STATE REPRESSION: ECONOMIC CONDITIONS AND ESCAPING PUNISHMENT

By

Nathaniel Charles Smith

A DISSERTATION

Submitted to

Michigan State University

in partial fulﬁllment of the requirements

for the degree of

Political Science – Doctor of Philosophy

2020

ABSTRACT

STATE REPRESSION: ECONOMIC CONDITIONS AND ESCAPING PUNISHMENT

By

Nathaniel Charles Smith

The use of repression by governments around the world serves as a key strategy for regimes

to maintain their power. When faced with a threat, states often choose to repress their

citizens as a means of quelling dissent. Recently, the literature has shown that states will also

use repression preemptively when facing a potential future threat. This dissertation provides

insights into this use of preemptive repression and analyzes instances where states will likely

use this strategy. Speciﬁcally, chapters two and three look at the economic conditions of

the country as a way to explain state use of preemptive repression. Chapter two determines

that the eﬀect youth bulges have on increased repression can be moderated by the economic

condition of the state in question. Speciﬁcally, economically better oﬀ states will fail to

use preemptive repression in the presence of a large youth cohort. Chapter three similarly

looks at the economic condition of the state and ﬁnds that countries experiencing negative

economic shocks will increase their preemptive repression as a means to insulate themselves

from the unrest these shocks often bring. Beyond the idea of preemptive repression, states

also desire to avoid the negative consequences often attached to their abuse of human rights

like economic sanctions and a possible withdrawal of foreign aid. Thus, the ﬁnal empirical

chapter of the dissertation explores how foreign states can use public relations (PR) ﬁrms

based in Washington D.C. to avoid the negative consequences of their use of repression

without having to modify their behavior. Taken together, this dissertation provides valuable

insights into state behavior and the use of preemptive repression. In addition, this project

explores an, as yet, underdeveloped way in which foreign states can escape being punished

for their human rights abuses.

To my wife and daughters for motivating me and to my parents for their support and

encouragement.

iii

ACKNOWLEDGMENTS

The dissertation serves as the culmination of the doctoral degree. Being such, I could not

have succeeded in such an endeavor without the help of countless mentors, peers, colleagues,

family, and friends. A true acknowledgement section must begin with those closest to me.

First, my wife, who has put up with many more years of educational attainment than any

wife should have to deal with. Throughout our twelve plus years of marriage, I have been

actively pursuing one degree or another. Completion of this degree will mark the ﬁrst time in

our relationship that I will not be in school — and for that I am thankful. Next, my children.

They too have ever known their father as a student. Their wholehearted belief that I am

the smartest person in the world helped spur me on when challenges came. Finally to my

family. First my siblings, who steadfastly acted as if my paper topics where interesting and

thereby representing a true act of love. And, of course, my parents for believing that I could

do this while simultaneously checking-in constantly to make sure I was. I truly could not

have done this without all of you!

To my mentors — your care and kindness helped bring out the best in me. Dr. Andrew

Long for suggesting I apply to Michigan State in an attempt to "get in on the way up". Dr.

Jakana Thomas for listening to my half-formed ideas and oﬀering feedback. Dr. Eric Juenke

for inspiring me to be a better teacher. Dr. Benjamin Appel for consistently helping me

become a better writer/researcher. Lastly, the Michigan State Political science department

faculty broadly who helped me in so many ways.

To my fellow graduate students whose support and encouragement proved essential to

my perseverance and (eventual) success. Alex Mardon for encouraging me to continue on

and providing me with the opportunity to talk about things other than political science.

Caleb Lucas for entertaining my methods questions with genuine interest. Jessica, Lora,

Elizabeth, Jamil, Marty, and many others for passing down valuable wisdom. Steve and

iv

Nick for providing me with a space at the end of the hall to come and chat. And to all my

other colleagues, you all made the MSU political Science department a special place.

v

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CHAPTER 1

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1 State Repression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.1 Contribution to the Literature . . . . . . . . . . . . . . . . . . . . . .

CHAPTER 2 THE CONDITIONAL RELATIONSHIP OF YOUTH BULGES AND

STATE REPRESSION . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1
2.2 Youth Bulges and the Potential for Violence . . . . . . . . . . . . . . . . . .
2.3 Youth Bulges and State Repression . . . . . . . . . . . . . . . . . . . . . . .
2.4 Youth, Repression, and Economic Opportunity . . . . . . . . . . . . . . . . .
2.5 Data and Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.1 Dependent Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.2
. . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.3 Control Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.5.4 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.6 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Independent Variables

CHAPTER 3 ECONOMIC SHOCKS AND STATE REPRESSION . . . . . . . . .
3.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Economic Conditions of the State and Repression . . . . . . . . . . . . . . .
3.3 Economic Shocks, Violence, and State Repression . . . . . . . . . . . . . . .
3.3.1 Example: Zimbabwe . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.1 Dependent Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.2
. . . . . . . . . . . . . . . . . . . . . . . . . .
3.4.3 Control Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.5 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Independent Variables

3.6.1 Do Economic Shocks Have an Interactive Eﬀect with Economic De-

. . . . . . . . . . . . . . . . . . . . . . . . .
3.7 Conclusion and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . .

velopment and Dissent?

CHAPTER 4 ESCAPING PUNISHMENT: PUBLIC RELATIONS FIRMS AND

STATE REPRESSION . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.1

x

1
1
4

9
9
12
15
17
23
23
25
26
27
28
36
37

38
38
39
40
44
49
49
49
52
54
55

57
64

66
66

vi

Sanctions

69
4.2 Lobbying . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
4.3 Punishing Human Rights . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
75
4.3.1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
77
4.3.2 The Allocation of Foreign Aid . . . . . . . . . . . . . . . . . . . . . .
79
4.4 Lobbying and the Decision to Punish . . . . . . . . . . . . . . . . . . . . . .
84
4.4.1 Examples: Azerbaijan and Egypt . . . . . . . . . . . . . . . . . . . .
87
4.5 Data and Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . .
87
4.5.1 Dependent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . .
90
4.5.2
. . . . . . . . . . . . . . . . . . . . . . . . . .
92
4.5.3 Control Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
4.5.4 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
97
4.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.6.1 Lobbying, Human Rights, and Sanctions . . . . . . . . . . . . . . . .
97
4.6.2 Lobbying, Human Rights, and Foreign Aid . . . . . . . . . . . . . . . 102
4.6.3 Lobbying, Human Rights, and Bypass Aid . . . . . . . . . . . . . . . 109
4.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

Independent Variables

CHAPTER 5 CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
CHAPTER 2 APPENDIX . . . . . . . . . . . . . . . . . . . 121
CHAPTER 3 APPENDIX . . . . . . . . . . . . . . . . . . . 132
CHAPTER 4 APPENDIX . . . . . . . . . . . . . . . . . . . 139

APPENDIX A
APPENDIX B
APPENDIX C

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

vii

LIST OF TABLES

Table 2.1: Summary Statistics: Dependent Variables and Explanatory Variables . . .

Table 2.2: Summary Statistics: Control Variables . . . . . . . . . . . . . . . . . . . .

Table 2.3: Moderating eﬀect of GDP growth on the relationship between youth

bulges and state repression . . . . . . . . . . . . . . . . . . . . . . . . . .

Table 2.4: Moderating eﬀect of youth unemployment on the relationship between

youth bulges and state repression . . . . . . . . . . . . . . . . . . . . . . .

Table 3.1: Correlations Between economic shock measures . . . . . . . . . . . . . . .

Table 3.2: Economic shock measures: Honduras - 1990 . . . . . . . . . . . . . . . . .

Table 3.3: Eﬀect of economic shocks on repression . . . . . . . . . . . . . . . . . . .

Table 3.4:

Interactive eﬀect of GDP on economic shocks and repression . . . . . . .

Table 3.5:

Interactive Eﬀect of Dissent on Economic Shocks and Repression . . . . .

Table 4.1: Top Ten Countries by Number Lobbying Contracts . . . . . . . . . . . . .

Table 4.2: Summary Statistics for Sanctions Analysis Sample . . . . . . . . . . . . .

Table 4.3: Summary Statistics for Foreign Aid Analysis

. . . . . . . . . . . . . . . .

Table 4.4: Summary Statistics for Bypass Aid Analysis Sample . . . . . . . . . . . .

Table 4.5: Eﬀect of lobbying and human rights on U.S. sanctions: 1978-2012 . . . . .

26

27

29

33

51

52

56

58

61

70

93

94

95

98

Table 4.6: Moderating eﬀect of lobbying on human rights and the allocation of

military aid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

Table 4.7: Moderating eﬀect of lobbying on human rights and the allocation of

economic aid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

Table 4.8: Eﬀect of lobbying and human rights on bypass aid: 2005-2009 . . . . . . . 110

Table A.1: Eﬀect of Youth Bulges on Repression by GDP Growth . . . . . . . . . . . 124

viii

Table A.2: Eﬀect of Youth Bulges on Repression Controlling for Education Spending

128

Table B.1: Eﬀect of Economic Shocks on Repression Using Currency Crises Measure

132

Table B.2: Eﬀect of Economic Shocks on Repression Using Xpolity Measure . . . . . 133

Table B.3: Eﬀect of Economic Shocks on Repression Controlling for Education Spending135

Table C.1: Eﬀect of Lobbying and Human Rights on U.S. Sanctions: 1978-2012 . . . 142

Table C.2: Foreign Aid: Using Latent Repression and Empowerment Rights Indica-

tors of Human Rights Abuses

. . . . . . . . . . . . . . . . . . . . . . . . 144

Table C.3: Eﬀect of Lobbying and Human Rights on U.S. Sanctions: 1978-2012 . . . 145

ix

LIST OF FIGURES

Figure 2.1: Percent Youth by Year . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13

Figure 2.2: Cameroon: Level of Repression by Year - Red Lines Indicate the Youth

Unemployment Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 2.3: Predicted level of repression by GDP growth across the range of per-
cent youth. Results from table 2.3, model 2. Bars indicate 90 percent
conﬁdence intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 2.4: Predicted level of repression by GDP growth across the range of per-
cent youth. Results from table 2.3, model 4. Bars indicate 90 percent
conﬁdence intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 2.5: Predicted level of repression by GDP growth across the range of per-
cent youth. Results from table 2.3, model 6. Bars indicate 90 percent
conﬁdence intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 2.6: Predicted level of repression by youth unemployment across the range
of percent youth. Results from table 2.4, model 2. Bars indicate 90
percent conﬁdence intervals

. . . . . . . . . . . . . . . . . . . . . . . . .

Figure 2.7: Predicted level of repression by youth unemployment across the range
of percent youth. Results from table 2.4, model 4. Bars indicate 90
percent conﬁdence intervals

. . . . . . . . . . . . . . . . . . . . . . . . .

Figure 2.8: Predicted level of repression by youth unemployment across the range
of percent youth. Results from table 2.4, model 6. Bars indicate 90
percent conﬁdence intervals

. . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.1: Level of repression by year: Zimbabwe. Shaded area represents period

of hyper-inﬂation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.2: Number of economic shocks by year. Note. Economic shocks measured

as a change in GDP growth of 5% or more. . . . . . . . . . . . . . . . . .

Figure 3.3: Predicted level of repression by economic shock. Results from table 3.3,

model 4. Bars indicate 90 percent conﬁdence intervals . . . . . . . . . . .

Figure 3.4: Conditional eﬀect of economic shocks and GDP per capita across the

range of development. Bars indicate 90% conﬁdence intervals.

. . . . . .

23

30

31

32

34

35

36

46

53

57

60

x

Figure 3.5: Predicted change in repression when experiencing an economic shock
with and without the presence of overt dissent. Bars indicate 90% con-
ﬁdence intervals.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 3.6: Predicted level of repression with and without dissent by size of eco-

nomic shock. Bars indicate 90% conﬁdence intervals.

. . . . . . . . . . .

Figure 4.1: FARA Report - Morocco: 2006 . . . . . . . . . . . . . . . . . . . . . . .

Figure 4.2: FARA Report - Equatorial Guinea: 2006 . . . . . . . . . . . . . . . . . .

Figure 4.3: Number of US Sanctions by Year . . . . . . . . . . . . . . . . . . . . . .

Figure 4.4: Distribution of Economic and Military Aid . . . . . . . . . . . . . . . . .

Figure 4.5: Proportion of Bypass Aid and the Logged Transformed . . . . . . . . . .

Figure 4.6: Number of Lobbying Contracts by Year . . . . . . . . . . . . . . . . . . .

Figure 4.7: Predicted probability of U.S. sanctions across range of lobbying. Results

from table 4.5, model 1.

. . . . . . . . . . . . . . . . . . . . . . . . . . .

62

63

71

71

88

89

90

91

99

Figure 4.8: Predicted probability of U.S. sanctions with high and low amounts of

lobbying across values of repression. Results from table 4.5, model 2.

. . 101

Figure 4.9: Predicted probability of severe sanctions across the range of lobbying.

Results from table 4.5, model 3. . . . . . . . . . . . . . . . . . . . . . . . 102

Figure 4.10: Predicted probability of receiving military aid with high and low levels

of lobbying across the range of repression. Results from table 4.6

. . . . 105

Figure 4.11: Predicted probability of receiving economic aid with high and low levels

of lobbying across the range of repression. Results from table 4.7.

. . . . 108

Figure 4.12: Predicted proportion of bypass aid by amount of lobbying. Results from

table 4.8, model 3.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

Figure 4.13: Eﬀect of increased repression on the predicted change in proportion of
bypass aid with low and high levels of lobbying. Results from table 4.8,
model 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

Figure 4.14: Predicted proportion of bypass aid by high and low amounts of lobbying

for high repression states. Results from table 4.8, model 4.

. . . . . . . . 114

xi

Figure A.1: Marginal Eﬀects Plot: Unemployment

. . . . . . . . . . . . . . . . . . . 122

Figure A.2: Marginal Eﬀect of GDP Growth on Repression. Bars indicate 90 percent

conﬁdence intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

Figure A.3: Marginal Eﬀect of GDP Growth on Repression. Bars indicate 90 percent

conﬁdence intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

Figure A.4: Marginal Eﬀect of GDP Growth on Repression. Bars indicate 90 percent

conﬁdence intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

Figure A.5: Marginal Eﬀect: From Model 2 . . . . . . . . . . . . . . . . . . . . . . . 129

Figure A.6: Marginal Eﬀect: From Model 2 . . . . . . . . . . . . . . . . . . . . . . . 130

Figure A.7: Marginal Eﬀect: From Model 2 . . . . . . . . . . . . . . . . . . . . . . . 131

Figure B.1: Component Plus Residual Graph for Price Shocks . . . . . . . . . . . . . 136

Figure B.2: Component Plus Residual Graph for Price Shocks . . . . . . . . . . . . . 137

Figure B.3: Predicted Level of Repression with Overt Dissent Present . . . . . . . . . 138

Figure C.1: FARA Report - Bahrain: December 2012 . . . . . . . . . . . . . . . . . . 139

Figure C.2: FARA Report - Mexico: 2012 . . . . . . . . . . . . . . . . . . . . . . . . 140

Figure C.3: Coding Rules for Sanction Severity . . . . . . . . . . . . . . . . . . . . . 140

Figure C.4: Predicted Probability of Severe Sanctions by Lobbying Across Values of

Repression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

Figure C.5: Predicted Probability of Severe Sanctions Across Range of Lobbying . . . 146

xii

CHAPTER 1

INTRODUCTION

Regimes around the world have often turned to repression when faced with a threat to

their power (Davenport 2007c). Thus the repression literature has a long and robust history.

This dissertation seeks to enter the literature on state repression by further analyzing the
economic determinants of human rights abuses.1 Chapter two ﬁnds that the eﬀect youth

bulges have on increased levels of repression can be moderated by the economic condition of

the state in question. Speciﬁcally, economically better oﬀ states will fail to use preemptive

repression in the presence of a large youth cohort. Chapter three similarly looks at the

economic condition of the state and ﬁnds that countries experiencing negative economic

shocks will increase their preemptive repression as a means to insulate themselves from

the unrest these shocks often bring. Finally, chapter four of the dissertation explores how

foreign states can use public relations (PR) ﬁrms based in Washington D.C. to avoid the

negative consequences of their use of repression without having to modify their behavior.

The dissertation proceeds as follows: ﬁrst I present an overview of the literature on state

repression followed by a discussion of how each chapter of the dissertation ﬁts within this

literature.

I then present my three empirical chapters each analyzing a diﬀerent aspect

of state repression. I follow these chapters with some concluding remarks which serve to

highlight the contribution of the dissertation. Finally, I provide an appendix for each chapter

with additional information and analyses.

1.1 State Repression

Quantitative research on state repression came about due, in part, to the availability of

data provided by outlets such as the U.S. State Department, Amnesty International, and
1Throughout the dissertation, I use the terms repression and human rights abuses inter-

changeably.

1

Freedom house. These organizations provided annual, national reports on human rights

conditions starting in the mid-1980’s (Hill and Jones 2014). One of the earliest and most

inﬂuential works to make use of this new data came from Poe and Tate (1994). Using re-

ports from Amnesty International and the U.S. State Department, Poe and Tate found that

democracy and GDP per capita were both negatively related to repression. Conversely, they

found that population size and civil wars were positively associated with state repression.

These ﬁndings have remained consistent within the literature on state repression (Hill and

Jones 2014). For the purpose of their study, Poe and Tate focused on human rights vio-

lations that dealt with the “integrity of the person” (Poe and Tate 1994). This includes

political imprisonment, torture, disappearances, and summary executions. A broader deﬁni-

tion of repression comes from Ritter and Conrad (2016) who say repression is "any realized

or threatened limit or coercive action taken by state authorities to control or prevent chal-

lenges that could alter the status quo policy or distribution of power.". In a similar vein,

Davenport and Inman (2012) argue that repression "deals with applications of state power

that violate First Amendment–type rights, due process in the enforcement and adjudication

of law, and personal integrity or security." These broad and somewhat varied conceptions of

state repression diﬀer from the data most often used in empirical studies – the overwhelming

majority of which follow Poe and Tate (1994) and use data on physical integrity rights.

The theoretical approach taken by Poe and Tate, and one that has been broadly adopted

within the literature, is that of rational choice. This view states that regimes decide whether

to use repression based upon a cost-beneﬁt calculation. Due to this, a state will engage in

repression when the beneﬁts of using such actions outweigh the costs. This approach, and

the resultant scholarly work, have led to a handful of conclusions that are consistent within

the literature on state repression. First, research has consistently shown that there is a neg-

ative relationship between democracy and state repression (Davenport and Armstrong 2004;

Conrad 2010; Davenport 2007c). This ﬁnding holds across diﬀerent democratic institutions

2

and across diﬀering types of repressive behavior leading to it being called the "domestic

democratic peace (Davenport and Moore 2012; Davenport 2007c). Democracy has a sig-

niﬁcant inﬂuence on repression with scholars ﬁnding that democracy can even diminish the

eﬀect violent dissent (which has been shown to signiﬁcantly and consistently increase human

rights abuses) has on repression (Carey 2010). However, some exceptions to the ﬁnding

that democracy reduces repression include democracies conducting counterinsurgency oper-

ations, foreign military occupation, or when facing violent dissent (Carey 2006; Conrad and
Moore 2010 cited in Chenoweth et al. 2017).2 Additional scrutiny of democracy’s inﬂuence

on repression comes by way of measurement. Some argue that using the Polity IV measure

creates issues as some of the components of the Polity index are correlated with repression

(Hill and Jones 2014).

Second scholars have shown that when a state is faced with political challenges in the

form of unrest and dissent, it often responds with some form of repressive action. Similar

to the relationship between democracy and repression, the relationship between dissent and

repression proves stable enough to be called the “Law of Coercive Responsiveness” (Daven-

port 2007a). However, within this relationship there exists some disagreement. Speciﬁcally,

scholars disagree on how dissent and repression interact.

Indeed, almost every possible

relationship (positive, negative, u-shaped) between repression and dissent has been found

(Davenport 1995; Moore 1998). Recent work has argued that this disagreement is largely

due to modeling error (Ritter and Conrad 2016). These authors contend that governments

use preemptive repression as a way to discourage dissent. Thus, the use of repression may

reduce the likelihood of dissent being seen. Additionally, if overt dissent manifests despite

the use of preemptive repression, the government’s choice of how to respond is unclear. This

is due to the fact that the overt dissent came about despite the use of preemptive repres-

sion. Therefore, the implementation of higher levels of repression may not work given that
2See also (Davenport 2007c) for a discussion of how the threat of violence can lead to the

potential breakdown of the domestic democratic peace.

3

the dissent already overcame the preemptive repression. Ritter and Conrad (2016) solve

this issue by using rainfall as an instrumental variable. So, while strides have been made

within the repression-dissent literature, challenges remain. It would seem then that almost

all scholars agree that repression inﬂuences dissent and vice versa, it is the exact nature of

this relationship, however, that remains unclear.

More broadly, the literature also has shown that domestic factors such as democracy

and dissent often outweigh international factors such as international human rights treaties,

sanctions, or trade dependence (Davenport and Inman 2012). Other factors like GDP per

capita (a measure of economic development) prove important but theoretically underdevel-

oped. The literature has failed, thus far, to adequately explain the mechanism through

which economic development aﬀects state repression. Some have argued that richer states

can provide more social services and other means of pacifying the population and this leads

to less of a need for repressive tactics. This explanation, though, is diﬃcult to test as we

do not know what exactly GDP per capita measures (Davenport and Inman 2012). Addi-

tional domestic factors that scholars have found to consistently be related to repression are

youth bulges which increase the likelihood of repression and an independent judiciary which

help reduce repression (Nordås and Davenport 2013; Linzer and Staton 2015). Despite the

key ﬁndings above and the robust nature of the literature on state repression, gaps remain.

Below I identify where the literature on state repression remains underdeveloped and show

how this dissertation helps to ﬁll those gaps.

1.1.1 Contribution to the Literature

One of the most consistent predictors of state repression proves to be large youth cohorts

often labeled youth bulges (Nordås and Davenport 2013). These youth bulges create security

issues for the state by increasing the probability that violence and unrest will be present

(Urdal 2006). Due to this increased likelihood of violence, states will take action to protect

4

themselves and their hold on power. As the literature on state repression has routinely shown,

states respond to potential or overt dissent with increased levels of repression (Davenport

2007c; Nordås and Davenport 2013; Danneman and Ritter 2014). Given this response, it

is no surprise that states respond to youth bulges with increases in repression due to the

violent potential they bring. Yet, these youth also bring with them an economic potential

that, if harnessed, could bring growth for the country. Using this logic, chapter two of the

dissertation argues that states will only increase their repression in the presence of youth

bulges when the economic situation of the state is such that the economy cannot incorporate

these youth within it. Without being able to incorporate these youth within the economy,

the state will instead repress them out of fear for the unrest they could bring. But why

would unemployed youth cause more fear to the regime than those with jobs? The answer

comes from the mobilization literature which argues that the lack of employment lowers the

cost of joining in violence because the potential recruit is giving up or risking far less by

participating if they have no job to lose (Becker 1968; Popkin 1979; Berman et al. 2011).

This situation of increased tension leads states to increase their level of unrest as a way to

insulate themselves from the potential threat. Through my analysis, I ﬁnd strong support

for the proposed moderating relationship between the economic conditions of the state and

the use of repression in the face of youth bulges. I measure the economic condition of the

state by looking at GDP growth as a means of assessing the general state of the economy

and the likelihood of new jobs being found. As a more speciﬁc measure of opportunity, I also

include the unemployment rate for youth age 15-24. Both measures show that states who

are better oﬀ economically use repression less in the face of youth bulges. This ﬁnding is

very signiﬁcant given the consistent relationship between youth bulges and state repression

within the literature (Hill and Jones 2014). Beyond this, chapter two also provides insight

to the policy and development community about how to improve human rights. If states

are more likely to crackdown on their citizens in the presence of a youth bulge but that

5

reduction in respect for human rights can be attenuated by the availability of jobs, perhaps

the international community can focus on employment initiatives and in this way improve the

human rights condition of the state. This ﬁnding also contributes to the literature on state

repression in general by bringing economic factors into the state repression literature but

in a way that moves beyond an important control and instead as a theoretically important

determinant.

Chapter three similarly looks at the impact the broader economy has on repression by
introducing the idea of economic shocks.3 Until now, economic shocks had only been con-

nected to increases in violence, protests, and coups (Miguel 2004; Hodler and Raschky 2014;

Dube and Vargas 2013; Galetovic and Sanhueza 2000; Kim 2016). These studies argue that

economic shocks create a situation where the population of a state has increased grievances

and a decreased cost of joining in conﬂict. This leads to a very volatile situation and one

ripe for unrest. I add to this literature by arguing that this situation of increased uncer-

tainty and potential for unrest will be seen by the state as a threat to its’ power and will

therefore be met with preemptive repression. The idea being that regimes desire to maintain

their power and meet threats to that power with repression. The atmosphere following an

economic shock, as previous literature has made clear, is deﬁned by an increased likelihood

for conﬂict. Knowing this, states will take actions to protest themselves and their hold on

power — i.e. preemptive repression. This use of preemptive repression is used as a means

of insulating the state and lessening the chance the regime is threatened with unrest. I ﬁnd

strong support for this assertion using multiple measures of economic shocks. This chapter

contributes directly to the state repression literature by providing a glimpse into how states

wield their repressive apparatuses.

It is not only when dissent is present that states will

increase repression. As Ritter and Conrad (2016) point out, the use of state repression is

made within the context of the strategic interaction between the regime and its opponents.
3For the purposes of this dissertation, economic shocks can broadly be thought of as a

sudden and sharp economic decline.

6

This chapter shows that strategic calculus in action and details how states use repression

preemptively to avoid a potential future challenge to their hold on power. Moreover, similar

to chapter two, this chapter also highlights how the broader economy can aﬀect the use of

repression. Three of my four measures of economic shocks are broad measures of the econ-

omy and therefore show the importance of how economic factors can drive government’s use

of their repressive apparatus.

Finally, my fourth chapter sheds light on another under-developed thread of the state

repression literature — the role lobbying plays in allowing states to avoid being punished

for their human rights abuses. There exists a robust literature on punishing human rights

violators. For instance, states impose sanctions, international non-governmental agencies use

naming and shaming, and states reduce the amount of foreign aid governments receive all

in an eﬀort to coerce regimes into changing their behavior (Hafner-Burton and Montgomery

2008; Peksen 2009; Murdie and Davis 2012. Yet, few have analyzed how states can avoid

such punishment. Chapter four enters this void and argues that states can avoid being

punished with sanctions or reductions in foreign aid through aggressive lobbying campaigns

in Washington D.C. Foreign governments have long hired PR ﬁrms in Washington D.C. to

lobby on their behalf. Yet the impact of this lobbying, especially with regards to human

rights abusing states, has only recently come to light. Lobbyist can provide information and

conduct media campaigns which help to mold public opinion and provide key information to

policymakers. I show that this inﬂuence can impact the likelihood sanctions are imposed on

states and can aﬀect the allocation of foreign aid for human rights abusing states. Through

these lobbying campaigns, repressive states can avoid the negative consequences of their

actions and can eﬀectively buy themselves out of being punished. This ﬁnding is important

to both the scholarly literature on human rights as well as the policy realm which is tasked

with punishing such states. This chapter contributes to the literature on human rights

by providing a better understanding of the relationship between human rights abuses and

7

punishment for such actions. Understanding how regimes can use lobbying to eﬀectively

avoid being punished for their repressive actions sheds light on an important aspect of human

rights. This insight also provides the policy realm with a clear summation of how repressive

states are avoiding the mechanisms set up to punish them. This understanding can help

scholars better model the relationship between human rights and punishment like sanctions

and foreign aid. Additionally, policymakers can use this information to help limit the ability

of such states to avoid the consequences of their actions.

Through these chapters, this dissertation provides important insights into both the eco-

nomic determinants of state repression as well as how states can avoid being punished for

their actions. In the following sections I present each empirical chapter in turn before provid-

ing a brief conclusion in which I again highlight the contribution and provide a summation

of the ﬁndings.

8

CHAPTER 2

THE CONDITIONAL RELATIONSHIP OF YOUTH BULGES AND STATE

REPRESSION

2.1

Introduction

The study of how demographic changes aﬀect state security has largely focused on the

idea of large youth cohorts or "youth bulges". These youth bulges serve as one of the most

important and consistent predictors of state repression (Hill and Jones 2014). Yet, the current

literature fails to consider the economic situation in which these youth bulges exist. This

paper seeks to address this issue and show that states see these large youth cohorts diﬀerently

depending on the economic situation of the country. The paper proceeds as follows: ﬁrst,

I deﬁne youth bulges and lay out the basic logic of my argument – highlighting how the

paper ﬁts with previous conceptions of youth, violence, and repression. I then show how

large youth cohorts have been linked to increases in political violence and how this link has

led to youth bulges being met with preemptive repression. Following this discussion, I lay

out the decision calculus of the state and show how the economic conditions of the country

inﬂuence how the regime sees these youth bulges. I then empirically test my hypothesis and

end with a discussion of the results.

