THE POLITICS OF DECENTRALIZATION UNDER DICTATORSHIPS
By
Hsin-Hsin Pan

.

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Political Science - Doctor of Philosophy
2015

ABSTRACT
THE POLITICS OF DECENTRALIZATION UNDER DICTATORSHIPS
By
Hsin-Hsin Pan

In this dissertation, I argue that dictators face two principal-agent problems of defection and
compliance when they find themselves in the dilemma of empowering local elites via
decentralization. On one hand, dictators wish to sufficiently empower local elites that they can
deliver effective and efficient governance tailored to local needs. On the other hand, dictators do
not wish to empower local elites too much so that they are capable of imposing threats on the
political survival of authoritarian rule. Thus, dictators (principal) have to face two
principal-agent problems in the management of local elites (agent).
First, dictators face the problem of defection if unsatisfied local elites’ governing apparatus
switch to aid opposition forces conditional on the party system. The dual installation of
decentralization and party system decides if the collision of unsatisfied local elites (insider) with
opposition elites (outsider) is incentivized in the equation of regime revision. Second, the
problem of compliance refers to the local elites’ violation of central policy. Elected local elites
are motivated to over-commit to international investors so that they fail to comply with central
policies of maintain credibility in bilateral investment treaties. The two problems are derived
from the lack of an ultimate authority to define, interpret and settle disputes between central and
local governments under dictatorship.
Using statistical models, I explain why some decentralized dictatorships are more resistant
to authoritarian breakdown as decentralization is simultaneously installed with single party

system; I also ask how decentralization influences authoritarian credulity in bilateral investment
treaties that central government signs with foreign countries. I find that decentralization
facilitates the signing of BITs in the case of China, but hurts the maintenance of them in
dictatorships. Local elites are strategic players in the problem of defection and compliance.
The findings are expected to shed light on the literature of political institutions,
democratization and international cooperation for authoritarian rule. As the literature of political
institutions under authoritarian regime applauds the institutional effect on the political survival of
dictatorship, the combined effect of institutions are left out of discussion. Chapter 2 makes a
small step forward in this regard. In addition, the linkage between the literature of political
institution and international cooperation is of consequence to global order as more and more
dictatorships are engaged in international institutions.

To Mei-jiou Soong, a loving mom

iv

ACKNOWLEDGEMENTS

I would never finish this dissertation without all the generous support throughout the
graduate study in MSU and dissertation writing in Taipei. My adviser of PhD dissertation, Jeff
Conroy-Krutz has been incredibly patient with a mom graduate student. In every aspect, he is the
kind of adviser whose care for students is a legacy to be remembered and passed on. The adviser
of guidance committee, Eric Chang never gives up on me in time of frustration. He comes to my
life as immense blessings. My committee members Mark Axelrod and Ben Appel provide
excellent criticism valuable for my dissertation chapters to step forwards to journal submission.
My gratitude also goes to the members of guidance committee members for advising on course
plans, Cristina Bodea, Bill Jacoby, Ani Sarkissian and Chuck Ostrom. Also, I appreciate the
cutting-edged insights through the courses provided by Sandy Schenider, Michael Colaresi,
Steven Kautz, and Corwin Smidt. The staffs in MSU, Karen Battin, Sarah Kraus, and Rhonda
Burns are superb in assisting me at all stages.
In the end of my journey to the US, I cannot help looking back and thank the people who
make all these possible in the beginning. My adviser of Masters' thesis in NCCU, Szu-yin Ho has
placed more faith in me than I do and has done more for me than I deserve. A lighthouse in my
life indeed. I am grateful for Yi-Chou Liu, Shing-Yuan Sheng, and Cheng-Tien Kuo who open
my eyes to the discipline and pushed me through the admission to higher learning.
The graduate study and dissertation writing are generously funded by student loan vouched
by Jia-An Chiang, scholarship of PhD program in the US by the Fulbright Foundation,
scholarship of studying overseas by the Ministry of Education, Taiwan, research/teaching

v

assistantship in Department of Political Science, MSU, research grant under the supervision of
Yu-Tzung Chang and Yun-han Chu in National Taiwan University, and the pre-doctoral program
in the Institution of Political Science, Academia Sinica under the directorship of Yu-shan Wu and
mentorship of Chin-en Wu. I greatly appreciate the financial support from these people and
institutions.
During my graduate studies, I am extraordinarily lucky to have professors' and cohorts'
intellectual stimulation, companions or advice on job hunting. They are Jih-wen Lin, Tom Kou,
Szu-chien Hsu, Tse-Kang Leng, Alex Chang, Wen-hsuan Tsai, Yousun Cheng, Feng-yu Lee,
Chealsea Chou, Hans Tung, Chiung-chiu Huang, Helen Lee, Shih-hao Huang, Seoyoun Choi,
Dominique Conway, Hasaaki Higashijima, Fang-yu Chen, Hsiao-chi Hsu, Tse-hsin Chen,
Szuning Ping, Wan-Ling Huang, Teresa Lee, Weiling Chen, Chiahsin Yeh, Luke Chu, Alice Hsu,
Yung-hsiu Tang, Tung-wu Hsieh, Po-chun Huang, Lenore Hsu, Bin Hwang, Sienna Hsieh,
Yu-hao Lee, Mica Wang, Pei-Lin Lee, Sean Ma, Ling Zhu, Nora Hung, Yuhan Hung, Yingju Lai,
Huang-hua Chen, Scott Chiu, Rose Chen, Leo Lin, James Baldus, Alex Hsueh, Nien-Chung
Chang Liao, Wei-ting Wu, Chun-chih Chang, Ronan Fu, Judy Wei, Yen-bin Su and Ya-wen Yu.
Lastly, I am grateful for family members' childcare: Sheng-Tong Wu, Sho-Rong Huang,
Chi-Ching Wu, Tsui-Chin Chen, Pei-Lan Pan, Hsien-chih Shen, Kate Pan, Sheng-ping Huang
and Tze-Zheng Pan. My husband, Wen-Chin Wu, has made all the differences along the less
trodden road we've taken. His and our son, Infinity Wu's love sustains my academic survival with
significance. This dissertation is dedicated to my mom, Mei-jiou Soong who raises four children
after my father's decease at my age of 6. This is the kind of achievement I will never surpass.

vi

TABLE OF CONTENTS

CHAPTER 1 ........................................................................................................................................ 1
INTRODUCTION ............................................................................................................................... 1
1.1 Decentralization under Dictatorship ...................................................................................... 2
1.2 The Two Principal-Agent Problems ....................................................................................... 4
1.2.1 The Problem of Defection ........................................................................................... 4
1.2.2 The Problem of Compliance........................................................................................ 4
1.3 Plan of the Dissertation .......................................................................................................... 5
CHAPTER 2 ........................................................................................................................................ 8
POLITICAL DECENTRALIZATION, PARTY SYSTEM, AND AUTHORITARIAN
BREAKDOWN.................................................................................................................................... 8
2.1 Introduction ............................................................................................................................ 8
2.2 Definition ............................................................................................................................. 12
2.3 Political Decentralization, Party System, and Elite Management ....................................... 13
2.4 Effects of Political Decentralization and Party System on Authoritarian Breakdown ........ 18
2.5 Research Design................................................................................................................... 25
2.5.1 Independent Variables ............................................................................................... 26
2.5.2 Control Variables ....................................................................................................... 28
2.6 Results .................................................................................................................................. 30
2.7 Conclusion ........................................................................................................................... 36
CHAPTER 3 ...................................................................................................................................... 38
POLITICAL DECENTRALIZATION AND VIOLATION OF
BILATERAL
INVESTMENT TREATIES ............................................................................................................. 38
3.1 Introduction .......................................................................................................................... 38
3.2 Fiscal Decentralization, FDI Inflow and BIT Violations under Dictatorship ...................... 43
3.3 Political Decentralization, Over-Commitment and Authoritarian Credibility in BITs ........ 47
3.4 Research Design................................................................................................................... 50
3.4.1 Independent Variable ................................................................................................. 52
3.4.2 Control Variables. ...................................................................................................... 54
3.4.2.1 Domestic Politics Controls........................................................................... 54
3.4.2.2 International Politics Controls ..................................................................... 56
3.5 Results .................................................................................................................................. 57
3.6 Conclusion ........................................................................................................................... 61
CHAPTER 4 ..................................................................................................................................... 63
DECENTRALIZATION AND CHINA’S BILATERAL INVESTMENT TREATIES............. 63
4.1 Introduction .......................................................................................................................... 63
4.2 Decentralization, FDI and Signing BITs under Dictatorship ............................................... 67
4.3 Leadership and Institution as Proxies to Decentralization and China’s BITs ...................... 70
4.3.1 Leadership ................................................................................................................. 71
4.3.2 Tax-sharing System ................................................................................................... 75
vii

4.4

Research Design................................................................................................................... 77
4.4.1 Independent Variables ............................................................................................... 78
4.4.2 Control Variables ....................................................................................................... 80
4.5 Results .................................................................................................................................. 83
4.6 Conclusion ........................................................................................................................... 88
CHAPTER 5 .................................................................................................................................90
CONCLUSION .................................................................................................................................. 90
APPENDIX ........................................................................................................................................ 95
BIBLIOGRAPHY ........................................................................................................................... 105

viii

LIST OF TABLES

Table 1-1: Plan of Dissertation ..........................................................................................................7
Table 2-1: Political Decentralization and Party System under Dictatorship ...................................21
Table 2-2: Effects of Political Decentralization and Party System on Authoritarian Breakdown ...32
Table 3-1: Effect of Political Decentralization on BITs Violation: Domestic Factors ....................58
Table 3-2: Effect of Political Decentralization on BITs Violation: International Factors ...............60
Table 4-1: Effects of Leadership and Institution on China’s BITs ..................................................84
Table 4-2: Competing Proxies of Decentralization on China’s BITs ..............................................87
Table A-1: List of Dictatorships in the Sample of Chapter 2 ..........................................................96
Table A-2: Summary Statistics of Chapter 2 ...................................................................................97
Table A-3: List of Dictatorship in the Sample of Chapter 3 ............................................................98
Table A-4: Summary Statistics of Chapter 3 ...................................................................................99
Table A-5: List of Countries Maintaining Effective BITs with China...........................................100
Table A-6: Summary Statistics of Chapter 4 .................................................................................104

ix

LIST OF FIGURES

Figure 2-1: Effect of Political Decentralization and Party System on Authoritarian Breakdown:
(High, Dominant) ..........................................................................................................34
Figure 2-2: Effect of Political Decentralization and Party System on Authoritarian Breakdown:
(Low, Single) .................................................................................................................35
Figure 3-1: The Trend of FDI Inflow and ICSID Cases Invoked by BITs for Dictatorship ...........39
Figure 3-2: The Mean Number of ICSID Cases by Decentralization for Dictatorships .................42
Figure 4-1: The Number of China’s BITs and FDI Inflow: 1982-2009 ..........................................64
Figure 4-2: Total Number of Countries Signed BITs by Regime Type ...........................................65
Figure 4-3: Accumulated China’s BITs by Leadership Turnover ...................................................73
Figure 4-4: Accumulated China’s BITs before and after Deng’s Decease ......................................75
Figure 4-5: Accumulated China’s BITs before and after Tax-Sharing System ...............................76

x

CHAPTER 1
INTRODUCTION

A well-groomed local elite by the ruling party, Hsu Hsin-liang claimed his electoral victory
in the magistrate of Taoyuan city, a second-tier municipal body under authoritarian Taiwan ruled
by the Nationalist Party (also known as Kuomintang, KMT) in 1977. The ruling party’s chair
Chiang Ching-Kuo sent his congratulations by revoking Hsu’s party membership in return for his
unauthorized candidacy in the local election. Hsu soon leveraged his electoral base in Taoyuan
city to the rising opposition party (Wu and Chen 1993). In 1997, one year after the authoritarian
breakdown, Hsu chaired the major opposition party, the Democratic Progressive Party (DPP),
united more local elites defecting from KMT and won the majority of local elections. In 2000,
DPP overthrew KMT and finished the first-turn-over test in Taiwan. Local elites’ defection can
threaten the political survival of authoritarian regime. Even if they remain loyal to the ruling
party, they can threaten authoritarian regime by violating central policies.
As one of the founding fathers of People’s Republic of China (PRC), Chen Yun (陈云)
complained about how local leaders in the coastal provinces were taking matters in their own
hands by discounting the central agenda for economic development (Lam 1999). The general
secretary of the Communist Party of Vietnam, Nguyễn Phú Trọng, cannot keep the provincial
leaders comply with central law because the local elites keep a large flow of foreign direct
investment to generate rents at their disposal (Malesky 2008). Dictators face the same dilemma:
how to share power with local elites and yet keep them under control.

1

1.1

Decentralization under Dictatorship
1

The politics of decentralization under dictatorship is different from that under democracies
in how conflict between leader and local elites is resolved. For democracies, conflict is
ultimately resorted for the judicial review for dispute resolution. The highest judicial court is
responsible for interpreting and/or reinterpreting how the constitution regulates the relationship
between central and local governments. But dictatorships lack an independent judicial power to
settle the disputes between dictators and local elites. Just like other political institutions under
dictatorships, the lack of an independent and ultimate authority drives the politics of
authoritarian rule (Svolik 2012b).
According to previous research, decentralization refers to a multi-tiered governing system
with a shift of authority from central governments to local governments in charge of issues
2

confined within specific geographical locales (Rodden 2004). The policy power granted to local
governments generally include fiscal, administrative and political decentralization (Falleti 2010;
Green 2005; Treisman 2002). Fiscal decentralization refers to the independent authority to make
and execute policies with regard to revenue and expenditure of local governments.
Administrative decentralization means that local governments have independent discretion over
the personnel. Political decentralization points to the electoral source of ruling legitimacy via
local elections.

1

Throughout the dissertation, I use dictatorship and authoritarian regime interchangeably to
refer to non-democracies defined by Geddes, Wright, and Frantz (2014).
2
Political decentralization and federalism are used in previous literature. Recently researchers
gradually replace federalism with political decentralization to broadly include the countries
without de jure entitlement of federalism, but with de facto practice of independent
decision-making power for local governments such as Spain.
2

Decentralization is understudied in the literature of authoritarian institutions. Nevertheless,
the theoretical framework of dictator’s power sharing with elites is useful for understanding the
dynamics of decentralization. Decentralization is an institution where dictators share power with
local elites in the interests of efficient and effective governance, which arguably benefits their
political survival.
Thus, local elites refer to the elites who possess in-depth knowledge, maintain personal
connections and even bear moral legitimacy in sub-national levels to stabilize social orders. The
identities of local elites are local people with the background of (1) family leaders, (2) local
bureaucrats, and/or (3) local faction leaders. To qualify as local elites, they have to maintain a
strong bond with local people. In other words, a sufficiently long period of stay in the region is
required. The longer local elites stay in sub-national levels, the more likely they accumulate
information about local governance and build up relationship with local people.
However, it is not easy to manage local elites because they are political players unique in
the geographical nature of their governing resources. It is very difficult to either rotate or replace
local elites because their in-depth knowledge, personal connections and moral legitimacy are all
rooted in specific regions. Rotation will cancel the very resources they can contribute to
dictatorships. On the other hand, replacing local elites will result in the governance crisis in
efficiency, efficacy or even legitimacy.
Why do dictators install decentralization? Dictators share power with local elites in
exchange for their political support. Dictators’ power-sharing helps local elites accumulate their
governing resources in regions. In other words, dictators are also invested in local elites’
governing resources while local elites are invested in dictators’ political survival. Given the
benefits in governance, the management of local elites is important for dictators.

3

1.2

The Two Principal-Agent Problems
The principal-agent problem refers to the situation when dictators (principal) have a hard

time motivating local elites (agent) to act in their best interests rather than in agents’ own
interests (Miller 2005). I argue that dictators have to face two principal-agent problems if they
decentralize: defection and compliance problem. The two problems are distinct in whether local
elites chip in their governing apparatus in the opposition party or not.
The former is an internal issue of elite cohesiveness within the ruling party while the latter
stems from incompatible preferences between dictator and local elites. The management of two
problems influences dictator’s duration and quality of governance, respectively. I will introduce
the two problems as follows.

1.2.1

The Problem of Defection
Facing the problem of defection, dictator fails to monopolize local elites’ resources invested

in the political survival of authoritarian rule. If local elites’ governing resources aid opposition
party, they are considered defecting from dictators or regime front to opposition force. In order
for dictators monopolize local elites’ governing apparatus, local elites are empowered with
limited governing apparatus. Also, they are blocked out of any party outside of regime front that
makes an available and feasible outside options to them.

1.2.2

The Problem of Compliance
The problem of compliance involves with dictator’s capacity to keep local elites follow

central policies. Given that local elites do not aid the opposition party, they can still threaten the

4

quality of governance in international cooperation for authoritarian regimes. The central
(principal) delegates local (agent) governments to comply with BITs which central government
signs with foreign countries. Political decentralization facilitates the signing of bilateral
investment treaties (BITs), but it results in more violation of BITs afterwards. The signing and
violation of BITs are indicators of dictator’s capacity to control local elites for the following
reasons.

1.3

Plan of the Dissertation
In Chapter 1, I introduce the two major principal-agent problems of defection and

compliance derived from the establishment of decentralization for dictatorships. The next three
chapters will elaborate the policy consequences of the two problems.
In Chapter 2, Political Decentralization, Party System and Authoritarian Breakdown, I
discuss about the combined effect of political decentralization and party system on authoritarian
breakdown. I contend that the simultaneous installation of political decentralization and party
system determines if unsatisfied insiders (local elites) and outsiders (opposition elites) are
incentivized to overthrow authoritarian regime. The level of political decentralization is
positively related to the size of governing apparatus under local elites’ control. The existence of
party outside of regime front provides a focal point for the unsatisfied local elites and opposition
elites to take collective action to revise the regime system.
In Chapter 3 and 4, I examine the policy consequences of the problem of compliance.
Whether local governments comply with central policies matters to authoritarian credibility in
international institutions. In Chapter 3 Political Decentralization and Violation of Bilateral
Investment Treaties, I investigate if decentralization increases the number of BITs violation. I

5

argue that political decentralization electorally empowers local elites are motivated to
over-commit with foreign investors in competition for limited inflow of FDI and later fail to
follow through the terms in BITs committed by the central government with foreign countries.
In Chapters 4, Decentralization and China’s Bilateral Investment Treaties, I conduct a
quantitative study on the case of China. I explain how dictators handle the problem of defection.
In China’s BITs, I explore how as proxies to decentralization, leadership and institution
influences China’s BITs. I argue that leader’s preferences about decentralization influence the de
facto practice of decentralization in signing as more BITs as possible.
Chapter 5 concludes the major findings along with the implications. The findings of the
dissertation are expected to further the understandings in the politics of decentralization, an
important and yet understudied power-sharing institutions under dictatorships. First, I intend to
further the understandings in the politics of decentralization, an important and yet understudied
power-sharing institutions under dictatorships. Specifically, I look at how power-sharing with
local elites matter to the maintenance of authoritarian rule in terms of duration and quality.
Second, I identify the principal-agent problems of defection and compliance that drive the
politics of decentralization under the framework of power-sharing. The problem of defection
unearth the mutually exclusive nature between decentralization and party system in prolonging
political survival, especially when party has been recognized beneficial to authoritarian resilience
(Geddes 2005a; Magaloni 2008; Svolik 2012a).
Third, I take a step forward on linking the two lines of literature of authoritarian institution
and international cooperation to discover the domestic source of international cooperation for
authoritarian regimes with an ever more importance in global order. A booming literature of
authoritarian regime closely examines how institutions influence the policy performance for

6

dictatorships. Meanwhile, the research on international institutions are widely recognized as a
credibility tool to prevent default on international cooperation in global order (Axelrod and
Keohane 1985; Martin and Simmons 1998; Simmons 2000; Simmons, Dobbin, and Garrett 2008).
A case study on China’s participation in international institutions enriches the understanding
across the two literatures by showing how informal leadership can determine the change and
continuity of institutional effect on international cooperation.
To sum up the outcome and explanatory variables, data, sample and method in each chapter,
I provide Table 1-1 to outline the structure of this dissertation as follows.

Table 1-1: Plan of Dissertation
Principal-Agent
Problems

Problem of
Defection

Problem of

Chapter

2

3

4

Outcome

Authoritarian
breakdown

BITs violation

BITs conclusion

Political
Explanatory Variable decentralization
and party system

Political
decentralization

Leadership and institution
as competing proxies to
decentralization

Data

Time-Series-Cross-Sectional

Sample

Method

Dictatorship

Compliance

China

Regression analysis Regression analysis Regression analysis
(BTSCS)
(NBRM)
(BTSCS)

Source: Author

7

CHAPTER 2
POLITICAL DECENTRALIZATION, PARTY SYSTEM, AND
AUTHORITARIAN BREAKDOWN

2.1

Introduction
In October 1977, Hsin-liang Hsu was elected the county magistrate of Taoyuan city, a

second-tier municipal body under authoritarian Taiwan ruled by the Nationalist Party (also
known as Kuomintang, KMT). Hsu was soon ousted by the KMT for his “unauthorized”
candidacy in local election and later recruited by non-party democrats who initiated the first
opposition party in 1986, Democratic Progressive Party (DPP) (Wu and Chen 1993). Two
decades later, the DPP won the majority seats in magistrate elections under Hsu’s chairmanship
and support of local elites previously chained to the KMT’s clientelism. The victory in local
elections paved the way to pass the first turn-over test for Taiwan’s democratization since 1996
by defeating the KMT in 2000 presidential election.

