch may be packed with loyalists who do not even attempt to oppose the Eroder’s agenda.
Both options are on the table if the Eroder commands a large enough majority in the parliament.
Structural factors at play include the procedure with which the constitution can be amended;
whether Supreme Court justices are appointed for life; etc.
Answering the following two questions is not necessary to understand the initial stages of
democratic erosion, but may nevertheless be instructive:
4. What are the variables that decrease the likelihood that the Eroder will have to change
course due to pressure from public opinion? The Eroder’s ability to shape public opinion as it
sees fit is challenged if there are political or societal actors that (a) have sufficient knowledge
about the government’s performance; (b) are able and willing to criticize it; and (c) have the
ability and resources to reach enough voters. Variables that determine the strength and
proactivity of these actors, as well as the average citizen’s propensity to listen to them, are of
obvious interest here.
5. What are the variables that decrease the likelihood that the Eroder will have to change
course due to pressure from foreign powers? A country may depend on a foreign power for a
number of reasons. Military occupation allowed the United States to shape Japanese politics after
World War Two. Nowadays, such direct interference is rare, especially in democracies.
119

Membership in – or the wish to join – an international organization may compel a county to
adhere to the organization’s values. For example, the Central and Easter European members of
the European Union had to fulfill the so-called Copenhagen criteria before their accession to the
bloc. Economic dependence may also leave countries susceptible to foreign influence.
In the overview of the literature (Chapter 2), a number of variables were proposed to play a
role in the onset of democratic erosion. Most of these variables function as answers to one or
more of the above questions. For example, the size of the legislative majority (Ginsburg and Huq
2018; Kaufman and Haggard 2019) determines control of the legislative branch (Question 2);
while the crises of the political system and the economy (Coppedge 2005; Erdmann 2011; De la
Torre 2017; Foa and Mounk 2017; Gauna 2017; Power 2019) contribute to the Eroder’s rise to
power (Question 1).
Polarization is especially interesting, since it is theorized to affect both initial and later stages
of democratic erosion. It generates support for illiberal actors (McCoy, Rahman, and Somer
2018; Kaufman and Haggard 2019; Haggard and Kaufman 2021), while also weakening norms
of tolerance and self-restraint (Levitsky and Ziblatt 2018; Kaufman and Haggard 2019). People’s
willingness to trade off democratic principles for partisan interests increases with polarization
(McCoy, Rahman, and Somer 2018; Svolik 2019, 2020). Since polarization seems to have
garnered more interest in recent years than any other variable in the context of democratic
erosion, it is the obvious candidate for quantitative testing. Accordingly, Part 3 will assess the
following hypothesis:
H1: Greater polarization increases the likelihood of democratic erosion.

120

5.3. Data Analysis
While the focus of this section is on the relationship between polarization and the onset of
democratic erosion, it seems judicious to put the variable into broader context by establishing
whether it has an effect on democracy itself (its quality). That initial part of the section is
followed by the actual testing of H1, which first requires the definition of the universe of
observations and the identification of the cases of democratic erosion.
5.3.1. Polarization and the Quality of Democracy
The question is whether a change in the degree of polarization (independent variable) causes a
change in the quality of democracy (dependent variable). The following variables will be used in
the analysis:
Dependent variable. I use the Liberal Democracy Index from the Varieties of Democracy (V-
Dem) dataset to represent the quality of democracy. Data is available for most countries, going
back to 1789 for some. Since my dissertation centers on the post-communist world, I only
consider the years since 1989. LDI is set on an interval scale, with values ranging from 0 (lowest
possible quality) to 1 (highest), meaning that its interpretation is straightforward.
Independent variables. Literature (e.g. Mason 2015) distinguishes between two types of
polarization: issue polarization, brought about by conflicting issue positions (e.g. pro-choice vs.
pro-life) and identity polarization, which is driven by partisan identity (e.g. Democrat vs.
Republican). I use V-Dem’s Polarization of society to operationalize the former, while Political
polarization stands for the latter. Both measures are originally ordinal, but have been converted
to interval by the measurement model. For Polarization of society, smaller values represent
greater polarization; for Political polarization, the opposite is true.
121

Control variables. I include two economic variables as controls. The affluence of a country
and the levels of income inequality have both been theorized to affect the quality of democracy. I
use World Bank’s GDP/capita and Gini coefficient values to this end. In addition, I test whether
the size of the ruling party’s majority in the legislature has an effect on the quality of democracy.
V-Dem’s Lower chamber election seat share won by largest party (or Seat) provides data for
election years. I work with the assumption that these values do not change significantly until the
subsequent elections.

Table 5.1. List of variables used in the analysis (linear regression)
Variable Source Values
Liberal Democracy Index Varieties of Democracy 0 to 1
Polarization (issue) Varieties of Democracy -2.666 to 4.041
Polarization (identity) Varieties of Democracy -3.527 to 3.816
GDP/capita World Bank 95 to 118,824
Gini coefficient World Bank 0 to 100
Seat Varieties of Democracy 0 to 100
Method. Linear regression is used, since the dependent variable is continuous.
Results. The results are summarized in the following table:

Table 5.2. Linear regression results

Our main variable of interest, polarization, is significant across all models at a significance
level of 0.01, with one exception, where it is significant at p = 0.05. The direction of the effect is
also as expected: the higher polarization is, the lower the quality of democracy. (Issue
Model 1a Model 1b Model 2a Model 2b Model 3a Model 3b
Polarization (issue) 0.0372758*** 0.0105322** 0.0147351***
Polarization (identity) -0.0851699*** -.0381961*** -0.0340186***
GDP/capita 0.00000834*** 0.00000948*** 0.00000757*** 0.00000867***
Gini coefficient 0.0023823*** 0.0032105*** 0.0030064*** 0.0032947***
Seat -0.0030439*** -0.0026371***
Constant 0.4013926*** 0.3649963*** 0.2997374*** 0.2344566*** 0.4323473*** 0.3733408***
Adjusted R2 0.0339 0.1726 0.4137 0.471 0.4675 0.5153
N 3560 4245 1267 1282 1245 1254
Significant at a significance level of * p = 0.1; ** p = 0.05; *** p = 0.01
122

polarization has a positive coefficient, since greater values of the measure stand for lower
polarization.) The adjusted R
2
value of Model 1b is high (0.17), which means that identity
polarization does play a significant role in explaining LDI’s variance. Indeed, a one-point
increase in polarization results in an 8.5 percent decrease of the LDI value.
Interestingly, as shown both by the respective R
2
scores and by the coefficients, identity
polarization seems to have a much greater effect on the quality of democracy than issue
polarization. This implies that disagreement on policy – and thus, for example, the growing
ideological distance between Republican and Democratic senators in the United States – is less
of a threat to democracy than the tribal approach to politics. It needs to be mentioned, though,
that there is substantial correlation between the two variables (r = -0.5737).
The two economic variables are also significant in Models 2 and 3 at a significance level of
0.01. The coefficients, however, suggest a limited impact, especially in the case of GDP/capita.
Surprisingly, the coefficients for Gini are positive, meaning that greater Gini values, and thus
greater income inequality, increase the quality of democracy. Seat is significant, too (p = 0.01),
and has a moderate impact on LDI: a 10 percent increase in the seat share won by largest party
results in a 3 percent (Model 3a) or a 2.6 percent (Model 3b) drop in the quality of democracy.
However, later research will need to establish whether there truly is causation between the two
variables, as well as its direction. (Countries with a low LDI score are autocracies, where the
ruling party tends to dominate the legislature.)
Overall, as evidenced by the adjusted R
2
values, the four variables explain 47 percent (Model
3a) or 52 percent (Model 3b) of the dependent variable’s variance. Therefore, while they do not
fully explain changes in the LDI, these models are certainly promising.

