BILATERAL INVESTMENT TREATIES - CONSEQUENCES ON HUMAN RIGHTS AND LABOR PROTECTION AND DETERMINANTS OF RATIFICATION  By  Fangjin Ye             A DISSERTATION  Submitted to Michigan State University in partial fulfillment of the requirements for the degree of  Political Science Ð Doctor of Philosophy  2016ABSTRACT  BILATERAL INVESTMENT TREATIES -  CONSEQUENCES ON HUMAN RIGHTS AND LABOR PROTECTION AND DETERMINANTS OF RATIFICATION  By  Fangjin Ye  This dissertation explores the domestic ratification process of bilateral investment treaties (BITs), and how these treaties and the asymmetric rights afforded by these treaties to investors impact human rights conditions and labor practices in developing countries. Chapter 2 investigates the domestic ratification process of BITs. It is puzzling that signed BITs take a long time to get ratified in some countries while they are ratified promptly in others. I model domestic ratification as a dynamic process of updating the costs and benefits of BITs after the signature. I argue that states update their assessment of the costs of BITs after signing by observing BIT claims against themselves as well as peer countries. Those claims delay ratification of BITs. In addition, host states are more likely to ratify BITs if ratification activity among their economic competitors is intense. Rapid ratification rates in countries that are economic competitors has the potential to increase fears of missed investment opportunities due to investment diversion, make domestic groups lean more favorably towards ratification and thus shorten the time between signing and ratifying BITs. Using all signed BITs as of 2007 and a BIT-year framework that allows to capture changing conditions after the signing stage, I find strong evidence for the arguments though the impact of BIT claims against states themselves is relatively weak, among others. Chapter 3 systematically theorizes and rigorously tests the effect of BITs on human rights practices of signatory countries. I argue that BITs have the potential to worsen human rights practices because they lock in initial conditions attractive to investors, both retrospectively and into the future. These conditions may include low standards for environmental protection or labor rights and tied hands with respect to provision of welfare benefits, basic infrastructure, investment in environmentally friendly technologies or land reform. The combined lock-in and constraining effects of BITs are sources of popular grievance and dissent in states that host foreign investment. Repression and human rights violations are key responses of states to the manifested or just anticipated protest and dissent that can result from such grievances. Furthermore, I argue that democracies have higher accountability and a lower threat perception for dissent, mitigating the negative effect of BITs. Using data on 113 developing countries between 1981 to 2009, I find support for my hypotheses. The last chapter examines the impact of BITs on collective labor rights that captures the freedom of association and collective bargaining rights and differentiates labor laws from the labor practices on the ground. I argue that BITs have little impact on collective labor laws while they worsen labor practices and widen the gap between labor laws and practices. Since foreign investors prefer to maintain the status quo level of regulations in capital-hosting states, BITs tend to stabilize labor laws. However, BITs are able to lock in initial low labor standards that are attractive to foreign investors, which may be a potential source of labor grievance and labor unrest. Also multinational corporations are found to be inviting targets for labor unrest. Given that all this anticipated or manifested labor unrest is likely to be challenged by foreign investors under stringent investment protective treaty clauses, host governments are forced to take measures to undermine the collective action capability of domestic labor groups, and reduce the risk of labor unrest. I argue that host governments may choose to undercut collective labor practices, resulting in a worsening of labor practices and a larger gap between labor laws and practices. Evidence from 120 developing countries from 1985 to 2002 supports my hypotheses.                                                                                  Copyright by                                                                        FANGJIN YE                                                                        2016         v To My Grandpa and Grandma           vi ACKNOWLEDGEMENTS   I am deeply indebted to many people for helping me at various stages of writing this dissertation. First and foremost, I would thank the members of my dissertation committee: Cristina Bodea, Benjamin Appel, Eric Chang, and Michael Colaresi. Being the chair of my doctoral dissertation committee, Cristina has always been supportive and encouraging since I began working on this project. She has been extremely generous with her time and valuable constructive criticisms. In addition to being a mentor, as a collaborator, she also taught me how to conduct rigorous research and publish work in journals. I am very fortunate to have Cristina as my advisor and I owe much of my success thus far to her guidance.  Ben has also provided invaluable advice and assistance. The first draft of the chapter 4 of this dissertation actually came out when I took his international law seminar in Fall 2013. He encouraged me to ask interesting questions and assisted me on various aspects of my research. I am also grateful for his time and assistance on the paper regarding the experimental study of investment treaties. Eric, since I first met him when I took his comparative political institutions seminar in Spring 2011, has always been very helpful in mentoring and encouraging me during my time in graduate school. I enjoyed his research seminars very much and learned an incredible amount of knowledge on comparative politics. Although frustration hits me from time to time when I took those seminars, it only makes me stronger and push myself harder. I am very grateful to meet Eric in my earlier stage of graduate study and have him in my dissertation committee. Mike was the chair of my guidance committee in my first two years of graduate school and continued to serve as a member of dissertation committee. He has always provided me with constructive comments on my chapters as well as other papers. His comments pushed me to think harder about the logic of my arguments, research design, and empirical strategies. I also thank Mike for his interesting proseminar course of international  vii relations in Spring 2013 which motives my interests in studying international relations as well as his much-needed assistance and encouragement during the downs of my graduate school. I truly couldnÕt have asked for more. Without the support and encouragement from my dissertation committee, I wouldnÕt have been able to finish my dissertation. Undoubtedly, the completion of this dissertation also benefits greatly from other members of MSU faculty and staff. In particular, I would like to thank Michael Bratton, Jeffrey Conroy-Krutz, Eric Juenke, Erica Frantz, Melinda Hall, Thomas Hammond, Richard Hula, Cristian Houle, Steven Kautz, Charles Ostrom, Corwin Smidt, Jakana Thomas, Karen Battin, Rhonda Burns, and Sarah Krause. I am also grateful to the Workshop on Conflict seminar in our department that provides me with opportunities to present my dissertation chapters as well as other papers.  In addition, I thank my fellow graduate students for their helpful comments, support, and most importantly, friendship. Gratitude goes to Fang-Yu Chen, Tolgahan Kinay Dilgin, Sung Min Han, Daniel Hansen, Masaaki Higashijima, Shih-Hao Huang, Hyunwoo Kim, Alon Kraitzman, Helen Lee, Hsin-Hsin Pan, Chunho Park, Peter Penar, Wen-Chin Wu, and many others.  Finally, I am deeply indebted to my family for being supportive and loving. My parents have always believed in me and given me much freedom to make decisions of my life on my own. I really appreciate it. I also thank my wife, Huidan, for her enormous support and encouragement. She was always available for chatting throughout the ups and downs of my graduate study. In addition, I am grateful to my old brother, Fangyun, who supports and takes care of our parents while I am away in the States.  Finally, I miss my grandpa and grandma. I lost both of them during my long journey of graduate study. I had lots of happiness in their house when I was a child. I hope happiness always stays with them in heaven. This dissertation is dedicated to them.   viii TABLE OF CONTENTS   LIST OF TABLES ...........................................................................................................................x LIST OF FIGURES ...................................................................................................................... xii CHAPTER 1 INTRODUCTION .....................................................................................................1    CHAPTER 2 THE DYNAMIC PROCESS OF INTERNATIONAL TREATY RATIFICATION: THE CASE OF BILATERAL INVESTMENT TREATY ..............................................................7    2.1 Introduction .............................................................................................................................7    2.2 Literature Review ..................................................................................................................11    2.3 Theory ...................................................................................................................................14 2.3.1 Costs and Benefits of BITs ............................................................................................14 2.3.2 ICSID Disputes and Ratification of BITs ......................................................................17 2.3.3 Competitive Pressure and Ratification of BITs .............................................................21    2.4 Data, Measurement and Research Design ............................................................................22  2.4.1 Dependent Variable .......................................................................................................23  2.4.2 Independent Variables ...................................................................................................23 2.4.3 Control Variables ...........................................................................................................26 2.4.4 Estimation Method .........................................................................................................28    2.5 Results and Discussion .........................................................................................................29  2.5.1 Robustness Check ..........................................................................................................34      2.6 Conclusion ............................................................................................................................43   CHAPTER 3 THE IMPACT OF BILATERAL INVESTMENT TREATIES ON HUMAN RIGHTS PRACTICES IN DEVELOPING COUNTRIES ...........................................................45 3.1 Introduction ..............................................................................................................................45    3.2 Bilateral Investment Treaties (BITs) ....................................................................................50    3.3 Literature Review ..................................................................................................................54 3.3.1 Foreign Direct Investment and Human Rights ..............................................................54 3.3.2 Preferential Trade Agreements and Human Rights .......................................................55    3.4 Theory ...................................................................................................................................58  3.4.1 Developing Countries and Policies Favoring Investors .................................................59 3.4.2 BITs Clauses Lock-in Favorable Policies ......................................................................61 3.4.3 BITs, Grievance and Human Rights Violations ............................................................65 3.4.4 The Mitigating Effect of Political Regime .....................................................................67    3.5 Data, Measurement and Research Design ............................................................................69   3.5.1 Dependent Variable .......................................................................................................69 3.5.2 Independent Variables ...................................................................................................70 3.5.3 Control Variables ...........................................................................................................72 3.5.4 Model Specification .......................................................................................................73    3.6 Results and Discussion .........................................................................................................74  3.6.1 Robustness Check ..........................................................................................................80      3.7 Conclusion ............................................................................................................................87    ix CHAPTER 4 THE IMPACT OF BILATERAL INVESTMENT TREATIES ON COLLECTIVE LABOR RIGHTS IN DEVELOPING COUNTRIES ....................................................................90    4.1 Introduction ...........................................................................................................................90    4.2 Literature Review ..................................................................................................................93 4.2.1 The Impact of FDI on Collective Labor Rights .............................................................93  4.2.2 The Impact of PTAs on Collective Labor Rights ..........................................................94    4.3 Theory ...................................................................................................................................95  4.3.1 The Impact of BITs on Collective Labor Laws .............................................................95 4.3.2 The Impact of BITs on Collective Labor Practices .......................................................97 4.3.3 The Impact of BITs on the Gap Between Labor Laws and Practices ..........................101    4.4 Data, Measurement and Research Design ..........................................................................102   4.4.1 Dependent Variable .....................................................................................................102  4.4.2 Independent Variables .................................................................................................104 4.4.3 Control Variables .........................................................................................................106 4.4.4 Model Specification .....................................................................................................107    4.5 Results and Discussion .......................................................................................................108  4.5.1 Robustness Check ........................................................................................................111      4.6 Conclusion ..........................................................................................................................121   CHPATER 5 CONCLUSION .....................................................................................................123 BIBLIOGRAPHY ........................................................................................................................125                         x LIST OF TABLES   Table 2-1. Costs and Benefits of BITs Perceived by Host Countries ............................................17   Table 2-2. Summary Statistics .......................................................................................................28  Table 2-3. Results of Cox Nonproportional Hazard Models 1970 Ð 2007 ....................................32  Table 2-4. Percentage Change of Hazard Rate: Vary ICSID Cases Filed and Ratified BITs in Economic Competing States, and Legislative Hurdles Variable ...................................................33  Table 2-5. Results of Logistic Models with a Cubic Polynomial Approximation 1970 Ð 2007 ....35  Table 2-6. Robustness of Cox Nonproportional Hazard Models Ð BITs Signed After 1989 ........36  Table 2-7. Robustness of Cox Nonproportional Hazard Models Ð ICSID Claims in Host Countries and Their Peer States .....................................................................................................39  Table 2-8. Robustness of Cox Nonproportional Hazard Models Ð Control for Half-decade Dummies ........................................................................................................................................40  Table 2-9. Robustness of Cox Nonproportional Hazard Models Ð Signed BITs with the ICSID Clause .............................................................................................................................................41  Table 2-10. Robustness of Cox Nonproportional Hazard Models Ð Signed BITs without the ICSID Clause .................................................................................................................................42  Table 3-1. Evidence That Host Countries with Low Credibility Sign BITs (Credibility Argument) ......................................................................................................................................49  Table 3-2. Summary Statistics .......................................................................................................74  Table 3-3. Effect of BITs on GovernmentsÕ Respect for Human Rights in Developing Countries 1981-2009 ......................................................................................................................................77  Table 3-4. Predicted Human Rights Conditions: Vary BITs and Democracy ...............................79  Table 3-5. Robustness Check for a Restricted Sample and Additional Control Variables Ð All Cumulative BITs ............................................................................................................................84  Table 3-6. Robustness Check for a Restricted Sample and Additional Control Variables Ð BITs with ICSID .....................................................................................................................................85   xi Table 3-7. Robustness Check for a Restricted Sample and Additional Control Variables Ð Adjusted North-South BITs ...........................................................................................................86  Table 3-8. Robustness Check for an Alternative Measure of Physical Integrity Index Ð PTS Scale ...............................................................................................................................................87  Table 4-1. Summary Statistics .....................................................................................................108  Table 4-2. Effect of BITs on Collective Labor Rights in Developing Countries 1985-2002 ......109  Table 4-3. Effect of BITs on Labor Laws and Practices Ð Seemingly Unrelated Regression (SUR Models) ........................................................................................................................................112  Table 4-4. Subsample Ð Above the Mean of Labor Laws ...........................................................114  Table 4-5. Subsample Ð Below the Mean of Labor Laws ............................................................115  Table 4-6. Sample of BITs Ratified After 1989 ..........................................................................115  Table 4-7. Robustness Check for Controlling PTAs and ILOs ...................................................117  Table 4-8. Robustness Check for Controlling IMF/WB Structural Adjustment Program ...........118  Table 4-9. Robustness Check for Controlling Human Rights and Rule of  Law .........................119  Table 4-10. Robustness Check for Controlling Labor NGOs and Media Freedom .....................120             xii LIST OF FIGURES  Figure 2-1. Distribution of Ratification Spell by BIT Status as of 2007 .........................................8  Figure 3-1. The Marginal Effect of BITs on GovernmentsÕ Respect for Human Rights Conditional on Regime Type .........................................................................................................78  Figure 4-1. Difference of Cross-equation BITs Coefficients (SUR Models) ..............................121                 1 CHAPTER 1 INTRODUCTION  Bilateral investment treaties (BITs) have become the most important international legal instruments for protecting and facilitating cross-border foreign direct investment (FDI) flows since their inception back in 1959 as a treaty between Germany and Pakistan. According to the current estimates from the United Nations Conference on Trade and Development (UNCTAD), 178 countries are involved in at least one BIT, and they have signed more than 2,900 BITs with one another1. BITs are created to encourage cross-border capital flows, especially from developed rich countries to developing states, by guaranteeing certain standards of investment protection in treaty language and implementation and hence reducing political and regulatory risks for foreign investors. Investment protection clauses include fair and equitable treatment, national treatment, repatriation of investment and return, no expropriation unless promptly compensated, and agreement to international dispute arbitration venues such as International Centre for Settlement of Investment Disputes (ICSID)2. One major part of the BIT literature examines how states participate in the global investment regime regulating foreign direct investment. Scholars mainly focus on three research areas Ð the signature of BITs, the design of BITs, and the ratification of these treaties. The first research area treats BITs as a Òblack boxÓ or undifferentiated treaties and examines the reasons countries come to sign them. This work tries to explain why host countries are willing to tie their hands and submit                                                 1 These statistics are obtained from UNCTADÕs International Investment Agreements (IIAs) database.  2 ICSID is a heavily used and widely observed international arbitral institution that is part of the World Bank. 156 countries have signed the ICSID Conventions (Allee and Peinhardt 2010). I will elaborate on those clauses later on in the dissertation. ! 2 investor-state disputes to international arbitration authorities. Two main mechanisms are argued to explain the signing of BITs: These investment treaties are argued to be substitutes for the low credibility of domestic commitment to investor rights3. Alternatively, they could be signed in response to a competition for capital among developing countries. While scholars find support regarding the mechanism of competition for capital (Elkins, Guzman, and Simmons 2006; Jandhyala, Henisz and Mansfield 2011; Neumayer and Plumper 2010; Lupu and Poast 2013), the empirical evidence is mixed for the other mechanism (Bergstrand and Egger 2013; Jandhyala, Henisz and Mansfield 2011; Lupu and Poast 2013; Swenson 2005; Elkins, Guzman and Simmons 2006; Neumayer and Plumper 2010).  Rather than consider BITs as a Ôblack boxÕ or undifferentiated treaties, another part of the literature explores the variation in BIT designs (e.g., the investor-state dispute settlement clause or the national treatment clause) and examines its causes (Allee and Peinhardt 2010, 2014; Blake 2013). Allee and Peinhardt (2010, 2014), for example, find strong evidence that the preference of home countries to tie host statesÕ hands (bargaining power) is driving the inclusion of stringent investor-state dispute settlement (ISDS) clause in BITs. Blake (2013) investigates the effect of government time horizons on the scope of national treatment clause in BITs4. He finds that governments with longer time horizons are more likely to preserve future policy autonomy by carving out policy areas in which they can deviate from the national treatment commitment.                                                 3 Multinational corporations face time inconsistency problem when investing in other countries. They are concerned about whether the host state would uphold their commitments about property rights protection agreed upon prior to investment. This is because bargaining power would shift towards the host state due to the sunk cost that foreign investors have encountered. So, host states may renege on their prior commitment and renegotiate their initial concession agreement afterwards.  4 National treatment commitment can be roughly defined as Òtreating foreign investors in a fair, equitable, and nondiscriminatory manner as domestic investorsÓ (Blake 2013).! 3 Additionally, some other researchers reveal the contagion mechanism in the literature of BIT designs (Neumayer, Nunnenkamp, and Roy 2014).  A third and the least voluminous research area on BIT participations emphasizes the ratification process of these treaties. This research explores domestic political conditions in order to address variation of the time between signing a treaty and ratifying it. Haftel and Thompson (2013) is the only work that directly examines why countries delay ratifying BITs after signing these treaties. Their theory focuses on three main explanatory variables: the institutional and political constraints that executives face domestically, transparency and predictability of political systems and the bilateral relationship between two states. Using a cross-section of 2,595 BITs signed between 1959 and 2007 and Cox models, and coding all independent variables at the year of signing BITs, they find that the difficulty of ratification increases with the political constraints the executive faces, legislative requirements for treaty ratification, and unpredictability of the political system, while it decreases with government capacity, and cultural and political affinity between two states. This approach, however, is static, looking at the conditions known as treaties were signed.  Chapter 2 of this dissertation contributes to this literature by modeling the ratification of BITs as a dynamic process both theoretically and empirically. Such an approach of modeling treaty ratification as a dynamic process is similar to Milner and Rosendorff (1997) and Mansfield and Milner (2012)Õs study of the ratification of preferential trade agreements (PTAs). I find strong evidence that BIT claims among peer states delay ratification, while rapid ratification rates among economic competing states facilitate a quicker ratification process of a signed BIT. Since BITs are created to solve the time-inconsistency problem that plagues the growth of foreign direct investment, the other major part of the BIT literature studies whether signing or ratifying BITs are able to increase FDI inflows into developing states. The empirical evidence on  4 the influence of BITs on investment flows is, however, mixed. While some scholars find a positive effect of BITs on FDI inflows into developing states (Neumayer and Spess 2005; Haftel 2010; Kerner 2009; Buthe and Milner 2009), some other works conclude that BITs have little impact on bringing FDI for developing countries (Hallward-Dreimeier 2003; Tobin and Rose-Ackerman 2005; Yackee 2007). In addition, there are some other researchers who argue that the effect of BITs is conditional on domestic institutional quality (Hallward-Dreimeier 2003; Tobin and Rose-Ackerman 2005; Yackee 2007) or subsequent good behavior of the governments who sign them (Allee and Peinhardt 2011). However, what is missing from this literature examining the consequences of BITs is that the Òbroad and asymmetricalÓ (Simmons 2014, p. 12) rights granted to investors may generate potential negative externalities for signatory developing states. Chapters 3 and 4 of this dissertation take on the task of investigating the impact of BITs on human rights practices and labor rights in developing countries. I find that BITs tend to worsen human rights practices and collective labor practices in developing countries. The findings of this dissertation also echo the concerns of domestic civil groups, NGOs, and political elites. ÒDespite signing this BIT5 more than 5 years ago and despite Canadian pressure, the Indian Parliament has declined to ratify this treaty and instead decided to reconsider some aspects of the Agreement that it found troubling,Ó said Elizabeth May, Green Party leader and member of the Canadian Parliament for Saanich-Gulf Islands.6 The resistance to ratification by the Indian Parliament is understandable. Vodafone served the Indian government with a Notice of Dispute under India-Netherland BIT in 2012 regarding retrospective tax legislation proposed by                                                 5 It is also called Canada-India Foreign Investment Protection and Promotion Agreement (FIPA) signed on June 16, 2007.  6 Source: http://www.greenparty.ca/en/media-release/2012-11-07/prime-minister-s-trip-fails-pressure-india-ratify-canada-india-fipa  5 the Indian government.7 In response to this threat, the Indian Parliament is currently reviewing all such BITs. In another example, when being asked about the potential negative consequences of Brazilian BIT ratification, Pedro Alberto Costa Braga de Oliveira, a Brazilian lawyer, noted ÒThe treaties would need to be so-called next generation treaties. In other words, they must be crafted to allow for sufficient governmental policy space, so that what has happened in Argentina8, for example, doesnÕt happen in Brazil.Ó The concerns regarding the potential costs of investor-state dispute arbitration clearly impact the ratification process of BITs.  Also human rights groups have charged that the hands of capital importing states are tied by investment treaties, generating important grievances and worsening governmentsÕ human rights and labor practices. NGOs9 have reservations about the ongoing negotiations on a US-India BIT, including about how the investor-state dispute mechanism can undermine the domestic policy space and domestic justice system. Civil groups10 and legal scholars (Prislan and Zandvliet 2013) are very cautious about the possibility that foreign investors may bring host governments to courts if governments would promote distinctly pro-labor policies (e.g., increase minimum wage and collective bargaining rights) or failed to put a swift end to labor unrest. Similar litigation has been observed regarding effort to promote public health and improve environmental protection. This dissertation has important implications. Chapter 2 suggests we model the ratification of international treaties in general as a dynamic process. It also sheds light on the role that                                                 7 Source: http://www.vodafone.com/content/index/media/vodafone-group-releases/2012/bit.html 8 According to Wellhausen (forthcoming), Argentina has the most public arbitrations filed against it (about 56) from 1995 to 2011. The majority of those lawsuits emerged due to its 2001Ð2002 default and financial crisis. 