A youth bulge is deﬁned as the proportion of the population aged 15-24 to the total adult

population (Urdal 2006). There remains some disagreement within the scholarly community

about what causes a youth bulge but most believe it comes as a result of increased devel-

opment. In most cases, this development decreases the infant mortality rate by increasing

access to health care. Yet if birth rates remain high (as they often do in developing coun-

tries), a youth bulge can occur as the number of children and young adults increase at a

faster rate than previous generations (Lin 2012). These increases in the younger generations

then create large youth cohorts that rise together through childhood and eventually reach

9

the age in which they are characterized as a youth bulge (15-24). At this point, these large

cohorts become an important factor in determining the future economic and security situ-

ation of the state. I argue that when the economic conditions of the state are such that

incorporating the youth into the workforce seems likely, these large youth cohorts are not

seen as a potential security threat and thus the state will not initiate preemptive repression.

This scenario is contrasted by that in which a state has little ability to economically incorpo-

rate a large youth cohort into the economy and thus chooses to increase repression in order

to preempt the possible violence and unrest that comes from a large amount of economically

underprivileged youth.

Previous research points to the potential for violence these cohorts bring. And indeed

this is often the case. Yet this body of research has overlooked an important factor that

aﬀects how the state interacts with and views these large youth cohorts. The economic

opportunities available to these youth proves to be an important factor that helps shape the

interaction between the youth and the state. When economic opportunity is low (a situation

characterized by a slow economy and high unemployment), we may see a situation such as

the Arab Spring.

In this scenario, large youth cohorts are met with decreased economic

opportunities and therefore take part in violence. John Bradley, in his book on the Arab

Spring, describes protesters as young people who often "knew nothing of political ideology.

They were brought into the streets, not by a burning desire for free and fair elections, but

by the dire economic circumstances in which they lived” (Bradley 2012, p. 201).

Others have argued similarly that the lack of economic opportunities for the youth of

Egypt, Tunisia, and the other states of the Arab Spring directly led to the violence and

unrest witnessed (Mulderig 2013). Due to this increase in potential unrest, scholars have

argued that the state is likely to increase repression in an attempt to curtail this potential

before it surfaces using preemptive repression (Nordås and Davenport 2013). As previously

mentioned, this argument is augmented by the ﬁnding that youth bulges are one of the most

10

important and consistent predictors of increases in repression (Hill and Jones 2014). Yet,

the relationship between youth, violence, and repression does not have to be characterized

by negative outcomes. In an important article on this topic, Wright and Moorthy (2018)

argue that when faced with a similar threat as youth bulges (in their case refugees) states

may decide to use preemptive repression as these groups often bring an increased likelihood

of violence. However, Wright and Moorthy ﬁnd that the eﬀect of refugees on repression is

moderated by the economic condition of the state. That is, when the state is more likely to

be able to absorb these refugees into the economic system, repression is less likely. It is only

when the economic condition of the state fails to allow for this integration that the regime

will fear these refugees and therefor increase preemptive repression. As I outline below, the

regime’s decision calculus when facing large youth cohorts operates in much the same way.

The main factors that shape a large youth cohort’s impact on the state have to do with

the economic situation of the country as a whole (as pointed out above). If the economic

situation of the state is one of opportunity and potential economic integration, then the state

may be able to harness the economic potential of the youth and save itself from potential

unrest (USAID 2013). The logic here is simple, as the number of working persons increases,

the dependency ratio (or ratio of non-working age population to working age population)

decreases. When this ratio declines, as it would if the majority of youth from the youth

bulge were employed, then income per capita should rise (Lin 2012). In this scenario, the

youth have a decreased likelihood to take part in unrest due to their ability to ﬁnd work

and make a living. This new workforce also provides potential beneﬁts to the state in the

form of increased taxation and economic growth. Many Asian economies have experienced

the beneﬁt of large youth cohorts and have seen a dramatic rise in economic growth due

to a large (and employed) youth workforce (Lin 2012).

It would follow then, that when

economic opportunities are available to these youth in the form of an economy with the

ability to incorporate the large number of new workers, the state would have little incentive

11

to increase repression as the threat of future violence is small.

These contrasting outcomes create the need to reexamine our expectations of how large

youth cohorts can impact a state’s economic and security situation. By taking economic

opportunity and incorporation into account, I show that the economic condition of the state

changes how states view large youth cohorts. When the probability of incorporating these

youth into the economy is low, states will view these youth as a potential future threat and

begin to increase repression preemptively. If, however, the state sees economic incorporation

as likely, the regime will fail to see the youth as a threat and will correspondingly not see a

need to increase repression preemptively.

2.2 Youth Bulges and the Potential for Violence

Large youth cohorts can have a detrimental impact on a state’s security situation. Figure

2.1 below shows the level of youth over time in some of the counties contained in the sample.

The average percent youth in a country at any given time is 29%. As the ﬁgure shows, states

see varying levels of youth across time. Some, like Nigeria, have maintained a consistently

above-average level of youth with values ranging from 32%-36% while others like Russia

and Argentina have maintained historically below-average percentages of youth with values

ranging from 17%-25%. Still others like Vietnam have seen their numbers go from far below-

average to above-average and back down.

These youth cohorts can dramatically increase the likelihood of terrorism, armed conﬂict,

riots, and violence generally (Urdal 2006; Elbadawi and Sambanis 2000). Beyond conﬂict,

scholars have found large youth cohorts to also be associated with increased political crises

and an increase in political activism (Goldstone 1991; Huntington 1996; Hart et al. 2004).

The mechanisms behind these ﬁndings point to the nature of youth as well as their eﬀect on

the institutions and systems of the state when present in large numbers. For instance Hart

et al. (2004) argues that it is an inherent distrust of the current political institutions that lead

youth to take part in political activism. Goldstone (1991) highlights the stress large youth

12

Figure 2.1: Percent Youth by Year

populations put on existing institutions and systems which lead to institutional bottlenecks.

Simply put, Goldstone argues that large numbers of youth overwhelm the system and create

high unemployment and a lack of economic opportunity due to the inability of the system to

incorporate such high numbers of youth. Collier (2000) and Urdal (2006) then extend this

argument and tie large youth cohorts to motivations and opportunity for violence.

Collier (2000), for example, argues that a high number of youth lead to more available

recruits for violence by lowering the barrier of entry into such activities. This comes as a

result of the institutional crowding and economic hardship mentioned earlier. If the number

of youth increases, causing institutional crowding, then things like wages and availability of

jobs will be lower. This lowers the opportunity cost of joining violence and thus makes the

number of recruits increase (Collier 2000). Based on this research, it is easy to see why these

youth bring such potential for volatility. In these conditions, you eﬀectively have a large

13

number of persons who distrust the current political order, are more likely to take part in

political activism, and have increased incentives for doing so given the strain they put on

the economic system which has a detrimental impact on their own economic situation. The

latter part of this equation, the negative impact youth have on the economic situation in a

state, proves vital to connecting youth to increases in potential for violence.

Urdal (2006), and his seminal work on this topic, relies heavily on the idea of youth

bulges, coupled with a decrease in economic opportunity, as a signiﬁcant factor in the reason

for an increased potential for violence. The logic behind this view follows the "opportunity"

argument discussed above but also draws on the "grievance" literature. Urdal argues that

youth increase the opportunity for violence through lowering the costs of joining these groups.

In order for this opportunity cost to be lowered, the opportunities oﬀered by rebels must

be greater than those oﬀered by other legitimate avenues. The inability to ﬁnd legitimate

employment can be a factor that tips the scale in favor of violence (Gates 2002). Much of the

literature on opportunity argues that violence will increase when the cost of joining in such

activity decreases. Large youth cohorts can create an overabundance of labor which can lead

to decreased wages. This then, lowers the cost of joining in violence because the wages you

give up by joining in violent activity are lower due to the overabundance of labor (Collier

2000; Fearon and Laitin 2003). Urdal (2006) cites Brett and Specht (2004) as ﬁnding that

poverty and low income opportunities lead to an increased chance of joining a rebel group

(610). Similarly, Korenman and Neumark (1997) ﬁnd that large youth cohorts lead to a

signiﬁcant increase in youth unemployment rates. These ﬁndings show that large youth

cohorts often face reduced costs to joining in violence due to the inability to ﬁnd economic

opportunities. There is, however, another side of the coin. While opportunity may play an

important role in violence, motives must also be taken into account (Collier and Hoeﬄer

2004).

The motive based literature, or "grievance", argues that violence happens as a result of

14

people being unhappy with the current system or their economic state. Similar to (and draw-

ing upon) the relative deprivation literature, this theoretical tradition argues that conﬂict

erupts as a rational response to economic or institutional discontent (Gurr 1970; Sambanis

2004). They argue that the motive for violence comes from grievances within the population.

Large youth cohorts can create institutional bottlenecks and can lead to high unemployment.

These increased grievances then lead to an increase in the potential for political violence

(Goldstone 1991). It should be pointed out as well that Urdal argues that the motive argu-

ment can be augmented by the opportunity argument in that the likelihood that violence is

used to redress grievances increases as the opportunity costs of joining in violence decreases

(610). That is, the economic factors that increase grievances (unemployment, low wages,

etc.) also reduce the opportunity costs of joining a violent movement such as lost wages.

Therefore, factors such as high unemployment, created by a large youth cohort, increase the

chance of violence by increasing the potential grievances of the population and decreasing

the opportunity costs of joining in a violent movement. According to this logic then, it is

necessary to couple grievances with the opportunity for violence discussed above in order for

the likelihood of violence to increase.

2.3 Youth Bulges and State Repression

Violence and unrest have long been linked to increases in repression. Some have argued

that the level of repression increases along with the level or nature of the threat (Gartner and

Regan 1996; Davenport 1995). Others argue that governments respond to dissent with either

repression or accommodation based upon previous interactions (Moore 2000). Davenport

(2007a) and Davenport and Inman (2012) aggregate these ﬁndings and term the results the

"Law of Coercive Responsiveness". This law states that when threatened, states will most

likely respond to that threat with repression. This “law” proves to be strong enough to even

overcome one of the other “laws” Davenport puts forward; the “Domestic Democratic Peace”

(Davenport 2007a). The domestic democratic peace comes from a large body of literature

15

that argues democracies repress less than other regime types. These ﬁndings are compiled

by Davenport and made into a “law”. However, as Davenport (2007b) ﬁnds, dissent, if severe

enough, can even break down this domestic democratic peace. So it stands that, in the face

of dissent, states will likely respond with repression. And this result holds even when looking

at democracies.

Nordås and Davenport (2013) build upon this ﬁnding and argue that large youth cohorts

and the potential for violence they bring can cause governments to use repression (often pre-

emptively) in order to deter the possible unrest these youth could engage in. The logic behind

this assumes that regimes want to stay in power, therefore states facing large youth cohorts

will see them as a threat and will engage in policies to control that threat. As we know, a

state facing an increased potential for unrest will likely respond with repression. Since large

youth cohorts are known to have an increased likelihood of future violence, and states can

identify large youth cohorts as they rise through the school systems; these states will engage

in repression – even in the absence of dissent – in order to control and attempt to stem

the increased potential for violence that accompanies these youth. Nordas and Davenport’s

argument centers on the idea of repression having both costs and beneﬁts. Repression can be

costly in democracies where citizens can hold leaders accountable and punish them for their

heavy-handed response. Repression can also have international ramiﬁcations such as naming

and shaming and sanctions (928). Additionally, repression also comes with real ﬁnancial and

resource costs. Engaging in increased repression requires more people and resources to be

provided by the state thus incurring real costs to a regime’s resources. However, potential

beneﬁts also exist, namely the ability to control a dissenting population, shield the regime

from potential future challenges, and remain in power. Therefore, Nordas and Davenport

argue that governments, when faced with large youth cohorts and the potential for future

violence that they represent, will repress preemptively to extend the status quo.

This argument, however, assumes that the threat posed by these large youth cohorts

16

outweighs the costs of increased repression. As Nordas and Davenport point out, repression

is not a cost-less action. Therefore, the threat posed to the regime must be one that out-

weighs the costs of increasing repression. Due to this, I argue that states will likely increase

preemptive repression only when they deem large youth cohorts to be threatening enough to

do so. How then do states deem a threat worthy of increased repression? By observing the

economic opportunities available to the youth population and calculating the probability of

future violence. When economic opportunity is high and the youth are likely to be incor-

porated into the economy, the state is less likely to respond with preemptive repression as

the threat of future violent dissent is lessened. This reduced likelihood of violence will lead

to the cost of increasing repression outweighing the potential threat posed to the regime.

In such a situation, we should observe little to no increase in repression in the face of large

youth cohorts. However, when the state deems economic incorporation of the youth into

the economy as unlikely, the regime will see the youth as having an increased likelihood of

participating in future unrest and will therefore begin to implement preemptive repression

as a means to insulate the regime from this violent potential. Below, I develop this logic

more fully and provide examples that highlight this decision calculus.

2.4 Youth, Repression, and Economic Opportunity

When should states fear large youth cohorts and in response to this fear increase levels

of repression? Drawing from the motive and opportunity literature discussed above, I argue

that states should fear large youth cohorts only when those youth pose a potential threat

to the state. Large youth cohorts are most likely to threaten a state when the motive and

opportunity for violence is at its highest. When these youth have less opportunities due

to the reduced economic capacity of the state to incorporate them into the system, the

youth populations have decreased costs of joining in violence yet have increased motives.

Less economic opportunities can create hardships that make joining in dissent more likely.

This rational draws upon the opportunity-cost approach ﬁrst described by Becker (1968)

17

and further developed by Popkin (1979). This theory argues that potential participators

in violence are rational actors and therefore respond to changes in the opportunity cost of

participating. Berman et al. (2011) lay out the theoretical framework formally by pointing

out that there exists some wage that actors would prefer to taking part in violence. This

threshold is inﬂuenced by things like risk aversion, commitment to the cause, and other

factors. Berman et al. (2011) goes on to point out that the number of potential participants

is "a decreasing function of wages in the legitimate economy."(500) This logic points to

a positive correlation between unemployment and the number of available participants to

unrest. This ﬁnding is echoed by recent work linking unemployment to conﬂict. In their

study of Indonesia, Barron et al. (2016) ﬁnd that unemployment signiﬁcantly increases the

chance that violence will occur. Similarly, Humphreys and Weinstein (2008), citing Abdullah

(1998), argue that "Unlettered, unemployed migrants formed the basis for Sierra Leone’s

insurgency in part because they were “cheap.”

For our purposes, suppose that a large number of youth exist in State A and State B.

State A is experiencing high economic growth and has a low unemployment rate while State

B is experiencing the opposite (low growth and high unemployment). The opportunity-

cost theory outlined above would say that the youth in State B pose a larger threat to the

regime than those in State A due to their availability and lower threshold for engaging in

unrest. Therefore, the government of State B is more likely to use preemptive repression

as a way to shield itself from the potential unrest these disaﬀected youth bring. State B

fears that future unrest is likely, due to the large availability of youth with a low threshold

for participation. Ideally, the state would like to incorporate these youth into the economy

and harness their economic potential but governments cannot simply grow the economy on

command. Therefore, states resort to what they can do — increase repression in the hope

that it will insulate the regime from a future threat.

This point is highlighted by Liberian President Ellen Johnson Sirleaf who pointed towards

18

unemployment as the cause of the violence within Liberia. She is quoted as saying “the level

of unemployment and the idleness of our youth have a propensity for social disenchantment.

For us, employment is synonymous to peace [emphasis added].” (Johnsson 2006, p. 3). Here,

then, is the leader of a state arguing that economic opportunity can bring about a reduction

in violence. This quote also shows that governments understand the relationship between

economic opportunity and the probability of violence, thus lending support to the argument

that they will try to preempt this violent potential by employing preemptive repression.

Additional anecdotal evidence for the relationship between economic opportunity and

threats to the state comes from the Arab spring movement. Houle (2017) cites Malik and

Awadallah (2013)as saying, "Arab revolutions were fueled by poverty, unemployment, and

lack of economic opportunity" and those that participated were "the youth who see little hope

for economic and social mobility" (p.296-297 cited in Houle (2017). Others have reiterated

this point saying "rather than the Arab world’s usual suspects — bearded Islamists or jaded

leftists — it is young people, angry at the lack of economic opportunity available to them,

who are risking their lives going up against police forces."[emphasis added] (Knickmeyer

2011) When economic opportunity is low, these youth cohorts are more likely to engage in

violence and unrest due to both the increased opportunity and motive for doing so. As the

case of Liberia shows, government leaders are aware of this as well which indicates that they

will likely take steps to avoid the threat to their power.

The use of preemptive repression by the state in the face of a potential future threat has

been established within the literature on state repression. As Rory Truex ﬁnds, the Chinese

regime increases repression preemptively during days in which it believes unrest to be more

likely such as national holidays. The regime purposely increases its’ level of repression in

order to insulate itself from the potential for unrest that come on these days regardless of

whether or not dissent is present (Truex 2019). Beyond the example of China, Danneman

and Ritter (2014) ﬁnd that states with neighbors who are actively engaged in a civil war use

19

preemptive repression at home as a strategy to insulate their country from the possibility of

conﬂict contagion. In both of these examples, states are deciding to use repression as a tool

to decrease the negative eﬀects of potential unrest without an increase in actual unrest.

This causal process is highlighted by the example of Liberia mentioned brieﬂy before.

Liberia in 1999 had a high youth population and had low economic opportunity. Liberia was

also experiencing a large amount of violent unrest. This violence, I would argue, stemmed in

part from having high numbers of youth without much economic opportunity. The inability

of the youth to ﬁnd jobs increased their grievances towards the state and decreased their cost

of joining in violence. This argument is echoed by the World Bank who, in a 2011 World

Development report, cite unemployment as the most frequent reason given to why one would

join a rebel movement (World Bank 2011). So when asked to give their reason for joining in

violence, the most common response stemmed from a lack of economic opportunity. Both the

World Bank and the Liberian President are arguing that the level of economic opportunity

directly impacts the likelihood of violence. The quote from the Liberian president citing

the pacifying eﬀect of employment shows that state leaders understand the impact economic

opportunity can have on youth and their propensity for violence. It follows, then, that these

same leaders can evaluate their own economic situations and gauge the potential threat posed

by large youth cohorts. If the economic situation is one of abundance, then state leaders

have little to fear and therefore little incentive to repress. If, however, the economic situation

is less abundant, then leaders understand the violent potential posed by large youth cohorts

and are therefore more likely to repress preemptively. The lack of economic opportunity

however, details only one potential situation. There also exists states in which economic

opportunity is abundant which leads to vastly diﬀerent outcomes.

Economic abundance has a direct impact on violence and unrest. For instance, when

economic opportunities are abundant, youth have an increased cost of joining in violence

and a decreased motive for doing so. A scenario of economic abundance can increase wages

20

(and the likelihood of having a job more generally) as well as inﬂuence one’s outlook on

the potential for future economic gain thereby decreasing grievances and increasing the
cost of joining in violence.1 Due to this, the probability of violence decreases as economic

opportunities increase. Drawing on this, I believe that economic opportunities can condition

the likelihood of preemptive repression as youth bulges increase. In states with economic

abundance, I expect the eﬀect of youth bulges on repression to be minimal due to the reduced

likelihood of violence. In states with economic scarcity, I expect the eﬀect of youth bulges

on repression to increase as youth bulges become larger and the threat of violence increases.

The case of Cameroon provides another example of how the economic situation of the

state can inﬂuence repression in the face of large youth cohorts. From 1990 to 2010,

Cameroon’s youth population maintained above average levels. The average percentage

of youth during this time was 36% with the yearly percentage never falling below 34.5%
throughout this time-frame.2 The president throughout this time period was Paul Biya who

assumed the presidency in 1982 and remains in oﬃce today. Throughout the 1990s and 2000s,

Cameroon saw a drastic change in its’ economic situation with GDP growth switching from

negative to positive in the mid 1990s and the youth unemployment rate dropping steadily

in the late 1990s and throughout the 2000s. The fact that the percentage of youth remained

steady throughout this time period provides a helpful illustration of the impact economic

opportunity has on repression in the face of a high percentage of youth. Below I provide a

graph that shows the level of repression in Cameroon from 1990 - 2010. During this time,

recall that youth percentages remained more or less unchanged while the economic situation

steadily improved. To show this, I include snapshots of the youth unemployment rate at

diﬀering times. What becomes clear from the graph is the eﬀect the economic opportunity

has on the level of repression. While youth unemployment numbers remain high, we see
1Again, see the opportunity-cost literature here Becker (1968); Popkin (1979); Blattman

and Miguel (2010)

2Note that the average percent youth in the sample is 29%.

21

the state responding with higher levels of repression. However, as the economic situation

improves, we see a drastic reduction in the level of repression. This trend continues as the

economic situation continues to improve. This represents a major change in repression with

Cameroon moving from being in the top 80th percentile of repression in the early parts of

the 1990s to being in the bottom 45th percentile by the end of 2010.

Additionally, Cameroon shows that repression in the face of large youth cohorts relies

more on the economic conditions of the state rather than on the realized level of dissent. The

correlation between the number of protests and the level of repression in Cameroon from

1990-2010 is .24 while the correlation between repression and youth unemployment during
the same time period is .84. Cameroon actually saw an increase in protests from 2005-20103

while their level of repression and youth unemployment was steadily decreasing. This gives

some support to the idea that governments fear the youth population when those youth do

not have the ability to be integrated into the economy. When the youth can be integrated,

the government reduces its’ repression even in light of higher levels of protest indicating that

it is not the level of dissent but the fear of youth driving the use of repression.

3Cameroon saw an average of just under one protest event a year from 1995-2004. That

number increased to three from 2005-2010

22

Figure 2.2: Cameroon: Level of Repression by Year - Red Lines Indicate the Youth Unemployment Rate

The discussion above leads to the following hypothesis:

Hypothesis: States will increase repression more in contexts of low economic opportunity

compared to situations of high economic opportunity, and this eﬀect will be stronger the

larger the youth bulge in that state.

2.5 Data and Research Design

2.5.1 Dependent Variable

To test this hypothesis, I use cross-national data from 1960-2012. Country-year serves

as the unit of analysis. To capture the level of state repression within a country I estimate

models using three diﬀerent dependent variables. The ﬁrst two come from the Political

Terror Scale (PTS) (Wood and Gibney 2010) and consist of categorical measures ranging

from one to ﬁve with higher values referring to increased levels of repression. The Political

23

Terror Scale deﬁnes a repression level of "1" as "Countries under a secure rule of law, people

are not imprisoned for their views, and torture is rare or exceptional. Political murders

are extremely rare." Similarly a "2" represents "There is a limited amount of imprisonment

for nonviolent political activity. However, few persons are aﬀected, torture and beatings

are exceptional. Political murder is rare." Most consider a PTS score of "3" to be where

repression begins (see Davenport and Appel (2014) for a discussion on repressive spells).

The Political Terror Scale deﬁnes a "3" as "There is extensive political imprisonment, or a

recent history of such imprisonment. Execution or other political murders and brutality may

be common. Unlimited detention, with or without a trial, for political views is accepted."

Finally, a "4" and a "5" are deﬁned as "Civil and political rights violations have expanded to

large numbers of the population. Murders, disappearances, and torture are a common part

of life. In spite of its generality, on this level terror aﬀects those who interest themselves in

politics or ideas." and "Terror has expanded to the whole population. The leaders of these

societies place no limits on the means or thoroughness with which they pursue personal or

ideological goals." respectively. When using the PTS coding of the dependent variable, the

data range is restricted due to missing values.

The PTS measure consists of two separate measures of repression based upon diﬀerent

reporting sources. The ﬁrst measure makes use of reports from Amnesty International and

codes the level of repression based upon the scale listed above. The second measure considers

reports from the United States State Department and categorizes a state’s level of repression

based upon these reports (again according the the 5 point scale deﬁned earlier).

The second measure of repression comes from Fariss (2014). This measure takes into

account the changing standards of accountability within the human rights community and

models this latent change into the coding of repression. Fariss (2014) argues that human

rights standards have changed over time and these changes are incorporated into the current

coding of human rights variables. Without modeling these changing standards, he argues,

24

our measures suﬀer from bias.

Incorporating these changing standards, Fariss maintains,

better reﬂects the human rights conditions across time. In the original coding of this latent

repression variable, higher values correspond to less repression (more respect for human

rights). In order to simplify interpretation of the coeﬃcients and to maintain consistency

with the PTS coding of repression, I invert this measure by multiplying by negative one so

that higher values indicate higher levels of repression.

2.5.2

Independent Variables

My main independent variables are youth bulges and GDP growth. Both of these vari-

ables come from the World Bank. Youth bulges are measured by dividing the total adult

population by the number of those who are 15-24. This coding is consistent with the litera-

ture on youth bulges and violence (Urdal 2006).

To proxy for the ability of a state to incorporate the youth into the economic system, I

follow Urdal (2006) and use economic growth. In that study, Urdal uses economic growth

to proxy economic opportunity or the ability of the state to economically handle these large

youth cohorts. Economic growth speaks to the overall economic situation of the state and

therefore provides a state-level proxy of the economic conditions of the country in regards

to future employment, income, and overall economic strength. These economic conditions

translate into how the regime views youth bulges. Economic growth itself, directly ties to

the country’s ability to incorporate new workers into the system. When economic growth is

high, demand for labor tends to be higher than when economic growth is low. Therefore,

economic growth should serve as a useful proxy for how well the system can absorb a youth

bulge. Table 2.1 provides some descriptive statistics for my dependent variables as well as

my variables of interest. As can be seen, GDP growth has some large outliers. To account

for this, as I point out later, when predicting levels of repression by percent youth I set a

proﬁle that uses percentiles rather than predicting over the minimum and maximum values

25

found in the data.

To further test the mechanism discussed, I also use the youth unemployment level to

capture the economic condition of the youth. This measure comes from the World Bank and

measures the unemployment level for youth 15-24 (which is the same conception of youth used

by Urdal 2006 and in this paper). Using the youth unemployment rate allows me to capture

how many youth may be available and have a reason to challenge the government. The higher

the level of unemployment amongst the youth population, the more likely governments are to

view those youth as a potential threat. Therefore, this measure captures both the economic

opportunities open to youth but also the likelihood the state could see them as posing a

potential threat.

Table 2.1: Summary Statistics: Dependent Variables and Explanatory Variables

2.5.3 Control Variables

To help isolate the eﬀect of my conditioning factors, I also include control variables that

the repression literature has proven to be important (Davenport 2007a;Hill and Jones 2014;

Davenport and Appel 2014). To account for the amount of potential dissent as well as the

ability of the state to repress I include a population measure (logged) as well as a measure

of GDP per capita (logged). Both variables come from the World Bank. As the literature

has shown, regime type can also be an important predictor of repression with democracies

experiencing much lower levels of repression. Due to this, I include a measure of regime type

into the model. I use the X-Polity data from Vreeland (2008). Additionally, I also control

26

for the amount of dissent within the state as well as the presence of a civil war. Both of

these factors have been shown to increase the probability of repression and are particularly

relevant given that they are also aﬀected by the percent youth in a country. The data on

dissent comes from the Cross-National Time-Series dataset (Banks and Wilson 2017). To

create my dissent measure, I ﬁrst sum the protest events within the CNTS data.

I then

create a dummy for whether or not dissent occured within that state in a given year. The

measure for civil war comes from UCDP conﬂict data and equals a 1 if there is a civil war

and a 0 otherwise. Lastly, I control for the amount of judicial independence. An independent

judiciary has been shown to dramatically reduce the amount of repression seen in a state, and

thus proves to be an important control. Data for this measure comes from Linzer and Staton

(2015). Table 2.2 provides a summary of the control variables contained in the analysis.

Table 2.2: Summary Statistics: Control Variables

2.5.4 Research Design

To test the conditional eﬀect of economic opportunity on youth bulges and repression

I run six sets of models interacting my independent variables (GDP growth and youth un-

employment) with the measure for youth bulges. All models use the same list of covariates

mentioned above in the control variable section and are regressed against each of my three

measures for state repression. To account for auto-correlation that may be present due to

the panel nature of my data, I then rerun each model with yearly time dummies.

I also

included a lagged dependent variable as some have shown that the lagged level of repression

27

is an important predictor of current repression (Poe and Tate 1994; Nordås and Davenport

2013). Lastly, I cluster my standard errors on country. For my models using the PTS coding

of repression I estimate an ordered logit model as this coding of repression utilizes an ordered

scale. For models utilizing Fariss’ coding of repression I use OLS following Hill and Jones

(2014). Given that my main theoretical interest lies in the interaction between GDP growth

and youth unemployment with youth bulges, I include a visualization of this conditioning

relationship.

2.6 Empirical Results

The results from table 2.3 give preliminary support for the hypothesis presented. Across

the diﬀerent models, we see that the control variables behave according to expectations

lending conﬁdence to the model. For instance, large populations, dissent, and civil wars are

all positively correlated to increased repression while an independent judiciary as well as

more democratic states are both negatively correlated to state repression. The testing of my

theoretical concept however, lies in interacting GDP growth and youth unemployment with

youth bulges. Interpreting such an interaction from a table such as this is of limited value
however. 4 To overcome this, I provide a visual representation of this relationship.

4See Brambor et al. (2006)

28

Table 2.3: Moderating eﬀect of GDP growth on the relationship between youth bulges and
state repression

29

Graphically, Figure 2.3 (based on model 2) shows that states with high values of GDP

growth fail to see an increase in the level of repression as the number of youth increase.

This is contrasted by those states with low GDP growth. To graph the conditioning eﬀect

shown in Figure 2.3, I predict values of repression using a proﬁle that leaves continuous

variables at their mean and the dummy variable for civil war set at zero. I set GDP growth

at the ﬁrst percentile (-13%) and the ninety-ninth percentile (21%) for low and high levels

of GDP growth respectively. As the graph shows, those states that have a limited ability

to incorporate the youth into the economic system see a dramatic increase in repression as

the number of youth increase. This graph lends support for the theoretical expectation that

the eﬀect of youth bulges on repression diﬀers across diﬀerent values of GDP growth. Those

states with high economic growth see very little change in repression while those with low

growth see their use of repression signiﬁcantly increase as the amount of youth increase.

Figure 2.3: Predicted level of repression by GDP growth across the range of percent youth. Results from
table 2.3, model 2. Bars indicate 90 percent conﬁdence intervals

30

We see this same relationship play out graphically when looking at the results from model

four depicted in Figure 2.4. Again, when there is high economic growth which represents

an economy that can accommodate large youth cohorts, we see very little change in the

probability that the state will have a high level of repression. However, this relationship

changes when a state has low values of economic growth.

In these instances, the state

cannot incorporate the large number of youth into the economy which leads to an increased

chance of violence. Knowing this, the state increases its level of repression as the number

of youth grow. Figure 2.5 from model six utilizes the state department coding of repression

and again shows a consistent result with the probability of high repression only increasing

when GDP growth in low and youth levels are high.

Figure 2.4: Predicted level of repression by GDP growth across the range of percent youth. Results from
table 2.3, model 4. Bars indicate 90 percent conﬁdence intervals

31

Figure 2.5: Predicted level of repression by GDP growth across the range of percent youth. Results from
table 2.3, model 6. Bars indicate 90 percent conﬁdence intervals

The results, as seen in Figure 2.3, Figure 2.4, and Figure 2.5 all support the hypothesis

that the level of economic growth within a state matters in how that state reacts to high

levels of youth. When there exists a high amount of economic opportunity, states have

little reason to fear these large numbers of youth and therefore fail to increase the level of

repression. However, when levels of economic opportunity are low, states have more reason

to fear these youth and therefore we see a decrease in respect for human rights.

To further test this relationship, I run the same set of models this time interacting youth

bulges with the youth unemployment rate. Table 2.4 presents the results and, as before, I

provide marginal eﬀects plots to interpret the interaction.

32

Table 2.4: Moderating eﬀect of youth unemployment on the relationship between youth
bulges and state repression

As with GDP growth, we again see that the unemployment level of youth conditions

33

the relationship between repression and youth bulges. Speciﬁcally, when the youth unem-

ployment level is low, we see little to no change in the predicted level of repression across

diﬀering levels of youth. Conversely, when youth unemployment is high, the predicted level

of repression increases in conjunction with the number of youth. This provides some support

for the assertion that states with high levels of youth will only see those youth as a threat

— and thereby increase repression — when the economic opportunities of those youth low.

When the youth can be successfully incorporated into the economy (indicated by a low un-

employment rate) states fail to see these youth as a threat which leads to no change in the

level of repression.

Figure 2.6: Predicted level of repression by youth unemployment across the range of percent youth. Results
from table 2.4, model 2. Bars indicate 90 percent conﬁdence intervals

34

Figure 2.7: Predicted level of repression by youth unemployment across the range of percent youth. Results
from table 2.4, model 4. Bars indicate 90 percent conﬁdence intervals

35

Figure 2.8: Predicted level of repression by youth unemployment across the range of percent youth. Results
from table 2.4, model 6. Bars indicate 90 percent conﬁdence intervals

Figures 2.6, 2.7, and 2.8 all show support for the theory that the economic opportunities

available to youth can condition the eﬀect these large youth cohorts have on state repression.

In each case, states with low unemployment see very little to no change in their level of

repression across varying levels of youth. This ﬁnding is contrasted by those states with high

unemployment which see a dramatic rise in repression as the level of youth increases.

2.7 Discussion

The models and graphs above lend support for the hypothesis that the eﬀect of youth

bulges on repression can be conditioned by the level of economic opportunity within the state.

It seems that states who ﬁnd themselves in situations of economic abundance fail to increase

repression in the face of large youth cohorts due to their ability to successfully incorporate

these youth into the economic system. This incorporation changes the perception of the

36

youth causing the government to not see them as a potential future threat. Contrasting this,

those states that are unable to provide this economic incorporation of the youth increase

repression when facing these large populations of young people in an eﬀort to prevent the

likely increase in potential for violence that they bring. While these ﬁndings shed light on an

overlooked part of repression, there remains work to be done. While this paper puts forward

a state-centric approach to the issue, how do the youth in question view their situation?

Perhaps survey data on youth perceptions of the government or perceptions of their own

economic situation may shed more light on how these youth bulges interact with and aﬀect

the state and its actions. Future work could do more work to identify the motives and

behaviors of the youth and thereby ascertain the motives and thinking behind their actions.

2.8 Conclusion

Research has shown that large youth cohorts serve as an important predictor of state
repression.5 States can see these youth as a threat and increase their level or repression

preemptively to avoid the negative consequences these youth have the potential to bring.

However, as I have shown, the economic conditions of the state can serve as a conditioning

factor on this use of repression. States with low levels of economic opportunity have more

to fear from large youth cohorts and therefore are more likely to use preemptive repression.

Conversely, states with high levels of economic opportunity have less of a reason to fear

these large youth cohorts and are therefore less likely to use preemptive repression. Regimes

calculate the cost of repression against the potential rewards it brings. When facing a large

number of youth, the state must decide the likelihood of future unrest caused by these youth.

To do this, the economic situation of the country must be taken into account. This decision

calculus proves essential in understanding the relationship between youth bulges and state

repression.

5See Nordås and Davenport (2013); Hill and Jones (2014)

37

CHAPTER 3

ECONOMIC SHOCKS AND STATE REPRESSION

3.1

Introduction

How do economic shocks aﬀect state repression? The literature on economic shocks has
long connected sudden and severe economic downturns with increases in violence.1 Yet, the

relationship between economic shocks and state repression has failed to garner the same

attention. This paper seeks to ﬁll that gap by arguing that states understand the violent

potential economic shocks can bring and therefore will seek to avoid this threat to their

power by preemptively increasing repression. My argument draws on other scholarship on

preemptive repression ( Nordås and Davenport 2013; Truex 2019; and Danneman and Ritter

2014) and contends that states experiencing an economic shock will see a corresponding

increase in repression as the government seeks to insulate itself from a future potential for

violence. Additional analysis determines whether this eﬀect could be moderated by the

economic development of the state or the presence of overt dissent. The paper progresses

as follows. I ﬁrst discuss state repression and outline the incentives states have for using

repressive measures and how the economic conditions of the state can impact this. I then

present a theory of how economic shocks lead to an increase in state repression highlighting

this logic with a short case study from Zimbabwe. Finally, I empirically test the mechanism

put forward and discuss the results. In brief, I ﬁnd strong support for the idea that states

will increase their repression in the wake of an economic shock in an eﬀort to avoid the

potential violence that often accompanies such shocks.

Examining this relationship proves crucial to understanding how economic shocks aﬀect

the security situation of a state. The eﬀect of economic shocks on a state’s repressive
1See Miguel (2004); Hodler and Raschky (2014); and Dube and Vargas (2013) among

others.

38

actions represents a vital aspect of the relationship between the state and its citizens. This

eﬀect is important for both policymakers and human rights advocates as they both seek to

better understand the strategic calculus used by states when deciding to employ repression.

Economic shocks represent a key, and until now, overlooked aspect of human rights violations

and can help further our understanding of state repression.

3.2 Economic Conditions of the State and Repression

The relationship between economic factors such as GDP and per capita income and re-

pression remains understudied. While some have argued that economic development, mea-

sured as GNP per capita, decreases repression — the mechanism behind this relationship

is unclear (Davenport and Inman 2012).

Indeed, much of the repression literature treats

GDP as an important control variable but not as an important predictor of state repression

(Hill and Jones 2014). While the literature connecting economic conditions of the state and

repression is limited, notable exceptions exist. A notable example of this comes from Dav-

enport (1995). Here, Davenport includes the domestic economy within his analysis. He cites

previous studies as arguing that when the domestic economy is running well, states are less

likely to perceive dissent as a threat and are therefore less likely to use repression (Goldstein

1983). Indeed, the conditional eﬀect of the economy is a growing theme within the repres-

sion literature. Recent work by Wright and Moorthy (2018) shows that state repression is

sometimes conditional on the economic condition of the state. The authors examine how

increased refugee ﬂows within a country aﬀect the level of state repression. They argue that

refugees could be seen as a potential threat and therefore increase the level of repression as

the state seeks to mitigate that threat. However, this expectation is conditional upon the

economic condition of the state. In states with high economic capacity (measured as high

GDP per capita) the threat posed by increased refugees is mitigated by the ability of the state

to economically adsorb these new arrivals within the economy (Wright and Moorthy 2018).

Here we see how the economic conditions of a state can inﬂuence how that state perceives

39

a threat and therefore how that state chooses to respond. Though the relationship between

high GDP and repression seems clear, the question of how economic downturns aﬀect repres-

sion remains less deﬁned. Higher levels of GDP help to reduce levels of repression as states

are better able to meet the needs of their citizens. This ﬁnding, as discussed earlier, proves

to be a consistent conclusion within the literature on state repression. But what happens

when the state is faced with an economic crisis? We know from the literature on conﬂict that

economic downturns are often associated with increases in violence and unrest (Miguel 2004;

Hodler and Raschky 2014; Dube and Vargas 2013). Yet it is unclear how these negative

economic conditions aﬀect the decision of states to apply repression. Economic downturns

(shocks) create an environment where people have increased grievances and decreased costs

in joining violent groups. This leads to an increase in the likelihood of violence and unrest.

I argue that states understand this reality and will seek to apply preemptive repression as a

way to mitigate the negative eﬀects of economic shocks.

3.3 Economic Shocks, Violence, and State Repression

Within the literature on unrest, economic shocks have been consistently connected to

increases in violence. States facing an economic shock understand that this can lead to

instability. To help mitigate this negative outcome, states are likely to turn to increases in

repression. While the literature on economic shocks has discussed the connection between

shocks and instability, it has largely overlooked the relationship between shocks and repres-

sion. I connect negative economic shocks to increases in state repression and show how states

use repression as a tool to insulate themselves from the negative eﬀects of economic shocks.

The economic situation of the state often dictates the amount of unrest that particular

state experiences. For instance, many civil wars begin due to the “greed” argument which

states that rebels seek to capture some economic good that is currently under the control of

the state (Collier and Hoeﬄer 2004). Others have shown that economic shocks (deﬁned as a

decrease in economic growth of ﬁve percentage points) drastically increase the likelihood of

40

civil war in that state (Miguel et al. 2004). This ﬁnding holds when looking at the regional

level in Africa with negative economic shocks (now measured as nighttime light intensity)

leading to an increase in civil conﬂict (Hodler and Raschky 2014). Consistent with this

literature, recent ﬁndings suggest that other forms of economic shocks (here measured as

commodity price increases) also increase the risk of conﬂict (Dube and Vargas 2013)). Beyond

civil conﬂict, economic shocks also impact leader survival. Many have shown that economic

shocks (whether measuring it as decreases in GDP or decreases in per capita income) lead to

an increase in the likelihood of a coup (Galetovic and Sanhueza 2000; Kim 2016). Economic

shocks, therefore, have a detrimental impact on a country’s overall stability and on regime

survival. Economic shocks are also quite visible to both the regime and the population.

Leaders are aware of economic shocks as they are happening and can take actions to try to

preemptively reduce their negative impact economically. Beyond economic action, leaders

can also take steps to limit the other, non-economic, negative outcomes that often accompany

economic shocks.

One of the strategies most readily available to leaders looking to reduce the uncertainty of

the future is to employ repression preemptively (Nordås and Davenport 2013; Danneman and

Ritter 2014; Truex 2019). These studies show that when regimes are faced with perceived

or future potential threats, they will engage in preemptive repression as a way to insulate

themselves from this threat. Yet leaders do not simply employ repression without ﬁrst

weighing the costs and beneﬁts.

In order to use repression, a state must ﬁrst identify a

threat. Then, the beneﬁt of using repression (increased control and likelihood of regime

survival) must outweigh the cost. When faced with a threat and the belief that the beneﬁts of

using repression outweigh the cost, states will employ their repressive apparatus (Davenport

2007a). We see this in the repression literature with political dissent being a consistent

predictor of increased state repression as states see this as a credible threat (Carey 2010).

The ﬁnding that dissent leads to increased repression is strong enough within the literature

41

that Davenport calls it the “Law of Coercive Responsiveness” (Davenport 2007a). Here,

states see dissent as a large enough threat to justify using repression.

Beyond realized threats though, regimes also employ repression when faced with per-

ceived threats or the potential for future unrest such as large youth cohorts or neighbors

experiencing civil war (Nordås and Davenport 2013; Danneman and Ritter 2014). States

also attempt to preempt unrest by increasing repression prior to events or dates likely to

see increases in dissent (Truex 2019). These increases in repression come in the absence of

overt dissent as the repression is taking place preemptively. The idea behind this use of pre-

emptive repression is to reduce the ability or capacity of opponents to mount a coordinated

movements against the regime (Tilly 1978). Additionally, preemptive repression seeks to

lower opponents willingness to protest or take part in unrest by increasing the price of such

actions with increased preemptive attacks against physical integrity rights (Galtung 1969).

Recent work in Zimbabwe shows the psychological eﬀect this sort of action can have on

potential dissidents. In her research, Young (2019) ﬁnds that the psychological fear created

by physical integrity rights violations can severely restrict the likelihood a person takes part

in unrest. She contends that this use of repression can greatly reduce the amount of overt

dissent which in turn insulates the leader from possible regime change. Her ﬁndings show

that the fear treatment given to those in the study reduced dissent by 14-23%. Similarly,

participants who received the fear treatment reported a reduced likelihood of taking part in

six diﬀerent high-risk dissent acts. This shows the power of preemptive repression in stiﬂing

future dissent. It follows, then, that states will respond to economic shocks with repression

in an attempt to mitigate their negative impact (i.e. the likelihood of seeing a challenge to

their authority).

Economic shocks are known to increase the likelihood of dissent and violence by increasing

the grievances of the population and decreasing the opportunity cost of joining in violence.

States understand that shocks bring about this increased potential for unrest and will likely

42

deem it to be a large enough threat to their power to initiate preemptive repression. Because

of this, we should see an increase in preemptive repression as a result of economic shocks as

states attempt to insulate themselves from these negative outcomes. The use of preemptive

repression allows states to limit the ability of opponents to dissent as well as the willingness

of people to take part in such action by raising the cost of participation. This rise in the

cost of participation promotes fear within the potential participants and thus reduces the

likelihood of unrest. Thus, preemptive repression serves as a powerful tool for regimes facing

an economic shock and the potential for violence that comes with it.

Indeed, within the literature on state repression we see some evidence of this relationship.

Economic sanctions are often used as a way to punish regimes for their human rights abuses

and as a way to encourage them to reduce this abuse. However, the literature shows that

these actions often lead to increases in repression rather than decreases (Wood 2008; Peksen

2009; Peksen and Drury 2009).

In a way, these externally imposed economic challenges

function in much the same way as economic shocks. External actors decide to create a

negative economic situation within the state as a way to coerce that state into changing its

behavior. This externally produced negative economic situation often creates an environment

that threatens the stability of incumbents and increases the likelihood of political violence

due to the negative economic conditions brought about by these external actors (Peksen

2009). Due to this increased instability and potential for unrest, states respond to economic

sanctions by increasing repression in order to stabilize the regime and reduce the threat posed

by potential challengers (Wood 2008). In this way, negative economic conditions brought

about by externally imposed sanctions create an environment ripe for possible unrest. This

potential for instability leads states to increase repression in order to mitigate the perceived

threat. Economic shocks function in much the same way as they also lead to an increase in

the likelihood of instability. Therefore, states increase their level of preemptive repression as

a way to avoid the potential challenge to their regime that comes from negative economic

43

conditions.

3.3.1 Example: Zimbabwe

An example of such a sequence of events can be seen in Zimbabwe. Zimbabwe gained in-

dependence in 1980 under its leader Robert Mugabe. Mugabe began many diﬀerent economic

projects including land reform and also attempted to attract FDI. Following independence,

the Zimbabwe economy enjoyed moderate success and was seen by many as the second

strongest economy in southern Africa (Parsons 2002). However, this success ended quickly

due mainly to Mugabe’s corrupt and economically unsound policies (Dashwood 2000). The

catalyst came in November of 1997 on a day known as “Black Friday” where the country’s

currency was decimated and lost 74% of its value in four hours (Bond 2007). On top of this

economic collapse, the entire country lost power for eight hours furthering the unease of its

population. This event led to a distinct increase in human rights abuses within the country

as the government fought to maintain its control.

In 1997, Zimbabwe experienced a very low level of repression as ﬁgure 3.1 shows. Mu-

gabe’s desire to attract outside investment led to little state sponsored repression and created

what many thought was a healthy economy. This is contrasted by the following year (1998)

when Zimbabwe saw an increase in repression passing the threshold that some argue repre-

sents a repressive spell (Davenport and Appel 2014). We see, then, how the crisis of 1997

and “Black Friday” impacted the human rights situation of the state. An article in the New

York Times in July of 1998 pointed out that riots were rare in Zimbabwe but since De-

cember (which was only a month after Black Friday), police and demonstrators had clashed

numerous times. As the situation worsened, the government responded by imposing harsher

policies. This pattern of deteriorating human rights continued as 1998 brought about high

interest rates and increased inﬂation. In 1999, amid political disagreements, the World Bank

and the IMF ended its aid to the country. This led to a continued economic downturn and

44

caused the regime to persist in increasing its repressive behavior.

In 2000, following this

steep downward economic spiral, Zimbabwe again saw their repression scores increase as the

state continued its suppression of people’s rights. The situation ﬁnally began to improve

following the end of hyperinﬂation in 2008. Zimbabwe abandoned its currency and saw pos-

itive economic growth numbers for the ﬁrst time in more than a decade. This change in

economic fortune led to a corresponding decline in repression. Figure 3.1 below shows this

relationship graphically. In the late 1970’s, Zimbabwe saw high levels of repression as the

guerilla war was in full force. The level of repression then started to decline as Mugabe took

power and began his bid to attract foreign investment. This decline in repression came to an

abrupt halt following the events of Black Friday in late 1997. This economic shock then led

to a decade long increase in repression as the government clung to its grip on power. The

economic downturn was ﬁnally reversed in 2008 following positive economic policies and as

a result, Zimbabwe saw a reduction in its level of repression. While the economic shock that

happened in 1997 is not the only causal factor that inﬂuenced Zimbabwe’s use of repression,

I do believe it served as a catalyst for increased levels of repression that were continually

increased over the next decade due to a failure of the economy to right itself. It is only in

2008 when positive economic growth return to Zimbabwe that we see a change in levels of

repression.

45

Figure 3.1: Level of repression by year: Zimbabwe. Shaded area represents period of hyper-inﬂation.

The economic and human rights situation in Zimbabwe details how economic shocks and

downturns negatively aﬀect human rights practices. As the Zimbabwean economy began to

fall apart, the government of Zimbabwe responded with increasing levels of repressive behav-

ior. Continued economic decline led to a corresponding increase in human rights violations

as the government struggled to maintain control amid protests and unrest. This example

leads to my hypothesis:

Hypothesis One: Economic shocks are associated with an increase in state repression.

The eﬀect of economic shocks may be conditioned by the economic makeup of the state

in question. Previous ﬁndings suggest that increased economic development is associated

with lower levels of repression (Poe and Tate 1994; Davenport 1995). Poe and Tate (1994)

argue that richer countries will have less grievances and therefore will have less of a need to

46

implement state repression. Similarly, Davenport (2007a) says that richer countries can oﬀer

alternative mechanisms of control like social services. He also cites Fearon and Laitin (2003)

and points out that fewer resources may make repression less eﬃcient which would further the

need for state action. However, despite the above discussion, Davenport (2007a) argues that

the mechanism through which economic development inﬂuences repression remains unclear.

The idea of economic development and human rights has also been tied to the globalization

literature with some arguing that globalization improves living standards which decreases

the need for repression (Gelleny and Mccoy 2001). But what happens when economic shocks

come into the equation and disrupt the above mechanisms? I argue that in poor countries,

economic shocks will have little eﬀect. The population of these states likely already have

numerous grievances against the government suggesting that a economic downturn will be

viewed by the government as constituting only a minor increased threat. Similarly, even if

an economic downturn is viewed as a threat, these governments will have limited capacity to

increase their level of repression given the scarce resources at their disposal. Rich countries

however, will likely see these economic downturns as a looming threat to their power and

also have the resources to meet that threat with increased repression. The same mechanisms

that reduce repression in rich countries will cause these governments to view economic shocks

as a threat. In the midst of a crisis, living standards and the ability of the government to

institute alternative avenues of control are severely reduced. This increases the likelihood of

unrest. Knowing this, rich governments are likely to increase repression preemptively and,

due to their resources, have the capacity to do so.

Hypothesis Two: Rich states experiencing an economic shock will see higher levels of

repression than poor states experiencing a shock.

Overt dissent may also play a role in the relationship between economic shocks and re-

pression. The literature on repression and dissent has long tried to pinpoint how protests and

other forms of unrest inﬂuence government repression. There is, however, some theoretical

47

consensus within the literature. Most scholars agree that governments respond to dissent

with increased levels of repression (Lichbach 1987; Davenport 1995; Moore 2000; Carey 2006;

Nordås and Davenport 2013). However, despite the overwhelming theoretical agreement on

the relationship between dissent and repression, Ritter and Conrad (2016) point out that

the empirical evidence is lacking. They argue that this is due to the endogenous relation-

ship between the two. It may be that people who fear repression will choose not to dissent

(Fearon and Laitin 2003). There is also the issue of preemptive repression discussed in this

paper. This sort of repression can stiﬂe dissent before it happens (Nordås and Davenport

2013; Danneman and Ritter 2014;Truex 2019). These issues cause the response to realized

dissent to be unclear (Ritter and Conrad 2016). Ritter and Conrad (2016) go on to ﬁnd that

dissent only leads to more repression when governments were not expected to use repression.

Their logic is that if a state uses preemptive repression to quell a threat before it mate-

rializes the best course of action, given the preemptive repression failed, is unclear. More

repression may end the dissent but it may also increase it. Following an economic shock

however, governments are likely to be wary of any overt dissent as that dissent would occur

within a period of increased tension and uncertainty making it all the more threatening. As

argued above, governments are likely to increase repression preemptively in response to an

economic shock due to the potential threat these shocks pose. The combination of overt

dissent matched with an economic crisis forms the basis for an even more volatile situation

— one in which the government will be increasingly trying to maintain its’ control. There-

fore, in the face of an economic shock we can expect governments to increase their level of

preemptive repression in response to the increased threat posed to the regime. Additionally,

if overt dissent materializes, the impact of shocks and dissent on repression will increase as

governments struggle to maintain control. This logic leads to my ﬁnal hypothesis:

Hypothesis Three: The eﬀect of economic shocks will be greater in situations where overt

dissent is present.

48

3.4 Data

3.4.1 Dependent Variable

To measure my dependent variable, state repression, I use a latent variable that combines

information from diﬀerent sources of human rights practices such as the Political Terror Scale

(Wood and Gibney 2010) and Cingranelli-Richards scales (Cingranelli et al. 2010) and takes

into account the increasing standards of accountability in human rights reporting (Fariss

2014). This measure tries to account for the latent amount of human rights in a country.

The scale is normalized with a mean of zero and positive numbers signifying an increase

in respect for human rights. To make the measure more comparable to other measures of

human rights, I inverse the scale by multiplying scores by negative one. Therefore, the

inverted measure now has higher numbers referring to an increase in repression.

3.4.2

Independent Variables

For my independent variable, I ﬁrst measure economic shocks by looking at decreases in

GDP growth. Following (Miguel et al. 2004) I construct a measure of economic shocks that

takes the value of one if GDP growth decreases by ﬁve percentage points from the previous

year. I then lag this variable and include it in the regression equation. GDP growth however,

proves to be a rather broad measure of economic shocks. To help with this, I employ three

other measures of economic shocks. Two of these measures look to proxy income shocks by

using trade data while my ﬁnal measure uses data on inﬂation to capture income changes.

The ﬁrst of these come from (Bazzi and Blattman 2014). These authors use export

price shocks to proxy for income shocks. The data range from 1957-2008 and cover all

countries in Africa, the Middle East, Latin America, and Asia. A shock is calculated as the

“annual diﬀerence in each country’s log commodity export price index.” (7). This number is

then multiplied by the ratio of commodity export values to GDP in order to capture how

49

commodity dependent a nation is. Weighting the measure in this way helps to connect the

export price shock to its eﬀect on the economy. This measure is then normalized making it

have a mean of zero with a maximum value of 9.68 and a minimum of -6.68. Higher scores

here represent a positive economic shock. In other words, positive values represent a better

economic situation for the state and its population. To better capture the phenomenon I

am interested in, negative economic shocks, I also inverse this measure. With the inverted

measure, positive numbers now refer to more serious economic shocks. This new measure

maintains a mean of zero and reverses the minimum and maximum with -9.68 now the

minimum and 6.68 now the maximum. I then lag this variable to better ﬁt with my theory

of how economic shocks aﬀect state repression.

To further test how broad trade shocks inﬂuence repression, I also include a measure of

economic shocks which is based on the International Monetary Fund’s (IMF) Commodity

Terms of Trade (CTOT) data-set (Gruss and Kebhaj 2019). This data-set contains country

speciﬁc indices for 182 countries from 1962-2018. The measure used in this analysis is an

index created from the international price change of up to 45 commodities weighted using

commodity-level trade data. These commodities cover all major industries from energy and

metals to food and agricultural products. The weights come from net exports. Thus, the

economic interpretation of variations in the commodity term of trade measure is the gains

or losses that stem from international price shifts. Speciﬁcally the authors state that "a

one percentage point change in the commodity terms-of-trade index can be interpreted as a

change in aggregate disposable income equivalent to one percentage point of GDP."(Gruss

and Kebhaj 2019). While similar to Bazzi and Blattman (2014)’s measure of economic

shocks, the CTOT measure provides the beneﬁt of not being geographically restricted. This

measure of economic shocks ranges in value from -.65 - .31 with an average of 0 and a

standard deviation of .04. Both of these broad trade measures provide insight into how the

broader economy can aﬀect the income levels of the population.

50

Finally, I include a measure of economic shocks that seeks to proxy the impact economic

shocks can have on individual incomes. Following Houle et al. (2016) I use data from Reinhart

et al. (2019), which measures whether a country experiences an inﬂation crisis in a given year.
2 This measure takes the value of 1 when the annual inﬂation reaches 20%. The average

inﬂation rate from 1914-2009 was about 5% making the 20% mark a reasonable threshold

in deﬁning a crisis (Reinhart and Rogoﬀ 2011). The data cover 70 countries in Africa, Asia,

Europe, Latin America, North America, and Oceania from 1962-2010.

Below I provide the correlations between these three measures of economic shocks. As

table 3.1 shows the correlation between these measures is generally quite low with only the

price shocks measure and the CTOT measure showing any real correlation at .23. This makes

sense given that both measures seek to proxy the eﬀect exports have on income changes.

∆ GDP Growth Price Shocks

CTOT

Inﬂation Crisis

∆ GDP Growth 1.00
.03
Price Shocks
.03
CTOT
Inﬂation
.09

1.00
.23
.003

1.00
-.02

1.00

Table 3.1: Correlations Between economic shock measures

An example of how these diﬀerent measures of shocks interact can be seen in the ex-

ample of Honduras in 1990. Following the election of Rafael Leonardo Callejas Romero in

1989, the Honduran economy began to deteriorate as the IMF and World Bank prescribed

austerity measures began to be implemented. On top of this, the United States Agency

for International Development ended disbursements of grants as a signal of its unhappiness

regarding the former regime’s economic policies. These events created the backdrop for a

negative economic shock in 1990. Table 3.2 displays the values of each shock measure for

Honduras in 1990. It also shows the average and standard deviation of each measure as a

way to gauge how each measure quantiﬁes the magnitude of the economic shock. Across
2In the appendix, I show that the results hold when using the measure for currency crisis

as well. See table B.1

51

most of the measures, we see that 1990 represented a signiﬁcant, negative economic event in
Honduras.3 Though, this example does show the diﬀerence in how each measure captures

the magnitude of an economic shock and provides some support for the incorporation of all

four measures into the analysis.

Economic
Shock Measure
Mean / SD
Honduras:

∆ GDP Growth Price Shock

CTOT

Inﬂation Crisis

3.9% / 6.7%
-4.23

0.0 / 1.0
1.99

0.0 / 0.04
0.015

NA
Yes

Table 3.2: Economic shock measures: Honduras - 1990

The distribution of economic shocks can be seen in ﬁgure 3.2. At any given time, ap-
proximately 15-40 countries experience a shock.4 Not surprisingly, 2008 saw the largest

percentage of economic shocks with 36.5% percent of observations experiencing an economic

shock that year. 2010 experienced the lowest number of shocks with only 1.5% of observa-

tions registering an economic shock. Figure one shows the number of shocks experienced by

year. As this shows, the number of economic shocks has remained fairly steady throughout

years covered by the data with the exception of 2008 which saw an abnormally large number

of shocks.

3.4.3 Control Variables

For control variables, I follow the literature on state repression and employ many of

the same controls (Poe and Tate 1994). I ﬁrst include a measure of civil war (taken from

UCDP), GDP and population (taken from Penn World Tables). Civil war has been shown

to drastically increase levels of repression and therefore serves an important control (Hill

and Jones 2014). GDP and the population of a state have less predictive power when
3Note that the ∆ GDP Growth measure just misses the 5% change cutoﬀ for being

categorized as an economic shock.

4Using the measure of change in GDP growth.

52

Figure 3.2: Number of economic shocks by year. Note. Economic shocks measured as a change in GDP
growth of 5% or more.

looking at repression yet do prove important when understanding economic inﬂuences on

state repression and therefore are included. I also control for the presence of protests and

youth bulges as both of these factors tend to increase state repression (Nordås and Davenport

2013)). My measure of protests comes from the Banks Cross National Database. I sum all

the diﬀerent types of protest activities (protests, riots, etc) then create a dummy coded one

if any of these were present within the state in year t (Banks and Wilson 2017). To measure

the type of governance, I include a measure of democracy taken from Cheibub et al. (2010)

which is a dichotomous variable coded as one if a state is a democracy. Another common

measure of regime type comes from the Polity IV measure (Marshall et al. 2016). However,

due to issues with the Polity IV measure and repression pointed out by Hill and Jones

(2014) I chose to instead use the measure from Cheibub et al. (2010). As a robustness check,

53

I re-run the analysis using the Xpolity measure from (Vreeland 2008).5 Lastly, I include

a measure for the amount of independence the judiciary enjoys as this has been shown to

decrease the amount of repression within the state (Linzer and Staton 2015). It could also be

that governments use other mechanisms of control besides increasing repression. To account

for this, I include, as a robustness check, a measure of education spending taken from the

World Bank. This measure records the amount of education spending as a percent of GDP.

I provide the results of this analysis in table B.3 in the appendix.

3.5 Research Design

To model the relationship outlined by hypothesis one, I run OLS regressions with each

of my four main independent variables measuring economic shocks being regressed upon my

measure for state repression. Each model includes ﬁxed eﬀects to account for any country

speciﬁc factors not captured in my control variables and reports standard errors clustered
on country. I lag all four measures of economic shocks and my control variables.6 The data

range from 1963-2008 for model one using the dichotomous variable drawn from change in

GDP growth. When using the coding of economic shocks drawn from Bazzi and Blattman

(2014) for model 2, the data range from 1960-2008. For model three, the time-frame is

1963-2008. And ﬁnally, using the inﬂation crises measure, the data range from 1962-2010.

To account for the possible interactions mentioned in hypotheses two and three, I interact

my measures of economic shocks with both overt dissent and regime type. This analysis is
presented in tables and 3.5.7

5Results from using the Xpolity measure can be found in the appendix. Results remain
largely unchanged with only the "change in GDP growth" measure of economic shocks just
missing conventional levels of statistical signiﬁcance. See table B.2 in the appendix.

6I lag all control variables except civil conﬂict following Wright and Moorthy (2018) who
argue that civil war can suddenly change a regime’s repressive output and therefore should
not be lagged in the model.

7The relationship between economic shocks and repression could be nonlinear (with re-
gards to the price shocks and CTOT measures. The logic here being that large positive
shocks could cause unrest with due to people not feeling like they are receiving the bene-

54

3.6 Results

The results from table 3.3 show support for hypothesis one stating that economic shocks

increase the level of repression within a state. Across the four models, the control variables

behave largely as we would assume and are consistent across models. Civil war, dissent

and youth bulges all signiﬁcantly increase state repression as previous literature has found.

An independent judiciary, as we would expect, signiﬁcantly reduces the amount of state

repression seen. Democracy varies in its signiﬁcance but shows that more democratic states

will have less repression. Similarly, economic development varies in its signiﬁcance levels

but shows that more developed states have lower levels of repression. Moving to my key

independent variables measuring economic shocks we see support for the hypothesized rela-

tionship across all four measures. Model one shows that a state experiencing a decrease in

GDP growth of 5% or more see a corresponding increase in repression. Similarly, when using

Bazzi and Blattman (2014)’s coding of economic shocks, we see in model two that economic

shocks (represented as positive numbers given the inverse nature of the coding) lead to an

increase in state repression. This result again holds in models three and four when using the

index measure from Gruss and Kebhaj (2019) and the inﬂation crises measure from Reinhart

et al. (2019). Therefore, across all models, we see that all four measures of economic shocks

produce similar results. States experiencing economic downturns increase their repression in

an attempt to insulate themselves from the negative eﬀects such shocks bring.

ﬁts of an economic surplus. I test for this possibility but ﬁnd no evidence of a nonlinear
relationship. See ﬁgures B.1 and B.2 in the appendix.

55

Table 3.3: Eﬀect of economic shocks on repression

The substantive eﬀects of an economic shock shows that this ﬁnding proves important.

For instance drawing upon model two, when increasing from the 10th to 90th percentile

in shock severity, the model predicts a 127% increase in the level of repression assuming

no civil war or protests and the other variables held at their means. Similarly, I show the

change in predicted level of repression (based on model 4) assuming their is no civil war

or overt dissent (with all other variables at their means) in ﬁgure 3.3. For a state with no

civil war or overt dissent that is also not experiencing an inﬂation crisis, the predicted level

of repression is -.53. Assuming now that the same state is experiencing an inﬂation crisis,

56

the model predicts the level of repression to increase to -.33. This increase in the predicted

level of repression lends support for the expectation that an economic shock will bring about

increased repression as states look to insulate themselves from the dangers of such negative

economic downturns.

Figure 3.3: Predicted level of repression by economic shock. Results from table 3.3, model 4. Bars indicate
90 percent conﬁdence intervals

3.6.1 Do Economic Shocks Have an Interactive Eﬀect with Economic Develop-

ment and Dissent?

Hypothesis two argues that states with higher levels of economic development will react

to economic shocks diﬀerently than states with less economic development. Speciﬁcally, that

rich states have a greater reason to fear unrest in the presence of an economic shock as well

as have the resources to preempt this with repression. Therefore, we should see the eﬀect of

57

economic shocks on repression be greater in more developed states. To test this, I interact

my four measures of economic shocks with my measure of economic development (GDP per

capita). All other control variables remain the same as the previous analysis and are lagged

one year (with the exception of civil war). I again estimate OLS regression with ﬁxed eﬀects.

Table 3.4 shows the results from this analysis.

Table 3.4: Interactive eﬀect of GDP on economic shocks and repression

The coeﬃcient on the interaction term between economic shocks and GDP fails to reach

signiﬁcance across all four models. However, some support for hypothesis two can be found

when looking at this relationship graphically which serves as a better investigation of the

58

results especially given all of the interactions have at least one continuous variable (Brambor

et al. 2006). In ﬁgure 3.4 (from model two) we see the interactive eﬀect of economic shocks

by level of GDP. To graph this, I calculate the interactive eﬀect (slope) of the coeﬃcient

for the interaction between economic shocks and GDP at diﬀering levels of GDP to see the

eﬀect across the values of economic development. The point marks and conﬁdence intervals

on the graph represent the eﬀect of the interaction term at diﬀering levels of GDP per capita.

Here, for below average levels of GDP (the average for the sample is just under 8) the eﬀect

on repression of the interaction between economic shocks and GDP is insigniﬁcant. This

suggests that in lower developed countries, the eﬀect of economic shocks on repression is

low. However, at above average levels of GDP we see the interactive eﬀect of shocks and

GDP on repression as positive and signiﬁcant. To quantify the eﬀect of economic shocks at

these high levels of development, I calculate the ﬁrst diﬀerences of an economic shock. To

get these ﬁrst diﬀerences, I calculate the predicted level of repression with and without an

economic shock in a state with a high level of economic development (represented as a 9 on

the GDP per capita measure) that is not experiencing a civil war and has no overt dissent.

The predicted level of repression for the above state without an economic shock is -.12. This

is contrasted by the predicted level of repression for the same state that is experiencing a

shock which is -.025. This .1 increase in the predicted level of repression is signiﬁcant at the

.05 level (p-value .006). Thus, the analysis lends support to the idea that rich states react

to economic shocks by increasing their level of repression to insulate themselves from the

potential threat these shocks represent.

59

Figure 3.4: Conditional eﬀect of economic shocks and GDP per capita across the range of development. Bars
indicate 90% conﬁdence intervals.

I turn now to hypothesis three which contends that the eﬀect of economic shocks on

repression is moderated by the presence of realized dissent. The hypothesis predicts that

the eﬀect of an economic shock on repression will be greater in the presence of overt dissent.

The results for this interaction are shown in table 3.5. Here we see mixed support for the

proposed relationship. Models one and three — which measure economic shocks as a 5%

change in GDP growth or using the Commodity Terms of Trade index, respectively — show

a signiﬁcant interaction coeﬃcient between economic shocks and dissent. The interaction

term in models two and four — using the price shock measure from Bazzi and Blattman

(2014) and data on inﬂation crises from Reinhart et al. (2019), respectively — fails to reach

conventional levels of statistical signiﬁcance. It should be noted that the coeﬃcient on the

interaction term in models one and three are statistically signiﬁcant but have the opposite

60

sign. Given this, I show both interactions graphically and discuss the results.

Table 3.5: Interactive Eﬀect of Dissent on Economic Shocks and Repression

The substantive eﬀect of this interaction is shown in ﬁgure 3.5. Here we see that the

eﬀect of an economic shock on repression is insigniﬁcant when overt dissent is present but

the ﬁrst diﬀerence is positive and signiﬁcant when there is no overt dissent. This means

that the interaction between economic shocks and dissent produces a positive and signiﬁcant

increase on repression when there is no overt dissent present. This gives some support to

the idea that economic shocks lead to preemptive repression. The eﬀect of economic shocks

61

on repression is only signiﬁcant when no overt dissent is present. However, it seems that

this interaction is muddied when overt dissent is present. When dissent is present, the

relationship between economic shocks and dissent on repression is unclear as the dissent

may be driving the repression or the repression driving the dissent similar to what Ritter

and Conrad (2016) argue.

Figure 3.5: Predicted change in repression when experiencing an economic shock with and without the
presence of overt dissent. Bars indicate 90% conﬁdence intervals.

The results from model three (shown in ﬁgure 3.6) display a slightly diﬀerent outcome. In

the table, model three shows the coeﬃcient on the interaction between economic shocks and

dissent is positive and signiﬁcant. Here we see that the interactive eﬀect of economic shocks

on repression is positive with or without the presence of overt dissent. And that the amount

of repression seen when overt dissent is present is higher than the level of repression with

no dissent present. This gives some support to hypothesis one stating that economic shocks

62

will increase preemptive repression (i.e. when no overt dissent is present). It also gives some

support to the hypothesis arguing that overt dissent will increase the amount of repression

seen in the context of an economic shock. The ﬁrst diﬀerences show that in both cases

(with or without dissent) an economic shock increases the level of repression. When no overt

dissent is present, an increase from the 10th percentile to the 90th percentile of economic

shock severity corresponds to a .024 increase in repression (p-value .00). Similarly, when

dissent is present, an increase in economic shock corresponds to a .076 increase in repression

(p-value .001). Thus, it seems that economic shocks trigger increases in repression as states

move to avoid the potential unrest these economic downturns can bring and this relationship

holds when no dissent is present. The presence of dissent increases the eﬀect of an economic

shock on repression.

Figure 3.6: Predicted level of repression with and without dissent by size of economic shock. Bars indicate
90% conﬁdence intervals.

63

3.7 Conclusion and Discussion

The literature connecting economic shocks to state repression is surprisingly underde-

veloped. As the data show, economic shocks are fairly common across time and can often

have quite negative outcomes on the security situation of the state. By connecting economic

shocks to state repression, I help ﬁll a gap in the literature and further our understanding

of when states decide to impose increased levels of repression. Additionally, I connect two

literatures that have, until now, developed separately from each other. Economic shocks

have been connected to many diﬀerent forms of unrest and states understand this connec-

tion. Therefore, it makes sense that they would want to avoid these challenges and often do

so through the use of preemptive repression. I also ﬁnd evidence that this relationship can

be moderated by the level of economic development within the state. Countries with low

levels of development are less likely to see economic shocks lead to increases in preemptive

repression. This ﬁnding is contrasted with richer countries who see a signiﬁcant increase in

preemptive repression when faced with an economic shock. This ﬁnding shows the complex

nature of the relationship between economic development and repression. As previously dis-

cussed, economic development has largely been found to reduce the amount of repression as

these states are better able to use alternative mechanisms of control such as social spend-

ing. The, until now, unforeseen consequence of this however, is the negative eﬀect economic

shocks have on this relationship. Economic shocks directly eﬀect the ability of governments

to continue these alternative mechanisms of control causing them to fear potential unrest.

This, in turn, causes these regimes to resort to increases in preemptive repression in the hope

of solidifying their control. Finally, I show that overt dissent also impacts this relationship.

The security situation of a state undergoing an economic shock is precarious as the literature

on violence and unrest has shown. Given this, I ﬁnd that economic shocks increase the level

of preemptive repression in the absence of dissent and that this eﬀect is ampliﬁed when faced

with overt dissent. While the results summarizing the relationship between shocks, dissent,

64

and repression are somewhat mixed, the overall ﬁnding that economic shocks lead to an

increase in preemptive repression prove to be robust and an important contribution to the

literature.

The ﬁndings presented above provide an initial look into how economic factors can impact

human rights but many questions yet remain. For one, what is the long-term eﬀect of

economic shocks on state repression? We know that once repression is initiated, it is often

diﬃcult to reverse (Davenport and Appel 2014). Therefore, is the eﬀect of these economic

shocks on state repression a long-term change in behavior or a short-term measure taken in

response to a speciﬁc threat? This project takes a ﬁrst step in connecting economic shocks

and repression which, in turn, opens up many new avenues for future research.

65

CHAPTER 4

ESCAPING PUNISHMENT: PUBLIC RELATIONS FIRMS AND STATE

REPRESSION

4.1

Introduction

Can human rights abusing states avoid being punished for their actions? We know that

countries with poor human rights records often spend large amounts of money on public

relations (PR) ﬁrms based in the United States to improve their image. A report by the

Center for Public Integrity found that between 2010 and 2015, the 50 countries with the

lowest scores on the Fragile States Index (and those with some of the worst human rights

records) spent $168 million on PR in the U.S. Additionally, the same report found that

between 2013 and 2015, this spending increased by 40 percent (Quinn 2015). But does this

strategy work?

The link between human rights violators and these PR ﬁrms can be clearly seen through

the case of Bahrain. In 2011, just as a domestic protest movement began to take shape,

the government of Bahrain signed a contract with Qorvis, a large PR ﬁrm in Washington

D.C. The deal, worth over $40,000 a month, led to Qorvis working to improve Bahrain’s

image including writing a pro-Bahrain column in the Huﬃngton Post. Qorvis also organized

a "youth delegation" to visit the United States. It said these youth represented the "silent

majority that is very moderate but wants to see limited and stable reform" (Elliott 2012).

While it was said that the delegation represented the pro-reform community, they spent

most of their time criticizing the protests and arguing that the protesters were "a fringe

group of rioters and vandals," according to Cole Bockenfeld an oﬃcial with the Project on

Middle East Democracy who met with the group. The youth delegation also attended a panel

discussion on Bahrain at Freedom House. Here, they challenged the speakers arguing that

the protesters were violent and that the panel was excluding their (the "silent majority")

66

voices. The contract with Qorvis outlines the PR ﬁrm’s work as "facilitated and coordinated
media and press activity" as well as drafting "opinion editorials" (FARA 2012). 1 These

actions, designed to cover Bahrain’s human rights abuses, illustrate how states can use PR

ﬁrms to mitigate the negative consequences of these abuses.

This paper seeks to analyze whether this strategy works. Can governments avoid the

negative consequences of abusing human rights by employing PR ﬁrms to lobby on their

behalf? To answer this question, I ﬁrst specify two ways in which human rights abusing

states can be punished for their behavior: sanctions and the allocation of foreign aid.

I

argue that PR ﬁrms are employed to improve the image of the client country amongst the

general population and directly lobby policymakers which can reduce the likelihood that

these regimes will be held accountable for their abuses. To make this argument, I ﬁrst give

an overview of lobbying and how it can aﬀect policy. I then connect lobbying to the literature

on human rights abuses, sanctions, and foreign aid showing the utility of using a PR ﬁrm.

In doing so, I describe how this can improve the image of the client country and directly

inﬂuence policymakers thereby reducing the likelihood of sanctions or a reduction in foreign

aid. Following this discussion, I present empirical evidence and end with a discussion of the

results.

This research contributes to the literature in a number of ways. First, it brings an

important player into the research on human rights and international punishment. Lobbying

has already been shown to have a direct eﬀect on the amount of foreign aid a country

receives (Licht 2010; Montes-Rojas 2013). The literature argues that this aid can be used

as a coercive measure to induce a change in human rights behavior by oﬀending states.

Therefore, understanding how lobbying can moderate this relationship and allow abusive

states to escape punishment proves an important contribution. The idea here is that if states

can avoid being punished for their actions by simply hiring a lobby ﬁrm, the usefulness of such
1A copy of the "Nature of Services" disclosure given to the Department of Justice by

Qorvis from this contract can be found in ﬁgure C.1 in the appendix.

67

tactics as tying aid to human rights fails to have as strong of an eﬀect. Additionally, I show

that lobbying can impact how aid is delivered. Governments either give foreign aid directly

to the government or do so through other channels like NGOs and other local actors. This

type of aid "bypasses" the government. Due to this, governments seek to limit the amount of

bypass aid because it eliminates their control over the money. I show that lobbying can put

pressure on lawmakers to reduce the amount of bypass aid and give the money directly to

the recipient government. This is an important contribution to the aid literature and has a

direct impact on policy as it exposes the power lobbying has on the allocation of foreign aid.

Beyond this, the paper also casts light on another mechanism of punishment: sanctions. Like

aid, sanctions are often used to coerce a change in behavior. However, if states are actively

lobbying against the imposition of sanctions, the eﬀect of human rights abuses in triggering

such actions could be muted. Introducing lobbying to this relationship helps to identify the

role human rights have on sanctions and how lobbying can aﬀect this interaction. Lastly,

this paper contributes directly to the growing literature on foreign lobbying. The American

politics literature has long understood the power of domestic lobbying but the inﬂuence of

international actors is just now coming to light. This paper engages with this idea and shows

how international actors can directly inﬂuence the foreign policy of the United States for

their own beneﬁt.

It should be noted that while this chapter is speciﬁcally about lobbying and U.S. foreign

policy, the results are by no means limited to the U.S. For instance, a recent report by

the British media watchdog Spinwatch described a "highly sophisticated, aggressive and

clandestine UAE political lobby has emerged to head oﬀ the threat of democracy in the
Middle East."2 The report goes on to show how the United Arab Emirates spent millions

of dollars to inﬂuence UK policy. Similarly, the Corporate Europe Observatory (CEO), a

research group based in Belgium, published a recent report highlighting foreign lobbying in
2http://spinwatch.org/index.php/issues/lobbying/item/6003-the-uae-lobby-subverting-

british-democracy

68

the European Union. In their report, the CEO outline the lobbying practices and inﬂuence

of 18 non-EU member states all of which have poor human rights records. The report

argues that foreign lobbyist’s work involves "laundering the reputations of dictators, seeking

lucrative trade and investment deals, pushing back against sanctions"(2). Many of the

countries mentioned in the report are the same ones outlined in this paper when discussing

U.S. lobbying such as Azerbaijan, the UAE, and Ukraine. Uncovering the impact of this

lobbying proves a diﬃcult process in the EU though as they do not have strict lobbying

laws like the U.S. For this reason, the report argues for the passing of such laws that will

provide transparency to a shady corner of EU politics. Thus, the ﬁndings of this chapter go

beyond the U.S. and are directly applicable to other countries like the UK and to broader

organizations like the EU.

4.2 Lobbying

Lobbying in the United States by foreign governments is not a new phenomenon. Regis-

tering such activities came about in 1938 as a response to fears over possible German agents

spreading propaganda in the U.S. in the lead up to WWII. This led to the Foreign Agents

Registration Act which speciﬁed that anyone attempting to inﬂuence U.S. policy on behalf of

a foreign client must register such activities with the Justice Department. It’s important to

note that lobbying for business interests need not be reported to the Department of Justice

(Cushman 2018). Table 4.1 shows the top ten countries for each decade as measured by the

number of lobbying contracts that year. As the table shows, many of the same countries

show up in the list over time. For example, Canada, the UK, Thailand, and India all show

up at least twice.

As a part of registering to the Justice Department, the U.S. based PR ﬁrm who is engaged

in lobbying must disclose the contract details and disclose the nature of all political activity

and any materials that may be distributed (Quinn 2015). This "Nature of Services" portion

of the contract often lists services provided by the PR ﬁrm in very general terms. While

69

Rank
1
2
3
4
5
6
7
8
9
10

1990
Germany
UK
Australia
France
Canada
Jamaica
Israel
Austria
Thailand
India

2000
Japan
Mexico
Saudi Arabia
Canada
UK
Turkey
Kazakhstan
Singapore
Switzerland
Gabon

2010
Canada
Iraq
Japan
UAE
Mexico
Georgia
South Korea
Turkey
India
Thailand

Table 4.1: Top Ten Countries by Number Lobbying Contracts

some reports list speciﬁc targets or actions of their lobbying — such as a 2012 report by

a ﬁrm representing Mexico which listed a proposed amendment to the Foreign Operations
Appropriations bill as its’ focus — most list the services provided in more general language.3

As an example, take the contract between the Government of the Kingdom of Morocco and

the lobbying ﬁrm Gabriel Company, LLC in 2006. As ﬁgure 4.1 shows, the ﬁrm was hired

to contact government oﬃcials and members of Congress to discuss terrorism, migration,

and human rights, amongst other things. The ﬁrm was paid $100,000 for the six-month

period of the report. This represents a rather generic reporting of activities with terms such

as "government oﬃcials" encompassing a wide swath of potential connections which make

discerning the exact nature of the lobbying diﬃcult.

A similar example, shown in ﬁgure 4.2, displays the contract between the lobbying ﬁrm

Cassidy & Associates, Inc. and the government of Equatorial Guinea. Here the lobbying

ﬁrm provided information on U.S. assistance programs and advised U.S. government oﬃcials

on the "progress that the government of Equatorial Guinea made on various key issues,".

The ﬁrm was paid $510,000 for the six-month period of this contract. Here we again see the

generalized nature of the language with "various key issues" representing all sorts of possible

talking points. These examples display the standard in regards to the amount of information

3The full text of this contract can be found in ﬁgure C.2 in the appendix.

70

Figure 4.1: FARA Report - Morocco: 2006

provided about the activities of the PR ﬁrm on behalf of the foreign principle within the

FARA data. While providing important information, it should be noted that enforcement

of reporting is often lax. A recent investigation by the Project on Government Oversight

found that enforcement in generally lacking with only 4 criminal cases of non-compliance

being brought forward from 2004-2014 (Freeman and Dennett 2014). Thus, the results of

this analysis, if anything, underestimate the inﬂuence of lobbying.

Figure 4.2: FARA Report - Equatorial Guinea: 2006

The results of such lobbying can be quite substantial. For example, Freeman (2009) ﬁnds

that states who lobbied in the U.S. were able to secure more foreign aid as a result of this

lobbying. Montes-Rojas (2013) ﬁnd similar results and argues that lobbying can aﬀect how

71

policymakers think about the United State’s relationship with the lobbying country. They, in

turn inﬂuence ﬁrms in an attempt to increase investment in these states. The result is more

foreign direct investment in states who lobby. This re-framing of the relationship between a

country and the U.S. proves to be a key aspect of the inﬂuence lobbyists have. Newhouse

(2009) points out that lobbyists can work in a way that is diﬀerent from traditional diplomats

and can operate more eﬀectively in Washington. The reason for this being that Washington

is a town of insiders and often these lobbying ﬁrms can operate more successfully than

traditional diplomats due to their connections and understanding of Washington politics.

This is an important point that must not be missed. When a foreign country sends its’

diplomatic corps to Washington, it is sending representatives from that country. For instance,

the Nigerian representatives here in the U.S. are Nigerian. They may have broad contacts

within Washington and may be well connected, but they are not the "insiders" that are

available through PR ﬁrms. Take the PLM group (discussed in detail later). This group is

made up of former members of Congress. These insiders have access to friends and associate

networks that most (if not all) foreign diplomats lack. Thus, when you hire a lobbying ﬁrm,

you are not hiring more diplomats, you are hiring the friends and former colleagues of the

very people you are trying to inﬂuence.

Importantly, beyond inﬂuencing policymakers as just discussed, lobbying ﬁrms can also

impact the general population of the United States. Lee (2007) ﬁnds that states with active
lobbying contracts see a signiﬁcant increase in media coverage.4 Carrying out media cam-

paigns represents an important way in which PR ﬁrms can inﬂuence foreign policy. Through

media campaigns and work with think-tanks, lobbyists can change the narrative and opinion

surrounding a country and thereby inﬂuence the policy towards that state. This inﬂuence

comes as the information provided through these media campaigns can help sway public

opinion about certain issues. This public opinion can then help pressure policymakers into
4Lee looked at coverage from the New York Times, Washington Post, ABC, NBC, CBS

and CNN.

72

making decisions beneﬁcial to the foreign principal who hired the PR ﬁrm.

This inﬂuence becomes especially important to regimes who repress their citizens and

generally ignore human rights. It is quite common for these states to hire lobbyist in Wash-

ington to help clean up their image. A 2015 report from the Center for Public Integrity

shows that the 50 worst human rights abusing states spent $168 million on lobbying since

2010 (Quinn 2015). And this trend is increasing with the number of PR contracts represent-

ing countries with poor human rights records increasing 40% from 2013-2015. According to

the Sunlight Foundation’s Bill Allison, lobbying ﬁrms serve "like renting a diplomatic corps"
for the governments they represent (Itkowitz 2014).5

Who do these ﬁrms lobby? The audiences of these PR ﬁrms can be quite diverse. A

FARA report from 2015 shows that Bangladesh hired two PR ﬁrms tasked with providing

"public relations services on behalf of the foreign principal to academics, news outlets, and

other individuals in the United States" (FARA 2015, 13) A similar contract between the

Sri Lankan government and a PR ﬁrm dictates that the PR ﬁrm "will develop a strategic

communications plan and conduct outreach to members of Congress and State Department

oﬃcials with the purpose of raising awareness of Sri Lanka’s strategic importance to the U.S."

(FARA 2014, 186). Just with these two reports we can see the breadth of inﬂuence lobbying

can have from the broad public through media campaigns to members of Congress and the

State Department. Through this broad inﬂuence, lobbying can have a signiﬁcant inﬂuence

on U.S. policy. These channels also highlight the diﬀerent avenues scholars have proposed

for how lobbying inﬂuences policy: direct lobbying, indirect lobbying, coalition-building,

and monitoring (Cushman 2018). Direct lobbying represents lobbyists meeting directly with

(or setting up meetings for their foreign principals) members of Congress, State Department

oﬃcials, federal agencies, or other policy-makers. Indirect lobbying seeks to inﬂuence through

op-eds and other press releases as well as grassroots movements who will speak on behalf of
5The Sunlight Foundation is a national non-partisan nonproﬁt focused on government

accountability.

73

the foreign government. Coalition-building seeks other, sympathetic nations or reaches out

to industry that may have similar goals. Finally, lobbying ﬁrms can monitor policy-makers

decisions and opinions on issues and report back to the foreign principal (Cushman 2018).

Before connecting this inﬂuence to foreign policy decisions, its important to outline how

states punish regimes with poor human rights records.

4.3 Punishing Human Rights

Scholars have long studied how states are punished for their human rights abuses. One

consistent theory is the eﬀect of international attention on behavior. Often, international

advocacy networks can help bring attention to human rights abuses (Keck 1998). For ex-

ample, international human rights organizations such as Amnesty International and Human

Rights Watch have routinely criticized states for their human rights practices. Often called

“naming and shaming” these organizations use public statements as a way to draw attention

to a state’s human rights abuses (Franklin 2008; Hafner-Burton 2008).

NGOs and other international organizations use naming and shaming as a way to bring

negative attention human rights abusers. This negative attention pressures other states

to then bring about a behavioral change from the targeted state often through the use of

sanctions or the allocation of foreign aid. We know that countries in general seek to maintain

international and domestic legitimacy and want to reduce the pressure put on them by other

states (Krain 2012). Therefore, by bringing attention to human rights abuses, these groups

hope to pressure governments to act in a way that will cause the regime to improve its’

behavior. This literature highlights the importance of information and public opinion on

policy formation. Here we see NGOs and others providing crucial information about human

rights abuses which, in turn, can lead to increased pressure on policymakers to act to punish

such behavior. However, if such information campaigns inﬂuence policy, it stands that similar

campaigns carried out on behalf of foreign governments by high-powered PR ﬁrms could also

impact policy, albeit in a way that directly opposes the outcome desired by human rights

74

organizations.

Similar to the work on NGO shaming, recent work has explored the eﬀect media attention

can have on punishing human rights violators. Peksen et al. (2014) ﬁnd that media attention

on human rights abuses can lead to sanctions being applied. They argue that the media can

highlight human rights abuses causing the public to push for action. Leaders, fearful of the

domestic cost on inaction, then employ sanctions to appease their domestic constituents and

interest groups. Thus, lobbying can directly engage the key players by inﬂuencing both key

players in the U.S. Congress, state department, and other U.S. agencies as well as the general

public through media and other campaigns.

The punishment of human rights abuses begins with identifying those in need of pun-

ishment. The above discussion shows how NGOs and the media can shine the spotlight on

human rights abusing states and use that negative attention to pressure governments into

punishing these oﬀenders. However, PR ﬁrms are hired to do the exact same strategy for

their client states. The only diﬀerence is that these ﬁrms seek the exact opposite outcome

than the shaming and negative media attention. These ﬁrms seek to bolster the image of

their client state and reduce the pressure put on policymakers to punish these regimes. As

the example of Bahrain in the introduction showed, actions by these PR ﬁrms can directly

inﬂuence the ability of human rights groups to publicize human rights abuses as the PR ﬁrms

work to counter such claims with an alternative narrative. Below, I outline two common

ways governments punish other states for their human rights abuses and then show how

lobbying can help foreign states avoid these punishments.

4.3.1 Sanctions

Economic sanctions are often implemented in an attempt to coerce states into changing

their behavior. The onset of such sanctions have been explained by looking at economic

interdependence (Hafner-Burton et al. 2009) and regime type (Goenner 2007) amongst other

75

factors. There exists a disagreement over how relations between the sender and receiver

states impacts the onset of sanctions with some arguing that it lessens the likelihood of

sanctions (Drury 2000; Drezner 1999) while others say it increases the likelihood sanctions

will be employed (Lektzian and Sprecher 2007). Given the relative frequency in the use of

sanctions, it is surprising to ﬁnd the literature on sanctions and human rights shows they

are often of limited value in changing the behavior of abusive states (Wood 2008; Peksen

2009). However, sanctions do oﬀer the sender state (especially the leader of the sender state)

certain beneﬁts even if they fail to alter the behavior of the human rights abusing regime.

Speciﬁcally, sanctions can serve as a international or domestic symbol for the sender state,

or can boost the domestic appeal of the sender state’s leader and act as a display of "strong

leadership" by threatening or imposing sanctions against human rights abusing countries

(Lindsay 1986; Whang 2011). This research empirically tests what Galtung (1967) calls the

"expressive function" of sanctions — that they help the sender nation send a clear signal of

disapproval and allow the sender state to avoid seeming complacent even if not serving the

instrumental purpose of changing behavior. Similar research has found that approval rates

and level of unemployment can impact the use of presidential sanctions (Drury 2001). This

shows that the domestic audience is an important player in deciding who is sanctioned for

their behavior and who is not. Due to this, research has also found that the media can be a

strong driver of economic sanctions.

The domestic nature of sanctions should not be missed as it forms the basis for how

lobbying can aﬀect this relationship. Expressive sanctions and those inﬂuenced by their

domestic appeal are aﬀected less by the actions of the oﬀending state and more by the

pressures and incentives of the policy-makers in the sending country. This opens the door

for sanctions to be greatly inﬂuenced by lobbying on behalf of the oﬀending state.

76

4.3.2 The Allocation of Foreign Aid

Past research has consistently connected U.S. foreign aid to human rights practices (Car-

leton and Stohl 1985; McKinlay and Little 1977). Though, the nature of this relationship

has been debated. Some argue that foreign aid is determined by strategic or economic inter-

ests with human rights being of little inﬂuence (Alesina and Dollar 2000; Neumayer 2003b).

Others, however, argue that human right do in fact inﬂuence aid allocation with states seeing

their foreign aid cut due to human rights abuses (Lebovic and Voeten 2009; Nielsen 2013).

Adding nuance to this line of research, Cingranelli and Pasquarello (1985) argue that U.S. for-

eign aid should be separated into two separate choices which both provide alternate signals.

First, the decision whether to provide aid to a country should be thought of as the ﬁrst-stage

of this process. Then, how much aid (assuming the ﬁrst decision was to grant aid) should

be allocated proves to be the second-stage. They ﬁnd that economic development inﬂuences

the ﬁrst stage while human rights practices aﬀects the second stage with improved human

rights practices leading to more aid. Further studies found that human rights practices were

indeed important in the allocation of aid but that political and strategic concerns often had

more inﬂuence. This ﬁnding has largely been attributed with the Cold-War era as recent

studies have found that in the post-Cold War time frame, human rights have become more

inﬂuential while security and strategic concerns have lessened (Demirel-Pegg and Moskowitz

2009). However, research suggests that human rights considerations still prove less inﬂuential

than economic and strategic inﬂuences (Alesina and Dollar 2000; Lebovic 1988; Neumayer

2003b). Recent research has begun to look at human rights and aid in more nuanced terms.

For instance, Esarey and DeMeritt (2017) argue that the relationship between the donor

and recipient state matters in how human rights abuses aﬀect aid. When the donor state

receives a direct political advantage from the aid (i.e. military advantage or perhaps some

domestic audience beneﬁt) the donor state will overlook human rights abuses. If, however,

the aid is more symbolic and largely for humanitarian purposes, donor state are more likely

77

to punish recipient states for human rights abuses. Also important is the distinction made

between economic and military aid. In the mid-1970s, the U.S. Congress passed a law linking

U.S. foreign aid to the human rights practices of the recipient government (Cingranelli and

Pasquarello 1985; Poe and Meernik 1995). This law, Boutton and Carter (2014) argue, had

a greater eﬀect on military aid rather than economic aid due to military aid’s direct relation-

ship with a state’s repressive capabilities. Beyond this, Boutton and Carter (2014) actually

ﬁnd that human rights abusing states actually receive more economic aid. They argue that

this is the case due to economic aid being used as more of a "carrot". This, coupled with the

fact that it may be easier for the U.S. Congress to send economic aid to regimes for strategic

reasons despite their human rights records, help to contextualize this result. Beyond this

distinction, others have found that diﬀerent types of aid are inﬂuenced by diﬀerent types

of human rights abuses. Neumayer (2003a) ﬁnds that military aid allocation is negatively

aﬀected by physical integrity rights violations while economic aid allocation is negatively

aﬀected by political and civil rights violations. The argument given by Neumayer (2003a) is

that these political and civil rights are more diﬃcult to discern with regards to the source

of the behavior. That is, political and civil rights violations may come as a direct result of

government strategy but may also be a more secondary result of poor governance tied to low

GDP or other broader metrics. Physical integrity violations, on the other hand, are more

easily tied to a coherent government strategy. Thus, since military aid is so closely tied to

the repressive apparatus of a state, it makes sense that this type of aid is more aﬀected by

physical integrity rights violations. Similarly, economic aid is aﬀected more by political and

civil rights as donor states seek to use this aid to coerce states into improving their human

rights practices.

There also exists, however, a domestic component to foreign aid. Many studies have

shown that aid allocation is directly aﬀected by the domestic politics of the donor state

(Alesina and Dollar 2000; Fleck and Kilby 2001; Irwin 2000). Speciﬁcally, Milner and Tingley

78

(2009) ﬁnd that policy-makers listen to both their constituents and interest-groups when

deciding how to allocate foreign aid. It is important to note that lobbying done on behalf of

foreign entities are not included within this idea of interest groups. Therefore, their inﬂuence

on foreign aid has been largely unstudied. The one notable exception to this is Pevehouse

and Vabulas (2019). Here, Pevehouse and Vabulas analyze the eﬀect lobbying has on human

rights scores. Speciﬁcally, they ﬁnd that State Department scores are signiﬁcantly lower

than Amnesty scores when lobbying is involved. They ﬁnd strong evidence that lobbying by

foreign governments inﬂuences State Department oﬃcials causing these scores to be biased

in a way that beneﬁts the lobbying country. In drawing out the implications of this ﬁnding,

Pevehouse and Vabulas argue that lobbying could also aﬀect policy outcomes like foreign

aid and sanctions. However, they do not provide a test of this empirically. Drawing on this

insight, I argue that lobbying can aﬀect the domestic audience of the donor state (in this

study, the U.S.) as well as the policy views of government oﬃcials and thereby aﬀect how

human rights aﬀects aid allocation and the likelihood of being sanctioned. Below, I outline

my theoretical expectations.

4.4 Lobbying and the Decision to Punish

Foreign policy decisions within the U.S. government (in this case sanctions and for-

eign aid) are inﬂuenced by a number of government agencies, interest groups, and other

actors. Sanctions, for instance, are issued by either the executive or legislative branches;

are maintained and enforced by the State Department, and are most often implemented by

the departments of Treasury and Commerce. Foreign aid is largely controlled by Congress,
though, the Executive branch has some inﬂuence. 6 The implementation of this aid however,

is carried out through twenty diﬀerent agencies. Therefore, we see how sanctions and foreign

aid both represent a complex web of inﬂuence and implementation. It is into this web where

the similarly complex networks of political connections found amongst lobbying ﬁrms can

6See a recent example of President Trump ending aid to Pakistan: (BBC News 2018)

79

function best. As Newhouse (2009) points out, lobbyists can function much diﬀerently than

a traditional diplomat. They are often former government oﬃcials themselves. The title of

a 2016 investigation by Politico says it all. The article, titled "Want to be a ‘foreign agent’?

Serve in Congress ﬁrst" outlines how many former Senators and House members now work

for powerful lobbying ﬁrms representing foreign countries. These high proﬁle ﬁgures, such

as former House Speaker Dennis Hastert (R-Ill.) and former CIA director and longtime

House member Porter Goss (R-Fla.), provide foreign governments something they could not

otherwise obtain — connections, insight, and inﬂuence. Those governments then use this to

directly aﬀect U.S. foreign policy.

The decision to sanction a country or to end its’ foreign aid due to human rights abuses

stems largely from the strategic calculus of policymakers from the donor state. I argue that

lobbying can inﬂuence these policymakers through two primary channels — by inﬂuencing

the general population who then pressure policymakers and by providing these policymakers

with key information regarding their client states.

The public can inﬂuence policymakers and foreign policy in general by demanding action

on certain issues. Due to their desire to be re-elected, these oﬃcials are likely to heed such

calls for action from their domestic constituents. The research on the public’s inﬂuence on

foreign policy shows that over time, oﬃcials are becoming more sensitive to the public’s

opinion on foreign policy (Powlick 1991). Yet, the public’s interest in foreign policy issues

is often characterized by what is disseminated through the media. In this way the media

has long been shown to have an eﬀect on foreign policy (Gilboa 2005; Baum and Potter

2008; Robinson 2011). Dubbed the "CNN eﬀect" this argument states that public opinion

is often inﬂuenced by what the news highlights and that this drives people to pressure

their governments into action (Sobel 2001; Jacobson 2007). This has largely increased as

more people receive their news from social media platforms (Baum and Potter 2019). This

eﬀect highlights what Baum and Potter (2008) call the "foreign policy marketplace" where

80

the two actors (the public and the policy-makers) both are interested in creating policy that

conforms to their interests and preferences. When these preferences are at odds, the literature

argues that information proves key. Thus, the media acts as a "trader of information" and

can provide the public with the resources they need to pressure their leaders (Baum and

Potter 2008, 56). Additionally, this inﬂuence goes beyond simply disseminating information.

Agenda-setting theory argues that the media can raise awareness of an issue and thereby

bring about policy changes (Cohen 1963; Baumgartner and Jones 1993; Iyengar and Kinder

1987). Scholars have also shown that international lobbying can inﬂuence a country’s media

coverage and public opinion in the United States (Lee 2007; Zhang and Cameron 2003).

Lobbying can enter this equation by inﬂuencing the information people receive. Through

these media campaigns, lobbyists, working on behalf of a foreign government, can disseminate

information that paints their client country in a positive light and thereby deﬂect any adverse

consequences that may come as a result of increased (negative) media attention.

Secondly, lobbying can inﬂuence foreign policy through providing information to Con-

gressional staﬀers, think-tanks, and other interested parties who then, in turn, inform and

inﬂuence policymakers and their decisions. As others have shown, policymakers often rely

on lobbying to ﬁll in information gaps on issues (Pevehouse and Vabulas 2019). Actors in

both Congress and the State department are faced with time constraints which inﬂuences

their ability to gather information. Due to this, research shows that these decision makers

often rely on lobbying to ﬁll in these information gaps (Austen-Smith 1993; Austen-Smith

and Wright 1994; Hansen 1991). Indeed, large lobbying ﬁrms can employ experts and gather

technical information and use this informational advantage to help inform policymakers (Lall

2012; Young 2012a; Helleiner 2011). These informational shortcuts allow lobbying ﬁrms to

exert a large amount of inﬂuence over U.S. foreign policy. Their connections within Washing-

ton D.C., coupled with their ability to gather and disseminated large amounts of information,

make lobbyists powerful ﬁgures in American foreign policy.

81

Through these two channels — pressure from the general public due to media campaigns

and through providing information to policymakers — lobbyists are able to steer foreign

policy objectives in a direction that favors their client state. It is with this inﬂuence that

lobbying can help states escape punishment for their human rights abuses. By increasing

positive media attention towards their client state, lobbying can help deﬂect possible calls for

action from the general population (such as sanctions or a reduction in foreign aid). Similarly,

lobbyists can use their informational advantage and general access to policymakers to sway

the opinions of those in charge towards viewing their client states in a more positive (or

perhaps more strategic) manner thus reducing the likelihood of punishment. The above

discussion leads to the following hypotheses:

Sanctions Hypotheses:

Hypothesis 1 An increase in lobbying will be associated with a decrease in the probability

of being sanctioned by the U.S.

Hypothesis 1b An increase in lobbying will moderate the eﬀect human rights abuses have

on the probability of being sanctioned.

Hypothesis 2 An increase in lobbying will be associated with a decrease in the severity of

U.S. sanctions.

Hypothesis 2b An increase in lobbying will moderate the eﬀect human rights abuses have

on the severity of U.S. sanctions

Foreign Aid Hypotheses:

Hypothesis 3 An increase in lobbying will lead to a corresponding increase in the allocation

of U.S. aid.

Hypothesis 3b An increase in lobbying will moderate the eﬀect human rights abuses have

on the allocation of U.S. aid.

Some of the literature on foreign aid and human rights ﬁnds that the overall amount of

aid is not inﬂuenced by human rights but that the type of aid is. This research shows that

82

when a country’s human rights abuses are made known through shaming from an INGO, that

country is more likely to see bypass aid than direct aid (Dietrich and Murdie 2017). This type

of aid, "bypasses" the government in favor of local NGOs and other organizations. Donor

states use this tactic to change the behavior of the recipient government in some way (Dietrich

2013). This coercive inﬂuence is due to the recipient state wanting the aid to ﬂow through

government channels rather than through local NGOs or other non-governmental avenues.

This is due to the government wanting control over the (often) large sums of money sent

to their countries through foreign aid. Lobbying can aﬀect this relationship by inﬂuencing

donor governments — in much the same way as outlined earlier — to limit bypass aid in

favor of more direct aid. Bypass aid is often due to the donor country seeing the recipient

state’s government as being poorly governed or, as previously discussed, due to human rights

abuses being made known (Dietrich 2013; Dietrich and Murdie 2017). Lobbyists working for

the recipient state can provide information to U.S. policymakers and assure them that the

recipient government is ready to receive direct aid and that it is in the interest of the U.S.

to do so. Lobbying can also convince those in charge of foreign aid to overlook human rights

abuses and focus on the strategic or other aspects of the donor-recipient relationship that

proves more to the recipient state’s advantage. Through this information and direct contact

with policy-makers, lobbyist can increase the likelihood foreign aid in delivered through

government to government channels rather than as bypass aid. This inﬂuence by lobbyists

leads to my ﬁnal hypothesis:

Bypass Aid Hypotheses:

Hypothesis 4 Increased lobbying will lead to a decrease in the proportion of bypass aid a

country receives.

Hypothesis 4b Increased lobbying will moderate the eﬀect human rights abuses have on the

proportion of bypass aid a country receives.

83

4.4.1 Examples: Azerbaijan and Egypt

To show the inﬂuence lobbying has on foreign policy decisions, I brieﬂy outline two ex-

amples. Both of these countries have long histories of lobbying within the U.S. and highlight

how lobbying can inﬂuence the likelihood of punishment.

Azerbaijan has become increasingly invested in lobbying operations within the U.S. A

report from the Sunlight Foundation found Azerbaijan to be in the top ten of foreign countries

in terms of money spent on lobbying in 2013. But what does this money go towards? In

February of 2015, Azerbaijan increased the monthly retainer for its’ main lobbying ﬁrm, the

Podesta Group, from $50,000 to $75,000 (Lozocsky 2015). Reports show this money was

used to to interact with members of Congress, executive branch agencies, media members,

and think tanks. This inﬂuence has led to many statements of support from members of

Congress being delivered on the House and Senate ﬂoors. For example, Rep. Gene Green

(D-Texas) called Azerbaijan a "beacon of democracy" while Rep. Ryan Zinke (R-Mont.)

claimed Azerbaijan and the United States shared "the same commitment to freedom and

liberty"(Lozocsky 2015). These statements came despite Azerbaijan being labeled as "not

free" by Freedom House.

In fact, in 2015, Azerbaijan was given a democracy score of

4.17/100. How do these eﬀorts connect with human rights? David Kramer, the former

president of Freedom House, argues that these public statements of goodwill can later be

used by the regime to help ﬁght against possible sanctions. If the possibility of sanctions

increases, Kramer says that the regime can reach out through its’ lobbying contacts and

“can say [to friendly legislators]: You were with us back then; we hope we can count on you

now.”(Lozocsky 2015) Evidence of this inﬂuence can be seen in a 2015 incident involving

the Podesta Group and Maran Turner, executive director of the human rights organization

Freedom Now. Turner had organized a brieﬁng for Capital Hill staﬀers on political prisoners

in Azerbaijan. As the brieﬁng neared, Turner reports that invitees told her the Podesta

group called each oﬃce to convince them not to attend. The Podesta group also called the

84

House Committee on Foreign Aﬀairs, who was hosting the event, and asked them to cancel

it (Lozocsky 2015).

Another example comes from Egypt which has a long history of lobbying in the United

States. One of the main lobbying ﬁrms hired by Egypt is the PLM Group which is a joint

venture between the Podesta Group, the Livingston Group, and the Moﬀett Group. These

ﬁrms are led by powerful Washington insiders such as former House Republican Majority

Leader Bob Livingston, Democratic campaign advisor Tony Podesta, and former Democratic

congressman Toby Moﬀett (Blumenthal 2011). All three have long lists of powerful contacts

and use those contacts to lobby directly for the Egyptian Government. For instance, in 2009,

the PML group was hired to contact "media representatives, civic organizations, members of

Congress, congressional staﬀers and various U.S. Government oﬃcials concerning the current

status of U.S./Egyptian relations, the proposed U.S. foreign aid program for Fiscal Year 2009,

human rights and religious freedom...[italics added]" (FARA 2009).

A speciﬁc example of the inﬂuence these lobbying ﬁrms have can be seen regarding the

delivery of 20 Lockheed Martin F-16 Fighter Planes to Egypt in 2012. The government of

President Hosni Mubarak agreed to purchase the planes in 2010. In the months leading up

to the announcement of the deal, lobbyists working for the Egyptian government set up 46

meetings with government oﬃcials with nearly a dozen of those meetings being with Pentagon

and State Department oﬃcials respectively including Rep. Rosa DeLauro, D-Conn., and Nita

Lowey, D-N.Y., who oversee foreign aid packages as a part of the Appropriations Committee.

The deal would take $213 million in aid given to Egypt by the United States and use that

money to buy the planes from Lockheed Martin (who, coincidentally, was represented by the

same lobbying group: PLM). The lobbying campaign preceding the Lockheed Martin deal is

not unique however. In 2008 alone, lobbyist arranged over 1,000 contacts with government

oﬃcals including Rep. Ileana Ros-Lehtinen, R-Fla., who was the chair of the House Foreign

Aﬀairs Committee and who also happended to oppose providing aid to Egypt (Young 2012b).

85

Similarly, in the lead-up to the F-16 deal, the PLM group made 366 contacts with government

oﬃcials from January to July of 2010. This included contact with 61 members of Congress

and 141 diﬀerent Congressional staﬀers(Blumenthal 2011).

During this same time, Egypt not only lobbied for foreign aid, but also lobbied to oppose

the passing of a pro-human rights and democracy resolution in the Senate (S. Res. 586 or

the Feingold Resolution) that called for the United States to support human rights in Egypt

and asked for an end to the emergency law in the wake of the deposing of Hosni Mubarak

(Feingold 2010). The lobbying eﬀort to oppose this resolution was intense with lobbyists

organizing 77 lobbying contacts with Senators — 52 of which explicitly mention the Senate

resolution. The Senator most contacted by the lobbyists was Roger Wicker (R-MS). Wicker

was contacted twenty times with seventeen of those listing the Resolution as the reason for

contact.

In an article published by Foreign Policy, Wicker was designated as one of the

main actors responsible for blocking the Feingold resolution (Rogin 2011). In the end, the

resolution was blocked from passage.

Following the Arab Spring, Egypt cut its’ ties with the PLM Group in January of 2012.

However, it was not long before Egypt again entered the Washington lobbying scene. Fol-

lowing the suspension of U.S. military aid in late 2013 (which included f-16 ﬁghter jets,

M1 Abrams tanks, and Apache attack helicopters), Egypt quickly hired the Glover Park

Group to lobby on its behalf. As reported by The Hill, the Glover Park Group will facilitate

communications “with U.S. government oﬃcials, business community, non-governmental au-

diences and the media” as well as help with “fostering and facilitating exchanges between the

U.S. and Egypt.” (Bogardus 2013). Following an intense lobbying campaign, $1.5 billion in

economic and military aid was restored to Egypt in early 2014.

As we can see, foreign lobbying can inﬂuence how policymakers perceive a country and

can impact actions towards the lobbying state. Additionally, lobbying can help change the

narrative within the general population by increasing attention to a particular state. Thus,

86

lobbying proves a powerful tool of inﬂuence for states seeking to aﬀect U.S. policy and

behavior.

4.5 Data and Research Design

4.5.1 Dependent Variables

Data on U.S. sanctions comes from Neuenkirch and Neumeier (2015). These data identify

instances where a sanction was imposed upon a country by the United States. Neuenkirch

and Neumeier (2015) draw upon earlier data from Hufbauer et al. (2009) and extend this

data to 2012. Within my sample, the percentage of states sanctioned by the U.S. is just

under 12%. Neuenkirch and Neumeier (2015) also code the severity of U.S. sanctions using

data from Wood (2008). Severity is measured from 1-3 with a value of one representing mild
sanctions and a value of three representing severe sanctions.7 Within the sample, there are

486 country-years with a U.S. sanction. Within these sanctions cases, the average severity

of a sanction is .66 with a standard deviation of .78. To understand how U.S. sanctions are

distributed across time, ﬁgure 4.3 shows the number of U.S. sanctions by year. Following

the end of the Cold War, we see the U.S. dramatically increase its’ number of sanctions.

However, over the last 20 years, the number of sanctions per year have been quite low.

To measure U.S. aid, I rely on data from Boutton and Carter (2014). They gather aid

data from the U.S. Agency for International Development (USAID). I use both a dichotomous

measure of aid as well as a continuous measure. The dichotomous measure is coded as a

one if a country received aid from the U.S. in that year and zero otherwise. The continuous

measure codes the amount of aid received that year by category. Speciﬁcally, it lists the

amount of military and economic aid a country receives. This continuous measure represents

the amount of aid in millions of 2009 constant dollars. The continuous measure of aid has a

high amount of variance with some countries receiving low amounts of aid in the hundreds

7For a full discussion of the coding rules, see ﬁgure C.3 in the appendix.

87

Figure 4.3: Number of US Sanctions by Year

of thousands of dollars while others receive billions of dollars. To account for this, I follow

previous literature (see de Mesquita and Smith 2007) and take the natural log of both types

of aid. Figure 4.4 shows the distribution of the log transformed aid variables. As the ﬁgure

shows, a large amount of countries receive no U.S. military aid. However, this ﬁgure is lower

when looking at U.S. economic aid.

Hypotheses 4 and 4b predict that lobbying will reduce the proportion of bypass aid a

country receives and that lobbying can moderate the eﬀect repression has on bypass aid.

To test this hypothesis, I use data on U.S aid from the OECD CRS aid activity database.

Starting in 2004, the OECD began collecting data on the channel through which aid was

being delivered (public or private).

I create a measure of the proportion of bypass aid

to total aid. Proportional dependent variables present some issues to normal estimation

procedures as discussed by Dietrich (2013) — speciﬁcally that the dependent variable is

bounded between 0 and 1 yet is more or less continuous within that bounded area. To

account for this, Dietrich (2013) follows Aitchison (1986) and creates a (J-1) log aid ratio

88

Figure 4.4: Distribution of Economic and Military Aid

where J is the total number of delivery channels (government to government or non-state

aid). I follow this approach and create a measure of bypass aid which equals the log of the

proportion of bypass aid for country i at time t divided by one minus the amount of bypass

aid for country i at time t. This creates a variable that is unconstrained and therefore able

to be estimated through OLS. The coeﬃcient on this variable displays how the log ratio

of bypass aid changes in comparison to non-bypass aid. Hypothesis four would predict a

negative coeﬃcient for lobbying signifying that an increase in lobbying corresponds to a

decrease in the proportion of bypass aid. Figure 4.5 shows the distribution of both the

bypass proportion measure as well the log transformed version. The proportional measure

of bypass aid has a mean of .54 with a min and max of 0 and 1 respectively while the log

transformed measure ranges from -8.2 to 9.4 with a mean of .15.

89

Figure 4.5: Proportion of Bypass Aid and the Logged Transformed

4.5.2

Independent Variables

Data on lobbying comes from the Foreign Agents Registration Act. As noted earlier, PR

ﬁrms working for foreign governments are required to report such activities to the Justice

Department. These data were collected by Pevehouse and Vabulas (2019). As a part of

the data collection process, Pevehouse and Vabulas (2019) record the number of lobbying

contracts per year by country. Across the sample, the average number of lobbying contracts

is 3.6. The measure of lobbying contracts, however, has a large amount of variance with some
countries having no lobbying contracts in a given year while others have dozens.8 Therefore,

I follow Pevehouse and Vabulas (2019) and take the natural log of the number of lobbying
8Japan has the highest number of lobbying contracts at 105 in 1996. Throughout the
sample, Japan averages 42 lobbying contracts per year — a signiﬁcantly higher average than
any other country.

90

contracts. Much of my theory focuses on the eﬀect lobbying has for states with diﬀering

levels of human rights abuses. Yet, it should be noted that all states tend to lobby at a

similar rate regardless of their human rights scores. For example, states with low levels of

repression (25th percentile and below) average 5.6 lobbying contracts per year while states

with high levels of repression (75th percentile and above) average 3.5 lobbying contracts per

year. In general, most states see a large variation in the amount of lobbying they engage

in across time. Figure 4.6 shows the number of lobbying contracts by year for six diﬀerent

countries.

Figure 4.6: Number of Lobbying Contracts by Year

My measure of repression comes from (Fariss 2014). I use this measure to connect human

rights abuses with military aid. Fariss uses a latent approach to model human rights scores

that takes into account the changing nature of accountability. This measure incorporates

the evolving nature of views on human rights and how that aﬀects human rights scores.

91

This approach allows for a more dynamic measure of repression. The data range in value

from -3.1 to 4.7 with higher values signifying higher respect for human rights. For ease of

interpretation, I invert this measure so higher values represent higher levels of repression.

My inverted measure has a mean of -.2 and a standard deviation of 1.3.

To account for the eﬀect human rights abuses have on the allocation of U.S. economic aid,
I also include a measure of empowerment rights taken from Cingranelli et al. (2014).9 This

measure is an additive index that draws upon the Foreign Movement, Domestic Movement,

Freedom of Speech, Freedom of Assembly & Association, Workers’ Rights, Electoral Self-

Determination, and Freedom of Religion indicators. It ranges from 0 which represents no

respect for these rights to 14 which represents full government respect for these rights with

an average value of 8.3. Similar to the repression measure from Fariss (2014) I invert this

measure so that higher values represent higher levels of human rights abuses.

4.5.3 Control Variables

When estimating the models predicting sanctions, I control for many other factors that

may aﬀect the likelihood of being threatened with sanctions or having them imposed. First,

I control for the amount of human rights related media coverage for a given country. Peksen

et al. (2014) show that the media can have a signiﬁcant eﬀect on the likelihood of U.S.

sanctions being applied. To control for this, I include a count measure of the number of
media mentions of human rights take from Nielsen (2013) for a given state.10 The type

of government a state has can also eﬀect the likelihood of sanctions (Cox and Drury 2006).

Thus I control for this by including a measure from PolityIV (Marshall et al. 2016). The main

eﬀect of regime comes from distinguishing democracies from non-democracies, therefore, I
9Following Neumayer (2003a), I expect political and civil rights (measured here as em-
powerment rights) to negatively inﬂuence economic aid while the repression variable from
Fariss (2014) I expect to aﬀect military aid.

10Speciﬁcally, this data codes media mentions of human rights from the New York Times.

92

create a dummy variable called democracy which represents all countries with a polity score

of six or above. I also control whether or not a country is allied with the U.S., the amount of

economic development, and population of the state as each of these are likely to impact the

likelihood of sanctions. Lastly, I control for the amount of trade between the country and

the U.S. as countries without any trade ties to the U.S. are less likely to be sanctioned given
there are no movement of goods to restrict. 11 Table 4.2 presents the descriptive statistics

for each control variable.

Controls:

NYT Shaming
Democracy
GDP per capita
Population
Trade
Alliance

Mean

6.9
0.39
7.29
16.13
3.93
0.25

SD

18.90
0.49
1.33
1.4
0.57
0.43

Min

0
0
4.32
13.25
1.9
0

Max

327
1

11.27
21.06
6.894

1

Table 4.2: Summary Statistics for Sanctions Analysis Sample

For my analysis on foreign aid it is important that other determinants of foreign aid be

controlled for. It has been shown that the level of economic development of a state can impact

the likelihood of receiving aid with less developed states more likely to receive aid (Demirel-

Pegg and Moskowitz 2009). I control for this using the natural log of GDP per capita taken

from Gleditsch (2002). It has also been shown that regime type matters in the allocation of

aid with democracies receiving more aid from the U.S. (Wright 2009). To account for this, I

include a variable for regime type taken from the Polity IV dataset (Marshall et al. 2016). It

could also be that the population size of a country matters. Therefore, I include the natural

log of population taken from the World Bank. U.S allies and those countries that trade more

with the U.S. may also be more likely to receive aid. Due to this, I control for the amount

of trade (taken from Gleditsch 2002) and whether or not the country is an ally of the U.S.
11For the trade variable, I follow the sanctions literature and take the total amount of

trade divided by the country’s GDP.

93

(Leeds et al. 2002). I also include dummy variables for the cold war period and the post

9/11 period as both of these could impact aid. Table 4.3 displays the descriptive statistics

for the control variables used in the foreign aid analysis.

Controls

Population
U.S. Ally
Polity
Trade
GDP per capita

Mean

15.93
0.414
0.583
6.040
7.994

SD

1.515
0.493
7.382
2.686
1.206

Min

12.11

0
-10
0

4.600

Max

20.99

1
10
13.10
11.09

Table 4.3: Summary Statistics for Foreign Aid Analysis

To estimate the eﬀect lobbying and human rights have on bypass aid, I include other

factors that the aid literature ﬁnds important. First, I include a measure of governance

quality as some have found that donor countries deliver more bypass aid to countries with

poor governance (Dietrich 2013). This data comes from Kaufmann et al. (2009). I also control

for regime type by including a measure from Polity IV as some argue that democracies may

have greater constraints on the ability of the government to capture aid (Marshall et al.

2016). Natural disasters may also inﬂuence the allocation of aid with more aid going to

disaster relief agencies rather than the government following a crisis. Thus, I include the

number of natural disasters in a country as recorded by the EM-DAT database. Related

to this, civil conﬂict may also cause donors to give more bypass aid. Therefore, I use data

from Gleditsch et al. (2002). I also include a measure of geographic distance (Scott Bennett

and Stam 2000) and trade intensity which is the logged sum of imports and exports taken

from the IMF-DOT (2009) database (Dietrich 2013). Security considerations could also play

a role in aid delivery which leads me to include in the model a measure of whether or not

the recipient state is a rotating member of the UN Security Council (Kuziemko and Werker

2006). Table 4.4 presents the descriptive statistics for each variable.

94

Controls:

Governance
Civil War
Natural Disaster
Distance
Trade Intensity
Security Council
Polity

Mean

1.95
0.24
0.8
0.54
8.5
0.06
2.89

SD

0.54
0.70
1.46
5.92
1.85
0.23
5.82

Min

0.72
0

-2.30
9.35
4.58
0
-10

Max

3.66
1
4.16

13.08

1
10

Table 4.4: Summary Statistics for Bypass Aid Analysis Sample

4.5.4 Research Design

The theory outlined above argues that states will use lobbying to avoid punishment

for their human rights abuses. To test this, I estimate a logistic regression with sanctions

as my dependent variable. The unit of analysis is country-year.

I lag all time variant

independent variables and cluster the standard errors by country. Given this speciﬁcation,

I run my analysis on 3407 country-year observations with my standard errors clustered over

111 countries. To test the severity hypothesis, I estimate an ordered logistic regression this

time using severity variable as my dependent variable. Here, I again lag all time-variant

controls and cluster the standard errors on country. For both dependent variables, I also

estimate models interacting lobbying and repression. This allows me to test the hypothesized

interactive eﬀect.

To analyze the eﬀect of lobbying and human rights on U.S. foreign aid, I follow previous

literature which suggests economic aid diﬀers from military aid and estimate separate models

for each type of aid (Apodaca and Stohl 1999; Boutton and Carter 2014). My unit of analysis

is the country-year. For the military aid equations, I have 4367 country- year observations.

This number is reduced to 3356 country-year observations in the economic aid equations

due to the use of the empowerment rights measure. Following the literature, I estimate a

two-stage Heckman model ﬁrst estimating the likelihood of receiving aid, then estimating

95

the amount of aid provided given that you received aid.12 I also include country ﬁxed-eﬀects

in the outcome equation to account for any unobserved country-speciﬁc attributes that may

be aﬀecting the relationship. Some have found that aid is inﬂuenced by what is called

"bureaucratic inertia" where aid in time t is inﬂuenced by aid in time t-1 (Lai 2003). To

account for this, I include in the selection equation a cubic time spline. I test the proposed

hypothesis of lobbying conditioning the eﬀect of human rights on aid by including within the

model an interaction term between lobbying and human rights. I include this interaction

in both the selection equation as well as the outcome equation as I believe lobbying can

moderate the eﬀect of human rights in the amount of aid received but also in the decision

to receive any aid at all.

Finally, to test hypotheses four, I run a series of OLS regressions predicting the proportion

of bypass aid a country receives based on the amount of lobbying contracts they have and

their human rights. I then interact lobbying and human rights to test whether or not lobbying

can moderate the relationship between human rights and bypass aid. Given the shorter

temporal period, I run the analysis on 437 country-year observations.

I include regional

dummies and year dummies to account for any unobserved factors that may inﬂuence this

relationship.

I also include robust standard errors. Following Dietrich (2013) I log the

measure of proportion of bypass aid as OLS assumes a continuous normal distribution.

However, this causes my estimates to be diﬃcult to interpret as they are the eﬀect lobbying

has on the log proportion of bypass aid. I could, of course, simply take the anti-log of the

coeﬃcient but this proves diﬃcult when doing predictions as the standard errors are incorrect.

To help with this, I draw on work by Papke and Wooldridge (1996) who use a fractional

logit as a means of exploring proportional data. These authors put forward the use of a user-

written fractional logit model however, canned options within STATA are available. Recent

work by Oberhofer and Pfaﬀermayr (2012) shows that STATA’s GLM command with the

"link(logit)" function successfully replicates Papke and Wooldridge (1996)’s original results.
12Lai 2003; de Mesquita and Smith 2007; and Boutton and Carter 2014 use similar models

96

I therefore estimate two GLM models ﬁrst estimating the eﬀect of lobbying on the proportion

of bypass aid then estimating the model with the interaction between lobbying and human

rights. Using the GLM models allows me to predict the eﬀect of lobbying on the proportion

of bypass aid with the predictions being bounded between 0 and 1 just as the proportion of

bypass aid is bounded by 0 and 1. Thus, the predicted probabilities serve as the predicted

proportion of bypass aid making interpretation more straightforward. Given that the number

of observations are limited, I also use the bootstrap option with the GLM model and draw

500 samples with replacement to better estimate my standard errors. As the analysis shows,

there is no substantive diﬀerence between the OLS and GLM models.

4.6 Results

4.6.1 Lobbying, Human Rights, and Sanctions

We will begin by looking at the relationship between lobbying, human rights, and sanc-

tions. The results are displayed in table 4.5. As displayed in models 1 and 3, the control

variables conﬁrm the previous literature on sanctions. We see that shaming by the New York

Times increases both the probability and severity of sanctions. Conversely, being a democ-

racy signiﬁcantly decreases both the likelihood and severity of sanctions. The models vary

some on the signiﬁcance of the other control variables with the alliance variable increasing

the probability of sanctions but not aﬀecting their severity. Trade switches this order and
increases the severity while not signiﬁcantly impacting the probability of being sanctioned.13
13Note that models 3 and 4 were run on the full sample of observations and therefore
compare no sanctions (0) to severe sanctions (3). I run the same analysis but this time only
on those country-year observations with a U.S. sanction. I re-code the severity variable so a
"0" now represents mild sanctions and a "2" represent severe. Some of the control variables
change in signiﬁcance but my main variables of interest remain unchanged. I present the
results of this analysis in table C.3 in the appendix.

97

Lobbying (t-1)

Repression (t-1)

Lobbying * Repression

NYT Shaming (t-1)

Democracy (t-1)

Alliance

GDP per capita (t-1)

Population

Trade (t-1)

Constant

Cut 1

Cut 2

Cut 3

(Sanctions)

Model 1
-0.11∗
(0.056)
1.44∗∗
(0.197)

0.02∗
(0.010)
-1.61∗∗
(0.363)
0.80∗∗
(0.362)
0.25
(0.171)
0.04
(0.128)
0.00
(0.002)
-5.40∗∗
(2.609)

Model 2
(Sanctions)
-0.16∗∗
(0.067)
1.54∗∗
(0.224)
0.06
(0.050)
0.02∗
(0.010)
-1.63∗∗
(0.367)
0.78∗∗
(0.366)
0.25
(0.172)
0.04
(0.129)
0.00
(0.002)
-5.43∗∗
(2.638)

Model 3 Model 4
(Severity)
(Severity)
-0.17∗∗
-0.14∗∗
(0.064)
(0.080)
1.54∗∗
1.47∗∗
(0.219)
(0.185)
0.04
(0.045)
0.01∗∗
(0.005)
-1.53∗∗
(0.337)
0.38
(0.368)
0.32
(0.196)
0.10
(0.130)
0.00∗∗
(0.001)

0.01∗∗
(0.005)
-1.52∗∗
(0.331)
0.41
(0.355)
0.31
(0.194)
0.10
(0.129)
0.00∗∗
(0.001)

6.76∗∗
(2.799)
7.88∗∗
(2.768)
9.03∗∗
(2.647)

6.84∗∗
(2.863)
7.96∗∗
(2.832)
9.11∗∗
(2.706)

-981.35
3407

Log Likelihood
N
Standard errors clustered on country.
∗ p < 0.10, ∗∗ p < 0.05
Table 4.5: Eﬀect of lobbying and human rights on U.S. sanctions: 1978-2012

-978.99
3407

-1456.62

3407

-1457.61

3407

98

The eﬀect of lobbying and repression remain consistent across models. As the table

shows, higher levels of repression consistently lead to a higher likelihood of being sanctioned

and higher levels of severity. Lobbying, however, shows the opposite eﬀect with increases

in lobbying leading to a reduction in the probability and severity of U.S. sanctions. This

gives support to hypotheses one and two stating that lobbying will decrease the probability

of being sanctioned and will decrease the severity. The substantive eﬀect of this relationship

can be seen in ﬁgure 4.7. Here we see the predicted probability of U.S. sanctions across the

range of lobbying. The predicted probability of being sanctioned changes from 8.8% to 3.6%

over the range of lobbying. This 5% change in probability of being sanctioned represents a

statistically signiﬁcant diﬀerence (p-value 0.062). Thus, this initial analysis provides support

for hypothesis one and shows that lobbying can indeed help states avoid being punished with

sanctions for their human rights abuses.

Figure 4.7: Predicted probability of U.S. sanctions across range of lobbying. Results from table 4.5, model
1.

99

Additionally, the analysis shows that lobbying can have a moderating eﬀect on the inﬂu-

ence of repression. Table 4.5 shows that the coeﬃcient for the interaction between lobbying

and repression is insigniﬁcant in both model two and model four. However, ﬁgure 4.8 shows

that at certain levels of repression, lobbying can have a moderating eﬀect on the probability

of being sanctioned. Based on model three, ﬁgure 4.8 shows the predicted probability of
sanctions for both high and low levels of lobbying. 14 Overall, we see that increased lob-

bying leads to a lower predicted probability of being sanctioned. However, the conﬁdence

intervals for these predictions overlap — except at average and slightly above average levels

of repression. Here we see that increased lobbying can signiﬁcantly reduce the probability of

sanctions. For example, at just above average levels of repression, the predicted probability

of being sanctioned, given you have low levels of lobbying, is eleven and a half percent. A

state with the same level of repression, but one who uses lobbyists, sees a reduction in the

probability of being sanctioned of six percent (p-value 0.03). Intuitively, this result makes

sense. At low levels of repression, states are unlikely to be sanctioned given the lack of

reasons for sanctions to be imposed. Thus, while lobbying still leads to a reduced likelihood

of being sanctioned, this ﬁnding fails to gain statistical signiﬁcance. The eﬀect of lobbying

proves strongest at average and slightly above average levels of repression. Here, a state’s

human rights abuses may make them a target of U.S. sanctions but since the abuses are

not egregious, lobbying can have a moderating eﬀect. At very high levels of repression, this

eﬀect is lessened as the egregious human rights abuses begin to overcome the eﬀect of lob-

bying. In these cases, the amount and severity of the human rights abuses cause the eﬀect

of lobbying to be lessened. Figure 4.8 shows this relationship in action. Across levels of

repression, lobbying consistently reduces the likelihood of being sanctioned. However, it is

only at moderate levels of repression where we see this eﬀect gain statistical signiﬁcance. At

these moderate levels of repression, lobbying can help states avoid being punished for their
14For high level of lobbying, I use the 99th percentile while low lobbying represents the

25th percentile.

100

human rights abuses. Thus, hypothesis 1b also receives some support from the analysis. It

seems that lobbying can indeed moderate the relationship between repression and sanctions,

especially at average to above average levels of human rights abuses.

Figure 4.8: Predicted probability of U.S. sanctions with high and low amounts of lobbying across values of
repression. Results from table 4.5, model 2.

The analysis of sanction severity shows similar results. Table 4.7 conﬁrms the predic-

tion made in hypothesis two that repression increases the severity of sanctions while lob-

bying reduces the severity. Figure 4.9 displays this relationship graphically showing that

as the number of lobbying contracts increases, the probability of having severe sanctions
decreases.15 The interaction between lobbying and repression again, like the results from
15Note the overall predicted probability of being severely sanctioned is quite low as this
graph shows the results from the full sample. Figure C.5 in the appendix shows the results
when using only those observations with a U.S. sanction. Here, as the graph shows, the
initial probability is much higher.

101

model 1, is insigniﬁcant in model 4. Graphically, the relationship shows that across the

range of repression, lobbying consistently reduces the probability of having severe sanctions.

However, unlike ﬁgure 4.8 showing the results from model 2, the results from model 4 fail to
gain signiﬁcance at any values of repression.16 Thus, I fail to ﬁnd a signiﬁcant moderating

eﬀect of lobbying on the relationship between repression and sanction severity.

Figure 4.9: Predicted probability of severe sanctions across the range of lobbying. Results from table 4.5,
model 3.

4.6.2 Lobbying, Human Rights, and Foreign Aid

Turning to the analysis of lobbying, human rights, and foreign aid. The results from

table 4.6 show the eﬀect lobbying has on the allocation of U.S. military aid for both the

probability of receiving aid (the selection equation) as well as the amount of aid a state

receives conditional on being a "1" in the selection equation. The control variables behave

largely in the manner we would expect. In determining who receives military aid, the analysis

16These results are show in ﬁgure C.4 in the appendix.

102

shows that more democratic states see an increased likelihood of receiving U.S. aid. The

model also shows that developed states are also less likely to receive military aid from the

U.S. As for the outcome equation, we see that again, less developed states are more likely to

receive military aid. In the outcome equation, we do see some changes in signiﬁcance. For

instance, the coeﬃcient on polity is no longer signiﬁcant meaning that democratic states,

while more likely to receive aid, do not see a signiﬁcant diﬀerence in the amount of aid they

receive. The population of a state also changes in its signiﬁcance. While larger states are

no more likely to receive aid, larger states are allocated less aid should they receive it. The

dummy variable for the cold war maintains a positive and signiﬁcant eﬀect both increasing

the likelihood of military aid and increasing the amount allocated.

The main variables of interest also largely behave according to expectation. We see

ﬁrst that repression (here measured using Fariss 2014’s measure) reduces the likelihood of

receiving aid. Given that military aid is sent, increased level of repression actually increase

the amount of military aid allocated. While seemingly counter to my expectations, this result

actually ﬁts within the theory presented. The fact that higher levels of repression actually

increase the predicted amount of aid a country receives likely stems from the countries the

U.S. provides military aid to. For instance, the countries receiving the most military aid

from the U.S. over the time-frame of the sample are Israel, Iraq, Egypt, and Afghanistan.

These states all receive large amounts of military aid from the U.S. and are likely driving the

ﬁnding that higher levels of repression increase the amount of aid one receives conditional

on being allocated military aid in the selection equation. Lobbying similarly shows the

predicted eﬀect with lobbing leading to an increase in the amount of aid a state receives.

In the selection equation, the coeﬃcient for lobbying is positive meaning more lobbying

increases the probability of receiving military aid but just fails to reach signiﬁcance with a

p-value of .11. Thus, hypothesis 3 receives some support with lobbying increasing both the

likelihood of receiving aid and the amount of aid a state receives. This ﬁnding also supports

103

past work that ﬁnds similar results (Montes-Rojas 2013).

Level of Aid Selection Level of Aid Selection

0.02∗
(0.010)
0.36∗∗
(0.048)
0.02∗∗
(0.009)

0.01
(0.006)
-0.37∗∗
(0.070)
-0.39∗
(0.219)
0.04
(0.034)
0.39
(0.240)
0.41∗∗
(0.081)
0.41∗∗
(0.064)

8.44∗∗
(3.449)

0.02
(0.012)
-0.04
(0.034)
0.03∗∗
(0.008)
0.01∗∗
(0.005)
-0.08∗∗
(0.042)
0.01
(0.028)
0.00
(0.021)
-0.07
(0.075)
0.18∗∗
(0.071)
0.13
(0.084)
-0.80∗∗
(0.029)
0.06∗∗
(0.003)
-0.00∗∗
(0.000)
1.79∗∗
(0.567)
-0.35∗∗
(0.053)

YES
4367

NO
4367

Military Aid

Lobbying (t-1)

Repression (t-1)

Lobbying * Repression

Polity (t-1)

Ln GDPPC (t-1)

Ln Population (t-1)

Ln Trade (t-1)

US Ally (t-1)

Cold War

Post 9/11

t

t-2

t-3

Constant

λ

Country Fixed Eﬀects
N
Standard errors in parentheses
∗ p < 0.10, ∗∗ p < 0.05

0.02∗
(0.010)
0.34∗∗
(0.046)

0.02
(0.012)
-0.06∗
(0.033)

0.01
(0.006)
-0.38∗∗
(0.070)
-0.38∗
(0.219)
0.04
(0.034)
0.38
(0.240)
0.42∗∗
(0.081)
0.40∗∗
(0.065)

8.41∗∗
(3.453)

YES
4367

0.01∗∗
(0.005)
-0.09∗∗
(0.042)
0.00
(0.028)
0.01
(0.020)
-0.08
(0.075)
0.17∗∗
(0.071)
0.13
(0.084)
-0.80∗∗
(0.029)
0.06∗∗
(0.003)
-0.00∗∗
(0.000)
2.02∗∗
(0.562)
-0.35∗∗
(0.052)

NO
4367

Table 4.6: Moderating eﬀect of lobbying on human rights and the allocation of military aid

104

The interactive eﬀect of lobbying and repression is also shown by table 4.6. For both the

selection and outcomes equations, the coeﬃcient for the interaction is positive and signiﬁcant.

However, as some have pointed out, interpreting the results of such interactions from a

regression table can be diﬃcult (Brambor et al. 2006). Thus, I also display the relationship

graphically in ﬁgure 4.10. Looking at this ﬁgure, I ﬁnd strong support for the moderating

eﬀect lobbying has on the relationship between human rights and the allocation of military

aid.

Figure 4.10: Predicted probability of receiving military aid with high and low levels of lobbying across the
range of repression. Results from table 4.6

As the ﬁgure shows, states with low amounts of lobbying (25th percentile) see a sig-

niﬁcant reduction in the probability of receiving military aid as their level of repression

increases. However, states with high amounts of lobbying (75th percentile) see no change

in the probability of receiving aid as their level of repression increases. By calculating the

105

ﬁrst diﬀerences, we can ascertain the substantive eﬀects of this relationship. For states with

low levels of lobbying, an increase in repression from the min to the max leads to a 19%
reduction in the probability of receiving U.S. military aid.17 For states with a high amount

of lobbying, the predicted change in probability of receiving U.S. military aid as the level

of repression increases is 0%. Thus, I ﬁnd initial support for hypothesis 3b predicting that

lobbying will moderate the relationship between human rights abuses and the allocation of

U.S. foreign aid.

Table 4.7 displays the results when using economic aid as the dependent variable and

empowerment rights as the measure of human rights abuses. The control variables are

similar to the military aid equation but some notable diﬀerences exist. Speciﬁcally, we see

a change in the polity variable with it now being signiﬁcant only in the outcome equation

with democratic countries receiving more aid. One explanation for this could be that the

allocation of economic aid could be used as a "carrot" for states to become more democratic

(Boutton and Carter 2014). This would cause economic aid to be given to those states

with less democratic tendencies. The population variable also shows a change with larger

states receiving more economic aid. This diﬀerence makes sense given that economic aid is

likely more tied to helping with issues that are also aﬀected by the population of a state like

infrastructure, education, and social spending.

17When moving from the 25th percentile to the 75th percentile of repression, the predicted

decrease in the probability of receiving U.S. military aid is 5% p-value .000.

106

0.01
(0.009)
-0.04∗∗
(0.010)

0.02∗∗
(0.006)
-0.38∗∗
(0.060)
0.59∗∗
(0.203)
-0.01
(0.025)
0.59∗∗
(0.184)
0.37∗∗
(0.068)
0.20∗∗
(0.053)

Economic Aid

Lobbying (t-1)

Empowerment Rights (t-1)

Lobbying * Empowerment

Polity (t-1)

Ln GDPPC (t-1)

Ln Population (t-1)

Ln Trade (t-1)

US Ally (t-1)

Cold War

Post 9/11

t

t-2

t-3

Constant

λ

0.00
(0.010)
-0.49∗∗
(0.065)
-0.02
(0.039)
0.00
(0.029)
0.12
(0.109)
-0.42∗∗
(0.098)
0.40∗∗
(0.110)
-0.64∗∗
(0.035)
0.04∗∗
(0.003)
-0.00∗∗
(0.000)
5.97∗∗
(0.937)
-0.44∗∗
(0.071)

0.06∗∗
(0.018)
-0.03∗∗
(0.010)
0.01∗∗
(0.002)
0.01∗∗
(0.006)
-0.37∗∗
(0.060)
0.53∗∗
(0.202)
-0.01
(0.025)
0.58∗∗
(0.183)
0.38∗∗
(0.068)
0.20∗∗
(0.053)

0.00
(0.010)
-0.49∗∗
(0.065)
-0.00
(0.039)
-0.01
(0.030)
0.18
(0.112)
-0.40∗∗
(0.099)
0.41∗∗
(0.111)
-0.64∗∗
(0.035)
0.04∗∗
(0.003)
-0.00∗∗
(0.000)
5.97∗∗
(0.939)
-0.44∗∗
(0.072)

Level of Aid Selection Level of Aid Selection
0.11∗∗
(0.029)
-0.01
(0.018)
0.01∗∗
(0.003)

0.03
(0.017)
-0.02
(0.017)

-4.14
(3.568)

-2.98
(3.569)

Country Fixed Eﬀects
N
Standard errors in parentheses
∗ p < 0.10, ∗∗ p < 0.05

YES
3356

NO
3356

YES
3356

NO
3356

Table 4.7: Moderating eﬀect of lobbying on human rights and the allocation of economic aid

107

Figure 4.11: Predicted probability of receiving economic aid with high and low levels of lobbying across the
range of repression. Results from table 4.7.

The variables of interest, lobbying and empowerment rights, show mixed support for

hypothesis 3. As the literature has previously shown, empowerment rights reduce the level

of aid a country receives. Lobbying, however, fails to gain signiﬁcance for both the outcome

and selection equations. Thus, I fail to ﬁnd support for hypothesis 3 when looking speciﬁcally

at economic aid. Lobbying does however, have a moderating eﬀect on human rights abuses

when looking speciﬁcally at economic aid. Similar to the military aid analysis, ﬁgure 4.11

shows the interactive eﬀect graphically. Again, states with low amounts of lobbying see

a signiﬁcant decrease in the probability of receiving economic aid as their level of abuse

increases. For those states with high amounts of lobbying, we see no signiﬁcant change in

the probability of receiving economic aid. The ﬁrst diﬀerences again show the magnitude

of this eﬀect. For states with a low amount of lobbying, an increase in repression from the

108

min to the max leads to a reduction in the predicted probability of receiving U.S. economic
aid of 9%.18 This ﬁnding, combined with the analysis of the allocation of U.S. military aid,

provides strong support for hypothesis 3b stating that lobbying can moderate the negative

eﬀect human rights abuses have on the probability of receiving U.S. foreign aid. Across both

economic and military aid analyses, there exists a signiﬁcant and consistent ﬁnding that

lobbying can indeed overcome the negative consequences of human rights abuses when it

comes to the allocation of U.S. foreign aid.

4.6.3 Lobbying, Human Rights, and Bypass Aid

Finally, I now turn to hypotheses 4 and 4b and the relationship between lobbying, hu-

man rights, and bypass aid. Table 4.8 provides the results from the analysis. The control

variables in models one through four follow closely to previous studies on bypass aid. States

experiencing a civil conﬂict see an increased amount of aid sent through bypass channels as

donor states fear the chaos often brought about by civil war. These donor states are also

likely to be responding to the crisis by sending more money to local NGOs for relief pur-

poses and thus more aid is delivered as "bypass" aid. Polity increases the amount of bypass

aid (though this eﬀect is only statistically signiﬁcant in models three and four). States are

also more likely to send aid directly to governments that are well run therefore we see the

"Governance" term as negative and statistically signiﬁcant implying that the proportion of

bypass aid decreases as the quality of governance increases. Finally, trade intensity reduces

the amount of bypass aid as donor states provide more direct government aid to those states

it shares a large amount of imports and exports with.

18A change in repression from the 25th percentile to the 75th percentile leads to a 4%

decrease in the probability of receiving U.S. economic aid (p-value .005).

109

Lobbying (t-1)

Repression (t-1)

Lobbying * Repression

Polity (t-1)

Civil Conﬂict (t-1)

Governance (t-1)

Natural Disaster (t-1)

Distance

Trade Intensity (t-1)

Security Council (t-1)

Constant

(Model 4)

GLM
-0.05∗∗
(0.022)

-0.08
(0.114)
-0.08∗∗
(0.025)
0.04∗∗
(0.013)
0.33∗∗
(0.095)
-1.03∗∗
(0.164)
0.07
(0.048)
-0.00
(0.120)
-0.10∗∗
(0.045)
-0.19
(0.234)
2.73∗∗
(1.075)

437

(Model 1)

OLS
-0.06∗
(0.031)

-0.04
(0.150)

0.04∗
(0.022)
0.33∗∗
(0.102)
-1.56∗∗
(0.248)
0.15∗
(0.076)
-0.01
(0.171)
-0.17∗∗
(0.067)
-0.00
(0.313)
4.40∗∗
(1.501)

437
0.278

(Model 2)

OLS
-0.07∗∗
(0.031)

-0.19
(0.168)
-0.09∗∗
(0.033)
0.04∗
(0.022)
0.41∗∗
(0.110)
-1.54∗∗
(0.247)
0.15∗∗
(0.076)
0.02
(0.168)
-0.16∗∗
(0.069)
-0.14
(0.316)
4.00∗∗
(1.490)

437
0.29

(Model 3)

GLM
-0.04∗
(0.021)

0.04
(0.107)

0.03∗∗
(0.013)
0.25∗∗
(0.084)
-1.04∗∗
(0.168)
0.06
(0.046)
-0.02
(0.121)
-0.11∗∗
(0.044)
-0.07
(0.235)
2.98∗∗
(1.079)

437

N
R2
Standard errors in parentheses. Bootstrapped standard errors for models 3 and 4.
NOTE: The coeﬃcients for the regional and time dummies are not reported to conserve space.
∗ p < 0.10, ∗∗ p < 0.05

Table 4.8: Eﬀect of lobbying and human rights on bypass aid: 2005-2009

As for the diﬀerent modeling strategy, we see the OLS models and the GLM models only

diﬀer in regards to the regime type variable with the coeﬃcient on polity reaching statistical

signiﬁcance in the GLM models but not the OLS models. All other ﬁndings are mirrored.

Looking speciﬁcally at model one we see the eﬀect of lobbying on the proportion of bypass

aid is negative and signiﬁcant. This shows that as the number of lobbying contracts increases

110

in a given year, on average, the proportion of bypass aid decreases thus lending support to

hypothesis four.

Figure 4.12: Predicted proportion of bypass aid by amount of lobbying. Results from table 4.8, model 3.

Figure 4.12 shows the predicted proportion of bypass aid a state receives across the

values of lobbying based on model three. The ﬁgure clearly shows that lobbying leads to

a decrease in the proportion of bypass aid a states receives with states at low levels of

lobbying predicted to receive just under 55% of their aid as bypass aid while states with

higher amounts of lobbying predicted to see that percentage drop by 9%. As an example

of the substantive eﬀect of this decrease in the proportion of bypass aid we can look at

Lebanon in 2008. In 2008, Lebanon received just over $45 million in bypass aid and had no

lobbying contracts. If Lebanon would have moved from the minimum amount of lobbying

to the 75th percentile of lobbying, it would see an expected decrease in the proportion of

bypass aid of 6%. In actual dollar amounts, Lebanon would see their actual bypass aid go

from $45 million to $40 million with the diﬀerence being transferred directly to government

111

aid channels. This means that the Lebanese government would have $5 million more dollars

under its control. This result lends support for the hypothesis that lobbying can decrease

the proportion of bypass aid a state receives.

Further, the interaction term in models two and four are negative and signiﬁcant. Graphi-

cally, ﬁgure 4.13 shows the moderating relationship lobbying has on the eﬀect of human rights

on proportion of bypass aid. Here I calculate the diﬀerence in the predicted proportion of

bypass aid moving from the 25th percentile of repression to the 75th percentile. I calculate
this change in bypass proportion for both high and low levels of lobbying.19 The graph

depicts the predicted change in proportion of bypass aid due to changes in the amount of

lobbying as the level of repression increases. I include the 90% conﬁdence intervals for these

predictions. For states with low amounts of lobbying, changes in the amount of repression

has a signiﬁcant and positive eﬀect on the proportion of bypass aid they receive. However,

in states with high levels of lobbying, an increase in repression fails to signiﬁcantly change
the proportion of bypass aid that a state receives.20 The predicted amount of change in

proportion of bypass aid equals .125 for states with low levels of lobbying as they move from

low to high levels of repression. Conversely, states with a high amount of lobbying fail to

see any increase in bypass aid as a result of their human rights abuses. Thus, the proposed

hypothesis that lobbying can moderate the relationship between bypass aid and repression

ﬁnds some support. The coeﬃcients for lobbying from models one and three clearly show

that states who hire lobbying ﬁrms receive less bypass aid. The eﬀect of human rights on this

relationship shows up clearly in ﬁgure 4.13. Here we see the change in predicted proportion

of bypass aid by amount of lobbying in high repression states and low repression states. In

these countries that have high amounts of repression, lobbying has a signiﬁcant and negative

eﬀect on the change in proportion of bypass aid.

19I again use the 25th and 75th percentiles as "low" and "high".
20The predicted change is actually negative as the graph shows but the conﬁdence interval

overlaps with zero thus making the predicted change indistinguishable from zero.

112

Again, as previously discussed, interpretation is often easier graphically. Thus, I present

the ﬁndings from model 4 in another way in Figure 4.14. This graph shows the eﬀect of lobby-

ing on the proportion of bypass aid by repression. Here I calculate the predicted proportion
of bypass aid when moving from the 25th percentile of lobbying to the 75th percentile.21

I calculate this change in lobbying for only high repression states (75th percentile) to see

the diﬀerence in the predicted proportion of bypass aid at diﬀerent levels of lobbying. As

hypothesis four predicts, the eﬀect of lobbying on bypass aid in high repression states shows

that states with higher levels of lobbying will see a signiﬁcant decrease in the predicted

proportion of bypass aid they receive.

Figure 4.13: Eﬀect of increased repression on the predicted change in proportion of bypass aid with low and
high levels of lobbying. Results from table 4.8, model 4.

21Figure 4.14 lists the 25th and 75th percentiles as "Low Amount of Lobbying" and "High
Amount of Lobbying" respectively as the number of lobbying contracts is logged and therefore
diﬃcult to interpret.

113

The diﬀerence in proportion of bypass aid due to lobbying depicted in ﬁgure 4.14 shows

a decrease of 16.4 (p-value .000) in states with high repression. This serves as not only

a statistically signiﬁcant ﬁnding but proves substantive as well. This decrease marks a

signiﬁcant change in the amount of aid a government is able to control rather than an NGO.

For instance, take the example of Haiti in 2009. The total aid for Haiti that year from

the U.S. was just over $253 million. The proportion of bypass aid was .67. Thus, Haiti

received about $170 million through bypass channels. A 16.4% reduction in the proportion

of bypass aid would represent just under $42 million dollars for the Haitian government in

2009. That means the Haitian government would have control over $42 million more dollars

than it would otherwise have had. As the example shows, this is a signiﬁcant decrease in

the proportion of bypass aid.

Figure 4.14: Predicted proportion of bypass aid by high and low amounts of lobbying for high repression
states. Results from table 4.8, model 4.

114

4.7 Discussion

The analysis presented here seeks to investigate the relationship lobbying has with the

likelihood of being punished for human rights abuses. The results show that lobbying can

indeed shield states from the negative consequences of their actions. Speciﬁcally, I ﬁnd that

lobbying can reduce the likelihood that a state has sanctions implemented against it and

reduces the severity of sanctions. Similarly, I ﬁnd that lobbying can moderate the eﬀect

human rights abuses have on the allocation of U.S. foreign aid. Lastly, I show that lobbying

can decrease the proportion of bypass aid a state receives and this eﬀect is stronger in

states with poor human rights records. The ﬁndings presented here are the ﬁrst attempt,

to my knowledge, to connect lobbying and punishment for human rights abuses. As noted

earlier, some studies have made mention of this relationship (Pevehouse and Vabulas 2019)

this study is the ﬁrst to empirically test the eﬀect lobbying has on punishment. Thus, this

chapter provides a strong contribution to the scholarly literature on human rights. These

ﬁndings are also incredibly important to the policy world. Understanding how states can

avoid being punished for their human rights abuses can dramatically alter their ability to

do so. A recent article in Foreign Policy magazine argues that increased attention towards

foreign lobbying, coupled with the scandal surrounding Paul Manafort, has compelled the
Justice Department to take enforcing FARA more seriously.22

Indeed, a report by the

Washington Post reveals that since the Mueller probe, which prosecuted Tony Podesta of

the aforementioned PLM lobbying ﬁrm, new and amended FARA registrations are at a 20-

year high signifying that PR ﬁrms are aware of the increased scrutiny and are amending

their reports accordingly (Zapotosky and Hamburger 2017). Beyond this policy relevance,

this chapter also contributes to the literature on punishing human rights abuses. Speciﬁcally,

the incorporation of lobbying into the discussion sheds new light on a powerful, yet, before

22https://foreignpolicy.com/2017/11/03/the-partys-over-for-washingtons-foreign-

lobbyists/

115

now under-studied, inﬂuence on how human rights inﬂuences sanctions and foreign aid. As

the ﬁndings suggest, lobbying can drastically alter the nature of punishment for human rights

abusing states.

Taken together, these ﬁndings paint a bleak picture of punishment for human rights

abuses. It seems that states are able to use the advocacy power of Washington lobbying

ﬁrms to champion their cause and thereby procure more favorable policies. Beyond this, these

lobbying ﬁrms are also able to shield their clients from the negative consequences of their

poor human rights records. These ﬁndings contribute both to the scholarly literature on the

punishment of human rights abuses as well as our conceptions on the inﬂuence of lobbyists.

As was pointed out earlier, the inﬂuence domestic lobbyists have has long been known in

American politics but the eﬀect of lobbying and foreign interests is far less developed. This

study helps to bridge that gap, drawing heavily on the literature of domestic lobbying, to

show that these same eﬀects can be seen when looking at foreign lobbying. Holding states

accountable for their human rights abuses is an important policy objective for the United

States if the government contends that it holds such actors accountable. But the inﬂuence

of Washington D.C lobbying ﬁrms greatly impacts this ability.

116

CHAPTER 5

CONCLUSION

The study of why states decide to torture and abuse their citizens has a long history of

scholarship. This dissertation contributes to that literature by showing the eﬀect the broader

economy has on decisions to repress. Chapter two provides insight into one of the most ro-

bust ﬁndings in the literature: youth bulges (Hill and Jones 2014). These large youth cohorts

have been shown to drastically alter the security situation of the state and thereby lead to in-

creases in repression as states move to insulate themselves from the potential threat (Nordås

and Davenport 2013). However, this dissertation shows that the relationship between youth

bulges and repression is moderated by the economic conditions of the state. Speciﬁcally,

chapter two shows that economic growth and unemployment rates amongst the youth mod-

erate the relationship between youth bulges and repression. The logic here is simply that

countries with more economic opportunity (proxied by GDP growth and unemployment rate)

can more easily incorporate the youth into the economic system. This incorporation makes

unrest less likely which in turn reduces the need to preemptively repress citizens. While the

logic is straightforward, the impact of this ﬁnding should not be understated. As mentioned,

youth bulges serve as one of the most robust predictors of repression (behind only civil war)

within the literature. Given this, to show that the economic makeup of the country can

moderate this relationship proves to be a key ﬁnding and important development in the

state repression literature. Beyond this, chapter two also contributes to the policy and in-

ternational development realms as well. As developed states seek to promote human rights

and economic growth within developing states, this ﬁnding shows that those two goals may

be attained using the same tools. In some cases, particularly those states with high youth

levels, it seems that economic development and increased respect for human rights, can be

simultaneously attained given continued stimulation of the local economy. This point to

117

important overlap between the development and human rights related agencies.

Chapter three continues with the economic approach to state repression and considers

another source of preemptive action: economic shocks. These sharp downturns in the econ-

omy, much like youth bulges, have the potential to cause mass unrest. This unrest proves

to be a threat to the regime as citizens become unhappy with the situation. Given this,

I show that states will employ preemptive repression as a way to shield themselves from

the potential upheaval these economic shocks can bring. Additionally, my analysis displays

a moderating relationship between economic shocks and level of development and realized

dissent. Here, I show that more developed states are more likely to increase their levels

of repression in the presence of an economic shock than democratic regimes. Similarly, I

ﬁnd that the marginal eﬀect of shocks is positive when there is no actual dissent further

suggesting that this repression is preemptive. This chapter provides another important con-

tribution to the repression literature by identifying a key predictor of preemptive repression.

If also identiﬁes an important limitation to economic development in reducing human rights

abuses.

It seems that economic development, in general, reduces repression.

Indeed, the

previously discussed chapter made this exact argument. However, chapter three of this dis-

sertation shows that these states, while generally less repressive, are vulnerable to increases

in repression during sharp economic downturns. The combined contribution of identifying a

previously unknown correlate of repression matched with linking that correlate to increased

repression in developed states, proves an important contribution to the existing literature

on state repression.

My ﬁnal empirical chapter shifts the focus slightly and investigates how economic factors

can be used to punish regimes for their human rights abuses.

Importantly though, this

chapter introduces the idea that foreign states can actually avoid these punishments by

hiring hi-powered Washington D.C. lobbying ﬁrms to lobby on their behalf. Through my

analysis I show that states can avoid having sanctions imposed and can reduce the severity

118

of sanctions through employing lobbyist. Beyond this, I also show that human rights abusing

states can increase their probability of receiving aid if they hire a lobbying ﬁrm and that this

lobbying can counteract the negative eﬀect human rights abuses have on foreign aid. Further,

these lobbyists can use their inﬂuence to make sure more of that the foreign aid is delivered

directly to the government rather than through "bypass" channels and this eﬀect is increased

in poor human rights countries. Thus, I ﬁnd that lobbying oﬀers human rights abusing states

a way to avoid being punished for their behavior. This is an important ﬁnding in many ways.

First, it shows the power of lobbying and sheds light on an, until now, under-studied aspect

of international human rights. It also provides important insight for policymakers as they

advocate for action against human rights abusers. With the information provided in this

dissertation, these policymakers can be more aware of the inﬂuence lobbying has and thereby

take steps to counteract that inﬂuence. Lastly, this chapter can help in shining a light on an

ugly corner of American foreign policy. By bringing to light the fact that American foreign

policy can eﬀectively be swayed by foreign governments, I hope to encourage action that

would make this process more transparent and thus less inﬂuential.

Taken together, this dissertation provides valuable insights into the relationship between

economics and state repression. This contribution not only helps to better understand the

determinants of preemptive repression but also show how states can avoid being punished

for using such tactics. While more work needs to be done, this dissertation provides unique

insight into an important aspect of human rights.

119

APPENDICES

120

APPENDIX A

CHAPTER 2 APPENDIX

A.0.1 Marginal Eﬀects Plot

Below I provide a graph showing the conditional eﬀect of youth unemployment across the

range of youth. As with my other marginal eﬀects plots, continuous variables were set at their

mean and civil war was set to 0. The ﬁgure displays the eﬀect of a one-unit increase in youth

unemployment across increasing levels of youth. The ﬁgure clearly shows that states with

high percentages of youth (the sample average is 29%) are likely to increase their repression

as the youth unemployment rate increases.

I display the results from interacting youth

unemployment rather than GDP Growth due to my agreement with the early discussion

that this better captures the economic situation of the youth. I have added this graph to

the appendix of the paper.

121

Figure A.1: Marginal Eﬀects Plot: Unemployment

A.0.2 Robustness Check with Alternative Dissent Data

Some have argued that the CNTS protest data used in this analysis under reports dissent

(Salehyan et al. 2012). In order to check the robustness of my results, I also use dissent data

from the Mass Mobilization Project colleced by Clark and Regan (2016). Clark and Regan

gather protest data from 162 countries for the years 1990-2014. To be considered a protest

there must be at least 50 people attending the event. This is contrasted with CNTS which

puts the cutoﬀ at 100. I run the models in my analysis using the same interaction term

and same controls variables (while using dissent data from Clark and Regan (2016) instead

of from Banks) in order to show the robustness of my results. The results are contained in

Table 3 and I graph the interaction based upon models one, three, and six below. The results

again show that the inclusion of the alternative measure for dissent does little to change the

122

substantive ﬁndings of the paper.

The results from Table A.1 again show the statistically signiﬁcant eﬀect the interaction

between youth bulges and GDP growth has on repression. This relationship is again captured

in ﬁgures A.2, A.3, and A.4 which all show little to no change in repression when economic

growth is high but a drastic increase in repression when economic growth levels are low.

Thus lending support to the hypothesis that the eﬀect youth bulges have on repression is

conditioned by the economic state of the country.

Figure A.2: Marginal Eﬀect of GDP Growth on Repression. Bars indicate 90 percent conﬁdence intervals

123

Table A.1: Eﬀect of Youth Bulges on Repression by GDP Growth

124

Figure A.3: Marginal Eﬀect of GDP Growth on Repression. Bars indicate 90 percent conﬁdence intervals

125

Figure A.4: Marginal Eﬀect of GDP Growth on Repression. Bars indicate 90 percent conﬁdence intervals

126

A.0.3 Alternative Mechanisms of Control: Education Spending, State Capacity,

and Civil Society

Economic development can be seen as an alternative to repression.

It could be that

governments face youth bulges by increasing spending on things like education as a way to

buy oﬀ the potential threat. Indeed the literature points to a connection between education
spending and general economic development.1 To account for such a strategy, I include

within the model spending on education taken from the World Bank. This measure is

the total amount of spending on education as a percent of that country’s GDP. Below I

include the regression table and a graph. As both show, accounting for an increase in

government spending on education does little to inﬂuence the increase in repression seen

when the economic situation of the state is weak. The general trend holds that when the

state is faced with a high number of youth and a poor economic outlook, the state chooses

to repress as a way to insulate itself from the potential future threat.

1See Hansson and Henrekson (1994)

127

Table A.2: Eﬀect of Youth Bulges on Repression Controlling for Education Spending

128

Figure A.5: Marginal Eﬀect: From Model 2

Government spending on education no doubt signals a regime with a certain amount of

capacity. As others have pointed out, state capacity is a multi-dimensional concept (Hendrix

2010). Following Hendrix (2010), one dimension of state capacity is already included in the

model: GDP per capita. Going beyond this, I also include another measure of state capacity

recommended by Hendrix (2010) – total taxes as a proportion of GDP. I use data compiled

by Hendrix (2010) and re-run my models using this measure of state capacity in place of my

former measure of GDP per capita. As the graph shows, the interactive eﬀect between youth

unemployment and youth bulges still moderates the relationship with repression. Those

states with high unemployment increase repression as the number of youth increases even

when controlling for the capacity of the state.

129

Figure A.6: Marginal Eﬀect: From Model 2

To help account for institutions that could harness the youth bulge, I control for civil

society in the state.

I use data from VDEM, speciﬁcally the question asking "Which of

these best describes the involvement of people in civil society organizations (CSOs)?" Higher

numbers indicate higher levels of participation (Coppedge et al. 2020). The models include

data from 1991-2012. As the graph below shows, the eﬀect of GDP growth and youth bulges

on repression remains unchanged when including civil society within the model.

130

Figure A.7: Marginal Eﬀect: From Model 2

131

APPENDIX B

CHAPTER 3 APPENDIX

B.0.1 Currency Crisis Results

Table B.1 shows the regression results when using currency crises as a measure of eco-

nomic shocks rather than inﬂation crises. This data, like the inﬂation data, was taken from

Reinhart et al. (2019). The results hold with the coeﬃcient on economic shocks positive

and statistically signiﬁcant meaning that countries experiencing a shock in time t-1 will, on

average, experience higher levels of repression than those countries who fail to experience an

economic shock, all else equal.

Table B.1: Eﬀect of Economic Shocks on Repression Using Currency Crises Measure

132

B.0.2 Results with XPolity

Below, table B.2 shows the results when using the Xpolity measure of regime type from

(Vreeland 2008). The results remain largely unchanged with only model one showing a

switch in signiﬁcance. Here, we see that the coeﬃcient on the measure of economic shock (a

reduction in GDP growth of ﬁve percent or more) just misses signiﬁcance with a p value of

.19.

Table B.2: Eﬀect of Economic Shocks on Repression Using Xpolity Measure

133

B.0.3 Robustness Check: Education Spending

Below I provide the results when including education spending in the model. Due to

the temporal range of the data, there are a signiﬁcant amount of observations lost over the

original analysis. Despite the incorporation of education spending, table B.3 shows that

economic shocks increase the amount of repression within the state. The only shock variable

that sees a change in signiﬁcance is the change in GDP growth variable which becomes

insigniﬁcant when including education spending. Given that this measure of economics

shocks is a bit broad, I am not overly concerned that it loses signiﬁcance. The eﬀect of

economic shocks can be clearly seen when looking at every other model.

134

Table B.3: Eﬀect of Economic Shocks on Repression Controlling for Education Spending

135

B.0.4 Tests of Non-Linearity

It could be that the relationship between economic shocks and repression is non-linear. To

test for this possibility I regress repression on economic shocks and examine the component

plus residual graph. As ﬁgures B.1 and B.2 show, there is no real evidence of a non-linear

relationship.

Figure B.1: Component Plus Residual Graph for Price Shocks

136

Figure B.2: Component Plus Residual Graph for Price Shocks

137

B.0.4.1

Interactive Eﬀect of Shocks and Dissent

Here I provide a graph of the predicted level of repression when overt dissent is present.

As the graph shows, there is no signiﬁcant diﬀerence in the predicted amount of repression.

Figure B.3: Predicted Level of Repression with Overt Dissent Present

138

APPENDIX C

CHAPTER 4 APPENDIX

C.0.1 Details of Lobbying Contracts with Foreign governments

Below, I provide a copy of the contract details between Qorvis and the Government of
Bahrain as reported to the Department of Justice. The contract shown covers the time period
when the youth delegation (highlighted in the introduction) was deployed to Washington
D.C. As you can see, the contract covers things like coordinating meetings with the media,
and NGOs as well as drafting opinion editorials and contacting members of Congress and
the media.

Figure C.1: FARA Report - Bahrain: December 2012

139

Here I provide a copy of the 2012 contract between Brownstein Hyatt Farber Schreck,
LLP and Mexico. As the report shows, the purpose of the lobbying is more speciﬁc in this
example than in others where the nature of services are far more generic.

Figure C.2: FARA Report - Mexico: 2012

Figure C.3 displays the coding rules used by Wood (2008) and Neuenkirch and Neumeier

(2015) to code sanction severity.

Figure C.3: Coding Rules for Sanction Severity

140

C.0.2

Interactive Results from Model 4 Table 4.5 Showing the Eﬀect of Lob-
bying and Repression on Probability of Severe Sanctions

Figure C.4 shows the interactive eﬀect of lobbying and repression. We see that across the
values of repression, lobbying has a consistently negative eﬀect on the probability of severe
sanctions, however, this ﬁnding fails to gain statistical signiﬁcance.

Figure C.4: Predicted Probability of Severe Sanctions by Lobbying Across Values of Repression

141

C.0.3 Sanctions Analysis with Fixed Eﬀects

Here I provide the results from the analysis of U.S. sanctions using a ﬁxed eﬀects logistic
regression. As table C.1 shows, the results remain unchanged from the models presented in
the body of the paper. The number of observations is less as those panels without any U.S.
sanctions were dropped. Regardless, lobbying still reduces the probability that a state will
be sanctioned by the U.S.

Lobbying (t-1)

Repression (t-1)

Lobbying * Repression

NYT Shaming (t-1)

Democracy (t-1)

Alliance

GDP per capita (t-1)

Population

Trade (t-1)

N
Fixed Eﬀects
Standard errors in parentheses.
∗ p < 0.10, ∗∗ p < 0.05

(Model 1)
-0.08∗∗
(0.035)
1.89∗∗
(0.190)

(Model 2)
-0.13∗∗
(0.048)
1.98∗∗
(0.201)

0.06

(0.044)
0.01∗∗
(0.004)
-1.12∗∗
(0.267)

0.07
(0.044)
0.01∗∗
(0.004)
-1.10∗∗
(0.268)

-1.32∗∗
(0.303)

0.22
(0.393)
0.01∗∗
(0.004)
1580
YES

-1.32∗∗
(0.303)

0.20
(0.390)
0.01∗∗
(0.004)
1580
YES

Table C.1: Eﬀect of Lobbying and Human Rights on U.S. Sanctions: 1978-2012

142

C.0.4 Foreign Aid Analysis: Military and Economic Aid with Latent Repression

and Empowerment Measure of Human Rights Abuses

The table below shows the results when using the measure of repression from Cingranelli
et al. (2014) coupled with military aid. Additionally, the table displays the results when using
the measure of repression from Fariss (2014) paired with economic aid. As the table shows,
economic aid behaves quite diﬀerently than military aid when it comes to human rights
abuses. We see that greater human rights abuses both increase the probability of receiving
economic aid and increase the amount assuming you receive aid. This proves counter to how
many thing about how human rights are connected to U.S. aid. Conversely, military aid acts
as we would expect. Here, human rights abuses (higher empowerment rights values) lead
to a reduced probability of receiving U.S. aid. It would seem then that while economic aid
is sensitive to empowerment rights abuses, military aid is sensitive to both empowerment
rights and physical integrity rights.

143

Type of Aid:

Economic

Economic

Military

Military

Lobbying

Repression (t-1)

Empowerment Rights (t-1)

Level of Aid

0.02∗∗
(0.008)
0.07∗∗
(0.034)

Selection Level of Aid Selection
0.03∗
(0.015)
0.14∗∗
(0.039)

0.00
(0.011)

0.01
(0.011)

Polity (t-1)

Ln GDPPC (t-1)

Ln Population

Ln Trade (t-1)

US Ally

Cold War

Post 9/11

t

t-2

t-3

Constant

λ

Country Fixed Eﬀects
N
Standard errors in parentheses
∗ p < 0.10, ∗∗ p < 0.05

0.03∗∗
(0.005)
-0.44∗∗
(0.052)
1.05∗∗
(0.157)
0.03
(0.022)
0.62∗∗
(0.162)
0.48∗∗
(0.065)
0.14∗∗
(0.052)

-11.30∗∗
(2.675)

Yes
4367

0.01∗∗
(0.006)
-0.46∗∗
(0.053)
-0.09∗∗
(0.032)
0.03
(0.023)

0.15
(0.092)
-0.51∗∗
(0.085)
0.47∗∗
(0.110)
-0.67∗∗
(0.029)
0.04∗∗
(0.003)
-0.00∗∗
(0.000)
6.71∗∗
(0.704)
-0.43∗∗
(0.068)

No
4367

-0.00
(0.015)
-0.00
(0.008)
-0.30∗∗
(0.077)
-0.69∗∗
(0.256)
-0.02
(0.038)

0.22
(0.265)
0.37∗∗
(0.084)
0.38∗∗
(0.066)

13.73∗∗
(4.139)

Yes
3356

-0.10∗∗
(0.012)
-0.00
(0.007)
-0.09∗∗
(0.039)
0.02
(0.026)
0.04∗
(0.020)
-0.21∗∗
(0.072)
0.15∗∗
(0.069)
0.01
(0.066)
-0.91∗∗
(0.075)
0.06∗∗
(0.007)
-0.00∗∗
(0.000)
0.11
(0.586)
-0.33∗∗
(0.130)

No
3356

Table C.2: Foreign Aid: Using Latent Repression and Empowerment Rights Indicators of
Human Rights Abuses

144

C.0.5 Sanctions Analysis: Severity Models with Restricted Sample

Model 3 Model 4
(Severity)
(Severity)
-0.20∗∗
(0.102)
0.81∗∗
(0.381)

-0.16
(0.236)
0.74
(0.586)
-0.03
(0.135)

Lobbying (t-1)

Repression (t-1)

Lobbying * Repression

NYT Shaming (t-1)

Democracy (t-1)

Alliance

GDP per capita (t-1)

Population

Trade (t-1)

Cut 1

Cut 2

N
Standard errors clustered on country.
∗ p < 0.10, ∗∗ p < 0.05

0.01
(0.006)
-0.86
(0.582)
-1.55∗∗
(0.697)
0.54∗
(0.315)
0.16
(0.225)
0.01∗∗
(0.003)
7.67
(5.642)
9.48∗
(5.670)

486

0.01
(0.006)
-0.85
(0.603)
-1.54∗∗
(0.717)
0.55∗
(0.298)
0.16
(0.228)
0.01∗∗
(0.002)
7.57
(5.824)
9.38
(5.847)

486

Table C.3: Eﬀect of Lobbying and Human Rights on U.S. Sanctions: 1978-2012

145

Figure C.5: Predicted Probability of Severe Sanctions Across Range of Lobbying

Here I show the results when using the sanctions severity dependent variable on the re-
stricted sample of only country-year observations with a U.S. sanction. We see that lobbying
still has a negative eﬀect on sanction severity when treating the variable as zero equalling
mild sanctions and 2 equalling severe sanctions. Figure C.5 shows the predicted probabil-
ity of severe sanctions over the values of lobbying. This clearly depicts the negative eﬀect
lobbying has on the probability of receiving severe sanctions.

146

BIBLIOGRAPHY

147

BIBLIOGRAPHY

Abdullah, I. (1998). Bush Path to Destruction: The Origin and Character of the Rev-
olutionary United Front/Sierra Leone. The Journal of Modern African Studies 36 (2),
203–235.

Aitchison, J. J. (1986). The statistical analysis of compositional data. Chapman and Hall.

Alesina, A. and D. Dollar (2000). Who Gives Foreign Aid to Whom and Why? Journal of

Economic Growth 5 (1), 33–63.

Apodaca, C. and M. Stohl (1999). United States Human Rights Policy and Foreign Assis-

tance. International Studies Quarterly 43 (1), 185–198.

Austen-Smith, D. (1993).

Information and Inﬂuence: Lobbying for Agendas and Votes.

American Journal of Political Science 37 (3), 799.

Austen-Smith, D. and J. R. Wright (1994). Counteractive Lobbying. American Journal of

Political Science 38 (1), 25.

Banks, A. and K. Wilson (2017). Cross-National Time-Series Data Archive. Databanks

International Jerusalem,.

Barron, P., K. Kaiser, and P. P. Menno (2016). Local Conﬂict in Indonesia: Measuring
Incidence and Identifying Patterns by Patrick Barron, Kai Kaiser, Menno Prasad Pradhan
:: SSRN. World Bank Policy Research Working Paper No. 3384 .

Baum, M. A. and P. B. Potter (2008). The Relationships Between Mass Media, Public
Opinion, and Foreign Policy: Toward a Theoretical Synthesis. Annual Review of Political
Science 11 (1), 39–65.

Baum, M. A. and P. B. K. Potter (2019). Media, Public Opinion, and Foreign Policy in the

Age of Social Media. The Journal of Politics 81 (2).

Baumgartner, F. R. and B. D. Jones (1993). Agendas and instability in American politics.

University of Chicago Press.

Bazzi, S. and C. Blattman (2014). Economic Shocks and Conﬂict: Evidence from Commodity

Prices †. American Economic Journal: Macroeconomics 6 (4), 1–38.

BBC News (2018, 1). Reality Check: How can a US president cut oﬀ foreign aid? - BBC

News.

Becker, G. S. (1968). Crime and Punishment: an Economic Approach. Journal of Political

Economy 76 (2), 169–217.

148

Berman, E., M. Callen, J. H. Felter, and J. N. Shapiro (2011). Do Working Men Rebel?
Insurgency and Unemployment in Afghanistan, Iraq, and the Philippines. Journal of
Conﬂict Resolution 55 (4), 496–528.

Blattman, C. and E. Miguel (2010). Civil war. Journal of Economic Literature 48 (1), 3–57.

Blumenthal, P. (2011, 1). Lobbying Contacts By Egypt’s Washington Lobbyists : Sunlight

Foundation.

Bogardus, K. (2013, 10). Egypt turns to K Street after U.S. aid cut | TheHill.

Bond, P. (2007). Competing Explanations of Zimbabwe’s Long Economic Crisis. The Journal

of South African and American Studies 8 (2), 149–181.

Boutton, A. and D. B. Carter (2014). Fair-Weather Allies? Terrorism and the Allocation of

US Foreign Aid. Journal of Conﬂict Resolution 58 (7), 1144–1173.

Bradley, J. R. (2012). After the Arab spring : how the Islamists hijacked the Middle East

revolts. Palgrave Macmillan.

Brambor, T., W. R. Clark, M. Golder, N. Beck, F. Boehmke, M. Gilligan, S. N. Golder,
and J. Nagler (2006). Understanding Interaction Models: Improving Empirical Analyses.
Political Analysis 14, 63–82.

Brett, R. and I. Specht (2004). Young soldiers : why they choose to ﬁght. Lynne Rienner

Publishers.

Carey, S. C. (2006). The Dynamic Relationship Between Protest and Repression. Political

Research Quarterly 59 (1), 1–11.

Carey, S. C. (2010). The Use of Repression as a Response to Domestic Dissent. Political

Studies 58 (1), 167–186.

Carleton, D. and M. Stohl (1985). The Foreign Policy of Human Rights: Rhetoric and

Reality from Jimmy Carter to Ronald Reagan. Human Rights Quarterly 7 (2), 205.

Cheibub, J. A., J. Gandhi, and J. R. Vreeland (2010). Democracy and dictatorship revisited.

Public Choice 143 (1-2), 67–101.

Chenoweth, E., E. Perkoski, and S. Kang (2017). State Repression and Nonviolent Resis-

tance. Journal of Conﬂict Resolution 61 (9), 1950–1969.

Cingranelli, D., D. Richards, and K. C. Clay (2014). The CIRI Human Rghts Dataset.

Cingranelli, D. L. and T. E. Pasquarello (1985). Human Rights Practices and the Distri-
bution of U.S. Foreign Aid to Latin American Countries. American Journal of Political
Science 29 (3), 539.

149

Cingranelli, D. L., D. L. Richards, S. Human, R. Quarterly, and N. May (2010). The Cin-
granelli and Richards (CIRI) Human Rights Data Project. Human Rights Quarterly 32 (2),
401–424.

Clark, D. and P. Regan (2016). Mass Mobilization Protest Data. Harvard Dataverse.

Cohen, B. C. (1963). Press and foreign policy. Princeton University Pres.

Collier, P. (2000). Doing Well Out of War: An Economic Perspective. In M. Berdal and
D. Malone (Eds.), Greed & Grievance: Economic Agendas in Civil Wars. Lynne Rienner.

Collier, P. and A. Hoeﬄer (2004). Greed and grievance in civil war.

fairs 56 (4), 757–777.

International Af-

Conrad, C. R. and W. H. Moore (2010). What Stops the Torture? American Journal of

Political Science 54 (2), 459–476.

Conrad, Courtenay R Moore, W. H. (2010). What Stops the Torture ? American Journal

of Political Science 54 (2), 459–476.

Coppedge, M., John Gerring, Carl Henrik Knutsen, Staﬀan I. Lindberg, Jan Teorell, David
Altman, Michael Bernhard, M. Steven Fish, Adam Glynn, Allen Hicken, Anna Luhrmann,
Kyle L. Marquardt, Kelly McMann, Pamela Paxton, Daniel Pemstein, Brigitte Seim,
Rachel Sigman, Svend-Erik Skaaning, Jeﬀrey Staton, Steven Wilson, Agnes Cornell,
Nazifa Alizada, Lisa Gastaldi, Haakon Gjerløw, Garry Hindle, Nina Ilchenko, Laura
Maxwell, Valeriya Mechkova, Juraj Medzihorsky, Johannes von Römer, Aksel Sundström,
Eitan Tzelgov, Yi-ting Wang, Tore Wig, and Daniel Ziblatt. (2020). V-Dem [Coun-
try–Year/Country–Date] Dataset v10.

Cox, D. G. and A. C. Drury (2006, 11). Democratic Sanctions: Connecting the Democratic

Peace and Economic Sanctions. Journal of Peace Research 43 (6), 709–722.

Cushman, C. (2018). Foreign Government Eﬀorts to Inﬂuence Congressional Foreign and
Defense Policy.
In J. A. Thurber, C. C. Campbell, and D. A. Dulio (Eds.), Congress
and Diaspora Politics : The Inﬂuence of Ethnic and Foreign Lobbying, pp. 19–43. Albany:
SUNY Press.

Danneman, N. and E. H. Ritter (2014). Contagious Rebellion and Preemptive Repression.

Journal of Conﬂict Resolution 58 (2), 254–79.

Dashwood, H. S. (2000). Zimbabwe : the political economy of transformation. University of

Toronto Press.

Davenport, C. (1995, 8). Multi-Dimensional Threat Perception and State Repression: An
Inquiry into Why States Apply Negative Sanctions. American Journal of Political Sci-
ence 39 (3), 683.

Davenport, C. (2007a). State Repression and Political Order. Annual Review of Political

Science 10 (1), 1–23.

150

Davenport, C. (2007b). State Repression and the Domestic Democratic Peace. Cambridge:

Cambridge University Press.

Davenport, C. (2007c). State Repression and the Tyrannical Peace. Journal of Peace Re-

search 44 (4), 485–504.

Davenport, C. and B. Appel (2014). Stopping State Repression: An Examination of Spells,

1976-2004.

Davenport, C. and D. A. Armstrong (2004, 7). Democracy and the Violation of Human
Rights: A Statistical Analysis from 1976 to 1996. American Journal of Political Sci-
ence 48 (3), 538–554.

Davenport, C. and M. Inman (2012). The State of State Repression Research Since the

1990s. Terrorism and Political Violence 24 (4), 619–634.

Davenport, C. and W. H. Moore (2012). The Arab Spring, Winter, and Back Again?
International Interac-

(Re)Introducing the Dissent-Repression Nexus with a Twist.
tions 38 (5).

de Mesquita, B. B. and A. Smith (2007). Foreign Aid and Policy Concessions. Journal of

Conﬂict Resolution 51 (2), 251–284.

Demirel-Pegg, T. and J. Moskowitz (2009). US Aid Allocation: The Nexus of Human Rights,

Democracy, and Development. Journal of Peace Research 46 (2), 181–198.

Dietrich, S. (2013). Bypass or Engage? Explaining Donor Delivery Tactics in Foreign Aid

Allocation*. International Studies Quarterly 57, 698–712.

Dietrich, S. and A. Murdie (2017). Human rights shaming through INGOs and foreign aid

delivery. Review of International Organizations 12 (1), 95–120.

Drezner, D. W. (1999). The sanctions paradox : economic statecraft and international rela-

tions. Cambridge University Press.

Drury, A. C. (2000). U.S. Presidents and the Use of Economic Sanctions.

Drury, A. C. (2001). Sanctions as Coercive Diplomacy: The U. S. President’s Decision to

Initiate Economic Sanctions. Political Research Quarterly 54 (3), 485.

Dube, O. and J. F. Vargas (2013). Commodity Price Shocks and Civil Conﬂict: Evidence

from Colombia. The Review of Economic Studies 80 (4), 1384–1421.

Elbadawi, E. and N. Sambanis (2000). Why are there so many civil wars in Africa? Under-

standing and preventing violent conﬂict. Journal of African Economies 9 (3), 244–269.

Elliott, J. (2012, 2). The American PR Firm Helping Out Bahrain’s Brutal Monarchy - The

Atlantic.

151

Esarey, J. and J. H. DeMeritt (2017). Political Context and the Consequences of Naming

and Shaming for Human Rights Abuse. International Interactions 43 (4), 589–618.

FARA (2009). FARA Report. Technical report.

FARA (2012). FARA. Technical report.

FARA (2014). FARA Semi-Annual Report. Technical report, Department of Justice.

FARA (2015). FARA Semi-Annual Report. Technical report, Department of Justice.

Fariss, C. (2014). Respect for Human Rights has Improved Over Time: Modeling the Chang-

ing Standard of Accountability. American Political Science Review 108 (2), 297–318.

Fearon, J. D. and D. D. Laitin (2003). Ethnicity, Insurgency, and Civil War. American

Political Science Review 97 (1), 75–90.

Feingold, R. D. (2010, 7). S.Res.586 - 111th Congress (2009-2010): A resolution supporting

democracy, human rights, and civil liberties in Egypt.

Fleck, R. K. and C. Kilby (2001). Foreign Aid and Domestic Politics: Voting in Congress
and the Allocation of USAID Contracts across Congressional Districts. Southern Economic
Journal 67 (3), 598.

Franklin, J. C. (2008). Shame on You: The Impact of Human Rights Criticism on Political

Repression in Latin America. International Studies Quarterly 52 (1), 187–211.

Freeman, B. and L. Dennett (2014). Loopholes, Filing Failures, and Lax Enforcement: How
the Foreign Agents Registration Act Falls Short. Technical report, Project on Government
Oversight.

Freeman, B. J. (2009). The Best Foreign Policy Money Can Buy?: An Investigation of

Foreign Lobbying and U.S. Foreign Policy. Ph. D. thesis, Texas A&M.

Galetovic, A. and R. Sanhueza (2000). Citizens, Autocrats, and Plotters: A Model and New

Evidence on Coups D’Etat. Economics and Politics 12 (2), 183–204.

Galtung, J. (1967). On the Eﬀects of International Economic Sanctions, With Examples

from the Case of Rhodesia. World Politics 19 (3), 378–416.

Galtung, J. (1969). Violence, Peace, and Peace Research. Journal of Peace Research 6 (3),

167–191.

Gartner, S. S. and P. M. Regan (1996, 8). Threat and Repression: The Non-Linear Rela-
tionship between Government and Opposition Violence. Journal of Peace Research 33 (3),
273–287.

Gates, S. (2002). Recruitment and Allegiance. Journal of Conﬂict Resolution 46 (1), 111–130.

152

Gelleny, R. D. and M. Mccoy (2001, 9). Globalization and Government Policy Independence:

The Issue of Taxation. Political Research Quarterly 54 (3), 509–529.

Gilboa, E. (2005). The CNN eﬀect: The search for a communication theory of international

relations. Political Communication 22 (1), 27–44.

Gleditsch, K. S. (2002). Expanded Trade and GDP Data. Journal of Conﬂict Resolu-

tion 46 (5), 712–724.

Gleditsch, N. P., P. Wallensteen, M. Eriksson, M. Sollenberg, and H. Strand (2002). Armed

Conﬂict 1946-2001: A New Dataset. Journal of Peace Research 39 (5), 615–637.

Goenner, C. F. (2007). Economic war and democratic peace. Conﬂict Management and

Peace Science 24 (3), 171–182.

Goldstein, R. J. (1983). Political repression in 19th century europe. Taylor & Francis.

Goldstone, J. A. (1991). Revolution and rebellion in the early modern world. University of

California Press.

Gruss, B. and S. Kebhaj (2019). Commodity Terms of Trade : A New Database.

Working Paper.

IMF

Gurr, T. (1970). Why Men Rebel. Princeton: Princeton University Press.

Hafner-Burton, E. (2008). Sticks and Stones: Naming and Shaming the Human Rights

Enforcement Problem. International Organization 62 (04).

Hafner-Burton, E., M. Kahler, and A. Montgomery (2009). Network Analysis for Interna-

tional Relations. International Organization 63, 559–592.

Hafner-Burton, E. M. and A. H. Montgomery (2008). Power or Plenty How Do International
Journal of Conﬂict Resolution 52 (2),

Trade Institutions Aﬀect Economic Sanctions?
213–242.

Hansen, J. M. (1991). Gaining access : Congress and the farm lobby, 1919-1981. University

of Chicago Press.

Hansson, P. and M. Henrekson (1994). A new framework for testing the eﬀect of government

spending on growth and productivity. Public Choice 81 (3-4), 381–401.

Hart, D., R. Atkins, P. Markey, and J. Youniss (2004). Youth Bulges in Communities.

Psychological Science 15 (9), 591–597.

Helleiner, E. (2011, 6). Understanding the 2007–2008 Global Financial Crisis: Lessons for
Scholars of International Political Economy. Annual Review of Political Science 14 (1),
67–87.

Hendrix, C. S. (2010). Measuring state capacity: Theoretical and empirical implications for

the study of civil conﬂict. Journal of Peace Research 47 (3), 273–285.

153

Hill, D. W. and Z. M. Jones (2014). An Empirical Evaluation of Explanations for State

Repression. The American Political Science Review 108 (3), 661–687.

Hodler, R. and P. A. Raschky (2014). Economic shocks and civil conﬂict at the regional

level. Economics Letters 124 (3), 530–533.

Houle, C. (2017). Social Mobility and Political Instability. Journal of Conﬂict Resolution,

1–27.

Houle, C., M. Kayser, and J. Xiang (2016). Diﬀusion or Confusion? Clustered Shocks and

the Conditional Diﬀusion of Democracy. International Organization.

Hufbauer, G., J. Schott, K. Elliott, and B. Oegg (2009). Economic Sanctions Reconsidered,

3rd Edition . Washington D.C.: Peterson Institute for International Economics.

Humphreys, M. and J. M. Weinstein (2008). Who Fights? The Determinants of Participation

in Civil War. American Journal of Political Science 52 (2), 436–455.

Huntington, S. P. (1996). The Clash of Civilizations and the Remaking of World Order.

Simon & Schuster.

Irwin, L. G. (2000). Dancing the Foreign Aid Appropriations Dance: Recurring Themes in
the Modern Congresses. Public Budgeting <html_ent glyph="@amp;" ascii="&amp;"/>
Finance 20 (2), 30–48.

Itkowitz, C. (2014, 5). Which foreign countries spent the most to inﬂuence U.S. politics? -

The Washington Post.

Iyengar, S. and D. Kinder (1987). News That Matters: Television and American News that

matters: Agenda-Setting and priming in a television age. University of Chicago Press.

Jacobson, G. C. (2007). A divider, not a uniter : George W. Bush and the American people.

Pearson Longman.

Johnsson, S. E. (2006). Speech at the 95th International Labor Conference. Geneva.

Kaufmann, D., A. Kraay, and Mastruzzi. Massimo (2009). Governance Matters VIII Aggre-

gate and Individual Governance Indicators. World Bank Working Paper.

Keck, Margaret E Sikkink, K. (1998). Keck, Margaret E., and Sikkink, Kathryn. Activists
beyond Borders : Advocacy Networks in International Politics. Ithaca, US: Cornell Uni-
versity Press, 1998. ProQuest ebrary. Web. 10 November 2016. Copyright c(cid:13) 1998. Cornell
University Press. All rights res. Number November.

Kim, N. K. (2016). Revisiting Economic Shocks and Coups. Journal of Conﬂict Resolu-

tion 60 (1), 3–31.

Knickmeyer, E. (2011). The Arab World’s Youth Army. Foreign Policy.

154

Korenman, S. and D. Neumark (1997). Cohort Crowding and Youth Labor Markets: A

Cross-National Analysis.

Krain, M. (2012). J’accuse! Does Naming and Shaming Perpetrators Reduce the Severity of

Genocides or Politicides? International Studies Quarterly 56 (3), 574–589.

Kuziemko, I. and E. Werker (2006). How much is a seat on the Security Council worth?
Foreign aid and bribery at the United Nations. Journal of Political Economy 114 (5),
905–930.

Lai, B. (2003). Examining the Goals of US Foreign Assistance in the Post-cold War Period,

1991-96. Journal of Peace Research 40 (1), 103–128.

Lall, R. (2012, 10). From failure to failure: The politics of international banking regulation.

Review of International Political Economy 19 (4), 609–638.

Lebovic, J. H. (1988). National Interests and US Foreign Aid: The Carter and Reagan Years.

Journal of Peace Research 25 (2), 115–135.

Lebovic, J. H. and E. Voeten (2009). The Cost of Shame: International Organizations and
Foreign Aid in the Punishing of Human Rights Violators. Journal of Peace Research 46 (1),
79–97.

Lee, S. (2007).

International public relations as a predictor of prominence of US news

coverage. Public Relations Review 33 (2), 158–165.

Leeds, B. A., J. M. Ritter, S. M. L. Mitchell, and A. G. Long (2002). Alliance treaty

obligations and provisions, 1815-1944. International Interactions 28 (3), 237–260.

Lektzian, D. J. and C. M. Sprecher (2007). Sanctions, Signals, and Militarized Conﬂict.

American Journal of Political Science 51 (2), 415–431.

Lichbach, M. I. (1987). from the SAGE Social Science Collections . All Rights Reserved .

Journal of Conﬂict Resolution 31 (2), 266–297.

Licht, A. A. (2010). Coming into Money: The Impact of Foreign Aid on Leader Survival.

Journal of Conﬂict Resolution 54 (1), 58–87.

Lin, J. Y. (2012). Youth Bulge: A Demographic Dividend or a Demographic Bomb in

Developing Countries? Technical report, The World Bank, Washington DC.

Lindsay, J. M. (1986). Trade Sanctions As Policy Instruments: A Re-Examination. Inter-

national Studies Quarterly 30 (2), 153.

Linzer, D. A. and J. K. Staton (2015). A Global Measure of Judicial Independence,

1948–2012. Journal of Law and Courts 3 (2), 223–256.

Lozocsky, I. (2015). How Azerbaijan and Its Lobbyists Spin Congress – Foreign Policy.

155

Malik, A. and B. Awadallah (2013). The Economics of the Arab Spring. World Develop-

ment 45, 296–313.

Marshall, M. G., T. R. Gurr, and K. Jaggers (2016). POLITY IV Project: Political Regime

Characteristics and Transitions, 1800-2015. Center for Systemic Peace 1st ed.

McKinlay, R. D. and R. Little (1977). A Foreign Policy Model of U.S. Bilateral Aid Alloca-

tion. World Politics 30 (1), 58–86.

Miguel, E. (2004). Tribe or Nation? Nation Building and Public Goods in Kenya versus

Tanzania. World Politics 56 (3), 327–362.

Miguel, E., S. Satyanath, and E. Sergenti (2004). Economic Shocks and Civil Conﬂict: An

Instrumental Variables Approach. Journal of Political Economy 112 (4), 725–753.

Milner, H. V. and D. H. Tingley (2009). The Political Economy of U.S. Foreign Aid: Amer-
ican Legislators and the Domestic Politics of Aid. Economics & Politics 22 (2), 200–232.

Montes-Rojas, G. V. (2013). Can Poor Countries Lobby for More US Bilateral Aid? World

Development 44, 77–87.

Moore, W. H. (1998). Repression and Dissent : Substitution , Context , and Timing.

American Journal of Political Science 42 (3), 851–873.

Moore, W. H. (2000). The Repression of Dissent: A Substitution Model of Government

Coercion. The Journal of Conﬂict Resolution 44 (1), 107–127.

Mulderig, M. C. (2013). An Uncertain Future: Youth Frustration and the Arab Spring.

Technical Report 16, Boston University.

Murdie, A. and D. Davis (2012). Shaming and Blaming: Using Events Data to Assess the

Impact of Human Rights INGOs. International Studies Quarterly 56 (1), 1–16.

Neuenkirch, M. and F. Neumeier (2015). The impact of UN and US economic sanctions on

GDP growth. European Journal of Political Economy 40, 110–125.

Neumayer, E. (2003a). Is respect for human rights rewarded? An analysis of total bilateral

and multilateral aid ﬂows. Human Rights Quaterly 25 (2), 510–527.

Neumayer, E. (2003b). The pattern of aid giving : the impact of good governance on devel-

opment assistance. Routledge.

Newhouse, J. (2009). Diplomacy, Inc.: The Inﬂuence of Lobbies on U.S. Foreign Policy.

Foreign Aﬀairs 88, 73–92.

Nielsen, R. A. (2013). Rewarding Human Rights? Selective Aid Sanctions against Repressive

States. International Studies Quarterly 57, 791–803.

Nordås, R. and C. Davenport (2013). Fight the Youth: Youth Bulges and State Repression.

American Journal of Political Science 57 (4).

156

Oberhofer, H. . and M. Pfaﬀermayr (2012). Fractional Response Models-A Replication

Exercise of Papke and Wooldridge. Journal of Applied Econometrics 6 (3), 619.

Papke, L. E. and J. M. Wooldridge (1996, 11). Econometric methods for fractional re-
sponse variables with an application to 401(k) plan participation rates. Journal of Applied
Econometrics 11 (6), 619–632.

Peksen, D. (2009). Better or Worse? The Eﬀect of Economic Sanctions on Human Rights.

Journal of Peace Research 46 (1), 59–77.

Peksen, D. and A. C. Drury (2009). Economic Sanctions and Political Repression: Assessing
the Impact of Coercive Diplomacy on Political Freedoms. Human Rights Review 10 (3),
393–411.

Peksen, D., T. M. Peterson, and A. C. Drury (2014). Media-driven Humanitarianism? News
Media Coverage of Human Rights Abuses and the Use of Economic Sanctions. Interna-
tional Studies Quarterly 58 (4), 855–866.

Pevehouse, J. C. W. and F. Vabulas (2019). Nudging the Needle: Foreign Lobbies and US

Human Rights Ratings. International Studies Quarterly 63 (1), 85–98.

Poe, S. C. and J. Meernik (1995). US Military Aid in the 1980s: A Global Analysis. Journal

of Peace Research 32 (4), 399–411.

Poe, S. C. and C. N. Tate (1994). Repression of Human Rights to Personal Integrity in the

1980s: A Global Analysis. American Political Science Review 88 (4), 853–872.

Popkin, S. L. (1979). The rational peasant : the political economy of rural society in Vietnam.

University of California Press.

Powlick, P. J. (1991). The Attitudinal Bases for Responsiveness to Public Opinion among

American Foreign Policy Oﬃcials. Journal of Conﬂict Resolution 35 (4), 611–641.

Quinn, E. (2015, 12). U.S. lobbying, PR ﬁrms give human rights abusers a friendly face –

Center for Public Integrity.

Reinhart, C., K. Rogoﬀ, C. Trebesch, and V. Reinhard (2019). Global Crises Data by

Country.

Reinhart, C. M. and K. S. Rogoﬀ (2011, 8). From ﬁnancial crash to debt crisis. American

Economic Review 101 (5), 1676–1706.

Ritter, E. H. and C. R. Conrad (2016). Preventing and Responding to Dissent: The Ob-
servational Challenges of Explaining Strategic Repression. American Political Science
Review 110 (1), 85–99.

Robinson, P. (2011). The CNN eﬀect reconsidered: mapping a research agenda for the future.

Media, War & Conﬂict 4 (1), 3–11.

157

Rogin, J. (2011, 2). How the Senate resolution on Egyptian democracy died – Foreign Policy.

Salehyan, I., C. S. Hendrix, J. Hamner, C. Case, C. Linebarger, E. Stull, and J. Williams
International Interactions 38 (4),

(2012). Social Conﬂict in Africa: A New Database.
503–511.

Sambanis, N. (2004). What Is Civil War?: Conceptual and Empirical Complexities of an

Operational Deﬁnition. Journal of Conﬂict Resolution 48 (6), 814–858.

Scott Bennett, D. and A. C. Stam (2000). EUGene: A conceptual manual. International

Interactions 26 (2), 179–204.

Sobel, R. (2001). The impact of public opinion on U.S. foreign policy since Vietnam :

constraining the colossus. Oxford University Press.

Tilly, C. (1978). From Mobilization to Revolution . McGraw-Hill.

Truex, R. (2019). Focal Points, Dissident Calendars, and Preemptive Repression. Journal

of Conﬂict Resolution 63 (4), 1032–1052.

Urdal, H. (2006). A Clash of Generations? Youth Bulges and Political Violence. International

Studies Quarterly 50, 607–629.

USAID (2013). The Potential of Youth For a Demographic Dividend: Investing in Health,

Education, and Job Creation. Technical report, USAID, Addis Ababa.

Vreeland, J. R. (2008). Political Institutions and Human Rights: Why Dictatorships Enter
into the United Nations Convention Against Torture. International Organization 62 (01),
65–101.

Whang, T. (2011). Playing to the Home Crowd? Symbolic Use of Economic Sanctions in

the United States1. International Studies Quarterly 55 (3), 787–801.

Wood, R. M. (2008). “A Hand upon the Throat of the Nation”: Economic Sanctions and

State Repression, 1976–2001. International Studies Quarterly 52, 489–513.

Wood, R. M. and M. Gibney (2010). The Political Terror Scale (PTS): A Re-introduction

and a Comparison to CIRI. Human Rights Quarterly 32 (2), 367–400.

World Bank (2011). World Development Report 2011: Conﬂict, Security, and Development.

Technical report, World Bank, Washington DC.

Wright, J. (2009). How Foreign Aid Can Foster Democratization in Autocratic Regimes.

American Journal of Political Science 53 (3), 552–571.

Wright, T. and S. Moorthy (2018). Refugees, Economic Capacity, and Host State Repres-

sion*. International Interactions 44 (1), 132–155.

158

Young, K. L. (2012a, 10). Transnational regulatory capture? An empirical examination of
the transnational lobbying of the Basel Committee on Banking Supervision. Review of
International Political Economy 19 (4), 663–688.

Young, L. (2012b, 12). Egyptian military aid still ﬂying high : Sunlight Foundation.

Young, L. E. (2019). The Psychology of State Repression: Fear and Dissent Decisions in

Zimbabwe. American Political Science Review 113 (1), 140–155.

Zapotosky, M. and T. Hamburger (2017, 11). Mueller probe appears to be prompting lob-

byists to register as foreign agents - ProQuest.

Zhang, J. and G. T. Cameron (2003). China’s agenda building and image polishing in the
US: Assessing an international public relations campaign. Public Relations Review 29 (1),
13–28.

159