123

5.3.2. Polarization and the Onset of Democratic Erosion
The previous section has confirmed that polarization does indeed affect the quality of
democracy. With that, it is time to consider this chapter’s original question: does greater
polarization (independent variable) increase the likelihood of democratic erosion (dependent
variable)? Since the following analysis requires a more in-depth look at countries, the universe of
observations is limited to the former EU 28 (the 27 current member states of the European Union
and the United Kingdom) as well as 18 Latin American countries. This set of countries
represents a great variety in size, population, wealth, democratic past, etc., and thus seems to
provide a good basis for quantitative analysis. Moreover, there are known cases of democratic
erosion in both regions. The choice of countries was also motivated by convenience: public
opinion research on relevant issues is available for both regions in the form of Eurobarometer
and Latinobarómetro publications.
Cases of democratic erosion. As mentioned previously, there is no generally accepted way of
operationalizing democratic erosion and, consequently, no definitive list of cases. In order to
proceed, we need to decide for each of the 28+18 countries included in the analysis whether
erosion has taken place there since 1989, and if so, when the process has started. The rules that
inform these decisions need to be based on the characteristics of the phenomenon:
(1) By definition, the regime needs to be democratic at the starting point of the process.
(2) As erosion progresses, the level of democracy needs to decrease substantially. There are
small fluctuations in democratic quality even in stable democracies.
(3) The decrease needs to be gradual; otherwise, the process would be closer to collapse than
to erosion.
124

With these principles in mind, the identification of positive cases goes as follows. I consider
the LDI scores for all 28+18 countries for the years 1989 to 2019. Democratic erosion is
assumed to take place in a sequence of years if the starting value is at least 0.5 (Characteristic 1);
the drop is at least 0.15 (Characteristic 2); and the decrease has taken place over at least two
years (Characteristic 3). Some level of flexibility is allowed to accommodate borderline cases,
especially if the process is likely to be ongoing. These rules yield the following positive cases:

Table 5.3. Cases of democratic erosion
Country Start Year -1 Start Value Last Year Last Value LDI Drop
Bolivia 2004 0.525 2013 0.381 0.144
Brazil 2013 0.790 2019 (ongoing) 0.508 0.282
Bulgaria 2016 0.602 2019 (ongoing) 0.488 0.114
Croatia 2015 0.686 2019 (ongoing) 0.545 0.141
Ecuador 2006 0.494 2016 0.300 0.194
Hungary 2009 0.760 2019 (ongoing) 0.402 0.358
Poland 2014 0.824 2019 (ongoing) 0.500 0.324
Romania 2016 0.595 2019 (ongoing) 0.434 0.161
Venezuela 1998 0.609 2019 (ongoing) 0.086 0.523
Of course, the numbers alone may not tell the whole story. A closer look is needed at each of
the cases to exclude those where the drop in LDI score is not due to democratic erosion. In seven
of these countries, the decline has started following the election of a new government, which
stayed in power for the whole duration of the process. This does point to democratic erosion. In
Romania, however, there have been five different prime ministers since the 2016 elections, the
first three from the center-left Social Democratic Party and the last two from the conservative
National Liberal Party. Such volatility is certainly not conducive to democratic erosion.
Brazil is a more complicated case. The tenure of President Dilma Rousseff (2011-2016)
witnessed no significant decline in democratic quality. Most of the drop in LDI score came in
2016, in connection with her impeachment and removal from office (0.132). Under her
successor, Michel Temer, there was again not much erosion to speak of. However, it would be
125

premature to completely dismiss Brazil as a case in light of the election of far-right populist Jair
Bolsonaro in October 2018. With him in office, the county’s LDI score dropped by 0.069 in
2019. While this value is far from the 0.15 threshold, the events since 2013 show patterns clearly
reminiscent of democratic erosion: first a succession of crises that undermine faith in the old elite
(Hunter and Power 2019), then the election of a (potential) Eroder. More data is needed for the
final verdict, but for now, Brazil will be classified as a positive case.
Dependent variable. Democratic erosion may conceivably start years after the Eroder has
gained power. In fact, eroding democracy may not even be his original intent, but rather an
answer to changing circumstances. For example, President Recep Tayyip Erdoğan of Turkey
began his premiership in 2003 as a relative moderate before his authoritarian turn years later.
However, both theoretical considerations and the cases identified above suggest that the start of
democratic erosion is, for the most part, closely connected to the election of the Eroder.
Consequently, democratic erosion cannot begin in non-election years, and thus the set of
observations needs to be limited to election years. The key question for each observation is
whether an Eroder has been elected that year or not. Accordingly, I use a binary dependent
variable with a value of 1 if an Eroder was elected and 0 otherwise. Once an election is assigned
the value of 1, observations after that one need to be discarded for the same country. (While
there might be multiple spells of erosion, and thus multiple starts, in practice, erosion is either
still ongoing in these countries, or if not, the quality of democracy has not recovered yet.) For
countries with parliamentary or premier-presidential systems (EU 28 minus Cyprus), legislative
elections are taken into account; whereas for presidential systems (Latin America plus Cyprus),
presidential elections are of interest.

126

Table 5.4. Year the Eroder was elected
Country Year
Brazil 2018
Bolivia 2005
Bulgaria 2017
Croatia 2016
Ecuador 2006
Hungary 2010
Poland 2015
Venezuela 1998
If the LDI value is lower than 0.5 for an election year, that year by definition could not bring
about the start of democratic erosion. The same is true for elections that happened in 2019, since
the drop in the quality of democracy has to take place over at least two years to count. As a
consequence, these observations need to be discarded.
Independent variables. Like in the previous section, I use V-Dem’s Polarization of society
and Political polarization to measure the two types of polarization (issue and identity).
Control variables. I keep World Bank’s Gini coefficient as a proxy for the level of income
inequality. Since voters are likely to be influenced by the country’s current performance and not
its affluence in general, I replace GDP/capita with the year-on-year GDP/capita change. V-
Dem’s Lower chamber election seat share won by largest party (or Seat) proved to be a
promising variable. However, in order to reflect the assumption that control of the legislature is
typically needed for the Eroder to pursue his agenda, I transform Seat into a binary variable,
Seat50. The value of Seat50 is 1 if the largest party has majority in the legislature, (Seat ≥ 50);
and 0 otherwise. I include one more control to check whether dissatisfaction with the political
elite does indeed play a role in the Eroder’s rise to power: Trust in parliament. Both
Eurobarometer and Latinobarómetro surveys collect data on the respondents’ trust in the
legislature. The values range from total mistrust (0) to absolute trust (100).
127

These variables do not show significant pairwise correlation: the correlation coefficient with
the highest absolute value (Gini and Trust in parliament) is -0.3529.