9 E.g., the Indian NGO ÒForum against Free Trade AgreementsÓ, a coalition of over 75 organizations.!10 See, eg, Bill Rosenberg, ÔLabour Rights and Investment AgreementsÕ (2 March 2012) <http://union.org.nz/sites/union.org.nz/files/Labour-Rights-and-TPPA-March2012.ppt>   6 international institutions (e.g., international investment arbitration institutions) play in providing information for domestic audiences. Furthermore, chapter 2 considers international treaty ratifications among states as being interdependent. States learn the consequences of international treaty commitments and ratification pace in other similarly situated states, and incorporate this information into the cost-benefit analysis of ratifying their own treaties. While the majority of the BIT literature studies whether signing or ratifying BITs are able to increase FDI inflows into developing states, chapters 3 and 4 draw the attention to the unintended consequences of concluding BITs. BITs have the potential to worsen human rights practices and collective labor practices in developing countries. The findings back the recent push by developing nation to incorporate human rights and environmental protection clauses as well as labor standards in the text of BITs. The remainder of the dissertation is organized as follows. The next three chapters present the logic of my arguments, hypotheses, research design, and empirical evidence for each of the three chapters. The final chapter concludes the dissertation.           7 CHAPTER 2 THE DYNAMIC PROCESS OF INTERNATIONAL TREATY RATIFICATION: THE CASE OF BILATERAL INVESTMENT TREATY  2.1 Introduction  Since its inception back in 1959 as a treaty between Germany and Pakistan, bilateral investment treaties (BITs) have become the most important and fast-growing international legal instrument underlying the growth of foreign direct investment. The process of agreeing to enter a bilateral investment treaty (BIT) between two countries involves two steps Ð signing the agreement in the international negotiation stage and then ratifying it within the domestic political arena. Importantly, a BIT enters into force only when ratified by both sides, that is, in the event of mutual ratification. Put differently, only ratified BITs are legally binding and therefore serve as a powerful mechanism for protecting foreign investors (Haftel 2010). But since only ratified BITs are binding, it is unclear why some states have a significant ratification delay after the BIT signing. For instance, some BITs are mutually ratified promptly while others are either delayed for a long time or never enter into force. As an example, Brazil signed 14 BITs in the early 1990s but none of them are ratified.11 Figure 2-1 shows the global distribution of the time between signing and ratifying BITs. For those BITs that have entered into force as of 200712, more than 80% have been ratified within 5 years of signature. However, for those treaties that were not mutually ratified as of 2007, about 80% of them are beyond 5 years of signature, indicating that leaders have                                                 11 Brazil signed BITs that are as of yet unratified with Portugal, Chile, UK, Ireland and Switzerland in 1994, Denmark, Finland, France, Germany, Italy, Venezuela in 1995, Cuba in 1997, Netherlands in 1998, and Belgium in 1999. 12 Year 2007 is the last year of my observations in the empirical analysis, so the data is right-censored at year 2007. ! 8 difficulties selling a majority of those BITs domestically even after signing more than five years prior. Furthermore, the average time between signing and ratifying for those BITs that are in force is about 3 years while the average time for those that are not in force by 2007 is about 8 years. What explains the length of time that it takes to ratify a BIT?          Figure 2-1. Distribution of Ratification Spell by BIT Status as of 2007   Previous research addresses this puzzle by mainly examining three aspects: (1) institutional (formal legislative requirements for treaty ratification) and political constraints that executives face domestically, (2) transparency and predictability of political systems, and (3) the nature of the bilateral relationship between the two states signing the BIT (Haftel and Thompson 2013). Because Haftel and Thompson (2013) have a static theoretical focus on statesÕ characteristics at the time of  9 signing, they use BIT as the unit of analysis and code all independent variables at the time of signing BITs. However, what is missing from this explanation is that the domestic ratification of BITs is a dynamic process. Domestic actors likely update their perception of costs and benefits of BITs as new information emerges following the BIT signing. This may affect the difficulty of domestic ratification. Therefore, a theory of conditions that alter statesÕ cost-benefit calculations and a research design that uses a BIT-year data structure rather than a cross-section of signed BITs are much better able to capture this dynamic process. I argue that signing BITs is a revealed preference and that, on balance, at the time of signing countries view benefits as outweighing the costs of BITs13. However, domestic actors in the signatory states have the opportunities to update their assessment of costs and benefits as time passes. I argue that investor-state disputes raised via ICSID update the perceived costs of BITs for domestic opposition groups. Due to the high observability of proceedings and awards as well as the effectiveness of enforcement mechanisms, observing ICSID disputes registered against states themselves as well as peer states increases the perceived costs of ratifying additional BITs for domestic opposition groups. The updated costs of BITs provide a focal point for domestic opposition and upset political executivesÕ cost-benefit equilibrium achieved during the treaty signing stage. This makes it difficult for executives to build broad coalitions in support of ratifying the signed BIT. Thus, observing its own disputes at ICSID or the disputes in peer countries should increase resistance to the ratification of signed BITs. In addition, I argue that states are more likely to go on and ratify BITs if their economic competitors have done so. Compared to merely signed BITs, ratified BITs are more likely to generate credible commitments for protecting foreign investment (Haftel 2010). As a result,                                                 "# I elaborate on the likely benefits and costs in the theory section.   10 ratified BITs raise the expected return on investment and level the playing field for host states as they compete for capital with similar countries. Therefore, I argue that the ratification practices of BITs in economic competitors update the observed stateÕs perception of the benefits of BIT ratification. A rapid ratification rate in countries that are economic competitors make domestic groups lean more favorably towards ratification and shorten the time between signing and ratifying BITs.  Using all signed BITs as of 2007 and a BIT-year setup I find strong evidence that BIT claims in peer states (either geographically close or in the similar risk profile of violating foreign investment protection) after the signature of a BIT delays its ratification while rapid ratification rates in economic competitors facilitate the ratification of this BIT. However, the impact of BIT claims against states themselves on the duration of ratification is not consistent and relatively weak.14 My empirical results are robust to an alternative estimation strategy, restricting the sample to BITs signed after 1989, measuring BIT claims at the host state level rather than in the dyadic level, controlling for time trend or time specific shocks, and exploring the heterogeneity of signed BITs by focusing on investor-state dispute settlement (ISDS) mechanism.  The chapter makes several contributions. First, the ratification process is important because only ratified BITs are legally binding and expected to attract foreign direct investment or lead to costly arbitration with foreign investors (Poulsen and Aisbett 2013). FDI is suggested to bring in capital and technological innovation for host states, and thus will promote economic growth (Jensen 2003). On the other hand, the costs of arbitration and the increased likelihood that investors will pursue arbitration have arguably slowed down the pace of signing BITs (Poulsen and Aisbett 2013). Therefore, a more comprehensive understanding of the ratification process appears                                                 "$ I will elaborate on this weak finding in the section of empirical result discussion.    11 necessary, in a literature that either ignores this final stage before a treaty becomes binding or focuses on static institutional features that political actors could have anticipated as hurdles in the ratification process. Second, this chapter joins scholars who treat the ratification of international treaties as a distinct outcome of international cooperation. This work employs survival models and examines the determinants of treaty ratification in the areas of international environmental treaties (Fredriksson and Gaston 2000, Neumayer 2002, Von Stein 2008), international labor treaties (Boockmann 2001, Chau et al. 2001), preferential trade agreements (Mansfield and Milner 2012), and bilateral investment treaties (Haftel and Thompson 2013). Third, the chapter speaks to the literature of how international conditions interact with domestic factors in influencing international cooperation (Chapman 2009, Dai 2005). Exogenous information following BIT signing provided by international investment arbitral institutions updates the costs of BITs for domestic oppositions and upsets the initial cost-benefit equilibrium for political executives during the signing stage. Finally, this paper contributes to the literature identifying a competitive pressure among states to attract FDI. The literature has already identified such pressure in signing BITs (Elkins, Guzman, and Simmons 2006), welfare spending (Rudra 2002), labor practices (Davies and Vadlamannati 2013), and tax breaks (Klemn and Parys 2012).  The remainder of the chapter is organized as follows. The next section reviews the literature on BITs with an emphasis on the BIT ratification. A second section presents the logic of the argument and the hypotheses. The third section discusses data and methodology. Section 4 presents the empirical evidence and discussion. The final section concludes the chapter.  2.2 Literature Review The overwhelming majority of the BIT literature studies the effect of these treaties on foreign direct investment (Neumayer and Spess 2005, Haftel 2010, Kerner 2009, Buthe and Milner 2009,  12 Hallward-Dreimeier 2003, Tobin and Rose-Ackerman 2005, Yackee 2007, Allee and Peinhardt 2011).15 Another important branch of the literature examines the reasons behind countriesÕ signing BITs or the kinds of clauses included in these treaties. Two main mechanisms are argued to explain the signing of BITs: These investment treaties are argued to be substitutes for the low credibility of domestic commitment to investor rights, and that they could be signed in response to a competition for capital among developing countries.16 The reasons behind the design of BITs are also examined, including why countries sign on to ICSID and national treatment clauses (Allee and Peinhardt 2010, 2014, Blake 2013, Simmons 2014).17 A third and the least voluminous literature on BITs emphasizes the ratification process of these treaties. This research explores domestic political conditions in order to address variation of the time between signing a treaty and ratifying it. Haftel and Thompson (2013) is the only work that directly examines why countries delay ratifying BITs after signing these treaties. Their theory focuses on three main explanatory variables: the institutional and political constraints that                                                 "% The evidence is mixed, although most recent studies using instrumental variable methods do find that BITs increase FDI.  "& While scholars find support regarding the mechanism of competition for capital (Elkins, Guzman, and Simmons 2006, Jandhyala, Henisz and Mansfield 2011, Neumayer and Plumper 2010, Lupu and Poast 2013), the empirical evidence is mixed for the other mechanism (Bergstrand and Egger 2013, Jandhyala, Henisz and Mansfield 2011, Lupu and Poast 2013, Swenson 2005, Elkins, Guzman and Simmons 2006, Neumayer and Plumper 2010).!17 Allee and Peinhardt (2014) compare three competing theories - the preference of host countries to tie its hands (depending on the severity of its credible commitment problem), the preference of home countries to tie host statesÕ hands (bargaining power), and rational design of international institutions to incorporate future uncertainty Ð in explaining the clause of investor-state dispute settlement (ISDS) in BITs, and find strong evidence for the theory of Òthe power and preference of home stateÓ. Allee and Peinhardt (2010) also find support for power and preference of home state mechanism for investor-state dispute resolution clauses in BITs rather than substitution for low credibility of domestic commitment mechanism. See also Simmons (2014). Blake (2013) investigates the effect of government time horizons on national treatment clauses in BITs. National treatment commitment can be roughly defined as Òtreating foreign investors in a fair, equitable, and nondiscriminatory manner as domestic investorsÓ.  13 executives face domestically, transparency and predictability of political systems and the bilateral relationship between two states. Using a cross-section of 2,595 BITs signed between 1959 and 2007 and Cox models, and coding all independent variables at the year of signing BITs, they find that the difficulty of ratification increases with the political constraints the executive faces, legislative requirements for treaty ratification, and unpredictability of the political system, while it decreases with government capacity, and cultural and political affinity between two states. Haftel and Thompson (2013) make a significant contribution to understating how countries assume their negotiated international legal commitments. This approach, however, is static, looking at the conditions known as treaties were signed. I propose modeling the ratification of BITs as a dynamic process both theoretically and empirically. Such an approach of modeling treaty ratification as a dynamic process is similar to Milner and Rosendorff (1997) and Mansfield and Milner (2012)Õs study of the ratification of preferential trade agreements (PTAs)18. Theoretically, assuming both governments negotiating a BIT decide to sign based on a rational calculation of the costs and the benefits, they have opportunities to update these expectations during the sometimes lengthy ratification process. Changes in the assessment of costs and the benefits after signing should influence then their incentive to ratify. Empirically, given my discussion of a time-varying cost-benefit evaluation of BITs, it appears more appropriate to use a research design that codes                                                 18 Milner and Rosendorff (1997) model the domestic determinants of the delay in ratifying North American Free Trade Agreement (NAFTA) and argue that legislative elections are a central source of uncertainty, which makes it difficult for the executive to anticipate the preferences of the median legislator when negotiating these agreements with foreign countries. Mansfield and Milner (2012) examine the duration between signing and ratifying a preferential trade agreement in a cross-national setting. They argue that the change in the number and composition of veto players after signing the treaty increases the delay of ratification because the interests of those new veto players are not anticipated, and thus not incorporated and well represented in the negotiation stage of the PTA.  14 BIT-year data in order to assess how some important changing covariates affect duration of time between signing and ratifying. 2.3 Theory 2.3.1 Costs and Benefits of BITs In general, BITs are signed to reduce political and regulatory risks for foreign investors. Hence, BITs are expected to increase FDI inflows for signatory states. Governments value FDI because it has the potential to generate economic growth, local employment, or transfers of technology (Jensen 2003). Other benefits of signing BITs may include: (1) signaling liberalization policy orientation to some domestic audiences without intention of ratification, (2) strengthening intergovernmental political ties and establishing diplomatic relations especially among BITs between developing countries (Haftel and Thompson 2013, Poulsen 2014), (3) behaving in a way that is consistent with the accepted norm of treating foreign investors (Jandhyala, Henisz and Mansfield 2011).  However, BITs are costly for signatory states. First, the wide-ranging nature of investment rules involves sovereignty costs and constrains the autonomy of domestic regulations including labor law, environmental law, health law, and property rights law (Allee and Peinhardt 2010). Such loss of policy autonomy can constrain sustainable development strategies significantly and prohibit governments from achieving environmental and social policy goals (Blake 2013). It also increases capital importing statesÕ probability of being reviewed and adjudicated by an international arbitration tribunal if foreign investors choose to bring claims to the tribunal.  There are many illustrations of countries considering the litigation costs involved with signing BITs. For instance, Brazil had fierce parliamentary debates over the sovereignty loss implied by delegation of arbitration to the ICSID during 1990s. This situation reflects the substantial costs of  15 exiting national courts and undermining the long-accepted Calvo Doctrine19 (Campello and Lemos 2015). In addition to granting foreigners equal treatment as domestic investors, this doctrine states that any disputes between foreigners and host states should be adjudicated by national courts of the host states. Similarly, the UK Ð Colombia BIT was signed in 2010 but is not yet ratified. Human rights and anti-poverty groups20 are concerned that this BIT containing international arbitration clauses21 will expose the Colombian government to costly lawsuits and impact Colombian land reform programs. These rights groups worry that the BIT would threaten the return of 5 million internally displace people. Another example is that French investor - Vivendi Universal Ð brought the government of Argentina to the international arbitral court in 1997 under the Argentina-France BIT signed in 1991. The company owned by this investor entered into a 30-year concession agreement in May 1995 with the Tucuman province for providing water and sewage services during Argentinean privatization campaign of state-owned public services in the early 1990s. However, later on the investor encountered opposition from the new Government of Tucuman and forced request of renegotiation of the agreement. In 1997, the investor filed a lawsuit in the ICSID                                                 19 ÒThe doctrine was advanced by the Argentine diplomat and legal scholar Carlos Calvo, in his International Law of Europe and America in Theory and Practice (1868). It affirmed that rules governing the jurisdiction of a country over aliens and the collection of indemnities should apply equally to all nations, regardless of size. It further stated that foreigners who held property in Latin American states and who had claims against the governments of such states should apply to the courts within such nations for redress instead of seeking diplomatic intervention. Moreover, according to the doctrine, nations were not entitled to use armed force to collect debts owed them by other nations. A Calvo clause in a contract between the government of a Latin American state and an alien stipulates that the latter agrees unconditionally to the adjudication within the state concerned of any dispute between the contracting parties.Ó Source: Britannica, on-line academic edition. '( E.g., Colombian NGO Cedetrabajo or British NGO Traidcraft. '" Those international arbitration venues include ICSID, Court of Arbitration of the International Chamber of Commerce, or ad hoc arbitration tribunal under Arbitration Rules of the United Nations Commission on International Trade Law (UNCITRAL). ! 16 venue on the grounds of unfair treatment and expropriation without compensation. An amount of 105 million (US dollars) was awarded to investors in 200022.  In a last illustrative example, investors from Luxembourg and Italy brought a claim against South Africa under the ICSID additional facility rules in 2007 saying that South AfricaÕs Mining and Petroleum Resources Development Act (MPRDA) expropriated their mineral rights, which is part of South AfricaÕs redistributive policy in favor of historically oppressed domestic social groups Ð Black Economic Empowerment (BEE) Policy. Among other things, MPRDA requires equity in mining companies be partly owned by ÔHistorically Disadvantage PersonsÕ. Furthermore, right after the settlement of this claim in 2010, South Africa terminated its BIT with Belgium-Luxembourg Economic Union on the grounds that this BIT limits the governmentÕs ability to pursue its constitutional-based transformation agenda. Due to concerns of costly sovereignty loss, some states including Ecuador, Bolivia, and Venezuela plan to withdraw their membership in ICSID (Allee and Peinhardt 2010).  A second important downside of BITs is related to the direct financial costs of appearing before international arbitration such as the ICSID. These costs can be substantial. Host states need to invest time and resources to get familiar with the international legal system. Instead, host states would rather prefer to solve disputes in the domestic legal systems in which they have more expertise and capability than multinational enterprises. Furthermore, some evidence suggests that in international arbitration tribunals there is some degree of legal asymmetry in favor of foreign investors (Van Harten 2012, Simmons 2014). Also the size of monetary awards has been significant including recent decisions against the governments of the Czech Republic ($350 million in 2001), Lebanon ($266 million in 2005) (Elkins et al. 2006), or Ecuador ($2.3 billion in                                                 22 This case is drawn from Wellhausen (2016). ! 17 2012)23. Finally, as shown by Allee and Peinhardt (2011), future foreign direct investment inflows decrease when investors register claims against host states at the ICSID venue. Host states experience the decline of FDI inflows even at the moment of simply being respondents in the lawsuits, which occurs years ahead of an outcome being decided. Table 2-1 shows a broad summary of the perceived costs and benefits of signing BITs.           Table 2-1. Costs and Benefits of BITs Perceived by Host Countries  Costs Benefits Loss of sovereignty: loss of regulatory policy autonomy to serve domestic purpose  (1) Attracting FDI, (2) signaling liberalization policy orientation to some domestic audiences without intention of ratification, (3) strengthening intergovernmental political ties, (4) behaving in a way that is consistent with the accepted norm of treating foreign investors. Potential costs of being sued including financial costs of litigations, loss of future FDI due to ICSID litigations  2.3.2 ICSID Disputes and Ratification of BITs I start from the assumption that political executives are satisfied with the balance of costs and benefits of the BIT at the time when the treaty is signed. Signature is a revealed preference that, on balance, governments view benefits as outweighing the costs of BITs. However, domestic actors are very likely to update their assessment of costs and benefits as time passes after the signature. If perceived costs of concluding a BIT start outweighing the benefit, domestic actors will be more likely to oppose moving to actually ratify the BIT. To the contrary, if benefits of concluding a BIT remain larger than costs, countries will be eager to ratify the BIT.                                                  23 This is the largest ICSID award to an investor and a direct consequence of the US-Ecuador BIT.  18 I argue that investor-state disputes before ICSID update the perceived costs of BITs for domestic opposition groups. ICSID is used far more than all other international arbitration venues combined (Allee and Peinhardt 2010). Its institutionalized structure with a centralized governing body enhances its effectiveness in resolving investor-state disputes, and its awards carry the same effect as the judgment of a national court (Blake 2013, Allee and Peinhardt 2010). In this chpater, I focus on a most important feature of ICSID disputes, which is the observability of proceedings and awards. ICSID routinely disseminates information through its Web site and various publications regarding details about the arbitration process of each ICSID dispute (Allee and Peinhardt 2011). While Allee and Peinhardt (2011) examine how foreign investors process this observed information, I argue that observing ICSID disputes registered after signing a BIT will also update the information regarding the costs of BITs for domestic opposition groups and increase their resistance to the ratification of the signed BIT. As a result, observing ICSID disputes should make the domestic ratification of BITs difficult.  ICSID disputes and their consequences may be inferred from statesÕ own experience or from the experience of peer countries. States may simply continue to be ambivalent about the costs of BITs even post signature. In other words, they might not realize these costs until they encounter them somewhat directly. Poulsen and Aisbett (2013) argue that states may not fully understand the risks of BITs until they themselves are hit by BIT-related claims. Poulsen and Aisbett (2013) find that states slow the pace of signing new BITs after experiencing their first BIT claims. If political executives and domestic actors follow BIT-related claims, ratification should similarly be affected. During this ratification process, if the host state itself is hit by investor-state dispute claims under other BITs that are concluded with other countries, it may update its assessment of the potential costs of ratifying new BITs.   19 First, host states may realize that BITs are indeed legally binding and biting. Violation of BITs becomes more clearly a menacing threat that is subject to legal actions. Second, even if executives still have an incentive to ratify a BIT despite increasing costs, it may take greater effort and more time to persuade domestic groups since the potential risk of BITs may be a salient issue in domestic politics. Finally, as already discussed, the financial cost of appearing before international arbitration such as ICSID can be substantial. Moreover, states need to invest time and resources to get familiar with international legal system and they may end up paying a substantial compensation if they lose the lawsuits.  An example of how BIT-related claims slow down the ratification process of signed BITs can be seen in the BIT24 between Canada and India signed on June 16, 2007. ÒDespite signing this BIT more than 5 years ago and despite Canadian pressure, the Indian Parliament has declined to ratify this treaty and instead decided to reconsider some aspects of the Agreement that it found troubling,Ó said Elizabeth May, Green Party leader and member of the Canadian Parliament for Saanich-Gulf Islands.25 The resistance to ratification by the Indian Parliament is understandable. Vodafone served the Indian government with a Notice of Dispute under India-Netherland BIT in 2012 regarding retrospective tax legislation proposed by the Indian government.26 In response to this threat, the Indian Parliament is currently reviewing all such BITs.  Because in general BIT-related claims against host states are relatively rare events27, states may not learn much from their own experience with BIT related arbitration. However, host states                                                 24 It is also called Canada-India Foreign Investment Protection and Promotion Agreement (FIPA). 25 Source: http://www.greenparty.ca/en/media-release/2012-11-07/prime-minister-s-trip-fails-pressure-india-ratify-canada-india-fipa 26 Source: http://www.vodafone.com/content/index/media/vodafone-group-releases/2012/bit.html 27 Between 1990 and 2012 there were at least 564 international arbitrations filed by investors  20 can also update their information of the potential costs of BIT claims by observing the consequence of BIT claims in peer countries, especially when those disputes are registered at the ICSID, a very transparent and observable international arbitration institution. Peer countries may be identified as groups of states that are geographically close or countries with a similar level of domestic property right protections. While the experience of peers may be considered somewhat less relevant, there are usually many more cases out of which countries could extract information. Also, information regarding the costs of BITs tends to spread well around neighboring countries as well as countries that are at a similar risk of violating investment protection for foreign investors and, thus, are likely encounter costly lawsuits in the future. For instance, Investment Treaty News published by a non-profit research institute named the International Institute for Sustainable Development (IISD) interviewed three lawyers in 2008 to seek their opinions on whether Brazil should begin ratifying BITs. When being asked about the potential negative consequences of Brazilian BIT ratification, Pedro Alberto Costa Braga de Oliveira, a Brazilian lawyer, noted ÒThe treaties would need to be so-called next generation treaties. In other words, they must be crafted to allow for sufficient governmental policy space, so that what has happened in Argentina28, for example, doesnÕt happen in Brazil.Ó29  My first two hypotheses follow this discussion: H1: The more claims countries face before the ICSID after signing a particular BIT, the more difficult the ratification of the BIT will be.                                                  against at least 110 host states (Wellhausen forthcoming).!28 According to Wellhausen (forthcoming), Argentina has the most public arbitrations filed against it (about 56) from 1995 to 2011. The majority of those lawsuits emerged due to its 2001Ð2002 default and financial crisis. 29 Source: https://www.iisd.org/itn/2008/11/30/investment-arbitration-in-brazil-yes-or-no/  21 H2: More ICSID claims faced by the peer group (that occur after two countries sign a BIT) are associated with a more difficult ratification.  2.3.3 Competitive Pressure and Ratification of BITs In addition to information from ICSID disputes, capital-importing states can also update their cost-benefit calculation by observing ratification practices of BITs in countries that are direct competitors for foreign direct investment. A capital-importing state may be eager to ratify a BIT once its economic competitors have done so. Haftel (2010) argues that ratified BITs (versus merely signed treaties) are more likely to generate ex ante and ex post costs30, which enhances the credibility of investment-friendly commitments and increases the inflow of FDI into countries that ratify the BITs. Thus, states may see that ratified BITs are a credible commitment as opposed to other similarly situated countries in the competition for foreign capital. From the perspective of foreign investors, countries that are economic competitors are many times close substitutes for investment. Hence, investors can choose to locate themselves in a country with a more credible commitment to favorable treatment of investors.                                                  30 The ex ante costs may include: (1) reducing autonomy of regulation and lawmaking for domestic purposes which may be politically costly for politicians who pursue BITs, (2) reducing competitive advantage of domestic and preexisting foreign investors who have invested in knowledge of local legal system while foreign investors following BITs operate under international legal system, which may generate opposition from domestic and preexisting foreign investors (Kerner 2009), (3) investing resource in negotiation of BITs despite that it may be small because countries may have standard format of BITs, (4) encountering barriers to ratification such as opposition from labor unions and other anti-globalization groups (Haftel 2010). The ex post costs may include: (1) financial cost of going through legal procedure in third-party adjudicating body when there are investor-state disputes, (2) compensation when lose in lawsuits of investor-state disputes in binding international arbitral institutions, (3) damage to diplomatic relations with home states, (4) others including international reputation costs, possibility of diminished credit rating (Allee and Peinhardt 2010, Elkins, Guzman, and Simmons 2006, Haftel 2010, Kerner 2009).  