3

Hsu is not the only local elite who bit the authoritarian hand that feed. In Mexico, the ruling
Institutional Revolutionary Party (PRI) has nurtured a generation of elites defecting to opposition
parties under the increasing electoral pressure (Greene 2007). Some of them participated in
state-level elections to compete with ruling the PRI, including Manuel Clouthier in Sinaloa and
Addy Coldwell in Quintana Roo for the National Action Party (PAN), and Andrés
Obrador, Cuauhtémoc Cárdenas, and Marcelo Ebrard in Mexico City for Party of the Democratic
Revolution (PRD). Successful or not in state elections, they’re recognized for significant
contributions to democratization in Mexico.
Local elites are important political players. Dictators face the dilemma of empowering local
3

Coined by Huntington (1991), turn-over test refers to peaceful power transition for democratic
consolidation.
8

elites and blocking their defection. On one hand, dictators install political decentralization via
local elections to enhance their legitimacy of rule by electorally empowering local elites with
governing apparatus. On the other hand, the hierarchical advancement to central positions with
greater advantages within the ruling party inevitably produces unsatisfied local elites capable of
crushing authoritarian regime.
Previous literature indicates that election and party are useful tools to manage elites via
information acquisition of their loyalty and performance (Blaydes 2008; Brownlee 2007;
Magaloni 2006, 2008), co-optation in accordance to their performance or sheer influence (Ames
1970; Boix and Svolik 2011; Bueno de Mesquita et al. 2003b; Diaz-Cayeros, Magaloni, and
Weingast 2007; Gandhi and Przeworski 2006; Geddes 2005a; Gehlbach and Simpser 2014;
Haber 2006; Landry 2008; Lazarev 2005; Lust-Okar 2005; Magaloni 2006; Wintrobe 2000;
Wright 2008; Zhong and Chen 2002), and deterrence into compliance with central policies
(Geddes 2005; Simpser 2013; Magaloni 2006; Greene 2007). However, how decentralization and
party system are interconnected is so under-specified that a synthetic look at the effect of two
institutions on elite management requires more academic attention.
This chapter investigates how political decentralization and party system incentivize
unsatisfied local elites to collide with opposition elites in overthrowing authoritarian rule when
they a dim prospect in the hierarchical career-ladder to central positions attached with greater
spoils. Specifically, I argue that it paves the way for unsatisfied insiders (local elites) within
ruling party to collude with outsiders (opposition elites) to simultaneously install high political
decentralization at executive and legislative branches and dominant-party system tolerating the
existence of party outside of regime front. The likelihood of authoritarian breakdown is highest
when high political decentralization greatly empowers local elites’ governing resources and

9

dominant-party system provides them with an outside option opposition elites have created. On
the contrary, the probability of authoritarian breakdown is lowest when low political
decentralization in legislature limits local elites’ resources and single-party system deprives them
of any outside option.
Political decentralization reveals local elites’ political preferences derived from their
electoral base characterized in economic status, ethno-lingual composition, natural endowments,
etc. Additionally, the degree of political decentralization provides local elites with governing
apparatus. When local elites are locally elected in executive and legislature bodies, they are
highly empowered. Not only do they represent people’s interests in the legislature, but also they
conduct the policymaking via executive resources including formal and informal personnel,
in-depth knowledge in local affairs, local social-economic networks and infrastructure. Local
elites are lowly empowered when they are only locally elected in legislature. Although local
elites remain influential to various policy outcomes via legislature (Gandhi 2008; Manion 1996;
Wright 2008), they are restricted to voice local people’s needs, which are ultimately up to
appointed executive’s discretion to meet their demand or not.
However, the hierarchical promotion within ruling party inevitably produces unsatisfied
local elites with a limited chance to occupy central position attached with greater privileges.
Party system incentivizes unsatisfied local elites to collude with opposition elites or not. As
dominant-party system tolerates the existence of party outside of regime front, unsatisfied and
highly empowered local elites are motivated to defect from ruling party to ally with opposition
elites and go against authoritarian regime. On the other hand, as single-party system forbids the
existence of party outside of regime front, unsatisfied and lowly empowered local elites are
trapped in the ruling party.

Accordingly, dictatorships with a dominant-party system and local

10

legislative elections are more likely to collapse as local elites have more opportunities to
challenge the regime with the opposition parties than other dictatorships without local elections.
To test this argument, I construct a dataset covering 41 party dictatorships from 1975 to
2008. The empirical evidence suggests that among various combinations of political
decentralization and party system, the combination of high decentralization and dominant party
system produces a higher likelihood of regime breakdown except for the set of high
decentralization and single party system. In addition, lower political decentralization at the
legislative branch and the existence of parties only in regime front is least likely to cause
authoritarian breakdown. These findings indicate that the dual installation of political
decentralization and party system as tools of elite management incentivize local elites to
maintain or overthrow authoritarian regimes.
The rest of this chapter is organized as follows. Next section discusses the definition of
political decentralization and different types of party system under dictatorships. In the third
section, I elaborate on a theory about the relationship among political decentralization, party
system, and authoritarian breakdown. I argue that both political decentralization and party
system are dictators’ tools of managing elites when sustaining their authoritarian rules. Then I
move to argue that different combinations of political decentralization and party system would
affect the probability of authoritarian breakdown.
Specifically, I argue that the degree of political decentralization and the existence of party
outside of regime front determines are most likely to experience authoritarian breakdown,
because unsatisfied local elites would collude with opposition elites of ruling party.
argument in empirical sections. The final section concludes.

11

I test my

2.2

Definition
Decentralization refers to a multi-tiered governing system with a shift of authority from

central governments to local governments in charge of issues confined within specific
4

geographical locales (Rodden 2004). The policy power granted to local governments generally
include fiscal, administrative and political decentralization (Falleti 2010; Green 2005; Treisman
2002). In this chapter, I focus on political decentralization, the installation of local elections
which foster a group of local elites whose capacity to mobilize the mass into their favor matters
to the consolidation or demise of authoritarian regimes (McDonough 1995; Scobell 2014; Tarrow
1995; Welzel 2006).

5

One may argue that authoritarian elections are mere window-dressing with limited
information about local elites’ capacity to mobilize the mass to bargain with central government.
But, some researchers argue otherwise. Elections not only potentially render ruling legitimacy at
home (Heberer 2006; Mozaffar 2002; Schatz 2006; Schedler 2002; Waterbury 1999), but also
abroad (Waterbury 1999). Even electoral fraud itself is a device to force bureaucracy into to
demonstrate their loyalty and capacity (Gehlbach and Simpser 2014). Therefore, electoral results
signal local elites’ capacity to mobilize and/or coerce the mass into their favor.
The significance of political decentralization rests in the unique nature of dictatorship. The
definition and settlement of disputes with regard to the distribution of power between central and

4

Political decentralization and federalism are used in previous literature. Recently researchers
gradually replace federalism with political decentralization to broadly include the countries
without de jure entitlement of federalism, but with de facto practice of independent
decision-making power for local governments such as Spain.
5
As Treisman (2007) points out, political decentralization is of different meanings for
researchers. However, the selection of local leaders with decision making power via local
election is widely considered a more decentralized form in a political system.
12

local governments is of great uncertainty. While democracies can always resort to a higher
authority, constitutional (or judicial) review to peacefully clarify and resolve the disputes
between central and local governments, dictatorships lacking such authority must eventually turn
to sheer violence to thwart the worst case scenario of being overthrown as all other means are
exhausted (Svolik 2012b).
Party system under dictatorship is described as one-party system, including single-party and
dominant-party systems. For both systems, the ruling party always rules as long as dictatorship
lives (Magaloni and Kricheli 2010). However, the two systems differ in the tolerance of
opposition party in election (Magaloni 2010; Sartori 1976a). As Magaloni and Kricheli (2010)
have summarized, the ruling party under single-party system prohibits the electoral participation
of party outside of regime front while the dominant-party dictatorships do not.

6

Therefore, I

adopt the distinction between single-party and dominant-party systems in this chapter to discuss
how party system structures elected local elites’ loyalty under dictatorship.

2.3

Political Decentralization, Party System, and Elite Management
Dictators count on violence to be their last resort of maintaining authoritarian rule (Svolik

2012b; Wintrobe 2000), but they have to share power with elites who may bite dictator’s feeding
hand. Thus, elite management has been dictators’ great concerns in the literature of comparative
authoritarianism (Svolik 2012b). In particular, existing studies have pointed out that dictators
could establish institutions to manage elites and achieve authoritarian survival. In other words,
the surviving dictatorships since the World War II have aroused intensive discussions on the
6

Dominant party system is also termed as hegemonic party (Sartori 1976b), electoral
authoritarian regime (Diamond 2002; Linz 2000; Schedler 2002), or competitive authoritarian
regime (Way and Levitsky 2002).
13

source of authoritarian resilience, especially after more and more of them adopt façade
democratic institutions such as election and party after the Cold War in the 1990s (Geddes
2005a; Magaloni 2006; Svolik 2012a).
For the last decade, approximately 70% of dictatorships hold elections and tolerate more
than one party to take part in the electoral market.

7

Scholars of comparative authoritarianism

suggest that dictators utilize election and party to manage elites to prevent authoritarian
breakdown via information acquisition, and cooptation and deterrence of elites (Gandhi 2008;
Geddes 2005; Brownlee 2007; Landry 2008; Magaloni 2006; 2008; Lust-Okar 2008; Gandhi and
Lust-Okar 2009; Gehlbach and Simpser 2014; Simpser 2013; Svolik 2012a).
First, election and party provide information about elite’s types. Dictators use electoral
results to differentiate loyal and competent party members from others in mobilizing voters into
electoral support (Blaydes 2008), detect popular grief by opposition party’s vote share (Brownlee
2007; Magaloni 2006), and split doves from hawks in the opposition force by tolerance of
electoral participation (Magaloni 2008).
Second, co-opting elites is feasible through elections and party. Dictators use elections to
co-opt party members (Magaloni 2006), elites from various social fabrics (Gandhi 2008; Gandhi
and Przeworski 2006; Wright 2008) and elites in general (Boix and Svolik 2013). Party is an
engine for clientelism, a career ladder for long-term commitment, a platform for bargaining, and
a credibility tool to power-sharing commitments.
First, party is an engine for clientelism. Ruling party provides elites with eco-social
privileges for party members to keep them invested in authoritarian regime (Bueno de Mesquita
et al. 2003a; Geddes 2005a; Wintrobe 2000). With the selective co-optation of opposition elites,
7

Calculated from Svolik (2012a).
14

dictators can bump up the difficulty of coordination among opposition elites (Diaz-Cayeros,
Magaloni, and Weingast 2007; Lust-Okar 2005; Magaloni 2006, 2008).
Second, the career ladder within a party ensures long-term commitment for party members
interested in exclusive privileges. Landry (2008) argues that merit-based hierarchical promotion
system within the ruling party keeps local elites loyal and motivated for good governance. Svolik
(2012a) emphasizes that party members are hostages for the sunk costs and the future benefits
they have invested over the course of seniority system so that their loyalty is fixed across time.
Third, party is a platform to contain power-bargaining. Dictators strategically adopt
multi-party system to divide and rule opposition forces (Blaydes 2011; Bueno de Mesquita et al.
2003a; Smith 2005). For instance, Blaydes (2011) argues that the ruling party in Egypt make
Muslim Brotherhood a major opposition power so that opposition forces fail to unite to cause
authoritarian breakdown. Also, party broadens the social appeals by collecting diversified ideas,
bargaining with opposition elites and addressing social demands (Gandhi and Przeworski 2006).
Fourth, party is a credibility tool to solve the commitment problem when dictators use party
to institutionalize a time-consistent promise of sharing power with potential opposition elites
(Brownlee 2007; Haber 2006; Lazarev 2005; Magaloni 2008) and elites in general (Boix and
Svolik 2011; Svolik 2012b). Boix and Svolik (2013) contend that the existence of party
guarantees mutual retaliation between dictator and elites once either party violates commitment
of power-sharing.
Lastly, some studies take a synthetic view at election and party in managing power elites in
general. Electoral results help ruling party to distribute spoils in accordance to party member’s
performance on vote collection and thus facilitate clientelism trapping voters reliant on state
transfers (Blaydes 2008, 2011; Greene 2007; Lust-Okar 2008; Pepinsky 2007; Wintrobe 2000;

15

Zhong and Chen 2002) or to adjust allocation of resources according to the revealed weak and
strong spots for ruling and opposition parties (Ames 1970; Brownlee 2007; Magaloni 2006).
Moreover, election and party deter rebellion. To prolong authoritarian rule, a landslide but
manipulated victory in election can deter coup d’état and opposition elites by showing how well
the government grips on power (Geddes 2005a; Simpser 2013), intimidate opposition elites by
how much resources ruling party can mobilize (Geddes 2005a; Greene 2007; Magaloni 2006) or
undermine the support base for opposition parties by adjusting policies with a more clear
direction to pour more resources (Magaloni 2006).
Nevertheless, the literature of party and elections on elite management under dictatorships
is clouded with the following issues. First, the level of election begs a deeper investigation. For
one thing, 46.8% of dictatorships install political decentralization by holding local elections.

8

When researchers take the existence of national election as a metric of institutionalization for
dictatorships (Blaydes 2011; Gandhi and Lust-Okar 2009b; Geddes 2005b; Pepinsky 2007), they
underestimate some of the most resilient dictatorships in the sample of analysis. For instance,
Qaddafi’s Lybia, East Germany, and Vietnam do not hold elections in the national level, but in
the sub-national level. Also, local election generates local elites unique in geographical nature for
dictators to accommodate with. Existing research indicates that local election can become a
greenhouse for political opposition because it builds up alternative patronage systems and
regional power base (Treisman 2007; Ochoa-Reza 2004).
Second, the type of governing resources granted to elected politicians is under-specified.
Control of executive and/or legislative branches decides how much elected local elites could chip

8

Calculated from Keefer et al. (2012) as dictatorships are defined by Geddes, Wright and
Frantz (2014).
16

in revolution. When local elites are elected in executive and legislature bodies, they are highly
empowered to represent people’s interests in the legislature and conduct the policymaking via
executive resources. Local elites are lowly empowered in elected legislature influential to
various policy outcomes (Gandhi 2008; Manion 1996; Wright 2008), but they are subject to
appointed executive’s discretion.
Third, the articulated effects of ruling party on elite management are problematic. Given
that central positions in the hierarchical career ladder are loaded with greater spoils (Landry 2008;
Svolik 2012a), it is not clear why the majority of local elites will remain loyal to the ruling party
when their future gains can never surpass sunk costs at their dead end of career. Moreover, the
fairness of merit-based promotion constantly calls into question due to low transparency in
authoritarian regime. In sum, the lack of procedural justice and transparency for high-end
leadership will drive out some insiders of ruling party in search for better opportunities.
Fourth, interactions between insiders and outsiders of authoritarian regime front are missing.
Some scholars focus on how election and party consolidate the loyalty of supportive insiders of
regime front (Greene 2007; Lust-Okar 2008; Pepinsky 2007; Wintrobe 2000), while others
emphasize how the two institutions keep opposition outsiders away from the regime front in
check (Boix and Svolik 2013; Gandhi and Przeworski 2006; Landry 2008; Smith 2005; Svolik
2012a).
However, the interactions and fluidity between the two kinds of elites constantly challenge
dictators’ task of elite management. As discussed in the beginning of this chapter, some
prominent opposition figures in Taiwan and Mexico were former party members of ruling parties,
and it is widely recognized that they make significant contributions to democratic transitions in
both countries.

17

2.4

Effects of Political Decentralization and Party System on Authoritarian Breakdown
In this chapter, I argue that the combination of political decentralization and party system

incentivizes unsatisfied insider (local elites) to collude with outsider (opposition elites) of ruling
party in regime revision under dictatorship. The degree of political decentralization determines
how much governing resources local elites take control. The openness of party system decides if
there exists a feasible and available outside option for local elites once they are not satisfied with
their career development within ruling party. Moreover, political decentralization enables
unsatisfied local elites to find their peers sharing career grievances because they reveal their
political preferences derived from their electoral base with economic status, ethno-lingual
composition and natural resources, etc. Together, the resources and options boost the chance in
the equation of regime revision for authoritarian regimes. The logic is elaborated in details as
follows.
Under political decentralization, local elites are highly empowered in elections when local
executive and legislature are both elected. They can not only update and voice people’s
preferences in legislature, but also steer the making and implementation of policy accordingly in
executive. Local elites are lowly empowered in election when only legislature is locally elected.
The elected local legislature can only collect and present people’s preferences to the centrally
appointed executive heads who retain the discretion to incorporate people’s preferences in the
policies or not.
Given that spoils trickle down from central to local positions, the hierarchy of career
development within party inevitably produces unsatisfied local elites with dim prospect to move
up. Their limited chance to occupy positions in the central government excludes them from

18

greater privileges. The majority of non-promoted local elites thus become a potential pool of
unsatisfied elites in search for outside options when they cannot balance the investment and
gains in ruling party. Equipped with electoral resources, they are capable to turn against ruling
party and impose a threat on authoritarian breakdown.
One may challenge that not all local elites seek central posts. They may care more about the
certainty of spoils rather than the size of them. Local elites may arguably be content with the
highest possible position they can get. But, once local elites stop maximizing their power, they
risk being replaced as their capacity to impose a threat on authoritarian rule is limited (Bueno de
Mesquita et al. 2003a; Svolik 2012b).
Party system incentivizes unsatisfied local elites to collude with opposition elites or not. For
local elites, defecting to party outside of regime front is a viable option conditional on the
existence of it in the first place. On one hand, single-party system forbids the existence of party
outside of regime front, lowly empowered local elites are trapped in the ruling party whether
they are satisfied or not. Organizing a party outside of regime front from scratch is too costly to
restrain unsatisfied local elites from defecting ruling party.
On the other hand, dominant-party system tolerates the existence of party outside of regime
front. When highly empowered local elites are not satisfied with their promotion within ruling
party, they have incentives to defect from ruling party to ally with opposition elites to go against
authoritarian regime once their prospect of promotion within the party is little. Their
participation in the opposition party can expect to leverage more political interests if ruling party
is replaced by the opposition party or the authoritarian regime is democratized. In short, a party
outside of authoritarian regime front provides a focal point for unsatisfied local elites to
cooperate with opposition elites.

19

For opposition elites, political decentralization reveals information about local elites’
political preferences and resources, which help classify local elites into potential friends or foes.
Moreover, they are incentivized to recruit or even solicit the defected local elites from dominant
parties because their expertise and resources are valuable assets for the opposition party to grow.
To precipitate authoritarian breakdown, a party outside of regime front makes an alternative
target of political investment for unsatisfied local elites and opposition ones to lower the costs of
collective action.
To summarize, the degree of political decentralization can range from low (locally elected
legislature) to high (locally elected executive and legislature). The types of party system covers
single party (existence of parties only in regime front) and dominant party (existence of party
outside of regime front). As Table 2.1 demonstrates, the two variables, political decentralization
and party system, constitute four sets, including (Low, Single), (Low, Dominant), (High, Single),
and (High, Dominant) denoting the combination of political decentralization and party system
respectively. I will discuss each of these four categories in details in the following paragraphs.
In Table 2.1, the set of (Low, Single) represents the combination of low political
decentralization and single-party system. This set is expected to be most resistant to authoritarian
breakdown. Local elites are only elected in legislature. In addition, the single-party system
forbids the existence of any party outside of regime front. Authoritarian rule is prolonged
because local elites are electorally empowered to a limited degree and they are deprived of any
feasible outside option. They are trapped in the ruling party so that authoritarian regime
demonstrates resilience to democratization.

20

Table 2-1: Political Decentralization and Party System under Dictatorship
Political Decentralization

Party System
Single-Party System
Dominant-Party System

Low
(Low, Single)

High
(High, Single)

(Low, Dominant)

(High, Dominant)

Note: The combination of (political decentralization, party system) in parenthesis.

Malaysia makes a good case to illustrate the logic of authoritarian resilience under low
political decentralization and single party system. Despite that state legislatures are elected in
thirteen Malaysian states, none of local executive heads are elected. Nine states are governed by
hereditary leaders who are entitled as Sultans (Johor, Kedah, Kelantan, Pahang, Perak, Selangor,
and Terengganu), Yamtuan Besar (Negeri) or Raja (Perlis). Four states are governed by centrally
appointed governors in Penang, Malacca, Sabah, and Sarawak (Shair-Rosenfield, Marks, and
Hooghe 2014). In terms of party system, Barisan National (BN) dictates Malaysian politics in
1973 and remains a ruling party alliance including three major parties, United Malays National
Organization (UMNO), Malaysian Chinese Association (MCA), and Malaysia India Congress
9

and ten satellite parties in regime front. The authoritarian rule in Malaysia has been challenged
as the MCA dropped out of cabinet due to race riot against Chinese Malaysians. As the Chinese
Malaysians have been under-represented and under-promoted in executive positions across all
levels of governments, unsatisfied Chinese leaders could have taken advantage of the race riot

9

The website of component party in Barisan Nasional show that the satellite parties are United
Traditional Bumiputera Party, Sarawak United People's Party, Malaysian People's Movement
Party, People's Progressive Party, Liberal Democratic Party, United Sabah People's Party, United
Pasokmomogun Kadazandusun Murut Organization, United Sabah Party, Sarawak Progressive
Democratic Party, and Sarawak People's Party. See
http://barisannasional.org.my/en/component-party (Access on February 5, 2015).
21

and imposed a threat on authoritarian breakdown (Freedman 2000). But the MCA quickly
returned to the ruling alliance and sustained authoritarian rule. Besides the minority status, the
unsatisfied Chinese in Malaysia lack executive resources and a feasible option outside of regime
front eventually retards its action against the government. The first hypothesis is thus derived as
follows:
H1: Compared with other sets of political decentralization and party system, low political
decentralization in the legislative level and single-party system tolerating only parties
within regime front denoted as (Low, Single) generates the lowest likelihood of
authoritarian breakdown.

On the other extreme, the set of (High, Dominant) represents the combination of high
political decentralization and dominant party system. It is expected to be the most vulnerable to
authoritarian breakdown. Local elites are not only elected in legislature, but also in executive
branch. Also, the dominant-party system tolerates the existence of party outside of regime front.
The demise of authoritarian Philippines makes an example for this category. Philippines
have suffered factionalism in local politics (Machado 1971). The party system in Philippines is
considered fragmented, personalized and under-institutionalized (Montinola 1999; Ufen 2008).
Under Ferdinand Marcos’ rule, the ruling party, New Society Movement (Kilusan Bagong
Lipunan, KBL), was confronted with challenges on the integrity of governance and elections.
One of the major opposition party was the People’s Power Movement-fight (Lakas Ng
Bayan-Laban) led by Benigno S. Aquino, Jr., originally a mayor of Concepcion City in the
province of Tarlac. After his assassination in 1983, the United Nationalist Democratic
Organization (UNIDO) was formed to unite prominent figures including local leaders like Cesar

22

Climaco (mayor of Zamboanga city), Evelio Javier (governor of Antique), and Joe Lingad
(governor of Pampanga). In 1986, Corazon Aquino, the widow of Benigno S. Aquino, Jr.,
assumed the presidency when the exile of Marcos brought democratization for Philippines. I
come up with the second hypothesis as follows.
H2: Compared with other sets of political decentralization and party system, high political
decentralization in the legislative and executive level and dominant-party system tolerating
the existence of party outside of regime front denoted as (High, Dominant) produces the
highest likelihood of authoritarian breakdown.

The probabilities of authoritarian breakdown for (Low, Dominant) and (High, Single) are
somewhere between those for (Low, Single) and (High, Dominant). (Low, Dominant) represents
the combination of low political decentralization and dominant party system. Local elites are
only elected in legislature. In addition, the single-party system forbids the existence of any party
outside of regime front.
The fate of authoritarian Egypt illustrate the case when authoritarian resistance to
democratization seems to persist, but eventually ends up in regime breakdown in 2011. Since
1960, governors are appointed for twenty-six regions and legislature is elected in Egypt
(Calingaert 2006). Although Egypt maintained a multi-party system with over thirty registered
parties across ideological spectrum, the National Democratic Party has ruled Egypt since 1978
(Albrecht 2005; Kienle 1998). Although unsatisfied local elites are electorally empowered to
local legislature, the lack of executive resources greatly shrinks the political capital local elites
can chip in collective action against authoritarian rule. Nevertheless, opposition party can
potentially take advantage of accumulated frustration to overthrow authoritarian regime with the

23

assistance of local elites because opposition parties in Egypt exist outside of ruling party’s
control.
On the other hand, (High, Single) represents the combination of high political
decentralization and single-party system. Local elites are elected in legislature and executive.
The ruling party may set up a few satellite parties within regime front. Any party outside of
ruling party’s control is strictly banned.
China has been widely discussed for its prospect of democratization or authoritarian
consolidation with regard to political decentralization and party system (Blanchard and Shleifer
2000; Landry 2008). There are nine parties registered in China.