Table 5.5. List of variables used in the analysis (logistic regression)
Variable Source Values
Start of erosion Own compilation 0, 1
Polarization (issue) Varieties of Democracy -2.564 to 2.768
Polarization (identity) Varieties of Democracy -3.527 to 2.123
GDP/capita change World Bank -8.541 to 12.919
Gini coefficient World Bank 0 to 100
Seat50 Varieties of Democracy 0 to 1
Trust in parliament Eurobarometer, Latinobarómetro 0 to 100
Method. Logistic regression (logit) is used, since the dependent variable is binary. The
analysis is intended to determine how the independent variables affect a probability: that of an
election being one that leads to democratic erosion.
Results. The results are summarized in the following table:

Table 5.6. Logistic regression results

As in the previous section, both types of polarization are significant across all models at a
significance level of 0.01, with one exception (Model 4b), where identity polarization is still
significant at p = 0.05. The direction of the effect is as predicted: greater polarization – i.e. lower
values of issue polarization and higher values of identity polarization – increases the likelihood
of the dependent variable being 1, or, in other words, of an Eroder coming to power and starting
Model 1a Model 1b Model 2a Model 2b Model 3a Model 3b Model 4a Model 4b
Polarization (issue) -1.713334*** -2.153957*** -2.040333*** -2.048717***
Polarization (identity) 1.209965*** 1.384044*** 1.327567*** 1.301407**
Trust in parliament -0.1008578** -0.0700968** -0.1249626** -0.0858511** -0.128031* -0.0751793*
Seat50 1.032194 1.058188 1.567197 0.9361016
GDP/capita change 0.2094524 -0.0273559
Gini coefficient 0.0606107 0.0466225
Constant -4.067569*** -3.042822*** -2.120744* -0.9336263 -1.868558* -0.9261287 -4.68955* -2.662044
Pseudo R2 0.3226 0.2163 0.4462 0.327 0.4587 0.3461 0.5045 0.3689
N 172 231 162 160 162 160 131 122
Significant at a significance level of * p = 0.1; ** p = 0.05; *** p = 0.01
128

democratic erosion. The results, therefore, confirm our hypothesis (H1). This time, both pseudo
R
2
scores and the respective coefficients suggest that issue, and not identity, polarization has a
greater catalytic effect on the onset of democratic erosion. Since the dependent variable is
different from the one in the previous section, the two sets of results are not necessarily
contradictory.
The control variables show a mixed picture. Seat50, as well as GDP/capita change and the
Gini coefficient seem to have no significant impact on the dependent variable. However, it is
worth noting that the number of observations is much lower than in the previous section: it is
between 122 (Model 4b) and 231 (Model 1b) out of a pool of 267 elections. Trust in parliament
is much more promising: it is significant at a significance level of 0.05 in Models 2a/b and 3a/b,
while significant at p = 0.1 in Models 4a and 4b. The sign of the coefficients is negative across
all models, implying, as one would expect, that greater trust in the legislature reduces the
likelihood of an Eroder being elected. Of course, a more in-depth assessment of the causality is
needed, as it is conceivable that the shock of a radical outsider coming to power may affect the
approval rating of the parliament.
Similarly, it is possible that instead of polarization helping to bring about the rise of the
Eroder, the opposite happens: with the anti-elite outsider being on the center stage and finally
beginning the implementation of his radical agenda, political divisions – issue as well as identity
– may deepen. In order to ascertain whether that is the case, I repeat the logistic regressions,
using this time the polarization values for the years before the elections (e.g. the 2005 values for
the Hungary 2006 observation). The results are as follows (with the differences in significance
marked in gray):

129

Table 5.7. Logistic regression results (Y-1 polarization)

While, predictably, the coefficients are somewhat different than before (they mostly have
smaller absolute values in Models 1/2/3/4a and both smaller and larger absolute values in Models
1/2/3/4b), the significance of the variables did not change. There is only one exception: in Model
2a, Trust in parliament is only significant at p = 0.1 instead of 0.05. Importantly, these results
suggest that it is greater polarization that increases the likelihood of democratic erosion, and not
the other way around – again, confirming our hypothesis.

5.4. Conclusion
Democratic erosion is possibly the greatest contemporary threat to democracies. With an
increasing number of cases around the world – including, according to some, the United States –
there is a growing body of research on this insidious phenomenon. Yet, research remains mostly
descriptive, with only a few attempts at going deeper. More comprehensive theories are needed
to truly understand the process, and thereby recognize the signs of democratic erosion and,
hopefully, prevent its onset.
A key question is why democratic erosion starts in certain countries, but not in others. In other
words, what are some of the variables that leave a country vulnerable to democratic erosion? The
aim is this chapter was to contribute to an eventual, comprehensive answer to this question in
Model 1a Model 1b Model 2a Model 2b Model 3a Model 3b Model 4a Model 4b
Y-1 Polariz. (issue) -1.353983*** -1.469537*** -1.387578*** -1.410894***
Y-1 Polariz. (identity) 1.29476*** 1.512792*** 1.467608*** 1.429475**
Trust in parliament -0.0765931* -0.0728562** -0.1002638** -0.0897945** -0.0981957* -0.0797923*
Seat50 1.166499 1.053437 1.466297 0.7947486
GDP/capita change 0.1503743 -0.0621028
Gini coefficient 0.0566425 0.0300563
Constant -3.469672*** -2.921708*** -1.643058* -0.752381 -1.499529 -0.7223741 -4.063587* -1.602619
Pseudo R2 0.2367 0.2175 0.3337 0.3349 0.3552 0.3534 0.4059 0.3743
N 167 230 160 160 160 160 129 122
Significant at a significance level of * p = 0.1; ** p = 0.05; *** p = 0.01
130

two ways. First, I introduced a theoretical framework that may help to structure thinking about
the early stages of the phenomenon. The main insight was that taking control of the executive
and legislative branches, and to a lesser extent the judiciary, is a necessary and sufficient
condition for the start of the process.
Second, I tested the variable that is most often mentioned in conjunction with democratic
erosion: polarization. The two types of polarization – issue and identity – both proved to have a
significant effect on the quality of democracy as well as on the likelihood of democratic erosion.
An additional takeaway of the quantitative analysis was the emergence of another variable of
interest: trust in the legislature. This finding is in line with the theoretical framework, which
hypothesized that disillusionment with the elite might play an important role in increasing the
probability that an Eroder would gain control of the executive branch.
Of course, polarization and trust in the legislature are just two of the potentially numerous
variables that leave a country more vulnerable to democratic erosion, and thus more research is
needed to gain a complete understanding of the initial stages of the phenomenon. The theoretical
framework outlined above provides a good starting point for conceptual work. As for the
empirical part, a major issue this chapter had to face was the paucity of observations. (The
overall number was 267, but many values were missing for the various variables.) In the future,
the universe of observations may be expanded by including countries outside of the EU and
Latin America, or by changing the way democratic erosion is operationalized.