22 Following Elkins, Guzman, and Simmons (2006)Õs theoretical logic that competition for capital drives the signing of BITs, I argue that rapid BIT ratification by economic competitor-states increases the observed stateÕs perception of the benefits of ratifying the signed BIT as a means of diverting foreign capital from those competitors (and thereby avoiding the costs of missed investment opportunities). An intense ratification activity in countries that are economic competitors will pressure the observed state to push for ratification in domestic politics and thus shorten the time between signing and ratifying BITs. For instance, the UK-Colombia treaty was approved by ColombiaÕs Congress (Law 1464 of 2011). In the document submitted to Congress for approval, the Colombian government noted that ÒÉ the competition for attracting FDI among developing countries becomes a major concern of post crisis (2008 financial crisis).Ó31  I propose therefore the last hypothesis: H3: A greater number of ratified BITs in the economic competing group increases the likelihood that the BIT is ratified after signing.  2.4 Data, Measurement and Research Design  Haftel and Thompson (2013) is the most comprehensive effort to systematically analyze the determinants of ratification spells between signing and ratifying BITs. In order to be consistent with their work and allow some comparisons, I include all their independent variables as control variables while testing my theoretical arguments. Also similar to their research design, the scope of my data is all BITs from 1959, the year in which the first BIT was signed, to 2007.32 There are 2,595 BITs in this time span. However, quite different from Haftel and Thompson (2013), the unit                                                 31 The original wording was in Spanish: ÒÉ la competencia por la atraccion de IED entre los paises en desarrollo se convierte en una de las principales preocupaciones de la pos crisis.Ó 32 The year coverage is also dictated by the availability of the key independent variable Ð ICSID claims. This variable is drawn from Allee and Peinhardt (2011)Õs replication data file in which this variable is measured until 2007.   23 of analysis is signed BIT-year in my analysis because my key independent variables vary yearly. This setup enables me to test my theoretical arguments regarding how states update their perceptions of costs and benefits of BITs as conditions changes during ratification process. The following part presents variables, data source, and empirical estimation methods. 2.4.1 Dependent Variable  Relying on UNCTAD sources, Haftel and Thompson (2013) use the number of months passed from signing to entry into force (mutual ratification) as the dependent variable. Although operationalizing the dependent variable in the same way would allow some comparisons between their empirical results and mine, I measure the number of years rather than months passed from signing to entry into force because my key independent variables of interest vary yearly. 2.4.2 Independent Variables  To test H1, I construct a variable that counts new BIT claims filed against both governments in the dyad in a given year. This variable is a simple sum of the annual claims for each of the two states in a given year. This operationalization captures how each government in the dyad absorbs the updated information regarding the costs of BITs based on their direct experience with BIT-related claims and thus affects the time span of mutual ratification.33  In order to examine H2 - that BIT-related claims in peer states may delay the ratification of a signed BIT - I identify those peer states in two broad ways: (1) the observed state may pay close attention to BIT claims brought against states in its immediate vicinity. Information regarding the risk of litigation transmits well to neighboring states. To capture the geographic closeness of information transmission of the costs of BIT claims, I calculate a stateÕs average regional ICSID                                                 33 Haftel and Thompson (2012) employ the same way to construct BIT claims variable in their dyadic data structure and examine its impact on renegotiation of BITs. However, my empirical result is similar when only BITs claims against host state in a given year are used (Table 2-7).   24 claims in a given year based on United Nation classification of region34 (exclude the observed country under study).35 (2) the observed state may also be attentive to BIT claims in countries to which its level of domestic property rights protection is similar. It is because it is likely to be in a similar foreign investor risk-profile, and hence a similar exposure to the risk of encountering BIT claims. I use the level of democracy and domestic judicial independence to locate states into similar risk categories of potentially violating investment protection rights in BITs. A more democratic country is associated with a lower political risk and higher level of property protection (Jensen 2003, Li and Resnick 2003). I argue that countries with similar levels of democracy are also likely to have similar levels of property rights protection. Using polity2 score from Polity IV dataset to capture the level of democracy, I divide states into four groups (polity2 scores between -10 and -6, -5 and 0, 1 and 5, and 6 and 10). Then I calculate annual group-average ICSID claims for each state (exclude the observed state). Additionally, an independent judicial system is likely to produce a high level of domestic property rights protection (Linzer and Staton 2011). I group countries into four categories based on the Latent Judicial Independence (LJI) index (Linzer and Staton 2011) using 25th, 50th, 75th, and 100th percentile. Then I calculate the annual average ICSID claims of the category to which a state belongs (exclude the observed state). Similar to testing H1, I construct the value of this variable at the dyad level by summing up the annual average claims in each of the                                                 34 Those regions are Australasia, CIS and SEE, Caribbean, East Asia, Europe, Latin America, North America, Northern Africa, Oceania, South-east Asia, South Asia, Sub-Saharan Africa, and West Asia. Poulsen and Aisbett (2013) also use this classification of region in their paper.  35 The result is robust when I use neighboring states to identify geographic closeness. Following Kerner (2009) I define neighboring countries using the Correlates of War coding for type 1 or type 2 contiguity, which includes countries that share a land border or are separated by 12 miles of water or less.  25 two statesÕ peer states in a given year.36 The data of ICSID case variables is drawn from Allee and Peinhardt (2011).37 In order to test H3 that host states compete to ratify BITs, I follow Elkins, Guzman, and Simmons (2006)Õs construction of export product similarity to identify host statesÕ economic competing countries. This index captures the degree to which states export the same basket of goods and I update this index until 200738. Following mainstream works that utilize dyadic data structure in BIT literature (Elkins, Guzman, and Simmons 2006, Haftel and Thompson 2013, Jandhyala, Henisz and Mansfield 2011), I define the country with a lower GDP per capita in a dyad as the host state that is likely to be on the recipient side of capital flow. The rationale behind this index in capturing the competition logic is that foreign investors may view host states with similar export products as close substitutes for their investment (Elkins, Guzman, and Simmons 2006). Following Elkins, Guzman, and Simmons (2006)Õs formula for generating spatial variables, I then use this export product similarity index as the weights to calculate the weighted number of new ratified BITs in the observed host stateÕs economic competitors in a given year39.                                                   36 The result is similar when only BITs claims in host statesÕ peers are used (Table 2-7).  37 The original data source includes the ICSID Web site, the International Institute for Sustainable Development, Investment Treaty Claims, American Society for International Law, and the Institute for Transnational Arbitration.  38 I calculate the correlation between countries for each year across fifteen indicators from World Development Indicators (WDI): agricultural raw materials exports; arms exports; communications, computer, etc.; food exports; fuel exports; high-technology exports; insurance and financial services; international tourism, receipts; manufactures exports; ores and metals exports; computer, communication, and other service; transport services; transport services; travel services; and travel services. The correlation runs from -1 to 1 and captures the similarity between two states with regard to their export product profile. 39 The result is robust when a simple proxy of annual average regional new ratified BITs is used to capture this competition logic. States in the same region are very likely to have similar economic structure, infrastructure, and factor endowments that are attractive to foreign investors. Hence, foreign investors may consider those regional states as substitutes for their investment.  26 2.4.3 Control Variables  Following Haftel and Thompson (2013), I control for the following variables: (1) legislative hurdle that captures the constitutional requirements for treaty ratification40. The data source is Hathaway (2008) and Haftel and Thompson (2013). (2) ratification ratio that is measured as  the percentage of signed BITs that are in force before the signing of the observed BIT. A good track of successful ratification in the past could be a good indicator of the ratification fate of the observed BIT. (3) level of democracy that is captured by polity2 score from Polity IV dataset. Decision-making processes in democracies are argued to be relatively open and transparent due to institutional mechanisms such as a free press, which allows the partner state to estimate preferences of democratic state and likelihood of successful ratification at the moment of negotiating the BIT. (4) natural logarithm of GDP (Penn World Tables 7.0). A higher value indicates a larger capacity that states can invest to collect information regarding potential ratification obstacles and then address them in the negotiation process, which would speed up subsequent ratification. (5) government expenditure (World Development Indicators). Similar to the GDP variable, this variable is also used to gauge the capability of states to anticipate potential ratification obstacles during the negotiation process.41 (6) home countryÕs net FDI outflows as a proportion of GDP42                                                 40 This variable is coded in the following way: 0 if no legislative approval is required, 1 if a simple majority in one house is required, 2 if a simple majority in two houses is required, and 3 if a supermajority in one or two houses is required.  41 Following Haftel and Thompson (2013), the first five control variables (legislative hurdle, ratification radio, level of democracy, GDP, and government expenditure) are constructed at the dyadic level by using the Òweakest linkÓ assumption. That is, the value of the variable for two countries in a dyad that would determine the eventual ratification is employed. For instance, a higher value of legislative hurdles in a dyad (a pair of countries) is used because it is this country would determine the eventual date of mutual ratification and entry into force of the BIT. 42 Similar to Haftel and Thompson (2013) as well as other works in the literature of BITs, I define the home country in a dyad as the one with a higher level of GDP per capita while host country as the one with a lower GDP per capita.   27 (UNCTAD). Home states with a higher volume of outflow FDI may find investment treaties more valuable in order to protect their investments abroad. (7) host countryÕs net FDI inflows as a proportion of GDP (UNCTAD). Host states with a higher volume of inflow FDI may face greater pressure from foreign investors and their home governments to quickly ratify the BITs. (8) common law system (Haftel and Thompson 2013). Some scholars argue that states with a common law system offer better domestic legal protection than other types of legal system, which renders external guarantees such as BITs unnecessary (Elkins, Guzman, and Simmons 2006, Goodliffe and Hawkins 2006, and Simmons 2009). Thus, these states would be less willing and pressured to ratify BITs. This variable is coded as 1 if at least one state in the dyad has a common law system, otherwise 0. In addition, I also include some dyadic controls: (i) common language (Haftel and Thompson 2013). Shared affinity may reduce legal ambiguities and divergent interpretation in the ratification process. It is coded 1 if the two countries in a BIT share a formal language, otherwise 0. (ii) colonial ties (Haftel and Thompson 2013). A shared colonial heritage may result in a similar legal and political culture and enhance mutual understanding. It is coded 1 if the two states in a BIT share a colonial heritage. (iii) alliance (Haftel and Thompson 2013). The existence of a formal alliance between two states in a BIT reduces the perceived risk of cooperation. This variable is coded 1 if two states in a BIT are in a formal alliance, otherwise 0. (iv) the development gap that is measured as the natural log of the difference in the GDP per capita of two states in a BIT (Penn World Tables 7.0). An uneven level of economic development measured as GDP per capita indicates a relative abundance of skilled labor in the home state, which is conventionally argued to be beneficial for vertical FDI. In addition, home states with much higher levels of economic development may have substantial bargaining leverage to pressure the host state to ratify the BITs in order to protect  28 foreign investors from home states. Thus, a larger development gap in a BIT may facilitate the speed of ratification. Additionally, I also include a control variable of cold war, which has a score of 1 before 1990, and 0 thereafter. There may be some systematic change in terms of increasing global FDI flows, the number of BITs, and globalization as cold war ends. The summary statistics for all variables is shown in Table 2-2.  Table 2-2. Summary Statistics  2.4.4 Estimation Method  An event history model is appropriate for this analysis because it deals with BITs that drop out of the analysis after they are ratified and right censored (BITs that never get ratified under the observation time period). Event history models estimates the ÒriskÓ that an event will occur as time elapses. I use a Cox model with robust standard errors in order to examine my BIT-year data structure. However, a Cox model assumes the relative hazard of covariates are proportional over  29 time. A violation of this assumption will lead to biased estimates for both the offending variables and other variables in the model (Box-Steffensmeier and Zorn 2001). Similar to Haftel and Thompson (2013), I test the proportional hazard assumption with Schoenfeld residuals. If this assumption is violated by some variables, following standard practice I interact those variables with the logged form of time duration.43  2.5 Results and Discussion  Table 2-3 shows the main results of Cox nonproportional hazard models. I first replicate Haftel and Thompson (2013)Õs main model and then introduce my key independent variables into the models. Coefficients are shown in all survival models in which a coefficient that is greater than 0 indicates a higher probability of ratifying a BIT as the values of independent variables increase. Model 1 replicates Haftel and Thompson (2013).44 Largely confirming Haftel and Thompson (2013)Õs empirical results, I find a dyad with a lower level of democracy, the ratification ratio, GDP, and development gap, and a higher requirement of legislative approval are more difficult to be ratified while a dyad that is an alliance facilitates ratification45. I then proceed to test my variables of interest.  Model 2 examines the impact of filed ICSID cases in the dyad while models 3-5 investigate the impact of filed ICSID cases in peer states on ratification speed of a signed BIT. As expected, in model 2 filed ICSID cases in the dyad variable takes on a negative sign indicating that an increase in the number of lawsuits filed in international arbitration institutions tends to slow down                                                 43 My key independent variables of interest do not violate this assumption. Alternatively, I also employ a logistic model with a cubic polynomial approximation to account for temporal dependence (Beck et al. 1998) in the robustness check. The results of these two statistical techniques are similar.   44 Model 1 in Haftel and Thompson (2013)Õs Table 1.  45 Cold war dummy is also a significant determinant of BITs ratification.   30 the ratification pace of a signed BIT. However, this variable falls just short of conventional statistical significance with a p-value of 0.117. This variable passes a 90% level of statistical significance only in model 11 when an alternative estimation technique of logistic model is used in the robustness check. We thus have limited confidence in making a definite conclusion that both governments in the dyad update the costs of BITs with the experience of being respondents in investment tribunals post signature, which in turn affects the ratification speed of the signed BIT. In contrast to the weak evidence of filed ICSID claims in the dyad variable, I find strong evidence that filed ICSID claims in the peer states would make it more difficult for states to ratify a BIT. Models 3 Ð 5 show the results of ICSID claims in three peer groups Ð regional peers, peers with similar level of judicial independence, and peers with similar level of Polity2 score, respectively. As expected, the coefficients of filed ICSID claims are negative and significant at above a 99% level of statistical significance. The contrasting findings on the impact of post signature filed ICSID claims and ICSID claims in peer states on the ratification fate of the signed BIT may seem somewhat surprising. Haftel and Thompson (2012) find that ICSID claims against both states in a BIT dyad increase the chance of renegotiating the BIT. Poulsen and Aisbett (2013) reveal that states tend to slow down the signing pace of new BITs after encountering their first BIT-related claim. In addition, Poulsen and Aisbett (2013) find that the average number of total claims in the region and the global number of BITs claims are negatively associated with participation in new BITs, suggesting that states may also learn policy outcomes from other countries in the same region and the world. Comparing to the decisions to sign or renegotiate BITs, ratification spells, however, are relatively short. Furthermore, since BIT-related claims against states are in general rare events, states may not have much chance of being hit by BIT claims during the relatively short ratification period after signing a BIT and then reconsider the costs of  31 BITs.46 However, states and domestic groups can update the costs of BITs from observing the consequence of BIT claims in their peer states that are either geographically close or in a similar risk of violating investment protections. This new post signing information from peer countries are relatively rich and updates the costs of BITs for states and domestic groups, leading to delays in ratification of a signed BIT.  Models 6 to 8 investigate these two hypothetical contentions regarding the impact of ICSID claims in the dyad and in the peer states together by including them in the same model. The results are very similar. Post signing ICSID lawsuits in peer states delay the ratification of a signed BIT while ICSID lawsuits in states themselves have no significant effect. With respect to Hypothesis 3, I find consistent evidence across all models (models 2-8). States respond to the new post signing information of ratification activities in economic competitors in their decisions and efforts to push for ratifying a signed BIT. The number of ratified BITs in economic competing states facilitates the ratification of the observed BIT.                                                          46 The distribution of values for the filed ICSID cases variable in the sample: 0 (86%), 1 (10.6%), 2 (2.6%), and all other values less than 1%.   32 Table 2-3. Results of Cox Nonproportional Hazard Models 1970 Ð 2007      33 To illustrate the substantive effect of my key independent variables, I calculate the percentage change in the hazard rate associated with a change in the covariates. The following formula is used for this calculation: !"#$%&'()*+,-.%&'()*+/-%&'()*+/-0122 where X1 and X2 are the values of each key independent variable at one standard deviation below the mean and one standard deviation above the mean, respectively47.   Table 2-4. Percentage Change of Hazard Rate: Vary ICSID Cases Filed and Ratified BITs in Economic Competing States, and Legislative Hurdles Variable  Percentage change of hazard rate !"# 95% confidence interval  Model 3 - ICSID cases filed (regional peer) -47.74 (-50.41, -32.29) Model 4 - ICSID cases filed (similar latent judicial independence index peer) -41.68 (-50.23, -32.15) Model 5 - ICSID cases filed (similar Polity2 score peer) -41.85 (-50.55, -32.21) Model 5 - Ratified BITs in economic competing states  22.61 (6.06, 40.63) Model 5 Ð Legislative hurdles  -16.17 (-24.44, -7.06) Note: Numbers are generated based on simulations with 10,000 draws from the estimated coefficient vector and variance-covariance matrix. The numbers in parentheses are 95 percent confidence intervals.  Based on models 3-5 in Table 2-3, Table 2-4 displays the percentage change in the hazard rate for my key independent variable with 95% confidence intervals48. When the variable of filed ICSID claims in peer states changes from one standard deviation below the mean to one standard deviation above the mean, the hazard rate of ratifying BITs decreases at least 41%. However, the hazard rate increases by about 22% when the number of ratified BITs in economic competing states changes from one standard deviation below the mean to one standard deviation above the                                                 47 When one standard deviation below the mean is negative, I substitute it with the minimum value (zero).   48 Each confidence interval is generated based on a simulation with 10,000 draws from the estimated coefficient vector and variance-covariance matrix.  34 mean. I also calculate the percentage change in the hazard rate for the legislative hurdles variable as a baseline for comparison with the substantive effect of my key independent variables. Haftel and Thompson (2013) find that domestic legislative hurdles have a significant and substantive impact on BITs ratification. The hazard rate decreases by about 16% when legislative hurdles variables move from one standard deviation below to one standard deviation above the mean. All the calculated percentage changes are statistically significant as both of the upper and lower bounds of confidence intervals exclude zero. It is notable that the substantive effect of my variable of interest, the variable of filed ICSID claims in peer states in particular, is much larger than that of legislative hurdles variable. This lends support to my emphasis on the importance of post signing new information (ICSID claims in peer states and ratification rates in economic competing states) on the ratification pace of a signed BIT.  2.5.1 Robustness Check I verify the robustness of the empirical results in the following ways. First, I employ an alternative estimation strategy - a logistic model that includes the length of time to ratify and cubic splines to control for time dependence (Beck et al. 1998). Standard errors are clustered by dyad. This model estimates the probability of ratifying BITs. As shown in Table 2-5, the main results are similar. Second, similar to Poulsen and Aisbett (2013), I restrict the sample to BITs that are signed after 1989, since it was not until the late 1980s that the vast majority of BITs began to give investors direct access to international arbitrations without first having to exhaust local remedies. Thus, it might be argued that the information of the costs of BITs deriving from ICSID disputes only impact the domestic ratification pace of those BITs in which binding international arbitration venues are most likely to be available. My results are robust (Table 2-6).    35 Table 2-5. Results of Logistic Models with a Cubic Polynomial Approximation 1970 Ð 2007       36 Table 2-6. Robustness of Cox Nonproportional Hazard Models Ð BITs Signed After 1989  Third, it is possible that only the side of the capital receiver in a dyad is likely to update the costs of BITs when observing the post signing information of international arbitral disputes because the risk of potential investor-states disputes may apply to them in the future. As a result, the ratification of the signed BIT will be more difficult to achieve in capital-hosting states, leading  37 to the delay in reaching mutual ratification of the BIT. Therefore, I measure ICSID cases in the host state as well as its peer states for each dyad. The main results remain as shown in Table 2-7. Fourth, I control for time trend or time specific shocks that might affect BITs ratification by including half-decade dummies. The result are very similar (Table 2-8)49.  Finally, I explore the heterogeneity of BITs and examine how my key independent variables affect the ratification speed of BITs with varying levels of stringency. Allee and Peinhardt (2010) focus on the variation of Investor-State Dispute Settlement (ISDS) mechanisms across BITs and investigate the determinants of statesÕ decision to include in BITs the ICSID clause in resolving investor-state disputes. The presence of the ICSID clause in BITs represents high level of delegation and hence are considered more stringent. Consistent with Allee and Peinhardt (2010)Õs coding schemes, I extend their coding50 to 2,090 BITs relying on UNCTADÕs International Investment Agreements (IIAs) database and Kluwer Arbitration OnlineÕs investment treaty database. I construct a binary variable to capture the presence of the ICSID clause as either one of many options for international arbitration or the only venue for international arbitration51. Table 2-9 displays the results for signed BITs with the ICSID clause. I find consistent evidence that new information of peersÕ experience with BIT-related claims delays ratification while rapid ratification rates in economic competing states52 facilitates ratification. The results for the                                                 49 The variable of the number of ratified BITs in economic competitors falls just short of 90 significance level with a p-value of 0.120 when Polity2 score is used to calculate ICSID cases in peer states.  50 The original dataset covers 1,473 BITs.  51 The proportion of the updated variable that takes on a value of 0 is highly comparable with Allee and Peinhardt (2010) Ð 17% vs 16.5%. 52 Neumayer, Nunnenkamp, and Roy (2014) find that host country competes with one another by signing BITs with similar level of strictness (e.g., Investor-State Dispute Settlement mechansim). That is, a host country is more likely to sign a strict (not strict) BIT if its competitor has signed a strict (not strict) BIT. Therefore, I measure the number of ratified BITs with the ICSID clause in economic competing states to capture the competition logic argument here. ! 38 ratification speed of signed BITs without the ICSID clause are shown in Table 2-10. The sample size drops substantially (182 BITs) comparing to those with the ICSID clause (1,077 BITs). Although we need to be cautious in interpreting the results due to the relatively small sample size, I find support for Hypothesis 3 that the more ratified BITs in economic competing53, the more likely the signed BIT is to be ratified. Interestingly, the information of ICSID lawsuits in peer states also tends to slow down the ratification pace of those BITs without the ICSID clause. A potential explanation would be that BIT-related claims in peer states make the costs of BITs salient and visible for a host state. In response to this potential risk of litigation, the host state may review all signed BITs carefully regardless of the contents of BITs and engage in more relatively lengthy parliamentary debates.                                                             53 Similarly, I construct the number of ratified BITs without the ICSID clause in economic competitors here (Neumayer, Nunnenkamp, and Roy 2014).   39 Table 2-7. Robustness of Cox Nonproportional Hazard Models Ð ICSID Claims in Host Countries and Their Peer States         40 Table 2-8. Robustness of Cox Nonproportional Hazard Models Ð Control for Half-decade Dummies           41 Table 2-9. Robustness of Cox Nonproportional Hazard Models Ð Signed BITs with the ICSID Clause      42 Table 2-10. Robustness of Cox Nonproportional Hazard Models Ð Signed BITs without the ICSID Clause       43 2.6 Conclusion    Various scholars have examined the ratification of international treaties as a distinct outcome of international cooperation (Milner and Rosendorff 1997, Mansfield and Milner 2012, Haftel and Thompson 2013, Campello and Lemos 2015, Fredriksson and Gaston 2000, Neumayer 2002, Von Stein 2008, Boockmann 2001, Chau et al. 2001, Baccini and Urpelainen 2014, Schneider and Urpelainen 2013). This chapter contributes to this broad literature by investigating the domestic ratification process of Bilateral Investment Treaties (BITs). Over the past fifty years, BITs have become the most important international legal mechanism for protections of foreign assets and the facilitation of cross-border capital flows, especially from developed rich countries to developing states. BITs are negotiated and signed by leaders at the international stage but must go through the domestic ratification process for approval. The ratification process is important because only mutually ratified BITs are legally binding and expected to provide protection for foreign investors. Therefore, we need to have a better understanding of ratification process of BITs.  This chapter examines the determinants of the time span between signing and ratifying BITs. Specifically, I model domestic ratification as a dynamic process of updating the costs and benefits of BITs after the signature. Signature is a revealed preference that, on balance, leaders view benefits as outweighing the costs of the BIT. However, domestic actors may update their assessment of costs and benefits as time passes by after the signature. The strong finding shows that domestic actors may, after signing a BIT, update their assessment of the associated costs by observing ICSID claims against peer countries that are either geographically close or in a similar risk profile of violating investment protection. This updated information delays ratification of the observed BIT. However, the impact of BIT claims against states themselves on the duration of ratification is relatively weak. A possible explanation would be that states are very unlikely to  44 experience BIT-related claims themselves during the relatively short ratification spells since BIT claims against states are in general rare events. Instead, states and domestic groups are likely to update the costs of BITs by referring to the relatively rich BIT claims from peer states, leading to delays in ratification of the signed BIT. In addition, a host state is more likely to ratify a BIT after signing as the number of BITs ratified in economic competitors increases. BITs ratification rates in economic competing states update the observed stateÕs perception of the benefits of ratifying the signed BIT as a credible commitment by raising the expected return on investment due to the possibility of investment diversion.  