10

All of them are under tight

control by the regime front, referring to the ruling party, the Chinese Communist Party (CCP) in
China. The provincial leaders are elected by the local legislature, which are directly elected
under the supervision of CCP. So far, All of them belong to the regime front. Unsatisfied local
elites are electorally empowered to local legislature and executive, but any feasible outside
option is not permitted under the CCP’s authoritarian rule. Before taking a post in the politburo
standing committee, the highest power organ in CCP, power elites must serve as provincial
leaders. However, powerful local elites are of potential to harm political stability in China,
especially when members of the politburo standing committee are involved.

11

Thus, a certain

level of risk of local elites involved in the scene of causing authoritarian breakdown does exist.

10

The nine registered parties are the Revolutionary Committee of the Chinese Kuomintang,
Chinese Democratic League, National Democratic Construction Association, China Association
for Promoting Democracy, Democratic Party for Chinese Peasant and Labor, and Jiu San Society.
See http://cpc.people.com.cn/BIG5/64162/122148/7430696.html (Access on March 5, 2015).
11
Allegedly, Bo Xilai, the former leader in Liaoning province and party chief in Chongqing, has
been involved with a conspiracy of coup along with Zhou Yongkang, a member of politburo
against the current general secretary, Xi Jingping.
http://www.bbc.com/news/world-asia-china-17673505 (Access on March 5, 2015).
24

2.5

Research Design
To test the effect of political decentralization and party system on the authoritarian

breakdown, I compile a time-series-cross-sectional (TSCS) dataset covering 41 party
dictatorships from 1975 to 2007 on the basis of several datasets. The first is Barbara Geddes,
Joseph Wright and Erica Frantz’s dataset (hereafter GWF) on authoritarian regimes. The second
dataset is Philip Keefer and the World Bank’s Development Research Group’s Database of
Political Institutions (hereafter DPI) for the existence of local elections defining the degree of
political decentralization. The third is Jose Antonio Cheibub, Jennifer Gandhi and James
Raymond Vreeland’s dataset (hereafter CGV) on party system under dictatorship.
The research interests of this chapter focus on the dual installation of political
decentralization and party system, so the sample is limited to dictatorships with local election
and party system. The former is defined by the DPI dataset while the latter is defined in the CGV
dataset. A list of dictatorships in my sample is in the Table A-1 in the Appendix.
The dependent variable in my empirical analysis is authoritarian breakdown. It is measured
by regime failure, a binary variable provided by GWF dataset. If an authoritarian regime should
fail in a given year, it is recorded as 1 and otherwise 0. According to GWF’s codebook, an
autocratic regime is considered as having failed if the following occurs: (1) “a competitive
election for the executive, or for the body that chooses the executive, occurs and is won by a
person other than the incumbent or someone allied with the incumbent; and the individual or
party elected is allowed to take office”, (2) “the government is ousted by a coup, popular
uprising, rebellion, civil war, invasion, or other violent means, and replaced by a different
regime”, or (3) “the ruling group markedly changes the rules for choosing leaders and policies
such that the identity of the group from which leaders can be chosen or the group that can choose

25

major policies changes” (Geddes, Wright, and Frantz 2014, 7–8).
Since the TSCS data structure can cause issues in temporal dependence and
heteroskedasticity, and since the dependent variable in my analysis is binary, I use binary
time-series-cross-sectional (BTSCS) models to conduct statistical analysis (Beck, Katz, and
Tucker 1998). According to Beck and coauthors, BTSCS models use time splines to detect the
temporal dependence for repeated observations on the same country. For the research interest of
this paper, the (non)occurrence of regime breakdown in the past can take place repeatedly in the
future for each country. In other words, history matters so that the observations are not
independent. Therefore, I follow the suggestion of Cartet and Signorino (2010) and include three
time polynomials for regime duration in the BTSCS models.
The other issue derived from TSCS data structure is the heteroskedasticity of standard errors
in cross-sectional comparison. In this chapter, each country experiences regime breakdown for
various reasons derived from its unique features in history, geography, demography, etc. These
factors are beyond the model’s reach so that I use clustered standard errors to deal with
heteroskedasticity in cross-sectional feature in BTSCS models.

2.5.1

Independent Variables
The first independent variable is the degree of political decentralization. It is measured by

the existence of local election ranging from 0 to 2. Sponsored by World Bank, the DPI codes an
ordered variable, STATE indicative of whether state/province governments are locally elected.
Note that the DPI does not code cases in which the executive is locally elected, but the local
legislature is appointed.
To accommodate with research interests, the sample only includes dictatorships with the

26

installation of political decentralization. Thus, I drop the observation with a value of 0 for STATE
means that neither local executive nor local legislature is locally elected. I keep the value of 1
denoting if the executive is appointed, but the legislature elected, and 2 denotes if they are both
locally elected (Beck et al. 2001, 20).
The variable, STATE, is chosen as an indicator of political decentralization for two reasons.
First, the highest level is defined as the “state/province” level across all levels of sub-national
governments. At this level, the total number of local elites for a dictatorship is limited so that
unsatisfied local elites’ costs of collective action are minimized for collectively opting out the
ruling party (Olson 1965).
Moreover, local election is coded when “indirectly elected state/province governments,
where directly elected municipal bodies elect the state/province level, are not considered locally
elected. Indirectly elected state/province governments elected by directly elected state/province
bodies are considered locally elected” (Keefer 2012, 20).
The second independent variable is the openness of party system. The CGV’s dataset
contains an ordered variable of defacto2 presenting the existence of parties outside of regime
front ranging from 0 to 2. Dictatorships are coded as 0 for the non-existence of party, 1 for “one
party or multiple parties, but they belong to regime front” and 2 “multiple parties” (Cheibub,
Gandhi, and Vreeland 2009, 5). Because the sample is limited to party dictatorship, I exclude
dictatorships without party from my sample. The party system is denoted as single when the
value of defacto2 is 1 and as dominant when the value is 2.
Since the variables for political decentralization and party system are both dichotomous and
qualitatively different, I generate four dummy variables, (Low, Single), (High, Single), (Low,
Dominant) and (High, Dominant) to exhaust the four possible combinations of political

27

decentralization and party system with distinct effects on authoritarian breakdown.
To compare the dual installation of political decentralization and party system, I compute
four dummy variables for four combinations. To denote the combinations in each parenthesis, the
former denotes the degree of political decentralization and the latter refers to the type of party
system in each parenthesis.

2.5.2 Control Variables
I include a battery of controls covering domestic and international factors indicated to
influence authoritarian breakdown in BTSCS models. The domestic factors include history of
past regime transition to dictatorship, endowment of natural resources, military expenditure, and
economic performance while the international factor models contain democratic diffusion in the
region and foreign direct investment. The major independent variable and control variables are
lagged by one year to mitigate reverse causality. I will introduce the variables in the two major
sets of models in the following paragraphs.
For domestic factor models, first of all, a record of regime breakdown in the past will result
in more regime breakdown in the future. The consolidation of either regime system is so difficult
that the probability of authoritarian breakdown is expected to be high. Thus, the history of
frequent regime transitions to dictatorship is controlled by incorporating stra, a count variable
from the CGV’s dataset.
Second, it takes resources to maintain a winning coalition in the interest of minimizing the
tendency of authoritarian breakdown, as Bueno de Mesquita et al. (2003a) suggest. Income from
natural resources is one of the most important types of income to maintain authoritarian rule.
Scholars debate the effect of natural resources on the longevity of authoritarian regime (Haber

28

and Menaldo 2011; Ross 1999; Singer 1988). Despite the mixed effect, I control this by using
Haber and Menaldo's variable coding the ratio of revenue from natural resources covering oil,
gas and metals over population. I also take log of it for smoothing great variation resulted from
extreme values.
Third, higher military spending represents intensive investment in state violence to suppress
any upheaval potentially leading to authoritarian breakdown for dictatorship. Previous research
finds that the maintenance of physical capacity is crucial to prevent authoritarian breakdown
(Albertus and Menaldo 2012). Heavy investment in state force is negatively associated with the
probability of authoritarian breakdown. Little investment works the opposite way. In the model, I
include military spending provided by the Correlates of War Project version 4.0 (Singer 1988).
To eliminate the great variation in dictatorships, I take the log of the variable plus one for each
dictatorship.
Fourth, a good economy arguably influences the likelihood of authoritarian breakdown
under dictatorships in both directions. Modernization theory also predicts that economic
development will breed middle class politically inclining for democracy (Lipset 1959). Some
argue irrelevance between the two variables (Przeworski 2000). To curtail the great variation, I
log the real GDP per capita, taken from the World Development Index, World Bank. I also
include a squared term of GDP per capita to control for potential non-linear effects of this
variable on regime breakdown.
For international factor models, the international environment is believed to command a
domino effect in authoritarian breakdown (Leeson and Dean 2009). The first variable deals with
the neighborhood effect of democratic diffusion. Being surrounded by more democracies will
raise the probability of democratization. The three waves of democratization are good examples

29

embodied the relationship of democratization and geographical proximity (Huntington 1991). I
incorporate the measure of percentage of democracies in region calculated by Haber and
Menaldo (2011) in the model.
The other international control is the volume of foreign direct investment, which is one of
the most important external revenues for dictatorships. The effect of foreign direct investment on
authoritarian breakdown is expected to be negative. With more inflow of foreign direct
investment per capita, the more likely dictatorship is safe from being overthrown (N. Jensen
2008). The World Development Index developed by World Bank contains a variable recording
the sum of foreign direct investment in current USD. I include the variable in the log form in the
model. The summary statistics is presented in Table A-2 in the Appendix.

2.6

Results
To test two hypotheses previously mentioned, I estimate the effect of political

decentralization and party system on authoritarian breakdown. Each hypothesis is tested in three
models containing domestic controls (including regime transition, natural resources, military
spending, and GDP per capita and squared GDP per capita), international controls (including
regional democracies and foreign direct investment) and full controls.
I first estimate the effect of political decentralization and party system on authoritarian
breakdown with the baseline of (High, Dominant) denoting the set of high political
decentralization in the legislative and executive level and dominant party system tolerating the
existence of party outside of regime front. According to H1, I expect that in comparison with
other sets of political decentralization and party system, (High, Dominant) produces the highest
likelihood of authoritarian breakdown. Next, I switch the baseline to the dual installation of low

30

political decentralization in the legislative level and single-party system tolerating only one party.
According to H2, (Low, Single) generates the lowest likelihood of authoritarian breakdown than
other sets of political decentralization and party system.
Table 2.2 reports the estimation results of BTSCS models with the battery of controls in
domestic and international factors. While Model 1 contains domestic controls, Model 2 covers
international controls. Model 3 is comprised of all domestic and international factors as controls.
From Model 1 to Model 3, (Low, Single) consistently shows a negative relationship with the
probability of authoritarian breakdown. The results are also significant at the p<0.1 level or
better. Meanwhile, (Low, Single) presents negative coefficients with significance at the p<0.1
level in Model 2 and Model 3, but not in Model 1. Although (High, Single) is not significantly
less likely to cause authoritarian breakdown than the baseline category, it shows negative signs in
line with the hypothesized causal direction.
To test H2, I switch the baseline to (Low, Single), which is expected to be least likely to
result in authoritarian breakdown. With (Low, Single) as the baseline category, Model 4 suggests
that (High, Single), (Low, Dominant) and (High, Dominant) are more likely to cause
authoritarian breakdown at the p<0.1 level with a full set of controls.
As for the control variables, the results for regime transitions and foreign direct investment
reveal expected sign with significance. However, military spending, natural resources, GDP per
capita, squared GDP per capita and regional democracies are not significant although the
expected direction fall square on expectation.
Meanwhile, the control of regime transition is positively correlated with the likelihood of
authoritarian breakdown at the p<0.5 level or better. A history of constantly switching back and
forth between democracy and dictatorship makes a dictatorship vulnerable to authoritarian

31

breakdown. Failure to consolidate either regime system raises the prospect for authoritarian
breakdown. The other control that shows significant outcomes across models is the volume of
foreign direct investment. The coefficient is negatively related to the authoritarian breakdown at
the p<0.5 level as previous literature concludes (N. Jensen 2008). The effect of foreign direct
investment prevents the collapse of authoritarian regime by benefiting its economy.

Table 2-2: Effects of Political Decentralization and Party System on Authoritarian Breakdown

(Low, Single)
(High, Single)

Model 1

Model 2

Model 3

-1.411*
[0.768]
-0.253
[0.753]
-0.818
[0.550]

-2.669***
[0.857]
-0.488
[1.065]
-0.905*
[0.494]

-2.921***
[1.077]
-0.207
[1.087]
-1.143*
[0.591]

Model 4

2.714**
[1.184]
(Low, Dominant)
1.778**
[0.865]
(High, Dominant)
2.921***
[1.077]
Regime transitions
0.865***
0.805**
0.805**
[0.263]
[0.396]
[0.396]
Military spending
-0.196
-0.211
-0.211
[0.183]
[0.169]
[0.169]
Natural resources
-0.036
0.143
0.143
[0.085]
[0.115]
[0.115]
GDP per capita
-0.008
-0.137
-0.137
[0.330]
[0.380]
[0.380]
GDP per capita (squared)
0.011
0.029
0.029
[0.033]
[0.033]
[0.033]
Democracies in the region
0.009
0.010
0.010
[0.017]
[0.013]
[0.013]
Foreign direct investment
-0.200**
-0.223**
-0.223**
[0.078]
[0.103]
[0.103]
N
41
38
38
38
No. of observations
665
530
518
518
Wald Test (Prob > chi2)
0.000
0.027
0.006
0.006
Log pseudolikelihood
-101.317
-77.778
-71.318
-71.318
Note: Clustered standard errors in brackets. Three time polynomials and constant not shown.
* p<0.1, ** p<0.05, *** p<0.01.

32

Based on Model 3, I simulate the predicted probabilities of effect of political
decentralization and party system on authoritarian breakdown with King, Tomz and Wittenberg’s
Clarify package (King, Tomz, and Wittenberg 2000). The baseline is set at (High, Dominant)
with all other variables fixed at their means. According to H1, (High, Dominant) is expected to
be most likely to lead to authoritarian breakdown, compared with other combinations of political
decentralization and party system.
The simulation results are presented in Figure 2-1. The predicted probabilities of (Low,
Single) and (Low, Dominant) on authoritarian breakdown are negative. The results show that the
effects of two sets on authoritarian breakdown are significantly negative in comparison with the
baseline of (High, Dominant).
The predicted probability of (High, Single) on authoritarian breakdown is weakly supportive
of H1. The results are negative as expected. But the 90 percent confidence intervals come across
the value zero. Thus, I cannot rule out the possibility of indifference in the effect on authoritarian
breakdown between (High, Single) and (High, Dominant). Nevertheless, the point estimate of
simulated effects for (High, Single) on authoritarian breakdown is negative in comparison with
the baseline of (High, Dominant).
In sum, the empirical evidence partially supports the argument that the dual installation of
high political decentralization and dominant party system makes dictatorships vulnerable to
regime breakdown. (High, Dominant) is more politically resilient than (Low, Single) and (Low,
Dominant).

Given high political decentralization, single party system does not necessarily

outperform dominant party system in terms of sustaining authoritarian rule. Thus, H1 is
statistically confirmed for (Low, Single) and (Low, Dominant), but not for (High, Single) in my
sample.

33

Figure 2-1: Effect of Political Decentralization and Party System on Authoritarian Breakdown:
(High, Dominant)
90% CI

0
-1

-0.207

-3

-2

-1.143

-2.921

-4

Pr(Authoritarian Breakdown)

1

Point Estimate

(Low, Single)
(High, Single)
(Low, Dominant)
The Sets of Political Decentralization and Party System

Note: The graph is drawn across 3 sets of (political decentralization, party system)

Based on the estimates of Model 4 with (Low, Single) as the baseline, I simulate the
predicted probability of regime breakdown across other three sets of political decentralization
and party system under dictatorship. As H2 proposes, the probability of authoritarian breakdown
is lowest for (Low, Single) among all sets of political decentralization and party system.
Figure 2-2 illustrates th2e simulated predicted probabilities of authoritarian breakdown with
the estimates of Model 4. As illustrated in Figure 2-2, (High, Single), (Low, Dominant) and
(High, Dominant) are more likely to result in authoritarian breakdown than (Low, Single). In
other words, (Low, Single) is mostly resistant to authoritarian breakdown as local elites are
limited in their resources and blocked out any alternative choice of defection.

34

Figure 2-2: Effect of Political Decentralization and Party System on Authoritarian Breakdown:
(Low, Single)
90% CI

2.921

2.714

-2

-1

0

1

1.778

-4

-3

Pr(Authoritarian Breakdown)

Point Estimate

(High, Dominant)
(Low, Dominant)
(High, Single)
The Sets of Political Decentralization and Party System

Note: The graph is drawn across 3 sets of (political decentralization, party system)

To conclude, the empirical evidence partially supports H1, but strongly confirms H2.
Specifically, (Low, Single) and (Low, Dominant) support H1, but (High, Single) does not. Given
high political decentralization, single party system does not necessarily outperform dominant
party system in terms of sustaining authoritarian rule. Meanwhile, (Low, Single) is least likely to
result in authoritarian breakdown, as suggested by H2. Thus, closing the door of outside option
with lowly empowered local elites best sustain authoritarian rule.

35

2.7

Conclusion
This chapter intends to take a synthetic view analyzing how different formula with regard to

political decentralization and party system influence authoritarian breakdown. No one can
govern alone, and dictators are no exceptions. To share power with elites who govern along,
dictators face the dilemma of empowering local elites and managing their loyalty via election
and party, which are believed to provide information, reward and punish elites. As a maximizer
of political interests, local elites become unsatisfied when they have little hope to climb higher in
the hierarchical career-ladder to central positions linked with greater advantages. The degree of
political decentralization determines how much resources unsatisfied local elites can offer for
overthrowing authoritarian rule. On the other hand, the type of party system decides how much
costs opposition elites can lower in the equation of regime revision. By just electorally
empowering local elites to voice local people’s needs under executive discretion and blocking
any alternative party outside of regime control makes a good tool to sustain authoritarian rule. As
unsatisfied local elites are limited in their organizational power to mobilize the mass and a
feasible outside choice to turn to, their loyalty is monopolized whether they are satisfied or not.
On the other hand, the empirical evidence partially supports the argument that the
simultaneous installation of high political decentralization and dominant party system increases
the probability for regime demise. Fixed at high degree of political decentralization, dominant
party system does not significantly results in more authoritarian breakdown. When unsatisfied
local elites are electorally empowered to voice and execute independent policies at people’s will
without discretion, the lack of feasible outside party may not necessarily stop them from
overthrowing dictatorships to bid for greater political gains.
Based on the statistical analysis of 41 dictatorships from 1975 to 2007, empirical evidence

36

supports the argument that the dual installation of high political decentralization and dominant
party system is detrimental to authoritarian regimes. In addition, the estimation results also
indicate that the dual installation of low political decentralization and single party system is
mostly resistant to authoritarian breakdown.
The findings imply that over power-sharing with supporting elites and tolerance for
opposition party can be lethal for dictators. The simultaneous installation of dictator’s over
power-sharing with insiders (local elites) of ruling party and their tolerance of outsiders
(opposition elites) to build up a party outside of regime front can ironically cause authoritarian
breakdown as they pave the way for two types of elite to pool in resources and to lower costs in
collective actions against authoritarian regimes. Current literature of authoritarian institutions
looks at the effect of a single institution on authoritarian breakdown, but overlooks how
institutions interplay with each other to influence elites’ strategic behaviors. A synthetic view of
institutions is expected to enrich our understanding in how elites act in the context of
authoritarian regime for future research.
Additionally, most scholars explain why dictators set up ruling party, but few of them
explain why they tolerate the existence of party outside of regime front. This chapter explains
that as dictators choose to share power with local elites, their choice of tolerating the existence of
party outside of regime front may not be the worst choice to the political survival of authoritarian
regime.

37

CHAPTER 3
POLITICAL DECENTRALIZATION AND VIOLATION OF
BILATERAL INVESTMENT TREATIES

3.1

Introduction
Dictatorships have gained greater significance in an increasingly competitive game for

global capital. According to the World Bank, global inflow of FDI has been up to 1.17 trillion
USD in 2010, of which dictatorships occupy 62.89 %.

12

But dictatorships are believed to be less

credible in attracting FDI due to the lack of executive constraints, political stability, transparency
in policy change, and protection of human rights compared with democracies (Jensen 2003; 2008;
Blanton and Blanton 2007; Rosendorff and Shin 2012).
To gain international investors’ trust, dictatorships need to rely on international institutions
to lend credibility to their treaty compliance (Fang and Owen 2011; Rosendorff and Shin 2012).
Given dictatorships’ preference for bilateral treaties over multilateral ones (Garriga 2009), BITs
make a reasonable choice of commitment tool for authoritarian regimes. The number of BITs has
increased to 3,060 engaging 174 signatory states from 1959 to 2014, of which 47.6 % involves at
least one dictatorship, as reported by the United Nations Conference on Trade and Development
(UNCTAD).