131

Chapter 6
Conclusion

Democratic erosion, or backsliding, is the greatest contemporary threat to democracies. Not even
established democracies are impervious to this insidious phenomenon, as evidenced by the
presidency of Donald Trump in the United States. Consequently, academic interest in democratic
erosion has grown exponentially in recent years. Yet, in spite of steady progress, significant gaps
remain in this nascent literature: no attempt has been made to provide a comprehensive account
of all the building blocks – or steps – of the process, and it is unclear which of the steps occur in
all cases. Moreover, the description of democratic erosion is not systematic enough, which
hinders conceptual work. Efforts to explain various aspects of the phenomenon are still in their
early stages, not helped by the fact that the number of positive cases is very limited.
Accordingly, few theories have been proposed. The aim of this dissertation was to contribute to a
more in-depth description of democratic erosion and to a better understanding of the variables
that lead to its onset – both directly, by introducing and evaluating original datasets, and
indirectly, by offering conceptual tools for future research.
Chapter 3. A magnifying glass can reveal details that the naked eye may miss. With that in
mind, the idea in this chapter was to put democratic erosion under a sort of magnifying glass, so
as to identify individual steps of erosion. The in-depth case study of Hungary – based on the
evaluation of tens of thousands of news articles – was meant to serve that purpose. The result is
an original dataset that contains the steps that the Orbán government resorted to since its election
in 2010. The presentation of the steps was structured around five analytical groups, derived from
132

an examination of the factors that might influence a government’s chances of being reelected,
and should therefore be principal targets. Having a small number of categories instead of
hundreds of steps facilitates conceptual work. In this chapter, they were used to establish
hypotheses about the timing of the various types of steps.
The data confirmed that the five analytical groups are necessary and sufficient to describe the
process; and that expectations about their relative timing are also true. The Hungarian case seems
to support the hypothesis that democratic erosion has two phases: a first one, in which the Eroder
changes the institutional setup (e.g. the power of political actors that may check or criticize the
government is curbed); and a second one, in which the new setup is exploited (e.g. state
resources are misused; and there are outright abuses of power).
Chapter 4. The description of political phenomena cannot be based on a single case.
Accordingly, the aim of this chapter was to extend the study of democratic erosion to five
affected countries. To compensate for the increased breadth, the focus was narrowed to one
specific area: the media. The fourth estate is a natural choice, since its importance in modern
politics cannot be questioned, and, due to its size, it should be targeted by more steps than
smaller actors, like the judiciary. This time, the case studies were built on Freedom House’s
Freedom of the Press reports, which guaranteed methodological uniformity across cases. Like in
the previous chapter, analytical groups were defined for a more structured presentation of the
steps. The classification was based on the analysis of the flow of political information and the
obstacles the Eroder might put in its way. The eight analytical groups, as well as a consideration
of the types of tools available to governments to effectuate their strategies, made it again possible
to establish hypotheses about the sequence of the steps.
133

The data provided empirical support for the classification, and thus for the underlying
assumptions. The picture was mixed when it came to the relative timing of the steps: some
analytical groups showed clear temporal patterns, while others did not. The main takeaway was
that democratic erosion is certainly not a uniform process, but the notable similarities between
certain countries (mainly Hungary and Poland; and Turkey and Venezuela) suggest that there
may be larger groups of cases – and therefore a number of “varieties of erosion.”
Chapter 5. One of the most consequential questions concerning democratic erosion is why it
happens in certain countries, but not in others. Identifying the variables that increase a country’s
vulnerability to backsliding could have major practical implications. As such, a growing number
of variables have been proposed in recent years, but most of them lack empirical support. The
theoretical contribution of this chapter was a framework based on the idea that variables of
interest must, in one way or another, help the (would-be) Eroder gain control of the executive
branch, then the legislature, and finally the judiciary. In other words, the search for variables
may be structured around their point of attack.
The empirical part of the chapter was designed to test the most promising variable:
polarization. The data, compiled from Varieties of Democracy, World Bank, and
Eurobarometer/Latinobarómetro indicators, confirmed that both issue and identity polarization
have an adverse effect on the quality of democracy in general, and increase the likelihood of
democratic erosion. While more work is needed to determine the exact mechanism through
which polarization harms democracy, interest in the variable is clearly warranted. An additional
finding of the chapter was that one of the control variables, trust in the legislature, may also play
a role in the onset of democratic erosion.
134

In sum, the dissertation introduced original datasets that might be used for future research.
However, their real value lies less in the actual data, and more in the way they were created,
which is straightforward and easy to apply to more cases. Furthermore, the dissertation made a
number of conceptual contributions to the literature: the idea that individual steps should be the
unit of analysis; the utility of all-encompassing analytical categories; the notion that different
types of steps may follow each other in a given sequence; and the consideration of the tools the
government has at its disposal at various points in time, and how that affects the timing of the
steps. These ideas provide a bridge between description and explanation. They may be used as
they are, or refined further with little difficulty.
Research agenda. While the case study of Hungary was meant to be as in-depth as possible,
only articles covering Hungarian domestic politics were considered. Given more time, the search
for steps of erosion could be extended to other – even marginally relevant – sections of the news
site, such as economic news. Only then would we be able to conclude with reasonable certainty
that all steps have been accounted for, though questions around editorial choices and the
identification of borderline cases may remain a cause for concern. The ultimate goal would be to
have a similarly in-depth case study of all countries affected by democratic erosion, based on
(comparable) primary sources, and not Freedom House data, which, though reliable and
thorough, is too limited in scope to include all episodes of interest. As a result, a catalogue of all
the steps of erosion could be assembled, which would be a gold mine for future research, helping
to compare and contrast cases and build general theories.
Of course, one needs to bear in mind that democratic erosion is not a static phenomenon: as
the world – and especially technology – changes, so do potential tools and targets of erosion.
Social media (and the Internet itself) are relatively new inventions, and yet their impact on
135

democratic quality cannot be overstated. Consequently, it is imperative to use sources that
capture newer tools and targets when cataloguing the various steps of erosion. A weakness of
Freedom House reports is their lack of focus on social media. Future research will need to make
sure that social media, and new technologies in general, receive the attention they deserve in a
methodologically consistent manner.
The work on analytical groups and the sequence of steps would also benefit greatly from an
expanded set of observations. The key question here is how similar the various cases are to each
other. The literature has identified similarities; and Chapter 4 has revealed that there are
differences, too; but whether there are larger groups, as suggested by that chapter, and thus
“varieties of erosion,” remains to be seen. If there are, working out the typology would in itself
constitute a whole new strand of literature, with a major impact on theory generation.
Another area worthy of further exploration is the relationship between democratic erosion and
populism. It has been mentioned above that there are reasons to expect some sort of connection,
but also that the relationship may not be straightforward. Clarifying the nature of this connection
would have obvious benefits: not only is it important conceptually, but incorporating the
extensive body of literature on populism may also provide a considerable boost to the study of
democratic erosion. A key question is whether populist tactics are part of the story of erosion.
There are apparent similarities: Eroders tend to choose confrontation over consensus, deriving
their actions from the will of the majority; the regime is headed by a strong leader; and there is a
clear distinction between the insiders (“us”) and the outsiders (“them”) who, in one way or
another, threaten the in-group. The outsiders may be migrants, followers of a different religion,
etc. (right-wing populism); or the elite, the domestic allies of foreign capitalist powers, etc. (left-
136

wing populism). Since there are different types of populism, the relationship between populism
and democratic erosion may also vary.
In Chapter 5, testing polarization’s effect on the likelihood of democratic erosion with purely
quantitative methods was made difficult by the relative shortage of observations, and thus both
positive and negative cases. An obvious way forward is to include more countries from all
around the world, not just the European Union and Latin America – a choice made out of
necessity. Being able to capture the start of democratic erosion without limiting observations to
election years would also be a step in the right direction. Otherwise, the use of more qualitative
methods, such as process tracing, might also be considered.