The implication of this study goes beyond the ratification process of BITs and can apply to international treaty ratification in general. Exogenous or unanticipated information following the signature may change political executivesÕ initial cost-benefit equilibrium. This study also sheds light on the role that international institutions (e.g., international investment arbitration institutions) play in providing information for domestic audiences54. Furthermore, this chapter considers international treaty ratifications among states as being interdependent. States learn the consequences of international treaty commitments and ratification pace in other similarly situated states, and incorporate this information into the cost-benefit analysis of ratifying their own treaties.                                                      54 Allee and Peinhardt (2011) study how foreign investors process the information from international investment arbitration institutions in their evaluations of capital-hosting statesÕ credibility of commitments.   45 CHAPTER 3  THE IMPACT OF BILATERAL INVESTMENT TREATIES ON HUMAN RIGHTS PRACTICES IN DEVELOPING COUNTRIES  3.1 Introduction In the absence of multilateral institutions, bilateral investment treaties (BITs) have been one of the most visible and powerful legal instruments governing the global growth of FDI (Elkins et al. 2006). Direct investment in a foreign country implies important sunk cost and BITs are designed to address investorsÕ concern about the future behavior of host states (Elkins et al. 2006). Common policy reversals that are adverse to investors include expropriation of foreign-owned assets and discriminatory changes to performance requirements, capital taxation and regulation, tariffs or social contributions. Since its inception in 1959 (Germany and Pakistan BIT), countries have ratified a large number of BITs. Currently, the United Nations Conference on Trade and Development (UNCTAD) estimates that 178 countries are involved in at least one BIT, with more than 2900 BITs signed among these countries.  In the hopes that BITs can promote much needed capital to poor countries, some view these investment treaties as a development tool (Leo 2010). Others, however, argue that the favorable treatment given to foreign investors has the potential to worsen the environmental or human rights practices of states (Peterson and Gray 2003). While the earlier literature has mixed findings (Hallward-Dreimeier 2003, Neumayer and Spess 2005, Tobin and Rose-Ackerman 2005, Yackee 2007), the most methodologically sophisticated recent evidence shows that indeed countries signing more BITs see a greater inflow of foreign direct investment (Buthe and Milner 2009,  46 Kerner 2009, Allee and Peinhardt 2011, Haftel 2010).55 Yet, researchers have given much less systematic attention to the potential negative externalities of the Òbroad and asymmetricalÓ56 rights obtained by investors through BITs. This chapter is the first systematic theoretical and empirical investigation into how and whether BITs may affect statesÕ human rights practices. Bilateral investment treaties include provisions that guarantee investor rights as well as mechanisms that investors can use to legally enforce such provisions (Elkins et al. 2006). To a great degree, then, developing countries have signed BITs to attract foreign investment and they have done so primarily because of competitive pressures (Elkins et al. 2006). Recent work shows that host states may have not fully anticipated the constraining effects and costs of such investment treaties (Jandhyala et al. 2011, Poulsen and Aisbett 2013), and started to overtly push back against the constraining effect of BITs on their domestic policy space (Simmons 2014, Poulsen and Aisbett 2013). In addition, human rights groups have charged that the hands of capital importing states are tied by investment treaties, generating important grievances and worsening governmentsÕ human rights practices. For example, the UK Ð Colombia BIT was signed in 2010 but is not yet ratified. Human rights and anti-poverty groups57 are concerned that this BIT containing an international arbitration clause will expose the Colombian government to costly lawsuits, impact Colombian land reform programs, and threaten the return of 5 million internally displaced people. Similarly, NGOs58 have reservations about the ongoing negotiations on a US-India BIT, including about how                                                 55 The legally binding favorable treatment accorded by BITs to investors and a growing number of costly litigations in the last decade have made host countries re-think the consequences of ratifying such treaties and slow-down the pace of treaty signing (Simmons 2014, Poulsen and Aisbett 2013).   56 Simmons (2014) p. 12.!57 E.g., Colombian NGO Cedetrabajo or British NGO Traidcraft. 58 E.g., the Indian NGO ÒForum against Free Trade AgreementsÓ, a coalition of over 75 organizations.! 47 the investor-state dispute mechanism can undermine the domestic policy space and domestic justice system. Yet, to date, there is no consistent evidence that the global investment regime has had such deleterious effects.   My key contention is that BITs have the potential to negatively influence human rights practices because they lock in legally enforceable conditions attractive to investors, both retrospectively and into the future. The lock-in effect of BITs can force the hand of the government to favor multi-national corporations or foreign investors even at the cost of violating the rights of their own citizenry. Retrospectively, many developing countries compete for investment and trade on issues ranging from environmental regulations to labor standards and welfare spending and tend to be destinations of vertical investment seeking cost efficiencies. BITs lock in these initial favorable conditions. In addition, BIT provisions constrain future policies, from the provision of welfare benefits, basic infrastructure and investment in environmentally friendly technologies to land reform. Locked-in low standards for environmental protection or labor rights and constrained policies are important sources of popular grievance in host states. The literature on the causes of repression suggests that human rights violations are key responses of states to the manifested or just anticipated protest that can results from these grievances. I will argue however, that statesÕ reaction to such potential dissent and the negative human rights consequences of BITs is moderated by regime type.  I test my hypotheses on a sample of 113 developing countries from 1981 to 2009. I find that countries that ratify a greater number of BITs have worse human rights practices. I also show that the effect of the cumulative number of ratified BITs is conditional on political regime: BITs are more likely to result in human rights violations in non-democracies. My results are robust to the  48 inclusion of a large number of control variables, coding of BIT specific clauses59 or focusing only on North-South BITs that likely govern over de facto investment flows, instrumental variable techniques, exclusion of outliers, variations in sample size or alternative measures of human rights practices. The chapter makes several contributions. It is the first to systematically theorize and test the effect of the global investment regime on statesÕ human rights practices. A plausible motivation for joining international treaties, including BITs is frail domestic institutions and low credibility with investors. Yet the evidence in favor of a credible commitment rationale for BITs is weak for variables ranging from political institutions (democracy or political constraints) to economic risks (property rights or law and order) (Table 3-1). Specifically, no evidence exists that countries with bad human rights records sign or ratify BITs for credibility reasons (Vadlamannati 2009, Neumayer 2006). My work however shows theoretical reasons and robust empirical evidence for an effect running from BITs to human rights violations. This research thus contributes to recent work investigating the effect of international economic treaties60 or international organizations61 on statesÕ human rights practices.  Second, recent work argues that states ratify BITs within a bounded rationality framework and do not appreciate the degree to which their hands are tied (Poulsen and Aisbett 2013). My evidence raises doubt with regards to these arguments, especially the strong claim of bounded rationality, because it appears that host states that have signed more BITs protect investor rights, even at the cost of violating the rights of their domestic populations.                                                  59 I code BITs that include a direct arbitration clause to the International Centre for Settlement of Investment Disputes (ICSID). 60 Hafner-Burton (2005b), Hafner-Burton (2009), Spilker and Bohmelt (2013). 61 Abouharb and Cingranelli (2006, 2007, 2009), Keith and Poe (2000).  49 Table 3-1. Evidence That Host Countries with Low Credibility Sign BITs (Credibility Argument)   Finally, I unpack the specific causal mechanisms through which FDI may affect human rights practices. In my empirical models I allow for a direct effect of FDI inflows which appears to aid human rights practices. This likely occurs via better economic development and growth, as discussed in the literature (Li and Liu 2005, Alfaro 2003, Alfaro et al. 2004). FDI has, however, been linked to human rights through opposing arguments that are unlikely to be captured by the variables used in the literature. My focus on BITs can capture directly the preferential and favorable treatment62 that many multinationals enjoy in developing countries and that is locked in by BITs.  The chapter proceeds as follows:  I first introduce BITs and then review relevant literature. I then explain how BITs may affect human rights practices and derive two hypotheses. Data and                                                 62 Simmons (2014), Van Harten (2012).! 50 research design are discussed next, followed by the empirical findings. The final section concludes this chapter.  3.2 Bilateral Investments Treaties (BITs) Broadly, BITs are concluded in order to facilitate the flow of cross-border foreign direct investment by guaranteeing certain standards of protections.63 Those treaties help to solve the time-inconsistency problem that plagues the growth of FDI, and enable capital-hosting states to make credible commitments of protecting investment rights. The first BIT was signed between Germany and Pakistan back in 1959. Since then, more than 2,900 BITs have been signed involving 178 countries. The pace of BITs signing grew rapidly with an average of more than 100 BITs a year during the 1990s compared to about 20 BITs per year from 1959 to the late 1980s (Jandhyala, Henisz and Mansfield 2011). However, the pace of BIT signing has slowed as the number of investment dispute arbitrations ruled against states has increased in the past decade (Jandhyala, Henisz and Mansfield 2011, Poulsen and Aisbett 2013).  Many BITs include investment protective clauses such as fair and equitable treatment, compensation for expropriation (direct and indirect), national treatment of foreign investors, and most favored nation treatment. Even more, investors can enforce their rights in a timely manner and through investor chosen venues that are unlikely to favor host states: Early BITs provided investor protection through state to state dispute resolution, via the establishment of tribunals or submission to the International Court of Justice. More recent BITs grant foreign investors the right                                                 63 Initially BITs were signed between capital importing developing countries on the one hand and capital exporting developed countries, on the other. More recently, there are also many BITs signed between two developing countries (especially during the 1990s) for reasons other than to encourage FDI (Jandhyala, Henisz, and Mansfield 2011).   51 to adjudicate alleged violation of rights in international tribunals64, without the need to exhaust local remedies, and, in case of non-compliance with the arbitration decisions, broad rights to request the confiscation of host governmentÕs property from around the world.65  By the late 1980s most BITs include such dispute settlement mechanisms and between 1990 and in 2012 there were at least 564 international arbitrations filed by investors against at least 110 host states (Wellhausen forthcoming).66 These investment arbitrations are directly related to the number of BITs ratified by states (Simmons 2014) and have important consequences. A first implication is for the budgets of host states: The size of monetary awards have been significant including recent decisions against the governments of the Czech Republic ($350 million in 2001), Lebanon ($266 million in 2005) (Elkins et al. 2006), or Ecuador ($2.3 billion in 2012)67. Second, future direct investment inflows decrease when investors allege violation of rights by the host government even at the moment of filing of an international arbitration (Allee and Peinhardt 2011). Third, because arbitration is a high-risk, high-cost option, the threat of arbitration on the side of investors can be effective in extracting concessions, even when governments may expect to win in                                                 64 The International Centre for Settlement of Investment Disputes (ICSID) in particular is a favorable venue for investors because of its very centralized structure as an international organization, affiliation with World Bank (imposing additional leverage on states who may anticipate future World Bank financing), and very limited grounds for appeal (Allee and Peinhardt 2010). 65 BITs typically also have grandfather clauses that ensure treaty obligations apply to FDI for 10 to 15 years after a treaty expires or is abrogated by a host government. 66 Most of the claims against host states are brought by investors to the International Centre for Settlement of Investment Disputes (ICSID), which is part of the World Bank. The next used option is the arbitration rule of United Nations Commission on International Trade Law (UNCITRAL) in venues such as the International Chamber of Commerce, the Stockholm Chamber of Commerce, and the London Court of International Arbitration.  67 This is the largest ICSID award to an investor and a direct consequence of the US-Ecuador BIT.  52 front of a tribunal. While most such threats remain confidential, several examples have emerged and experts estimate that the practice is not uncommon.68  Not only do foreign investors have favorable choices to enforce their rights, but host states may be disadvantaged over investors in the litigation process. Most BITs are signed by developing and middle income states, which lack the legal capacity and experience to effectively counter the investor-states claims. Even more, in actual litigations, there is some degree of legal asymmetry towards foreign investors. For example, Van Harten (2012) examines jurisdiction issues that by definition could be interpreted either way and litigated precisely because parties cannot readily anticipate the outcome. He finds a bias towards expanding investorÕ rights of interpreting jurisdictions that applies to 76 percent of the cases (out of 140 cases as of 2010). Simmons (2014) uses Van HartenÕs data and finds that the poorer the state, the more likely an arbitration panel will rule expansively in favor of investorsÕ rights. She also finds that, compared to multilateral agreements such as NAFTA, BITs-based complaints are more likely to be associated with larger monetary awards. While BITs offer strong protection to foreign investors, human rights provisions in BITs are marginal at best, painting a stark contrast to preferential trade agreements (PTAs). In practice very few, if any, BITs mention human rights or associated fields (Jacob 2010) and many developing countries would like to see BITs include more obligations for investors (Milner 2014). For instance, no explicit reference to human rights is to be found in the country model BIT of Germany                                                 68 Recent examples include investor arbitration threats from open pit mining companies in Indonesia and Costa Rica regarding regulation changes; telecom operators in Zimbabwe over the cancellation of their license; or an energy company in India over changes to taxation policy.  Also, Luke Eric Peterson, publisher of the Investment Law Reporter notes that ÒI would not be the least bit surprised if there were dozens upon dozens of such informal treaty-uses for every claim that actually gets arbitratedÓ (Gallagher and Shrestha 2011, p.5).!! 53 (2008), France (2006), China (2003), India (2003), the United Kingdom (2005), or the United States (2004). An exception is the 2007 draft of the Norwegian model BIT (Jacob 2010), which mentions human rights practices in preambular language.69 However, such preambular wording does not amount to substantive provisions on human rights standards and, by itself, is too weak to compel compliance with human rights standards for either foreign investors or host states. While there is a lack of references to human rights, related provisions (labor standards, environmental protection) are occasionally and increasingly mentioned in several recent BITs or template for BITs. For instance, article 11 of the 2004 Canadian model BIT mentions that ÒThe Parties recognize that it is inappropriate to encourage investment by relaxing domestic health, safety or environmental measures. Accordingly, a Party should not waive or otherwise derogate from, or offer to waive or otherwise derogate from, such measures as an encouragement for the establishment, acquisition, expansion or retention in its territory of an investment of an investor. If a Party considers that the other Party has offered such an encouragement, it may request consultations with the other Party and the two Parties shall consult with a view to avoiding any such encouragementÓ.70 This language can hardly be construed as similar to ÒhardÓ PTAs that explicitly link material benefits to compliance with human rights standards.                                                    69 ÒReaffirming their (treaty partiesÕ) commitment to democracy, the rule of law, human rights and fundamental freedoms in accordance with their obligations under international law, including the principles set out in the United Nations Charter and the Universal Declaration of Human RightsÓ. The treaty text is obtained from International Investment Agreements (IIA) database maintained by UNCTADÕs IIA Section. 70 The treaty text is obtained from International Investment Agreements (IIA) database maintained by UNCTADÕs IIA Section. The 2004 model BIT of United States contains similar provisions Ð domestic environmental law (article 12) and domestic labor laws (article 13).  The 2012 US model BIT states that states should not waive labor and environmental standards in order to encourage investment (Simmons 2014).!! 54 3.3 Literature Review  3.3.1 Foreign Direct Investment and Human Rights The economic rationale for the globalization of trade and investment is a more efficient allocation of resources or gains from specialization and economies of scale. Supporters of globalization have also argued that it improves human rights practices, while detractors voiced concerns about human rights degradation. The evidence on the effect of FDI on human rights is mixed, although recent studies support a positive impact.71 Work that traces itself to dependency theories and Òrace to the bottomÓ ideas argues that foreign investors co-opt local elites and extract local resources (Bhattacharya et al. 1997) or, alternatively, use exit threats as leverage for tax breaks, favorable labor policies and wages, and fewer welfare programs (Haggard and Maxfield 1996). To sustain such investor-friendly policies governments arguably need to control the masses, including through "repression and curtailment or denials of civil and political rights for the populations of developing countries" (Meyer 1996, p.379). On the other hand, FDI is suggested to improve governmentsÕ respect for human rights through several mechanisms. A key argument is that market liberalization brought by trade openness or FDI boosts economic development, which promotes better human rights practices (Apodaca 2001). Although not unequivocally, FDI is associated with better growth outcomes in a wide range of situations72 and the improvement in human rights may happen because development sustains the middle class, which demands more respect for political and human rights (Richards et                                                 71 A positive effect of FDI on human rights practices is identified in Richards et al. (2001), Hafner-Burton (2005a). Sorens and Ruger (2012), however, find no effect. 72 Li and Liu (2005) find a direct positive effect of FDI on growth and an indirect through human capital. Alfaro (2003) shows a positive effect on growth, coming from FDI in manufacturing and a negative effect from FDI in the primary sector. Alfaro et al. (2004) also finds that FDI is likely to promote growth in countries with well-developed financial markets. Still, other work, finds no direct effect of FDI on economic growth (Carkovic and Levine 2005).!! 55 al. 2001, Meyer 1996). Alternatively, economic development produces political stability by improving living standards, reducing the need to repress on the side of governments (Gelleny and McCoy 1999).73 FDI is thus linked to human rights in complex ways and through opposing arguments that are unlikely to be captured by the variables used in the literature measuring the stock or flows of FDI. In the context of the link between FDI and human rights, my focus on BITs usefully captures and empirically insulates the aspects of FDI flowing into developing countries pointed out by the Òrace to the bottomÓ literature. As I make the case in the theory section, BITs lock-in any favorable initial conditions granted to multinational corporations to attract FDI to host countries. BITs are thus one direct way to capture the effect of such alleged favorable treatment of foreign investors. Empirically, then the measures of FDI included in our empirical models, should capture the residual effect of direct investment through channels like economic development and improved living standards.  3.3.2 Preferential Trade Agreements and Human Rights Various scholars study the impact of PTAs on human rights protection since a growing number of PTAs embed human rights standards into rules governing market access. More specifically, they differentiate two types of PTAs regarding human rights standards Ð soft PTAs (benefits or sanctions of PTAs are unconditional on statesÕ human right practices) and hard PTAs (benefits or sanctions of PTAs are conditional on statesÕ human right practices), and investigate their respective impact on human right protection (Hafner-Burton 2005, Spilker and Bohmelt 2013, Cao, Greenhill and Prakash 2013). However, the evidence is mixed. Hafner-Burton (2005b) argues that coercion                                                 73 Moreover, FDI may promote better human rights practices because FDI is attracted by countries with already better human rights records (Blanton and Blanton 2007, Barry et al. 2013).   56 is more effective than persuasion in changing actorsÕ repressive behaviors.74 Comparing to human right agreements and soft PTAs that might mainly influence statesÕ human rights practices via persuasion (lack enforcement mechanisms), hard PTAs can change statesÕ repressive behaviors through coercion since PTAs with hard human right standards are able to withhold economic benefits or impose economic sanctions (benefits of market access and threats or acts to disrupt exchange with violating members) in the case of human rights violations. Therefore, hard PTAs are expected to change actorsÕ incentive to respect human rights protection that would not otherwise be implemented. In the empirical analysis, she finds that hard PTA indeed improves statesÕ respect for human rights while human right agreements and soft PTA have no significant impact.  However, Hafner-Burton (2005b)Õs finding on Òhard PTAsÓ is challenged by Spilker and Bohmelt (2013) and Cao, Greenhill and Prakash (2013). Spilker and Bohmelt (2013) argues that states anticipate the Òshadow of futureÓ and expect to comply with human rights standards in the formation stage of PTAs (selection issue)75. Therefore, it might be the case that only those with expectation of compliance and good human right practices are more likely to join hard PTAs in the first place, which renders the effect of ex post joining Òhard PTAÓ disappear once we control for the selection effect of including hard human rights clauses in a PTA. Following Hafner-Burton                                                 74 Four reasons behind this rationale are discussed: (1) it does not require changing actorsÕ deeply held preference for repression (just increase cost of repression and benefit of human right protection); (2) coercion can take place in a much shorter time horizon; (3) coercion can change various repressive actorsÕ (such as future actors) behaviors as long as they value the gains of cooperation more than that of repression; (4) coercion does not requires repeated access to target repressors. 75 Actually, Hafner-Burton (2005b) is aware of this endogeneity issue. Relying on several cases studies, she concludes that repressive states are no less likely than non-repressive states to join hard PTAs and those PTAs indeed reduce repressive behaviors in repressive states. Furthermore, she employs two-state least squares to systematically deal with endogeneity of Òhard PTAÓ and finds that the results of hard PTAs still hold.   57 (2005b)Õs way of constructing the key Òhard PTAsÓ variable and employing genetic matching techniques to deal with selection problem, they find hard PTAs have no impact on human right practice once we account for selection effect of joining hard PTA in the first place. While Cao, Greenhill and Prakash (2013) investigate whether there is a transmission belts for the diffusion of human rights practices from importing to exporting states, they control for hard and soft PTAs in their empirical analysis. While being consistent with Hafner-Burton (2005b)Õs coding and operationalization of hard and soft PTA variables, they find that neither a countryÕs membership in a hard PTA nor its membership in a soft PTA is positively associated with physical integrity rights76.  In contrast to PTAs, little attention has been paid to the impact of BITs on governmentsÕ human right practice. Unlike PTAs that embed human rights standards into rules governing market access (conditional benefits or sanctions), BITs tend to lack human rights requirements. In practice very few, if any, BITs make express references to human rights or associated fields (Jacob 2010). Therefore, there is virtually no Òhard BITsÓ that can coerce states to comply with human rights practices as being observed in Òhard PTAsÓ. However, it does not necessary indicate that BITs would not have any impact on human right practice in capital-hosting countries. The following section leverages the strong protection offered by BITs to investors against the lack of provisions in these treaties with regards to human rights. Hafner-Burton (2005b) forcefully argues that Òchange in repressive behavior almost always requires legally binding obligations that are enforceableÓ (p.595).  I apply this logic in reverse to suggest that BIT sanctioned binding legal                                                 76 They suspect that the non-finding of the effect of hard PTAs may be due to two reasons: (1) fail to account for selection problem (similar to Spilker and Bohmelt 2013), and (2) a simple binary indicator of PTA membership does not distinguish between countries that are members of multiple PTAs and those that belong to a single PTA.  ! 58 commitments can incentivize the government to favor foreign investors (versus vulnerable domestic populations), with a net result of worsening human rights practices. 3.4 Theory The last two decades have seen a great deal of informal polemics on whether the global investment regime and, more specifically, BITs harm human rights. Human rights organizations are particularly vocal about the effect of BITs on governmentsÕ policy autonomy to create social peace and justice.77 This chapter is the first one, however, to engage in systematic theorizing and rigorous empirical testing.  As I just explained, BITs include both provisions that guarantee investor rights and mechanisms for investors to legally enforce such provisions.78 My key contention is that BITs have the potential to harm human rights practices because they lock-in conditions attractive to investors, both retrospectively and into the future.79 That is, as many developing countries predominantly attract vertical FDI and compete for investment and trade, BITs lock in the initial conditions favorable to investors on issues ranging from environmental regulations to labor standards and welfare spending. BITs not only act retrospectively, however. They also constrain the future policy choices of states for sustainable development, from the provision of basic infrastructure and investment in environmentally friendly technologies to land reform. The overt favoring of foreign investors and the constraints on development policies are important sources of popular grievance in host states. The governments of such states can then either anticipate dissent or experience                                                 77 For example, human rights NGOs opposed the Multilateral Agreement on Investment negotiated in the late 1990s by the OECD (Peterson 2009). NGOs also express concern in individual BITs negotiations, as for example the UK-Colombia BIT or the US-India BIT.  78 Even if investors have extensive legal protection, states still break contracts with multinational enterprises (Blake 2013, Allee and Peinhardt 2011, Wellhausen 2015).  79 Others note the lock-in effect of BITs (e.g., Milner 2014). We discuss why the conditions preserved in developing countries by BITs tend to be favorable to investors. ! 59 outright protest as a result of grievance. Repression and human rights violations are key responses of states to manifested or just anticipated threats. I argue however, that statesÕ reaction to threats and the negative human rights consequences of BITs are moderated by regime type. I detail my argument below.  3.4.1 Developing Countries and Policies Favoring Investors Many developing countries continue to compete in trade or for foreign investment by offering low environmental regulations, taxes, and lax labor standards, and reducing welfare spending. BITs de facto lock-in such favorable treatment granted to investors.  Recent scholarship shows that there are specific issues on which developing countries engage in Òraces to the bottomÓ to attract foreign investment. For example, Davies and Vadlamannati (2013) find that developing countries compete for FDI by relaxing de facto labor practices. Similarly, Klemm and Parys (2012) show that developing countries compete for capital with other countries in the same region with regards to the corporate income tax rate, as well as by offering corporate income tax holidays.  The existence of races to the bottom may further depend on the way in which countries are integrated in the global economy (Mosley 2007). My theory thus fits best flows of vertical FDI80, as this type of investment is particularly interested in conditions that cut production costs and recent research shows conclusive evidence that much of the investment in developing countries is indeed vertical. Supporting my contention, recent work examining country characteristics (market size, quality and quantity of labor, location, tax rates) to infer the nature of FDI (Blonigen and Wang 2005, Hanson et al. 2003). A similar conclusion is reached by UNCTAD (2004) which                                                 80 FDI that establishes facilities in multiple countries, each producing different inputs for the firm's production process.  