13

12

Calculated from the inflow of foreign direct investment by World Development Indicators,
World Bank accessed by STATA.
13
Calculated from International Investment Agreement on UNCTAD website:
http://investmentpolicyhub.unctad.org/IIA (Access on March 4, 2015).
38

0

0

2

FDI, net inflow in million USD
10000
20000

The Number of ICSID Cases
4
6
8
10

12

30000

Figure 3-1: The Trend of FDI Inflow and ICSID Cases Invoked by BITs for Dictatorship

1987 1992 1997 2002 2007 2012
Year

1987 1992 1997 2002 2007 2012
Year

Source: UNCTAD and WDI

Many believe that signing BITs not only signals the host country’s intention to protect FDI
(Haftel 2010; Neumayer and Spess 2005), but also punishes the BIT’s violation by reducing FDI
in the future (Allee and Peinhardt 2011).
However, dictatorships and international investors do not live happily ever after BITs are
concluded. As I present in Figure 3-1, a growing inflow of FDI to dictatorships is accompanied
by an increasing number of ICSID lawsuits against them. Moreover, some dictatorships have
been accused of BIT violations often, while others have not. For instance, Egypt and Venezuela
have been sued for BIT violations at least ten times before International Center for Settlement of
Investment Disputes (ICSID), while Singapore and Cuba maintain a perfect record of not being
sued. The question is what makes some dictatorships more likely to be sued before ICSID cases

39

than others after BITs are signed? For those signing BITs as international institutions committed
to international market, what is the domestic determinant that makes some countries good
compliers to their commitment and some not?
Existing literature of fiscal decentralization sheds light on this puzzle. Camps of scholars
debate about whether decentralization matters to the increase and protection of FDI, which is the
purpose of enforcing BITs. Some scholars argue that decentralization facilitates FDI because it
produces a group of local elites in fierce competition for limited inflow of FDI and thus fosters
an efficient capital market, innovative polices, and limited central intervention (Montinola, Qian,
and Weingast 1995; Oates 1972; Qian and Weingast 1996, 1997; Tiebout 1956; Xu and Zhuang
1998). Those conditions are attractive for international investors. Some refute this claim and
argue that decentralization does not benefit FDI. Instead of facilitating market efficiency,
competition for capital among local elites incentivizes them to help enterprises to find the
loopholes in central government’s policy (Cai and Treisman 2004, 2005, 2006; Treisman 2007).
However, the debates fall short to develop a general argument, inadequately apply the
measure of fiscal decentralization to dictatorships, and weakly address the distinction between
developing countries and dictatorships. To contribute to the literature, I employ a quantitative
study to examine the effect of political decentralization on BIT violation under dictatorships.
Political decentralization refers to the existence of local elections which electorally empowers a
group of local elites capable of amassing the votes into their favor in specific regions of political
interests (McDonough 1995; Scobell 2014; Tarrow 1995; Welzel 2006). Given that local elites
are electorally connected with interest groups, they tend to protect interest groups’ profits at the
cost of violating BIT. The conflict of interest between central and local governments takes place
in democracies as well. However, authoritarian regime lacks an ultimate authority to regulate the

40

relationship between central and local governments. Therefore, the conflict can be resort to
judicial review in democracies, but not dictatorships. The unsolved conflict eventually results in
international litigation.
As illustrated in Figure 3-2, the general trend shows that dictatorships with political
decentralization are more likely to be sued in ICSID cases invoked by BITs than those that are
not. Under the auspice of World Bank, the Dataset of Political Institutions provides the variable
of political decentralization. The number of ICSID cased invoked by BIT is calculated from the
record published on the ICSID website

14

. The t-test shows that the installation of

decentralization does significantly lead to a higher number of ICSID cases at the p< 0.001 level.

14

ICSID website:
https://icsid.worldbank.org/apps/ICSIDWEB/icsiddocs/Documents/List%20of%20Contracting%
20States%20and%20Other%20Signatories%20of%20the%20Convention%20-%20Apr%202014.
pdf (Access on April 4, 2015).
41

0

Mean No. of ICSID Cases Invoked
.01
.02
.03
.04

.05

Figure 3-2: The Mean Number of ICSID Cases by Decentralization for Dictatorships

No

Politial Decentralization

Yes

Source: DPI and ICSID

In this chapter, I argue that political decentralization exacerbates the principal-agent
problem between central and local governments. After concluding BITs the central government
(principal) delegates the mission of BIT compliance to local elites (agents), as FDI takes physical
roots in regions. Under the framework of BIT, electorally minded local elites tend to attract
limited inflow of FDI by bidding up terms, such as tax reduction, subsidy, land grant and etc. The
extra and detailed terms automatically qualify the target of protection by BIT. elected local elites
seek rents for provision of public/private goods in the interest of sustaining their political
survival. FDI is one of the sources to boost economic development from which local elites can
extract rents.
However, Local elites are incentivized to over-commit the extra terms to outbid their
counterparts. Those terms are conditional on local elites’ capacity and intention to fulfill.
42

However, local elites’ capacity and intention to carry out the terms may change over time so that
they have time-inconsistency issue with regard to BIT compliance. When local elites are caught
in the dilemma between BIT compliance delegated by central government and interest groups’
needs, electorally minded local elites will succumb to the latter as they face the
time-inconsistency issue.
This is particularly serious for dictatorships because they lack an ultimate authority to
define, interpret and settle the disputes between central and local governments with regard to the
BIT violations. Thus, BIT violations become a possible outcome jeopardizing a dictatorship’s
international reputation, a common good among local elites who are incentivized to free ride
each other’s efforts to maintain it.
This chapter is organized as follows. In the next section, I review the literature on how
decentralization influences the protection of FDI, the very purpose of signing BITs with
dictatorship in the first place. Next, I argue that political decentralization leads to more BIT
violations where local elites tend to overcommit to international investors in BITs. BIT violation
is a tempting choice in face of time-inconsistency problem. In the empirical sections, I first
discuss the research design and then present empirical evidence to support my argument. The
final section concludes the findings and implications to literature.

3.2

Fiscal Decentralization, FDI Inflow and BIT Violations under Dictatorship
The introduction of international institutions imports credibility for dictatorships lacking

credibility in international cooperation (Fang and Owen 2011; Rosendorff and Shin 2012). To
clear the doubts of hold-up problems, home states tend to sign investment treaties to signal their
intention to protection the property rights of international investors’ investments. BITs are

43

international agreements to prevent home counties of FDI from facing “obsolescing bargain”
with host countries holding up investor’s assets for greater profits after FDI is settled (Vernon
1985). Since dictatorships favor bilateral over multilateral treaties (Garriga 2009), BITs are
commitment tools to engage international investors’ home countries.
To protect FDI from being held up by an authoritarian regime, BITs stipulate the terms and
conditions for protection and exercise of property rights, preferential tariff levels, free transfer of
production means, resolution of disputes, access to loans, tax terms, and so on. The hold-up
problem concerns international investors because the costs for reallocation of assets are huge.
Once foreign investments are settled in specific regions, relocation takes extra costs, which may
not be economically efficient. BIT is an important tool for dictatorships to build up international
credibility to relieve international investors from the concerns over hold-up problem (Allee and
Peinhardt 2014; Kerner 2009; Neumayer and Spess 2005).
Although the ongoing debate shows mixed results in whether signing BITs induces more
FDI (Tobin and Busch 2010; Tobin and Rose-Ackerman 2011; Büthe and Milner 2008; Kerner
2009; Salacuse and Sullivan 2005; Haftel 2010; Egger and Pfaffermayr 2004), it is confirmed
that being sued can lead to significant reduction in FDI (Allee and Peinhardt 2011). According to
Allee and Peinhardt’s calculation (2011, 423–426), the net FDI inflow reduces by $55 million
USD for each pending case in an annual base on average. The figure goes up to an average of
$791 million USD if a respondent country has lost an ICSID case within the last two years.
Since FDI is affected by BIT violations, I look at how decentralization results in BIT violation.
Previous literature presents two competing lines of argument about the effect of
decentralization on international cooperation in terms of protecting FDI, for which BITs are
devised. For the camp taking a positive view on the effect of decentralization for securing FDI,

44

scholars argue that decentralization facilitates FDI because it produces a group of local elites in
fierce competition for limited inflow of FDI and thus fosters an efficient capital market attractive
for international investors (Montinola, Qian, and Weingast 1995; Qian and Weingast 1996, 1997;
Tiebout 1956; Xu and Zhuang 1998).
Some authors attribute double-digit economic growth rate in China to market-preserving
federalism, which adds three more criteria upon Riker’s two criteria (Montinola, Qian, and
Weingast 1995; Qian and Weingast 1997; Riker 1964). Riker (1964) proposes two criteria, “(1) a
hierarchy of governments, that is, at least "two levels of governments rule the same land and
people," each with a delineated scope of authority so that each level of government is
autonomous in its own, well-defined sphere of political authority, (2) the autonomy of each
government is institutionalized in a manner that makes federalism's restrictions self-enforcing”.
Weingast and his coauthors (1995) adds three more criteria to define market-preserving
federalism:” (3) subnational governments have primary regulatory responsibility over the
economy, (4) a common market is ensured, preventing the lower governments from using their
regulatory authority to erect trade barriers against the goods and services from other political
units, and (5) the lower governments face a hard budget constraint, that is, they have neither the
ability to print money nor access to unlimited credit.”
For instance, Weingast and his coauthors (1995; 1996) argue that the decentralization
system in China is of market-preserving features since the start of economic open policy in 1979.
The Chinese central government limits the degree of central planning in economy and
empowered local elites to experiment on innovative methods of boosting economic development
in regions. To ensure a constant inflow of FDI, fiscal decentralization empowers local elites
dedicated to market security hinged on binding BITs China has concluded with international

45

investors’ home countries.
On the contrary, some scholars posit that the installation of decentralization corrodes market
efficiency (H. Cai and Treisman 2004, 2006; Treisman 2007). Cai and Treisman contend that the
market-preserving federalism argument is overly optimistic about the central government’s
capacity to collect tax or to impose regulations. When the central government is incapable of
enforcing policy goals, market efficiency is not derived when local governments compete for
capital. Local elites are motivated to attract investment by aiding tax or regulation evasion for
enterprises. Under this logic, local elites’ competition harms market efficiency. Thus,
decentralization carries negative effect on the inflow of FDI.
In the case of China, Weingast and his coauthors argue that fiscal decentralization is
irrelevant to China’s dazzling economic development. First, the causal sequence is wrong. The
initiation of economic openness precedes the practice of decentralization. Second, fiscal
decentralization does not improve fiscal discipline in state enterprises. Also, taxation consumes
the premium of revenue. The Chinese economic miracle is actually rooted in competing central
factions with ideological rivalry and local networks
Nevertheless, the debates are not satisfying in the following regards. First, the debates are
limited to the discussion on the cases of China and Russia. The literature begs an examination on
dictatorships in general. In addition, the sole focus on China and Russia neglects greater
variation within dictatorships, especially since China and Russia are both territorially huge states
with the legacy of communism. Most dictatorships do not share the characteristics unique in the
two countries.
Second, previous literature focuses on the measure of fiscal decentralization, which is
arguably inappropriate to address the issue of BIT compliance. Rampant corruption and low

46

transparency under authoritarian regime makes it hard to evaluate the continuance and change of
fiscal and administrative decentralization. For instance, under-the-table revenues and informal
hired hands can under-estimate how much organizational power local elites can leverage to
bargain with central government. It is also likely that the central government still keeps an upper
hand in economic planning even when it maintains a high tax share for local governments. Either
way compromises the integrity of dataset in fiscal or administrative decentralization.
Lastly, the debates neglect the distinction between developing countries and dictatorships.
Despite that most dictatorships are in developing countries, not all developing countries are
dictatorships. The nature of regime type is of great political significance for any country to
engage in international commitment like BITs. The terms in BITs are left for bilateral
negotiations between two countries. Once deposited to ICSID for a third-party intervention in
time of disputes, the judgement is made upon the given terms of BITs accordingly. Therefore, the
nature of regime type is of greater importance than the level of economic development as I
elaborate the discussion on BITs violation.

3.3

Political Decentralization, Over-Commitment and Authoritarian Credibility in BITs
In this chapter, I focus on how treaty compliance with international commitments under

dictatorships is conditional on political decentralization, where local elections exacerbate the
principal-agent problem between central and local governments. To be specific, I argue that
political decentralization results in more BIT violations because elected local elites are motivated
to over-commit in competition for inflowing FDI and ignore BIT compliance, the
implementation of policy central government delegates to local elites.
The principal-agent problem refers to the situation when dictators (principals) have a hard

47

time motivating local elites (agents) to act in principal’s best interests rather than in agent’s own
interests (Miller 2005). In the case of BITs, the principal (central government) and the agent
(local governments) both agree that signing of BITs is in both parties’ best interests because it
can bring more FDI beneficial to economic development (Borensztein, De Gregorio, and Lee
1998; Zhang 2001). To attract FDI, local agents support their central government’s signing BITs
with the terms of favorable subsidy, land appropriation, tax reduction, and various policy
preferential treatments to foreign investors. Once a BIT is signed, the central government
delegates the policy of BIT compliance to local elites, as FDI takes physical roots in regions. But
the maintenance of credibility in BITs is problematic after the BITs are concluded.
The existence of local elections diverge central and local governments on the issue of
complying with BITs. For the central government, complying with BITs ensures a constant
inflow of FDI. Once international reputation is tarnished by being sued for BIT violation before
international court, the volume of inflow FDI will decrease (Allee and Peinhardt 2011).
FDI is one of the sources to boost economic development from which local elites can
extract rents. After central government signs BIT with foreign countries, local elites are
motivated to outbid each other with extra terms in the interest of maximizing the inflow of FDI
in the region of their rule. The extra terms are automatically enrolled in the protection of BITs.
With more inflow of FDI, local elites benefit from extracting more taxes derived from growing
economy. In turn, they can use tax income to provide supporters with provision of public/private
goods in the interests of sustaining their political survival.
As elected local elites face the issue of time inconsistency, local elites are tempted to
prioritize local needs at the cost of BITs violation, resulting in tension between central and local
governments. The issue of time inconsistency refers to the change in local elites’ preferences in

48

BIT compliance derived from the change in conditions of signing BITs from time t to (t+1). For
instance, political or economic shocks, such as recession, coups, wars and so on so forth, can
severely shrink local elites’ capacity or incentive to fulfill the promises they make in BITs.
The time-inconsistency issue is particularly serious for dictatorships due to the lack of an
ultimate authority to define, interpret and settle the disputes in BITs compliance between central
and local governments. Given the logic of over-commitment, BITs violation jeopardizes a
dictatorship’s international reputation, a common good among local elites who are incentivized
to free ride each other’s efforts to maintain it.
For instance, the provincial leaders are elected by local legislatures, which are directly
elected under the supervision of CCP. Under the political decentralization system, local elites in
China are agents who need to comply with 108 BITs the Ministry of Commerce in central
government has so far concluded. Provincial governments violate BITs so that China has
engaged in two lawsuits because they are accused of violating BITs. The first case is Ekran
Berhad v. People's Republic of China (ICSID Case No. ARB/11/15). Ekran Berhad, a
Malaysia-based company, accused Hainan province of violating Malaysia-China BIT signed in
1988 (Shan and Su 2015). Hainan province committed a land usage to Ekran Berhad, but later
revoked the right of lease on land use by asserting that Ekran Berhad failed to develop the land
under local regulations stipulated in local legislation. The case was registered in 2011. The other
lawsuit is invoiked by South Korea-China BIT signed in 2007. Ansung Housing Co., Ltd. v.
People's Republic of China (ICSID Case No. ARB/14/25) is an investment dispute in property
development for a golf country club and condominiums in Jiangsu province.

15

15

Ansung accuses

Herbert Smith Freehills Law Corporate’s website:
http://hsfnotes.com/arbitration/2014/11/10/china-sued-by-south-korean-property-developer-at-ics
49

Jiangsu province of intruding on its right to delay the schedule of an investment plan.
Based on these discussions, I empirically test the following hypothesis in this chapter:
Hypothesis: The installation of local election is positively related to the number of ICSID cases
invoked by BITs for dictatorships.

3.4

Research Design
In this chapter, I examine the effect of political decentralization on BIT violations among

dictatorships. To fit the research interest, the sample is restricted to dictatorships engaged in
effectively enforced BITs. I include cases where a dictatorship is sued and drop the cases where
non-state actors are sued.

16

I also adopt Geddes, Wright and Frantz’s variable to identify

non-democracies (2014). The effectively enforced BITs come from the dataset of International
Investment Agreements, which was collected and updated under the auspices of UNCTAD. The
duration of BIT is determined by the time span of a signed and enforced BIT. The unit of
analysis is the country-year. The dataset covers 81 dictatorships dating from 1959 to 2013.
The dependent variable is a count variable coding the total number of cases invoked by
BITs registered to ICSID for a given dictatorship in a given year. The data on this variable is
collected via the historical records of dispute settlements published on the website of ICSID.

17

Until 2014, 150 states have signed the Convention on the Settlement of Investment Disputes
between States and Nationals of Other States and deposited their instrument of ratification to the

id/ (Access on April 4, 2015).
16
I drop two observations where respondent is a non-state actor in the case. The case numbers
are ARB/76/1 (Société Serete S.A ) and ARB/07/3 (PT Kaltim Prima Coal and others).
17
ICSID website:
https://icsid.worldbank.org/apps/ICSIDWEB/Pages/default.aspx (Access on January 14, 2015).
50

ICSID at the World Bank.

18

The coverage of state members and BITs in ICSID is

comprehensive enough to cover the legal actions initiated with regard to signed BITs. The
legitimacy of the ICSID is confirmed by statistical examinations by empirical legalists (Franck
2010).
One may argue that it is inappropriate to operationalize BIT violations as the number of
being sued in ICSID case because claimants can use lawsuits as a strategic measure to blackmail
host states of FDI, in efforts at re-negotiation for better terms. In other words, respondents in
lawsuits for violating BITs may not actually have violated them. Besides, lawsuits represent the
claimant’s subjective evaluation and unilateral action against the respondent states. Treating the
sheer existence of lawsuits as violation of BITs can overestimate the likelihood of BIT violation.
However, this objection could not be sustained by empirical evidence. On average, a
registered case takes 4.6 years and millions US dollars before a case is concluded. This is quite a
toll for private nationals to file a lawsuit against a state.

19

During the proceedings, the assets and

capital under disputes are greatly limited for liquidation, which further prevents promiscuous and
groundless accusations against the host states.
In this regard, claimants would have exhausted all possible means to settle the disputes
before resorting to international trials. It is highly likely that host states would have violated BITs
before they are filed for dispute settlement in the ICSID by claimants. The observations on

18

Calculated from ICSID website. See footnote 17.

19

The Figures are calculated from ICSID website. The ICSID rules published on website as
well. According to the costs of proceedings governed by ICSID rules, claimants are private
nationals who have to make payments including the party’s expenses, the advances paid to
ICSID and the lodging fee paid by the party instituting proceedings. The average years are
calculated from the duration from the date registered to the date of outcome of proceeding for
concluded cased published on ICSID website.
51

lawsuits thus would only under-estimate rather than over-estimate the occurrence of BITs
violation. As stated previously, lawsuits invoked by BITs before international court may
underestimate the presence and prevalence of BITs violation as high costs induced by lawsuits
may deter claimants from seeking legalist solutions.
In addition, lawsuits invoked by BITs are arguably an objective indicator of treaty violation.
A direct measurement and identification of BIT violation cannot be objective because home and
host states engaged in BITs disputes may provide biased interpretations of terms and selective
presentation of evidence to their favor. Therefore, the ICSID is an objective measure not only it
is recognized as an objective third-party in BITs deposition, but also the outcomes of litigations
prove the objectivity of litigation in an ex post way. In the ICSID cases against dictatorships,
69.2 percent of concluded cases end up with outcomes favorable to claimants, including awards
and settlements. 21.2 percent of cases end up in discontinuance of proceeding without
information with regard to disputes. And only 9.6 percent end up proving authoritarian
respondent innocent with an outcome of annulment.
I choose negative binomial regression models (NBRM) based on the nature of dependent
variable. Moreover, being sued at time t may change the likelihood of being sued in time (t+1).
Compared with Poisson or ordinary least square (OLS) regression models, NBRM accounts for
the contagious dynamic in the dataset, produces standard errors without downward bias, and
avoids downward biased p-values (Brandt et al. 2000; King 1988; Cameron and Trivedi 1986;
1998).

3.4.1 Independent Variable
The key independent variable of this chapter is a binary variable that measures the existence

52

of political decentralization. I operationalize political decentralization as the existence of local
elections. Philip Keefer and colleagues constructed the Database of Political Institutions (DPI)
that contains an ordered variable, STATE, to indicate whether state/province governments are
locally elected (Beck et al. 2001, 20). The variable of STATE is coded 0 for non-existence of
local election, 1 for elected legislature, and 2 for elected executive and legislature. Note that the
DPI does not code cases in which the executive is locally elected, but the local legislature is
appointed. I re-compute the 3-category variable into a dichotomous variable where the existence
of local election is coded as 1 and 0 otherwise. According to my previous discussions, the
existence of political decentralization is expected to increase the number of being sued in ICSID
cases invoked by BITs for a dictatorship, since politically decentralized dictatorships are more
prone to encounter the problem of over-commitment incurred by elected local elites.
The use of political decentralization has two major advantages. First, the dataset of fiscal
and administrative decentralization is suspiciously contaminated by the lack of transparency or
government manipulation, whether local elections are installed or not is linked to local elites
electoral concerns.
Second, the significance of political decentralization rests in the unique nature of
dictatorships. The definition and settlement of disputes with regard to the division of power
between central and local governments is of great uncertainty. While democracies can always
resort to a higher authority, there exists constitutional (or judicial) review to peacefully clarify
and resolve the disputes between central and local governments. But, dictatorships lack such
authority so that the disputes may end up deadlock. On account of the reasons stated above, the
choice of political decentralization is a better measure to evaluate whether local elites are
electorally connected to interest groups whose need may conflict with central government's

53

policy with regard to BIT compliance.

3.4.2 Control Variables.
I include a battery of political, economic, and institutional controls from domestic and
international factors (Simmons 2014; Allee and Peinhardt 2011; 2014). When the presumed
conditions for states to engage in BITs change, states’ preferences change so that they fail to
adhere to international commitments. Domestic and international institutions are installed to
facilitate states’ observation of BITs (Allee and Peinhardt 2011). I will elaborate the
operationalization of each control in the following paragraphs.

3.4.2.1

Domestic Politics Controls

I include political, economic, and institutional controls in the models of domestic politics.
To account for political instability in domestic politics, I borrow the unweighted and annual
count of domestic political shocks ranging from 1 to 8 for a given dictatorship from Banks and
Wilson’s Domestic Conflict Event Data (n.d.). The count adds up the number of general strikes,
government crises, anti-government protests, riots, purges, revolutions and guerilla warfare,
which corrodes the legitimacy of leadership or even the dictatorship. Allee and Peinhardt also
include the same variable in the evaluating the reputational effects of investment treaty disputes
on FDI (2011). Dictators are tempted to confiscate foreign properties to finance ongoing civil
war. I calculate the total number of civil war from intrastate war dataset from the Correlates of
War Project (COW) in an annual fashion (Sarkees and Schafer 2000). The effect of civil war is
predicted to be positive on the count of BIT violations.
As for domestic economic factors, I use Laeven and Valencia’s dataset (2013) to incorporate

54

the number of domestic economic shocks in the form of banking, currency or debt crises a
dictatorship experiences each year. Economic shocks can dampen dictatorships’ ability to keep
up with the terms to which they commit in the BITs, as Allee and Peinhardt address in their study
(2011). The higher the number of domestic economic shocks in a dictatorship, the more likely it
violates BITs.
Moreover, economic fluctuations revise local elite’ preferences or capacity to the
maintenance of BITs. For instance, a lesser-developed economy or a soaring inflation rate raises
the risk of hold-up problem for FDI, the target of protection by BITs. The GDP per capita
measures the level of development whereas the inflation rate is an indicator of the stability of
general economy (Aizenman and Glick 2008). I adopt the logged variables of GDP per capita
and the percentage of inflation rate over GDP from the World Bank (Sarkees and Schafer 2000).
For the two economic variables, the expected direction of effect on BIT violations is mixed.
Next, domestic institutions play a crucial role in treaty adherence. Some of them are even
specifically established to produce credibility in the global market in order to conclude more
BITs. They include policies and practices with regard to protection of property rights, capital
openness, and types of dictatorships. Taken from the PRS group’s International Country Risk
Guide, I use the untransformed index of property rights protection ranging from 0 to 30, with
higher scores indicate better protection of property rights. The index is composed of four
indicators evaluating investment profiles, bureaucratic quality, corruption, and law and order
(2015). Another economic policy is capital openness, which measures the openness of capital
account (Ito and Chinn 2008). The variable is expected to decrease the BIT violations.
Party dictatorships are believed to perform better in various policy outcomes, such as
leadership constraints and policy transparency, that are beneficial to international cooperation

55

(Lai and Slater 2006; Mattes and Rodríguez 2013; Weeks 2008, 2012). I generate three dummies
of military, monarchy, and personal dictatorships and treat party dictatorship as a base from the
Geddes, Wright, and Frantz’s dataset (2014). Compared to the base, party dictatorship, the three
dummies are expected to show positive signs on BITs violation.