137

References

Acemoglu, Daron and James A. Robinson. 2006. Economic Origins of Dictatorship and
Democracy. New York: Cambridge University Press.
Akser, Murat. 2018. “News Media Consolidation and Censorship in Turkey: From Liberal Ideals
to Corporatist Realities.” Mediterranean Quarterly 29(3): 78-97.
Akyuz, Kadir and Steve Hess. 2018. “Turkey Looks East: International Leverage and
Democratic Backsliding in a Hybrid Regime.” Mediterranean Quarterly 29(2): 1-26.
Alvares, Claudia and Peter Dahlgren. 2016. “Populism, extremism and media: Mapping an
uncertain terrain.” European Journal of Communication 31(1) 46-57.
Anria, Santiago. 2016. “More Inclusion, Less Liberalism in Bolivia.” Journal of Democracy
27(3): 99-108.
Aro, Jessikka. 2016. “The cyberspace war: propaganda and trolling as warfare tools.” European
View 15(1): 121-132.
Arutunyan, Anna. 2009. The Media in Russia. Maidenhead: Open University Press.
Bajomi-Lázár, Péter. 2013. “The Party Colonisation of the Media: The Case of Hungary.” East
European Politics and Societies 27(1): 69-89.
Bajomi-Lázár, Péter. 2017. “Particularistic and Universalistic Media Policies: Inequalities in the
Media in Hungary.” Javnost 24(2): 162-172.
Bakke, Elizabeth and Nick Sitter. 2020. “The EU’s Enfants Terribles: Democratic Backsliding in
Central Europe since 2010.” Perspectives on Politics First View: 1-16.
138

Bánkuti, Miklós, Gábor Halmai, and Kim Lane Scheppele. 2012. “Hungary’s Illiberal Turn:
Disabling the Constitution.” Journal of Democracy 23(3): 147-155.
Bátorfy, Attila and Ágnes Urbán. 2020. “State advertising as an instrument of transformation of
the media market in Hungary.” East European Politics 36(1): 44-65.
Bennett, W. Lance. 2016. News: The Politics of Illusion, 10
th
ed. Chicago, IL: University of
Chicago Press.
Bermeo, Nancy. 2016. “On Democratic Backsliding.” Journal of Democracy 27(1): 5-19.
Boix, Carles. 2003. Democracy and Redistribution. New York: Cambridge University Press.
Burke, Paul J. and Andrew Leigh. 2010. “Do Output Contractions Trigger Democratic Change?”
American Economic Journal: Macroeconomics 2(4): 124-157.
Buzogány, Aron. 2017. “Illiberal democracy in Hungary: authoritarian diffusion or domestic
causation?” Democratization 24(7): 1307-1325.
Castaldo, Antonino. 2020. “Back to Competitive Authoritarianism? Democratic Backsliding in
Vučić’s Serbia.” Europe-Asia Studies 72(10): 1617-1638.
Central Intelligence Agency. 2020. The World Factbook 2020. Washington, DC. Accessed on
October 11, 2020.
https://www.cia.gov/library/publications/resources/the-world-factbook/index.html
Čepo, Dario. 2020. “Structural weaknesses and the role of the dominant political party:
democratic backsliding in Croatia since EU accession.” Southeast European and Black Sea
Studies 20(1): 141-159.
Chadwick, Andrew. 2017. The Hybrid Media System: Politics and Power, 2
nd
ed. New York,
NY: Oxford University Press.
139

Conaghan, Catherine M. 2008. “Ecuador: Correa’s Plebiscitary Presidency.” Journal of
Democracy 19(2): 46-60.
Conaghan, Catherine M. 2016. “Ecuador Under Correa.” Journal of Democracy 27(3): 109-118.
Conaghan, Catherine M. 2017. “Contra-Associational Strategy in a Hybrid Regime: Ecuador,
2007-2015.” Bulletin of Latin American Research 36(4): 509-525.
Coppedge, Michael. 2005. “Explaining Democratic Deterioration in Venezuela through Nested
Inference.” In Frances Hagopian and Scott Mainwaring (Eds.). The Third Wave of
Democratization in Latin America (pp. 289-316). Cambridge: Cambridge University Press.
Coppedge, Michael. 2017. “Eroding Regimes: What, Where, and When?” Working paper.
Coppedge, Michael, John Gerring, Carl Henrik Knutsen, Staffan I. Lindberg, Jan Teorell, David
Altman, Michael Bernhard, M. Steven Fish, Adam Glynn, Allen Hicken, Anna Lührmann,
Kyle L. Marquardt, Kelly McMann, Pamela Paxton, Daniel Pemstein, Brigitte Seim, Rachel
Sigman, Svend-Erik Skaaning, Jeffrey Staton, Steven Wilson, Agnes Cornell, Lisa Gastaldi,
Haakon Gjerløw, Nina Ilchenko, Joshua Krusell, Laura Maxwell, Valeriya Mechkova, Juraj
Medzihorsky, Josefine Pernes, Johannes von Römer, Natalia Stepanova, Aksel Sundström,
Eitan Tzelgov, Yi-ting Wang, Tore Wig, and Daniel Ziblatt. 2019. “V-Dem Country-Year
Dataset v9” Varieties of Democracy (V-Dem) Project. https://doi.org/10.23696/vdemcy19
Corke, Susan, Andrew Finkel, David J. Kramer, Carla Anne Robbins, and Nate Schenkkanet.
February 3, 2014. “Democracy in Crisis: Corruption, Media, and Power in Turkey.” A
Freedom House Special Report.
Corporación Latinobarómetro. 2018. Informe 1995-2018. Santiago de Chile. Accessed on
October 11, 2020. https://www.latinobarometro.org/lat.jsp
Corrales, Javier. 2015. “Autocratic Legalism in Venezuela.” Journal of Democracy 26(2): 37-51.
140

Corrales, Javier and Michael Penfold-Becerra. 2007. “Venezuela: Crowding Out the
Opposition.” Journal of Democracy 18(2): 99-113.
Dahl, Robert A. 1971. Polyarchy. New Haven, CT: Yale University Press.
D’Alessio, Dave and Mike Allen. 2000. “Media Bias in Presidential Elections: A Meta-
Analysis.” Journal of Communication 50(4): 133-156.
Daly, Tom Gerald. 2020. “Understanding Multi-directional Democratic Decay: Lessons from the
Rise of Bolsonaro in Brazil.” Law & Ethics of Human Rights 14(2): 199-226.
Deibert, Ronald J., John G. Palfrey, Rafal Rohozinski, and Jonathan Zittrain (Eds.). 2010. Access
Controlled: The Shaping of Power, Rights, and Rule in Cyberspace. Cambridge, MA: MIT
Press.
de la Torre, Carlos. 2013. “Technocratic Populism in Ecuador.” Journal of Democracy 24(3):
33-46.
de la Torre, Carlos. 2017. “Hugo Chávez and the diffusion of Bolivarianism.” Democratization
24(7): 1271-1288.
de la Torre, Carlos and Andrés Ortiz Lemos. 2016. “Populist polarization and the slow death of
democracy in Ecuador.” Democratization 23(2): 221-241.
Diamond, Larry. 2010. “Liberation Technology.” Journal of Democracy 21(3): 69-83.
Diamond, Larry. 2015. “Facing Up to the Democratic Recession.” Journal of Democracy 26(1):
141-55.
Diamond, Larry. 2021. “Democratic regression in comparative perspective: scope, methods, and
causes.” Democratization 28(1): 22-42.
Dragomir, Marius. 2018. “Control the money, control the media: How government uses funding
to keep media in line.” Journalism 19(8): 1131-1148.
141