60 shows that for both manufactured goods, but also services, FDI in developing countries is increasingly vertical. In addition, Buthe and Milner (2008) find that trade flows and trade policy, on the one hand, and FDI are complements, supporting indirectly their expectation that, at least for developing countries, FDI is largely vertical, part of intra-firm cross border transactions.  The link between investment in developing countries and their trade has further implications. Multi-national corporations use developing countries as part of their global production chain, by importing inputs and exporting processed goods (Buthe and Milner 2014). Because developing countries serve as export platforms for multi-national corporations (Frieden 1991, Ruane and Ugur 2005), export promoting policies are very relevant and attractive to foreign direct investors. Again, BITs have the potential to lock-in such export promoting policies and evidence exists that there are conditions when trade competition leads to lower environmental and labor standards. Cao and Prakash (2010), for example show that countriesÕ water pollution regulation mirrors the regulation of trade competitions. They also find that, while for water pollution regulation both a race to the bottom and a race to the top can be observed, air pollution regulation only responds to downward policy changes in trade competitors. Rudra (2011) finds that especially the size of exports has a negative effect on access to potable water, an effect that is mediated by income inequality and Mosley and Uno (2007) show that trade openness worsens collective labor rights.81  In addition to locking in past investor friendly policies, BITs constrain future government choices for sustainable development and welfare improvement. Rudra (2002) for example finds that globalization (trade openness and capital flows) lowers welfare spending in countries where labor enjoys little bargaining power vis-‹-vis the government. There are also numerous anecdotes                                                 81 Mosley and Uno (2007) also find that foreign direct investment flows improve the legal (not de facto) rights of workers to freedom of association and collective bargaining.   61 that support the loss of sovereignty due to BITs. For example, Blake (2013) argues that DenmarkÕs subsidies to local firms in the electricity sector to develop environmentally friendly technologies may potentially contravene the national treatment clauses in BITs. In an another example, in 2007 investors from Luxembourg and Italy brought an ICSID claim against South Africa arguing that the 2002 Mining and Petroleum Resources Development Act (MPRDA) expropriated their mineral rights. MPRDA is part of South AfricaÕs redistributive policy in favor of historically oppressed domestic social groups (the Black Economic Empowerment Policy), requiring that mining companies be partly owned by Ôhistorically disadvantaged personsÕ.82 Furthermore, right after the settlement of this claim in 2010, South Africa terminated its BITs with Belgium and Luxembourg, arguing that they limited the governmentÕs ability to pursue a transformative agenda. Additionally, there are pending BIT arbitrations arising out of disputes over the provision of water and sewage services such as Vivendi/CGE vs Argentina83, Aguas del Tunari vs Bolivia84, and Suez Corporation vs Argentina85. Also, an increasing number of requests for annulment86 of arbitration awards are argued by host states in relation to the provision of basic utilities Ð water, gas, electric power and infrastructure (Simmons 2014). 3.4.2 BITs Clauses Lock-in Favorable Policies Several specific BIT provisions act to both lock-in initial conditions and constrain the future choices of governments. I point to three of the most prominent ones: national treatment,                                                 82 See: http://www.lexology.com/library/detail.aspx?g=daf93855-71f9-425e-92d3-5368d104f8ff 83 CompaŒ™a de Aguas del Aconquija S.A. and Vivendi Universal v. Argentine Republic.  84 See for example: William Finnegan, ÒLeasing the RainÓ, The New Yorker, April 8, 2002, pg. 43 85 ÒSuez to Take US $496MN Charge Ð ArgentinaÓ, Business News America, July 1, 2002 86 The tribunal decisions in BIT arbitrations are final and binding, with no ground for appeal. Annulment is the only option (other than noncompliance) available to a party if it disagrees with the arbitration decision (Simmons 2014).!! 62 stabilization clauses under the umbrella clause, indirect expropriation, and fair and equitable treatment (FET).  One of the most important standards of investor treatment included in BITs is the national treatment clause, which prohibits host government from making negative differentiations between national and foreign investors (Blake 2013). As argued by Blake (2013), the national treatment clauses are Òperhaps the single most important standard of treatment enshrined in international investment agreementsÓ. Such prohibitions can constrain development strategies significantly.87 For instance, host governments are restricted in asking foreign multi-national corporations to locate investment in underdeveloped regions or employ domestic inputs in their production. In addition, BITs restrict host governments from favoring domestic firms that match the governmentÕs environmental or social policy goals. Take, India-Australia BIT (signed in 1999 and ratified in 2000), for example. Article 4 of this treaty requires ÒEach Contracting Party shall, subject to its laws, regulations and investment policies, grant to investments made in its territory by investors of the other Contracting Party treatment no less favorable than that which it accords to investments of its own investorsÓ. Domestic policy autonomy is also restricted by the inclusion of stabilization clauses in investment contracts88 that prevent host states from changing domestic law as it stands at the time of investment. Such clauses aim to reassure foreign investors in projects that demand a large amount of investment, especially in infrastructure or natural resource exploration that domestic                                                 87 While the majority of countries do not carve-out exceptions to the national treatment clause, some investment treaties do include instances when the host countries can deviate from national treatment.  88 It is difficult to estimate the number of international contracts containing stabilization clauses. Amnesty International (2006) estimates that stabilization clauses are more prevalent in the contracts of poorer countries facing groups of large multi-national corporations.  https://www.amnestyusa.org/sites/default/files/pdfs/hrtradeinvestmentmatters.pdf  63 law will not change to the detriment of the investor. Investors can invoke the stabilization clause from contracts with the host states under the Òumbrella clauseÓ of BITs and directly go to international arbitration. The umbrella clause in BITs provides the possibility of elevating contract breaches into treaty breaches. About 40% BITs include such umbrella clauses (Gill et al. 2004).89 Here are two illustrative treaty examples: (1) Article 11 of Switzerland-Pakistan BIT (signed in 1995 and ratified in 1996) - ÒEither Contracting Party shall constantly guarantee the observance of the commitments it has entered into with respect to the investments of the investors of the other Contracting Party.Ó (2) Article 10 of Poland-Australia BIT (signed in 1991 and ratified in 1992) - ÒA Contracting Party shall, subject to its law, do all in its power to ensure that a written undertaking given by a competent authority to a national of the other Contracting Party with regard to an investment is respectedÓ.  With direct expropriation on the decline, BIT clauses related to indirect expropriation have become key constraints on government policies. Besides guarding against weak enforcement of property rights, indirect expropriation refers to less clear-cut and potentially very broad measures such as changes in taxation, revocation of licenses or denial of access to infrastructure. The vast majority of investment treaties include language referring to indirect expropriation and governments have to compensate investors for indirect expropriation. The uncertainty over the meaning and scope of indirect expropriation has the potential to deter states from taking actions that, while in the public interest, may be regarded as indirect expropriation and requite significant investor compensation (Nikiema 2012). An example of treaty language is the Article III in Jordan Ð US BIT (signed in 1997 and ratified in 2003), requiring ÒNeither contracting party shall                                                 89 The Italian model BIT includes directly stabilization clauses as an international treaty obligation.   64 expropriate or nationalize a covered investment either directly or indirectly through measures tantamount to expropriation or nationalization (ÒexpropriationÓ) except for a public purpose; in a non-discriminatory manner; upon payment of prompt, adequate and effective compensation; and in accordance with due process of law and the general principles of treatment provided forÉÓ. Finally, fair and equitable treatment (FET) clause intends to protect investors Òagainst serious instances of arbitrary, discriminatory or abusive conduct by host statesÓ (UNCTAD 2012). As one of the most frequently invoked clauses in investor-state disputes arbitration, the terms of ÒfairnessÓ and ÒequityÓ are very vaguely defined, significantly constraining domestic policy autonomy of host governments. The absence of a clear definition of ÒfairnessÓ and ÒequityÓ and hence broad and expansive interpretations of FET clause makes it possible for investors to challenge many domestic policies in host states. As Prislan and Zandvliet (2013) point out, Òthe apparent indeterminacy of this standard has allowed investment tribunals to read into it an extensive list of disciplines, including the obligations to provide stability and predictability of the legal framework and to protect investorsÕ legitimate expectations Ð obligations which, in their nature, are geared towards generating stabilizing (and potentially chilling) effects.Ó For instance, US Ð Azerbaijan BIT (signed in 1997 and ratified in 2001) includes the treaty language of FET in Article 2.3, which requires ÒEach Party shall at all times accord to covered investments fair and equitable treatment and full protection and security, and shall in no case accord treatment less favorable than that required by international law. Neither Party shall in any way impair by unreasonable and discriminatory measures the management, conduct, operation, and sale or other disposition of covered investmentsÓ.      65 3.4.3 BITs, Grievance and Human Rights Violations The lock-in effect of BITs can force the hand of the government to favor multi-national corporations or foreign investors even at the cost of having to violate the rights of their own citizenry. First BIT provisions can simply ask the government to directly intervene and physically protect multi-national companiesÕ investment. Second, the same conditions that were designed to be favorable to investors and attract multi-national corporations have the potential to create popular dissent or the expectation of dissent, followed by repressive counters on the side of the government.  Very directly, a common BIT provision includes an obligation to provide foreign investors with Òfull protection and securityÓ. This clause commits host states to exercise Òdue diligenceÓ in protecting foreign assets and can be invoked by foreign investors when they encounter protests against their operations and practices in host countries. In a recent arbitration case90, for example, a Spanish multinational enterprise sued Mexico for failing to uphold the full protection and security clause by claiming that the authorities did not act as quickly and thoroughly as possible to Òprevent or put an end to the adverse social demonstrationsÓ regarding the investorsÕ controversial hazardous-waste treatment facility (Amnesty International 2006). The arbitration tribunal dismissed the claim related to the Spain-Mexico BITÕs Òfull protection and securityÓ clause saying that there is Ònot sufficient evidence supporting the allegation that the Mexican authorities É have not reacted reasonablyÓ (Amnesty International 2006). However, the tribunal concluded that the government had to compensate the enterprise on grounds of Òindirect                                                 90 T”cnicas Medioambientales Tecmed, S.A. v United Mexican States (ICSID Case ARB(AF)/00/2).  66 expropriationÓ, because, following popular protests, the government refused to renew the Spanish companyÕs permit to operate the landfill. Second, as explained above important BITs clauses act to preserve any incentives offered to attract investors, including low environmental protection standards, limited welfare provision and basic infrastructure development, as well as limitations on labor rights. Such locked-in conditions can create enormous dissatisfaction and the governmentÕs response to popular grievance may turn violent or abusive.91 For example, in the 1980s, USÕ Coca-Cola subsidiary in Guatemala faced street protest and hunger strikes over low-pay and refusal to allow the unionization of workers (Frundt 1987). In this case, while protecting foreign investorÕs property, the government used violence against Coca-Cola employees. In another example, in 1997, peaceful protest against the social and environmental consequences of the power-plant built in India by the Dabhol Power company, a joint venture of three US multinationals, were met with harassment, arbitrary arrest and preventive detention.92 Also, in the past decade, Cambodia has granted land concessions through the Economic Land Concession scheme to Cambodians with ties to government and foreign private investors (Amnesty 2011). In one high-profile case of forced eviction in 2007, about 4,200 families in the Boeung Kak Lake area were relocated to build luxury housing. The government employed intimidation and violence to force residents to accept paltry compensation and removal to a remote location.93  The grievance that results from policies overtly favoring investors can lead to overt dissent or the expectation on the part of government of future dissent. Dissent in the form of protest or the expectation of dissent as a consequence of popular grievance increase the perceived threat for                                                 91 Abouharb and Cingranelli (2009).  92 http://www.refworld.org/docid/3ae6a9884.html; Peterson and Gray (2003). 93 http://www.hrw.org/world-report-2012/world-report-2012-cambodia  67 governments seeking to preserve the status quo and generate motivation to employ repressive measures. Empirically, actual protest or potential dissent has been shown to be a key determinant of state employment of coercive measures against its own citizens (Davenport 1995, Moore 1998, Davenport and Armstrong 2004, Nordas and Davenport 2013).  The use of state repression in response to manifested or anticipated dissent is the result of authoritiesÕ assessment of the costs and benefits of rights violations versus other tools at their disposal (Davenport 1995, Nordas and Davenport 2013).  Repression is thus not an automatic response. I explore the variation in the costs of repression in the next subsection. Here I argue that, ceteris paribus, when the governmentÕs hands are tied by BITs, it becomes relatively expensive to address the root causes of popular grievance by taking measures against the legally protected investors. In addition, not only are investorsÕ rights legally protected, but these rights limit severely some of the non-repressive options that governments normally use to buy off the potential opposition, including increasing social benefits, cheap access to infrastructure services like water or electricity or providing side payments through domestic companies.  My first hypothesis follows this discussion: BITs are associated with a worsening of human rights practices (Hypothesis H1). 3.4.4 The Mitigating Effect of Political Regime Several conditions can mitigate the incentives of host states to use repression. I focus on democratic institutions, which, in the repression literature, are a key consistent variable that is shown to increase the cost of rights violations (Davenport and Armstrong 2004).  Importantly, democracies may, to begin with, be less likely to offer initial conditions that are favorable to foreign investors. Thus, Li (2006) finds that countries with better rule of law, which  68 tend to be democracies, offer lower levels of tax incentives.94 Also, I argued earlier that favorable trade policies are valued by foreign investors who use developing countries as export platforms. In this context, Cao and Prakash (2012) find that in countries with low political constraints, which tend to be non-democracies, the response to trade pressures is lower de facto standards for air pollution.  Moreover, while BITs tie the hands of all governments in a similar fashion, I argue that democracies and dictatorships vary in two key dimensions that affect statesÕ calculus of the costs and benefits of repression. First, democratic leaders and dictators are likely to have different assessments of the level of threat to their rule posed by popular grievance and dissent. All else equal, the greater the perceived threat to their rule, the more likely that governments will make use of repression (Davenport 1995, Nordas and Davenport 2013). It is likely, however, that the level of perceived threat emerging from conflict between the interests of multi-national corporations and domestic groups is higher in dictatorships. Protest or mass demonstrations, either manifested or just expected, are more likely to be seen challenges to regimes that severely limit citizensÕ freedom of speech and association, or voting for the political competition as outlets to express grievance.  Second, at a very fundamental level, democracies and dictatorships face different levels of accountability. In their review of the literature, Davenport and Armstrong (2004) note that Òin democracies political leaders who use repression against their citizens can be removed from office through the popular vote and, at the same time, these governments contain numerous institutional checks and balances on government activityÓ (p. 538). Thus, in political regimes that face real political opposition and a free media, episodes of human rights abuses can be expected to be                                                 94 Li and Resnick (2003) also argue that democratic accountability reduces the ability of governments to offer Ôsweet bargainsÕ to foreign investors, which, in turn, reduces the incentives for multinationals to pick democracies as investment locations.!! 69 quickly and widely acknowledged, raising the political and electoral costs of repression. In democracies both mechanisms Ð a low threat perception and high accountability Ð are likely to balance the favorable treatment afforded to investors by BITs with a high cost of repression.   I propose therefore a second hypothesis: The negative impact of BITs on governmentsÕ respect for human rights is mitigated in democracies (Hypothesis H2).       3.5 Data, Measurement and Research Design I test the hypotheses using data for 113 developing countries from 1981 to 2009.95 The start year is dictated by the availability of the key dependent variable. The sample includes only developing countries because rich countries have a different position in the global economy and are both sources of FDI and FDI recipients. Human rights practices also tend to be better in rich countries, making it likely that the causal process and governmentÕs trade-offs are different in the developing world.    3.5.1 Dependent Variable I use Cingranelli and RichardsÕ (1999) measure of governmentsÕ respect for physical integrity rights (CIRI dataset updated to 2012) as the dependent Variable96. I choose CIRI data over Political Terror Scale (PTS) that is used by Hafner-Burton (2005b) because the former explicitly captures governmentsÕ human rights practices while the latter only captures overall human rights conditions (Cingranelli and Richards 2010, Cingranelli and Filippov 2010, Richards, Gelleny, and Sacko 2001)97. Thus, the choice of CIRI fits better my theoretical arguments focusing on the actions of                                                 95 Similar to Poulsen and Aisbett (2013), these are countries that the World Bank does not classify as high-income for the majority of our sample period. 96 The CIRI data are freely available on the website www.humanrightsdata.org.  97 According to Cingranelli and Richards (2010), CIRI index defines human right practices as Òthe actions of government officials and actions by private groups if instigated by government directly affecting the degree to which citizens can exercise various types of human rightsÓ. However, there might be some other non-governmental actors such as revolutionaries, gangsters,  70 governments than PTS. It includes protection from torture, disappearance, extrajudicial killing, and political imprisonment. This index ranges from 0 (no respect for any of the four physical integrity rights) to 8 (full respect for all four physical integrity rights). The data sources include both Amnesty InternationalÕs Annual Report and the U.S. State Department's annual Country Reports on Human Rights Practices. When there is a conflict between these two sources, Amnesty International assessment is treated as authoritative in order to remove a potential bias in favor of US allies (Cingranelli and Richards 2010).98 In cases where only one source assessment is available, that sourceÕs is used.  3.5.2 Independent Variables I use three key independent variables. To test hypothesis one I use the cumulative number of BITs ratified by a country in a given year. The total or cumulative number of BITs that a country is subject to in a given year makes sense because our focus is on the total leverage that foreign investor interests have on the host states via the conclusion of BITs. The logic is that the greater the number of BITs a host state ratifies, the greater the potential for popular grievance and repressive tactics on the side of the government. I also use ratified BITs rather than signed BITs because only ratified BITs are legally binding (Haftel 2010, Haftel and Thompson 2013), which                                                 or terrorists that violate human rights conditions in a given country independently from governmental actions especially in many cases of domestic turmoil. For instance, for the last few years, the PTS places Afghanistan and Iraq at maximum terror (level 5 in PTS) while the CIRI scores that only consider the behavior of the governments reveal less governmental violation of human rights (Cingranelli and Richards 2010). Furthermore, Cingranelli and Richards (2010) argue that their coding scheme is more transparent and reliable. Finally, they suggest that using the PTS and CIRI data as mutual robustness checks in empirical analysis is unnecessary and invalid due to different focus of these two indexes Ð PTS (human right condition) while CIRI (human right practice of governments). However, following various works in the human right literature that still do the mutual robustness checks, I will conduct a robustness check by using PTS database.!!98 However, since 1981, there has been substantial and increasing agreement between these two reports (Poe et al. 2001).   71 indicates the leverage that foreign investors have on the host states. This variable is constructed using the International Investment Agreements (IIA) database on the UNCTAD website. The variable ranges from 0 to 104 and, over the period considered here, developing countries were on average subject to about 11 BITs. I further explore the heterogeneity of BITs to get at the causal mechanism in our theoretical explanation. First, I emphasize the strength of investorsÕ legal protection. To explore the stringency of BITs, this variable codes exclusively the BITs for which the International Centre for Settlement of Investment Disputes (ICSID) is the only option for dispute arbitration. This means that investors can use the ICSID for dispute settlements directly, without exhausting local remedies. The ICSID has leverage over developing countries because of its status as an international convention affiliated with the World Bank, provides for very limited grounds for appeal and its awards have the same effect as the judgments in national courts (Blake 2013, Allee and Peinhardt 2010). The measure is based on Allee and Peinhardt (2010) and my original coding of 420 additional BITs. The variable ranges from 0 to 24, with an average of 4. Second, I want to capture those BITs that regulate de facto investment, such that it is plausible that grievance created in host countries by favorable conditions granted to real investors. It is very likely that the BITs between developed countries on the one hand and developing nations capture an investment relationship characterized by de facto flows of capital to the capital poor developing country. I follow Poulsen and Aisbett (2013) and use as a second measure only North- South BITs, as well as the BITs of major capital exporting developing countries: Brazil, Russia, South Africa, China, Argentina, Panama, Mexico, Malaysia, Saudi Arabia, Indonesia, Hungary, Chile, and India. The variable ranges from 0 to 66, with an average of 7.99                                                  99 Our results are similar when we code those BITs for which ICSID is the only option or one of  72 To test my second hypothesis, we use the polity2 score from Polity IV dataset that tries to capture the level of democracy. It ranges from -10 to 10 where a larger value indicates a higher level of democracy. Regimes with democratic forms of government are less likely to violate human rights of their citizens because institutional features of democracies such as elections, free media and civil liberties enable citizens to be informed of those human right abuse actions, remove oppressive leaders, and constrain the power of the governments. I include an interaction term between the polity2 score and the cumulative ratified BITs to test the conditional effect of investment treaties. 3.5.3 Control Variables Since I am interested in comparing the impact of two types of economic treaties - BITs and PTAs - on human rights and exploring whether they will exert different or even contrasting influences on human rights, I rely on Hafner-Burton (2005b)Õs empirical analysis and try to use her control variables as well as her main key independent variables (such as Hard PTAs) as my control variables. In my main model, I use the following control variables: (1) foreign investment measured as the net FDI inflows as percentage of GDP (UNCTAD)100; (2) the sum of a stateÕs total exports and imports as a share of gross domestic product (logged) (World Bank World Development Indicators - WDI); (3) GDP per capita in constant US dollars (logged) (WDI); (4) regime durability measured as the number of years since a state has undergone a structural regime transition, defined as a movement on the Polity scale of three points or more (Polity IV); (5) population per squared kilometer of land area (WDI); (6) international human right agreements capture whether countries have ratified the International Covenant on Civil and Political Rights                                                 the options, or, alternatively only North- South BITs. 100 Using log (FDI in millions) our results are robust in all model specification. However, the FDI variable is only statistically significant and positive in the instrumental variable models.   73 and the Convention Against Torture (Spilker and Bohmelt 2013); (7) soft Preferential Trade Agreements measures whether a state belongs to any PTAs with soft human rights standards (Spilker and Bohmelt 2013); (8) hard PTAs is a dummy variable that takes a value of 1 if a state belongs to any PTAs with hard  human rights standards (Spilker and Bohmelt 2013). In addition to this baseline specification, we also include key covariates from the repression literature: Two dummy variables code ongoing civil war and interstate war (Armed Conflict Dataset). Also, following Nordas and Davenport (2013) we include political dissent coded as the sum of antigovernment protest, riots, and general strike (Banks CNTS). Additional relevant variables are discussed in the robustness checks. The summary statistics for all variables is shown in Table 3-2. 3.5.4 Model Specification I use an OLS regression with panel-corrected standard errors (Beck and Katz 1995) and adjust for first-order auto-correlation by specifying an AR(1) process101. All models include country dummies to capture country-specific unobserved heterogeneity. I also include half-decade period dummy variables to account for time-specific shocks or time trends that may influence both human rights violations and BIT ratification. All independent variables are lagged one year. The empirical model is shown as:   I expect that  is negative indicating that BITs work to worsen countriesÕ human rights practices.  should be positive, indicating that democracies mitigate the negative effect of BITs. Finally, based on the literature on repression  should be positive: Democracies tend to have better human rights.                                                  101 The results are broadly similar when autocorrelation is corrected by (i) using a panel-specific AR1 autocorrelation structure; (ii) adding a lagged dependent variable to our estimations.  tititititititiControlsPolityBITsPolityBITsCIRIµ!"####+++++++=,,,4,3,21,][*2!4!3! 74 Table 3-2. Summary Statistics  3.6 Results and Discussion Table 3-3 presents our results. Models 1 and 2 use the cumulative number of all BITs as the key independent variable. Models 3 and 4 use an instrumental variable approach to estimate the effect of BITs on human rights practices. Models 5 to 8, on the other hand, use our alternative measures for relevant BITs, based on the ability of investors to litigate at the ICSID (BITs with ICSID; Models 5 and 6) and the likely existence of actual investment flows (Adjusted North-South BITs; Models 7 and 8). The empirical estimations support my two hypotheses. Models 1, 3, 5 and 7 include the un-interacted cumulative number of BITs and the polity2 score. Across all the models, as expected, the coefficient on the cumulative number of investment treaties is negative  75 and statistically significant. A greater number of BITs ratified by a country reduces the CIRI index, showing a worsening of human rights conditions. Models 2,4, 6 and 8 include an interaction term between the cumulative number of ratified BITs and polity2 score. The coefficient on the cumulative number of investment treaties continues to be negative and statistically significant, but the interaction term is positive and statistically significant. This supports my second hypothesis: Democracy mitigates the negative effect of BITs on host governmentÕs respect for human rights. Next I discuss our instrumental variable estimation and the scope condition when BITs influence human rights conditions.  