3.4.2.2 International Politics Controls
The source of BITs violation can also result from international factors, including political,
economic and institutional controls. To capture the political volatility in international society, I
employ the total number of interstate wars engaged by a dictatorship in a yearly fashion. The
prevalence of international war in the region and world is measured by the dataset of the
Militarized Interstate War (MID, version 4.1) of COW (Palmer et al. 2015). I generate two
variables to measure the count of interstate wars in the region and world to account for the
spillover effect of conflicts. International economic shocks carry spillover effects as well. I
aggregate the number of domestic economic shocks in the region and world, including banking,
financial and economic crises (Laeven and Valencia 2013). In addition, I take the log of the
inflow of FDI, reserves including gold, and the percentage of remittances over GDP from World
Bank (Sarkees and Schafer 2000). The inflow of FDI represents international investors’ vote of
confidence in authoritarian credibility. A larger inflow of FDI makes the maintenance of BITs
more likely, thus rewarding the honoring of international commitments. The relationship between
FDI and BIT violation is expected to be positive.
The volume of international reserve is a tool to combat financial turmoil (Ouyang, Rajan,
and Willett 2008). Large volume of international reserve decreases the likelihood of BITs as the
dictatorship ensures its capacity of liquidation. On the other hand, remittances can be a

56

replacement for FDI. The larger the percentage of remittance over GDP for an authoritarian
regime, the less essential FDI is. Therefore, the significance of BITs is lower. A negative
relationship exists between remittance and ICSID cases. To reduce variation, I log the three
variables in the models.
Lastly, international institutions exert binding effects. From ICSID website (ICSID, World
Bank 2015), I calculate the total numbers of enforced BITs signed by a given dictatorship in a
given year and in the given region as well. I also calculate the total number of lawsuits involved
in the region. The more BITs a dictatorship concludes, the more costs it has to pay in order to
digest all the preferential terms included in BITs (Tobin and Rose-Ackerman 2010). The signing
and violating BITs can diffuse in neighboring states to capture unobserved factors. In my models,
I expect a positive relationship between the three variables and BIT violation. Table A-4 in the
Appendix presents the summary statistics for all variables in models.

3.5

Results
Table 3-1 and Table 3-2 present the results of statistical analysis controlling for domestic

and international factors, respectively. The two groups of models are further divided by political,
economic, and institutional controls. In Table 3-1, I examine the relationship between political
decentralization and BIT violations with controls in domestic factors. The empirical results
suggest that political decentralization is positively related to a higher number of ICSID cases at
the p<0.05 level in five models. Financial openness significantly increases the number of ICSID
cases at the p<0.05 and p<0.01 level in Model 4 and Model 5, respectively. Greater financial
openness increases the probability that dictatorships will violate BITs, as they are too financially

57

Table 3-1: Effect of Political Decentralization on BITs Violation: Domestic Factors
DV: ICSID cases
IV: Political decentralization

Model 1
1.029**
[0.489]

Domestic political shocks
Intra-state conflicts

Model 2
1.376**
[0.591]
-0.044
[0.071]
-1.693
[1.069]

Domestic economic shocks

Model 3
1.264**
[0.503]

-0.740
[1.038]
0.034
[0.166]
0.587***
[0.205]

Inflation
GDP per capita
Property rights
Financial openness
Military dictatorship
Monarchy dictatorship
Personal dictatorship
N
Countries
Log likelihood

Model 4
1.258**
[0.637]

1312
81
-168.603

1203
78
-142.493

1083
76
-153.952

0.020
[0.087]
1.728***
[0.722]
0.547
[0.803]
0.565
[0.858]
0.065
[0.628]
742
57
-125.029

Model 5
1.417**
[0.689]
-0.066
[0.078]
-1.487
[1.098]
-0.264
[1.074]
0.153
[0.198]
0.124
[0.347]
-0.047
[0.068]
1.707**
[0.647]
0.818
[0.853]
0.494
[0.992]
0.108
[0.764]
605
54
-111.621

Note: Clustered standard errors in brackets. Three time splines and constant not shown.
* p<0.1, ** p<0.05, *** p<0.01

open to global economic fluctuations. The rest of controls are not of significant effect on the
dependent variable across models
In Table 3-2, the models contain international factors. Again, the empirical results confirm
that the relationship between the existence of political decentralization and ICSID cases is
positive at the p<0.05 level or better. NBRMs contain international factors, including political
controls (inter-state conflicts, inter-state conflicts in region, and inter-state conflicts worldwide),
58

economic controls (economic shocks worldwide, FDI inflows, foreign reserves, and remittances,
and institutional controls (BITs signed, BITs signed in the region, and ICSID cases in the region).
For controls in international political instability, none of them remain significant in all
models. As for international economic instability, the number of economic shocks in the world is
negatively correlated to ICSID cases at the p<0.05 level. Global financial crisis actually
decreases the likelihood of being sued in ICSID cases. In order to avoid the escape of FDI inflow
due to global economic crisis, dictatorships have to try harder to maintain their credibility in
BITs as they are by nature less credible in global economy (Saiegh 2005; Archer, Biglaiser, and
DeRouen 2007).
Foreign reserves and remittances are positively related to ICSID cases at the p<0.01 level.
Foreign reserves and remittances are substitute sources for FDI. Therefore, higher foreign
reserves and remittances will reduce the importance of FDI. For controls international institution,
more BITs signed leads to more ICSID cases at the p<0.1 level or better. As more BITs are
signed, more costs are incurred to maintain the preferential policies in international investors’
favor. The higher costs can go beyond a dictatorship’s capacity causing a breach in BITs. The rest
of controls are not of significant effect on the dependent variable across all models with domestic
controls in politics and economy.

59

Table 3-2: Effect of Political Decentralization on BITs Violation: International Factors
DV: ICSID cases
IV: Political decentralization
Inter-state conflicts
Inter-state conflicts (region)
Inter-state conflicts (world)

Model 6
1.025**
[0.488]
-0.379
[0.267]
-0.007
[0.044]
0.010
[0.009]

Model 7
1.364**
[0.635]

Economic shocks (world)

-0.119**
[0.052]
-0.328**
[0.153]
0.932***
[0.221]
0.356**
[0.163]

FDI inflows
Foreign reserves
Remittances
BITs signed
BITs signed (region)
ICSID cases (region)
N
Countries
Log likelihood

Model 8
0.948**
[0.471]

1312
81
-166.818

798
62
-101.266

0.130*
[0.071]
-0.002
[0.002]
0.043
[0.034]
1312
81
-166.469

Model 9
1.370***
[0.521]
-0.590**
[0.278]
0.011
[0.057]
0.004
[0.011]
-0.165**
[0.068]
-0.219
[0.149]
0.779***
[0.180]
0.541***
[0.154]
0.156**
[0.075]
-0.002
[0.002]
0.041
[0.032]
798
62
-96.515

Note: Clustered standard errors in brackets. Three time splines and constant not shown.
* p<0.1, ** p<0.05, *** p<0.01

In sum, the empirical results support the hypothesis of this chapter. With the existence of
local elections, dictatorships are more likely to commit more BITs violations. This finding goes
in line with my argument that the installation of political decentralization is more likely to
increase the likelihood of being sued for authoritarian regimes.

60

3.6

Conclusion
International institutions are constantly employed as a tool to avoid the hazard of default on

international cooperation consequential to the global economic order (Axelrod and Keohane
1985; Martin and Simmons 1998; Simmons 2000; Simmons, Dobbin, and Garrett 2008). With
more and more dictatorships engaging in globalization along with a booming literature on
authoritarian institutions, a dialogue is needed between the two lines of literature to discover the
domestic source of international cooperation for authoritarian regimes.
In this chapter, I use the case of BIT violations to examine the domestic source of treaty
compliance for dictatorships. Specifically, I look into how the number of BIT violations is
influenced by political decentralization where the existence of local elections exacerbates the
principal-agent problem between central and local governments to maintain international
commitment under dictatorships.
I argue that elected locally minded elites over-commit to foreign investors in competition
for limited inflow of foreign direct investment and later fail to fulfill the commitments stipulated
in BITs, which central governments sign with the home countries of FDI. Dictatorships lack an
ultimate authority to settle the disputes between central and local governments with regard to
BIT violations. Electorally empowered local elites will prioritize local needs at the costs of BITs
violation in the face of time-inconsistency in their preferences in complying BITs.
Based on the dataset on the ICSID litigations from 1987 to 2013, the empirical evidence
supports my argument that the installation of political decentralization increases the number of
being sued in BITs lawsuits registered in ICSID.
This chapter contributes to understanding authoritarian credibility in international
cooperation in two ways. First, I echo Allee and Peinhardt’s call (2011) to focus on the ex post

61

behavioral records after treaty is concluded to evaluate the credibility of treaty compliance. As
more and more authoritarian regimes intensively engage in international cooperation, they have
to pay attention if the arrangement of domestic institutions makes them more credible.
Second, I emphasize the importance of political decentralization, as it is one of the
important institutions left under-investigation. Political decentralization is of great political
significance. It produces local elites with geographical concerns different from major political
actors such as dictators or central elites, who are constantly mentioned in previous literature.
However, the literature of authoritarian institutions has not yet fully investigated the cause and
consequences of decentralization on international cooperation.

62

CHAPTER 4
DECENTRALIZATION AND CHINA’S BILATERAL INVESTMENT TREATIES

4.1

Introduction
Since 2011, China replaces Japan and becomes the second largest economy in the world. To

maintain the momentum of economic growth, China has been an aggressive in attracting foreign
direct investment (FDI) in global capital market. Based on the World Development Indicators
(WDI) under the auspice of World Bank (WB), China is the greatest holder of over 347 billion
USD of FDI inflow, which is about 19.9% of global share as of 2013.
To attract FDI inflow, China follows what academic recipes prescribe. In discussion of
international institution and regime type, dictatorships imports credibility from international
institutions in protecting FDI via investment treaties to overcome institutional handicaps such as
the lack of executive constraints, political stability, transparency in policy change, and protection
of human rights (Jensen 2003; 2008; Blanton and Blanton 2007; Rosendorff and Shin 2012).
Also, dictatorships prefer bilateral rather than multilateral treaty due to the ease of smaller
win-set in pairwise international cooperation (Garriga 2009). BITs make a commitment devise to
derive credibility in hosting FDI for authoritarian regimes. Lastly, signing bilateral investment
treaties (BITs) not only signals the host country’s intention to protect FDI (Haftel 2010;
Neumayer and Spess 2005), but also introduces the international forum to execute punishment on
the BITs violation by reducing FDI in the future (Allee and Peinhardt 2011).
The empirical evidence shows that signing BITs with China is an enthusiastic trend. As
reported by the United Nations Conference on Trade and Development (UNCTAD),

20

Calculated from International Investment Agreement on UNCTAD website:
http://investmentpolicyhub.unctad.org/IIA (Access on April 4, 2015).
63

20

Figure

0

Total Number of Signed BITs
3
6
9
12

100
0
20
40
60
80
Accumulated Number of Signed BITs

15

Figure 4-1: The Number of China’s BITs and FDI Inflow: 1982-2009

1980

1985

1990

1995
Year

2000

2005

2010

Total Number of BITs Siged in Year t
Accumulated Number of BITs in Year t

Source: UNCTAD

4-1 shows that the peak in the annual number of China’s BITs falls around 1993 when
Chinaenters the Convention of International Center for Settlement of Investment Disputes
(ICSID).

21

Although ongoing debates show mixed results in whether BITs induces more FDI

(Tobin and Busch 2010; Tobin and Rose-Ackerman 2011; Büthe and Milner 2008; Kerner 2009;
Salacuse and Sullivan 2005; Haftel 2010; Egger and Pfaffermayr 2004), BITs draws more FDI
for China as the trend of BITs is accompanied by that of FDI in a positive way. Note that the
slope of accumulated number of BITs appears flatter somewhere around 1997 as there are not
many countries that have not yet signed BITs with China in the world system.

21

Calculated from ICSID website. See footnote 17.
64

150

Figure 4-2: Total Number of Countries Signed BITs by Regime Type

Germany (136)
Switzerland (119)

0

Total Number of BITs
50
100

China (107)
Egypt (98)

Non-Democracy (Mean = 20.4)

Democracy (Mean = 39.3)

Source: UNCTAD

Through the comparative lens, China has signed 107 BITs than most democracies and all
dictatorships. According to Figure 4-2, the mean number of BITs signed for democracies is 39.3
while that for non-democracies is 20.4. For democracies, only Germany and Switzerland surpass
China in terms of total number of BITs. China makes an exceptional case for existing research
testing the effect of regime type on singing BITs (Elkins, Guzman, and Simmons 2006; Tobin
and Rose-Ackerman 2010). Among dictatorships, China signs the highest number of BITs.
But, what makes China more capable of signing BITs? Specifically, what are the domestic
determinants in China that makes it uniquely capable of being credible to engage other countries
in BITs? Current literature deploys a debate with regard the attribution of decentralization to
economic miracle of double-digit growth in China with a focus on FDI, the purpose of singing
65

BITs in the first place.
Some scholars posit that decentralization increases FDI because local elites are empowered
to avoid central intervention and foster an efficient capital market via competing for footloose
FDI (Montinola, Qian, and Weingast 1995; Oates 1972; Qian and Weingast 1996, 1997; Tiebout
1956; Xu and Zhuang 1998). To outbid counterparts in capital game, local elites are incentivized
to embrace innovative polices. Weingast and his coauthors coined the term, market-preserving
federalism, to describe the decentralization system in China. They attribute the economic miracle
to the pro-market features attractive to foreign investors (Montinola, Qian, and Weingast 1995;
Qian and Weingast 1996, 1997).
Some contend otherwise. Decentralization system in China is not the key to the booming
FDI inflow. Instead of facilitating market efficiency, competition for capital among local elites
dampens it. As a matter of fact, local elites in China help corporations evade central tax
regulations (Cai and Treisman 2004; 2005; 2006; Treisman 2007). They argue that power
struggle among central factions rather than empowered local elites decide the devise, change and
continuity of innovative policies in provincial level. Others contend that social homogeneity
matters to the issue of distribution so that it conditions the effect of decentralization on FDI
(Rodden and Rose-Ackerman 1997; Treisman 1999; Cai and Treisman 2004; 2005; 2006).
In this chapter, I argue that informal Chinese leader’s preference to decentralization policy
determines whether local elites are empowered enough to sign BITs for the sake of attracting FDI.
To be specific, Deng Xiaoping (Deng) preferences on empowering local elites determine the de
facto practice of China’s decentralization policy, leading to the high number of China’s BITs. To
test the arguments in the literature of authoritarian institutions and dynamics of power elites, I
use leadership turnover and installation of tax-sharing system (TSS) as proxies to the measure of
66

decentralization in China. Based on the empirical evidence from 1983 to 2009, the number of
China’s BITs is significantly influenced by Deng’s influence from the start of economic openness
policy in 1978 to his decease in 1997. TSS does not decrease the number of China’s BITs until
Deng died.
The layout of this chapter is as follows. Next, I review the literature on how decentralization
in China influences the protection of FDI, the purpose of signing BITs with dictatorship in the
first place. Next, I argue that decentralization leads to more China’s BITs where Deng exerts
influence on empowering local elites. After he passed away, the recentralization institution, TSS,
decreases the number of China’s BITs as local elites are less empowered. To present the
empirical evidence, I first explain the research design and show empirical results to support my
argument. The final section discusses the findings and the implications to existing literature.

4.2

Decentralization, FDI and Signing BITs under Dictatorship
Whether decentralization facilitates China’s double-digit economic growth, two branches of

competing literatures present contradictory arguments about the effect of decentralization on
international cooperation in terms of protecting FDI, for which BITs are devised. For the camp
with a positive view on the effect of decentralization for securing FDI in China, scholars argue
that decentralization is benign to attracting FDI because empowered local elites are forced to be
innovative enough to outbid their counterparts in competition for limited inflow of FDI (G.
Montinola, Qian, and Weingast 1995; Qian and Weingast 1996, 1997; Tiebout 1956; Xu and
Zhuang 1998). An efficient capital market is thus born to benefit the virtuous circle of aiding
economic growth.
Weingast and his coauthors describe decentralization policy in China as market-preserving
67

federalism, which contributes to the double-digit economic miracle in China. The
market-friendly features heaps three more criteria upon Riker’s definition (Montinola, Qian, and
Weingast 1995; Qian and Weingast 1997; Riker 1964). They include Riker’s (1964) two criteria,
“(1) a hierarchy of governments, that is, at least "two levels of governments rule the same land
and people," each with a delineated scope of authority so that each level of government is
autonomous in its own, well-defined sphere of political authority, (2) the autonomy of each
government is institutionalized in a manner that makes federalism's restrictions self-enforcing”.
In addition to Riker’s two criteria, Weingast and his coauthors (1995) adds three more:” (3)
subnational governments have primary regulatory responsibility over the economy, (4) a
common market is ensured, preventing the lower governments from using their regulatory
authority to erect trade barriers against the goods and services from other political units, and (5)
the lower governments face a hard budget constraint, that is, they have neither the ability to print
money nor access to unlimited credit.”
Under the market-preserving federalism, Chinese central government refrains itself from
central intervention and empowered local elites to boldly test innovative and custom-made
policies to facilitate economic development in specific regions. To ensure a constant inflow of
FDI, they adopt fiscal decentralization to measure how much local elite are empowered.
The head-to-head line of argument maintain that fiscal decentralization hurts market
efficiency (Cai and Treisman 2004; 2006; Treisman 2007). Cai and Treisman contend that in the
case of China, fiscal decentralization is irrelevant to China’s economic performance. First, the
causal sequence is wrong. The economic openness reform starts earlier than fiscal
decentralization. Second, fiscal decentralization induces cheating over tax sharing and
encourages extraction of corporate revenue. The Chinese economic miracle is actually derived

68

from ideological struggles among central factions with various local networks.
Nevertheless, the debates are not satisfying in the following regards. First, the quality of
data in fiscal decentralization in China is no satisfying. In general, the datasets of dictatorships
contain a lot of missing data (Hollyer, Rosendorff, and Vreeland 2011). Treisman’s dataset on
fiscal decentralization is seriously missing. Computing from a given timespan, Treisman
provides one observation for China. The variation is extremely limited. The dataset in China are
suspiciously manipulated by officials (Holz 2004, 2005; Wallace 2014). Fiscal decentralization is
no exception.
Second, de jure decentralization does not necessarily account for the de facto function. The
prevalent corruption and lack of transparency makes China a difficult case to observe evaluate
the continuance and change of decentralization. For instance, under-table revenues and informal
hired hands can under-estimate how much organization power local elites can leverage to bargain
with central government. On the contrary, it is also likely that central government still keeps an
upper hand in economic planning even when it maintains a high tax share for local governments.
Leaders’ preferences are arguably to be greatly consequential to the function of decentralization
in China. Their influences do not necessarily reflect on the sheer existence of institutions.
Lastly, developing countries and dictatorships are fundamentally different. Although most
dictatorships are developing countries, not all developing countries are dictatorships. The nature
of regime type is of great political significance for any country to engage in international
commitment like BITs. The lack of transparency makes foreign investors hesitate on investing in
dictatorships, but not developing countries.

69

4.3

Leadership and Institution as Proxies to Decentralization and China’s BITs
In this chapter, I examine the policy effect on China’s BITs via leadership or institution as

proxies to decentralization in China. To be specific, I test which proxy best keeps the policy
consequences of signing BITs.
A burgeoning literature of authoritarian institutions suggests that institutions reflect
temporal equilibrium of leadership dynamics under authoritarian regimes (Brownlee 2007; Lai
and Slater 2006; Weeks 2012). The same institution functions differently as the dynamics of
leadership change across time. Under thumb of the Chinese Communist Party, China is
categorized as a party dictatorship (Boix and Svolik 2013; Geddes, Wright, and Frantz 2014;
Svolik 2012a). The form of authoritarian rule produces the most desirable policy outcomes.
On the other hand, if institutions are tools of power-sharing commitment to overcome the
time-inconsistency issue (Boix and Svolik 2013; Svolik 2012b), institutions remain effective
regardless of leadership change. China has revised the central-local relations via several
transitions in institutions.
Existing literature on China’s decentralization policy takes two approaches. The first one is
leadership approach. The literature examines top leader’s preferences about empowering local
elites. Also, there are two types of top leaders in China: the de jure leaders and the de facto ones.
The two types of leaders are not necessarily the same person. For de jure leaders, the general
secretary of CCP Deng is recognized as the top rule of China. However, Deng is considered the
highest mater without the crown of general secretary of CCP. The other approach takes an
institutional look by investigating the devise and consequences of TSS, a recentralization
institution to weaken local elites’ power in 1994. In the following paragraphs, I will elaborate
these two lines of arguments.

70

4.3.1 Leadership
In China, prime minister of central government answers to general secretary of CCP, the top
leader in the party-state. However, de jure institution functions differently across leaders.
However, formal leader in China is not necessarily the true leader, especially during Deng’s era
since he is recognized as the “core of leadership”. The discussion of leadership will take two
folds. The first is about the CCP General Secretaries while the other is about Deng with regard to
their preferences about empowering local elites.
Since the start of economic openness reform since 1978 until the latest BITs China signs in
2009, there are five CCP top leaders with various preferences in decentralization policy. They are
Hua Guofeng (Hua, 华国峰, 1976 - 1981) , Hu Yaobang (Hu1, 胡耀邦, 1982 - 1987), Zhao
Ziyang (Zhao, 赵紫阳, 1988 - 1989) , Jiang Zemin (Jiang, 江泽民, 1989 - 2002) and Hu Jintao
(Hu2, 胡锦涛, 2003 - 2012). Whether they are called either chair or general secretary, they are
the de jure ruler of the party-state.
Among the five general secretaries, Hua was removed from his general chairmanship of
CCP because he fails his ideological battle with Deng (Fontana 1982; Xu and Zhuang 1998). He
insisted on the central planning economy over liberal economy. Therefore, his prefers not to
empower local elites with regard to economic policies. During his tenure, China concluded zero
BIT because power elites was still in struggle for ideological differences. Hu1 ensures that local
elites are sufficiently empowered to conduct policy experiments in regions (Lin 1992). Zhao
supports the liberal economic development in the coastal provinces by loosening the central
control over local fiscal policies (Bachman 1986). Local elites have autonomy in promoting FDI
inflows with free hands in conducting custom-made policies in coastal regions.
71

After the outbreak of Tiananmen Incidence in 1989, conservatives in CCP urged for major
revision of central-local relations as local elites are too strong to go against a weak central
government as the booming economic development emboldens local elites’ economic power.