Duffy, Natalie. 2015. Internet Freedom in Vladimir Putin’s Russia: The Noose Tightens.
American Enterprise Institute. Accessed on February 10, 2021.
www.jstor.org/stable/resrep03199
Eberl, Jakob-Moritz, Hajo G. Boomgaarden, and Markus Wagner. 2017. “One Bias Fits All?
Three Types of Media Bias and Their Effects on Party Preferences.” Communication
Research 44(8): 1125-1148.
Egorov, Georgy, Sergei Guriev, and Konstantin Sonin. 2009. “Why Resource-Poor Dictators
Allow Freer Media: A Theory and Evidence from Panel Data.” American Political Science
Review 103(4): 645-668.
Engesser, Sven, Nicole Ernst, Frank Esser, and Florin Büchel. 2017. “Populism and social
media: how politicians spread a fragmented ideology.” Information, Communication &
Society, 20(8): 1109-1126.
Erdmann, Gero. 2011. “Decline of Democracy: Loss of Quality, Hybridisation and Breakdown
of Democracy.” In Gero Erdmann and Marianne Kneuer (Eds.). Regression of Democracy?
(pp. 21-58) Wiesbaden: VS Verlag fur Sozialwissenschaften GmbH.
Esen, Berk and Sebnem Gumuscu. 2016. “Rising Competitive Authoritarianism in Turkey.”
Third World Quarterly 37(9): 1581-1606.
European Commission. 1989 to 2019. Standard Eurobarometer 31 to 91. Eurobarometer
Surveys. Accessed on October 11, 2020.
https://ec.europa.eu/COMMFrontOffice/publicopinion/index.cfm
Fish, M. Stephen. 2002. “Islam and Authoritarianism.” World Politics 55(1): 4-37.
Foa, Roberto Stefan and Yascha Mounk. 2017. “The Signs of Deconsolidation.” Journal of
Democracy 28(1): 5-15.
142

Forelle, Michelle, Philip N. Howard, Andrés Monroy-Hernández, and Saiph Savage. 2015.
“Political Bots and the Manipulation of Public Opinion in Venezuela.” Accessed on February
11, 2021. http://dx.doi.org/10.2139/ssrn.2635800
Freedom House. 2019. “Freedom in the World.” Accessed on December 20, 2019.
https://freedomhouse.org/report/freedom-world/freedom-world-2017
Freedom House. 2019. “Freedom of the Press.” Accessed on December 20, 2019.
http://www.freedomhouse.org/report-types/freedom-press
Gandhi, Jennifer. 2019. “The Institutional Roots of Democratic Backsliding.” The Journal of
Politics 81(1): e11-e16.
Gauna, Anibal F. 2017. “Revisiting the Issue of Democratic Deterioration in Venezuela, 1974-
1998.” Journal of politics in Latin America 9(1): 33-58.
Gerbaudo, Paolo. 2018. “Social media and populism: an elective affinity?” Media, Culture &
Society 40(5): 745-753.
Gibler, Douglas M. and Kirk A. Randazzo. 2011. “Testing the Effects of Independent Judiciaries
on the Likelihood of Democratic Backsliding.” American Journal of Political Science 55(3):
696-709.
Ginsburg, Tom and Aziz Z. Huq. 2018a. “Democracy’s Near Misses.” Journal of Democracy
29(4): 16-30.
Ginsburg, Tom and Aziz Z. Huq. 2018b. How to save a constitutional democracy. Chicago;
London: The University of Chicago Press.
Griffen, Scott. 2020. “Hungary: a lesson in media control.” British Journalism Review 31(1): 57-
62.
143

Grzymala-Busse, Anna. 2019. “How Populists Rule: The Consequences for Democratic
Governance.” Polity 51(4): 707-717.
Gunitsky, Seva. 2015. “Corrupting the Cyber-Commons: Social Media as a Tool of Autocratic
Stability.” Perspectives on Politics 13(1): 42-54.
Haggard, Stephan and Robert Kaufman. 2021. Backsliding: Democratic Regress in the
Contemporary World. Cambridge: Cambridge University Press.
Hallin, Daniel C. and Paolo Mancini. 2004. Comparing Media Systems: Three Models of Media
and Politics. New York: Cambridge University Press.
Hallin, Daniel C. and Paolo Mancini (Eds.). 2012. Comparing Media Systems Beyond the
Western World. New York: Cambridge University Press.
He, Qinglian. 2008. The Fog of Censorship: Media Control in China. New York: Human Rights
in China.
Heller, Patrick. 2020. “The age of reaction: Retrenchment populism in India and Brazil.”
International Sociology 35(6): 590-609.
Hidalgo, Manuel. 2009. “Hugo Chávez’s ‘Petro-socialism.’” Journal of Democracy 20(2): 78-
92.
Hunter, Wendy and Timothy J. Power. 2019. “Bolsonaro and Brazil’s Illiberal Backlash.”
Journal of Democracy 30(1): 68-82.
Huntington, Samuel P. 1996. “Democracy for the Long Haul.” Journal of Democracy 7(2): 3-13.
Kaufman, Robert R. and Stephan Haggard. 2019. “Democratic Decline in the United States:
What Can We Learn from Middle-Income Backsliding?” Perspectives on Politics 17(2): 417-
432.
144

Kenny, Timothy and Peter Gross. 2008. “Journalism in Central Asia: A Victim of Politics,
Economics, and Widespread Self-censorship.” The International Journal of Press/Politics
13(4): 515-525.
Kerpel, Aron. 2015. “Pole and Hungarian Cousins Be? A Comparison of State Media Capture,
Ideological Narratives and Political Truth Monopolization in Hungary and Poland.” Slovo
29(1): 68-93.
Krastev, Ivan and Stephen Holmes. 2018. “Explaining Eastern Europe: Imitation and Its
Discontents.” Journal of Democracy 29(3): 117-128.
Lanskoy, Miriam and Dylan Myles-Primakoff. 2018. “Power and Plunder in Putin’s Russia.”
Journal of Democracy 29(1): 76-85.
Lanskoy, Miriam and Elspeth Suthers. 2013. “Outlawing the Opposition.” Journal of Democracy
24(3): 75-87.
Lee, Martin A. and Norman Solomon. 1991. Unreliable Sources: A Guide to Detecting Bias in
the News Media. Secaucus, NJ: Lyle Stuart.
Lehoucq, Fabrice. 2008. “Bolivia’s Constitutional Breakdown.” Journal of Democracy 19(4):
110-124.
Levitsky, Steven and Lucan Way. 2002. “The Rise of Competitive Authoritarianism.” Journal of
Democracy 13(2): 51-65.
Levitsky, Steven and Lucan Way. 2010. Competitive Authoritarianism: Hybrid Regimes and the
Cold War. Cambridge: Cambridge University Press.
Levitsky, Steven and Daniel Ziblatt. 2018. How democracies die. New York: Crown.
Linz, Juan J. 1978. The Breakdown of Democratic Regimes: Crisis, Breakdown, &
Reequilibration. Baltimore: The John Hopkins University Press.
145