Very important, the relationship between BITs and human rights violations is open to the charge of endogeneity and may be spurious. It may be that states with low level of respect for human rights are in a greater need for the credibility afforded by ratifying BITs in the first place. Although there is no clear evidence in the literature that states with high level of human right violation are more likely to conclude BITs (Neumayer 2006, Vadlamannati 2009, Table 3-1), I show instrumental variable estimations to address the potential endogeneity problem. I use two instruments for my key independent variable - the cumulative number of ratified BITs. The first instrument measures the average of the total ratified BITs in neighboring states in a given year. Following Kerner (2009) I define neighboring countries using the Correlates of War coding for type 1 or type 2 contiguity, which includes countries that share a land border or are separated by 12 miles of water or less. This instrument aims to capture the competitive nature of BITs signing (Elkins et al. 2006) and the correlation of the instrument with our independent variable is 0.69. The second instrument uses the three year lagged (year t-2, t-1 and t) total of new BITs ratified in other countries in the world. This instrument intends to capture the trend of BITs ratification and the opportunities of concluding BITs. The correlation of this instrument with the key independent  76 variable is 0.29 in the sample used in my model. I use Stata command xtivreg2 and the results shown in Table 3-1 (Models 3 and 4) continue to support my hypotheses. The chosen instruments perform well: The Hansen test of over-identifying restrictions tests the overall validity of the instruments (including the choice of exogenous variables) and failure to reject the null hypothesis gives support for the model. For Model 3 and 4, the Hansen J statistic Chi-sq(2) p-value is 0.38 and, respectively,  0.54 , so I can not reject the null hypothesis. In instrumental variable models, while chosen instruments may be exogenous they may be weak, biasing the estimated coefficients. For both Models 3 and 4, the weak identification Kleibergen-Paap rk Wald F statistic is above 42. This value easily passes the Òrule of thumbÓ (Staiger and Stock 1997) that the F statistic should be at least 10 for weak identification not to be considered a problem.102  In addition to models 3 and 4, I use similar instrumental variables for my modified independent variables (BITs with ICSID and Adjusted North-South BITs) and the results shown in Table 3-1 are largely robust.  More than just examining multiplicative interaction terms for direction and statistical significance, Brambor et al. (2006) prescribe that inference should be done with meaningful marginal effects and standard errors to determine the conditions under which the variable of interest has a statistically significant effect. Figures 3-1(a, b, c, d) show the marginal impact of BITs on governmentsÕ respect for human rights conditional on the level of democracy. Figures are based on Model 2, 4, 6 and 8, respectively.                                                   102 My results are robust to the use of a third instrument following Buthe and Milner (2008): Based on Jandhyala, Henisz, and Mansfield (2011)Õs dyadic dataset of BIT signing, I first calculate for each dyad-year the probability that the two countries in the dyad will become members of a BIT in that year. I then add up the predicted probabilities for each country and year and divide that (monadic) sum by the number of possible BIT partners the country could have had for the given year. This calculation is my third instrument, as it yields a measure of the average probability that a country signs a BIT with all other countries in the world in the given year.    77 Table 3-3. Effect of BITs on GovernmentsÕ Respect for Human Rights in Developing Countries 1981-2009                78 Figure 3-1. The Marginal Effect of BITs on GovernmentsÕ Respect for Human Rights Conditional on Regime Type             Note: Figure 3-1 (a) is based on Model 2, Figure 3-1 (b) is based on Model 4, Figure 3-1 (c) is based on Model 6, Figure 3-1 (d) is based on Model 8.    The marginal effect of cumulative BITs is negative and highly statistically significant in less democratic states, whereas in democracies the effect is much smaller and only marginally statistically significant. Very relevant for our argument, the size of the marginal effect from Figure 3-1(a) is smaller than that in Figures 3-1(c & d). This means that the effect of BITs with ICSID is about three times as large as the effect of all BITs, while the effect of adjusted North-South BITs Fig.3-1 (a) Fig. 3-1 (b) Fig.3-1 (d) Fig.3-1 (c)  79 is about 1.3 as large.103 This is consistent with my theoretical story that governments react to the legal protection of investors in international tribunals and that the effect of BITs should be larger as these treaties actually regulate de facto inflows of capital to developing countries. To further assess the substantive effect of our key independent variable, I use Models 2, 6 and 8 to predict human rights conditions as we vary the cumulative number of BITs and the level of the Polity democracy score (Table 3-4). I vary the cumulative number of all BITs, BITs with an ICSID clause and the adjusted North-South BITs, as well as the Polity 2 score one standard deviation above and below the mean. All other variables are held at mean values.     Table 3-4. Predicted Human Rights Conditions: Vary BITs and Democracy Note: Predictions are calculated using Stata command margins. The numbers in parentheses are 95 percent confidence intervals.  In non-democracies, varying the number of BITs by one standard deviation around the mean accounts for about 8% of the variation in the CIRI dependent variable. For example, for BITs with the ICSID clause, moving from no BITs to 7 BITs that include the restrictive clause on investment                                                 103 When use North-South BITs rather than the adjusted North-South BITs,  the marginal effect is more than 3 times larger than in Model 2.!  Democracies Polity2 score 1 SD above mean (8) Non-Democracies Polity2 score 1 SD below mean (-6)   Model 2 Ð all BITs BIT count 1 SD above mean (25 BITs) 4.68 3.33 BIT count 1 SD below mean (0 BITs) 4.96 3.98 Column Difference -0.27 (-0.48, -0.064) -0.64 (-0.90, -0.39)   Model 6 Ð BITs with the ICSID clause BIT count 1 SD above mean (7 BITs) 4.62 3.28 BIT count 1 SD below mean (0 BITs) 4.95 3.94 Column Difference -0.33 (-0.57, -0.094) -0.66 (-0.98, -0.34)   Model 8 Ð Adjusted North-South BITs BIT count 1 SD above mean (15 BITs) 4.64 3.34 BIT count 1 SD below mean (0 BITs) 5.00 3.98 Column Difference -0.36 (-0.63, -0.095) -0.64 (-0.92, -0.36)  80 arbitrations reduces the predicted level of physical integrity rights from 3.94 to 3.28, or about 8% of the 0 to 8 range of the dependent variable. For democracies, on the other hand, the same variation in the number of BITs has a smaller effect of about 4% of the variation in the CIRI dependent variable. In democracies, for BITs with the ICSID clause a move from zero to 7 reduces the predicted level of physical integrity rights from 4.95 to 4.62. The differences in predictions are statistically significant at the 95% confidence level and, for non-democracies in particular, the size of this effect is important. For comparison, I compute the effect on the human right conditions of democracy, a key determinant of human rights in the repression literature. A move of one standard deviation above and below the mean in the Polity2 score (keeping BITs and all other variables at mean values), accounts for about 13% of the 0 to 8 range of the dependent variable (from 4.86 to 3.75). Finally, regarding our control variables, I find that soft PTAs may increase repression levels. On the other hand, we find no impact for hard PTAs. In addition, ratification of human right treaties is associated with worse practices across our models. Also, I find that the level of democracy, trade openness, GDP per capita, population density, civil war, and political dissent are all significant predictors of human rights practices and take on expected signs. Importantly, I find that net FDI inflows are associated with better human rights. This effect, and the results supporting my hypotheses, maintain when we use the ten-year lagged moving average of FDI inflows to mitigate the potential endogeneity between FDI and human rights practices.  3.6.1 Robustness Check I verify the robustness of the empirical results against additional threats to inference. These include the timing of when BITs started to give investors access to international arbitration without  81 the requirement to exhaust local remedies; the operationalization of the dependent variable; and, finally, the presence of outliers and the effect of additional variables.  First, I restrict the sample to begin from 1990 onward because it was not until the late 1980s that BITs began to give investors access to investor-state arbitration without first having to exhaust local remedies. This strategy is similar to Poulsen and Aisbett (2013), who note that, after 1990, the vast majority of BITs include a binding consent to investor-state arbitration. My theory centers on the leverage of multinational corporations have on host states and therefore the magnitude of lock-in effect of initial policies. It may be argued therefore that focusing on BITs with access to investor-state arbitration is a more appropriate way to capture the foreign investors leverage through BITs. Model 9 and 10 in the Table 3-5 restrict the sample to BITs ratified after 1990 and show that my results are robust.  Second, I use an alternative measure of human rights violations - the Political Terror Scale (PTS) (Gibney et al. 2012) 104. The PTS database has two separate indexes Ð one based on the source from US State Department Country Reports and the other from Amnesty InternationalÕs Annual Report. An ordinal variable of five categories that capture personal integrity rights is constructed for each PTS index105. A larger value indicates a higher level of repression. To be consistent with the interpretation of CIRI data used in my main analysis, I rescale each of both PTS indexes so that a larger value reflects a higher respect of human rights. These two sources have slightly diverging coverage106, and both have been criticized for possible biases. State Department Report protects itself and its allies as well as countries that US has an interest (Poe et                                                 104 Besides CIRI index, PTS is the other commonly used human right index to capture physical integrity rights.  105 For detailed coding scheme, please refer to Gibney, Cornett, Wood, and Haschke 2012 as well as http://www.politicalterrorscale.org.  106 The State Department Report has a broader coverage than the Amnesty International Report.!! 82 al. 2001) while Amnesty International Report often did not rate countries (socially regimes) with few human rights violation in earlier years such as 1970s and 1980s (Wood 2008). Therefore, scholars differ in how to use these two PTS indexes. Some use them separately Ð either US State Department Country Reports (Nordas and Davenport 2013, Conrad and DeMeritt 2013, Englehart 2009, Peksen 2009) or Amnesty InternationalÕs Annual Report (Nordas and Davenport 2013, Wright 2014) while others use the average of these two indexes to counterbalance biases (Blanton 2000, 2005, Apodaca 2001, Daxecker and Hess 2013, Demirel-Pegg and Moskowitz 2009). To address these issues and controversy, I employ all three ways of dealing with the PTS data Ð an average PTS index107, an Amnesty International index108, and a US State index109. Table 3-8 shows my estimates. The coefficient for cumulative BITs remains negative, but loses statistical significance in the non-interaction model. However, the interacted models remain very similar to my findings in Table 3-3, supporting Hypothesis 2. I maintain, however, that the CIRI data is more appropriate for testing my hypotheses because the PTS indexes cannot differentiate the human rights practices of governments from broader human right conditions (Richards 2001, Cingranelli and Richards 1999).  Finally, I exclude outlier countries and include additional control variables. My results are substantively similar if I exclude the countries that are above the 99th percentile in terms of cumulative BITs ratified (China, Romania, the Czech Republic) or above the 95th percentile (China, Romania, the Czech Republic, Turkey, India, Egypt). I also test the robustness of my                                                 107 Where US State Department or Amnesty International values are missing, the value that is available is taken as the average. This PTS average index will take on some values such as 1, 1.5, 2, 2.5,É, 4, 4.5, 5.  108 To compensate for the missing data due to potential biases in both indexes, I follow Poe and Tate (1994), Poe, Tate, and Keith (1999, 2009) by substituting a State Department score for a missing Amnesty International score.  109 Similarly, I substitute an Amnesty International score for a missing State Department score.!! 83 hypotheses by including additional control variables: (1) Simmons (2014) finds that states are more likely to sign restrictive BITs during economic downturns. Economic crisis may also induce governments to repress social unrest. I control for the 3 year lagged average economic growth (WDI). (2) Abouharb and Cingranelli (2006, 2007, 2009) find that IMF or World Bank adjustment programs tend to worsen human rights in loan receiving countries. I control for the number of years that countries are under either IMF or WB programs. Following Abouharb and Cingranelli (2006, 2007, 2009) I also control for British or French colonial legacy. (3) The ÒshamingÓ activities of human rights international NGOs may also improve statesÕ human rights practices. I control for this by using a new dataset of shaming events of more than 400 human right NGOs towards governments (Murdie and Davis 2012). Table 3-5 shows that my results are largely robust to the inclusion of additional control variables.              84 Table 3-5. Robustness Check for a Restricted Sample and Additional Control Variables Ð All Cumulative BITs           85 Table 3-6. Robustness Check for a Restricted Sample and Additional Control Variables Ð BITs with ICSID            86 Table 3-7. Robustness Check for a Restricted Sample and Additional Control Variables Ð Adjusted North-South BITs            87 Table 3-8. Robustness Check for an Alternative Measure of Physical Integrity Index Ð PTS Scale  3.7 Conclusion  This chapter is a first theoretical and empirical investigation into how and whether the global investment regime, and, in particular the ratification of BITs, affects human rights in developing countries. In these countries, I argue that BITs have the potential to worsen human rights practices because they lock in initial conditions attractive to investors, both retrospectively and into the future. Retrospectively, many developing countries still compete for investment and trade on issues ranging from environmental regulations, taxes, labor standards, and welfare spending and  88 BITs lock in these initial favorable conditions. In addition, BITs provisions can constrain the future policy choices of states for sustainable development, from the provision of basic infrastructure and investment in environmentally friendly technologies to land reform. Low standards for environmental protection or labor rights and constraints on development and social policies can be important sources of popular grievance. Moreover, repression and human rights violations are key responses of states to the manifested or just anticipated protest and dissent that can result from such grievances. I argue however, that statesÕ reaction to threats and the negative human rights consequences of BITs will be moderated by regime type. Democracies are less likely to offer investors more initially favorable conditions, as seen in tax incentive policies or de facto environmental standards. Also, relatively low perceived threat of protest or dissent to the regime stability and a high level of political accountability in democracies increase the cost of state repression and are more likely to balance the favorable treatment afforded to investors by BITs.  Using a sample of 113 developing countries from 1981 to 2009, I find support for our theoretical arguments. Countries that have ratified a higher number of BITs have worse human rights practices. This effect holds and is larger when we restrict our BITs count to only those treaties that have stringent arbitration clauses (ICSID arbitration) or are likely to govern over de facto investment flows (North-South BITs). In addition, I find that the effect of the cumulative number of ratified BITs is conditional on political regime: BITs are more likely to result in human rights violations in non-democracies. The results are robust to alternative modeling techniques, inclusion of controls and variations in the sample. This chapter draws attention to the unintended externalities of concluding BITs. Investment treaties were drafted to facilitate cross-border capital flows and promote development through foreign investment. Yet we bring robust evidence that ratifying BITs tends to worsen the human  89 rights practices of developing countries, very likely because they tie the hands of governments, supporting thus the concerns of human rights NGOs. The findings also back the recent move to incorporate human rights standards in the content of BITs, either by explicitly referencing human rights110 or by including related provisions with regards to labor standards or environmental protection.111                                                                 110 2007 Norwegian model BIT mentions human rights practices in preambular language  (Jacob 2010).  111 See 2004 Canadian model BIT, 2004 US model BIT.! 90 CHAPTER 4 THE IMPACT OF BILATERAL INVESTMENT TREATIES ON COLLECTIVE LABOR RIGHTS IN DEVELOPING COUNTRIES  4.1!Introduction  Bilateral investment treaties (BITs) have been the most visible and powerful legal protective mechanism underlying the growth of cross-border capital flows since its inception back in 1959 between Pakistan and Germany. Since BITs are initiated to overcome the time-inconsistency problem plaguing cross-border foreign direct investment (FDI) by providing legal protection for foreign investors, most of scholarships in the literature examine whether conclusion of BITs are able to bring in foreign direct investment for developing countries (recent works include Buthe and Milner 2009, Kerner 2009, Allee and Peinhardt 2011, Haftel 2010). However, little attention has been paid to the potential negative externalities of the Òbroad and asymmetricalÓ (Simmons 2014, p. 12) investment protection clauses on capital-hosting developing countries. In this chapter, I explore one of those potential negative externalities and examine whether BITs influence labor rights in developing countries. More specifically, I focus on the collective labor rights that capture the freedom of association and collective bargaining rights and differentiate labor laws from the labor practices on the ground112.                                                 112 Although other important labor rights such as minimum wage, working conditions (e.g., safety and health, overtime) are not addressed here, along with other scholars I assume that higher levels of collective labor rights lead to improvements in other labor rights (Aidt and Tzannaos, 2002; Huber and Stephens 2001; Mosley and Uno 2007).   91 Although investor-state disputes arbitrations regarding labor rights are rare113, civil groups114 and legal scholars (Prislan and Zandvliet 2013) are very cautious about the possibility that foreign investors may bring host governments to courts if governments would promote distinctly pro-labor policies (e.g., increase minimum wage and collective bargaining rights) or failed to put a swift end to labor unrest. Similar litigation has been observed in the fields of promoting public health and improving environmental protection115. However, we lack systematic theory and rigorous empirics on how and whether BITs affect labor rights in host states. Importantly, it is crucial to differentiate labor laws and actual labor practices in developing countries because there is substantial discrepancy between these two related but distinct dimensions.116  I argue that BITs should have no significant impact on collective labor laws in developing countries. Even though BITs have the potential to reduce host governmentsÕ capability and incentive in passing regulations aiming at improving collective labor rights, they need not necessarily lead to deteriorating conditions of collective labor laws. Foreign investors prefer to maintain the status quo in terms of regulations in host states. In contrast, I argue that BITs are                                                 113 Investor-state international arbitrations, however, are relatively rare events in general. Theoretically speaking, if investors and host states are rational at the bargaining process, we will not observe any outbreaks of litigation in the international adjudication bodies because arbitration is costly for both investors and host states. However, in reality, we do observe some cases of bargaining failures that are published to the public. According to Wellhausen (2015)Õs publicly known international arbitration record, multinational corporations have sued at least 100 host states for at least 535 times from 1990 to 2012. 114 See, eg, Bill Rosenberg, ÔLabour Rights and Investment AgreementsÕ (2 March 2012) <http://union.org.nz/sites/union.org.nz/files/Labour-Rights-and-TPPA-March2012.ppt>  115 Public health regulation: Philip Morris Asia Limited v. Australia in 2012, and Philip Morris Brands S‹rl, Philip Morris Products SA and Abal Hermanos SA v. Oriental Republic of Uruguay in 2010 regarding host statesÕ stringent tobacco labelling legislation. Environmental protection regulation: Vattenfall v. Federal Republic of Germany in 2012 regarding GermanyÕs nuclear phase-out policy.  116 When utilizing Mosley (2011)Õs collective labor law and collective labor practice indexes, the correlation is only about 0.27 for developing countries.  92 likely to worsen collective labor practices and widen the gap between labor laws and practices in host states. BITs are able to lock in initial low labor conditions that tend to be attractive to foreign investors, which may be a potential source of labor grievance and labor unrest117. Also although inflows of FDI are suggested to generate economic development and increasing incomes for workers, multinational corporations are found to be inviting targets for labor unrest. All this potential or manifested labor unrest makes host states vulnerable to international arbitration from foreign investors given the legal protections incorporated in BITs (e.g., full protection and security). Consequently, I argue, host governments have incentives to undermine the collective action capability of labor groups by undercutting their collective labor rights, reducing collective action and thus minimizing potential labor unrest. If given the choice of altering the law or just adapting labor practices, host governments may find it cheaper and easier to disguise to opportunistically undermine de facto practice. Another implication is then that BITs should be related to a larger gap between labor laws and labor practices. Using a sample of about 120 developing countries from 1985 to 2002, I find support for my hypotheses.  This chapter makes several contributions. First, although most of the studies have focus on the link between BITs and FDI inflows in developing countries, this study draws attention to the potential unintended consequences of concluding BITs by revealing that BITs tend to worsen collective labor practices and enlarge the gap between labor laws and labor practices in host                                                 117 Some may argue that workers actually benefit from inflow of foreign direct investment (FDI). For instance, based on the distributive effect of FDI, Pinto (2013) assumes that workers are beneficiaries of FDI in the factor market (e.g., higher wages). He then argues that leftist governments whose core supporters are labor groups have incentive to implement FDI-friendly policies. However, as Robertson and Teitelbaum (2011) point out, FDI inflows are also likely to  generate industrial strikes and labor unrest in capital-hosting developing countries though inflows of foreign direct investment are projected to bring in economic development and many well-paid jobs.    93 governments. Second, the non-finding of the impact of BITs on collective labor laws does not imply that BITs do not influence labor laws at all in developing countries. While I find no overt negative effect, BITs appear able to deter host governments from improving domestic collective labor rights and other labor-enhancing policies (e.g., minimum wage). Finally, this chapter backs the recent movement of BITs by including clauses safeguarding labor laws as well as labor practices (e.g., United States Ð Uruguay 2005, Belgian-Luxembourg Economic Union Ð Republic of Mauritius 2005)118.    The remainder of the chapter is organized as follows. I first review the relevant literature on foreign direct investment (FDI), preferential trade agreements (PTAs) and labor rights. Next I elaborate my theoretical argument and derive two hypotheses. I then discuss the research design and empirical results. The final section concludes the chapter.  4.2 Literature Review 4.2.1 The Impact of FDI on Collective Labor Rights Mosley and Uno (2007) find that inflows of FDI have the potential to improve collective labor standards in developing countries. It is because subsidiaries of multinational corporations may tend to transmit better labor standards from parent companies and avoid violation of labor rights that may hurt their reputations among consumers and other audiences in their home countries. However, other studies reveal no significant evidences (Greenhill, Mosley, and Prakash 2009;                                                 118 Two main modes of mentioning labor rights in the main texts of recent BITs are proposed: (1) prevention of Òrace to the bottomÓ in labor standards: Òinappropriate to encourage investment by weakening or reducing the protections afforded in domestic labor lawsÓ (e.g., United States Ð Uruguay 2005, Belgian-Luxembourg Economic Union Ð Republic of Mauritius 2005); (2) preserve policy autonomy in labor-related fields: ÒRecognizing the right of each Party to establish its own domestic labor standards, and to adopt or modify accordingly its labor laws and regulationsÓ (e.g., Belgian-Luxembourg Economic Union Ð Republic of Mauritius 2005).   94 Blanton, Blanton, and Peksen 2015a; Blanton, Blanton, and Peksen 2015b119; Neumayer and de Soysa 2006). Additionally, Blanton and Blanton (2012) find that FDI inflows reduce the level of collective labor rights. However, this effect varies by sectors. FDI inflows into services sectors tend to worsen collective labor rights while inflows into manufacturing sectors have the potential to increase collective labor rights. Due to the potential link between BITs and FDI, I control for FDI inflows in the empirical analysis. In this way, I am able to test my theoretical argument regarding the impact of BITs that are not potentially confounded through the channel of FDI inflows. 4.2.2 The Impact of PTAs on Collective Labor Rights Since many preferential trade agreements (PTAs), especially between developing countries and developed countries (e.g., United States and European Union) incorporate labor-related clauses, scholars have examined whether those PTAs are able to improve domestic labor rights in partner countries. A positive relationship between PTAs and labor rights in partner countries is revealed in Kim (2012) on PTAs with United States and Postnikov and Bastiaens (2014) on PTAs with European Union. In contrast to PTAs, such labor-related clauses have not been included in BITs until very recently. It is likely to be the case that legal scholars or policy makers assume that increased inflows of FDI brought by BITs may inevitably result in improvement of labor rights in host states although the empirical evidence on the link between FDI and labor rights is not clear. However, in this chapter I examine the impact of BITs on collective labor rights (labor laws and labor practices) by leveraging the strong investment protections granted to foreign investors against the lack of clauses safeguarding labor-related issues.                                                     119 Blanton, Blanton, and Peksen (2015b) find some evidence that FDI increases collective labor laws but has no impact on collective labor practices.   95 4.3!Theory Following recent studies on labor rights (Blanton, Blanton, and Peksen 2015a, 2015b; Greenhill, Mosley, and Prakash 2009; Payton and Woo 2014), I differentiate labor law from labor practice. Specifically, I examine the collective labor rights that measure the freedom of association and collective bargaining rights (Mosley 2011). It is important to differentiate these two aspects of labor rights particularly in developing countries because governments may Òput on a good faceÓ by passing strong labor protective laws while they may inadequately enforce those laws either due to lack of capability or willingness (Davies and Vadlamannati 2013). The labor practice index from Mosley (2011)Õs dataset for developing countries is indeed only weakly correlated with the labor law index, with a correlation coefficient of 0.27. In the following section, I will theorize the impact of BITs on labor law, labor practice and the gap between labor law and practice, respectively, and argue that BITs tend to worsen collective labor practice and widen the gap between labor law and practice while they do not have significant impact on collective labor law.  4.3.1 The Impact of BITs on Collective Labor Laws I argue that BITs are likely not associated with either improving or deteriorating collective labor laws in developing countries. Since foreign investors have utilized BITs to challenge regulations in host states at international arbitration institutions regarding promotion of public health or protection of the environment, civil groups or NGOs have expressed the fear that foreign investor may bring host states to the international investment arbitration courts as a response to the introduction of labor-enhancing policies or fail to end industrial strikes. Similarly, in the context of collective labor laws the existence of BITs allows foreign investor to deter host states from improving those laws as it may increase labor costs and hence affect their profits. This Òregulatory chillingÓ effect empowered by BITs can be achieved through at least two specific  96 treaty clauses: (1) stabilization clause embedded in investment contracts combined with umbrella clause in BITs: Many multinational corporations sign investment contracts containing a Òstabilization clauseÓ with host governments, which require the governments maintain the same domestic regulations throughout the length of investment. When BITs have the umbrella clauses, foreign investors can elevate the breach of investment contracts into breach of BITs and have access to all the protections in BITs. About 40% BITs include such umbrella clauses (Gill, Gearing, and Birt 2004)120. (2) indirect expropriation clause: The vast majority of BITs contain the clause of Òindirect expropriationÓ that may be invoked by foreign investors when host governments change domestic regulations in the field of collective labor law. It is because foreign investors can frame the behaviors of changing domestic laws as expropriating their property indirectly and bring host governments to international courts for investor-state disputes arbitration. The uncertainty over the meaning and scope of indirect expropriation enables foreign investors to deter government from implementing labor-improving policies though it may be in the interest of domestic labor welfare. In a word, strong investment protections empowered by BITs have the ÒstabilizingÓ or ÒchillingÓ effects on collective labor laws in host states. Although BITs tend to constrain host statesÕ capability or willingness in improving collective labor laws, it does not imply that BITs will necessarily lead to worsening collective labor laws in host states. Many capital-hosting developing states may not have incentive to strengthen collective labor laws even in the absence of BITs. Thus, I posit the first hypothesis: Hypothesis 1: BITs are likely to have little impact on collective labor law in developing countries.                                                   