22

The regional disparity in economic development between coastal and inland provinces grows the
tension between provincial elites and central governments (Bachman 1986). In latter phase of
Jiang’s tenure, Jiang initiates recentralization by shrinking local elites’ fiscal control and
strengthen central government’s capacity to subsidize inland provinces (Zheng 2000).
In Figure 4-3, I present the accumulated number of China’s BITs by leadership turnover.
China signs the first BITs in 1982. During Hua’s rule, the ideological struggle battles over the
economic openness policy including attracting more FDI in China. Under his tenure, China did
not sign any BITs. The slope is steep in the tenures of Hu1, Zhao and the first half of Jiang’s
tenure so that their preferences for decentralization reflect on China’s BITs behavior. The slope
turns flat in the latter half of Jiang’s tenure and even flatter for Hu2’s rule. Note that there are not
many countries left to sign BITs with China during Hu2’s tenure.

22

See http://www.epochtimes.com/b5/1/7/19/n111179.htm (Access on April 4, 2015).
72

Hua

Hu1

Zhao

Jiang

Hu2

0

The Accumulated No. of China's BITs
25
50
75
100

Figure 4-3: Accumulated China’s BITs by Leadership Turnover

1978

1982

1986

1990

1994
Year

1998

2002

2006

2010

Source: UNTAD

Judging the leader’s preferences in the governing circumstances in China, I derive the first
hypothesis as follows.
H1: All else equal, general secretaries’ rule increases the likelihood of maintenance of
effective BITs with China since 1978.

In the context of Chinese politics, formal leaders are not necessarily true leaders. Deng is
believed to be the master mind behind the economic openness reform in China since 1978 until
his decease (Baum 1996). He is the true holder of power despite that he has never been crowned
the general secretary of CCP, the official ruler in China. According to authorized history of the
CCP documenting the concluding remarks in the Third Plenary Session of the 11th Central
73

Committee of CCP, Deng is the “core of second-generation leadership”.

23

Deng is an informal

leader who makes ally with provincial elites sharing his preferences in liberal economic reform
to battle against the conservatives in CCP (Bachman 1986).
Four general secretaries of CCP have lived under his influence. The full terms of Hua, Zhao
and Hu1 from 1978 to 1988 are under Deng’s influence. Deng remains alive for the first half of
Jiang’s term from 1989 to 1997. After Deng’s death in 1997, Jiang serves his second half of
general secretary until 2002 without Deng’s shadow. Hu2’s rule is free from Deng’s direct
influence. His decease brings the contraction of local elites’ power (Zheng 2000).
Recentralization policy is executed to salvage the weakening central governments.
Figure 4-4 shows that the slope of accumulated China’s BITs is steep before Deng’s decease.
After Deng’s decease in 1997, the slope becomes flat. Based on the previous discussion, I
develop my second hypothesis as follows.
H2: All else equal, Deng’s informal rule increases the likelihood of maintenance of effective
BITs with China from before his decease in 1997 than after 1997.

23

http://www.chengmingmag.com/t344/select/344sel15.html (Access on April 4, 2015).
74

The Accumulated No. of China's BITs
25
50
75
100

Figure 4-4: Accumulated China’s BITs before and after Deng’s Decease

Before Deng's Decease

0

After Deng's Decease

1978

1982

1986

1990

1994
Year

1998

2002

2006

2010

Source: UNCTAD

4.3.2 Tax-sharing System
In 1994, China installed TSS to shrink local elites’ fiscal power. Before TSS is installed,
local elites are greatly empowered by “separate stoves where each level of government is granted
the power to manage the revenue and expenditures (Fenzao Chifan) (Wong 1991; 2000). With
the soaring economic growth, local elites in coastal provinces benefit a flood of revenue under
their control. Without a proportionate share of it, central government becomes weak in face of
local elites from rich provinces. The worsening economic gap between rich coastal and poor
inland provinces also intensifies the relationship between central and local relations.

75

100

After TSS

25

50

75

Before TSS

0

The Accumulated No. of China's BITs

Figure 4-5: Accumulated China’s BITs before and after Tax-Sharing System

1978

1982

1986

1990

1994
Year

1998

2002

2006

2010

Source: UNCTAD

In 1994, the Chinese government launched a series of reform on fiscal decentralization by
putting TSS into practice since 1994. The recentralization policy intends to restore a stronger
central government by bringing a cap on local government’s discretion of revenue and
expenditure. Figure 4-5 presents the trend between accumulated China’s BITs before and after
TSS. The slope changes around the year of 1994 when TSS is put into practice.
Since the practice of TSS rescale the power balance between central and local governments
with regard to the fiscal management, the third hypothesis is derived as follows.
H3: All else equal, the installation of TSS is more likely to decrease the likelihood of
maintenance of effective BITs with China after 1994 than before 1994.
Lastly, I compare the two competing proxies of leadership and institution to decentralization

76

in China. I argue that informal leader’s preferences outperform the institutional effect on
decentralization when the two factors have contrary effect on the number of China’s BITs. Thus,
Deng’s preferences make a better proxy to test the effect of decentralization on the number of
China’s BITs. The fourth hypothesis is set as follows.
H4: All else equal, Deng’s preferences outperform the installation of TSS in changing the
likelihood of maintenance of effective BITs with China.

4.4

Research Design
The research interest in this chapter inspects how the effect of decentralization influences

whether a country signs a BIT with China. The dataset is time-series-cross-sectional structure
where all countries exist in the global system defined and generated by Eugene (Bennett and
Stam 2000). The time spans from 1978 to 2008 when the economic openness policy in China is
initiated.
The dependent variable denotes whether a given country maintains an effective BIT with
China in a given year. Based on the records of International Investment Agreements collected
and updated by UNCTAD, China has signed BITs with 104 countries from 1978 to 2008. Also,
the BITs must be signed and enforced to be recognized as effective. For the dependent variable, I
assign the value as 1 when a given country maintains an effective BIT with China and 0
otherwise. A list of countries maintains effective BITs with China in my sample is in the Table
A-5 in the Appendix.
Given

the

binary

dependent

variable

and

the

TSCS

data

structure,

binary

time-series-cross-sectional (BTSCS) model is an appropriate choice to proceed statistical
analysis (Beck, Katz, and Tucker 1998). With regard to the detection of temporal dependence for

77

repeated observations on the same country, Beck, Katz and Tucker propose that researchers
should include time splines in BTSCS models. In this chapter, the research interest is the
maintenance of effective BITs with China. Once a BIT is signed, it can maintain effective until it
is terminated. Also, a good record of maintaining effective BITs in the past can lead to
continuance of effective BITs in the future. Thus, the value of observations can repeat itself
across time. The significance of history makes interdependent observations. Upon Cartet and
Signorino’ advice (2010), I employ three time polynomials in the BTSCS models to estimate the
maintenance of effective BITs with China for each country.

4.4.1 Independent Variables
The independent variable of research interest is decentralization in China. As previously
elaborated in argument, decentralization in China is hinged on leader’s preferences and the
institutionalization of decentralization. For leader’s preferences, I develop formal and informal
measure on leaders’ preferences on decentralization from Model 1 to Model 3.
For formal measure, general secretary’s decentralization policy is believed to shape the
level of decentralization in China. But, decentralization policy is fixed on general secretary’s
tenure. The level of decentralization changes as the next general secretary’s decentralization
policy changes. There have been four general secretaries in CCP whose tenure covers the time
span from 1978 to 2008. They are Hua from 1976 to 1981, Hu1 from 1982 to 1987, Zhao from
1988 to 1989, Jiang from 1990 to 2002, and Hu2 from 2003 to 2012. I assign the value of 1 in a
given year for a given general secretary in power and 0 otherwise. In Model 1, I evaluate formal
power holders’ tenure as a proxy of decentralization policy on the maintenance of effective BITs
of China. The baseline is set at Hu2.

78

For informal measure, Deng Xiaoping (Deng) has been the master mind behind the back of
general secretaries of CCP since the initiation of economic openness policy in 1978 till his death
in 1997. To capture Deng’s influence on decentralization policy, I generate two measures.
The first measure is designed to capture Deng’s influence on general secretary’s
decentralization policy in Model 2. Since the initiation of economic openness policy in 1978,
Deng remains the mater mind behind general secretary, the highest throne in CCP until he dies in
1997. The full tenures of Hua, Hu1 and Zhao are under Deng’s influence. In addition, Deng
appointed Jiang and Hu as the next two consecutive general secretaries after Zhao. Deng remains
alive and influential in the first half of Jiang’s tenure until his decease in 1997. The second half
of Jiang’s tenure and the full tenure of Hu2 are both free from Deng’s personal guidance. I
generate two dummy variables, Jiang before Deng's decease from 1990 to 1997 and Jiang after
Deng's decease from 1998 to 2002. Therefore, I compute the dummies variables of leaders’
tenure, during which I assign the value of 1 for each leader’s tenure and 0 otherwise. The
baseline is set at Hu2. Again the model dropped Hua because of non-variation in the dependent
variable.
In Model 3, the second measure is a direct measure of Deng’s influence since 1978, the first
year of economic openness policy. I compute two variables. Deng takes the value of 1 from 1978
to 1997 and 0 if not whereas After Deng takes the value of 1 after 1998 and 0 otherwise. The
baseline is thus set at After Deng.
Next, I use the installation of TSS as a proxy for decentralization policy in Model 4. The
TSS is installed since 1994 so that I code two variables. For After TSS, I assign the value of 1
after 1994 and 0 if before. The variable of Before TSS takes the value of 1 from 1978 to 1993 and
0 if not. The base line is set at Before TSS.

79

Lastly, I estimate the two competing proxies of decentralization in Model 5 and Model 6.
The practice of TSS is closely related to Deng’s influence. To control Deng’s influence, I
compute three dummy variables, Deng without TSS from 1978 to 1993, Deng with TSS from
1994 to 1997 and TSS without Deng from 1998 to 2008, taking the value of 1 for the timespan
and 0 otherwise.
In Model 5, the baseline is set at Deng with TSS, representing the mix of institutional and
leadership effects in TSS and Deng’s influence respectively. Based on the differences between
the baseline and the other two variables, the statistical results represent the institutional and
leadership effects. In comparison with Deng with TSS, the result of Deng without TSS, presents
the institutional effect of tax-sharing system whereas the result of TSS without Deng, the estimate
means the leadership effect of Deng’s influence.
Lastly, I compare the effects of Deng’s influence and TSS as proxies to decentralization on
the maintenance of effective BITs in Model 6. As Deng prefers empowering local elites, TSS
shrinks local elites' power. I will test which factors outperforms in influencing the number of
China's BITs. I compute the variable of Deng, taking the value of 1 from 1978 to 1997, the time
span of Deng's lifetime since the economic reform and 0 otherwise. I assign the value of 1 since
the installation of TSS in 1994 and 0 if before.

4.4.2 Control Variables
I include a battery of institutional, economic and political controls influential to the
maintenance of effective BITs with China. For institutional control, scholars find that treaties
have binding effects on a state’s behaviors (Abbott and Snidal 2000; Alter 2001; Dolzer and
Schreuer 2008; Rosendorff and Shin 2012; Simmons 2000, 2014). The competition for capital

80

encourages the diffusion of BITs (Elkins, Guzman, and Simmons 2006). Therefore, the existing
number of BITs in the past influences the probability of signing one more BIT in the future. To
control the institutional effect, I take the log value of the total number of BITs signed with China
from the dataset of International Investment Agreements by Economy under the auspice of
UNCTAD. One-year lag is operated to capture the effect of historical record of pre-existing BITs.
A positive relationship is expected between the number of China’s BITs in the previous year and
the number of concluding a BIT with China.
As for economic controls, the level of economic development and economic interactions
with China are crucial to the likelihood of signing BITs with China. A thriving economy does not
have an urgent need to sign BITs because FDI can become a lesser important source of capital to
host countries. Therefore, investors are willing to take a higher risk of expropriation before BITs
are concluded (Jensen 2008). Also, home countries are usually developed countries with
sufficient capital to shop for targets of investments worldwide. Based on the Penn World Table
(PWT, v8.0), I adopt the variable of GDP per capita for China and its BIT partner (Feenstra,
Inklaar, and Timmer 2013).
As for the economic interactions, singing investment treaties is arguably related to the
volume of trade (Büthe and Milner 2014; Simmons 2014; J. L. Tobin and Busch 2010). A higher
volume of trade induces more urge for FDI. Consequently, more BITs are needed to protect FDI.
I take the log value of each country’s trade volume with China from the dataset of international
trade 1870-2009 (v3.0) under the Correlates of War Project (Barbieri, Keshk, and Pollins 2009).
The volume of trade with China is positively related to the maintenance of effective BITs with
China.
Political controls cover the regime type of China’s BIT partner, affinity in international

81

politics, and China’s grand strategy in geographical politics. First of all, democracies are more
likely to engage in international treaties as a form of international cooperation due to the legalist
norm and domestic audience costs (Jensen 2003; 2008). Thus, I incorporate polity2 score of
China’s BIT partner (Marshall and Jaggers 2002). For polity2 ranging from -10 (most
authoritarian) to 10 (most democratic), the expected sign for the coefficient is positive.
Next, political and economic interactions are highly correlated. Political allies leads to more
economic cooperation. United Nation (UN) is an international forum with the most
comprehensive state membership and the variety of global issues. Voting in general assembly of
UN is deemed as publicly revealed state preferences in international politics. Based on the voting
records, Gartzke (2006) generates the affinity index to measure the ideological distance between
dyads of states. Given the research interest in this chapter, I retain the affinity index between
China and the rest of the UN members in the world. For the variable, the values are on an
interval scale from –1 to 1 between China and all other UN members from 1946 to 2002. The
relationship between affinity index and maintenance of effective BITs is expected to be positive.
Geographic proximity is arguably crucial to signing BITs due to the pressure of competing
for global capital with neighboring countries (Elkins, Guzman, and Simmons 2006). In addition,
China has been an active investor in the continents of Africa, Asia and Latin America for the last
decade (Cai 1999; Chan and Wong 2003). African countries are of abundant natural resources
whereas Latin American countries are backyard of US. Either continent is of great significance to
China’s strategy to expand its influence via BITs. To control the political and economic
significance embedded in geography in the models, I follow the MARC Code List for
Geographic Areas, provided by the Library of Congress to categorize three dummy variables for
Asia, Africa and Latin America. Among all three continents, China itself is situated in Asia where

82

countries are most likely to maintain effective BITs with China for geographical and strategic
reasons. Table A-6 in the Appendix reports the summary statistics of variables used in the
empirical analysis of this chapter.

4.5

Results
In this chapter, I test four hypotheses of how decentralization influences the maintenance of

effective BITs with China with two proxies of leadership and institution in six models. All
models cover the same set of control variables including China's total BITs in previous year,
GDP per capita for China, GDP per capita for other states, log of trade with China for other states,
polity score for other states, affinity index, and three regional dummies (Africa, Asia and Latin
America).
The first line of argument utilizes leaders’ preference as a proxy to decentralization policy
and tests the effect of decentralization on the maintenance of effective BITs with China in H1
and H2 from Model 1 to Model 3 with the baseline of Hu2. The statistical results are presented in
Table 4-1. The influence of leadership is measured by formal and informal leadership. The
formal leadership is measured by the successive leaderships in the level of general secretary of
CCP. From 1978 to 2008, there are five leaders, including Hua, Hu1, Zhao, Jiang and Hu2. But,
the observations for Hu1 are dropped in all models because since China signed its first BITs in
1982. In other words, the dependent variable shows no variation in Hu1.
In Model 1, the coefficients for Hu1, Zhao, and Jiang are not significant, compared with
Hu2. The empirical results do not support H1 that general secretaries’ rule increases the
likelihood of maintenance of effective BITs with China since 1978, given all else equal.

83

Table 4-1: Effects of Leadership and Institution on China’s BITs
DV: China’s BITs
Baseline
IV1: Leadership
Hu1
Zhao
Jiang

Leadership
General
Deng’s Influence
Model 1
Model 2
Model 3
Hu2
After Deng
-0.015
[0.740]
-0.006
[0.756]
0.279
[0.605]

Jiang before Deng's decease

2.972**
[1.267]
3.016**
[1.275]
2.896***
[1.104]
0.726
[0.833]

Jiang after Deng's decease
Before Deng
IV2: Institution
TSS
Controls:
China's total BITs in t-1
GDPpc: China
GDPpc: others
ln(Trade with China): others
Polity2: other states
Affinity index
Africa
Asia
Latin America

Institution
TSS
Model 4
Before TSS

2.146***
[0.631]
0.055
[0.592]
0.629***
[0.222]
-4.047***
[0.624]
0.166
[0.113]
0.083
[0.061]
0.008
[0.021]
-0.619
[0.708]
-0.680
[0.444]
0.683**
[0.266]
-0.115
[0.387]

0.193
[0.237]
-1.847**
[0.826]
0.181
[0.113]
0.095
[0.062]
0.011
[0.021]
-0.494
[0.716]
-0.698
[0.435]
0.664**
[0.272]
-0.157
[0.391]

0.270
[0.216]
-2.266***
[0.785]
0.178
[0.112]
0.093
[0.061]
0.011
[0.021]
-0.510
[0.711]
-0.698
[0.434]
0.665**
[0.272]
-0.154
[0.389]

0.728***
[0.221]
-4.179***
[1.008]
0.162
[0.114]
0.082
[0.062]
0.009
[0.021]
-0.632
[0.719]
-0.677
[0.445]
0.690***
[0.266]
-0.112
[0.388]

N
3598
Countries
152
Wald chi2
119.72
122.07
119.05
116.92
Log pseudolikelihood
-343.690
-335.775
-336.199
-344.030
Pseudo R2
0.121
0.142
0.141
0.120
Note: Clustered standard errors in brackets. Three time polynomials and constant not shown.
Hua dropped for few obs. in all models. *<0.1, **<0.05, ***<.0.01.

84

For Model 2 and Model 3, I introduce the measure of Deng’s influence as Deng is an
informal leader in China during his life span. In Model 2, Deng influences the rule of three
general secretaries, Hu1, Zhao and Jiang. Jiang’s rule is further divided into two parts: Jiang’s
rule before and after Deng’s decease. With the reference of Hu2, Hu1, Zhao and Jiang before
Deng’s decease present positive coefficients at p<0.05 level or better. The coefficient of Jiang
after Deng’s decease is not significant. Deng’s preference for a decentralized China greatly
influences formal leaders’ decentralization policy. Thus, a decentralized facilitates the
maintenance of BITs with China.
In Model 3, I use a direct measure of Deng’s influence and set the baseline of After Deng.
Compared with After Deng, the coefficient of Before Deng is positively related to the
maintenance of effective BITs with China at p <0.01 level. Based on the statistical results in
Model 2 and Model 3, H2 is confirmed that Deng’s informal rule increases the likelihood of
maintenance of effective BITs with China from before his decease in 1997. Deng as an informal
leader in China is crucial to the decentralization policy in China. His preference for empowering
local elites consolidates the maintenance of China’s BITs.
Another proxy of decentralization is the installation of TSS since 1994 in H3. The
installation of TSS is controlled by Deng’s influence in Model 4. In comparison with baseline
Before TSS, the coefficient of After TSS is not significant. The results in Model 4 does not
support H3 that the installation of TSS is more likely to decreases the likelihood of maintenance
of effective BITs with China after 1994 than before. The existence of counter decentralization
does not significantly decrease the number of China’s BITs.
As for the control variables, only two variables present significant results with the predicted
direction of effects. GDP per capita for China is negatively related to the maintenance of

85

effective BITs with China at the p<0.05 level or better. Also, the regional dummy in Asia shows a
positive coefficient at the p<0.05 level or better. Asian countries and China are mostly likely to
maintain effective BITs with each other because they are deeply engaged with political and
economic interests derived from geography.
In Table 4-2, I present the results for the competing proxies of leadership and institution. I
test which line of argument empirically outperforms the other via the two types of proxies. The
baseline is Deng with TSS, the mixed effect of leadership and institution in Model 5. In
comparison with Deng with TSS, Deng without TSS is not significant. But, TSS without Deng
maintains a negative sign at the p<0.01 level. Without Deng, TSS is detrimental to the
maintenance of effective BITs with China. When informal leader’s preferences on
decentralization disappear, institutional effect of TSS starts to shrink the power of local elites so
that the number of China's BIT decreases.
I incorporate separate variables of Deng and TSS in Model 6 to directly compare the effect
of leadership and institution as proxies to decentralization policy. Deng presents a positive sign
at p <0.01 level whereas coefficient of TSS is not significant. Deng increases the probability of
maintaining effective BITs with China. Deng’s influence is the decisive factor to empower local
elites who strive to sign more BITs in competition for footloose capital in global market. Based
on the findings in Model 5 and Model 6, I conclude that Deng’s influence outperforms TSS when
the two factors are in comparison. After Deng’s decease, TSS resumes the function to shrink
local elite’s power so that the number of China's BITs decreases. The empirical evidence
supports H4 that all else equal, Deng’s preferences outperform the installation of TSS in
changing the likelihood of maintenance of effective BITs with China.

86

Table 4-2: Competing Proxies of Decentralization on China’s BITs
DV: China’s BITs
Baseline
IV: Leadership and Institution
Deng without TSS

Competing Proxies
Model 5
Model 6
Deng with TSS
0.393
[0.478]
-2.304***
[0.700]

TSS without Deng
TSS

-0.393
[0.478]
2.304***
[0.700]

Deng
Controls:
China's total BITs in t-1 year

0.192
[0.218]
-1.684
[1.074]
0.175
[0.112]
0.091
[0.062]
0.011
[0.021]
-0.525
[0.711]
-0.698
[0.433]
0.665**
[0.272]
-0.152
[0.387]

GDPpc: China
GDPpc: other countries
ln(Trade with China): partner
Polity2: partner
Affinity index
Africa
Asia
Latin America

0.192
[0.218]
-1.684
[1.074]
0.175
[0.112]
0.091
[0.062]
0.011
[0.021]
-0.525
[0.711]
-0.698
[0.433]
0.665**
[0.272]
-0.152
[0.387]

N
3598
Countries
152
Wald chi2
121.24
121.24
Log pseudolikelihood
-335.836
-335.830
Pseudo R2
0.141
0.141
Note: Clustered standard errors in brackets. Three time polynomials and constant not shown.
*<0.1, **<0.05, ***<0.01.

As for the control variables in Table 4-2, only one variable shows significant result with the
predicted direction of effects. The regional dummy in Asia shows a positive coefficient at the
87

p<0.05 level or better. Asian countries are mostly likely to maintain effective BITs with China for
strategic concerns over political and economic interests.
To sum up the findings of this chapter, Table 4-2 confirms H2, but not H1 and H3. Deng’s
preference to decentralization is consequential to the probability of maintenance of effective
China’s BITs. Table 2 settles the competing proxies of leadership and institutions as leader’s
preference makes a more reliable indicator to the decentralization policy than the institution of
decentralization in China. Without leader’s vouch, the sheer existence of TSS does not function
as expected. A decentralized China is beneficial to the development and consolidation of BITs
policy. As a proxy to decentralization policy, the influence of Deng as an informal leader during
his lifetime outperforms the effect of TSS. Once Deng passed away, TSS resumes the function of
limiting local elites’ power and decreases the number of China’s BITs.