Linz, Juan. 1990. “The Perils of Presidentialism.” Journal of Democracy 1(1): 51-69.
Lipman, Masha and Michael McFaul. 2001. “‘Managed Democracy’ in Russia: Putin and the
Press.” The International Journal of Press/Politics 6(3): 116-127.
Lippmann, Walter. 1922. Public Opinion. New York, NY: Harcourt, Brace and Company.
Lorch, Jasmin. 2021. “Elite capture, civil society and democratic backsliding in Bangladesh,
Thailand and the Philippines.” Democratization 28(1): 81-102.
Lueders, Hans and Ellen Lust. 2018. “Multiple Measurements, Elusive Agreement, and Unstable
Outcomes in the Study of Regime Change.” The Journal of Politics 80(2): 736-741.
Lührmann, Anna and Staffan I. Lindberg. 2019. “A third wave of autocratization is here: what is
new about it?” Democratization 26(7): 1095-1113.
Luo, Zhaotian and Adam Przeworski. 2019. “Democracy and Its Vulnerabilities: Dynamics of
Democratic Backsliding.” Working paper.
Lust, Ellen and David Waldner. 2013. “Parties, Polarization, and Democratic Development in the
Middle East.” Working paper.
Lust, Ellen and David Waldner. June 11, 2015. “Unwelcome Change: Understanding,
Evaluating, and Extending Theories of Democratic Backsliding.” United States Agency for
International Development.
Maeda, Ko. 2010. “Two Modes of Democratic Breakdown: A Competing Risks Analysis of
Democratic Durability.” The Journal of Politics 72(4): 1129-1143.
Mainwaring, Scott. 1993. “Presidentialism, Multipartism, and Democracy: The Difficult
Combination.” Comparative Political Studies 26(2): 198-228.
Mainwaring, Scott and Aníbal Pérez-Liñán. 2015. “Cross-Currents in Latin America.” Journal of
Democracy 26(1): 114-18.
146

Mason, Lilliana. 2015. “‘I Disrespectfully Agree’: The Differential Effects of Partisan Sorting on
Social and Issue Polarization.” American Journal of Political Science 59(1): 128-145.
Mazzuca, Sebastian. 2013. “The Rise of Rentier Populism.” Journal of Democracy 24(2): 108-
122.
McCoy, Jennifer, Tahmina Rahman, and Murat Somer. 2018. “Polarization and the Global Crisis
of Democracy: Common Patterns, Dynamics, and Pernicious Consequences for Democratic
Polities.” American Behavioral Scientist 62(1) 16-42.
Mickiewicz, Ellen. 2008. Television, Power, and the Public in Russia. New York: Cambridge
University Press.
Müftüler-Baç, Meltem and E. Fuat Keyman. 2012. “The Era of Dominant-Party Politics.”
Journal of Democracy 23(1): 85-99.
Mukherji, Rahul. 2020. “Covid vs. Democracy: India’s Illiberal Remedy.” Journal of Democracy
31(4): 91-105.
O’Donnell, Guillermo. 1988. “Challenges to Democratization in Brazil: A Threat of a Slow
Death.” World Policy Journal 5(2): 281-300.
O’Donnell, Guillermo. 1995. “Democracy’s Future: Do Economists Know Best?” Journal of
Democracy 6(1): 23-28.
Öniş, Ziya. 2016. “Turkey’s Two Elections: The AKP Comes Back.” Journal of Democracy
27(2): 141-154.
Över, Defne. 2021. “Democratic backsliding and the media: the convergence of news narratives
in Turkey.” Media, Culture & Society. Accessed on February 11, 2021.
https://doi.org/10.1177/0163443720975879
147

Pallin, Carolina Vendil. 2017. “Internet control through ownership: the case of Russia.” Post-
Soviet Affairs 33(1): 16-33.
Pemstein, Daniel, Kyle L. Marquardt, Eitan Tzelgov, Yi-ting Wang, Juraj Medzihorsky, Joshua
Krusell, Farhad Miri, and Johannes von Römer. 2019. “The V-Dem Measurement Model:
Latent Variable Analysis for Cross-National and Cross-Temporal Expert-Coded Data”. V-
Dem Working Paper No. 21. 4th edition. University of Gothenburg: Varieties of Democracy
Institute.
Pérez-Liñán, Aníbal, Nicolás Schmidt, and Daniela Vairo. 2019. “Presidential hegemony and
democratic backsliding in Latin America, 1925–2016.” Democratization 26(4): 606-625.
Plattner, Mark F. 2012. “Media and Democracy: The Long View.” Journal of Democracy 23(4):
62-73.
Postill, John. 2018. “Populism and social media: a global perspective.” Media, Culture & Society
40(5): 754-765.
Powell, G. Bingham. 1982. Contemporary Democracies: Participation, Stability, and Violence.
Cambridge, Mass.: Harvard University Press.
Przeworski, Adam. 2019. Crises of Democracy. Cambridge: Cambridge University Press.
Przybylski, Wojciech. 2018. “Explaining Eastern Europe: Can Poland’s Backsliding Be
Stopped?” Journal of Democracy 29(3): 52-64.
Puglisi, Riccardo and James M. Snyder, Jr. 2015. “Empirical Studies of Media Bias.” In Simon
P. Anderson, Joel Waldfogel, and David Strömberg (Eds.). Handbook of Media Economics
(pp. 647-667). Amsterdam: Elsevier/North-Holland.
Putnam, Robert. 1994. Making Democracy Work: Civic Traditions in Modern Italy. Princeton:
Princeton University Press.
148

Reynolds, Andrew. 2011. Designing Democracy in a Dangerous World. Oxford: Oxford
University Press.
Riaz, Ali. 2021. “The pathway of democratic backsliding in Bangladesh.” Democratization
28(1): 179-197.
Robertson, Graeme and Samuel Greene. 2017. “The Kremlin Emboldened: How Putin Wins
Support.” Journal of Democracy 28(4): 86-100.
Rodan, Gerry. 2004. Transparency and Authoritarian Rule in Southeast Asia. London:
RoutledgeCurzon.
Rogenhofer, Julius Maximilian and Ayala Panievsky. 2020. “Antidemocratic populism in power:
comparing Erdoğan’s Turkey with Modi’s India and Netanyahu’s Israel.” Democratization
27(8): 1394-1412.
Rupnik, Jacques. 2007. “From Democracy Fatigue to Populist Backlash.” Journal of Democracy
18(4): 17-25.
Rupnik. Jacques. 2018. “Explaining Eastern Europe: The Crisis of Liberalism.” Journal of
Democracy 29(3): 24-38.
Said-Hung, Elías and Francisco Segado-Boj. 2018. “Social Media Mobilization in Venezuela: A
Case Study.” Social and Economic Studies 67(4): 235-259.
Saka, Erkan. 2018. “Social Media in Turkey as a Space for Political Battles: AKTrolls and other
Politically motivated trolling.” Middle East Critique 27(2): 161-177.
Schedler, Andreas. 1998. “What is Democratic Consolidation?” Journal of Democracy 9(2): 91-
107.
149