120 The Italian model BIT actually includes directly stabilization clauses in the BIT text.   97 4.3.2 The Impact of BITs on Collective Labor Practices I argue that BITs have the potential to worsen collective labor practice in developing countries. BITs have the potential to lock-in initial low labor conditions that are attractive to foreign investors that may be an important source of labor grievance and hence labor protest. Additionally, FDI inflows are found to generate industrial strikes and labor unrest. Facing potential or manifested labor unrest in host governments, highly protective foreign investors empowered by BITs may force host governments to address the grass root of grievance by taking measures against domestic workers and cutting back the collective capability of labor groups. First of all, although whether developing countries compete with one another in lowering labor standards is debated in the literature (Blanton and Blanton 2012; Davies and Vadlamannati 2013; Mosley and Uno 2007; Payton and Woo 2014), some developing countries may maintain a low level of labor costs in order to be attractive to foreign investors. Take, the minimum wage, for example. In response to the demand of increasing minimum wage from current 9 dollars to 12 dollars a day (US dollars) by labor unions in 2012, the President Benigno Aquino of the Philippines argued against this demand during his speech on labor day celebrations on the ground that it would make Philippines less competitive in attracting foreign investment as other neighboring nations including Indonesia, Vietnam, and Cambodia maintain a very low level of minimum wage that is less than 5 dollars a day (Payton and Woo 2014). In another example, tens of thousands of workers protested against the African National Congress (ANC) government of South Africa in early 2012 and demand an end to labor brokers that supply temporary workers rather than full time employees to businesses. While the practice of labor brokers evades the benefits of full employments and generates dissatisfaction from workers, Òthe ANC government is also struggling to find a balance between workers' demands and economic growth. Officials fear that feeding too much power to  98 the unions could scare off foreign investors who are seen as vital to South Africa's economyÓ.121 I argue that BITs are able to lock in those initial low level of regulations in the labor-related field and deter states from improving those regulations in the future through Òregulatory chillingÓ122.  In addition to the stabilization clause in investment contracts combined with the umbrella clause and the indirect expropriation clause that are already discussed, other BITs clauses such as the national treatment and fair and equitable treatment clauses are also likely to constrain host statesÕ capability of changing initial labor regulations. These two commonly seen BITs clauses prohibit host states from treating foreign investors unfairly. Foreign investors may challenge legislative changes in host states if those changes are perceived to affect foreign investors more than their domestic counterparts. Although investor-state disputes that are registered on the ground of labor-related issue are rare in international arbitration institutions, there are several publicly known lawsuits including French utility company Veolia v. Egypt in 2012 (partly due to increase in minimum wage), Russian investors v. Mongolia in 2007 (partly due to requirement for employing local workers)123, and Investors from Italy and Luxembourg v. South Africa in 2007 (partly due to requirement for hiring Òhistorically disadvantage South Africans)124 and Canadian mining company Centerra v. Kyrgyz Republic in 2006 (due to increasing high altitude premiums                                                 121 ÒThousands of Workers Strike in South Africa.Ó http://www.npr.org/2012/03/14/148617387/thousands-of-workers-strike-in-south-africa 122 After entering into force, BITs will protect foreign investors for 10 to 15 years. Furthermore, these investment agreements will apply to existing foreign investors for another 10 to 15 years after they expire or are terminated by a host government (survival clause).  123 Under the new 2006 Mineral Law, a mining company will be penalized by paying ten times the minimum monthly salary for each foreign national it employs above the maximum quota (10% of its workforce).  124 It requires about 40% participation rate of those historically disadvantage citizens in the management of mining companies under the new Mining Charter.   99 to workers). I then argue that the initial investment-attractive low labor conditions that are locked in by BITs are potential sources of domestic labor grievance and labor unrest.  Additionally, while inflows of foreign direct investment are projected to bring in economic development and many well-paid jobs for developing countries, it also tends to generate industrial strikes and labor unrest as a result of labor-related issues. For instance, about 1,900 workers went on a strike against Japanese multinational corporation Honda in Foshan, a southern city of China in May 2010, asking for increasing wage and the right to form an independent labor union125. Similar requests leading to labor unrest can be seen in protests against U.S. car multinational corporation Ford in Vsevolozhsk, Russia in 2005 (Greene and Robertson 2009). In the context of a cross-national study, Robertson and Teitelbaum (2011) find that inflows of FDI tend to generate labor protest in low- and middle-income countries. This finding is theorized with the following three reasons: (1) multinational corporation tend to locate in cities and hire a substantial amount of migrant workers from rural areas. Those workers are likely to be abused by employers and have less alternative ways (e.g., traditional social networks) to express their grievance other than protest. (2) workers intentionally target multinational corporation to make their demand through protest because they understand that those foreign companies will not repress the protest due to the concerns of international reputation. (3) the structure of multinational production with little built-in redundancy makes those companies vulnerable to the disruption of production in one country. Furthermore, workers in a country can make the comparison with the workers in the same multinational corporations in other countries and frame any different treatment (e.g., wage) as Òinternational injusticeÓ to mobilize protest.                                                  125 http://www.nytimes.com/2010/06/14/business/global/14honda.html?_r=0  100 When facing the threat of anticipated (deriving from labor grievances due to investment-attractive low labor conditions) or manifested labor unrest (generated by inflows of FDI), how do BITs affect host governmentsÕ calculation to balance the interests of foreign investors and domestic labors groups? Multinational corporations can directly refer to the Òfull protection and securityÓ clause for protection and challenge host governments in the presence of failing to put an end to the labor unrest. This clause is very common in the language of BITs. For instance, one of US company Noble VenturesÕs claims against Romania in international tribunal in 2001 was that Romanian government did not provide full protection and security to its investment during a period of labor unrest in spring and summer of 2001. In this matter the company challenged the government on the ground that local police refused to exercise adequate measures to stop the unlawful acts of labor unrest (e.g., theft, occupation and seizure of facilities, acts of intimidation)126. Furthermore, the existence of Òindirect expropriationÓ clauses can be invoked by foreign investors when host governments attempt to provide concessions to labor groups by improving higher labor-related standards in order to address the grass root of unrest. The less clear-cut and potentially very broad measures of indirect expropriation together with the lack of legal expertise in those low- and middle-income host countries may deter those governments from taking measures against foreign investors. With the strong protection of BITs for foreign investors, host governments may then refer to strategies that can reduce the collective action capability of labor groups, and hence relieve the threat of labor mobilization. I argue that host governments can either undercut the collective labor laws (e.g., ban the formation of independent labor union and the right to strike) or choose to undermine the practices of collective labor rights (e.g., interference of employers, authorities interfere with union rights of assembly and collective bargaining                                                 126 http://www.italaw.com/documents/Noble.pdf  101 processes, employ violence against union members or organizers) with the latter being a more preferable option. In general, passing legislations that reduce workersÕ collective rights is very difficult because it is very visible not only to domestic but also to international audiences. It is very likely to generate strong domestic oppositions and international criticisms127. Furthermore, this explicit and noticeable policy change may generate mobilization among opposed groups and backfire by promoting more labor protest and unrest. However, host governments can employ a less visible and more feasible strategy by reducing the enforcement of collective labor rights. They can either turn a blind eye on the violation of employers or actively interfere in workersÕ collective labor rights.128 Lack of enforcement of collective labor rights can substantially reduce the collective action capability of labor groups, and reduce the risk of potential or manifested labor grievance. I therefore propose the second hypothesis: Hypothesis 2: BITs are associated with a worsening of collective labor practices in developing countries.  4.3.3 The Impact of BITs on the Gap Between Labor Laws and Practices As already discussed, the strong BIT-empowered leverage that foreign investors have over host states force those states to undercut the collective action capability of labor groups in order to address the grass root of potential or manifested labor unrest. It is also important to investigate the impact of BITs on the gap between collective labor laws and labor practices. A low level of existing labor laws deters labor groups from engaging in collective action activities and already reflects a                                                 127 Cao and Prakash (2012) make a similar argument that when facing pressures from trade competitions, rewriting environmental regulations in order to lower environmental standards is very difficult. It is very visible and easily identifiable to domestic environmental constituencies and NGOs and likely to generate mobilization from opposition groups. Therefore, governments tend to relax enforcement of environmental standards in order to be competitive in the global trade.  128 Both dimensions are captured by Molsey (2011)Õs collective labor practices index.   102 relatively low level of their collective action capability. Therefore, in those countries the governments do not need to interfere in the practices of labor rights very much in order to undercut the labor groupÕs collective action capability, rendering the gap between labor law and labor practice relatively small. However, in the countries with relatively high level of labor laws, the potential threat of labor group mobilization is large because legally speaking they are allowed to have freedom of association and collective bargaining rights. It is in those countries that governments have more incentive to violate labor rights on the ground and hence deteriorate the de facto collective action capability of labor groups. Therefore, we might expect a larger gap between labor laws and labor practices. Following the discussion, I posit the last hypothesis: Hypothesis 3: BITs are associated with a larger gap between collective labor law and labor practices in developing countries.  4.4 Data, Measurement and Research Design Since my theory regarding BITs and collective labor rights applies to capital-receiving countries, it is appropriate to test my hypotheses in developing countries129. I define countries as developing countries if they are classified by the World Bank as non-high income for the majority of the sample period. The time period is from 1985 to 2002 dictated by the availability of the labor rights variables from Mosley (2011). The sample includes about 120 developing countries and the unit of analysis is country-year.  4.4.1 Dependent Variable I use one of the most fine-grained and widely used collective labor rights dataset constructed by Mosley (2011) that captures the freedom of association and collective bargaining rights. It                                                 129 Greenhill, Mosley, and Prakash (2009) and Mosley and Uno (2007) also investigate collective labor rights in developing countries.   103 covers almost all states from 1985 through 2002. This dataset records thirty-seven types of violations in the six broad categories: freedom of association, the right to form and join unions, rights to other union activities (e.g., to elect union leaders, enact union laws, control union finances), the right to bargain collectively, the right to strike, and the rights of workers in export-processing zones. Based on Kucera (2002)Õs methodology, a weighting is assigned to each violation depending on the severity of violations. More serious violations (e.g. general prohibitions on unions) receive larger weights than others (e.g. a requirement of previous authorization in order for a union to join a confederation of unions). The sum of the weighted scores produces the annual index of collective labor rights violations. For ease of interpretation, I use the reversed scale of this index, so that higher (lower) values indicate better (worse) collective labor rights. Mosley (2011) constructs this dataset based on detailed content analyses of reports on labor standards, drawing from three different sources - the U.S. State Department annual Country Reports on Human Rights Practices, reports from the International Labor OrganizationÕs Committee of Experts on the Applications of Conventions and Recommendations (CEACR) and the Committee on Freedom of Association (CFA), and the International Confederation of Free Trade UnionsÕ (ICFTU; now part of the International Trade Union Confederation) Annual Survey of Violations of Trade Union Rights. Since there are oftentimes discrepancies between the extent to which labor laws on the paper safeguarding workerÕs rights and the degree to which workerÕs rights are violated in practice especially in developing countries, Mosley (2011) codes collective labor law (legal provision of labor law) and collective labor practice (labor practice on the ground) separately and create two different labor rights indexes, making them particularly useful for testing my hypotheses. The correlation between these two indexes is just about 0.27, suggesting that they capture different  104 aspect of collective labor rights in developing countries. Furthermore, I employ the standardized version of these two measures130. The variable of collective labor law ranges from -4.17 to 0.89 while collective labor practice variable runs from -4.44 to 1.04 in my sample. A higher value indicates more protection of labor rights. The details of the coding can be found in the appendix of Greenhill, Mosley, and Prakash (2009). In order to test my Hypothesis 3, the dependent variable is constructed by subtracting labor practices score from labor laws score for a country in a given year. This difference serves as a proxy for the extent to which host governments need to undercut the collective action capability of domestic labor groups and relieve the potential threat posed by labor groups. Higher values represent more necessity and room to deteriorate the collective action capability of workers. This variable runs from -5.22 to 4.21.  4.4.2 Independent Variables  To test these three hypotheses, I measure the cumulative number of BITs ratified by a country in a given year. I prefer to use a total or cumulative count of BITs because it captures the amount of leverage that foreign investors have on host states and the potential severity of constraining effect on labor-related issues. The greater the number of BITs a host state ratifies, the greater the level of potential labor grievance or unrest and incentives of the host government to undermine the collective action capability of labor groups. I also opt for employing ratified BITs rather than signed BITs because only ratified BITs are legally binding (Haftel 2010). More specifically, I use three different versions of this variable. First of all, I do not differentiate different types of BITs. Instead I simply count all ratified BITs for a host state in a given year.                                                  130 The standardized variable has a mean of 0 and standard deviation of 1. These two standardized variables are employed mainly to construct my dependent variable for the last hypothesis Ð the gap between labor laws and labor practices. I will elaborate on it shortly.   105 Second, I explore the stringency of BITs by focusing on International Centre for Settlement of Investment Disputes (ICSID) mechanism and only count the number of BITs in which ICSID is an available option (either as one of many options or as the only option) for investor-state dispute resolution. These BITs capture the strength of investorsÕ legal protection. With the help of ICSID mechanism, foreign investors can bring host governments to binding international courts directly, without exhausting local remedies when disputes arise. Furthermore, the ICSID has leverage over developing countries due to its affiliation with the World Bank, limited grounds for appeal and high enforcement of awards (Blake 2013, Allee and Peinhardt 2010). I believe that the availability of ICSID for dispute arbitration enables foreign investors to have strong leverage over host states and have their interests protected by host states, making it an appropriate test for my theory. The measure is based on Allee and Peinhardt (2010) and my original coding of 420 additional BITs.  A final version of this independent variable explores another heterogeneity of BITs to get at the causal mechanism of my theoretical argument. I only include ÒNorth-SouthÓ BITs and disregard all ÒSouth-SouthÓ BITs.131 Only those BITs that regulate de facto investment flows are plausible to capture the lock-in effect of the initial labor conditions that are attractive to foreign investors and hence potential labor grievance as well as legal protection granted to investors when facing labor unrest. North-South BITs are very likely to govern de facto capital flows from rich developed countries to poor developing countries. It has also been argued that ÒSouth-SouthÓ BITs are concluded for reasons other than investment protection, making those BITs irrelevant for my theory (Elkin, Guzman, and Simmons 2006; Jandhyala, Henisz, and Mansfield 2011; Poulsen and                                                 131 A North-South BIT is defined as a BIT between a developed country and a developing country while a South-South BIT is a BIT between two developing countries.   106 Aisbett 2013)132. All the information of BITs is taken from the International Investment Agreements (IIA) database on the UNCTAD website. 4.4.3 Control Variables Following previous studies on labor rights in developing countries (e.g., Mosley and Uno 2007; Greenhill, Mosley, and Prakash 2009), I include a battery of standard economic and political controls. To account for the impact of overall economic integration on labor standards and a potential dynamic of Òrace to the bottomÓ (Mosley and Uno 2007; Neumayer and De Soysa 2006), I control trade openness and foreign direct investment inflows. The trade openness variable is measured as the sum of a stateÕs total exports and imports as a percentage of GDP (logged), capturing the overall dependence on trade for a country in a given year (World Bank World Development Indicators Ð WDI). I also incorporate a second variable that measures net foreign direct investment inflows as a share of GDP (UNCTAD). In addition to these two international factors, I also control for several domestic-level determinants of labor standards. Economic growth and GDP per capita in constant 2005 US dollars (logged) variables are included to account for the potential impact of economic conditions and development on domestic labor standards (Mosley and Uno 2007; Blanton, Blanton, and Peksen 2015). Both variables are drawn from WDI database.  As for political factors, I include the level of democracy variable to control for the influence of political regime type on collective labor rights. I utilize the Polity2 score from Polity IV dataset to capture the level of democracy. This variable takes on values between -10 to +10, a higher value                                                 132 It might be argued that a South-South BIT in which one of the country is a major capital exporting developing countries (e.g., China) is also able to capture the de facto cross-border flows. I follow Poulsen and Aisbett (2013) and construct a second measure of North-South BITs by taking into major capital exporting developing countries: Brazil, Russia, South Africa, China, Argentina, Panama, Mexico, Malaysia, Saudi Arabia, Indonesia, Hungary, Chile, and India. My main result is robust to this alternative measure of North-South BITs.  ! 107 of which represents higher level of democracy. Scholars generally find that democracy is positively associated with collective labor standards and respect for human rights (Mosley and Uno 2007; Neumayer and De Soysa 2006; Blanton, Blanton, and Peksen 2015; Davenport and Armstrong 2004; Davenport 2007; Bueno de Mesquita et al. 2005). Additionally, I control the size of population by measuring population per squared kilometer of land area (WDI). We might expect that countries with larger populations are likely to have more opportunities for various categories of labor rights violations, all else being equal. Finally, my models include a binary variable that captures the presence of civil war. Labor rights and human rights violations in general tend to increase during periods of domestic instabilities as governmentsÕ incentives and capabilities of enforcing laws are impaired (Mosley and Uno 2007; Hafner-Burton and Tsutsui 2005; Poe, Tate, and Keith 1999). The data source is from Peace Research Institute OsloÕs (PRIO) Armed Conflict Database (Harbom, Melander, and Wallensteen 2008). I also control for some additional variables in the robustness check. The summary statistics for all variables is shown in Table 4-1.  4.4.4 Model Specification  I employ an OLS regression with panel-corrected standard errors (Beck and Katz 1995) and adjust for first-order auto-correlation by specifying an AR(1) process. I also control for country-specific unobserved heterogeneity by including country dummies. Half-decade period dummies are also incorporated to account for the potential time-specific shocks or time trends that drive both the dependent variables and the key independent variable. All independent variables are lagged one year. The empirical model is shown as follows:  3456789:;#<=>?@$ABCADEFG=>?@CH6I<76J=CK>?@CL>CM@  108 I expect that  is negative when using collective labor practices variable while positive when using the gap between laws and practices variable, indicating that BITs tend to lower labor practice and enlarge the gap between laws and practices in developing countries. However,  will not be statistically significant from zero when collective labor laws index is employed in the model, suggesting that BITs do not have significant impact on collective labor laws.   Table 4-1. Summary Statistics  4.5 Results and Discussion Table 4-2 shows the main results of my empirical analysis. Models 1 to 3 display the impact of three different types of operationalization of the key independent variable Ð all types of BITs, BITs in which ICSID clause is available, and North-South BITs Ð on collective labor law.      2!2! 109 Table 4-2. Effect of BITs on Collective Labor Rights in Developing Countries 1985-2002  I find that ratified BITs (model 1) tends to increase the level of collective labor law while BITs with ICSID and North-South BITs do not have significant effect on collective labor law. Although this finding seems to be contrary to the concerns raised by civil groups and NGOs that BITs may have the potential to constrain the improvements of or even deteriorate legal labor standards in host countries, there are two caveats regarding the interpretation of these findings. First of all, the operationalization of BITs in model 1 also includes many South-South BITs that are ratified by host governments. Those types of BITs are concluded for reasons other than facilitating cross-border capital flow (Jandhyala, Henisz, and Mansfield 2011). I believe these BITs may not have much constraining effect on host statesÕ labor regulations comparing to North-South BITs. Thus, I prefer to rely on models 2 and 3 to make a conservative interpretation that BITs do not have a significant impact on collective labor law in host states. Second, the non-finding of the link between BITs and collective labor law does not necessarily imply that BITs will not affect labor laws at all in developing countries. When host governments have the political will or incentive to improve domestic labor laws (collective labor laws or other labor-enhancing policies such as  110 minimum wage), BITs are able to deter host governments from taking those actions. In contrast to weak result of collective labor laws, I find strong evidence that BITs tend to worsen collective labor practices in host states. The coefficients of BITs are negative and statistically significant across all three types of BITs variable operationalization (Models 4 to 6). Furthermore, as shown in models 7 through 9, I find that the coefficients are positive and statistically significant at at least 95% level across all three versions of BITs variables. It provides strong evidence that BITs also widen the gap between labor laws and labor practices in developing countries.  Regarding the control variables, I find that the level of democracy is positively associated with collective labor laws and labor practices. Mosley and Uno (2007) finds that trade openness reduces collective labor rights which is a composite index of labor law and labor practice. By differentiating labor law from labor practice, I find that trade openness does not have any significant effect on collective labor laws but it reduces collective labor practices. However, in contrast to Mosley and Uno (2007), the variable of FDI inflows has no significant impact on either measures of collective labor rights133. Also, population is negatively related with collective labor laws and labor practices while it only achieves statistical significance in the latter. I find no significant impact for GDP per capita and GDP growth. Finally, while civil war tends to reduce the capability and incentive to enforce labor laws and hence results in a low level of labor practice and a large gap between labor laws and labor practices, it is surprising to observe that civil war increases the level of collective labor laws.                                                    133 This non-finding is consistent with Greenhill, Mosley, and Prakash (2009), and Neumayer and de Soysa (2006).  111 4.5.1 Robustness Check I verify the robustness of the results in several ways. First of all, it is likely that the practice of labor rights on the ground is associated with labor laws on the paper, suggesting that the two separate equations of labor laws and labor practices may not be independent of one another. To account for the possibility that the error terms in these two equations may be correlated, I estimate seemingly unrelated regression (SUR) models. As is shown in Table 4-3, the main results are similar. Based on the Breusch-Pagan test of independence, the cross-equation error terms are significantly correlated, with correlations of about 0.12 in all these three SUR estimations. I further examine whether cross-equation coefficients of BITs variables are statistically different from each other in each two-equation system of SUR estimation. In all three SUR estimations, I find that the coefficients of BITs variable in labor laws equation and labor practices equation are statistically different at above 99% significance level, suggesting that the impact of BITs on labor laws and labor practices are very different from each other. A visual display of difference in cross-equation coefficients of BITs variables is shown in Figure 4-1. The difference in coefficients is calculated by subtracting the coefficient of BITs variable in labor laws equation from that in labor practices equation for each of the three SUR estimations. As it stands out, all the calculated differences are statistically significant as both of the upper and lower bounds of 95 percent confidence intervals exclude zero.       112 Table 4-3. Effect of BITs on Labor Laws and Practices Ð Seemingly Unrelated Regression (SUR Models)                            Figure 4-1. Difference of Cross-equation BITs Coefficients (SUR Models)   113 Second, I explore further by splitting countries into two categories: those for which their mean collective labor rights laws over the sample period (1985-2002) is above the mean and those for which their mean is below the mean of all developing countries134. The rationale behind this attempt is trying to investigate whether the impact of BITs on labor practices and the gap between labor laws and practices differs between countries with relatively strong labor laws and those with relatively weak labor laws on the paper. We might expect that host governments in countries with strong labor laws have more incentive to maneuver the labor practices on the ground when facing the pressure and leverage from foreign investors empowered by BITs, and undercut the collective action capability of labor groups. However, this might not be the case for the countries with weak labor laws. It is because in those countries the collective action capability of workers has already been relatively restricted by the weak labor laws. Tables 4-4 and 4-5 display the results for these two subsamples, respectively. As expected, the impact of BITs on labor practices is more evident in countries with relatively strong labor laws than those with weak labor laws. However, I find that BITs lead to wider gap between labor laws and practices in both subsamples.  Third, I reconstruct my key independent variables by only counting BITs that are ratified after 1989. Poulsen and Aisbett (2013) argues that it was not until the late 1980s that the vast majority of BITs gives foreign investors access to binding investor-state arbitration without exhausting domestic remedies. My theory centers on the leverage that multinational corporations have on host governments, which in turn incentivizes those governments to undermine the collective action capability of labor groups. Therefore, it may be argued that only counting those BITs in which                                                 134 My main results are similar if I use the mean of all countries in the world as the threshold to categorize countries into these two groups Ð above and below the mean.    114 binding consent to investor-state arbitration is available is a more appropriate test to capture the foreign investorsÕ leverage empowered by BITs. Table 4-6 shows that my results are robust.   Table 4-4. Subsample Ð Above the Mean of Labor Laws                        115 Table 4-5. Subsample Ð Below the Mean of Labor Laws   Table 4-6. Sample of BITs Ratified After 1989    116 I then test the robustness of my hypotheses by controlling some potential confounding factors: (1) I control for preferential trade agreements (PTAs) and ratification of International Labor Organization (ILO) standards. Hafter-Burton (2005) finds that ÒhardÓ PTAs in which human rights conditions are bundled with materialistic enforcement via trade sanction are able to force governments to respect domestic human rights. On the contrary, ÒsoftÓ PTAs in which materialistic punishment is not available for human rights violations are less likely to improve human rights in trade partners. Furthermore, improvement in general respect for human rights is likely to be associated with labor rights in host governments. Thus, I include two binary variables to capture whether each host state belongs to at least either one hard PTA or one soft PTA in a given year. The data is obtained from (Spilker and Bohmelt 2013). I also control for the ratification of the two core ILO conventions pertaining to freedom of association and collective bargaining rights135. It is possible that ratification of these two ILO conventions is likely to affect collective labor rights in host states136. I construct a ILO variable that accounts for the number of these two core ILO conventions ratified by a host state in a given year. The result is shown in Table 4-7. (2) I control for the number of years that countries are under either IMF or WB programs. IMF or World Bank structural adjustment programs are found to worse physical integrity rights (Abouharb and Cingranelli 2006, 2007, 2009) and collective labor rights (Blanton, Blanton, Peksen 2015a) in loan receiving countries. Table 4-8 displays the result. (3) I also include controls for the general respect for human rights Ð physical integrity rights, and rule of law. We may expect that host governments                                                 135 These two core ILO conventions are C87 (Freedom of Association and Protection of the Right to Organize Convention) created in 1948 and C98 (Right to Organize and Collective Bargaining Convention) created in 1949. The data source is from the ILOÕs NORMLEX database at: http://www.ilo.org/dyn/normlex/en/f?p=1000:12001:0::NO::: 136 However, Peksen and Blanton (2016) actually find that ratification of ILO core conventions tends to worsen domestic labor practices.   117 that respect human rights more broadly are more likely to have a better respect for labor rights as well (Payton and Woo 2014). I use the Cingranelli and Richards (2010)Õs measure of governmentsÕ respect for physical integrity rights. Furthermore, a host government with a strong rule of law may well respect and protect the rights of workers. The latent judicial independence index developed by Linzer and Staton (2011) is used as a proxy for the rule of law in host states. The result is shown in Table 4-9.  Table 4-7. Robustness Check for Controlling PTAs and ILOs            118 Table 4-8. Robustness Check for Controlling IMF/WB Structural Adjustment Program                            119 Table 4-9. Robustness Check for Controlling Human Rights and Rule of Law   Finally, I control for the number of labor-oriented non-governmental organizations (NGOs) and the level of media freedom in host states. It is possible that the observation of labor rights violations on the ground is not random and potentially biased, casting doubt on my main findings regarding the impact of BITs on labor practices and the gap between labor laws and practices. Violations of labor rights are more likely to be spotlighted in countries where press is free or a large number of labor-oriented NGOs is present. As a result, the measure of labor practices index may be biased as it is constructed based on textual reports. To adjust for this potential source of bias, I control for labor NGOs by counting the number of pro-labor rights organizations with international ties in a host state in a given year. This variable is drawn from Peksen and Blanton (2016)137. To account for the media freedom, I use the press freedom index from Freedom House                                                 137 The original data source is the Yearbook of International Organizations, the standard reference for international NGOs data. It is available at http://www.uia.org/yearbook  120 as a proxy. This index covers both printed and broadcast media. This variable is constructed in three categories: (i) Ònot freeÓ (0), (ii) Òpartly freeÓ (1), or (iii) free (2)138. The number in the parenthese is the value that I assign to each catogory139. Thus, a higher value indicates freer media. Table 4-10 displays the results. My main results are robust to all those additional control variables.    Table 4-10. Robustness Check for Controlling Labor NGOs and Media Freedom                                                     138 Since freedom house assigns two separate ratings to print and broadcast media between 1981 and 1988, I simply calculate the average of these two indexes for those time periods.  139 Besides the coding of press freedom into three categories, detailed coding on a scale of Ò0-100Ó is also available from Freedom House. But it only starts in 1993. It will restrict our analysis to the period of 1993-2002 and omit a substantial part of time period (1985-1992) from our original sample period. However, I also test the robustness of my results with this alternative measure of press freedom index. My main results are similar.   121 4.6!Conclusion While the vast majority of BITs literature investigates the link between BITs and FDI, this paper explores the potential negative externalities of concluding BITs by examining their impact on collective labor laws and labor practices. I argue that BITs have little impact on collective labor laws while they worsen collective labor practices and widen the gap between labor laws and practices in developing countries. Since foreign investors prefer to maintain the status quo in terms of regulations in host states including collective labor laws, conclusions of BITs do not necessarily lead to improving or worsening collective labor laws. However, BITs reduce collective labor practices in host states. BITs are able to lock in initial low labor conditions that are attractive to foreign investors, which may be a potential source of labor grievance and labor unrest. Also even if foreign direct investment may have a net positive impact on host states in terms of increasing income for workers (better paid comparing to domestic counterparts) and economic growth, foreign direct investment is found to increase domestic labor unrest as well (Robertson and Teitelbaum 2011). Given that all this anticipated or manifested labor grievance is likely to be challenged by foreign investors under stringent investment protective treaty clauses, host governments are forced to take measures against domestic labor groups in order to address the grass root of grievance. They can choose to undermine the collective action capability of domestic labor groups, and reduce the potential risk of labor unrest. I argue that host governments prefer to undercut the enforcement of labor law rather than rewrite to lower the collective labor laws because the former is less visible and less likely to introduce domestic opposition and international criticisms. As a result, we can also observe a larger gap between labor laws and labor practices. Using a sample of 120 developing countries from 1985 to 2002 and differentiating collective labor law from collective labor practice, I find support for my arguments.   122 However, the non-finding on collective labor laws does not suggest that BITs have no constraining effect on domestic policy autonomy in the field of labor rights. BITs appear able to deter host governments from improving domestic collective labor rights and other labor-enhancing policies (e.g., minimum wage). Furthermore, this study draws the unintended consequence of concluding BITs. This finding also backs the recent movement of BITs by incorporated labor-related clauses (de jure and de facto) in the main texts of BITs.                                  123 CHAPTER 5 CONCLUSION  Since its inception in 1959, bilateral investment treaties (BITs) have become the prominent legal instrument underlying the growth of cross-border foreign direct investment (FDI). Chapter 2 furthers our understanding of how developing countries integrate into this global economic treaty.  Specifically, this chapter examines the duration between signing and ratifying the BITs. It models domestic ratification of BITs as a dynamic process of updating the costs and benefits of BITs after the signature. I find strong evidence that BIT-related claims in peer states (either geographically close or in a similar foreign investor risk-profile) delays ratification of the observed BIT while rapid ratification rates in economic competitors facilitate the ratification of this BIT. The findings imply that international institutions (e.g., international investment arbitration institutions) act as the information provider for domestic audiences. Furthermore, this chapter considers international treaty ratifications among states as being interdependent as states learn the consequences of treaty commitments and ratification rates in other similarly situated states. The presence of BITs alleviates the time inconsistency problem plaguing the promotion of FDI. Therefore, developing countries view BITs as commitment devices and potential remedies for time inconsistency problem, leading to the expectation that signing or ratifying BITs are likely to increase inflows of foreign investment. While the empirical evidence on the positive impact of BITs on investment flows is quite mixed, the chapters 3 and 4 of this dissertation draw the attention to the potential negative externalities of concluding BITs for signatory developing countries. Given the Òbroad and asymmetricalÓ (Simmons 2014, p. 12) rights granted to foreign investors, chapter 3 finds that BITs have the potential to worsen human right practices in developing countries.  124 However, this negative impact is mitigated by democratic regime type. In chapter 4, I investigate the impact of BITs on collective labor rights in capital-hosting developing countries. The findings reveal that ratified BITs tend to deteriorate collective labor practice and widen the gap between collective labor laws and labor practices while they do not have significant impact on collective labor laws. These two studies draw the attention of the unintended consequences of concluding BITs and also back the recent movements of incorporating human right protected clauses in the text of BITs.          125                     BIBLIOGRAPHY                     126 BIBLIOGRAPHY  Abouharb, Rodwan, and David Cingranelli. 2006. ÒThe Human Rights Effects of World Bank Structural Adjustment, 1981-2000.Ó International Studies Quarterly 50(2): 233Ð62.  Abouharb, Rodwan, and David Cingranelli.  2007. Human rights and structural adjustment. New York: Cambridge University Press.  Abouharb, Rodwan, and David Cingranelli. 2009. ÒIMF Programs and Human Rights, 1981-2003.Ó Review of International Organizations 4(1): 47Ð72.  Aidt, Toke, and Zafiris Tzannatos. 2002. Unions and Collective Bargaining: Economic Effects in a Global Environment. Washington, DC: World Bank.   Alfaro, Laura. 2003. ÒForeign Direct Investment and Growth.Ó Harvard Business School Working Paper.  Alfaro, Laura, Areendam Chanda, Sebnem Kalemli-Ozcan, and Selin Sayek. 2004. ÒFDI and Economic Growth.Ó Journal of International Economics 64(1): 89-112.  Allee, Todd, and Clint Peinhardt. 2010. ÒDelegating Differences: Bilateral Investment Treaties and Bargaining over Dispute Resolution Provisions.Ó International Studies Quarterly 54(1): 1Ð26.  Allee, Todd, and Clint Peinhardt. 2011. ÒContingent Credibility: The Impact of Investment Treaty Violations on Foreign Direct Investment.Ó International Organization 65(3): 401Ð32.  Allee, Todd, and Clint Peinhardt. 2014. ÒEvaluating Three Explanations for the Design of Bilateral Investment Treaties.Ó World Politics 66(1): 47-87.  Amnesty International report. 2006. Human rights, trade and investment matters.  Amnesty International report. 2011. Eviction and resistance in Cambodia.   Apodaca, Claire. 2001. ÒGlobal Economic Patterns and Personal Integrity Rights After the Cold War.Ó International Studies Quarterly 45(4): 587Ð602.  Baccini, Leonardo, and Johannes Urpelainen. 2014. ÒBefore ratification: understanding the timing of international treaty effects on domestic policies.Ó International Studies Quarterly 58(1): 29-43.   Barry, Colin, Chad Clay, and Michael Flynn. 2013. ÒAvoiding the Spotlight: Human Rights Shaming and Foreign Direct Investment.Ó International Studies Quarterly 57(3): 532Ð544.  127 Beck, Nathaniel, and Jonathan Katz. 1995. ÒWhat to Do (and Not to Do) with Time Series Cross-Section Data.Ó American Political Science Review 89(3): 634 - 47.  Beck, Nathaniel, Jonathan Katz, and Richard Tucker. 1998. ÒTaking Time Seriously.Ó American Journal of Political Science 42(4): 1260 Ð 1288.  Bennett, Scott. 1997. ÒTesting Alternative Models of Alliance Duration, 1816-1984.Ó American Journal of Political Science 41(3):846-78.  Bergstrand, Jeffrey and Peter Egger. 2013. ÒWhat determines BITs?Ó Journal of International Economics 90(1): 107-122.  Bhattacharya, Amar, Peter Montiel, and Sunil Sharma. 1997. ÒHow Can Sub-Saharan Africa Attract More Private Capital Inflows?Ó Finance & Development 34:3-6.  Blake, Daniel. 2013. ÒThinking Ahead: Government Time Horizons and the Legalization of International Investment Agreements.Ó International Organization 67(4): 797-827.  Blanton, Shannon. 2005. ÒForeign Policy in Transition? Human Rights, Democracy, and U.S. Arms Exports.Ó International Studies Quarterly 49(4): 647Ð667.  Blanton, Shannon, and Robert Blanton. 2007. ÒWhat Attracts Foreign Investors? An Examination of Human Rights and Foreign Direct Investment.Ó Journal of Politics 69 (1): 143Ð55.  Blanton, Robert, and Shannon Blanton. 2012. ÒLabor rights and foreign direct investment: Is there a race to the bottom?Ó International Interaction 38(3): 267Ð94  Blanton, Robert, Shannon Blanton, and Dursun Peksen. 2015a. ÒThe Impact of IMF and World Bank Programs on Labor Rights.Ó Political Research Quarterly 68(2): 324Ð36.  Blanton, Robert, Shannon Blanton, and Dursun Peksen. 2015b. ÒFinancial Crises and Labor: Does Tight Money Loosen Labor Rights?Ó World Development 76: 1Ð12.  Blonigen, Bruce, and Grace Wang. 2005. ÒInappropriate Pooling of Wealthy and Poor Countries in Empirical FDI Studies.Ó In Does Foreign Investment Promote Development? Ed. Theodore Moran, Edward Graham, and Magnus Blomstrom. Washington, DC: Institute for International Economics, 221-44.  Boockmann, Bernhard. 2001. ÒThe ratification of ILO conventions.Ó Economics & Politics 13 (3): 281-309.  Brambor, Thomas, William Clark, and Matt Golder. 2006. ÒUnderstanding Interaction Models.Ó Political Analysis 14(1): 63Ð82.    128 Bueno de Mesquita, Bruce, George Downs, Alistair Smith, and Feryal Cherif. 2005. ÒThinking Inside the Box: A Closer Look at Democracy and Human Rights.Ó International Studies Quarterly 49(3): 439Ð57.  Buthe, Tim, and Helen Milner. 2008. ÒThe politics of foreign direct investment into developing countries.Ó American Journal of Political Science 52(4):741Ð762.  Buthe, Tim, and Helen Milner. 2009. ÒBilateral Investment Treaties and Foreign Direct Investment.Ó In The Effect of Treaties on Foreign Direct Investment: Bilateral Investment Treaties, Double Taxation Treaties, and Investment Flows Ed. Karl Sauvant and Lisa Sachs. New York: Oxford University Press, 171-224.  Buthe, Tim, and Helen Milner. 2014. ÒForeign Direct Investment and Institutional Diversity in Trade Agreements.Ó World Politics 66(1): 88Ð122.  Campello, Daniela, and Leany Lemos. 2015. ÒThe non-ratification of bilateral investment treaties in Brazil.Ó Review of International Political Economy 22: 1-32.  Cao, Xun, and Aseem Prakash. 2010. ÒTrade competition and domestic pollution.Ó International Organization 64(3): 481Ð503.  Cao, Xun, and Aseem Prakash. 2012. ÒTrade competition and environmental regulations: Domestic political constraints and issue visibility.Ó The Journal of Politics 74(1): 66-82.  Chapman, Terrence. 2009. ÒAudience Beliefs and International Organization Legitimacy.Ó International Organization 63(4): 733Ð64.  Carkovic, Maria, and Ross Levine. 2005. ÒDoes Foreign Direct Investment Accelerate Economic Growth? In Does Foreign Direct Investment Promote Development? Ed. Theodore Moran, Edward Graham, and Magnus Blomstrom. Washington, DC: Institute for International Economics, 195-220.  Chau, Nancy, Ravi Kanbur, Ann Harrison, and Peter Morici. 2001. ÒThe Adoption of International Labor Standards Conventions.Ó Brookings Trade Forum. Brookings Institution Press: 113Ð56.  Cingranelli, David, and David Richards. 1999. ÒMeasuring the Level, Pattern, and Sequence of Government Respect for Physical Integrity Rights.Ó International Studies Quarterly 43(2):407-417.  Cingranelli, David, and David Richards. 2010. ÒThe Cingranelli and Richards (CIRI) Human Rights Data Project.Ó Human Rights Quarterly 32(2): 401Ð24.  Cingranelli, David, and Mikhail Filippov. 2010. ÒElectoral Rules and Incentives to Protect Human Rights.Ó The Journal of Politics 72(1): 243-257.   129 Conrad, Courtenay, and Jacquelline DeMeritt. 2013. ÒConstrained by the bank and the ballot.Ó Journal of Peace Research 50(1): 105-119.  Dai, Xinyuan. 2005. ÒWhy Comply? The Domestic Constituency Mechanism.Ó International Organization 59(2): 363Ð98.  Davenport, Christian. 1995. ÒMulti-Dimensional Threat Perception and State Repression.Ó American Journal of Political Science 39(3):683Ð713.  Davenport, Christian. 2007. State Repression and the Domestic Democratic Peace. New York: Cambridge University Press.  Davenport, Christian, and David Armstrong II. 2004. ÒDemocracy and the Violation of Human Rights.Ó American Journal of Political Science 48(3): 538Ð54.  Davies, Ronald, and Krishna Vadlamannati. 2013. ÒA race to the bottom in labor standards? An empirical investigation.Ó Journal of Development Economics 103 (July): 1Ð14.   Elkins, Zachary, Andrew Guzman, and Beth Simmons. 2006. ÒCompeting for Capital: The Diffusion of Bilateral Investment  Treaties, 1960Ð2000.Ó International Organization 60(4): 811Ð46.  Fredriksson, Per, and Noel Gaston. 2000. ÒRatification of the 1992 climate change convention.Ó Public Choice 104(3): 345-368.  Freeman, Nathan. 2009. ÒThe International Investor Rights Regime: Domestic Sources of Commitment and Compliance.Ó Diss. University of Georgia.  Frieden, Jeffry. 1991. ÒInvested interests.Ó International Organization 45(4): 425Ð451.  Frundt, Henry. 1987. Refreshing Pauses. New York: Praeger.  Gallagher, Kevin, and Elen Shrestha. 2011. ÒInvestment Treaty Arbitration and Developing Countries.Ó Global Development and Environment Institute Working Paper No. 11-01.   Gelleny, Ronald, and Matthew McCoy. 1999. ÒTaxing Capital in a Global Economy.Ó Paper presented at the Midwest Political Science Association Meeting, Chicago.  Gibney, Mark, Linda Cornett, Reed Wood, and Peter Haschke. 2012. ÒPolitical Terror Scale 1976-2012.Ó Data from the Political Terror Scale Web site: http://www.politicalterrorscale.org/  Gill, Judith, Matthew Gearing, and Gemma Birt. 2004. ÒContractual Claims and Bilateral Investment Treaties.Ó Journal of International Arbitration 21(5): 397-412.  Ginsburg, Tom. 2005. ÒInternational substitutes for domestic institutions.Ó International Review of Law and Economics 25(1): 107-123.  130 Gleditsch, Nils, Havard Strand, Mikael Eriksson, Margareta Sollenberg, and Peter Wallensteen. 2002. "Armed Conflict 1946-2001." Journal of Peace Research 39 (5): 615-37.  Goodliffe, Jay, and Darren Hawkins. 2006. ÒExplaining Commitment.Ó Journal of Politics 68 (2):358Ð371.  Greene, Samuel, and Graeme Robertson. 2010. ÒPolitics, Justice and the New Russian Strike.Ó Communist and Post-Communist Studies 43(1): 73Ð95.  Greenhill, Brian, Layna Mosley, and Aseem Prakash. 2009. ÒTrade-Based Diffusion of Labor Rights: A Panel Study , 1986 Ð 2002.Ó American Political Science Review 103(4): 1986Ð2002.  Hafner-Burton, Emilie. 2005a. ÒRight or Robust? The Sensitive Nature of Repression to Globalization.Ó Journal of Peace Research 42(6): 679Ð98.  Hafner-Burton, Emilie. 2005b. ÒTrading Human Rights: How Preferential Trade Agreements Influence  Government Repression.Ó International Orgainzation 59(3): 593-629.  Hafter-Burton, Emilie. 2009. Forced to Be Good. Ithaca, NY: Cornell University Press.  Hafner-Burton, Emilie, and Kiyoteru Tsutsui. 2005. ÒHuman Rights in a Globalizing World: The Paradox of Empty Promises.Ó American Journal of Sociology 110(5): 1373-411.  Haftel, Yoram. 2010. ÒRatification Counts: US Investment Treaties and FDI Flows into Developing Countries.Ó Review of International Political Economy 17(2): 348Ð77.  Haftel, Yoram, and Alexander Thompson. 2012. ÒThe Renegotiation of International Agreements: The Case of Bilateral Investment Treaties.Ó Working Paper.   Haftel, Yoram, and Alexander Thompson. 2013. ÒDelayed Ratification: The Domestic Fate of Bilateral Investment Treaties.Ó International Organization 67(2):253Ð275.  Haggard, Stephan, and Sylvia Maxfield. 1996. ÒThe Political Economy of Financial Internationalization in the Developing World.Ó In Internationalization and Domestic Politics. Ed. Robert Keohane and Helen Milner. Cambridge: Cambridge University Press, 209-242.  Hallward-Dreimeier, Mary. 2003. ÒDo Bilateral Investment Treaties Attract Foreign Direct Investment? A Bit . . . and They Could Bite.Ó World Bank Working Paper No. 3121.  Hanson, Gordon, Raymond Mataloni, and Matthew Slaughter. 2003. ÒVertical Production Networks in Multinational Firms.Ó NBER Working Paper No.9723.  Harbom, Lotta, Erik Melander, and Peter Wallensteen. 2008. ÒDyadic Dimensions of Armed Conflict, 1946!2007.Ó Journal of Peace Research 45:697Ð710.   131 Hathaway, Oona. 2008. ÒTreatiesÕ End: The Past, Present, and Future of International Lawmaking in the United States.Ó Yale Law Journal 117 (7):1236 - 372.  Huber, Evelyne, and John Stephens. 2001. Development and Crisis of the Welfare State. Chicago: University of Chicago Press.   Jacob, Marc. 2010. ÒInternational investment agreements and human rights.Ó INEF Research Paper Series on Human Rights, Corporate Responsibility and Sustainable Development 03/2010.  Jandhyala, Srividya, Witold Henisz, and Edward Mansfield. 2011. ÒThree Waves of BITs: The Global Diffusion of Foreign Investment Policy.Ó Journal of Conflict Resolution 55(6): 1047Ð73.  Jensen, Nathan. 2003. "Democratic governance and multinational corporations." International Organization 57(3): 587-616.  Keith, Linda, and Steven Poe. 2000. ÒThe United States, the IMF, and human rights.Ó In The United States and human rights. Ed. David Forsythe. Lincoln, NE: University of Nebraska Press, 273-299.  Kerner, Andrew. 2009. ÒWhy Should I Believe You? The Costs and Consequences of Bilateral Investment Treaties.Ó International Studies Quarterly 53(1): 73Ð102.  Kim, Moonhawk. 2012. ÒEx Ante Due Diligence: Formation of PTAs and Protection of Labor Rights.Ó International Studies Quarterly 56(4): 704Ð19.  Klemm, Alexander, and Stefan Parys. 2012. ÒEmpirical evidence on the effects of tax incentives.Ó International Tax and Public Finance 19(3): 393Ð423.  Kucera, David. 2002. ÒCore Labour Standards and Foreign Direct Investment.Ó International Labour Review 141(1!2):31!69.  La Porta, Rafael, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2008. ÒThe Economic Consequences of Legal Origins.Ó Journal of Economic Literature 46 (2): 285Ð332.  Leeds, Brett, Jeffrey Ritter, Sara Mitchell, and Andrew Long. 2002. ÒAlliance Treaty Obligations and Provisions, 1815Ð1944.Ó International Interactions 28 (3): 237Ð 260.  Leo, Benjamin. 2010. ÒWhere are the BITs? How U.S. Bilateral Investment Treaties with Africa Can Promote Development.Ó Center For Global Development Essay.   Li, Quan. 2006. ÒDemocracy, autocracy, and tax incentives to foreign direct investors.Ó Journal of Politics 68(1): 62-74.   132 Li, Quan, and Adam Resnick. 2003. ÒReversal of Fortunes.Ó International Organization 57(1):175-211.  Li, Xiaoying, and Xiaming Liu. 2005. ÒForeign Direct Investment and Economic Growth.Ó World Development 33(3): 393-407.  Licht, Amanda. 2011. ÒChange comes with time: Substantive interpretation of nonproportional hazards in event history analysis.Ó Political Analysis 19 (2): 227-243.   Linzer, Drew, and Jeffrey Staton. 2011. ÒA measurement model for synthesizing multiple comparative indicators: The case of judicial independence.Ó Working paper.  Lupu, Yonatan and Paul Poast. 2013. ÒAre bilateral investment treaties really bilateral?Ó Working paper.   Mansfield, Edward, and Helen Milner. 2012. Votes, Vetoes, and the Political Economy of International Trade Agreements. New Jersey: Princeton University Press.  Meyer, William. 1996. ÒHuman Rights and MNCs.Ó Human Rights Quarterly 18(2): 368Ð397.  Milner, Helen. 2014. ÒIntroduction Ð The global economy, FDI, and the regime for investment.Ó World Politics 66(1): 1-11.  Milner, Helen, and Peter Rosendorff. 1997. ÒDemocratic Politics and International Trade Negotiations.Ó Journal of Conflict Resolution 41(1): 117-146.  Moore, Will. 1998. ÒA Repression and Dissent.Ó American Journal of Political Science 45(3): 851Ð73.  Mosley, Layna. 2011. ÒReplication data for: Collective Labor Rights DatasetÓ.   Mosley, Layna, and Saika Uno. 2007. ÒRacing to the Bottom or Climing to the Top? Economic Globalization and Collective Labor Rights.Ó Comparative Political Studies 40(8): 923Ð48.  Murdie, Amanda and David Davis. 2012. ÒShaming and Blaming: Using Events Data to Assess the Impact of Human Rights INGOs.Ó International Studies Quarterly 56(1): 1-16.  Neumayer, Eric. 2002. ÒDoes Trade Openness Promote Multilateral Environmental Cooperation?Ó The World Economy 25 (6): 815-832.  Neumayer, Eric. 2006. ÒSelf!Interest, Foreign Need, and Good Governance.Ó Foreign Policy Analysis 2(3): 245-268.  Neumayer, Eric, and Indra De Soysa. 2006. ÒGlobalization and the Right to Free Association and Collective Bargaining: An Empirical Analysis.Ó World Development 34(1): 31Ð49.   133 Neumayer, Eric, and Laura Spess. 2005. ÒDo Bilateral Investment Treaties Increase Foreign Direct Investment to Developing Countries?Ó World Development 33 (10): 1567-85.  Neumayer, Eric, Peter Nunnenkamp, and Martin Roy. 2014. ÒAre stricter investment rules contagious? Host country competition for foreign direct investment through international agreements.Ó Kiel Working Paper No.1910.  Neumayer, Eric and Thomas Plper. 2010. ÒSpatial effects in dyadic data.Ó International Organization 64(1): 145-166.  Nikiema, Suzy. ÒBest Practices Ð Indirect Expropriation.Ó International Institute for Sustainable Development working paper March 2012.   Nordas, Ragnhild, and Christian Davenport. 2013. ÒFight the youth.Ó American Journal of Political Science 57 (4): 926-940.   Payton, Lockwood, and Byungwon Woo. 2014. ÒAttracting Investment: GovernmentsÕ Strategic Role in Labor Rights Protection.Ó International Studies Quarterly 58(3): 462Ð74.  Peksen, Dursun, and Robert Blanton. 2016. "The Impact of ILO conventions on worker rights: Are empty promises worse than no promises?" Review of International Organization.   Peterson, Luke. 2009. ÒHuman Rights and Bilateral Investment Treaties.Ó Right and Democracy Working Paper.   Peterson, Luke, and Kevin Gray. 2003. ÒInternational human rights in bilateral investment treaties and in investment treaty arbitration.Ó International Institute for Sustainable Development Working Paper.   Poe, Steven, Sabine Carey, and Tanya Vazquez. 2001. ÒHow Are These Pictures Different? A Quantitative Comparison of the US State Department and Amnesty International Human Rights Reports, 1976Ð1995.Ó Human Rights Quarterly 23(3): 650Ð77.  Poe, Stephen, Neal Tate, and Linda Keith. 1999. ÒRepression of the Human Right to Personal Integrity Revisited: A Global Cross-National Study Covering the Years 1976Ð1993.Ó International Studies Quarterly 43 (2): 291Ð313.  Postnikov, Evgeny, and Ida Bastiaens. 2014. ÒDoes Dialogue Work? The Effectiveness of Labor Standards in EU Preferential Trade Agreements.Ó Journal of European Public Policy 21(6): 923Ð40.  Poulsen, Lauge. 2014. ÒBounded rationality and the diffusion of modern investment treaties.Ó International Studies Quarterly 58(1): 1Ð14.  Poulsen, Lauge, and Emma Aisbett. 2013. ÒWhen the Claim Hits: Bilateral Investment Treaties and Bounded Rational Learning.Ó World Politics 65(2): 273Ð313.  134 Prislan, Vid, and Ruben Zandvliet. 2013. ÒLabor provisions in international investment agreements: Prospects for sustainable development.Ó Grotius Centre Working Paper Series 003.  Richards, David, Ronald Gelleny, and David Sacko. 2001. ÒMoney with a Mean Streak? Foreign Economic Penetration and Government Respect for Human Rights in Developing Countries.Ó International Studies Quarterly 45(2): 219Ð39.  Robertson, Graeme, and Emmanuel Teitelbaum. 2011. ÒForeign Direct Investment, Regime Type, and Labor Protest in Developing Countries.Ó American Journal of Political Science 55(3): 665Ð77.  Rosendorff, Peter, and Kong Joo Shin. 2015. ÒRegime type and international commercial agreements.Ó International Journal of Economic Theory 11(1): 107-119.  Ruane, Frances and Ali Ugur. 2005. ÒExport Platform FDI and Dualistic Development.Ó Transnational Corporations. Forthcoming.   Rudra, Nita. 2002. ÒGlobalization and the decline of the welfare state in less-developed countries.Ó International Organization 56(2): 411-445.  Rudra, Nita. 2011. ÒOpenness and the Politics of Potable Water.Ó Comparative Political Studies 44(6): 771-803.   Schneider, Christina, and Johannes Urpelainen. 2013. ÒDistributional conflict between powerful states and international treaty ratification.Ó International Studies Quarterly 57(1): 13-27.  Simmons, Beth. 2009. Mobilizing for Human Rights: International Law in Domestic Politics. New York: Cambridge University Press.  Simmons, Beth. 2014. ÒBargaining over BITs, Arbitrating Awards: The Regime for Protection and Promotion of International Investment.Ó World Politics 66(1): 12Ð46.  Sorens, Jason, and William Ruger. 2012. ÒDoes Foreign Investment Really Reduce Repression?Ó International Studies Quarterly 56(2): 427Ð436.  Spilker, Gabriele, and Tobias Bohmelt. 2013. ÒThe Impact of Preferential Trade Agreements on Governmental Repression Revisited.Ó Review of International Organizations 8(3): 343Ð61.  Staiger, Douglas, and James Stock. 1997. ÒInstrumental variables regression with weak instruments.Ó Econometrica 65(3):557Ð586.   Swenson, Doborah. 2005. ÒWhy do developing countries sign BITs?Ó U.C. Davis Journal of International Law and Policy 12(1): 131-155.   135 Swenson, Deborah. 2009. ÒWhy Do Countries Sign BITs?Ó In The Effect of Treaties on Foreign Direct Investment: Bilateral Investment Treaties, Double Taxation Treaties, and Investment Flows. Edited by Karl P. Sauvant and Lisa E. Sachs. New York: Oxford University Press, 437-457.  Tobin, Jennifer, and Marc Busch. 2010. ÒA BIT is better than a lot.Ó World Politics 62(1): 1-42.  Tobin, Jennifer and Rose-Ackerman, Susan. 2005. ÒForeign direct investment and the business environment in developing countries.Ó Yale Law School Center for Law, Economics and Public Policy Research Paper No. 293.  UNCTAD (United Nations Conference on Trade and Development). 2004. World Investment Report. New York Ð Geneva: United Nations.  UNCTAD. 2012. Fair and Equitable Treatment. Technical report New York  Vadlamannati, Krishna. 2009. ÒEconomic interest versus social conscience.Ó Working paper.   Van Harten, Gus. 2012. ÒArbitrator Behavior in Asymmetrical Adjudication.Ó Osgoode Hall Law Journal 50(1): 211Ð68.  Von Stein, Jana. 2008. ÒThe international law and politics of climate change.Ó Journal of Conflict Resolution 52 (2): 243-268.  Wellhausen, Rachel. 2015. ÒInvestor-state disputes: when can governments break contracts?Ó Journal of Conflict Resolution 59(2): 239-261.   Wellhausen, Rachel. 2015. ÒBondholders vs. Direct Investors? Competing Responses to Expropriation.Ó International Studies Quarterly 59(4):750Ð764.  Wellhausen, Rachel. 2016. ÒRecent Trends in Investor-State Dispute Settlement.Ó Journal of International Dispute Settlement 7(1): 117-135.  Wood, Reed. 2008. ÒA Hand upon the Throat of the Nation.Ó International Studies Quarterly 52(3): 489Ð513.  Wright, Thorin. 2014. ÒTerritorial revision and state repression.Ó Journal of Peace Research 51(3):375Ð387.  Yackee, Jason. 2007. ÔÔDo BITs Really Work?ÕÕ University of Wisconsin Legal Studies Research Paper No. 1054, Madison, WI.