4.6

Conclusion
To defend a peacefully rising China, the adoption of international institutions remains a

debate as a tool to regulate international cooperation consequential to the stability of world order
(Axelrod and Keohane 1985; Ikenberry 2008; Kent 2007; Martin and Simmons 1998; Simmons
2000; Simmons, Dobbin, and Garrett 2008). With China’s aggressive engagement in
globalization, it becomes an important case challenging the literature of authoritarian rule and
international institutions and begging a reexamination of domestic factors in international
cooperation.
In this chapter, I use the case of China’s BITs to investigate the domestic source of China’s
BITs. To be specific, I look into how as proxies to decentralization, leadership and institution
influences China’s BITs. I argue that leader’s preferences about decentralization influence the de
88

facto practice of recentralization producing local elites in signing BITs.
The empirical findings from 1978 to 2008 suggest that Deng’s preference for
decentralization outperforms the installation of TSS to maintain China’s BITs. With leader’s
influence, TSS does not shrink local elite’s power as expected. Once leader’s influence decreases,
TSS decreases the number of China’s BITs as local elites are less empowered.
In this chapter, I make a case of decentralization as a source to China’s BITs behaviors. The
findings are expected to shed light on the domestic source of international cooperation for
authoritarian regimes in two ways. First, I emphasize the drive of leadership behind the domestic
institutions crucial to international behaviors under dictatorship. Previous literature uses the
measure of political, fiscal and administrative decentralization to debate about the degree of
decentralization in China (Montinola, Qian, and Weingast 1995; Rodden and Rose-Ackerman
1997; Treisman 2007). However, the measures of de jure regulations and the dynamic of
leadership has been recognized as an important ingredient of Chinese politics as informal
influences can sustain or hollow out the functions of institutions.
Second, as the competing proxies to decentralization, the dynamics between leadership and
institutions in China provide a piece of evidence in the debate in the literature of authoritarian
regimes. If institutions reflect temporal equilibrium of leadership dynamics (Brownlee 2007; Lai
and Slater 2006; Weeks 2012), the same institution functions differently as the dynamics of
leadership change across time. If institutions are tools of power-sharing commitment to
overcome the time-inconsistency issue (Boix and Svolik 2013; Svolik 2012b), institutions remain
effective regardless of leadership change. The case study of China supports the former rather
than the latter. The findings suggest that the effect of TSS is conditional on Deng’s influence
whereas Deng’s influence can exert independent effect without the presence of TSS.
89

CHAPTER 5
CONCLUSION

Some dictators install decentralization to share power with local elites who can provide
efficient and effective governance beneficial to the duration and quality of authoritarian rule.
Local elites are well-informed of local affairs and trusted by local people so that dictators have a
hard time rotating or even replacing them because their resources in governance are
geographically grounded. In managing local elites under authoritarian regime without an ultimate
authority to define, interpret and settle disputes between central and local governments, dictators
encounter two principal-agent problems: defection and compliance.
To prevent the problem of defection, dictators have to monopolize local elites’ loyalty to
prolong the political survival of authoritarian regime. They should block any available outside
option in the party system to prevent local elites’ defection as only a few of them can obtain the
limited top positions within the ruling party. Thus, decentralization benefits dictator’s political
survival on the condition of single-party system.
On the other hand, the problem of compliance deals with if local elites comply with central
policies. For instance, BITs are for central governments to sign with foreign countries, but for
local governments to comply with the terms to which central government commits. Foreign
countries are less willing to engage with decentralized dictatorships without local elections
because electorally minded local elites prioritize local voters’ preferences over dictators’.
Ironically, decentralization facilitates the signing of BITs in China as empowered local elites are
motivated to comply with central policy of engaging in international institutions.
I argue that local elites are influential to dictators’ duration and quality of governance
because local elites are capable of leveraging their political support from the mass to
90

increase/dampen dictators’ interests. The two problems are derived from the lack of ultimate
judicial authority to settle disputes between the central and local governments under dictatorship.
The management of two problems influences regime resilience and international cooperation
respectively for dictatorship. To further elaborate on the case of China, I identify the informal
leader’s preferences trump institutional effects in decentralization consequential to the number of
China’s BITs. I use statistical methods to present empirical evidence supportive of my general
arguments.
This dissertation intends to make the following contributions to the literature of authoritarian
resilience and democratization. First, I intend to further the understandings in the political
institutions under dictatorship with a focus on decentralization, an important and yet
understudied institution under the framework of dictator’s power-sharing with elites.
Conventional approach investigates the effect of individual political institution on In terms of
dictator’s political survival. But, this dissertation takes a unique approach in examining the
interactions of political institutions.
In Chapter 2, I investigate the interconnected power-sharing institutions in party system and
decentralization to shed light on the relationship between the installation of power-sharing
institutions and democratization. Since power-sharing institutions facilitate dictator’s political
survival as current literature suggests, the greatest puzzle is that dictatorships with all the
power-sharing institutions are actually on the way to democratization, which ironically shortens
dictator’s political survival.
The investigation of political institutions with democratic façade under dictatorships is an
on-going literature with great potential to unravel the sophisticated nuances between
democracies and dictatorships. The more we understand the functions of political institutions
91

under dictatorships, the closer we approach the answer of a long-time puzzle in political science
from an institutional perspective: why and how democratization takes place and why not.
Second, I reexamine the domestic determinant in international cooperation for dictatorships.
As dictatorships are ever more engaged in investment treaty as a form of international
cooperation, it is important to link the literatures of authoritarian institutions and international
cooperation. In Chapter 3 and 4, I investigate the decentralization and authoritarian credibility.
The existence of empowered local elites is beneficial to the signing of BITs in the case of China,
but not the maintenance of BITs in dictatorships.
In Chapter 3, political decentralization breeds a group of electorally-minded local elites who
over-commit to international investors in bidding limited inflow of foreign direct investment, but
end up violating BITs. The opportunistic behaviors in the maintenance of credibility lead to
greater likelihood of being sued before international forum. It is under the context of
authoritarian regime that the conflicts in the interests between central and local governments lack
an ultimate resolution in domestic politics. The unresolved domestic disputes spill over to
international cooperation.
In Chapter 4, the change and continuity of decentralization is empirically attributed to
informal leader’s preferences in China. Empirical tests confirm that Deng’s preferences in
empowering local elites lead to the sustainability of economically engaging foreign countries via
BITs. In terms of policy consequences, the effect of TSS is dependent on Deng’s influence while
that of Deng’s influence is independent from TSS. As China is deeply and widely involved with
globalization, the understanding about its domestic source of credibility in authoritarian regimes
prescribes the way how democracies evaluate the hazard which may backfire on economic
voting.
92

However, this dissertation falls short to provide a fine-grained study in several regards. First,
this dissertation takes the institution of decentralization as it is and discusses the policy
consequences of it. However, the origin of decentralization matters to the policy consequence.
For instance, Latin American countries have legacies of feudalism where local elites have taken
control of resources in specific regions for generations even before modern states take shape.
In authoritarian Taiwan, some local elites are generated by superimposed local elections. The
political costs of removing the former is way higher than the latter as the former has fostered a
strong and informal coalition across local elites to threaten any type of rule, but not the latter.
Without tracing down the origin of decentralization, it is biased to equate the weights of political
threat local elites can impose on the duration and quality of authoritarian rule under the same
institution of decentralization.
Second, the origin of local elites is left out of discussion with regard to policy consequences.
Influential local elites can be a representative or an agent of ethnic identity, religious orientation,
or kinship. For the former, the shared privilege or grievance of political-economic identity is of
greater potential to mobilize the collective action among their “own people”. For the latter, the
lack of shared identity sets a higher bar for collective action against the central government.
Given the constraints, future research can be extended to a fine-grained distinction of
whether the existence of local elites precedes the installation of decentralization. On the account
of informal and formal institutionalization of local elites, we can further examine to what extent
local elites can credibly impose a threat of overthrowing authoritarian rule. How do the dynamics
of informal institutionalization of local elites influence the formal institutions of decentralization?
What makes some dictatorships establish or cancel the institution of decentralization, but not
others?
93

Moreover, the origin of local elites can be regarded as a result of bargaining between central
and local government. When local elites are outsiders assigned by central government to join
local elections, do they outperform in sustaining authoritarian rule? When local elites are insiders
of local residence, are they more likely to answer local needs even if they have to go against
central policy? The extended studies are expected to enrich the understanding of how
authoritarian regime functions as the distribution of political authority is electorally connected.

94

APPENDIX

95

Table A-1: List of Dictatorships in the Sample of Chapter 2
Albania

Laos

Algeria

Madagascar

Angola

Malaysia

Armenia

Mexico

Azerbaijan

Mozambique

Benin

Nigeria

Botswana

North Korea

Brazil

Panama

Bulgaria

Philippines

Burundi

Poland

Cambodia

Romania

China

Senegal

Republic of Congo

Yugoslavia (Serbia and Montenegro)

Cote d'Ivoire

South Africa

Egypt

Spain

Ethiopia

Sri Lanka

Gambia

Tajikistan

Ghana

Togo

Guinea-Bissau

Tunisia

Iran

Venezuela

Kyrgyzstan

96

Table A-2: Summary Statistics of Chapter 2
Variables

N

Mean

Std. Dev.

Min

Max

1976

2007

Year

665

Regime failure (DV)

665

0.04

0.19

0

1

Political decentralization

665

1.29

0.45

1

2

Party system

665

1.55

0.50

1

2

(Low, Single)

665

0.27

0.44

0

1

(High, Single)

665

0.18

0.38

0

1

(Low, Dominant)

665

0.44

0.50

0

1

(High, Dominant)

665

0.11

0.31

0

1

Regime transitions

665

0.24

0.51

0

2

Military spending

654

12.99

2.10

6.91

18.28

Natural resources

645

4.05

2.53

0

7.67

GDP per capita

645

3.86

2.67

0.42

13.67

GDP per capita (squared)

645

22.04

26.51

0.17

186.85

Democracies in the region

645

16.47

17.36

0

95.45

Foreign direct investment

523

18.60

2.85

0

25.86

97

Table A-3: List of Dictatorship in the Sample of Chapter 3
Albania
Algeria

Ghana
Guinea

Philippines
Poland

Angola

Haiti

South Korea

Armenia

Hungary

Romania

Azerbaijan

Indonesia

Russia

Bangladesh

Iran

Rwanda

Belarus

Iraq

Saudi Arabia

Benin

Jordan

Senegal

Botswana

Kuwait

Serbia

Bulgaria

Kyrgyzstan

Sierra Leone

Burundi

Laos

Singapore

Cameroon

Lesotho

South Africa

Central African Republic

Libya

Sri Lanka

Chad

Madagascar

Sudan

China

Malaysia

Swaziland

Congo

Mali

Syria

Cote d'Ivoire

Mexico

Tajikistan

Cuba

Mongolia

Tanzania

Democratic Rep. of Congo

Morocco

Thailand

North Korea

Mozambique

Togo

Ecuador

Nepal

Tunisia

Egypt

Nigeria

Turkey

El Salvador

Oman

Turkmenistan

Eritrea

Pakistan

Uganda

Ethiopia

Panama

United Arab Emirates

Gabon

Paraguay

Venezuela

Gambia

Peru

Zimbabwe

98

Table A-4: Summary Statistics of Chapter 3
Variable
Year

N
1312

Mean

Std. Dev.

Min
1975

Max
2010

Non-democracy

1312

1

0

1

1

ICSID cases

1312

0.03

0.20

0

3

Political decentralization

1312

0.46

0.50

0

1

Domestic political shocks

1203

1.85

4.36

0

49

Intra-state conflicts

1312

0.14

0.37

0

2

Domestic economic shocks

1312

0.07

0.26

0

1

Inflation

1089

2.20

1.32

-3.11

10.19

GDP per capita

1245

6.98

1.34

4.68

10.91

Property rights

795

13.91

3.93

3

26.5

Financial openness

1202

0.36

0.33

0

1

Military dictatorship

1312

0.10

0.30

0

1

Monarchy dictatorship

1312

0.16

0.36

0

1

Personal dictatorship

1312

0.27

0.44

0

1

Inter-state conflicts

1312

0.66

1.05

0

7

Inter-state conflicts (region)

1312

5.38

3.92

0

23

Inter-state conflicts (world)

1312

25.22

17.64

0

86

Economic shocks (world)

1312

4.70

5.35

0

24

FDI inflows

1078

18.40

2.62

2.37

25.95

Foreign reserves

1215

20.48

2.37

14.53

28.31

Remittances

906

-0.22

2.17

-7.74

4.67

BITs signed

1312

0.82

1.59

0

15

BITs signed (region)

1312

154.15

122.35

0

796

ICSID cases (region)

1312

9.89

8.19

0

42

99

Table A-5: List of Countries Maintaining Effective BITs with China
Country
Albania
Algeria
Argentina
Armenia
Australia
Austria
Azerbaijan
Bahrain
Bangladesh
Barbados
Belarus
Belgium
Bolivia
Bosnia and Herzegovina
Bulgaria
Cambodia
Canada
Cape Verde
Chile
Colombia
Croatia
Cuba
Cyprus
Czech
Democratic People's Republic of Korea
Denmark
Ecuador
Egypt
Estonia
Ethiopia
Finland

Status of BITs with China
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force (Terminated)
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force (Terminated)
In force
In force
In force
In force
In force
In force
In force (Terminated)

100

Start Year End Year
1993
1996
1992
1992
1988
1985
1994
1999
1996
1998
1993
1984
2009
1992
2002
1989
1996
2012
1998
1994
2008
1993
1995
2001
1991
2006
2005
1985
1994
1994
1993
1998
1984
2006
（Continuing）

Table A-5: (cont’d)
Country
France
Gabon
Georgia
Germany
Ghana
Greece
Guyana
Hungary
Iceland
India
Indonesia
Iran
Israel
Italy
Jamaica
Japan
Kazakhstan
Kuwait
Kyrgyzstan
Laos
Latvia
Lebanon
Lithuania
Luxembourg
Macedonia
Madagascar
Malaysia
Mali
Malta
Mauritius
Mexico
Moldova
Mongolia

Status of BITs with China
In force (Terminated)
In force
In force
In force (Terminated)
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force
In force (Terminated)
In force
In force
In force
In force
In force
In force
In force
In force
In force

101

Start Year
1984
1997
1993
1983
1989
1992
2003
1991
1994
2006
1994
2000
1995
1985
1994
1988
1992
1985
1992
1993
2004
1996
1993
1984
1997
2005
1988
2009
2009
1996
2008
1992
1991

End Year
2010
2009
（Continuing）

Table A-5: (cont’d)
Country
Morocco
Mozambique
Myanmar
Netherlands
New Zealand
Norway
Oman
Pakistan
Papua New Guinea
Peru
Philippines
Poland
Portugal
Qatar
Republic of Korea
Romania
Russia
Saudi Arabia
Serbia
Singapore
Slovakia
Slovenia
South Africa
Spain
Sri Lanka
Sudan
Sweden
Switzerland
Syria
Tajikistan
Thailand
Trinidad and Tobago
Tunisia

Status of BITs with China
In force
In force
In force
In force (Terminated)
In force
In force
In force
In force
In force
In force
In force
In force
In force (Terminated)
In force (Terminated)
In force (Terminated)
In force
In force (Terminated)
In force
In force
In force
In force
In force
In force
In force (Terminated)
In force
In force
In force
In force (Terminated)
In force
In force
In force
In force
In force

102

Start Year
1995
2001
2001
1985
1988
1984
1995
1989
1991
1994
1992
1988
1992
1998
1992
1994
1990
1996
1995
1985
1991
1993
1997
1992
1986
1997
1982
1986
1996
1993
1985
2002
2004

End Year
2004
2008
2000
2007
2009
2008
2010
（Continuing）

Table A-5: (cont’d)
Country
Turkey
Turkmenistan
Ukraine
United Arab Emirates
United Kingdom
Uruguay
Uzbekistan
Viet Nam
Yemen
Zimbabwe

Status of BITs with China
In force
In force
In force
In force
In force
In force
In force (Terminated)
In force
In force
In force

Source: UNCTAD

103

Start Year
1990
1992
1992
1993
1986
1993
1992
1992
1998
1996

End Year
2011
-

Table A-6: Summary Statistics of Chapter 4
Variable
Year

N
3598

Mean

Std.

Min
1983

Max
2008

Signing BITs with China

3598

0.023

0.149

0

1

Hu1 ('82-'87)

3598

0.133

0.339

0

1

Zhao ('88-'89)

3598

0.237

0.425

0

1

Jiang ('90-'02)

3598

0.519

0.500

0

1

Hu2 ('03-'12)

3598

0.244

0.430

0

1

Jiang while Deng's alive ('89-'97)

3598

0.314

0.464

0

1

Jiang after Deng's death ('98-'02)

3598

0.205

0.404

0

1

Before Deng's decease ('78-'97)

3598

0.551

0.498

0

1

After Deng's decease ('98-)

3598

0.449

0.498

0

1

Before TSS ('78-'94)

3598

0.614

0.487

0

1

After TSS ('95-'08)

3598

0.386

0.487

0

1

Deng without TSS ('78-'93)

3598

0.386

0.487

0

1

Deng with TSS ('94-'97)

3598

0.165

0.371

0

1

TSS without Deng ('98-)

3598

0.449

0.498

0

1

China's total BITs in t-1 year

3598

3.768

1.110

0

4.625

GDPpc: China

3598

7.647

0.603

6.524

8.702

GDPpc: partner

3598

8.375

1.336

5.081

11.607

ln(Trade with China): partner

3598

4.954

2.709

0

13.096

Polity2: partner

3598

2.190

7.115

-10

10

Affinity score

3598

0.712

0.250

-0.724

1

Africa

3598

0.319

0.466

0

1

Asia

3598

0.251

0.433

0

1

Latin America

3598

0.123

0.328

0

1

104

BIBLIOGRAPHY

105

BIBLIOGRAPHY

Abbott, Kenneth W., and Duncan Snidal. 2000. “Hard and Soft Law in International Governance.”
International Organization 54(3): 421–56.
Aizenman, Joshua, and Reuven Glick. 2008. “Pegged Exchange Rate Regimes: A Trap?” Journal
of Money, Credit and Banking 40(4): 817–35.
Albertus, Michael, and Victor Menaldo. 2012. “Dictators as Founding Fathers? The Role of
Constitutions Under Autocracy.” Economics & Politics 24(3): 279–306.
Albrecht, Holger. 2005. “How Can Opposition Support Authoritarianism? Lessons from Egypt.”
Democratization 12(3): 378–97.
Allee, Todd, and Clint Peinhardt. 2011. “Contingent Credibility: The Impact of Investment
Treaty Violations on Foreign Direct Investment.” International Organization 65(3):
401–32.
———. 2014. “Evaluating Three Explanations for the Design of Bilateral Investment Treaties.”
World Politics 66(1): 47–87.
Alter, Karen J. 2001. Establishing the Supremacy of European Law: The Making of an
International Rule of Law in Europe. Oxford: Oxford University Press.
Ames, Barry. 1970. “Bases of Support for Mexico’s Dominant Party.” The American Political
Science Review 64(1): 153–67.
Archer, Candace C., Glen Biglaiser, and Karl DeRouen Jr. 2007. “Sovereign Bonds and the
‘Democratic Advantage’: Does Regime Type Affect Credit Rating Agency Ratings in the
Developing World?.” International Organization 61(2): 341–65.
Axelrod, Robert, and Robert O. Keohane. 1985. “Achieving Cooperation under Anarchy:
Strategies and Institutions.” World Politics 38(1): 226–54.
Bachman, David. 1986. “Differing Visions of China’s Post-Mao Economy: The Ideas of Chen
Yun, Deng Xiaoping, and Zhao Ziyang.” Asian Survey 26(3): 292–321.
Banks, Arthur, and Kenneth A. Wilson. “Cross-National Time-Series Data Archive.”
http://www.databanksinternational.com/71.html (March 21, 2015).
Barbieri, Katherine, Omar M. G. Keshk, and Brian M. Pollins. 2009. “Trading Data Evaluating
Our Assumptions and Coding Rules.” Conflict Management and Peace Science 26(5):
471–91.

106

Baum, Richard. 1996. Burying Mao: Chinese Politics in the Age of Deng Xiaoping. Princeton
University Press.
Beck, Nathaniel, Jonathan N. Katz, and Richard Tucker. 1998. “Taking Time Seriously:
Time-Series-Cross-Section Analysis with a Binary Dependent Variable.” American
Journal of Political Science 42(4): 1260–88.
Beck, Thorsten et al. 2001. “New Tools in Comparative Political Economy: The Database of
Political Institutions.” The World Bank Economic Review 15(1): 165–76.
Bennett, D. Scott, and Allan C. Stam. 2000. “Eugene: A Conceptual Manual.” International
Interactions 26(2): 179–204.
Blanchard, Olivier, and Andrei Shleifer. 2000. Federalism with and without Political
Centralization: China versus Russia. National Bureau of Economic Research. Working
Paper.
Blanton, Shannon Lindsey, and Robert Blanton. 2007. “What Attracts Foreign Investors? An
Examination of Human Rights and Foreign Direct Investment.” Journal of Politics 69(1):
143–55.
Blaydes, Lisa. 2008. “Authoritarian Elections and Elite Management: Theory and Evidence from
Egypt.”
———. 2011. Elections and Distributive Politics in Mubarak’s Egypt. New York: Cambridge
University Press.
Boix, Carles, and Milan Svolik. 2011. “The Foundations of Limited Authoritarian Government:
Institutions and Power-Sharing in Dictatorships.”
———. 2013. “The Foundations of Limited Authoritarian Government: Institutions,
Commitment, and Power-Sharing in Dictatorships.” The Journal of Politics 75(2):
300–316.
Borensztein, E., J. De Gregorio, and J-W. Lee. 1998. “How Does Foreign Direct Investment
Affect Economic Growth?” Journal of International Economics 45(1): 115–35.
Brancati, Dawn. 2014. “Democratic Authoritarianism: Origins and Effects.” Annual Review of
Political Science 17(1): 313–26.
Brandt, Patrick T., John T. Williams, Benjamin O. Fordham, and Brain Pollins. 2000. “Dynamic
Modeling for Persistent Event-Count Time Series.” American Journal of Political
Science 44(4): 823–43.
Brownlee, Jason. 2007. Authoritarianism in an Age of Democratization. New York: Cambridge
University Press.
Bueno de Mesquita, Bruce, Alastair Smith, Randolph M Siverson, and James D. Morrow. 2003a.
107

The Logic of Political Survival. Cambridge, Mass: MIT Press.
Bueno de Mesquita, Bruce, Alastair Smith, Randolph Siverson, and James D. Morrow. 2003b.
The Logic of Political Survival. Cambridge, Mass: MIT Press.
Büthe, Tim, and Helen V. Milner. 2008. “The Politics of Foreign Direct Investment into
Developing Countries: Increasing FDI through International Trade Agreements?”
American Journal of Political Science 52(4): 741–62.
———. 2014. “Foreign Direct Investment and Institutional Diversity in Trade Agreements:
Credibility, Commitment, and Economic Flows in the Developing World, 1971–2007.”
World Politics 66(1): 88–122.
Cai, Hongbin, and Daniel Treisman. 2004. “State Corroding Federalism.” Journal of Public
Economics 88(3-4): 819–43.
———. 2005. “Does Competition for Capital Discipline Governments? Decentralization,
Globalization, and Public Policy.” The American Economic Review 95(3): 817–30.
———. 2006. “Did Government Decentralization Cause China’s Economic Miracle?” World
Politics 58(4): 505–35.
Cai, Kevin G. 1999. “Outward Foreign Direct Investment: A Novel Dimension of China’s
Integration into the Regional and Global Economy.” The China Quarterly (160): 856–80.
Calingaert, Daniel. 2006. “Election Rigging and How to Fight It.” Journal of Democracy 17(3):
138–51.
Cameron, A. Colin, and Pravin K. Trivedi. 1986. “Econometric Models Based on Count Data.
Comparisons and Applications of Some Estimators and Tests.” Journal of Applied
Econometrics 1(1): 29–53.
Cameron, Adrian Colin, and P. K. Trivedi. 1998. Regression Analysis of Count Data. New York:
Cambridge University Press.
Carter, David B., and Curtis S. Signorino. 2010. “Back to the Future: Modeling Time
Dependence in Binary Data.” Political Analysis 18(3): 271–92.
Chan, Sarah, and John Wong. 2003. “China’s Outward Direct Investment: Expanding
Worldwide.” China: An International Journal 1(2): 273–301.
Cheibub, Jose Antonio, Jenifer Gandhi, and James Raymond Vreeland. 2009. “Democray and
Dictatorship Revisited.” Public Choice 143(1-2): 67–101.
Diamond, Larry Jay. 2002. “Thinking About Hybrid Regimes.” Journal of Democracy 13(2):
21–35.
Diaz-Cayeros, Alberto, Beatriz Magaloni, and Barry R Weingast. 2007. “Tragic Brilliance:
108

Equilibrium Party Hegemony in Mexico.”
Dolzer, Rudolf, and Christoph Schreuer. 2008. Principles of International Investment Law.
Oxford: Oxford University Press.
Egger, Peter, and Michael Pfaffermayr. 2004. “The Impact of Bilateral Investment Treaties on
Foreign Direct Investment.” Journal of Comparative Economics 32(4): 788–804.
Elkins, Zachary, Andrew T. Guzman, and Beth A. Simmons. 2006. “Competing for Capital: The
Diffusion of Bilateral Investment Treaties, 1960-2000.” International Organization 60(4):
811–46.
Falleti, Tulia Gabriela. 2010. Decentralization and Subnational Politics in Latin America. New
York: Cambridge University Press.
Fang, Songying, and Erica Owen. 2011. “International Institutions and Credible Commitment of
Non-Democracies.” The Review of International Organizations 6(2): 141–62.
Feenstra, Robert C., Robert Inklaar, and Marcel P. Timmer. 2013. “The Next Generation of the
Penn World Table.” www.ggdc.net/pwt.
Fontana, Dorothy Grouse. 1982. “Background to the Fall of Hua Guofeng.” Asian Survey 22(3):
237–60.
Franck, Susan D. 2010. “ICSID Effect - Considering Potential Variations in Arbitration Awards,
The.” Virginia Journal of International Law 51: 825.
Freedman, Amy L. 2000. Political Participation and Ethnic Minorities: Chinese Overseas in
Malaysia, Indonesia, and the United States. New York: Routledge.
Gandhi, Jennifer. 2008. Political Institutions Under Dictatorship. New York: Cambridge
University Press.
Gandhi, Jennifer, and Ellen Lust-Okar. 2009a. “Elections Under Authoritarianism.” Annual
Review of Political Science 12(1): 403–22.
———. 2009b. “Elections under Authoritarianism.” Annual Review of Political Science 12:
403–22.
Gandhi, Jennifer, and Adam Przeworski. 2006. “Cooperation, Cooptation, and Rebellion under
Dictatorships.” Economics & Politics 18(1): 1–26.
Garriga, Ana Carolina. 2009. “Regime Type and Bilateral Treaty Formalization.” Journal of
Conflict Resolution 53(5): 698–726.
Gartzke, Erik. 2006. “The Affinity of Nations Index: Similarity of State Voting Positions in the
UNGA.”

109

Geddes, Barbara. 2005a. “Why Parties and Elections in Authoritarian Regimes?”
———. 2005b. “Why Parties and Elections in Authoritarian Regimes?” In Annual Meeting of
the American Political Science Association,.
Geddes, Barbara, Joseph Wright, and Erica Frantz. 2014. “Autocratic Breakdown and Regime
Transitions: A New Data Set.” Perspectives on Politics 12(2): 313–31.
Gehlbach, Scott, and Alberto Simpser. 2014. “Electoral Manipulation as Bureaucratic Control.”
American Journal of Political Science 59(1): 212–24.
Green, Amanda. 2005. “Managing Human Resources in a Decentralized Context.” In East Asia
Decentralizes: Making Local Government Work in Asia, eds. Ronald White and Paul
Smoke. Washington, D.C: World Bank, 129–55.
Greene, Kenneth F. 2007. Why Dominant Parties Lose: Mexico’s Democratization in
Comparative Perspective. New York: Cambridge University Press.
Haber, Stephen. 2006. “Authoritarian Government.” In Handbook of Political Economy, eds.
Barry R Weingast and Donald A. Wittman. New York: Oxford University Press, 693–707.
Haber, Stephen, and Victor Menaldo. 2011. “Do Natural Resources Fuel Authoritarianism? A
Reappraisal of the Resource Curse.” American Political Science Review 105(1): 1–26.
Hadenius, Axel, and Jan Teorell. 2007. “Pathways from Authoritarianism.” Journal of
Democracy 18(1): 143–57.
Haftel, Yoram Z. 2010. “Ratification Counts: US Investment Treaties and FDI Flows into
Developing Countries.” Review of International Political Economy 17(2): 348–77.
Heberer, Thomas. 2006. Institutional Change and Legitimacy via Urban Elections? People’s
Awareness of Elections and Participation in Urban Neighbourhoods (Shequ). Duisburger
Arbeitspapiere Ostasienwissenschaften.
Hollyer, James R., B. Peter Rosendorff, and James Raymond Vreeland. 2011. “Democracy and
Transparency.” The Journal of Politics 73(4): 1191–1205.
Holz, Carsten A. 2004. “China’s Statistical System in Transition: Challenges, Data Problems, and
Institutional Innovations.” Review of Income and Wealth 50(3): 381–409.
———. 2005. OECD — China Governance Project. Paris: Organisation for Economic
Co-operation and Development. OECD Statistics Working Papers.
Huntington, Samuel P. 1991. The Third Wave: Democratization in the Late Twentieth Century.
Norman: University of Oklahoma Press.
ICSID Cases. https://icsid.worldbank.org/apps/ICSIDWEB/cases/Pages/AdvancedSearch.aspx
(Access on March 22, 2015).
110

Ikenberry, G. John. 2008. “The Rise of China and the Future of the West: Can the Liberal System
Survive?” Foreign Affairs 87(1): 23–37.
International Country Risk Guide.” The PRS Group.
http://www.prsgroup.com/about-us/our-two-methodologies/icrg (Accessed on March 21,
2015).
International Investment Agreements by Economy.
http://investmentpolicyhub.unctad.org/IIA/IiasByCountry#iiaInnerMenu (Accessed on
March 21, 2015).
Ito, Hiro, and Menzie Chinn. 2008. “A New Measure of Financial Openness.” Journal of
Comparative Policy Analysis 10(3): 309–22.
Jensen, Nathan. 2008. “Political Risk, Democratic Institutions, and Foreign Direct Investment.”
The Journal of Politics 70(4): 1040–52.
Jensen, Nathan M. 2003. “Democratic Governance and Multinational Corporations: Political
Regimes and Inflows of Foreign Direct Investment.” International Organization 57(3):
587–616.
Keefer, Philip. 2012. Database of Political Institutions: Changes and Variable Definitions.
Development Research Group, World Bank.
Kent, Ann E. 2007. Beyond Compliance: China, International Organizations, and Global
Security. Stanford University Press.
Kerner, Andrew. 2009. “Why Should I Believe You? The Costs and Consequences of Bilateral
Investment Treaties, Why Should I Believe You? The Costs and Consequences of
Bilateral Investment Treaties.” International Studies Quarterly 53: 73–102.
Kienle, Eberhard. 1998. “More than a Response to Islamism: The Political Deliberalization of
Egypt in the 1990s.” Middle East Journal 52(2): 219–35.
King, Gary. 1988. “Statistical Models for Political Science Event Counts: Bias in Conventional
Procedures and Evidence for the Exponential Poisson Regression Model.” American
Journal of Political Science 32(3): 838–63.
Laeven, Luc, and Fabián Valencia. 2013. “Systemic Banking Crises Database.” IMF Economic
Review 61(2): 225–70.
Lai, Brian, and Dan Slater. 2006. “Institutions of the Offensive: Domestic Sources of Dispute
Initiation in Authoritarian Regimes, 1950-1992.” American Journal of Political Science
50(1): 113–26.
Lam, Willy Wo-Lap. 1999. The Era of Jiang Zemin. New York: Prentice Hall.
Landry, Pierre F. 2008. Decentralized Authoritarianism in China: The Communist Party’s
111

Control of Local Elites in the Post-Mao Era. New York: Cambridge University Press.
Lazarev, Valery. 2005. “Economics of One-Party State: Promotion Incentives and Support for the
Soviet Regime1.” Comparative Economic Studies 47(2): 346–63.
Leeson, Peter T., and Andrea M. Dean. 2009. “The Democratic Domino Theory: An Empirical
Investigation.” American Journal of Political Science 53(3): 533–51.
Lin, Sen. 1992. “A New Pattern of Decentralization in China The Increase of Provincial Powers
in Economic Legislation.” China Information 7(3): 27–38.
Linz, Juan J. 2000. Totalitarian and Authoritarian Regimes. Boulder, CO: Lynne Rienner
Publishers.
Lipset, Seymour Martin. 1959. “Some Social Requisites of Democracy: Economic Development
and Political Legitimacy” eds. Kent Weaver and David M. Rocke. The American Political
Science Review 53(1): 69–105.
Lust-Okar, Ellen. 2005. Structuring Conflict in the Arab World: Incumbents, Opponents, and
Institutions. Cambridge: Cambridge University Press.
———. 2008. “Competitive Clientelism in Jordanian Elections.” In Political Participation in the
Middle East and North Africa, eds. Ellen Lust-Okar and Saloua Zerhouni. Boulder, CO:
Lynne Rienner Publishers.
Machado, K. G. 1971. “Changing Aspects of Factionalism in Philippine Local Politics.” Asian
Survey 11(12): 1182–99.
Magaloni, Beatriz. 2006. Voting for Autocracy: Hegemonic Party Survival and Its Demise in
Mexico. New York: Cambridge University Press.
———. 2008. “Credible Power-Sharing and the Longevity of Authoritarian Rule.” Comparative
Political Studies 41(4-5): 715–41.
———. 2010. “The Game of Electoral Fraud and the Ousting of Authoritarian Rule.” American
Journal of Political Science 54(3): 751–65.
Magaloni, Beatriz, and Ruth Kricheli. 2010. “Political Order and One-Party Rule.” Annual
Review of Political Science 13(1): 123–43.
Malesky, Edmund J. 2008. “Straight Ahead on Red: How Foreign Direct Investment Empowers
Subnational Leaders.” The Journal of Politics 70(1): 97–119.
Manion, Melanie. 1996. “The Electoral Connection in the Chinese Countryside.” The American
Political Science Review 90(4): 736–48.
MARC Code List for Geographic Areas: Code Sequence.
http://www.loc.gov/marc/geoareas/gacs_code.html (Accessed on March 3, 2015).
112

Marshall, Monty, and Keith Jaggers. 2002. “Polity IV Project: Political Regime Characteristics
and Transitions, 1800-2002.”
Martin, Lisa L., and Beth A. Simmons. 1998. “Theories and Empirical Studies of International
Institutions.” International Organization 52(4): 729–57.
Mattes, Michaela, and Mariana Rodríguez. 2013. “Autocracies and International Cooperation.”
International Studies Quarterly.
McDonough, Peter. 1995. “Identities, Ideologies, and Interests: Democratization and the Culture
of Mass Politics in Spain and Eastern Europe.” The Journal of Politics 57(3): 649–76.
Miller, Gary J. 2005. “The Political Evolution of Principal-Agent Models.” Annual Review of
Political Science 8(1): 203–25.
Montinola, Gabriella, Yingyi Qian, and Barry R. Weingast. 1995. “Federalism, Chinese Style:
The Political Basis for Economic Success in China.” World Politics 48(1): 50–81.
Montinola, Gabriella R. 1999. “Parties and Accountability in the Philippines.” Journal of
Democracy 10(1): 126–40.
Mozaffar, Shaheen. 2002. “Patterns of Electoral Governance in Africa’s Emerging Democracies.”
International Political Science Review 23(1): 85–101.
Neumayer, Eric, and Laura Spess. 2005. “Do Bilateral Investment Treaties Increase Foreign
Direct Investment to Developing Countries?” World Development 33(10): 1567–85.
Oates, Wallace E. 1972. Fiscal Federalism. New York: Harcourt Brace Jovanovich.
Ochoa-Reza, Enrique. 2004. “Multiple Arenas of Struggle: Federalism and Mexico’s Transition
to Democracy.” In Federalism and Democracy in Latin America, ed. Edward L. Gibson.
Baltimore: Johns Hopkins University Press.
Olson, Mancur. 1965. The Logic of Collective Action; Public Goods and the Theory of Groups.
Cambridge, Mass: Harvard University Press.
Ouyang, Alice Y., Ramkishen S. Rajan, and Tom Willett. 2008. “Managing the Monetary
Consequences of Reserve Accumulation in Emerging Asia.” Global Economic Review
37(2): 171–99.
Palmer, Glenn, Vito D’Orazio, Michael Kenwick, and Matthew Lane. 2015. “The MID4 Data Set:
Procedures, Coding Rules, and Description.” Conflict Management and Peace Science
forthcoming.
Pepinsky, Thomas. 2007. “Autocracy, Elections, and Fiscal Policy: Evidence from Malaysia.”
Studies in Comparative International Development 42(1-2): 136–63.
Przeworski, Adam. 2000. Democracy and Development: Political Institutions and Well-Being in
113

the World, 1950-1990. New York: Cambridge University Press.
Qian, Yingyi, and Barry R. Weingast. 1996. “China’s Transition to Markets: Market-Preserving
Federalism, Chinese Style.” The Journal of Policy Reform 1(2): 149–85.
———. 1997. “Federalism as a Commitment to Perserving Market Incentives.” The Journal of
Economic Perspectives 11(4): 83–92.
Riker, William H. 1964. Federalism: Origin, Operation, Significance. Boston: Little Brown.
Rodden, Jonathan. 2004. “Comparative Federalism and Decentralization: On Meaning and
Measurement.” Comparative Politics 36(4): 481–500.
Rodden, Jonathan, and Susan Rose-Ackerman. 1997. “Does Federalism Preserve Markets?”
Virginia Law Review 83(7): 1521–72.
Rosendorff, B. Peter, and Kongjoo Shin. 2012. “Importing Transparency: The Political Economy
of BITs and FDI Flows.”
Ross, Michael L. 1999. “The Political Economy of the Resource Curse.” World Politics 51(2):
297–322.
Saiegh, Sebastian M. 2005. “Do Countries Have a ‘Democratic Advantage’? Political Institutions,
Multilateral Agencies, and Sovereign Borrowing.” Comparative Political Studies 38(4):
366–87.
Salacuse, Jeswald W., and Nicholas P. Sullivan. 2005. “Do BITs Really Work: An Evaluation of
Bilateral Investment Treaties and Their Grand Bargain.” Harvard International Law
Journal 46: 67.
Sarkees, Meredith Reid, and Phil Schafer. 2000. “The Correlates of War Data On War: An
Update To 1997.” Conflict Management and Peace Science 18(1): 123–44.
Sartori, Giovanni. 1976a. Parties and Party Systems: A Framework for Analysis. New York:
Cambridge University Press.
———. 1976b. Parties and Party Systems: A Framework for Analysis. New York: Cambridge
University Press.
Schatz, Edward. 2006. “Access by Accident: Legitimacy Claims and Democracy Promotion in
Authoritarian Central Asia.” International Political Science Review 27(3): 263–84.
Schedler, Andreas. 2002. “The Menu of Manipulation.” Journal of Democracy 13(2): 36–50.
Scobell, Andrew. 2014. “China and Taiwan: Balance of Rivalry with Weapons of Mass
Democratization.” Political Science Quarterly 129(3): 449–68.
Shair-Rosenfield, Sarah, Gary Marks, and Liesbet Hooghe. 2014. “A Comparative Measure of
114

Decentralization for Southeast Asia.” Journal of East Asian Studies 14(1): 85–107.
Shan, Wenhua, and Jinyuan Su, eds. 2015. 2 China and International Investment Law: Twenty
Years of ICSID Membership. Beaverton, United States: Ringgold Inc.
Simmons, Beth A. 2000. “International Law and State Behavior: Commitment and Compliance
in International Monetary Affairs.” The American Political Science Review 94(4):
819–35.
———. 2014. “Bargaining over BITs, Arbitrating Awards: The Regime for Protection and
Promotion of International Investment.” World Politics 66(1): 12–46.
Simmons, Beth A, Frank Dobbin, and Geoffrey Garrett, eds. 2008. The Global Diffusion of
Markets and Democracy. New York: Cambridge University Press.
Simpser, Alberto. 2013. Why Governments and Parties Manipulate Elections: Theory, Practice,
and Implications. New York: Cambridge University Press.
Singer, J. David. 1988. “Reconstructing the Correlates of War Dataset on Material Capabilities of
States, 1816–1985.” International Interactions 14(2): 115–32.
Smith, Benjamin. 2005. “Life of the Party: The Origins of Regime Breakdown and Persistence
under Single-Party Rule.” World Politics 57(3): 421–51.
Svolik, Milan W. 2012a. “Chapter 6: Why Authoritarian Parties? The Regime Party as an
Instrument of Co-Coptation and Control.” In The Politics of Authoritarian Rule, New
York: Cambridge University Press.
———. 2012b. The Politics of Authoritarian Rule. New York: Cambridge University Press.
Tarrow, Sidney. 1995. “Mass Mobilization and Elite Exchange: Democratization Episodes in
Italy and Spain.” Democratization 2(3): 221–45.
Tiebout, Charles M. 1956. “A Pure Theory of Local Expenditures.” Journal of Political Economy
64(5): 416–24.
Tobin, Jennifer L., and Marc L. Busch. 2010. “A BIT Is Better Than a Lot: Bilateral Investment
Treaties and Preferential Trade Agreements.” World Politics 62(1): 1–42.
Tobin, Jennifer L., and Susan Rose-Ackerman. 2010. “When BITs Have Some Bite: The
Political-Economic Environment for Bilateral Investment Treaties.” The Review of
International Organizations 6(1): 1–32.
Tobin, Jennifer, and Susan Rose-Ackerman. 2011. “When BITs Have Some Bite: The
Political-Economic Environment for Bilateral Investment Treaties.” The Review of
International Organizations 6(1): 1–32.
Treisman, Daniel. 1999. “Political Decentralization and Economic Reform: A Game-Theoretic
115

Analysis.” American Journal of Political Science 43(2): 488–517.
———. 2002. “Defining and Measuring Decentralization: A Global Perspective.”
———. 2007. The Architecture of Government: Rethinking Political Decentralization. New
York: Cambridge University Press.
Ufen, Andreas. 2008. “Political Party and Party System Institutionalization in Southeast Asia:
Lessons for Democratic Consolidation in Indonesia, the Philippines and Thailand.” The
Pacific Review 21(3): 327–50.
Vernon, Raymond. 1985. “Sovereignty at Bay : Ten Years After.” Multinational corporations :
the political economy of foreign direct investment, ISBN 0-669-11241-0: 247–59.
Wallace, Jeremy L. 2014. “Juking the Stats? Authoritarian Information Problems in China.”
British Journal of Political Science: 1–19.
Waterbury, J. 1999. “Fortuitous Byproducts.” In Transitions to Democracy, ed. Lisa Anderson.
New York: Columbia University Press, 261–83.
Way, Lucan, and Steven Levitsky. 2002. “The Rise of Competitive Authoritarianism.” Journal of
Democracy 13(2): 51–65.
Weeks, Jessica L. 2008. “Autocratic Audience Costs: Regime Type and Signaling Resolve.”
International Organization 62(1): 35–64.
———. 2012. “Strongmen and Straw Men: Authoritarian Regimes and the Initiation of
International Conflict.” American Political Science Review 106(2): 326–47.
Welzel, Christian. 2006. “Democratization as an Emancipative Process: The Neglected Role of
Mass Motivations.” European Journal of Political Research 45(6): 871–96.
Wintrobe, Ronald. 2000. The Political Economy of Dictatorship. Cambridge: Cambridge
University Press.
Wong, Christine P. W. 1991. “Central–Local Relations in an Era of Fiscal Decline: The Paradox
of Fiscal Decentralization in Post-Mao China.” The China Quarterly 128: 691–715.
———. 2000. “Central-Local Relations Revisited: The 1994 Tax Sharing System Reform and
Public Expenditure Management in China.”
Wright, Joseph. 2008. “Do Authoritarian Institutions Constrain? How Legislatures Affect
Economic Growth and Investment.” American Journal of Political Science 52(2):
322–43.
Wu, Nai-Teh, and Ming-Tong Chen. 1993. “Regime Transition and Elite Mobility: The Historical
Formation of Local Elites in Taiwanese Politics.” In An Early History of Post-Colonized
Taiwan, ed. Tse-Han Lai. Taipei: Institute of Sociology, Academia Sinica, 303–34.
116

Xu, Chenggang, and Juzhong Zhuang. 1998. “Why China Grew: The Role of Decentralization.”
In Emerging from Communism: Lessons from Russia, China, and Eastern Europe, eds.
Peter Boone, Stanislaw Gomulka, and P. R. G. Layard. Cambridge, Mass: MIT Press.
Zhang, Kh. 2001. “Does Foreign Direct Investment Promote Economic Growth? Evidence from
East Asia and Latin America.” Contemporary Economic Policy 19(2): 175–85.
Zheng, Yongnian. 2000. “Institutionalizing de Facto Federalism in Post-Deng China.” In China
under Jiang Zemin, eds. Hung-mao Tien and Yun-han Chu. Boulder: Lynne Rienner
Publishers.
Zhong, Yang, and Jie Chen. 2002. “To Vote or Not to Vote An Analysis of Peasants’ Participation
in Chinese Village Elections.” Comparative Political Studies 35(6): 686–712.

117