Scheppele, Kim Lane. 2013. “The Rule of Law and the Frankenstate: Why Governance
Checklists Do Not Work.” Governance: An International Journal of Policy, Administration,
and Institutions 26(4): 559-562.
Seawright, Jason. 2012. Party-System Collapse: The Roots of Crisis in Peru and Venezuela.
Stanford: Stanford University Press.
Sestanovich, Stephen. 2007. “Putin’s Invented Opposition.” Journal of Democracy 18(2): 122-
124.
Shevtsova, Lilia. 2015. “Forward to the Past in Russia.” Journal of Democracy 26(2): 22-36.
Shirky, Clay. 2011. “The Political Power of Social Media: Technology, the Public Sphere, and
Political Change.” Foreign Affairs 90(1): 28-41.
Smyth, Regina and Irina Soboleva. 2014. “Looking beyond the economy: Pussy Riot and the
Kremlin’s voting coalition.” Post-Soviet Affairs 30(4): 257-275.
Somer, Murat and Jennifer McCoy. 2018. “Déjà vu? Polarization and Endangered Democracies
in the 21st Century.” American Behavioral Scientist 62(1): 3-15.
Stockmann, Daniela. 2013. Media Commercialization and Authoritarian Rule in China. New
York: Cambridge University Press.
Surowiec, Paweł and Václav Štětka. 2020. “Introduction: media and illiberal democracy in
Central and Eastern Europe.” East European Politics 36(1): 1-8.
Surowiec, Paweł, Magdalena Kania-Lundholm, and Małgorzata Winiarska-Brodowska. 2020.
“Towards illiberal conditioning? New politics of media regulations in Poland (2015–2018).”
East European Politics 36(1): 27-43.
Svolik, Milan W. 2015. “Which Democracies Will Last? Coups, Incumbent Takeovers, and the
Dynamic of Democratic Consolidation.” British Journal of Political Science 45(4): 715-738.
150

Svolik, Milan W. 2019. “Polarization versus Democracy.” Journal of Democracy 30(3): 20-32.
Svolik, Milan W. 2020. “When Polarization Trumps Civic Virtue: Partisan Conflict and the
Subversion of Democracy by Incumbents.” Quarterly Journal of Political Science 15(1): 3-
31.
Tang, Wenfang and Shanto Iyengar. 2011. “The Emerging Media System in China: Implications
for Regime Change.” Political Communication 28(3): 263-267.
Tolz, Vera and Yuri Teper. 2018. “Broadcasting agitainment: a new media strategy of Putin’s
third presidency.” Post-Soviet Affairs 34(4): 213-227.
Turam, Berna. 2012. “Are Rights and Liberties Safe?” Journal of Democracy 23(1): 109-118.
Vachudova, Milada Anna. 2020. “Ethnopopulism and democratic backsliding in Central
Europe.” East European Politics 36(3): 318-340.
van de Walle, Nicolas. 2003. “Presidentialism and Clientelism in Africa’s Emerging Party
Systems.” Journal of Modern African Studies 41(2): 297-321.
Varol, Ozan O. 2015. “Stealth Authoritarianism.” Iowa Law Review 100(4): 1673-1705.
Vartanova, Elena. 2012. “The Russian Media Model in the Context of Post-Soviet Dynamics.” In
Daniel C. Hallin, and Paolo Mancini (Eds.). Comparing Media Systems Beyond the Western
World (pp. 119-142). New York: Cambridge University Press.
Waldner, David and Ellen Lust. 2018. “Unwelcome Change: Coming to Terms with Democratic
Backsliding.” Annual Review of Political Science 21(1): 93-113.
Walker, Christopher and Robert W. Orttung. 2014. “Breaking the News: The Role of State-Run
Media.” Journal of Democracy 25(1): 71-85.
Weeks, Brian E. and Homero Gil de Zúñiga. 2021. “What’s Next? Six Observations for the
Future of Political Misinformation Research.” American Behavioral Scientist 65(2): 277-289.
151

Weyland, Kurt. 2013. “The Threat from the Populist Left.” Journal of Democracy 24(3): 18-32.
Wolfsfeld, Gadi, Elad Segev, and Tamir Sheafer. 2013. “Social Media and the Arab Spring:
Politics Comes First.” The International Journal of Press/Politics 18(2): 115-137.
Yesil, Bilge. 2016. Media in New Turkey: The Origins of an Authoritarian Neoliberal State.
Urbana: University of Illinois Press.
Yıldırım, Kerem, Lemi Baruh, and Ali Çarkoğlu. 2020. “Dynamics of Campaign Reporting and
Press-Party Parallelism: Rise of Competitive Authoritarianism and the Media System in
Turkey.” Political Communication Volume ahead-of-print: 1-24.
Zhuravskaya, Ekaterina, Maria Petrova, and Ruben Enikolopov. 2020. “Political Effects of the
Internet and Social Media.” Annual Review of Economics 12(1): 415-38.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A theory of status-quo terrorism: democracies in conflict and their proclivity to outsource repression

PDF

The fabled fourth estate: media freedom, democracy and human rights

PDF

The political psychology of democracy promotion, or how democracies democratically promote democracy promotion

PDF

Power to the masses? The rise of direct democracy in Latin America

PDF

Identity salience and political violence: centering social identity in the study of intrastate conflict

PDF

Politics behind appointments: policy, patronage, and party systems in the Andes

PDF

Securitizing the democratic peace: democratic identity and its role in the construction of threat

PDF

The partial Good Samaritan states: China and Japan in the international relations of autocracy and democracy

PDF

Challenges from below: the origins of status competitions in world politics

PDF

Volatile nationalism: nationalism and its influence on maritime disputes

PDF

Institutional variance of the democratic peace, 1816-2002: electoral, executive, and federal institutions in time and space

PDF

The strategy of firms towards political elites and consumers in polarized western democracies

PDF

Economic policy-making in a changing environment: the politics of success in Peru vis-à-vis China

PDF

Feeling polarized: emotional partisan polarization and its reinforcement through campaign advertising

PDF

Developing countries in the digital era: state and business interactions for industrial upgrading

PDF

The changing state of tolerance and tolerance in changing states

PDF

Emergency-war machine: national crisis, democratic governance, and the historical construction of the American state

PDF

Habits and policy: the socio-cognitive foundations of foreign policy stability

PDF

The politics of transformation: mass media and the Northern Ireland peace process

PDF

Are you as old as you feel? A quantitative and qualitative analysis of leader age and foreign policy decision-making

Asset Metadata

Creator Somogyi, David (author)

Core Title The erosion of democracy in the post–Cold War world: understanding the phenomenon and its causes

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Political Science and International Relations

Degree Conferral Date 2021-08

Publication Date 07/18/2021

Defense Date 03/18/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag democratic backsliding,Ecuador,erosion of democracy,Hungary,media freedom,OAI-PMH Harvest,poland,polarization,steps of erosion,trust in the legislature,Turkey,Venezuela

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Munck, Gerardo (committee chair), Crigler, Ann (committee member), Cull, Nick (committee member), James, Patrick (committee member)

Creator Email dsomogyi@usc.edu,somogyidavid@hotmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15609013

Unique identifier UC15609013

Legacy Identifier etd-SomogyiDav-9785

Document Type Dissertation

Format application/pdf (imt)

Rights Somogyi, David

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu