BELT AND ROAD INVESTMENTS AND ETHNIC POLITICS IN MALAYSIA By Alexander R. Mardon A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Political Science – Doctor of Philosophy 2022 ABSTRACT Chinese investments under the auspices of the Belt and Road Initiative (BRI) have accelerated throughout Asia in recent years, introducing new economic and political forces into host countries’ domestic political spaces. This study explores the effects of the BRI on an aspect of host country domestic politics that has received little attention to date: the strategic transformation of BRI investment projects into ethnic issues by political actors and the receptiveness of voters to this process. This study leverages interview data and an original survey of 1,308 Malaysian voters collected over five months of fieldwork in Malaysia. It employs qualitative techniques as well as a set of two conjoint experiments to identify the process through which BRI projects become domestic ethnic issues and to isolate how Malaysian voters use ethnic cues to guide their preferences on investment projects and candidates. The results find strong effects of the ethnic and national origins of investment projects and of respondents’ own ethnic identity in determining Malaysian voters’ preferences for investments and candidates. These results hold important insights for scholars of ethnic politics, international relations, and observers of growing Chinese influence in countries that receive BRI funds. To my family iii ACKNOWLEDGEMENTS This dissertation is the product of many years’ worth of meeting challenges that were as much mental as related to the research itself. I can confidently say that it is only due to the constant support and guidance of my network of mentors, colleagues, family, and friends that I was able to complete a dissertation that I am proud of and that, I hope, can offer some useful contributions to political science research. First, I would like to thank my dissertation committee for their advice, guidance, and patience. My committee chair, Jeff Conroy- Krutz, dedicated countless hours in helping me carry out my research, critiqued and edited draft after draft of each chapter, and gave me confidence in my work. Eric Chang, Erica Frantz, and Christian Houle taught me how to conduct political science research in their courses, provided focus to my dissertation, and were always helpful with their advice. I am truly grateful to the members of my dissertation committee for their mentorship and friendship throughout my time at Michigan State. Additionally, the rest of the faculty and staff in the political science department were vital to this process. Without their help, navigating graduate student life simply would not have been possible. I thank them for helping me to keep this project on track and for the generous funding they helped me to obtain for my research. I am also indebted to many of my fellow graduate students and colleagues during this period. Thank you in particular to Nate Smith, iv Caleb Lucas, Kangwook Han, Tara Iseneker, Steve Anderson, and Chrissy Scheller for the years of discussion, encouragement, and friendship. And finally, I thank my family for their endless support. To Mom and Dad, thank you for the inspiration, encouragement, and patience. To Dyana and Andrew, thanks for always being there for me. To Aunt Nore and Grandma, thank you for giving me a home in Michigan that offered a break from the graduate school grind. And last, to my wife Sarah, thank you for your love and belief in me. v TABLE OF CONTENTS Chapter 1: Introduction .............................................. 1 Chapter 2: Theory ................................................... 28 Chapter 3: Malaysian Politics, Ethnic Relations, and the Belt and Road Initiative .......................................................... 53 Chapter 4: Citizen Preferences for Investment Projects .............. 94 Chapter 5: Chinese Investments and Candidate Preference ............ 130 Chapter 6: Conclusion .............................................. 162 BIBLIOGRAPHY ....................................................... 174 vi Chapter 1: Introduction The Asia-Pacific region, home to over half of the world’s population, has in recent years witnessed a transformational shift in the regional balance of power in the ascendancy of China. China has invested heavily to advance its strategic and economic goals in the region. Under the leadership of Xi Jinping, in 2013 China launched its flagship foreign policy endeavor, the Belt and Road Initiative (BRI), spurring rapid growth in Chinese investment activity across Asia and parts of Africa. The vast scale of investments under the BRI umbrella has altered the regional financial landscape, encouraging leaders in host countries to reorient their foreign policies, and giving birth to a proliferation of speculation as to the economic and geopolitical consequences of a more assertive China by the media and in academic circles. The BRI’s close official and symbolic association with China has the potential to interact with fragile ethnic relations in host nations. Many BRI recipient countries, particularly throughout the Asia-Pacific, are home to a historical ethnic Chinese diaspora whose distinct cultural, religious, linguistic, and historical experiences have often been used to encourage political mobilization. Counter- mobilizations of “native” or “indigenous” ethnic groups have also exploited strategies of “othering” Chinese groups as a way of constructing political coalitions (Baginda 2016, Holst 2012). In Malaysia, which receives among the largest infusions of BRI capital, the dominant political parties are constructed along explicitly ethnic lines between “indigenous” and “Malaysian Chinese” 1 parties, which perpetuate systems of “ethnicized” policies (Holst 2012). In this dissertation, the ethnicization of the BRI and other international issues refers to the process by which an issue becomes ethnic in character. Rising Chinese influence via the BRI has the potential to interact with preexisting ethnic tensions in Malaysia and other BRI recipient countries, and Malaysian elites have already evoked BRI to rally political support. The increasing influence of China via initiatives such as the BRI has the potential to exacerbate ethnic tensions. The transformation of BRI into an ethnic issue in Malaysia is compelling because, on its face, BRI lacks a strong ethnic component. It is an economic program designed to boost regional economic development, or to enhance Chinese influence abroad, depending on one’s point of view. Chinese leaders have even gone to lengths to present the BRI in a benevolent light abroad and to dispel suspicions of anything more than economic interest in its partner countries (Rolland 2017). But Malaysian political discourse has defied these efforts, and tied finance from mainland China to ethnic competition between the Bumiputera and Malaysian Chinese ethnic groups. This process of transfiguration from an international economic program to a domestic ethnic issue forms the basis of this dissertation. Malaysia’s system of ethnic politics, and the ability of the BRI to be tied to it, contribute to a situation in which an international issue (the BRI) has become ethnicized in Malaysia. What are the roles played by candidates and voters in this process, and why do they pursue an ethnic narrative? How does ethnicity factor into public 2 opinion and preferences for foreign investment projects? And ultimately, how does it affect vote choice? In addressing these questions, this study attempts to establish the extent that BRI investments resonate with voters as a political issue, and to identify the dimensions of investment projects that matter most in forming voter opinions. Using primary data from a pair of conjoint experiments and survey of 1,308 Malaysian citizens, this study consolidates a wide-ranging literature with a test that allows for consideration of multiple explanatory factors simultaneously. It directly examines an understudied yet central factor in determining the effects and future prospects of Chinese ascendancy and the BRI: voter opinion in democratic host countries. In doing so, this study can shed light on how citizens process and react to a major geopolitical and economic change in their countries. I find that the respondent’s ethnicity is influential in determining support or opposition to investments: ethnic Malay citizens are more likely to exhibit strong disapproval of foreign investments connected to foreign Chinese investors, while Malaysian Chinese citizens are less likely to distinguish between foreign Chinese investments and other foreign investments, but prefer fellow Malaysian Chinese investors over all other investors. This provides support for the idea that large-scale Chinese investments under initiatives such as the BRI activate ethnic identities among citizens in BRI recipient countries and stands as a potentially powerful political issue in host countries’ domestic politics. In contexts where ethnic identity is salient, which is common in the Asia-Pacific 3 region, BRI investments may contribute to rising use of ethnic issues in campaigns. This presents a situation to monitor as Chinese foreign policy becomes increasingly assertive in regions where Chinese identity is politically salient. This study benefits from and contributes to several major bodies of academic research that help to explain the ethnicization of BRI in Malaysia. One such literature is known as the “second image reversed,” in which scholars have examined the ability of international economic forces to alter the nature of domestic political coalitions and voting behavior (Gourevitch 1978, Milner and Keohane 1996). They point out that international forces have differential effects on local actors that alter their incentives and opportunities in systematic ways. Much of the work in this field has identified the conditioning effect that host country institutions have on this relationship. This field generally has focused on domestic economic cleavages, and operates under assumptions that individuals are motivated by their own rational self-interest. Yet the idea that international economic issues can affect domestic politics can be expanded to include identity cleavages that are not necessarily bound by economic interest and that can be motivated by a combination of material and nonmaterial interests. An additional contribution to this literature is in my exploration of the ability of the BRI to be used as a tool for ethnic mobilization based on both material and expressive rationales. To date, there has been relatively little exploration of how international factors can affect domestic ethnic politics. Some studies have noted that international economic crises 4 have contributed to ethnic conflict, such as in Indonesia following the 1997 Asian Financial Crisis (Khoo 2004). However, to my knowledge, these studies have yet to offer a set of theoretical scope conditions that specifically describe when international issues are likely to activate ethnic tensions. I argue that international factors that are less “extreme” than crises can be shown to affect ethnic politics, in ways that may be more subtle than outright ethnic conflict. My study therefore offers a significant theoretical contribution to the second image reversed literature by calling attention to the relationship between international economic developments and domestic ethnic politics, which may be more robust and commonplace than previously thought. Additionally, this study is closely tied to the field of ethnic politics. Scholars have noted the ability of ethnic identity to arouse powerful emotional responses among ethnic communities, and induce people to act in solidarity with their coethnics, at times even against their own material interest (Horowitz 1985, Dickson and Scheve 2006). Recent advances in this field have further established that ethnicity carries a strong instrumental incentive for political mobilization, as it can be used to win public office, monitor coethnics (Kamara 2007), and distribute the spoils of political power (Bates 1974). While these two general approaches offer divergent interpretations of the role that ethnicity plays politically, they likely both offer relevant insight into the case at hand. An international issue can become part of a country’s ethnic narrative via materialistic or expressive pathways. Politicians have clear 5 material incentives to manipulate the issues to be perceived as closely tied to an ethnic group’s fortunes, and citizens can be motivated to accept such narratives out of either an expectation of personal benefit or as a means of showing group solidarity. Most studies focus on domestic forces or issues that elicit this type of ethnicized response. This study contributes to this literature by identifying that the mechanisms of issue ethnicization also can be expanded to the international arena, even with issues that are not inherently ethnic in nature. In this way, it shows the breadth and adaptability of ethnic political forces, and suggests that in a globalized economic context, ethnic politics may show a robustness and resilience to encompass a wider array of topics than previously thought. A third area to which this study contributes is the growing body of literature that has specifically outlined the expansion of Chinese activities and influence via BRI. Promulgated by academics, international organizations, and policy think tanks, recent studies outline the patterns and effects of Chinese initiatives on regional trade and investment, the potential benefits and costs to BRI engagement for recipient countries, or the risks of corruption or environmental damage associated with megadeals (Rolland 2017, Hilton 2019, WWF 2017). These studies have made strides to improve public understanding of the character and form of Chinese deals with its partners, elucidating the priorities of Chinese actors and domestic partners, and highlighting various adverse effects of the initiative. 6 However, this is a new area of research, and many of the existing studies consider only a narrow angle of the projects or provide no empirical testing for their claims. Consequently, the literature is fragmented with multiple competing assertions as to the effects of BRI on host countries. This project seeks to contribute theoretically to the BRI literature by testing many of its claims. It also calls attention to BRI’s effects on ethnic politics by exploring the effects of Chinese investment activities in Malaysia, one of China’s main BRI partners. The ethnic aspect of BRI investments should not be overlooked, because issues of ethnicity are prevalent in the politics of many BRI host countries. This project also contributes methodologically to BRI research in two main ways. Scarce attention has been paid to the role of public opinion in relation to BRI investments and the aspects of investment projects that drive public perception of them. This is problematic because public opinion is a critical factor in assessing the effects of BRI investments for average citizens and in determining the future prospects of the BRI as a whole. Public support can place limits on the types of deals that leaders can negotiate with investors. In democracies, voter sentiment can directly influence the political fortunes of the political elites who manage BRI projects. Public opinion is a core component of BRI analysis, so I turn my focus to the voter, and provide a primary dataset that delves deeply into the factors that drive public opinion of projects. A second methodological contribution is in research design. Chapters 4 and 5 provide the project’s main quantitative analyses, and both chapters 7 use conjoint experiments. The advantages of using a conjoint experiment are that it allows for random assignment of experimental treatments for respondents, and therefore has greater ability to make causal claims as to which aspects of investment projects matter most to voters, and that it has the capacity to test multiple alternative hypotheses simultaneously. In sum, my analysis of the BRI in Malaysia presents an opportunity to bridge and contribute to three core literatures: the “second image reversed,” ethnic politics, and research focused specifically on the BRI. In a substantive sense, I contribute to the “second image reversed” literature by explicitly expanding its scope from previous focus on economic cleavages to incorporate ethnic identity, a political cleavage defined not by economic status but one usually defined by descent (Chandra 2006). The ethnic politics literature also can profit from directly considering international influences and examining how they can be manipulated into ethnic-based appeals and become a driver of domestic ethnic politics. And finally, I expand on the BRI literature by examining public opinion, a vital, yet overlooked component of BRI’s ultimate success, in depth. I also call explicit attention to the interactions between the BRI with local ethnic relations and politics as a means of ethnic political mobilization. Methodologically, the use of a conjoint design offers the ability to test multiple potential drivers of public opinion on projects simultaneously, allowing comparison of many disparate parts of a fragmented body of literature. Under certain conditions, discussed in 8 further detail below and in Chapter 2, international factors can be shown to impact domestic ethnic politics. The interaction of the international sphere with the study of local ethnic effects is an area that has been largely overlooked, yet can have significant effects on strategies of mobilization, voting, public opinion on the international issue, and by extension the issue’s future prospects for success. My findings suggest that there is a strong relationship between ethnicity in Malaysia and preferences for international investment projects. A final contribution of the study is its timeliness: the BRI is a massive foreign policy endeavor currently being undertaken by a global power that has potentially large ramifications for partner countries. The effects of the BRI, not only economically, but also politically, are not yet fully understood. The project’s focus on the BRI’s ability to become ethnicized ties in with issues of democracy and stability in host countries. The ethnicization of the BRI can contribute to antagonism between ethnic groups. In some cases, ethnic blame in reaction to these issues can lead to ethnic violence (Rodrik 1999). Research into the mechanisms and use of ethnicity to serve political purposes can help to anticipate and prevent such instances from occurring as a result of the BRI. What is BRI? Why focus on BRI? BRI is a massively ambitious endeavor in foreign affairs, with profound implications for China, its partners, and other stakeholders in regional politics and economics, such as the United States. The 9 benefits and consequences of BRI, and reaction to it, have been the subject of much speculation. However, systematic exploration of the topic is lacking. This study aims to enhance our understanding of how BRI can affect domestic politics in host countries, particularly along ethnic lines. In this study I refer to investments from mainland China after 2013 as BRI investments. Although there is no official declaration from the Chinese government labeling investment projects as part of the BRI, common usage has followed this practice. I focus on the BRI for two main reasons. The first is the scale of the Initiative and its importance to regional politics. The BRI is a cornerstone of modern Chinese foreign policy. First promoted in 2013 in policy speeches given by leader Xi Jinping, and formalized as part of the charter of the Chinese Communist Party in 2017, BRI represents a comprehensive plan to build and enhance economic, transportation, and communication linkages between China and partner nations. Chinese officials have touted it as a $1 trillion economic program to provide trade and infrastructure capacity between Asia, Europe, and Africa. Independent media reports have estimated that the BRI could in fact exceed $8 trillion in total spending (Hurley et al. 2018, 1). As of 2018, BRI programs were present in seventy-eight partner countries, home to over 4.4 billion people who produce over a third of global GDP (Hamzah 2018, 19). The BRI’s proponents have been eager to root the modern initiative in a historical past. Connections to the ancient Silk Road travelled by Marco Polo and to maritime routes used by legendary 10 Chinese Admiral Zheng He evoke feelings of China’s restoration as a global power (Lim 2016, Rolland 2017). The ambition of the BRI matches this lofty aspiration, as shown by its massive financial support and its sprawling geographic reach. The BRI has projects spanning six “economic corridors,” from Eastern Europe to Africa and across Asia. These corridors are divided into the “Silk Road Economic Belt,” which is comprised of land links between China, Central Asia, and Eastern Europe, and the “21st Century Maritime Silk Road,” which is primarily concerned with improving sea links between China, South Asia, and Southeast Asia. The geographical breadth of the BRI and its robust financial infrastructure and backing have led observers to label it as “the most ambitious example of global economic statecraft in the twenty-first century” (Alon et al. 2018). According to Chinese officials, the purpose of BRI is to improve connectivity between China and its partner nations via the construction of industrial and transportation infrastructure, financed through Chinese lending and investment. Critics of the BRI claim that its purposes are not rooted in benevolence, but that it serves China’s own economic and geopolitical interests (Stokes 2015). With problems of excess capital and industrial capacity at home, and with a desire to project greater power regionally and globally, the BRI serves as a vehicle to advance Chinese goals on multiple fronts. The scale of the Initiative and its growth in recent years is in large part a response to strong demand for investment in recipient countries. Developing countries in BRI regions require vast amounts of infrastructure investment to maintain economic growth and achieve 11 vital goals related to alleviating poverty while mitigating the effects of climate change. Estimates by the Asian Development Bank (ADB) call for over $26 trillion of investment over the 15-year period from 2016 to 2030 for Asian countries to meet these goals. In the Southeast Asian subregion alone (represented by membership in ASEAN), infrastructure investment needs over this same period are estimated to be between $2.8 and $3.1 trillion (ADB 2017). The investment needs for the ASEAN region’s densely populated and developing economies are vast, compelling economic actors and political leaders to pursue sources of capital to meet the demands of their economies. The second reason I examine BRI is that BRI satisfies an assumption of my theory that international issues must have a relevant ethnic component in order to affect domestic ethnic politics. The BRI has a strong attachment to China that allows it to be exploited by ethnic political entrepreneurs in Malaysia, who may connect the BRI’s influence to that of domestic ethnic Chinese populations. In this way the issue can be shifted to fit a host country’s ethnopolitical narrative and become a factor that encourages political competition along ethnic lines. Further elucidation of this process is given in Chapter 2. Intersection of Ethnicity and Foreign Investments in Electoral Politics Existing studies have identified a number of possible effects of large-scale foreign investment projects in BRI host countries. BRI projects, like other forms of FDI, offer economic benefits in capital- 12 scarce regions, such as growth, job creation, or technology transfer. Many recipients of BRI investments have a strong need for capital and technical expertise for development, and BRI can help to fill the gap by buying materials from local businesses or providing jobs to local workers (SERC 2017). These potential benefits form the basis of official justifications for the projects. On the other hand, BRI and other large-scale investment projects have been associated with a number of negative externalities as well. Chinese projects have been criticized for importing Chinese materials and labor, in effect bypassing the local markets and inhibiting wage growth and technology transfer (Todd and Slattery 2018; Yean 2018). In addition to this, several projects in Malaysia have been criticized for significant environmental damage, lack of economic viability, or serious allegations of corruption (Beech 2019, Wright and Hope 2019, Doig 2019). Increasingly, the position of the BRI as a major policy tool of the Chinese government has associated it with issues of national sovereignty in host nations. Mounting public debts to Chinese state- backed companies can pressure host countries into granting economic or geopolitical concessions in exchange for debt relief.1 This impact on state sovereignty has encouraged some opponents of BRI investments to caution against overreliance on Chinese interests. Malaysia, one of the main destinations for BRI funds due to its relatively advanced 1 In Sri Lanka, the government granted a 99-year year lease and rights to dock military vessels at a port built by a Chinese state-owned company when it was unable to service loans on the port project. 13 economy and strategic location, has witnessed mounting critiques against BRI investment projects, often tinged with nationalist or anti-imperialist rhetoric. For example, then-former Prime Minister Mahathir Mohamad (who was later to serve again as PM) accused a BRI- associated condominium project in a 2018 campaign speech of being “not Chinese investment but a settlement” (Beech 2018). Mahathir’s opponents responded by claiming that he was intentionally manipulating the investment into an anti-Chinese ethnic issue to curry favor with his ethnic Malay base. The ethnic framing of BRI projects by political entrepreneurs, which I refer to as the ethnicization of the issue, is possible due to the strong ethnic component of BRI. Since many projects are undertaken by mainland Chinese state-owned enterprises, and others are funded by Chinese state entities, there is space for political actors to turn public attention to the ethnic aspect of BRI. The BRI’s close affiliation with Chinese influence has the potential to entrench ethnic politics as the mode of political competition and increase ethnic tensions in countries in which ethnic Chinese wield some form of political power. I consider two main types of actors that contribute to the process of ethnicization of international investment projects. Candidates for political office frame investment projects in ethnic terms, and citizens choose to accept or reject the ethnic frame. Candidates are driven by rational motivators, such as selecting what they believe are the optimal campaign strategies to win office. Voters are motivated by a likely combination of rational factors, such 14 as personal gains flowing from having a coethnic in office, and emotional factors, such as utility gained from acting in solidarity with coethnics. This study is not designed to directly test these mechanisms, but rather to identify the effects of these processes as manifested in the correlation of individual ethnic identity and opinion of various types of investment projects. I expect this theory to operate under the constraints of two scope conditions. The first is that the host country must have existing ethnic cleavages that are politically relevant. Candidates will not frame investment projects in an ethnic light if they do not anticipate that voters will be receptive to such appeals. If ethnicity is not a viable strategy for political mobilization, candidates will pursue alternate means of framing investment projects, such as by emphasizing economic interests. However, if ethnicity is a viable mobilization strategy, candidates may opt to cast investment projects in terms of their ethnic effects, or how they pertain to specific ethnic groups. If a host country has an existing form of political competition that is centered around ethnic appeals, that makes the manipulation of investment projects into ethnic issues more likely. The second scope condition is that the investment projects must have an identifiable ethnic component that can be used by candidates and voters to create a credible ethnic story around the investments. If the projects can be readily tied to ethnic competition in a way that is plausible to voters, such ethnicized appeals are more likely to be an effective mobilizational tool. In the absence of a credible 15 ethnic component of the investment projects, political rhetoric regarding the projects is more likely to take the shape of non-ethnic appeals. The presence of both of these scope conditions in a foreign investment initiative and host country political context lead to my expectation that voters will be exposed to political appeals that paint the investment projects in an ethnic manner, and that they will be receptive to such appeals. An implication of this theory at work is that voters will exhibit various levels of preference for investment projects based on the ethnic identity that is associated with the projects. These preferences will be conditioned by the ethnic identity of both the individual voter as well as the ethnic association of the investment project. In general, I expect individuals to show greater preference for projects associated with their coethnics. Specifically, in regards to BRI projects (those associated with mainland Chinese investors), I expect that ethnic Malays will exhibit greater aversion to these projects than will Malaysian Chinese. Tests of these propositions are carried out in Chapters 4 and 5, and find support for my claims regarding respondent preferences for projects and respondent voting patterns. Why Malaysia? Malaysia is an ideal case to study the relationship between BRI investments and domestic ethnic politics for several reasons. The first is that it is a major destination for BRI funds. Malaysia is part of the Maritime Silk Road, one of the two main branches of the 16 BRI. Estimation of the total value of BRI projects in Malaysia is complicated by the fact that there is no official declaration of which projects fall under the BRI umbrella by either China or Malaysia (Grassi 2020). However, a conservative estimate of the total value of various BRI megaprojects (each valued at over $1 billion) puts Chinese investments from 2014-2020 at over $150 billion. These projects are concentrated in infrastructure construction and span numerous sectors, from rail and highway construction, power generation, gas and petrochemical pipelines, industrial parks, port facilities, and real estate. Some of these projects have been the subject of controversy. The East Coast Rail Link, designed to connect the east and west coasts of the Malayan peninsula by rail, has been criticized for being overbudget. The Melaka Gateway port complex has been labeled a white elephant, lacking demand for its services to justify its ambitious scope and for environmental damage associated with the construction of artificial islands. And the Forest City real estate complex has been accused of pricing out Malaysians and marketing itself to mainland Chinese buyers. The second reason for selecting Malaysia is that it is multiethnic and has a political context where ethnic salience is high, thereby satisfying one of the scope conditions of my theory. The ethnic composition of Malaysia falls along three main groups. The largest of these is the Bumiputera ethnic group, approximately 65% of the total population, which is a collection of “indigenous” groups (Bumiputera translates roughly to “sons of the soil”). The largest constituent group among the Bumiputera are the ethnic Malays, but it 17 also includes multiple additional groups living in both the Malayan peninsula and the East Malaysian states located on the island of Borneo. While the Bumiputera group is highly diverse culturally, linguistically, and religiously, it is bound by political interests in competition against the other main ethnic groups. The reasons behind the formation of the Bumiputera identity are discussed further in Chapter 3. The two other significant ethnic groups are the Malaysian Chinese and the Malaysian Indian groups. Malaysian Chinese today make up about 25% of the population, and Malaysian Indians a further 8%. While there is great diversity within these groups as well, they are commonly grouped together in political matters. The size of the Bumiputera and Malaysian Chinese groups has contributed to a system of ethnic competition in Malaysia that is often contested along the Bumiputera-Malaysian Chinese divide. This study is primarily concerned with the dynamic between these two largest ethnic groups. Political competition in Malaysia is highly ethnicized along these lines (Holst 2012). Ethnic politics are institutionalized in articles of the Constitution and in legislation that grants special political and economic rights to individuals and groups based on their ethnic identity. The party system is dominated by political parties that either explicitly advocate for ethnic interests or are closely associated with particular ethnic groups. These conditions make Malaysia a fertile ground for the ethnicization of international investment projects such as those that fall under the BRI umbrella. 18 Research Design and Data My theoretical framework and data collection strategies profited from five months spent in the field in Kuala Lumpur, Malaysia.2 During this time, I conducted interviews with Malaysian politicians and researchers to better understand the local political context and to guide the parameters of my theory. I also explored the feasibility of launching an online survey, recruited a translation service so that the survey could be offered in both English and Bahasa Malaysia, and drafted questionnaire items with the insights offered by my interview and other local contacts. Sampling This study is mainly focused on host country public opinion on investment projects and Chinese influence. The primary tests are conducted using a survey questionnaire and two unique conjoint experiments, with supplementary data gathered in interviews conducted by the author. Data for the survey as well as the two conjoint experiments were collected from an online sample3 of 1,308 Malaysian voting-age citizens. Due to the nonrandom nature of sampling, I took several steps to mitigate sources of sampling bias. To enhance the representativeness of the sample to mirror Malaysian census figures as 2 The first two months were January-February 2020. With the onset of the Covid-19 pandemic, my fieldwork was suspended until a second period in- country from January-March 2022. The pandemic also contributed to a shift in my research strategy to rely less on interview data, and more on survey data. 3 Sampling was carried out by Qualtrics using locally based partners with my consultation and supervision. Subject were recruited by email or online advertisements, and accessed the survey and experiments by using the Qualtrics portal. 19 closely as possible, the sample was pre-stratified in an attempt to match respondents to the underlying Malaysian population based on gender, urbanization, and ethnic identification. This was accomplished by implementing maximum quotas for respondents based on these three criteria and using screening questions at the beginning of the survey. After specific quotas were filled, additional respondents in those same categories were blocked from participating. A comparison of selected descriptive statistics for the sample and the actual Malaysian population can be seen in Table 1.1. For the purposes of this study, ethnic Malay and Bumiputera investors are grouped together, mirroring official Malaysian policies that often treat the groups as one. This decision is consistent with that of other researchers who are concerned about differences between Bumiputera and other ethnic groups, and not within the Bumiputera groups themselves, while enabling a simplified interpretation of my results. Table 1.1: Comparison of Descriptive Statistics Between Sample and Malaysian Population Sample Malaysia4 Ethnicity Bumiputera 831 (64%) 69% Malaysian Chinese 340 (26%) 23% Malaysian Indian 99 (8%) 7% Gender Male 674 (52%) 51% Female 634 (48%) 49% Urbanization Urban 1,035 (79%) 75% Rural 270 (21%) 25% Age (median) 30 30 Monthly Income RM 3,000-6,000 RM 5,873 (median) ($715-$1,430) ($1,400) 4 Source: Malaysia Department of Statistics, 2020. Income data are for 2019. 20 Descriptive statistics show that my sample is materially similar to figures for the full population. Respondents who identified as Malay or Bumiputera made up 64% of the sample, Malaysian Chinese were 26%, and Malaysian Indians made up 8%. Compared to census figures, my sample slightly underrepresents Bumiputera respondents and slightly overrepresents Malaysian Chinese respondents. In regards to gender, the sample skews slightly male, with 52% male respondents compared to 51% in the general population. The sample is also more urban than the general population. 79% of respondents reported living in an urban area and 21% in rural areas, while the Malaysian census reports that in 2020 its population was 75% urban and 25% rural. Although my sample is not fully random, it closely resembles national figures across these three criteria. Still, the fact that this is an online convenience sample may be cause for concern that the sample respondents differ from the external population in systematic ways. This can be a problem if treatment effects are heterogeneous (Berinsky, Huber, and Lenz 2012). I do hypothesize that treatment effects are conditional on respondent ethnicity. By ensuring that the ethnic composition of my sample closely resembles that of the general population, the risk to external validity for pooled analyses is mitigated. This is also a hedge against bias induced by possible heterogeneous effects based on respondent gender and urban status. In addition to this, respondents may differ from the general population in other ways. For example, internet users may be wealthier or younger than those without internet 21 access. Internet coverage is widespread in Malaysia. As of 2021, the Malaysian Department of Statistics estimated that 96.8% of Malaysian citizens over 15 years of age used the internet. A look at the sample statistics shows that the median age and monthly income of the sample is similar to that of the general population. Median age for both the sample and population is 30, while the reported income range for the median respondent includes the figure given by official Malaysian sources for country-level median income. Of course, there could be additional sources of heterogeneous treatment effects not listed or examined here that could impact the external validity of the sample to the Malaysian population. But across these major criteria, the sample closely resembles the population. A second concern to external validity arising from the use of an online sample is whether respondents behave differently when taking a survey online. Some respondents may be serial respondents who habitually participate in online surveys, and their extensive participation in surveys can contaminate their responses. Berinsky, Huber, and Lenz (2012) examine whether chronic study participants give substantively different responses than respondents who had only participated in one study, and find no difference between the groups. In addition, the authors successfully replicated multiple studies that had used random sampling with their online samples. While the authors looked specifically at MTurk, which is most commonly used to find online samples in the United States, there is little reason to expect that online respondents in Malaysia behave in significantly different ways. Coppock and McClellan (2019) find similar results to Berinsky 22 et al. (2012) with samples drawn from the Lucid platform. This evidence suggests that online samples, like the sample collected for this study, can yield reliable and minimally biased results. Precautions have also been taken against risks to internal validity. First, recruitment for the study prevented respondents from taking the survey more than once. Access to the survey required a unique email, phone number, and IP address, which established a high barrier to repeat responders. Second, several checks were implemented during the survey to protect against low-quality responses. Respondents who sped too quickly or who took too long to complete portions of the survey were rejected. Respondents also were not given detailed information about my hypotheses prior to their participation in order to minimize the risk of survey satisficing. Overall, while online convenience samples are more prone to bias than random sampling, they can produce samples that closely resemble the population they are drawn from across important criteria, often to a higher degree than other convenience samples (such as samples drawn from students or passersby). They have also been shown to yield similar results for major findings in political science as have other widely accepted sampling methods (Berinsky et al. 2012). Modeling In recent decades, political scientists have extended the logic of conjoint designs from their origins in marketing to the decision- making process that individuals make in social and political life. Citizens routinely make political decisions and judgments that can be 23 similarly modeled to those of consumer choice. Political opinion is multidimensional, dependent on numerous factors acting simultaneously. Conjoint analysis in political science has gained traction and acceptance over the past decade, reflected by the method’s increased publication in notable political science journals (Hainmueller et al. 2013). The application of conjoint design to political science extends to multiple areas. These designs have been used to explore major questions in the field such as party selection of candidates (Doherty et al. 2019), the classification of violent events as terrorism (Huff and Kertzer 2017), and voter sanctioning of electoral violence (Gutierrez-Romero and LeBas 2020), to name a few. The application of conjoint experimental designs to political science offers a promising and fruitful tool for researchers to examine a wide range of political evaluations and decisions. Conjoint experiments offer a number of analytical benefits over alternate methods of assessing public opinion. Random assignment of treatments (attribute levels) strengthens the study’s ability to make causal claims about the attributes’ effects. Additionally, the simultaneous randomization of all attributes allows for the effects of each attribute to be assessed relative to one another. This is particularly useful regarding BRI investment projects because the projects vary along multiple dimensions. And finally, conjoint designs allow respondents to assign their choice between investment profiles along multiple paths, thereby mitigating the risks of social desirability bias (Hainmueller et al. 2013). 24 For the analysis in Chapter 4, all respondents were asked to complete a choice-based conjoint task in which they specified whether they preferred to have a hypothetical investment project in Malaysia. Each of these hypothetical choices was fully randomized according to seven attributes that commonly differentiate investment projects from one another and are argued to affect voter sentiments. These attributes include the investment type, location, associated benefits of the project, its associated costs, the type of investor, investment size, and the ethnic/national origin of the investor.5 The investor’s origin is the attribute of primary interest, as it represents the effect of ethnicity on respondent preference. It identifies the investor as a member of one of five groups: Malay, Chinese Malaysian, Mainland Chinese, Western, or Japanese. Two of these levels are domestic, while the other three represent foreign investors of different origins, allowing a comparison of both foreign and domestic actors on citizen sentiment. The analysis in Chapter 5 employs a separate conjoint experiment and dependent variable. In this design, respondents were asked to choose between two candidates who had been randomized according to eight attributes. This turns the focus of the study to examine the degree to which investment projects matter in vote choice, while Chapter 4 focuses on which features of investment projects matter most to voters. The primary attribute of interest is the candidate’s position on foreign investments, and offers variations on whether the 5 A full list of the attributes and attribute levels included in this study can be found in Chapter 4. 25 candidate supports increased investments from China, the United States, Japan, or is opposed to foreign investments. It is analyzed alongside seven additional attributes that are commonly thought to affect voter preference for candidates or were major issues in Malaysia’s 2018 general election: the candidate’s partisan affiliation, history of corruption, position on taxes, candidate ethnicity, profession, age, and gender.6 Plan of the dissertation The following chapters describe my argument in greater detail and test its theoretical implications. Chapter 2 situates the study into relevant bodies of literature that provide the foundations for the causal pathways and mechanisms that can lead to ethnicization of an international economic issue. It also discusses the flaws and omissions of these research areas and the techniques and considerations I employ to attempt to improve upon them. Finally, it details the two scope conditions that I argue lead to a high likelihood of the ethnicization of an international issue in domestic politics. The core of the empirical analysis is contained in Chapters 3, 4, and 5. The empirical strategy features a mix of qualitative and quantitative techniques using data collected from primary and secondary sources. Chapter 3 provides evidence of the two scope conditions in the BRI-Malaysia context by tracing the development of 6 The attributes and attribute levels for this study can be found in Chapter 5. 26 modern-day ethnic politics in Malaysia and the relevant ethnic aspects of BRI. It uses survey data collected for the purpose of this study to support my claims regarding the Malaysian political system and the BRI. As mentioned above, Chapters 4 and 5 turn to two separate conjoint experimental designs to test the observable implications of the theory. Chapter 4 asks respondents to consider multiple randomly generated investment project profiles to determine which aspects of investment projects matter most to them, and how. These projects vary the origin of the project investor by ethnic group and nationality to measure the direction and magnitude of ethnic affiliations in respondent preference. It also allows for a comparison of the many alternate effects of projects in the BRI literature. Chapter 5 changes the focus from investment projects to candidates, and asks respondents to choose between hypothetical candidates whose stances on various types of investment projects vary. In this way, I examine the degree that candidate investment stance affects vote choice. Chapter 6 concludes the study and suggests areas of future research in light of my findings. 27 Chapter 2: Theory Scholars have long understood that international and domestic politics are intertwined. A country’s laws and freedoms within its borders are driven in a number of ways by international forces. For example, at Malaysia’s independence in 1957, its security institutions were formulated in light of a homegrown communist movement with financial and ideological ties to China. Strong restrictions on popular expression and mobilization were enshrined in the country’s constitution and legislation to counter the perceived communist threat. Even political competition was consciously structured around ethnic identity politics instead of ideological divides, at least partly in reaction to international influences (Holst 2012). Numerous examples in other political contexts mirror the Malaysian reaction to the international communist movement. The arrow of causation is reversed as well. Domestic forces also affect international relations, as political and economic structures can encourage or discourage the transmission of people, ideas, or finances across borders. Market-based economies, for example, are driven to pursue higher returns on investment that are often found abroad. Deregulation of finance and capital markets has enabled firms to establish themselves internationally in pursuit of profit. The result has been an increase in trade and financial ties in a more globalized world. Even as political scientists acknowledge the dense linkages between the domestic and international arenas, in practice, these ties are often overlooked in academic studies. Comparativists sometimes 28 focus too narrowly on the domestic context of a few case studies and ignore or downplay international influences, while international relations scholars sometimes paint states with too broad a brush and treat them as cleanly comparable units with an idealized set of motivations and capabilities. The perspective of structural realism in international systems goes as far as treating the state as a “black box,” in effect denying agency of domestic forces to impact state actions in the international arena (Waltz 1979). To be sure, these simplified models have proven fruitful in formulating testable theories and in advancing our knowledge in each field. However, in doing so they often overlook the effects of political actors outside of the artificially truncated scope of their models. Expanding the field of view can enrich our understanding of political linkages. For instance, liberal and constructivist scholars in IR have identified the tangible impact of international actors who operate in the domestic realm or as part of non-governmental organizations (Keck and Sikkink 1998, Kelley 2004). Certain fields of comparative politics, such as ethnic politics, can similarly benefit from an expansion of the causal factors under consideration to include international forces. This study seeks to bridge the literatures on the “second image reversed” to the ethnic politics literature, operating from the perspective that these two major bodies of political science literature rarely speak to one another, even though the linkages between them are profound. I provide evidence that international forces, namely economic forces flowing from foreign investment, can 29 affect host country politics along ethnic lines under the right conditions. There are two scope conditions that enable a foreign investment project to be ethnicized and to have this type of effect: first, the host country must have pre-existing politically salient ethnic cleavages. Second, the investment projects must have a credible connection to these existing ethnic cleavages. If both of these conditions are satisfied, citizens and political elites of the host country are more likely to link the projects to existing domestic ethnic biases. The following sections explore this basic premise in greater depth. The experience of the BRI in Malaysia serves as the case study of this theory at work. Malaysian politics are heavily influenced by issues of ethnicity, and BRI investments have increased the presence and visibility of China in Malaysia. Since independence, questions of the relative influence of ethnic groups in the government and economy, particularly between the ethnic Bumiputera and Malaysian Chinese, and of affirmative action policies constructed to benefit Bumiputeras in these areas, have dominated the political landscape (Lemiere 2018). As an initiative backed by the Chinese government, some politicians have drawn connections between the BRI and the Malaysian Chinese diaspora, regardless of whether the connections made are a fair representation of the facts. Media and academic reports on the implementation and consequences of greater Chinese investment have increased since the Initiative’s arrival, and political elites have increasingly used the BRI as an issue to mobilize and countermobilize support (Beech 2018, Wright and Hope 2019). These developments create 30 a prime situation for the examination of this dissertation’s main concerns: how and why international issues can be transformed into domestic ethnic issues, and the roles played by political actors and average citizens. It is also a good context for discerning the contours of how ethnicity affects how individuals think about international investments and the extent to which it affects their vote. The rest of this chapter explores the aspects of the literatures on ethnic politics and the second image reversed that are relevant to the case at hand. It then connects the two literatures to present a theory of how an international economic program such as the BRI can be manipulated into an ethnic political issue, and why such ethnic appeals are attractive to candidates and voters. Finally, I discuss the conditions of international issues and of political systems that make the occurrence of the ethnicization of programs like the BRI more likely, and I lay out several empirical expectations of my theory. Ethnic Salience in Host Countries This study borrows heavily from the literature on ethnic politics. Ethnicity is an important dimension of political competition across much of the developing world because it can provide insight into how people form associations and are motivated to various political behaviors. Ethnic considerations are impactful in many areas of politics, from extreme cases of ethnic violence (Horowitz 1985, Fearon 1998), to more routine aspects of elections (Posner 2005), collective action (Habyarimana et al. 2007), and governance. 31 Party politics, wealth distribution, public goods provision (Alesina et al. 1999), and voting behavior are all areas in which ethnicity has been shown to have an effect (Kalin and Sambanis 2018). There are several major explanations for the role of ethnicity in political systems. One branch of thought argues that ethnic identification offers individuals psychological benefits that they gain from belonging to a group (Horowitz 1985). Individuals gain “expressive” utility by situating themselves as part of a larger ethnic identity and from the sense of belonging such membership entails. Through a sense of shared culture, allegiance to a group, historical grievances against other ethnic groups, or emotional benefits from a feeling of solidarity, individuals can be motivated to engage in collective political action. However, expressive theories have been criticized as offering a rather undertheorized and static conception of ethnic identity and are not well positioned to explain variations in the political salience of ethnicity across time and space (Fearon 2006). A second branch of thought attempts to explain spatial and temporal variation of ethnic salience by emphasizing the instrumental nature of ethnicity. In this conceptualization, voters are rational actors who are driven primarily by concerns for their own material interest (Bates 1974, Fearon and Laitin 1996, Chandra 2006). Ethnic identities are important primarily because they are useful to voters and elites. They are more influential on political behavior in contexts where voters anticipate that electing coethnics will lead to future material benefits (Bates 1974, Posner 2005). Although 32 ethnicity is just one of many possible dimensions around which voters can be mobilized, it is a potentially potent form of political coalition formation. Ethnic groups frequently have a shared language and culture, easing group formation (Bates 1983), and the ascriptive and descent-based nature of ethnicity is useful in allocating resources to members while excluding outgroups (Fearon 1999, Chandra 2004). These contexts are fluid and situational, and as a result the salience of ethnicity is variable over time and space. Operating from this perspective, instrumentalist scholars often focus on identifying the conditions that impact ethnic salience. In elucidating the process through which international investments are ethnicized, this study profits from the perspectives offered by both the rationalist and expressive accounts of the politicization of ethnicity. While the rationalist perspective establishes the socially constructed and instrumental aspects of ethnicity in political behavior, there is still good reason to believe that expressive motivations can perform a parallel function (Kalin and Sambanis 2018). Expressive utility gained from giving one’s allegiance to a group or conforming to social norms can rival or mediate the effects of material payoffs (Akerlof and Kranton 2000; Bassi et al. 2011). In some cases, individuals forego material benefit to enhance group esteem (Shayo 2009). As such, it is likely that ethnicity affects individuals along both dimensions to various degrees simultaneously. As Kalin and Sambanis (2018) note, “identities shape interests, interests shape identities, and, together, they both shape actions” (p. 250). This study is agnostic 33 as to whether instrumentalist or expressive effects prevail and is not designed to adjudicate that debate in the case at hand. Using the tools provided by both the instrumental and expressive camps, it intends to identify and explain the effects of a factor that has largely been overlooked in studies of the determinants of ethnic salience: international investments. While these perspectives offer explanations for why ethnic salience varies over time and space, a consistent feature of these theories is that they deal with explanatory factors that operate over a long time horizon. Even rationalist perspectives, while acknowledging the situational nature of ethnic salience, often rely on explanatory factors that change slowly over time. For example, aspects of ethnicity such as its ease of identification for membership, or institutions that encourage the politicization of ethnicity, are often sticky, in the sense that they are often stable over time. Yet even in stable polities that are accustomed to ethnic political competition, the salience of ethnicity varies. At certain points in time, or in certain elections, ethnic issues can flare up, while at other points the role of ethnicity can seem to temporarily wane. Likely driving this short-term variation are timely political issues that are not always ethnic in nature. Some of these issues can readily fit into a country’s ethnic narrative, but others cannot. The presence of issues that can be readily ethnicized can impact the short-term importance of ethnic issues in a political system. It is in this space that international issues, such as the BRI, can become 34 instruments of ethnic politics and drivers of ethnicity’s short-term salience. International Economic Affairs and Domestic Politics This study is concerned with international investment projects and international investors, and speaks to an established literature that examines the effects of international economic factors on domestic politics known as the second image reversed. As Gourevitch (1978) has outlined, international forces can have a considerable, and often underappreciated, impact on domestic politics. In the extreme case, international military interventions fundamentally alter the policy options for domestic leaders.7 However, more mundane policies in the security realm or in the international economy also can alter the capabilities or incentives of political actors in the domestic arena. In the decades since these authors called for greater attention to the interaction of the international and domestic spheres, globalization of the world economy has only increased the flows of finance and information across borders (Milner and Keohane 1996). Internationalization of the economy can have a strong effect on a country’s domestic policies as well as on its citizens’ policy preferences (Milner and Keohane 1996). Opening up the economy to higher levels of trade and international capital bears disproportionate benefits and costs to domestic actors. Determining 7 The recent example of Russia’s military invasion of Ukraine exhibits the policy constraints currently facing the Ukrainian government. 35 who wins and who loses in this process depends on several factors. Scholarly work in this area primarily has identified how internationalization of the economy realigns domestic actors’ incentives and the coalitions that stand to benefit from and hence advocate for further openness. The level of engagement with the international economy and structural features of the domestic economy, such as the country’s comparative advantage and factors of production, determines which societal groups benefit most from opening the economy (Solingen 2001). In turn, these accounts argue, they shape the nature of political competition. However, the effects of the international economy on domestic policies and policy preferences are not uniform across countries. They are conditioned in large part by host country institutions, such as electoral rules, labor unions, and political parties, which play a key role in determining the types of political coalitions that form and the types of appeals candidates make to voters (Garrett and Lange 1995, Hallerberg and Basinger 1998). Institutions condition the effect of the international economy on domestic preferences through several mechanisms: first, by obscuring price signals, and thereby actors’ interests; second, by providing stability to existing political coalitions that resist change; and third, by incentivizing the type of response political leaders have to economic change (Milner and Keohane 1996). While the majority of literature on economic internationalization examines the effects on political coalitions formed around economic cleavages, relatively little attention has been paid to its effects on 36 political groups that are not necessarily economic in nature. I expand the framework of the second image reversed literature to incorporate the conditions under which international factors impact ethnic identity at home. The institutional arguments already advanced by the second image reversed literature can inform how this process occurs. A country’s institutional makeup constrains the actions of candidates and voters, and channels them toward strategies that are successful under existing institutions. For example, ethnic party systems feature parties that are constructed on and excel at competition along ethnic lines. Party platforms for these parties are often not built on a broad programmatic basis, and coalitions may be hard to hold together if the platform is extended. Individual candidates are incentivized to campaign on established ethnic cleavages, and going outside of the playbook can be costly. The electorate can also be affected by the institutional landscape and be primed and conditioned by years of ethnic competition to perceive political issues along ethnic lines. In these ways, international issues can become ethnicized, that is, transformed to fit a host country’s ethnic narrative. The following section delves deeper into this process, expanding the scope of both the ethnic politics literature and the second image reversed and linking them to one another. Tying Ethnic Politics and Foreign Investments Together Why are international investment projects able to be ethnicized? Why are candidates able to turn these projects into ethnic issues, and 37 why do voters accept these types of appeals for issues that may not immediately appear ethnic in nature? The accounts of ethnicity and internationalization of the economy discussed above inform my argument of when an international economic issue is likely to be ethnicized in the domestic arena. I now turn to a discussion of the process through which international economic issues can be linked to domestic ethnic politics. A mechanism that rationalist scholars have proposed through which ethnicity affects political behavior is its ability to provide informational “shortcuts” to voters (Chandra 2004; Posner 2005; Ferree 2006). Ethnic identity is a simplifying mechanism that provides information on the intentions and capabilities of political actors to voters, and can reduce uncertainty as voters navigate complex political waters. Voters often have poor access to information in political settings where parties are not well established or do not offer thorough or stable policy platforms, or where sources for political information such as the media or civil society are not independent or are insufficient. When voters have limited access to political information, they turn to alternative methods to infer political information, including the use of group cues as heuristics (Brady and Sniderman 1985, McDermott 1998). Ethnic identity of candidates is a readily available and low-cost source of information, as it is often public knowledge or can be inferred from the candidate’s name or physical appearance. Voters use ethnicity to infer the preferences, capabilities, and strategic electoral viability of candidates if they believe that 38 members of the same ethnic group tend to share political preferences and beliefs (Chandra 2004, Birnir 2006). Experimental evidence has shown that voters are more likely to use ethnicity as a voting cue in information-poor environments (Conroy-Krutz 2012). Most studies that look at the role of information in driving voters toward the use of heuristics do so by examining how voters use these cues to evaluate candidates. This study innovates on this design by applying it to voter evaluations of a political issue. I argue that the process that voters use to judge candidates from their ethnic identity can be readily applied to how they judge major political issues that can credibly be assigned an ethnic component, such as investment projects. As they do with candidates, when voters evaluate investment projects, they operate from a position of imperfect information, and they must weigh a complex set of potential benefits and costs. The complexity of investment projects means that individuals are unlikely to know critical details of the implementation of the project. Much of the information regarding the projects is not public, such as the sourcing of labor and materials, the steps taken to mitigate environmental damage, or even the bidding and payment processes (Rolland 2017). Even for information that is public, gathering information on projects is a costly task that requires time and attention that voters may be unwilling to expend. It is reasonable to expect individuals to resort to cognitive shortcuts in the evaluation of major political issues like investment projects in a similar fashion as they do political candidates. 39 Since information on projects can be hard to obtain, the few sources that do supply information on the projects become increasingly impactful on citizens’ evaluations. Political leaders are one such source of information. Politicians can have a variety of goals, from advancing ideological policies to controlling the spoils of power. To make progress on their goals, they must first win and retain political office. To obtain the support of a winning coalition, politicians appeal to their electoral base and demonstrate or promise to deliver benefits to that base. However, politicians have latitude in how they respond to changes brought on by international capital flows. They may choose to pursue policies that deviate from the aggregate good or from economic efficiency. Brader (2005) finds that politicians can effectively cue both positive and negative emotional responses that subvert rational decision making. One way that they can do so is by couching their rhetoric in nationalist terms, which can successfully divert public attention from material self-interest toward a rally- around-the-flag response (Feinstein 2016). Although Feinstein directly examines nationalism, a parallel logic can be applied to ethnic politics contested between domestic groups. deFiguerido and Weingast (1999) formulate a model in which rational politicians mobilize ethnic support by manipulating information asymmetries to instill fear among their coethnics, and then use that fear to mobilize support and resources. As noted above, politicians can effectively elicit either positive or negative emotional responses among voters (Brader 2005). In an ethnic polity, politicians are driven to reward their ethnic base of support and to 40 shore up support based on ethnic appeals. To do so, they may be incentivized to ethnicize issues by calling attention to the issue’s differential ethnic effects. Making ethnic appeals can be an attractive strategy for candidates because it allows them to mobilize support based on powerful concepts of the defense of ethnic identity, thereby capitalizing on often emotional and duty-bound feelings of sacrifice for the group against outsiders (Fearon 2006). In doing so, it can allow candidates to bypass or downplay appeals based on the net material benefit of investment projects to voters, which in some situations may be a more opaque, complicated, and less effective type of appeal. The masking or diversionary effect of ethnic appeals can also mean that the ethnic connection made by politicians does not necessarily have to be grounded in truth in order for it to be effective. Political issues with high informational barriers are susceptible to exaggeration or misinformation. These incentives of voters and candidates in response to the problem of incomplete information can be applied to the issue of foreign investment projects. While the average voter may not be aware of the characteristics of investment projects, politicians are likely to have greater access to this information, and can strategically divulge information to voters that benefit their cause. Campaign rhetoric can be specifically designed to prime voters according to their ethnic identities as a means of securing ethnic support, encouraging voters to prioritize social identity confirmation even if it may go against material interest (Dickson and Scheve 2006). Political elites may find it in their interest to call attention to 41 the ethnic component of investment projects to cast the projects as for or against the interests of their ethnic group, thereby leveraging the power of ethnic identity to mobilize support along that dimension. Even if candidates view investment projects as potential issues around to mobilize ethnic followers, their audience must go along with the linking of the projects to ethnic interest. Voters are likely to be receptive to appeals that ethnicize investment projects for either rational or emotional reasons. Following instrumentalist reasoning, they can accept arguments from political elites that the investments will materially advantage or disadvantage members of their ethnic group, and perceive it as in their individual and group interest to vote with their ethnic bloc. Alternatively, they can be influenced by ethnic appeals based on expressive utility derived from norms of solidarity with their ethnic group. Bassi, Morton, and Williams (2011) find that in contexts with low information, individuals are willing to support members of their own identity groups even when it does not lead to their material benefit. This suggests that even if individuals think they may benefit from investment projects, they may still be open to voting against projects if they prioritize allegiance to their ethnic group. If politicians decide to pursue a strategy of mobilization based on ethnic identity, they are likely to find a receptive audience if they can credibly tie international issues to the existing structure of ethnic competition. The political ethnicization of investment projects is more likely to occur with the presence of two scope conditions. The first condition is that there must first be an existing political ethnic 42 cleavage in the host country. Otherwise, citizens of the host country will interpret the issue along another dimension, such as class or partisan lines. Instrumentalist accounts highlight that ethnicity is most salient under conditions of relatively low information and when ethnicity stands as a viable pathway to winning political power. In these situations, voters and political elites should choose to align themselves on ethnic lines only if they anticipate that other voters will as well. In the absence of a realistic possibility of a political coalition organized around ethnicity that has a chance of winning political power, citizens will interpret international issues along alternative dimensions. The Malaysian case satisfies this condition. Malaysian politics are commonly understood through the lens of ethnic competition between the majority Bumiputera ethnic group and various minority groups. Of the ethnic groups that make up the minority, the Malaysian Chinese are the next largest group. Many facets of the political system are impacted by ethnic considerations. For example, many political parties, including the most successful ones, are ethnic parties organized around promoting ethnic interests. Several major policies in education, religion, and the economy are designed to advance or protect the rights of ethnic groups. In the economy, although Malaysian Chinese are a minority group, they control a disproportionately large percentage of the economy, which can be a cause of ethnic tension (Khalid and Li 2019). The Malaysian Chinese community, as a diaspora, exists as full citizens in Malaysia, while maintaining culturual, linguistic, and 43 familial ties to its country of origin, China. This creates a three- part relationship between the Malaysian government, the Chinese government, and the Malaysian Chinese community. At points in Malaysian history, the connections between mainland China and the Malaysian Chinese have been exploited by Malay leaders to question the allegiance of the Malaysian Chinese to Malaysia (Holst 2012, Sin 2015). Although this study examines the rationale behind such accusations and the downstream effects of this sentiment, it is not my position that Malaysian Chinese citizens, or their Bumiputera counterparts, are monolithic or bent upon ethnic conflict. Successive generations of Malaysian Chinese, like most Chinese diaspora groups in Southeast Asia, have crafted a self-image that is distinct from the mainland Chinese and fully Malaysian (Tan 1997, Gomez 2006). While cultural ties may give Malaysian Chinese an affinity for mainland China as a place of origin, they should not be assumed to mean that Malaysian Chinese act as agents of mainland Chinese interests. However, modern-day Malaysia still sometimes sees accusations against Malaysian Chinese citizens, usually from the majority Bumiputera group, that insinuate such a relationship. Even though the diversity of Malaysian Chinese interests is well established, the Bumiputera “generally view them as a cohesive or homogenous group” (Gomez 2006). As discussed in this chapter and in Chapter 3, the persistence of such claims of Malaysian Chinese disloyalty can be partially explained by the self-interest of candidates seeking office or policies of the state (Gomez 2006, Noh 2014). These types of claims can be effective in mobilizing ethnic support in their target 44 audiences by manufacturing a feeling of threat from outsider groups. Further justification for the extent that ethnicity guides Malaysian politics is provided in Chapter 3. Given that a host country already has politically salient ethnic cleavages, the second scope condition for an international issue to impact domestic ethnic politics is that the issue must be able to be credibly tied to the host country’s existing ethnic narrative. In the absence of this characteristic, while investments may still become a political issue, it is unlikely that they will be ethnicized. An investment made by an American investor in Indonesia, for example, lacks a credible ethnic connection to the Indonesian political landscape. While such investments may become politicized in other ways, rhetoric surrounding the issue is unlikely to be couched in ethnic terms.8 But if an international investment has features that can be credibly tied to the host country’s ethnic competition, then it is more likely that the investment becomes ethnicized, and rhetoric surrounding the project takes on an ethnic tone. In these cases, ethnic appeals are more likely to be persuasive to voters, thereby enhancing the value of an ethnic approach to strategic candidates. An investment program such as the BRI fits this bill in certain contexts. The BRI is a Chinese initiative, backed by Chinese funding, executed by Chinese state-backed and private firms, and a cornerstone of Chinese foreign policy. This close affiliation with Chinese 8 An example can be seen in the Grasberg gold mine in West Papua, Indonesia. The mine is run by an American mining company and has been the subject of controversy regarding its environmental impact. However, criticism surrounding the mine and its practices have not been ethnic in nature. 45 influence has the potential to increase ethnic tensions in countries in which ethnic Chinese wield some form of political power. Multiple BRI host countries are home to a Chinese diaspora population with a distinct identity and ability to form viable political coalitions. While these diaspora communities are established citizens of their home countries and often stress their independence from and unique identity in contrast to mainland China, it has been a well-trodden strategy of their ethnic opponents to paint these communities with various degrees of enduring connection to mainland China, ranging from accusations of sympathies to China to allegations of acting as proxies to mainland Chinese interests (Holst 2012). The immense mainland Chinese presence and influence that the BRI brings present an opportunity for ethnic political entrepreneurs to rehash these old charges as a way of shoring up support among their own ethnic group. Since ethnic politics in countries like Malaysia are commonly contested between the Bumiputera and Malaysian Chinese groups, the Chinese association with the BRI lends itself to rhetoric that stirs up feelings of ethnic competition that voters are already accustomed to and that have a track record of past success. For example, former Prime Minister Mahathir Mohamad (who was later to serve again as PM) accused a BRI-associated condominium project in a 2018 campaign speech of being “not Chinese investment but a settlement” (Beech 2018). If both of these scope conditions are satisfied, that is, a country has pre-existing ethnic political salience and the international issue is relevant to the country’s ethnic cleavages, this can entice political elites of the host country to present the 46 international issue in an ethnic light, and citizens to process it as such. The theoretical framework can be diagrammed with a two-by-two table that sets out the expectations on the ethnicization of the issue based on configurations of domestic ethnic politics and the degree to which the international issue can fit the country’s ethnic landscape. A visual representation of this is presented in Table 2.1. Table 2.1: Factors Affecting the Likelihood of Ethnicization of an International Issue Low Ethnic Salience High Ethnic Salience in Country in Country Int’l Issue Lacks No ethnicization of Low likelihood of Relevant Ethnic issue ethnicization of Component issue Zambia - BRI Int’l Issue Has Low likelihood of High Likelihood of Relevant Ethnic ethnicization of Ethnicization of Component issue issue Singapore - BRI Malaysia - BRI The first condition of a country’s pre-existing ethnic salience is represented by the values in the columns. There are two values in this simplified representation: low ethnic salience in-country and high ethnic salience. The second condition, the relevance of the international issue to the country’s ethnic politics, is represented in the rows. The two levels of this dimension are, first, if the international issue lacks an ethnic component relevant to the host country’s ethnic competition and second, if the international issue has a relevant ethnic component to the host country’s ethnic politics. 47 This creates four scenarios for the likelihood of the issue to be ethnicized. In the case that a country’s politics are not characterized by ethnic competition and the international investment lacks a relevant ethnic component, ethnicization of the issue will not occur. International investments may still become politicized, but it will most likely not be ethnicized. Ethnic appeals may appear nonsensical in these situations, and would likely be unable to gain traction with voters. Political elites would then opt for rhetoric around the investment along an alternative dimension. In the quadrant that denotes the presence of international investment in a country with high ethnic salience, but the investment lacks a relevant ethnic component, I expect a nonzero, but low likelihood of ethnicization of the investment. Experience from past campaigns in an ethnicized political system may encourage political elites to rely on an ethnic playbook, but in the absence of directly relevant ethnic ties in the investment, ethnic appeals are likely to have their effectiveness diminished. Strategic candidates then may decide that nonethnic approaches will more effective. Consider the example of BRI investments in Zambia. Ethnicity plays a strong role in Zambian political life (Posner 2005), and the country has been a recipient of significant Chinese investment under the BRI. The BRI has become a political issue in Zambia, but criticism of the deals has focused on the lack of transparency in the deals and on economic concerns such as Zambia’s mounting debts or labor concerns (Rapanyane and Shai 2020, Shieh et al. 2021). The rhetoric surrounding the issue 48 lacks the ethnic dimension seen in Malaysia because Chinese ethnicity is not a major political cleavage in the Zambian context. As such, even as BRI has been politicized, it has not been ethnicized in a way that divides Zambia’s domestic ethnic groups. Considering the quadrant in which an international investment has a relevant ethnic component and is present in a country with low ethnic salience, I expect a low likelihood of ethnicization of the issue. This is because politicians and voters are not likely to be accustomed to ethnic appeals. An example of this kind of context can be seen with BRI investments in Singapore. Although Singapore is multiethnic, with an ethnic Chinese majority, Singaporean politics are not dominated by issues of ethnicity to the degree found in neighboring Malaysia. As such, political rhetoric surrounding the BRI has not been ethnic in nature. Finally, the last quadrant details the country and issue context that is the focus of this study in which ethnic salience is high and the investment has a relevant ethnic component. This study explores this situation in detail, using the example of the BRI in Malaysia. When both of these conditions are present, there is a high likelihood of ethnicization of the issue. Candidates and voters are experienced in ethnic campaigns, making ethnic appeals more likely to resonate. The dynamics of these two conditions in the Malaysia–BRI case are explored in greater detail in Chapter 3. This visualization is intended to be suggestive, and this study only performs a test of the theory for political and issue contexts that match the corner in the bottom-right (bolded). I do not explore 49 the cases of high ethnic salience/issue lacks ethnic component or low ethnic salience/relevant ethnic component in depth in this study. The inclusion of these two alternative cases are meant to illustrate the possible outcomes of cases that do not satisfy both conditions of the theory. The empirical evaluations of the theory are carried out in the Malaysian/BRI context in the chapters that follow. The above discussion establishes the plausibility of an ethnic effect of international economic forces on domestic politics. International investment projects are an issue in which the prospect of direct material incentives resulting from the projects to individual voters is weak and information regarding the projects can be low and costly to attain. These factors increase the ability of ethnicity to drive political behavior as citizens seek out alternative sources of information on the project. Candidates are likely to cast BRI in an ethnic light as they campaign, due to the relevant ethnic component of the Initiative and the mobilizational benefits of ethnic identity. To test this proposition, I identify the following hypotheses: H1: Respondents will favor projects with investors who share their own ethnic identity. This hypothesis tests whether respondents show different preferences for investment projects based on the identity of the project’s investor. Ethnic identities can serve as cues that give information to individuals about a project’s merits, thereby guiding preferences for projects in situations without perfect information. The observable implication of this argument is that, on average, 50 respondents will prefer projects run by their co-ethnics. In the case at hand, the expectation is that Bumiputera respondents will favor Bumiputera investors, and Malaysian Chinese respondents will favor Malaysian Chinese investors. H2: Bumiputera respondents will have less preference for projects with mainland Chinese investors than Malaysian Chinese respondents. In a political context that is characterized by ethnic competition, the close connection between BRI investments and ethnic Chinese will contribute to ethnic identification around these projects. Bumiputera respondents may be more likely to associate BRI presence with a perceived loss of their own ethnic group’s influence vis-à-vis that of the Malaysian Chinese, and punish mainland Chinese investments as a result. H3: Bumiputera respondents will have less preference for candidates who promote investments from mainland China compared to Malaysian Chinese respondents. I expect that if voter preferences on investment projects are conditioned by ethnicity, this relationship will translate to their vote preferences as well. A candidate’s issue stance will likely affect vote choice if the voter perceives that stance to be against his ethnic group’s interests. Conclusion This chapter argues that the linkages between the ethnic politics and the second image reversed literatures should be expanded. The use 51 of ethnicity as a tool of political mobilization can apply in situations where political entrepreneurs see it in their interest to exploit international investments relationship to their country’s ethnic cleavages to their political advantage. This can increase the importance of ethnicity in a host country’s political sphere. Similarly, the second image reversed literature would benefit from expanding its scope to look directly at how international economic factors can affect domestic political groups, even those that are not explicitly economic. In contexts where candidates and voters are accustomed to ethnic political competition and ethnic campaign appeals, candidates and voters can be incentivized to present and interpret the effects of international investments such as those under the BRI as ethnic issues. The effects of this ethnicization process can be observed in the preferences of respondents for investment projects from mainland China compared to investments from other sources, and in their preferences for candidates who support investments from mainland China versus from other sources. If respondent preferences are conditional on the respondent’s own ethnic identity, then I interpret this as evidence of the ethnicization of investment projects in Malaysia. The tests conducted in Chapters 4 and 5 are designed to empirically test the theoretical implications discussed here. 52 Chapter 3: Malaysian Politics, Ethnic Relations, and the Belt and Road Initiative Malaysian political competition often falls along ethnic lines. Three major ethnic groups form the basis of Malaysian political competition. These are the Bumiputera (69.6% of the total population),9 Malaysian Chinese (22.6%), and Malaysian Indians (6.8%). The diversity of origin suggested by these group monikers reveals the main lines of distinction and potential sources of political and social conflict. Indeed, the main areas in which these three groups have fought include issues of language, culture, and religion (Milner 2011). Walking down a street in Kuala Lumpur, one can hear conversations in Malay, Mandarin, Cantonese, Hokkien, Tamil, or English, all while passing by Islamic mosques, Christian churches, and Hindu or Buddhist temples. These examples of cultural diversity are vigilantly guarded by each ethnic group, and perceived cultural incursions stand as a highly sensitive and ever-present political issue. Despite the presence of multiple distinct ethnic groups, Malaysia “has always been recognized as a bi-ethnic society, in terms of its intergroup power relations” (Yeoh 2008). These two groups who wield 9 The Bumiputera group is composed of primarily ethnic Malays (who make up approximately 82% of Bumiputera), but also includes individuals identifying as Dayak, Peranakan, and many other “indigenous” groups. The literal translation of “bumiputera” is “sons of the soil.” In this study, I use the terms Bumiputera and Malay interchangeably, though it should be acknowledged that there are significant cultural, linguistic, and religious differences between Malay and non-Malay Bumiputera groups. However, they are grouped together politically in Malaysia, and for the purposes of this study, aggregating the groups provides analytical clarity. This is a practice followed by other major surveys such as the Asian Barometer. 53 political power are the Bumiputera and the Malaysian Chinese. Domestic policies play an active role in managing the distributions of power and wealth between these groups. The arrival of BRI brings new economic opportunities for Malaysian economic development, but its close association with China and Chinese influence has the potential to have disruptive effects on Malaysian domestic politics, particularly by becoming entangled in preexisting ethnic grievances between Bumiputera and Malaysian Chinese citizens. The aim of this chapter is two-fold: first, to discuss the political and ethnic landscape of Malaysia in order to provide context on the behavior of its citizens and political elites in reaction to BRI. It will then discuss the development and growth of BRI and its implications to China and Malaysia. It does so by examining how BRI investments resonate with Malaysian citizens, drawing mainly from primary data that I gathered in a survey of 1,308 Malaysian voting age citizens in March 2021 and interviews conducted for the purpose of this study. Further support is derived from various secondary sources. These analyses provide evidence that the BRI has been ethnicized by Malaysian politicians and establish that the BRI and Malaysia satisfy the two requirements set out in Chapter 2 that lead to the ethnicization of an international issue. Malaysia’s brand of ethnic politics interacts with the ethnic aspects of BRI in such a way that elites strategically appeal to voters by priming ethnic identities, and that voters’ cognitive process regarding mainland Chinese investments is seen through an ethnic lens. 54 Malaysian Politics and Ethnic Relations Any broad study of Malaysian politics must account for the high salience of ethnicity in its political structure, the sources of which are historical, demographic, and institutional. Historical patterns of between-group interactions crafted a sense of distinctive identity between the major ethnic groups and influenced the structure of political institutions, including the party and electoral systems (Osborne 2010). In turn, these institutions have molded the set of political incentives and opportunities within the political system to reinforce competition along the existing ethnic divide (Holst 2012). Finally, demographic patterns establish the viability of ethnicity to mobilize voters to win office and control power (Yeoh 2008). The following section delves deeper into the processes that have elevated the role of ethnicity in Malaysian politics, before transitioning to an examination of how the political narrative around BRI fits neatly into Malaysia’s ethnic political system. Despite the ethnically charged political landscape, Malaysian society should be credited for maintaining a relatively high level of social harmony since its independence. With the exception of a few incidences of political violence, which will be discussed below, Malaysia has not experienced the levels of ethnic or sectarian violence that have plagued many of its regional neighbors in Myanmar, Indonesia, or Thailand. Malaysian citizens express willingness to mix socially and work with other ethnic groups, even if close friendships between ethnic groups are reportedly less common than within ethnic groups (Lee 2017). 55 Nonetheless, an examination of the political realm reveals a country dominated by issues of ethnic identity. Political entrepreneurs from all major ethnic groups have reliably turned to communal politics to mobilize electoral support, and major political parties have found the most success through ethnic platforms. Today, many of the major parties are set up to compete first and foremost on an ethnic basis, contributing to institutional inertia that perpetuates political conflict along the lines of ethnic competition (Holst 2012). Even parties of the opposition that have historically competed on non-ethnic platforms are closely associated with a particular ethnic group, based on the ethnicity of party leadership or support. In response to this system, Malaysian citizens have become conditioned to expect and respond to ethnic cues. In response, political elites often attempt to manipulate and capitalize on perceptions of ethnic threat (Holst 2012). The introduction of a massive infusion of capital from the BRI and China can create opportunities for political entrepreneurs to persuade voters to judge the BRI through an ethnic perspective if citizens are attuned to the ethnic connection of these investments. The first step to examining this process is to lay out the historical contours of ethnic competition and conflict in Malaysia to establish the central role of ethnicity in modern-day politics. Historical sources of ethnic tension According to my theory, for an international economic issue to become a domestic ethnic issue, the host country must have preexisting 56 ethnic cleavages that can be activated by the effects the issue has on the country. As mentioned above, Malaysian politics are in many ways dominated by competition between two groups: the Bumiputera ethnic group and the ethnic Malaysian Chinese. The roots of the use of ethnicity as the main form of political identity lie in historical relations between groups, demographics, and institutional structures. To justify this claim and show that Malaysia fits the theoretical criteria for this study, I delve into the ethnic-political context of modern Malaysia. Prior to European colonization, forms of identity on the Malayan peninsula were centered on a system of distinct sultanates that exercised political authority. The sultan occupied the top of a rigidly hierarchical social structure. Although the sultanates shared similar religious, linguistic, and cultural characteristics, subjects did not yet conceive of themselves as part of a larger “Malay” ethnic identity, but rather associated themselves with the particular sultan with whom they owed allegiance (Milner 2011). Individual sultanates considered themselves as unique peoples and used these divisions to craft a sense of difference and superiority over outsiders, to forge loyalty, and mobilize forces for conquest. These sultanates, such as those in Melaka or Johor, had rather fluid geographic boundaries, and subjects commonly moved from one area to another in search of better prospects (Baker 2008). At best, then, in the pre-colonial period the common form of social identification was as subject of an individual sultanate, and likely did not expand beyond this to a conception of a pan-Malayan identity (Milner 2011, Husain et al. 2021). 57 However, even in this period the experience of Chinese immigrants can be clearly differentiated from that of the indigenous Malayan population. These migrants brought and maintained their own economic, cultural, and political traditions, and did not readily assimilate with the sultanates (Chin 2020). The earliest settlers from these groups possessed a capacity and affinity for commercial activity that surpassed the existing standard in the sultanates (Osborne 2010). Foreign observers noted that Chinese settlers in the peninsula generally lived in separate areas and retained architectural styles, Chinese laws and customs, and religious practices (ibid). The pursuit of wealth was openly accepted, in contrast to Malayan subjects of the sultans, for whom accumulation of wealth was apt to be viewed as threatening the sultan’s position. Some observers view this dynamic – in which a Chinese merchant class was nurtured but an indigenous one stifled – as formative in the race-based political-economic structure that persists to this day (Milner 2003). The clear distinction between Malay and Chinese groups was apparent from the arrival of the earliest traders and settlers. While awareness of ethnic differences and ethnic mentalities existed in the pre-colonial setting, colonial era policies accelerated and solidified ethnicity as a basis of categorization and division (Milner 2011). British colonial administrators created a political system that “was organized around an ethnic division of labor and administrative policies of divide and rule” (Nair 1999). The British instituted a regular census, which required racial classification of residents, with the effect of drawing clear lines between ethnic 58 groups, and encouraged the homogenization of groups around these ethnic lines. Malayan political society and economy were organized along ethnic lines, with Chinese and Indians working primarily in mining and commerce, while Malays kept mostly to traditional roles such as agriculture (Nair 1999). During the colonial period, the British encouraged a large influx of Chinese labor and capital to the Malayan peninsula to provide labor for the growing tin and rubber industries (Kong 2016, Osborne 2010). These settlers existed alongside ethnic Malays, but in separate communities. There was little mixing of living spaces, with most Chinese and Indians settling in commercial and urban areas in Kuala Lumpur, Ipoh, or Seremban (Baker 2008). Economic differences were reinforced by strong business and financial ties within these groups (Abdullah 2013). Politically, members were more likely to refer to their own traditional sources of authority, such as clans or secret societies, than to the authority of local sultans. This bifurcated system persisted in part because a significant portion of the new Chinese and Indian settlers did not view themselves as permanent settlers in Malaya. Many still considered their homeland to be their places of origin. As a result, these settler populations eschewed participation in local politics (Baker 2008). Chinese migrants, for instance, were more concerned with political issues in mainland China such as the growing importance of the communist/nationalist divide there (Osborne 2010). The Malay sultans generally accepted this system, as it insulated their political realms from competition from the growing settler population. However, the 59 effects of World War II and the Communist victory in China fundamentally altered this situation. Chinese settlers, who by this time made up fully 40% of the population in Malaya, faced a new reality in which they could not, or were unwilling to, realistically return to a Chinese homeland (Baginda 2016). In addition to cultural, religious, and linguistic distinctions between the main ethnic groups, demographic factors contribute to the political importance of ethnicity in Malaysia. While the percentage of Malaysian citizens who are ethnically Chinese has fallen over time, this group still represents a quarter of the total population. Malaysian ethnic divisions are also geospatial in nature (Holst 2012). Patterns of Chinese and Indian immigration led to the concentration of immigrants in various urban and commercial centers, while ethnic Malays remained in large part in the rural hinterlands, known as kampong (Baker 2008). Over time, this contributed to economic and occupational divergence between the ethnic groups, and discouraged interaction, mixing, and intermarriage between the races. Malaysian Chinese remain concentrated in urban areas today, which contributes to the high salience of ethnicity in politics. Elections for the Malaysian parliament use geographically demarcated single-member districts, which means that in many constituencies the winner-take-all characteristics pressure parties to compete on ethnicity. The concentration of ethnic groups in different geographic electoral districts creates incentives for political entrepreneurs to mobilize ethnic groups to form winning coalitions. 60 The separation of settler and indigenous communities and their economic, political, and social differences contributed to the formation of ethnic grievances. Economically, Malays resented the commercial success of the Chinese and Indian migrants in relation to their own (Daniels 2005).10 Many of these migrants possessed business acumen and commercial sophistication that outpaced that of the local population (Osborne 2010). They expanded the use of selling on credit, personal and business loans, and the use of middlemen to areas that had previously seen little of it. They also benefited from trans-border networks that provided them with capital and access to markets. These features set the Chinese and Indian groups at an economic advantage over local Malays and contributed to an increasingly discernable income gap between the groups. In the political dimension, Chinese and Indian groups had strong links to the British colonial government. In case of disputes between these groups and local Malays, the migrant groups often relied on colonial laws and enforcement mechanisms over traditional mechanisms of the sultanates (Daniels 2005). In the eyes of local Malays, this led to a close association between the new migrants and the foreign colonial authorities (Osborne 2010, Beech 2018). Additionally, stark social and religious differences endured in each community, such as the Chinese community’s reluctance to embrace Islam or to give up their traditional customs (Baker 2008). The colonial era may not have created the sense of alienation between Malaya’s major ethnic groups, 10 It is important to note that while the Chinese migrant community enjoyed commercial success, many migrants to Malaysia came with virtually no resources or networks and labored for low wages. 61 but the colonial experience crystallized an awareness of ethnic difference and grievance and set the stage for future political conflict to be fought on ethnic lines. These historical relations between the ethnic groups influenced their future interactions as Malaysia neared independence. The various indigenous groups of the Malayan peninsula and Borneo coalesced around a shared identity centered on a loose sense of shared historical presence, religion, language, and custom (Milner 2011). While these individual groups are quite diverse, political expediency has encouraged political cooperation under the Bumiputera label. Similar processes encouraged disparate Chinese communities to cooperate politically. This process homogenized diverse groups within the Bumiputera and Malaysian Chinese communities, while simultaneously hardening the distinctions between them. Independence and the Development of Modern-day Institutions The Federation of Malaya officially gained independence from Britain in 1957 in a peaceful handover of power. The new state was organized under a federal constitutional monarchy, which still exists today. The head of state, known as Yang di-Pertuan Agong, is chosen from one of the nine sultans in the country, who rotate holding the position every five years. Although their formal powers are limited, the sultans hold traditional significance as symbols of Malay and Islamic power. Perhaps the most visible institution representing ethnic Malay power, the position of the sultans is considered a non- negotiable topic by Malay nationalists, and questioning their position 62 is prohibited by law. Day-to-day executive powers rest with the Cabinet, headed by the prime minister. Parliamentary proceedings follow the Westminster system, with the Cabinet deriving its authority from the confidence of the lower house of parliament. The nature of political competition and ethnic relations in the new country was strongly influenced by the colonial experience. British authorities, who had spent much of the last decade of their rule in Malaya countering the Malayan Communist Party (MCP) in a period known as the “Emergency,” engineered the transfer of power by working with partners from each of the three ethnic groups, with the aim of establishing a stable polity in which all three groups had a stake (Baker 2008). Against the backdrop of the Emergency, British authorities supported a political system formulated around ethnic identity instead of class identity (Nair 1999) and negotiated with center-right political leaders from the Alliance, comprised of the United Malays National Organisation (UMNO), the Malaysian Chinese Association (MCA), and the Malaysian Indian Congress (MIC). This represented a consociational arrangement, in which a “government by elite cartel” is “designed to turn a democracy with a fragmented political culture into a stable democracy” (Lijphart 1969, Mauzy 1978). The “Bargain of 1957,” which formed the basis of Malaya’s interethnic social contract, was essentially a quid-pro-quo in which the Malay leadership legitimized the minority Chinese and Indian population as full citizens of the federation, in contrast to their previous status as temporary settlers, and recognized their economic and religious rights (Holst 2012). In return, the agreement reserved 63 a privileged political and cultural role for Malays (Osborne 2010).11 Malay institutions such as the sultanates were protected and given the right to govern cultural and religious issues. The result of Malaysia’s colonial experience and the nature of its regime transition was the continuation and the affirmation of the use ethnicity as the dominant political cleavage moving forward. The gentleman’s agreement securing the privileged position of the Malays soon came under severe pressure. Two incidents in the 1960s revealed the tenuous nature of the agreement and led to major institutional transformations, which had the ultimate effect of solidifying ethnic cleavages. The first incident was the short-lived merger with Singapore from 1963 to1965. Colonial Singapore was governed by the British as part of the Straits Settlements, which incorporated several coastal settlements such as Melaka and Penang (today part of Malaysia). In 1946, British possessions in the area were reorganized. Melaka and Penang joined the rest of the peninsular holdings as part of the new Malayan Union, while Singapore remained a separate colony. As Britain withdrew from the region over the next decade, Singapore faced existential threats in its geopolitical isolation and resource vulnerability. Malayan leaders also viewed Singapore warily, as they were concerned with the possibility of the city-state falling under the influence of communist elements or of an increasingly assertive Indonesia (Sadka 1962, Time 1962). 11 The privileged position of the Malays is enshrined in Article 153 of Malaysia’s 1957 Constitution. 64 In 1963, to allay these threats, Singapore joined Malaysia,12 along with the territories of Sarawak and Sabah in Borneo. However, some Malay leaders were concerned over the addition of Singapore’s heavily ethnic Chinese population to the federation. The leaders of UMNO, the main Malay nationalist party, feared that the addition of Singapore’s predominantly Chinese population could upset the country’s demographic balance, and with it, Malay dominance of the political sphere (Baker 2008). They expressed concern that Singaporean political parties, particularly the People’s Action Party (PAP) under its able leader Lee Kuan Yew, could upset the existing balance of political power and rival UMNO’s Chinese allies in the MCA (Fitzgerald 1965). Notably, the PAP advocated for non-communal politics, while UMNO called for special rights and affirmative action based on Bumiputera privilege. UMNO viewed the addition of the Bornean states as essential to counterbalance the influx of ethnic Chinese citizens from Singapore, since Sabah and Sarawak were predominantly Bumiputera. Additionally, Malay leaders expressed doubts over the loyalty of ethnic Chinese, and feared potential connections between this group and the CCP in China (Baginda 2016). Conflict quickly arose between the PAP, who resisted Malay political dominance, and the Alliance parties. PAP success in the 1964 election, in which it won a seat in the Kuala Lumpur area, was perceived as an encroachment on Alliance territory by UMNO and the MCA (Barr 1997). Economic disputes persisted regarding the establishment 12 It was at this point that the name “Malaysia” was adopted. From this point on, I will refer to the country as Malaysia. 65 of a free customs union. A series of bloody 1964 ethnic riots through the streets of Singapore were caused in part by provocative speeches made by UMNO politicians that accused the PAP of oppressing Malays (Lee 2000, Baker 2008). The fundamental differences between Malaysian and Singaporean interests and vision, dominated by the ethnic dimension, overpowered the benefits of integration with Singapore and convinced Alliance leaders to expel Singapore in 1965 (Toh 1999, Holst 2012).13 The extreme act of expelling the territory reveals how deeply issues of ethnic identity and ethnic threat guided Malaysian politics, and continue to do so today. Ethnic rioting in 1969 formed the second major incident that transformed Malaysian politics from an ethnic standpoint. The direct cause of the riots, which raged for four days and caused at least 200 deaths by official counts, was the 1969 election, in which opposition parties associated with Malaysian Chinese made significant gains against the Alliance parties (Baker 2008).14 Official accounts claim that opposition supporters marched through the streets in celebration, where they came into conflict with their political rivals. While the elections provided the spark for the riots, the tinder was provided by the ethnic grievances of the Malay and Malaysian Chinese populations. Malay citizens generally resented high levels of poverty and the endemic levels of economic inequality in comparison to the Chinese. 13 The Malaysian Parliament voted 126-0 in favor of expulsion. Members of Parliament from Singapore, who wanted to stay in the union, were not present. 14 The Democratic Action Party (DAP) and the Parti Gerakan Rakyat Malaysia (Gerakan) won 31.4% and 17.5% of the vote, respectively. DAP’s party membership is composed mostly of Chinese and Indian citizens. Gerakan, while nominally non-communal, draws its support from Chinese voters, and has its stronghold in heavily Chinese Penang. 66 On the Chinese side, anger was fueled by Malay dominance of politics and the civil service, pro-Malay affirmative action, and cultural issues such as the use of Malay as the national language (Osborne 2010). The riots led to the declaration of a state of emergency and the suspension of parliament, and provided the Malay leadership with motivation and political momentum to institute a number of reforms that reinforced the Malay political position. These incidents convinced the Malay leadership that their political and economic positions would need to be strengthened via legislation. New laws and amendments enacted in the wake of the 1969 riots solidified structural divisions between the major ethnic groups that persist today (Holst 2012). The Sedition Act of 1948, originally passed by the British to limit opposition to colonial rule, was amended in the wake of the 1969 riots to prohibit discussion of “sensitive” provisions of the Constitution, namely those pertaining to the privileged Malay position in Articles 152, 153, and 181. The Sedition Act remains in effect and was invoked as recently as 2020 in response to a Facebook post in which several students questioned the political role of the sultans (Zack 2020). Further restrictions on political expression include the University and University Colleges Act of 1971 and the Official Secrets Act of 1972, which criminalized student involvement in political parties and the possession of secret documents for journalists, respectively. These acts are interpreted by many to restrict critical expression on ethnic policies by civil society and the media (Amnesty International 2020). 67 In the economy, UMNO spearheaded the passage of the New Economic Policy (NEP) in 1971. The NEP was adopted to alleviate widespread poverty and to restructure the country’s socioeconomic situation, in which the Bumiputera were disproportionately poor. The NEP expanded the “special position” of the Bumiputera from areas that were already reserved in the constitution (such as special rights in land ownership and quotas in the civil service and public education) to include provisions that mandated a minimum Bumiputera equity stake in publicly traded companies and exclusive subsidies to purchase automobiles and real estate. These pro-Malay affirmative action policies continue to be a point of contention between ethnic groups in Malaysia (Jomo 2004). Malaysia’s intergroup ethnic relations and its system of ethnic politics are a product of its past, which I have outlined in this section. Critical moments in its history, such as its experiences under colonialism, its short incorporation of Singapore, and ethnic rioting have resulted in legislation and institutions that have perpetuated this system. The following section looks in detail at how the ethnic political system operates in the electoral arena. Ethnic Electoral Politics and the Role of Elites One of the strongest manifestations of the power of ethnicity in Malaysian politics is in the electoral and party systems. Since independence, political elites from Bumiputera and Malaysian Chinese groups have operated along ethnic lines. The transition of power from colonialism to independence vested political power in the Alliance 68 parties, who together represented the three main ethnic groups. The Alliance parties were each organized around ethnic identity instead of alternative political cleavages such as class or ideology, which had the effect of reinforcing ethnic cleavages. As discussed in Chapter 2, the advantages of political mobilization around ethnicity include the ascriptive nature of ethnic identities, as well as its ability to exclude outside groups from the spoils of power (Bates 1974, Chandra 2004). During the 1960s and 1970s, the threats that faced Malaysia at independence began to fade. The communist threat originally posed by the MCP decreased as relations with China improved (Baginda 2016). Relations with Indonesia improved as well. However, the Alliance parties were organized around communal lines and were experienced in campaigning around communal interests and thus were not set up to compete on other terms (Holst 2012). In turn, this reinforced the ethnic dynamic to voters. Elections since independence have been contested and won primarily by ethnic parties. The most successful party has been UMNO, which generally campaigns on center-right economic and social positions in support of Malay nationalism. Until 2018, every prime minister of Malaysia was also the president of UMNO. UMNO’s historic coalition partners, the MCA and the MIC, represent similar center- right factions for each of their respective ethnic groups. Until 1973, UMNO, the MCA, and the MIC won each of the country’s elections after independence as constituent parties of the Alliance coalition.15 The coalition’s relatively weak results in the 1969 15 In Malaysia, electoral coalitions are usually formed prior to elections. 69 elections, in which it retained a majority of seats but did not win a majority of the popular vote, and the subsequent 1969 riots, which caused the suspension of parliament until 1971, led to negotiations to expand the Alliance and resulted in the co-optation of several opposition members into a rebranded coalition known as Barisan Nasional (BN)16. Core membership of BN has been made up of UMNO, MCA, and the MIC, but has incorporated other parties periodically such as the Malaysian Islamic Party (PAS), Gerakan, and the People’s Progressive Party (PPP), or parties from the Bornean states. Current membership as of 2022 is UMNO, MCA, MIC, and the United Sabah People’s Party (PBRS). As was the case with the Alliance, UMNO is the leading partner in BN.17 BN has had dominant electoral success since independence. Cracks appeared at certain elections, such as in 1969 and 2008, but for most of its existence, BN has been able to garner the two-thirds electoral majorities strong enough to amend the constitution. As mentioned above, the opposition parties have faced problems of platform coordination and of attracting broad-based electoral support, and their strongest returns generally have come in the wake of scandals or crises. In the past decade and a half, however, the electoral fortunes of the opposition have improved. In 2008, BN did not win a majority of the popular vote, but still retained a majority of seats 16 Barisan Nasional translates to “National Front.” 17 As of 2022, UMNO holds 37 of BN’s 41 seats in Malaysia’s lower house of parliament, while the MCA and MIC hold two and one seat, respectively. The PBRS currently holds one seat. 70 in parliament due to issues of malapportionment and gerrymandered districting. In 2018, BN faced significant pressure from corruption scandals, partly tied to BRI deals. The fallout from the scandals contributed to BN’s first electoral loss (Minter 2018). The traditional opposition coalition won power, organized under the Pakatan Harapan (PH) coalition. They did so in large part by campaigning with the former prime minister of Malaysia, Dr. Mahathir Mohamad. Mahathir, well-known for his ability to rally Malay support as the leader of UMNO, had become dissatisfied with his former party since leaving office in 2003. In 2017, he formed a new Bumiputera nationalist party called the Malaysian United Indigenous Party (BERSATU) with a number of defectors from UMNO, and joined PH as its prime ministerial candidate for the 2018 election. The addition of BERSATU enhanced PH’s appeal to ethnic Malays, and combined with the corruption scandals dogging BN, was sufficient to oust BN from power for the first time. However, the marriage between BERSATU and its new partners in DAP and PKR was unstable. Disputes quickly surfaced when Mahathir appeared to delay honoring a pre-election agreement he had made with DAP and PKR to hand power to his long-time rival, Anwar Ibrahim, the leader of PKR. In February 2020, parliamentary maneuvers resulted in Mahathir’s resignation, the defection of BERSATU from PH, and the formation of a new governing coalition called the Perikatan Nasional (PN). PN consisted of a coalition of BERSATU, PAS, and Gerakan, and initially cooperated with BN to form a government in 2020. However, further instability was caused by the handling of the 71 Covid-19 pandemic, and in 2021 UMNO withdrew support from the PN government, causing it to collapse. UMNO was able to form a new majority and regain power. The current prime minister of Malaysia, Ismail Sabri Yaakob, is also vice president of UMNO. Major opposition parties include the Democratic Action Party (DAP), and People’s Justice Party (PKR).18 The DAP and PKR generally advocate for center-left policies. While they often try to avoid engaging their rivals on ethnic platforms, they nevertheless draw their support from ethnic bases and are closely associated with them. The DAP is closely associated with Malaysian Chinese, while the PKR is associated with ethnic Malays. Historically, cooperation between opposition parties has been lower than the level of cooperation within the BN. The first time they formed an electoral coalition was in 1999, as the Barisan Alternatif (Alternative Front). In part, the difficulties of forming a stable opposition coalition stem from the fundamental policy differences between the center-left parties and the hardline religious policies of PAS. As noted above, PAS was part of BN from 1972-77, but its hardline position in support of political Islam has also met resistance from the more secular and ideologically central BN. Despite the gains made by opposition parties in recent decades and general instability since 2020, the Malaysian political system retains its ethnic character. As seen in this section, the issue of ethnicity pervades the Malaysian political context due to significant 18 Additional parties also are partners in the ruling coalition, the opposition coalition, or are independent. For brevity, I focus only on the most relevant national level players. 72 historical, institutional, and demographic factors. Political parties formed along ethnic lines are incentivized electorally to reinforce “us versus them” mentalities. Official policies, such as pro-Malay affirmative action policies in the economy and public sector, have remained in place since the 1970s and directly tie ethnic identity to material prospects. While levels of overall economic inequality have decreased over time, ethnic Malaysian Chinese still maintain economic advantages, while ethnic Malays hold a dominant political position. This situation means that ethnic conflict is still quite present in Malaysian politics today (Noor 2009), and Malaysian voters remain defensive of ethnic rights and perceived encroachments by other ethnic groups (Holst 2012). Through this lens, BRI investments have been perceived by many Malay political elites as threatening (Beech 2019). The following section lays out the scope and implications of the BRI for recipient countries. In Malaysia, where ethnic relations between Bumiputera groups and ethnic Chinese have led to conflict in the past and where the power relations between them are frequently a point of contention, the BRI’s connection with China and by extension its connection with Malaysian Chinese creates the potential for the BRI to exacerbate ethnic tensions and impact Malaysian politics. The Belt and Road Initiative When Chinese Premier Xi Jinping introduced the BRI in a 2013 speech, it signaled the beginning of a monumental foreign policy thrust to integrate regional economies via infrastructure construction and investment. The initiative’s close personal association with Xi 73 elevates its role to that of core foreign policy and raises the stakes for its success (Rolland 2017, Lew and Roughead 2021). The level of funding behind the BRI reflects its significance to Beijing (Stokes 2015, Rolland 2017). Chinese officials have touted it as a $1 trillion economic program to enhance trade and infrastructure capacity between Asia, Europe, and Africa. Independent media reports have estimated that the BRI could in fact exceed $8 trillion in total spending (Hurley et al. 2018, 1). As of 2018, BRI programs were present in 78 partner countries, home to over 4.4 billion people who produce over a third of global GDP (Hamzah 2018, 19), and Chinese investments under the BRI umbrella had already surpassed $183.1 billion.19 BRI projects have been concentrated in construction and span a wide range of sectors. Projects have been funded to boost transportation infrastructure, including roads, railways, real estate, and ports, energy with power plants, dams, and pipelines, extractive capacity with mining projects, and information technologies with telecommunications and fiber-optic cables. The BRI is a massively ambitious endeavor in the global economy with profound economic and geopolitical implications for China, its partners, and other stakeholders in regional politics and economics, such as the United States (Lew and Roughead 2021). Given the importance of the BRI to China and its partner countries, I turn this chapter’s focus now to the drivers of the BRI in China and host countries such as Malaysia, 19 While there is no official declaration of which projects fall under BRI auspices, common practice among researchers is to consider all Chinese investment in the period since 2013 as BRI investment (see, e.g., Yean 2018). 74 as well as the potential costs of the program to host countries and the potential interaction of ethnic politics with the BRI. The purpose of BRI to China The BRI advances core Chinese economic and strategic interests both domestically and abroad. Beijing places great value on maintaining domestic stability and growth, but its economy faces several hurdles. China’s rapid economic development has disproportionately benefitted coastal urban centers, outpacing growth in its rural hinterlands. The BRI is intended to alleviate these issues by building infrastructure to China’s rural provinces and integrating them into the national economy, with the hope that development breeds stability and unity (Lew and Roughead 2021, Shullman 2019). In addition, the Chinese economy remains reliant on state-owned enterprises (SOE), which Beijing has continued to support for several reasons. First, these companies provide jobs, which is critical to maintaining domestic order. Second, these companies carry Chinese influence abroad, and Chinese officials envision a stronger role for the renminbi in international transactions. Third, SOEs have a greater tolerance than the private sector to operate at thinner margins or at a short-term loss, since they have state financial backing. This can boost the competitiveness of Chinese SOEs in international infrastructure since they can leverage lower bids in return for influence or market share in strategic sectors like telecommunications. Fourth, SOEs are amenable to top-down control, an attractive feature for Beijing (Rolland 2017). 75 However, in part due to the prominent role of SOEs in the economy, China faces economic headwinds. Without the pressures of market competition, these companies are susceptible to bloated payrolls, inefficiency, and overcapacity. The BRI offers an outlet for these ills by pursuing returns in new markets (Lew and Roughead 2021). Shifting its domestic industrial overcapacity to foreign markets helps to relieve problems of excess capital and industrial materials at home. It can also boost Chinese labor if local workers are used for projects abroad. BRI offers China an alternative to a short-term overhaul of its economic system, heading off potential layoffs and instability. As Rolland notes, “BRI is an attempt to patch China’s most pressing economic problems without fundamentally altering its development model” (p. 108). There is also a strategic rationale to BRI that matches the importance of the economic aspect. Domestically, China hopes to integrate its less-developed provinces via economic development boosted by BRI infrastructure, with the hope that increased development will quell unrest in its western provinces. In order to continue its economic growth, China must secure its oil and gas imports. China currently imports approximately three-fourths of its total oil and gas needs, most of which comes from the Middle East, East Africa, and Southeast Asia. Much of this must be shipped via the natural chokepoint of the Straits of Malacca, which connects the Indian Ocean to the South China Sea and is the busiest shipping lane in the world. China relies on continued access to this strategic maritime route, which is controlled by Malaysia and Indonesia, and is 76 routinely patrolled by the US navy. By increasing connections with regional countries, and increasing Beijing’s influence therein, BRI is intended to give China leverage in securing its access to vital shipping lanes and over other strategic bodies such as the South China Sea (Kong 2016, Minter 2018). Additional BRI projects such as overland pipelines serve as a hedge against potential naval blockades. As BRI increases Chinese presence in partner countries, it also increases its diplomatic influence. Chinese foreign policy views the United States as a potential existential threat due to the latter’s military presence and network of regional alliances. BRI fits into Beijing’s strategy of countering US influence in the region without risking an escalation to military conflict (Rolland 2022). It also pushes back against American pressure on regional regimes to democratize or improve their human rights records. In contrast to Western aid, China does not routinely attach conditionality to its loans that require respect for democratic norms and practices, human rights, or transparency (Balding 2018). Thus, BRI loans offer an alternative and potentially more attractive source of funding to partner countries, which could in turn move these countries closer to Beijing’s orbit. Beyond these points, BRI is symbolically powerful, and asserts Beijing’s claim to great power status and global leadership (Stokes 2015). It recalls eras of Chinese centrality over geographic spheres of influence over the Asian landmass and maritime regions. Evocative connections to the ancient Silk Road travelled by Marco Polo and to maritime routes used by legendary Chinese Admiral Zheng He inspire 77 feelings of reinvigorating old connections and glory (Kong 2016, Lim 2016). Observers refer to the importance of BRI as taking on a “moral narrative” that can help to correct “national humiliation” and “injustices” done upon China in modern times (Rolland 2017). BRI’s connection to national pride and geopolitical influence should not be underestimated, and elevates it from a common economic program to the status of national vision for China’s place in the world, endowed with purpose. To support the initiative, China has set up a robust institutional framework, both financial and intellectual. It has established the Asian Infrastructure Investment Bank (AIIB) as a new vehicle to provide investment for infrastructure development in the region, apart from the World Bank. The United States and Japan did not join the AIIB, leaving China as the dominant stakeholder. China committed $100 billion in funding to the bank and holds its largest voting share (Rolland 2017). The Silk Road Fund also provides funding for BRI projects. It draws 65% of its capital from the Chinese State Administration of Foreign Exchange and an additional 15$ from the Chinese sovereign wealth fund (Fallon 2015). Additional Chinese institutions involved in BRI funding include the Chinese Development Bank and the Chinese Export-Import Bank. Beyond the active government role in financing BRI projects, China has developed intellectual resources geared toward promoting BRI in think tanks, universities, and the media. The goal of these institutions is to enhance soft power resources by improving the image of BRI projects as well as awareness of them at home and abroad 78 (Rolland 2017). Chinese authorities are wary of how BRI projects are received in host countries, and state media works closely with the Chinese propaganda apparatus to put forward a non-threatening image of BRI. Universities aim to improve technology transfers and innovation, while building student exchange programs that can put forward a benign image of China in partner countries. These financial and intellectual institutions reveal the depth of Beijing’s commitment to the success of BRI’s grand vision. Its ambitious plans and financial commitments combine hard and soft power approaches that support the argument that BRI constitutes a core aspect of Chinese foreign policy with potentially far-reaching implications for regional balance of power and its individual countries. Benefits of BRI in Malaysia Malaysia is a major destination for Chinese investments under BRI. Over the period from 2010-2017, Chinese FDI flows in Malaysia increased from approximately $224 million (1.0% of total Malaysian FDI inflows) to $1.51 billion (9.0% of inflows). Much of this investment growth is concentrated in the construction sector, with roughly 42% of the total value of foreign investments in this sector awarded to Chinese companies by 2016 (Todd and Slattery 2018). Malaysia is an attractive destination for Chinese investment funds for several reasons. It is rich in natural resources such as tin, timber, and rubber. It is a middle-income economy, with industrial capacity and human capital levels that make it attractive 79 to multinational firms. Finally, Malaysia sits astride the Straits of Malacca, which is the most heavily trafficked maritime passage in the world. This strategic location encourages global powers such as China to establish a foothold in the region for commercial and strategic reasons (Grassi et al. 2020). The rapid rise in Chinese finance under BRI has spurred debate as to the motives and goals of the policy. According to the Chinese official line, BRI is intended to enhance relations with partner countries in five main ways: policy coordination, infrastructure connectivity, open trade, financial integration, and people-people communication. By this account, BRI is a benign endeavor that will bring benefits to all partners. Outside of the official line, observers view BRI as a tool to increase China’s presence abroad, to secure its territorial claims and possibly to establish itself as the dominant power in the Asia-Pacific region (Lew and Roughead 2021) and have identified a number of areas that Chinese investment and presence can be less savory. The increase in Chinese investments in BRI recipient countries has given way to vigorous debate on its costs and benefits. In terms of its benefits, BRI investment funds contribute capital to a region that has strong demand for it. Estimates by the Asian Development Bank (ADB) call for over $26 trillion of investment over the 15-year period from 2016 to 2030 for Asian countries to meet these goals. In the Southeast Asian subregion alone (represented by membership in ASEAN), infrastructure investment needs over this same time period are estimated to be between $2.8 and $3.1 trillion (ADB 2017). However, 80 the World Bank and ADB together spend about $20 billion annually for this purpose, leaving a clear need for additional funds and funding structures, and funds and institutions under the BRI umbrella can help to address it. Additional benefits associated with BRI projects include jobs, transportation infrastructure (and its accompanying economic multiplier effects), and commercial opportunities for local businesses (Rolland 2017). If properly managed, technologically advanced projects can also improve local workers’ technical knowledge and facilitate technology transfer to firms (Todd and Slattery 2018). Malaysian citizens express openness to Chinese investments in their country. As discussed in Chapter 1, I conducted a survey of 1,308 Malaysian voting-age citizens to assess their opinions of Chinese investments in their country. Respondents were asked to rate on a 0-100 scale the degree that Chinese investments in Malaysia were harmful or beneficial to the economy, with values between 60 and 80 corresponding to “somewhat beneficial” and those above 80 corresponding to “very beneficial.” In the aggregate, roughly two- thirds of respondents rated Chinese investments in Malaysia as “somewhat beneficial” or better (65%). Only 13% of respondents rated Chinese investments as “somewhat harmful” or “very harmful” to Malaysia, and the remaining 22% had a neutral opinion. When the respondents from each ethnic group are examined independently, slight differences between the Bumiputera respondents and the Malaysian Chinese become apparent. 76% of Malaysian Chinese respondents said that Chinese investments are “somewhat beneficial” or “very 81 beneficial” to Malaysia, with 7% saying they are “somewhat” or “very harmful.” Among Bumiputera respondents, approximately 61% said that Chinese investments are at least somewhat beneficial, and 15% rated them as at least somewhat harmful. These results suggests that ethnic identity plays a role in how respondents appraise the relative benefits and risks of Chinese investments. The connection between ethnicity and preference for investment projects will be explored in greater depth in Chapters 4 and 5. However, even among Bumiputera respondents, the majority believe that the net effect of Chinese investments is positive. This suggests that Malaysians are attuned to the various benefits that BRI investments can bring. Costs of BRI in Malaysia However, BRI projects have also been associated with significant costs for recipient countries. One group of these potential costs is economic. Concerns exist as to whether local businesses can compete with more sophisticated Chinese firms (Todd and Slattery 2018). The Chinese firms that win investment contracts boast a high operational capacity with experience in running several large-scale projects concurrently. Additionally, multinational Chinese firms are often directly associated with the Chinese state or have close ties to state financial institutions, which can translate to access to cheaper finance (Lew and Roughead 2021). Ultimately, this can enable large Chinese firms to operate on thinner margins than their local competition. 82 Additional economic concerns exist as to the position of local laborers and suppliers in Chinese investment projects. Chinese firms have opted in the past to carry out their projects with imported Chinese labor and project materials. In an interview I conducted with former Deputy Minister of International Trade and Industry Ong Kian Ming, he explained, “What we saw, at least in the initial part of the Chinese investments coming in, whether property or manufacturing, was that Chinese investors would also bring in their own supply chain…so the rhetoric out there was that even [Chinese workers’] toilet paper would be imported from China” (Ong 2020). While Chinese investors may bring in their own labor and materials in order to circumvent differences in language, work culture, or gaps in worker skillsets, this eliminates prospects for technology and skill transfers, and leaves local labor in positions of low pay and low skill. A second type of cost associated with BRI projects are those tied to the quality of local governance, as BRI has been associated in several cases with corruption or environmental damage. BRI projects have been prone to corruption due to a notable lack of transparency. Past BRI projects in Malaysia have been criticized as having price tags inflated above market value. For example, the East Coast Rail Link was originally awarded at a value of over $16 billion, but alternative estimates of the project estimated that it could be completed at half the price (Wright and Hope 2019). Excess funds from these bloated contracts are suspected to be cycled back in part to Malaysian politicians as a means of illicitly funding their political activities or for personal gain. Indeed, many BRI projects have been 83 questioned as to their financial viability and their resulting impacts on Malaysian sovereignty. The Malaysian sovereign development fund 1MDB became entangled in scandal when it was revealed that Chinese leaders had negotiated with Malaysian leaders to bail out the fund in return for stakes in BRI projects. The proposed concessions were highly lucrative and included the ECRL (over $16 billion) and the Trans-Sabah Gas Pipeline ($2.5 billion). There is some evidence that Malaysian respondents consider Chinese investments to be more corrupt than foreign investments from other countries. My survey asked respondents to compare Chinese investments directly to American investments in terms of the likelihood that they would be associated with corruption. Respondents were more than twice as likely to choose the Chinese investments as more corrupt over American investments (24% to 11%). Still, over half of respondents (51%) believed that investments from these two countries are equally likely to be associated with corruption. Interestingly, when Chinese investments were directly compared to domestic Malaysian investments, respondents labelled the Malaysian investments as more likely to be associated with corruption in comparison to Chinese investments (23% selected Malaysian investments; 18% Chinese). These figures provide some support for the arguments made by observers that the lack of transparency associated with BRI finance can contribute to corruption. Malaysian citizens associate Chinese investments with corruption over some alternative foreign investments. However, they are also wary of the susceptibility of domestic investments to lead to corruption. Corruption is a major 84 risk with large-scale investment projects, but the expectations of Malaysian citizens based on Chinese investments are somewhat mixed. BRI projects have also been accused of causing severe environmental damage. Most of the power plants constructed with BRI funds have been coal-fired plants. China has been willing to continue to finance coal-fired plants even as the United States, South Korea, and Japan have restricted their own financing of coal plants abroad (Lew and Roughead 2021). There have also been concerns connected to the construction of artificial islands around port and commercial facilities. These practices and scandals have led to skepticism among observers as to whether the potential benefits of BRI projects outweigh the strong associations it has to corruption and poor governance. These governance concerns are connected to a third group of concerns stemming from BRI – that Chinese social and political influence can infringe on local sovereignty and strategic concerns (Scissors 2018). The growing influence and assertiveness of China has been the topic of a large body of research in academic and strategic circles (Yu 2017). There is a near consensus that China under Xi Jinping has increased its capacity and willingness to apply pressure in the region to advance its foreign policy, and that BRI is one of the main components of its strategy. Average citizens seem to agree that Chinese influence has increased in recent years. Asian Barometer data from 2019 shows that about 48% of Malaysian respondents think 85 that China has a “great deal” of influence in the country, up from 40% who thought so in the previous wave conducted in 2014.20 Recipient country debt crises are one mechanism that can connect BRI to increased Chinese influence. Incurring high levels of debt to companies that are either Chinese state-owned or that have close ties to the Chinese government has led some observers to highlight the effect of this leverage on recipient country officials and policy (Page and Shah 2018). High debts could feasibly lead to concessions to China that are based not on financial or developmental merit but are granted due to strategic reasons. Malaysian leaders in 2016 courted BRI investments in return for a bailout of the domestic development fund 1MDB. An estimate of the dealings between the Malaysian and Chinese governments in this time claimed that the Malaysian government promised approximately $34 billion in investment deals to China in exchange (Wright and Hope 109). Connections between BRI and Ethnicity As discussed in Chapter 2, for ethnicization of an international issue to occur, the issue must be able to be credibly tied to the host country’s ethnic politics. This means that the international issue 20 Data analyzed in this chapter were collected by the Asian Barometer Project (2018-2021), which was co-directed by Professors Yun-han Chu and received major funding support from Taiwan’s Ministry of Education, Academia Sinica and National Taiwan University. The Asian Barometer Project Office (www.asianbarometer.org) is solely responsible for the data distribution. The author appreciates the assistance in providing data by the institutes and individuals aforementioned. The views expressed herein are the author's own. 86 needs to be relevant to the host country’s particular brand of ethnic politics. In Malaysia, the primary ethnic divide is structured between the Bumiputera and the Malaysian Chinese. BRI has direct connections to the Chinese state, and through its implementation represents growing mainland Chinese influence in Malaysia. If segments of the Malaysian population believe that mainland Chinese influence will translate to increasing influence of Malaysian Chinese domestically, then the BRI can be successfully ethnicized by elites and voters. Even if such a transformation does not actually occur, the mere perception of disproportionate benefit or influence being conferred onto Malaysian Chinese as a result of BRI can be sufficient to lead to ethnic mobilization as a result of BRI. Political elites play an important role in driving an ethnic narrative. As discussed in Chapter 2, candidates need to secure coalitions to win office, and they often pursue strategies of ethnic mobilization when ethnic cleavages present viable winning coalitions. Examining campaign rhetoric around the 2018 general election shows that Malaysian politicians have adopted a strategy that ties ethnicity to BRI. Doing so is not always a straightforward task, because politicians are wary of the risks of presenting themselves as ethnically divisive. As mentioned above, there are legal restrictions on speech that increases ethnic tensions. To circumvent these risks, politicians often use “dog whistles,” veiling their speech in ways as not to cross a line that could invite sanction from legal institutions or outgroups, but are persuasive for the ingroup (Albertson 2015). Commentary from local experts and observers is useful in identifying 87 the underlying ethnic angle of masked appeals, and I refer to several of these analyses in support of my interpretation. One approach used to present ethnic appeals in more ambiguous terms is to couch the appeals in terms of nationalism or patriotism. This is particularly effective in relation to the BRI because of two of the risks associated with the Initiative: concerns around its effects on host country sovereignty and on wealth distribution. Using nationalist terminology allows the speaker to present appeals in a way that, on the surface, can appeal to all ethnic groups, but taken in light of historical and political context, encourage the audience to ultimately interpret the message along well-trodden ground of domestic ethnic competition. These types of appeals have been made by candidates around BRI projects that are high profile due to the nature of the project or its costs, or that can be easily tied to sensitive issues of ethnic balance. For example, the Forest City development in the state of Johor served as a frequent punching bag for Mahathir during his campaign for the 2018 election. As Malaysia’s most well-known political figure, and one known for his strong Malay nationalist credentials, his rhetoric regarding the project is likely to resonate with supporters who are accustomed to interpret a political issue according to its connection to domestic ethnic rivalries. In discussing the Forest City development, a megaproject centered around a massive condominium, Mahathir claimed that the project was “not Chinese investment but a settlement” and further, “We do not want a situation where there is a new version of colonialism happening” 88 (quoted in Beech 2018). Notably, Forest City is funded by a mainland Chinese investor, and the majority of its high-priced residential units had been purchased by foreign Chinese buyers. On the surface, Mahathir’s quotes seem to reflect opposition to international influence instead of a commentary on the project’s effect on domestic ethnic cleavages. However, his word choice is suggestive of historical ethnic tensions between Malays and Malaysian Chinese. The use of the word “settlement” is reminiscent of colonial labels that categorized the immigrant Chinese and Indian communities. Members of these communities during this time were viewed as settlers, conveying a sense of impermanence and strong ties to foreign places and people, and setting them sharply apart from the “indigenous” Malays. The relevance to modern politics is that it still suggests to Malay nationalists a sense of grievance against outsider groups such as the Malaysian Chinese, and suggests the risk of further erosion of Malay influence. Mahathir’s caution against the specter of colonialism also has connections to domestic ethnic tensions. Because he is referring to a project run with Chinese finance and with primarily mainland Chinese residents, the “colonial” power he is referring to is presumably China. Through these thinly veiled criticisms of Chinese influence brought on by BRI, Mahathir is able to signal to voters that these projects have deleterious effects on the domestic balance of power between ethnic groups. This interpretation of campaign rhetoric is consistent with analyses from various other observers in the media and in local politics. The Sultan of Johor, Ibrahim Ismail, responded to 89 Mahathir’s comments by accusing him of “creating fear, using race, just to fulfil his political motives...Dr. Mahathir thinks it is easy to play up race because these investors happen to be from China” (quoted in Jaipragas 2017).21 Journalists and independent researchers have similarly interpreted statements such as those made by Mahathir as serving a strategy of ethnic mobilization “amid suspicions that a private Chinese property developer was somehow secretly plotting to reshape Malaysia’s delicate ethnic balance” and “dilute the Malaysian national identity” (Beech 2018). Statements that cast BRI as a threat to national sovereignty by way of growing Chinese influence are readily tied to issues of Malay power vis-à-vis the Malaysian Chinese. As stated by Welsh, “the attack on Chinese business could be seen to be a criticism of those involved in engaging and profiting from Chinese business as selling out the Malays…The criticism about the role of Chinese business in Malaysia goes to the heart of Malay identity and its position in Malaysia” (quoted in Jaipragas 2017). Within the context of Malaysian politics, with its history of ethnic competition between ethnic Malays and Malaysian Chinese, the issue of BRI has been used in high-profile ways to turn the focus of the campaign to ethnic identity politics, despite the actual substance of BRI seeming only tangential to identity politics. Distributive concerns make up one of the prime areas that the differential ethnic effects of BRI can resonate with voters. The growing role for BRI projects in the economy and their close connection to China inevitably bring up issues of disproportionate 21 It should be noted that Sultan Ismail is an investor in Forest City. 90 gain between ethnic groups, especially in a country with a history of ethnic competition. According to Kong (2016), BRI will “trigger the political sensitive of Malays ethnics [sic] to doubt whether this development could impact their political position and erode their economic condition.” To examine this type of claim, my survey directly asked respondents if they thought foreign Chinese investments conferred greater benefits to ethnic Malays or Malaysian Chinese. The responses showed a discernable expectation that Malaysian Chinese would be the greater beneficiaries of Chinese projects in comparison to ethnic Malays: 36% of respondents said that Chinese investments benefit Malaysian Chinese more than Malays, compared with 10% who said the opposite. Even so, a plurality of respondents (46%) said that these investments benefit both ethnic groups equally. Segmenting respondents based on their own ethnicity enhances the picture. Among Bumiputera respondents, approximately 40% expect Chinese investments to disproportionately benefit Malaysian Chinese, while 12% of these respondents said that Chinese investments benefit ethnic Malays more. A further 41% responded that the benefits confer equally. When looking at Malaysian Chinese respondents alone, only 30% of respondents said that Chinese investments benefit their own ethnic group more than ethnic Malays, while 4% believed that Malays benefit more. 57% of Malaysian Chinese stated that the benefits of Chinese investment are equal between the ethnic groups. Once again, these results suggest a connection between a respondent’s ethnic identity and opinions of the relative benefit of BRI projects between Malaysian 91 ethnic groups. Analyses in Chapters 4 and 5 examine the role of ethnic identity in how respondents appraise projects and how projects affect vote choice. Conclusion The goal of this chapter has been two-fold: to establish the high level of ethnic salience in contemporary Malaysia, and to outline the ethnic connections that can readily be drawn to the BRI. The discussion of ethnicity in Malaysia reveals the complexity of ethnic power relations in domestic politics, and the confluence of forces that have made inter-ethnic competition the dominant dimension of political contestation. Historical interactions, institutions, and demographics condition the incentives and opportunities of candidates and voters to mobilize along ethnic cleavages to maximize their chances of success. Rising influence from mainland China, arriving through the vehicle of BRI, brings with it the promise of much-needed international investment and economic development. But it may be a double-edged sword, as authors have claimed, causing a host of economic, environmental, or governance issues. In addition to these risks, I highlight the potential of BRI to exacerbate domestic ethnic tensions. I have shown that candidates use campaign rhetoric to describe BRI projects in ways that are designed to direct voters to the ethnic aspect of the projects, thereby strategically manipulating the playing field from a purely economic one into an ethnicized one. Survey results from primary data support the idea that voters are 92 receptive to these ethnic appeals, and reveal that voters believe the projects confer differential benefits on citizens based on their ethnic identity. I have provided evidence that the BRI in Malaysia satisfies the two minimal conditions discussed in Chapter 2. The combination of high ethnic salience and a relevant international issue creates a situation in which there is high likelihood of ethnicization of the issue. The next chapter more rigorously pursues the public opinion aspect of this relationship with an experimental design that tests my claim that voters formulate their opinions of BRI projects according to the projects’ ethnic connection as well as their own ethnic identity. 93 Chapter 4: Citizen Preferences for Investment Projects The growth of the Belt and Road Initiative has brought a complex set of competing benefits and costs to Malaysia. In many ways, BRI projects affect the local economy in a similar manner as alternative forms of FDI from other countries, such as Japan or the United States. The degree to which they boost local development, provide jobs and contracts for local firms, or contribute to corruption or environmental issues is subject to intense debate. However, Chinese investment under BRI is notable in its regional geopolitical implications and its connections to Chinese diaspora. Compared to some of Malaysia’s other major foreign investors, such as Singapore, the Netherlands, and Japan, China possesses greater military capacity and strategic ambition. Chinese regional territorial claims in the South China Sea and its assertive foreign policy give additional meaning to the prospect of expanded Chinese economic influence via the BRI. Even in comparison to other major investment partners such as the United States that do possess significant military capacity and regional strategic interests, the concentration of Chinese military power in the region sets it apart. An additional complication for Malaysia and other Southeast Asian countries with a large Chinese diaspora lies in the prospect of increasing influence from mainland China aggravating already sensitive ethnic relations. These characteristics of BRI set it apart from other major foreign investors and warrant a close look at BRI’s effects on recipient countries. 94 Bringing in Voters In this chapter I evaluate the effects of the BRI mainly from the perspective of voters. While policy decisions on the implementation of BRI in Malaysia, such as the awarding and funding of projects, are made at higher levels of government, ministers remain subject to public opinion. This is especially true in democratic settings, where political elites must continue to win electoral support and have incentive to consider public opinion regarding major developmental policies such as the BRI. While a number of existing studies have argued that BRI projects have profound effects on average citizens in recipient countries, these studies tend to do so from a birds-eye perspective (Rolland 2017, Malik et al. 2021). The level of analysis is commonly at the country level or even at that of the international system. Consequently, the conclusions they draw speak to the BRI’s effects on macro-level factors such as economic performance, quality of governance, or strategic concerns. But while some studies have intimated that domestic social forces in recipient countries can form an obstacle to BRI’s implementation, few have directly examined public opinion in relation to BRI. This leaves a gap in our knowledge of the ways that individuals in recipient countries form opinions of Chinese projects and how voter perspectives can impact BRI moving forward. The previous chapter established that a substantial proportion of Malaysian voters consider Chinese investments under BRI to be an important voting issue and that their opinions are influenced by ethnic identity. This chapter extends this analysis with a more fine- 95 grained look at voters’ preference toward foreign investment projects. It looks at multiple aspects of investment projects that have been commonly argued by researchers to affect voters, and identifies which aspects have a significant effect on individual preference for projects. The analysis also features a direct test of the effect of a project’s ethnic connection on how voters rate investment projects. This gauges whether voters factor in the origin of the project investor into their assessment of the project as a whole. Due to the close affiliation of BRI projects to China and the ethnic context in Malaysia, in which power relations between ethnic Bumiputera and ethnic Chinese are commonly contested, it is important to examine whether the ethnic component of BRI and other investment projects resonates with voters. I expect the ethnic and national origin of the investor to affect how the public views investment projects. Investment projects are complex and multifaceted, with the details of bidding and construction often shrouded by a lack of transparency. Additionally, it can be difficult for individuals to fully understand the impacts of the investments. The complexity of investment projects and their competing potential benefits and costs impede the ability of average citizens to discern the net impact of the projects. In situations of low information, or an inability to assess the true impacts of the investments, individuals often resort to the use of heuristic cues to guide their opinion. This is reinforced by the actions of political elites, who have competed on the basis of ethnic identity since 96 independence in Malaysia (Holst 2012, Lemiere 2018). This encourages citizens to be sensitive to the ethnic attachment of an investment project, and to refer to ethnic stereotypes and biases in their assessment of the projects. Existing research has shown that co-ethnic bias is prevalent in many political contexts around the world. A prominent body of scholarship posits that such biases exist because ethnicity provides a readily accessible informational cue to voters, especially in low- information settings (Chandra 2004, Posner 2005, Conroy-Krutz 2012). Even in cases where information limitation is less severe, the effects of ethnic identification can be robust. For example, individuals can process information in ways that favor co-ethnics and punish other ethnic groups (Adida et al. 2017). Research into consumer behavior has documented ethnocentric biases in consumer product preferences (Kaynak and Kara 2002). In my research design, respondents are presented with a moderate amount of information on investment projects: investor origin, investment type, costs, benefits, location, if the investor is state-backed, and investment size. Even so, information on investment projects is imperfect, and I expect the information limitations to encourage Malaysian citizens to rely on ethnicity as a heuristic and consequently to prefer co-ethnic investors. That is, ethnic Bumiputera respondents should prefer projects run by fellow Bumiputera investors, and Malaysian Chinese respondents should favor Malaysian Chinese investors. Additionally, as a Chinese diaspora group, Malaysian Chinese respondents share historical, cultural, and ethnic ties with mainland Chinese. 97 Presumably, these ties can enhance the familiarity that Malaysian Chinese respondents have with mainland Chinese investors, so Malaysian Chinese respondents may prefer mainland Chinese investors over other foreign investors with whom they do not share ethnic ties. In addition to a preference for coethnics, individuals often display a preference for local investments over foreign ones, due to the greater familiarity and availability of information on local investments (Coval and Moskowitz 1999, Massa and Simonov 2006). This relationship should hold when considering foreign or domestic investors for large-scale infrastructure projects. Local citizens have greater exposure, familiarity, and experience with domestic investors than they do with foreign investors. Domestic investors may also be subject to a higher level of accountability from the media or legal institutions. For these reasons, investment projects carried out by domestic actors (such as Bumiputera or Malaysian Chinese investors) should be preferred by respondents over foreign investments, on average. Compared to other investments, BRI investments are unique, in that they may be dually penalized: for their international origin, and for their historical association in the minds of some Malaysian citizens with a rival ethnic group. Additionally, BRI investments have been associated in the media and by political elites with various problematic issues, such as their tendencies to use mainland Chinese labor and materials, accusations of corruption, and accusations of costliness and debt implications. I therefore expect investments from mainland China to be least preferred by respondents compared to other 98 project types. This effect will be amplified for Bumiputera respondents, who do not share national or ethnic identity with mainland Chinese investors. For Malaysian Chinese respondents, the effect will still be present, but weakened since they do not share national identity but have a closer ethnic identification with mainland Chinese investors. Research Design This study uses an experimental design to measure the effect of investor origin on citizen preference for foreign investments. This is a methodological contribution to the literature on BRI, which to date has mostly focused on descriptive narratives and anecdotal evidence to discuss how citizens receive BRI investments. These past studies are susceptible to inferential errors such as selection bias, as they tend to focus on Chinese projects that have been high profile or accused of corruption or ethnic bias. An experimental design offers improved ability to draw causal inferences because variation of the treatment is explicitly controlled. Observed differences post- treatment between respondents are attributable to the experimental treatments and not to unobserved variation. The experimental design also protects against inferential errors of reverse causality. It is possible that respondent preferences for specific types of investment projects drive which type of investors invest in the first place. Specifically, this study employs a conjoint survey experiment. Conjoint designs offer several advantages over standard survey questionnaires or more narrow survey vignettes. For example, they are 99 well positioned to assess public opinion on a multidimensional entity like BRI, for which respondents must weigh competing benefits and risks (Rao 2014). The model’s strength lies in its ability to isolate and compare the effects of individual project characteristics. Because it varies multiple attributes simultaneously, conjoint designs can evaluate competing hypotheses and assess their relative explanatory power, something that more narrow survey vignettes are unable to do (Hainmueller, Hopkins and Yamamoto 2014). This is especially valuable when evaluating BRI, as many hypotheses have been advanced regarding BRI’s effects, but often do so in isolation, without accounting for the effects of other factors. An additional benefit of the design is that it mitigates social desirability bias by presenting the respondents with multiple pathways for their selection. This is necessary when assessing the role of ethnicity in the Malaysian context, where ethnic biases are frequently treated as taboo (Ngeow and Tan 2018). If the respondents are asked to state their preference for investments based on ethnicity alone, they may choose to give a socially “desirable” response, in this case, by underreporting the role of ethnicity in determining their preference. Since the conjoint design enables respondents to mask any of their potential ethnic biases behind alternative treatments, the effect of social desirability bias on the dependent variable, respondent preference, will be reduced (Nederhof 1985). The conjoint design used here first generates randomized profiles of investment projects. Each of these project profiles consists of multiple attributes of investment projects. These attributes 100 represent major characteristics of investment projects in Malaysia. In turn, each attribute consists of multiple levels as set by the researcher. Table 4.1 provides the attributes and levels used to generate investment project profiles. Table 4.1: Attributes and Levels for Investment Projects Conjoint Attributes Levels 1. Investor 1. Malay Origin 2. Chinese Malaysian 3. Mainland Chinese 4. Western 5. Japanese 2. Investment 1. Road Construction Project Type 2. Rail Construction Project 3. Power Plant 4. Real Estate Development 5. Telecommunications Project 6. Mining Project 3. Benefits 1. Beneficial to the National Economy 2. Beneficial to the Local Economy 3. Employs many Malaysian Workers 4. Buys Materials from Malaysian Businesses 4. Costs 1. Associated with Environmental Pollution 2. Employs many Foreign Workers 3. Buys Materials from Foreign Companies 4. Associated with High Costs and Debt 5. Associated with Corruption 5. Location 1. In your Home State 2. In a Wealthy State in Malaysia 3. In a Poor State in Malaysia 6. Investor Type 1. Private Investor 2. State-Backed Investor 7. Investment 1. RM 300 Size (in 2. RM 700 Millions) 3. RM 1,500 4. RM 5,000 5. RM 10,000 The seven conjoint attributes include: investor origin, investment type, benefits, costs, location, investor type, and investment size. Each of these attributes contains between two and 101 six levels. From each attribute, one attribute level is selected at random, and these seven levels are then presented simultaneously to the respondent as a full project profile. The first attribute describes the origin of the primary investor. Investment projects in Malaysia commonly become associated with investors from a particular location. The inclusion of this attribute is intended to discern whether citizens judge the projects based on their association with a particular ethnic group or nationality. If they do, then I argue that there is evidence of the ethnicization of these projects and that respondents characterize the investments based on their ethnic affiliation. The first two attribute levels in this category are investors that would be considered as domestic Malaysian actors, and the final three are various international investors. The first level is Malay, which indicates a domestic investor from the largest ethnic group in the country. The second level is Malaysian Chinese, which instructs the respondent that the investor is domestic, but ethnically Chinese. The next three attribute levels are for international investors, which can illuminate any differences in respondent selections based on whether the investor is domestic or international. The first of these international attribute levels stipulates that the investor is from mainland China. This attribute level is designed to elicit a response from any respondents who are highly concerned about Chinese projects specifically, such as BRI projects. It also allows me to see if respondents differentiate between Chinese investments made by mainland Chinese versus those made by domestic Malaysian Chinese. The next two international attribute 102 levels are for Western and Japanese investors. These levels have been included because they provide non-Chinese foreign investors, which enables a comparison of distinct foreign investors to see if respondents prefer a specific type of foreign investor over others. It also allows me to see if any preference for or against mainland Chinese investments are due to the investments simply being foreign in origin, or if there is particular support or backlash against mainland Chinese investors in particular. The second attribute is investment type. Foreign investment projects, including those under the BRI umbrella, serve a number of different purposes. For instance, projects can be intended to enhance transportation capacity, communications abilities, industrial performance, real estate, and so on. The attribute levels selected for investment type reflect the most common types of BRI projects in Malaysia or those that receive high levels of public and media attention (Rolland 2017, Todd and Slattery 2018). The six attribute levels within the investment type attribute include: road construction, rail construction, power plant, real estate development, telecommunications, and mining. This selection of general investment project types covers a broad range of major projects and includes those that are more extractive in nature as well as those that more directly impacts average voters. Including this range of project types enables analysis of whether specific types of projects resonate positively or negatively with voters. Inclusion of this attribute is also important to isolate the effects of investor identity and ensure that potential effects of investment type are not misattributed to 103 investor identity. It could be possible that respondents associate certain types of investments with BRI investors. Certain BRI investments, such as the ECRL rail link or the Forest City real estate development, have had especially high public profile (Grassi 2020). Failing to include an attribute for investment type could mask potential effects that this attribute has on respondent preference. The third attribute details the benefits associated with the project. Observers have frequently commented on the economic benefits that accompany foreign investment projects in Malaysia (OECD 2018). Including an attribute that dissects the different types of possible benefits associated with projects enables a direct comparison of the different benefits. It also controls for the effects of various project benefits on respondent preferences, isolating the effect and decreasing the risk of confounding the effects of project benefits with the effects of the other conjoint attributes. This attribute has four levels: the project is beneficial to the national economy, beneficial to the local economy, employs many Malaysian workers, and buys materials from Malaysian businesses. The first pair of attribute levels provides a test between benefits that accrue to either the country as a collective unit versus to a more targeted group in which the individual may stand to benefit more personally or at least to a more localized community. The third and fourth attribute levels, test commonly made arguments in support of foreign investment projects and tests if the stated presence of these benefits affects voter preferences for the project. These two factors include whether the projects provide benefits to Malaysian laborers in the form of jobs, 104 or if they provide benefits to Malaysian companies by purchasing materials for the projects from domestic companies. In the case of BRI projects, these two features are particularly salient because detractors of these projects have frequently cited the projects’ heavy reliance on mainland Chinese labor and materials as reasons to shift away from foreign BRI projects (Todd and Slattery 2018). The fourth attribute describes costs that are commonly associated with foreign investment projects, including those of the BRI. Analysts and observers often claim that there are major indirect costs associated with foreign investment projects, of which five of the most common costs are tested here. The first attribute level in this category stipulates that the hypothetical project in question is associated with environmental pollution. This is a common refrain for major investment projects, as they may involve harmful effects for local ecosystems; pollution to air, water, or soil; extraction of raw materials; and so on. This attribute level serves as a catch-all for projects that are noted for their deleterious effects on the environment and captures whether the respondents are sensitive to environmental costs. The second attribute level stipulates that a project employs many foreign workers. A notable criticism of BRI projects in Malaysia has been that they have imported labor from China to work the projects, at the expense of Malaysian workers. The inclusion of this level tests the importance of the origin of labor to the respondents.22 The next level for project cost deals with sourcing 22 It is important to construct conjoint attribute levels such that randomized profiles are realistic and make sense. Some readers may be 105 of project materials, and stipulates that the project buys materials from foreign companies. This level addresses the common claim that BRI projects procure a large number of materials from Chinese sources, at the expense of domestic Malaysian producers. The next attribute level associates the investment project with high costs and debt. This criticism has been frequently leveled at BRI investments, with some high-profile consequences for various projects in the region that have incurred unsustainable levels of debt. The inclusion of this level in a project profile tests the degree to which costly and debt- prone projects affect citizens’ openness to them. The final level in the costs category stipulates that the project is associated with corruption. Foreign investment projects bring considerable sums of money to bear, and are often fertile ground for financial malfeasance. These large-scale projects are often associated with corruption, and BRI projects are no exception. In fact, BRI projects have frequently been tied to increased corruption, with some observers linking heightened corruption to the large sums of money coming in as well as to the relatively permissive requirements loans originating from mainland China in comparison to Western or Japanese loans (Rolland 2017). The inclusion of this attribute protects against the modeling risk that respondents may associate certain costs with certain investors. For example, criticisms of BRI projects in the media often concerned that a project profile that contains the attribute levels that the project employs Malaysian workers as well as employs many foreign workers is contradictory. However, this is not the case, as projects could reasonably employ both foreign and domestic workers in large numbers. 106 focus on high debts and imported labor and materials. Modeling this attribute allows for the effects of each attribute level to be isolated so that their effects are not misattributed to alternative variables. The fifth attribute is the location of the project. There are three attribute levels: in the respondent’s home state, in a wealthy Malaysian state, and in a poor Malaysian state. The level that specifies a project in the respondent’s home state suggests that any benefits or costs associated with the projects will be more localized, and any effects of the projects amplified. The two other options gauge whether respondents engage in a sociotropic evaluation of the project’s effects, that is, whether or not they take into account the effects of any projects on society at large. Projects that are located in wealthy states may be seen as less beneficial than those in poor states, where jobs and infrastructure may have greater impact. Obviously, the attribute levels included here can only measure effects related to general wealth of a state and so they cannot offer a finer- grained assessment of sociotropic effects, but they can offer a general indication of whether general economic evaluations play a major role in evaluations. Including this attribute also serves the purpose of controlling for any masking effects of project location on the other tested attributes. The sixth attribute specifies the private or public nature of the investor. It is possible that citizens react differently to projects based on the private or public nature of the investor, especially because claims have been made of infringements on national sovereignty 107 in some cases. In these cases, citizens may link state-backed investors more closely to foreign governments, and thereby be more wary of the intentions of the project and the overall benefit to Malaysia. This is a potential factor to consider with BRI, since BRI projects receive significant backing from the Chinese government in finances and project implementation. The final attribute provides information on the size of the investment, denoted in millions of Malaysian Ringgit (RM). The five attribute levels specify five distinct levels, including RM 300, RM 700, RM 1,500, RM 5,000, and RM 10,000. At the exchange rate at the time of the survey of 4.1 ringgit to the US dollar, these levels correspond to approximately $73 million, $171 million, $366 million, $1.2 billion, and $2.4 billion, respectively. This attribute measures whether Malaysian citizens consider the size of investment projects when assigning their preferences. It is possible that larger projects are associated with higher risk of debt repayment issues or wastefulness. Since BRI projects are noted for the scale and ambition of projects, this is an important variable to include. Doing so also differentiates the effect of project size from the other attributes. The choice-based conjoint experiment used in this study presents two fully randomized and independent profiles to the respondent side- by-side. The respondent then selects which investment project profile he or she would prefer to have in Malaysia. This selection process is referred to as a “task,” and the binary outcome (determined by whether a given project profile is selected or not selected) serves as the outcome of interest. It represents the respondent’s preference for a 108 particular hypothetical investment project in relation to an alternative. The choice task was repeated for each respondent 12 times. Since each task contains two profiles, each respondent in the study generates a maximum of 24 ratings. An example of a single conjoint task is presented in Figure 4.1. Figure 4.1: Example of a single task The practice of assigning multiple tasks to each respondent in a conjoint experiment is common (Johnson and Orme 1996, Bansak et al. 2018). There are several benefits to doing so. First, it 109 substantially increases the number of observations, enabling conjoint designs to be a cost-effective tool for eliciting public opinion. Second, respondents may undergo a learning process when participating in a conjoint experiment. The amount of time elapsed for respondents to complete the first or second tasks in a study is often much higher than the time needed for subsequent tasks (Johnson and Orme 1996). By including multiple tasks, researchers can test whether early tasks produce anomalous results due to a respondent’s learning process. However, requiring respondents to engage in multiple tasks carries methodological risks as well. For example, increasing the number of conjoint tasks for each respondent can induce fatigue. Each comparison between two investment project profiles requires the respondent to consider multiple features of two alternatives and to select a preference. Repeating this procedure many times carries the risk that respondents will tire and decrease their attention paid to later tasks. Practitioners generally advise that conjoint designs limit the number of attributes to fewer than ten, and the number of levels within each attribute to no more than about eight (Johnson and Orme 1996). This is to prevent overburdening the respondent. As such, the number of tasks contained in this study are in line with other conjoint experiments in the field, and well within limits of laboratory experiments that have examined changes in response patterns in conjoint experiments as the number of tasks increases (Bansak et al. 2018). It is unlikely that the number of tasks given in this conjoint experiment will induce fatigue to the extent that responses become significantly affected. Furthermore, the choice- 110 based nature of the tasks is comparatively more simple, and less fatigue-inducing, than other conjoint designs. Conjoint designs that require respondents to rate each profile, or that include a written vignette, ask more of respondents than do the relatively simple choice-based tasks included here. Additional design considerations used in this study include randomizing the order in which attributes appear to respondents. It is plausible that respondents may place greater emphasis on attributes that appear higher up on the list, for example. This introduces order bias into the results, and would artificially magnify the importance of certain attributes. In this study, the order of the attributes is randomized for each individual in order to remove order bias. Within each individual’s set of tasks, however, the order of the attributes is unchanged between tasks. This means that for an individual respondent, the order of the attributes will be presented in the same way for all tasks, but for another respondent the order of the attributes will be randomized for his tasks as a whole but not changed between tasks. The purpose of this approach is to minimize the risk of respondent fatigue over many tasks. Data and Results The completed choice tasks are analyzed using the conjoint approach described by Hainmueller, Hopkins and Yamamoto (2014). This approach estimates the average marginal component effect (AMCE) of each of the conjoint attributes on respondent preferences. The AMCE represents the average change in the probability of a profile being 111 chosen when it includes a particular attribute level compared to the baseline level, accounting for all possible values of the other components tested. It denotes the causal effect of each of the study’s attributes on the respondents’ preferred investment project. These effects reflect preferences for the population of respondents and can also account for interaction effects between attributes. Estimates for the AMCE from conjoint data result from regressing the dependent variable (whether the respondent selected the investment profile) on a series of dummy variables for each level of the various investment project attributes. Reference categories for each attribute are omitted from the regression equation. Since each respondent performs multiple tasks, it is reasonable to expect nonindependence of preferences within tasks performed by the same respondent. Therefore, standard errors are clustered at the respondent level. Overall, there are 31,560 observations of investment project profiles collected from 1,308 respondents. Details regarding data collection and sampling are discussed in Chapter 1. The predicted statistical power of this model is over 98%, which is greater than the conventionally recommended power of 80% and indicates that the model is sufficiently powered.23 The results of the fully pooled regression model can be seen in Figure 2. The AMCE plots represent the change in the probability of a respondent selecting a profile with the attribute level noted, 23 All power analyses in this chapter were conducted using the tool created by Martin Lukac and Alberto Stefanelli. https://mblukac.shinyapps.io/conjoints- power-shiny/ 112 compared to the baseline category. AMCEs to the right of the dotted line at ‘0’ signify that respondents are more likely to prefer an investment project with that specific attribute level, compared to investment projects that contain the baseline category attribute level. AMCEs to the left of the line signify that respondents are less likely to choose project profiles with that attribute level in comparison to profiles that contain the baseline attribute level. The points for each AMCE represent the regression estimate, and the lines represent 95% confidence intervals. The baseline categories are omitted from the regression. Figure 4.2: Effects of Investment Project Attributes on Respondent Preference 113 The primary analysis tests the claim that Malaysian voters are sensitive to the origin of the project investor when forming their opinions of projects. At the same time, it tests various alternative hypotheses on what aspects of projects are most salient, enabling a comparison of the relative effects of various project attributes on respondent choice. As seen in Figure 4.2, the origin of the investor has a strong effect on respondent preference. Respondents prefer projects with a Malay/Bumiputera investor over other types of investors, including other domestic investors (the Malaysian Chinese attribute level). Respondents also have a clear preference for Malay/Bumiputera investors over foreign investors from the West or Japan. The effects of Western and Japanese investors on respondent preference are both negative and statistically significant. However, relative to all other investors included in the study, respondent dislike for mainland Chinese investors is far stronger. Respondents were approximately 12% less likely to choose an investment profile that had a mainland Chinese investor as opposed to a Malay/Bumiputera investor. This effect is matched only by voters’ strong antipathy to projects that are noted for their association to corruption (see below). Further examination is necessary to assess whether respondents are punishing mainland Chinese investments simply due to ethnically based anti-Chinese sentiment (whether foreign or domestic) or if mainland Chinese investments are punished for additional reasons. The experimental design’s inclusion of unique attribute levels for 114 mainland Chinese and Malaysian Chinese investors mitigates the risk of misinterpretation along these lines. If respondents were judging projects purely on ethnicity, I would expect to see similar levels of antipathy toward domestic Malaysian Chinese investments. However, the results show that while both mainland Chinese and Malaysian Chinese investors are not preferred to Malay/Bumiputera investors, mainland Chinese are much less preferred to their Malaysian Chinese counterparts. As noted above, respondents are 12% less likely to pick an investment with a mainland Chinese investor compared to a Malay/Bumiputera one. In contrast, projects with a Malaysian Chinese investor are only about 2% less likely to be selected in the same comparison. This shows that Malaysian respondents are not just reacting to Chinese ethnicity in general, but clearly differentiate between foreign Chinese and domestic Malaysian Chinese investors. Mainland Chinese investments, such as those under BRI, are significantly less attractive to Malaysian citizens. This shows that respondents are basing their choices on additional considerations beyond purely ethnic ones, and that there are specific aspects of mainland Chinese investments, like their attachment to foreign Chinese influence, that discourage preference for those projects. While this experiment cannot state with certainty exactly what respondents prefer in domestic Chinese investments compared to foreign Chinese investments, it is likely a combination of factors. As discussed in Chapter 2, these can include a preference for local actors due to respondents’ greater familiarity with these actors’ preferences and capabilities, and tendencies. It could also be a reaction to 115 perceived high costs of mainland Chinese (BRI) investments, such as the associations made by politicians and the media that attach these investments to debt issues, corruption, or labor and materials sourcing. Beyond the ethnic element of BRI investments, it is also possible that respondents are punishing mainland Chinese investments because they are against foreign investments in general. The AMCE estimates for other foreign investors do not support this position. Western investors were also less preferred in comparison to Malay/Bumiputera investors, as respondents were about 6% less likely to select these projects. However, Western investors were significantly more likely to be selected when compared directly to mainland Chinese investors. Similarly, Japanese investors, while approximately 4% less likely to be selected in comparison to Malay/Bumiputera investors, were significantly more likely to be selected in comparison to mainland Chinese investors. In light of these findings, respondents’ preferences against mainland Chinese investors cannot be explained by their foreign status alone. Malaysian citizens prefer other major foreign partners, at least from the West or Japan. While this relationship holds for Western and Japanese investors, the design is limited in making claims outside of the tested attribute levels. It cannot rule out the possibility that Malaysians hold low opinions of untested partners, such as those from Indonesia or Russia, for example. Regardless, the results of this model support the argument that mainland Chinese investments are significantly less attractive to Malaysian respondents than investments that are tied to other major 116 domestic and foreign investors, and that the cause of their preference is not due to Chinese ethnicity alone. These results provide evidence that Malaysian citizens are highly responsive to the origins of investors for large projects in their country. Compared to some commonly cited factors that affect individuals’ preferences regarding investment projects, investor origins have a relatively greater effect. This suggests that ethnic factors associated with projects can stand as politically salient issues. I have taken steps to differentiate between respondent preferences regarding Chinese ethnicity in general and foreign Chinese investments. The significant difference in respondent preferences between mainland Chinese and Malaysian Chinese investors suggests that preferences are not simply based on Chinese ethnicity, but that there is a clear preference for domestic Chinese investors compared to foreign Chinese investors. Furthermore, I have taken steps to show that respondents react differently to foreign Chinese investments as opposed to foreign investments generally. The greater dislike that respondents stated for foreign Chinese investors compared to both Western and Japanese investors shows that it is not just a preference for domestic investors – there is an additional penalty for foreign Chinese investors. Additional attributes further reveal the preferences that Malaysians have toward projects. The project type attribute shows that respondents prefer certain types of investment projects over others. There is no statistically significant difference between preferences for road construction (the baseline), rail projects, power 117 plants, and telecommunications. However, respondents were significantly less likely to prefer two project types: real estate and mining. On average, respondents were about 3% less likely to prefer a real estate project compared to a road construction project. Real estate was singled out by Mahathir as a sector that was benefiting foreign citizens more than Malaysians. As referred to in Chapter 1, the Forest City condominium development was selling in large numbers to foreign citizens, especially from mainland China. The high level of attention received by real estate may be contributing to its negative reception by respondents here. Mining projects are also less preferred by respondents, and are about 6% less likely to be selected by respondents when compared to a road project. This might be due to the extractive nature of mining projects, as well as environmental concerns. Among the levels within the benefits attribute, respondents do not show a strong preference between projects that benefited the national economy versus the local economy, suggesting that they do not differentiate strongly between benefits or costs that accrue in their own area over those that are nationally dispersed. Respondents do prefer projects that are noted to employ Malaysian workers. On average, they are 2% more likely to select these investment project profiles over the baseline project that specifically benefits the national economy. The respondents’ preference for projects that employ local workers may be due to the attention in the Malaysian media and among politicians on this issue specifically. As noted in Chapter 3, there is a strong narrative in Malaysian discourse that 118 imported Chinese workers for BRI projects are harmful to the Malaysian economy. Finally, respondents showed little preference for projects that were described as buying Malaysian materials. When interpreting this result, it is important to remember that all attribute levels described beneficial aspects of projects, so interpretation of the AMCE estimate on local materials should be considered only in comparison to the baseline category that the project benefits the national economy. Respondents here were about 2% less likely to prefer projects that sourced local materials in comparison to those that benefited the national economy. The various costs associated with investment projects are frequently brought up in BRI research. As with the benefits attribute, when analyzing the costs attribute it is important to note that AMCE effects are all estimated in relation to the baseline category, and positive or negative effects are in reference to how respondents feel about a particular cost in relation to the baseline. Compared to projects that are associated with environmental pollution (the baseline), respondents preferred projects that employed foreign workers (2% more likely to be selected) or bought materials from foreign companies (9%). Respondents exhibit statistically similar preferences for projects that incur high costs and debt as those that pollute the environment. A particularly strong effect on respondent preference is seen for projects associated with corruption. Respondents were 13% less likely to select these projects in comparison to those associated with pollution. Connections to 119 corruption are clearly a strong motivator of opinion on investment projects. The location of the project and its size also affect respondent preferences. Against a baseline investment project in the respondent’s home state, respondents have less preference for projects based in a wealthy state, while showing similar preference for projects based in a poor state. This suggests that respondents are attuned to the developmental benefits that projects can bring to their localities, and have more interest in projects that boost either poorer regions, or their home regions. The size of the project has some effect on preferences as well. Respondents showed the least preference for the smallest project, with size denoted by monetary value. The baseline category of RM 300 million ($70 million) received uniformly lower support in comparison to the four attribute levels attached to larger projects. However, there was no statistical difference between these four larger project levels themselves. The final attribute, investor status, revealed no statistical difference in respondent preference between private investors and state-backed investors, suggesting that respondents place relatively little weight on the investment’s direct connections to state institutions. These findings show the relative preference that respondents have for each attribute level relative to the baseline category. Taken in isolation, they cannot show that respondents do or do not want investment projects with specific attribute levels, but are limited in showing which levels are preferred over others. For example, the negative AMCE estimate for mining projects does not convey that 120 Malaysians are against mining projects in general, but that they prefer road projects over mining projects. Nonetheless, the conjoint design provides valuable insights into the various aspects of investment projects that matter most to voters. Ethnic Identity While respondents as a whole exhibit a preference for various features of investments, such as those undertaken by Malay or Bumiputera investors, these effects might be conditional on the respondent’s own ethnic identity. As discussed in Chapter 3, ethnicity is a highly salient political cleavage in Malaysia, and the unconditional model shown above reveals that voters judge the attractiveness of investment projects in part on the project’s ethnic component. The next models condition the effects of the conjoint attributes on investment project preference on respondent ethnicity, and present results for Bumiputera and Malaysian Chinese respondents, the two largest ethnic groups in Malaysia. Respondents self- identified their ethnic identity as either Malay, Bumiputera, Malaysian Chinese, Malaysian Indian, or Other. Malay and Bumiputera respondents are grouped together for this analysis as in the previous chapter. This analysis was conducted by interacting the respondent’s ethnicity with each of the conjoint attributes. As in the pooled analysis, the conditional analysis clusters standard errors on the respondent. Plots of the AMCEs can be seen in Figures 4.3 and 4.4. Figure 3 shows the estimates and 95% confidence intervals for Malay and Bumiputera respondents, calculated from 20,112 observations 121 derived from 831 respondents. Power analysis for the Bumiputera model with 831 respondents and 12 tasks per respondent yields a predicted statistical power over 95%. Figure 4 shows the estimates and 95% confidence intervals for Malaysian Chinese respondents, with data from 8,160 observations generated by 340 respondents. This model is sufficiently powered with a predicted statistical power of 85%. Figure 4.3: Effects of conjoint attributes on respondent preference conditional on Malay or Bumiputera ethnicity 122 Figure 4.4: Effects of conjoint attributes on respondent preference conditional on Malaysian Chinese ethnicity When looking at each ethnic group separately, the effect of the investor’s origin on respondent preference matters more sharply, and in opposing ways for the two largest ethnic groups in the study: Malay/Bumiputera and Malaysian Chinese. Bumiputera respondents favor investments done by their co-ethnics over all other alternatives. Projects identified as having a Malay investor were at least 11% more likely to be chosen by Malay respondents than projects having any other investor type. This provides evidence that ethnic identity plays an influential role in individuals’ sentiment regarding 123 investment projects. The reason for ethnic favoritism in this area could be due to several reasons. Respondents may believe that Bumiputera investments bring greater benefits to the Bumiputera community, for example. Although Bumiputera respondents show a clear affinity for co- ethnic investors, their preferences vary among the alternatives tested. In particular, compared to projects with mainland Chinese investors, these respondents were over 22% more likely to select a project with a Malay investor. As is the case in the unconditional model, the results suggest that it is not simply a matter of preferring domestic investors. Bumiputera respondents have a discernable preference for projects run by their domestic co-ethnics over domestic investors with Chinese ethnicity. In addition, they are indifferent between domestic Malaysian Chinese investors and foreign investors from either the West or Japan. The direct comparison between mainland Chinese investors and Western or Japanese investors shows that respondents discern between different foreign investors and that they are particularly wary of mainland Chinese investors. It is also not the case that Bumiputera attitudes toward Chinese ethnicity in general can explain their preferences. They prefer Malaysian Chinese investors significantly more than they do mainland Chinese investors. Respondents may be punishing mainland Chinese investors due to issues of transparency and high costs commonly associated with BRI. Interestingly, Malaysian Chinese respondents exhibit starkly different preferences regarding investor origin than their Bumiputera 124 counterparts. The most apparent trend when looking at Malaysian Chinese respondents is a particular antipathy toward Malay/Bumiputera investors. Malaysian Chinese respondents penalize investment projects with Malay/Bumiputera investors more harshly than those with any other investor. Compared to the baseline project with a Malay/Bumiputera investor, Malaysian Chinese respondents were about 11% more likely to select a project with a mainland Chinese investor, 13% more likely to select a project with a Japanese investor, and 11% more likely to select a project with a Western investor. Malaysian Chinese respondents also prefer their own domestic co-ethnics as investors, with a greater likelihood (~17%) of selecting a project with a Malaysian Chinese investor over a Bumiputera investor. This suggests that rival ethnic tensions in the domestic arena are especially salient for Malaysian Chinese respondents, and could represent a reaction against special rights and privileges afforded to Bumiputera citizens in Malaysia. One might expect that Malaysian Chinese respondents may exhibit a greater preference for mainland Chinese investments due to possible cultural ties. The evidence presented here does not offer strong support for that position. While mainland Chinese investments are preferred over Bumiputera investments, they are evaluated equally to the other foreign investments tested (mainland Chinese, Japanese, and Western). The differences in the AMCE estimates for mainland Chinese, Japanese, Western, and Malaysian Chinese investors do not achieve statistical significance when compared to one another, but all do when compared directly to Bumiputera investments. The main driver of 125 Malaysian Chinese opinion regarding the origin of the investor is in the domestic arena. After accounting for the divergent responses for investor origin across both Bumiputera and Malaysian Chinese groups, these results appear to suggest that mainland Chinese investments arouse the strongest responses among Bumiputera citizens. Political operatives looking to capitalize on BRI may find the most receptive audience among this group. Indeed, many of the loudest campaign appeals against BRI investments have come from politicians courting Bumiputera votes. In the campaign for the 2018 election, Mahathir Mohamad was especially strident in criticizing BRI investments from an ethnic angle. It not likely coincidence that Mahathir has advocated Malay nationalist platforms over his career, first with UMNO, and in 2018 under the newly created, but similarly Malay nationalist party BERSATU. Looking at the other conjoint attributes, there is not a profound change in the AMCE estimates when segmenting by respondent ethnicity. Bumiputera and Malaysian Chinese respondents hold similar views on the type of project, least preferring mining projects. In the benefits category, Malaysian Chinese respondents show a preference for projects that are noted to benefit the national economy over projects that are noted to buy materials from Malaysian businesses, indicating that the issue of sourcing materials from local companies does not particularly resonate with them. For both Bumiputera and Malaysian Chinese groups, preferences in the costs attribute are similar. They view projects that buy 126 materials from foreign companies as most preferable to the other costs tested. This suggests that the issue of materials sourcing may be the “lesser of evils.” Without doubt, corruption is the greatest of evils among those tested here, evoking strong negative reactions from respondents regardless of ethnic identity. This suggests that if BRI projects continue to be associated with corruption moving forward, they will encounter broad-based resistance from across the Malaysian voting spectrum. Regarding location of investment, Bumiputera respondents show a greater dislike for projects located in wealthy states, while Malaysian Chinese respondents do not. The Bumiputera group was over 4% less likely to select a project in a wealthy state compared to one in the respondent’s home state, while the model for the Malaysian Chinese group failed to detect a statistical difference between these two attribute levels. One explanation for this difference could arise from the difference in where these groups live, as discussed in Chapter 3. Malaysian Chinese citizens are more likely to live in wealthier, urban areas, while Bumiputera citizens are more concentrated in the relatively poorer rural hinterlands. The stronger reaction among Bumiputera respondents against projects in already wealthy areas could represent their assumption that projects in their home states would lead to greater benefit for Bumiputera citizens, while the same relationship does not hold for Malaysian Chinese. Finally, the attributes describing the attachment of the investor to the state or private sector and describing the size of the investment show little variation based on respondent ethnicity. 127 Respondents from both groups are indifferent to the private or state- backed status of the investor, suggesting that this is not a major factor in forming preferences, at least in a situation where these other project factors are considered simultaneously. And as in the unconditional model, the models segmented by ethnicity show that both major groups generally prefer larger projects over the smallest baseline project size. Conclusion While the growing literature on BRI has examined the numerous effects the BRI has on host countries, strong evidence for these assertions has been lacking. This study improves on this by using original data and a conjoint experimental design to identify the preferences that Malaysian citizens have for investment projects in their country. The findings of the experiment establish a strong effect of investor identity on respondents’ preference for large investment projects in Malaysia. In line with expectations of previous work on ethnic politics, Malaysian voters tend to prefer investments to be undertaken by their own coethnics. This result holds for ethnic Malays as well as Malaysian Chinese. However, Malaysian voters are not monolithic, and respondents’ own ethnic identity also plays a pivotal role in their project evaluations. Ethnic Malays have a notably lower preference for projects with mainland Chinese investors than do ethnic Malaysian Chinese. On the other hand, Malaysian Chinese respondents do not show a significantly different preference 128 between mainland Chinese investors and other foreign investors. Instead, they exhibit the least preference for domestic Malay/Bumiputera investors. This chapter clearly establishes that ethnicity plays a major role in how Malaysian citizens formulate their preferences for investment projects. These effects can have consequences for the future of BRI investments in the Asia-Pacific region and can signal headwinds for Chinese economic endeavors in the future due to the risk of increasing ethnic identities and pushback in host countries. Additionally, the differences between the Malays and Malaysian Chinese suggest that the Chinese ethnic component of BRI may resonate differently for each group, with consequences for how domestic political elites choose to devise their campaigns. The next chapter continues to explore the interaction of ethnicity with investment projects by examining whether this relationship translates when citizens enter the voting booth. 129 Chapter 5: Chinese Investments and Candidate Preference The previous chapter outlined the various features of investment projects that affect the average Malaysian citizen’s preference for projects. Based on the conjoint experiment employed in that endeavor, there is evidence that the ethnic affiliation of investment projects has a significant effect on respondents’ choice of which project they prefer. There is also evidence that the direction and magnitude of the effect on respondent preference is contingent on the respondent’s own ethnic identity. Given that respondents are sensitive to ethnic effects in their evaluation of large investment projects, this chapter extends the analysis to examine these effects in a direct connection to political outcomes in a democracy. It looks beyond respondent preferences for investment project features into how support for projects affects a citizen’s vote. As in Chapter 4, the analysis is drawn from a choice- based conjoint design. Instead of generating randomized profiles of investment projects, the current design generates profiles of candidates for political office. The critical attribute that ties the analysis to ethnicity and the BRI is one that stipulates a candidate’s support for various origins of investments. This enables me to quantify the level that a candidate’s support for a particular investment affects respondent preference for that candidate. The BRI and Voters’ Preferences for Candidates As a whole, this dissertation is devised to test whether the BRI as a political issue in host countries that affects political behavior 130 via ethnic pathways. While the experiment in Chapter 4 is necessary to directly examine how citizens respond to aspects of the BRI, it does not speak to how impactful of a political issue BRI is in determining how Malaysians vote. In several interviews that I conducted in Malaysia in 2020 with politicians and academics, interviewees were generally in agreement that BRI investments had been made into a political issue, but were less sure of the efficacy of campaign appeals in voters’ minds (Ong 2020, Gomez 2020). Determining the BRI’s impact on voting is important to ascertain whether Malaysians view the BRI as a major political issue, or if they see it as more peripheral when compared to other issues. It is one thing for voters to have opinions on a political issue, and another for that opinion to matter enough when they ultimately cast their vote. The analysis in this chapter attempts to provide an answer to this question. Similarly to Chapter 4, this chapter measures public opinion. To my knowledge, it is the only study that directly examines the effect of a candidate’s position on BRI (and other foreign investments) on vote choice. Paying attention to how individuals process and weigh candidate affiliations with BRI offers an important step forward in the BRI literature, allowing us to situate the issue in direct comparison to other major electoral issues. Why would a candidate’s position on investments affect how citizens vote? There are several reasons why this may be the case. First, the scale and high visibility of BRI investments have pushed the issue to become a major issue in Malaysian political discourse. 131 Coverage by the local news media and mentions by local politicians have increased public awareness of the BRI and brought conversations regarding its various benefits and costs to the fore. As citizens become familiar with investments and accustomed to hearing political appeals around them, they may be more likely to give the issue greater weight when deciding how to vote. Second, the fact that Malaysian politicians have used the BRI as a campaign issue suggests that they think it is a potentially effective motivating issue. Assuming that politicians are rational actors, and that campaign choices are constrained by limited time and money, politicians are likely to focus on issues that they think will gain traction with voters. The use of the BRI as a campaign issue suggests that politicians think that the issue can bring in votes among at least a subset of the population. Which voters are more likely to consider a candidate’s investment stance when casting their vote, and what voting patterns emerge? The theoretical foundation for how a candidate’s position on BRI investments will affect votes for that candidate again mirror many of the arguments described in Chapter 4. First, I expect that respondents will favor their coethnics. Respondents are operating in a low information environment and are likely to use ethnic cues as a voting heuristic. Since the candidate profiles generated in this experiment are of hypothetical candidates, respondents will need to infer candidate capabilities and intentions. In these situations, ethnicity is often used as a signal to voters (Chandra 2004, Posner 2005, Conroy-Krutz 2012). 132 There are two attributes in the design that convey ethnic signals. The first is straightforward and notes the candidate’s ethnic identity. The second attribute with ethnic connotation describes the candidate’s stance on investments. Within this attribute, the levels are attached to specific investor origins. In one of those levels, the candidate supports increased investments from China. This attribute level is of primary interest for discerning how BRI support affects support for a candidate. As discussed in Chapters 3 and 4, investments from China can be readily tied to Malaysia’s existing structure of ethnic political competition and consequently have the potential to activate ethnic sentiments in Malaysia. Making a respondent aware that a candidate supports more foreign Chinese investment can lead respondents to connect this to a pro-Chinese stance, and to refer to domestic ethnic rivalries when voting. I expect that respondent preferences will be affected by both of these candidate attributes. Regarding the candidate’s investment position, Malaysian respondents have shown, in the analysis presented in Chapter 4, that they have strong preferences on their ideal investment projects. Furthermore, these preferences are conditional on the respondent’s own ethnic identity, signifying a strong effect of ethnicity. Bumiputera respondents show a general dislike for investment projects headed by a mainland Chinese investor, while Malaysian Chinese respondents did not distinguish between mainland Chinese investors and other foreign investors. I expect this relationship to carry into the voting booth. Given the discussion above, there are several testable implications of this theory. 133 Respondent preferences for candidates should be contingent on the respondent’s own ethnic identity. If, on average, respondents are more likely to favor coethnics, and candidates encourage respondents to view investments from China in an ethnic light, then respondents may consider mainland Chinese investments to translate to increased Chinese influence in general. Reacting to this ethnic cue, respondents who identify as ethnic Bumiputera may be more likely to punish candidates who support higher levels of investment from China. On the other hand, Malaysian Chinese respondents are unlikely to perceive the same level of threat from increased Chinese influence. These respondents are less likely to punish Chinese investments due to their ethnic connections. Research Design The research design used for this chapter shares many similarities with that used in Chapter 4. It retains the basic framework of the choice-based conjoint experiment. Respondents again had the task of comparing two randomly generated conjoint profiles and simply selecting which of the two profiles they preferred. What is materially different in this chapter is the focus on candidate profiles instead of investment profiles, and therefore the content of the attributes and attribute levels. This changes the outcome of interest to the probability that a respondent will select a candidate who supports particular policies regarding foreign investments rather than looking at the individual characteristics of investment projects themselves. The benefit of this approach is that it can address 134 whether voter opinions on investment projects matter when they ultimately cast their votes. This enables a more complete picture of how BRI and other investment projects affect domestic politics in a democratic setting such as Malaysia. Table 5.1: Attributes and Levels for Candidates Conjoint Attributes Levels 1. Foreign 1. Supports Increased Investments from China Investments 2. Supports Increased Investments from the USA 3. Supports Increased Investments from Japan 4. Opposed to Increased Foreign Investments 2. 1. Pakatan Harapan (DAP/PKR/AMANAH) Party/Coalition 2. Barisan Nasional (UMNO/MCA/MIC) 3. Perikatan Nasional (BERSATU/PAS) 3. Corruption 1. Never Accused of Corruption 2. Accused of Corruption 3. Prosecuted for Corruption 4. Convicted of Corruption 4. Taxes 1. Supports Higher Sales and Services Tax 2. Supports Lower Sales and Services Tax 5. Ethnicity 1. Malay/Bumiputera 2. Malaysian Chinese 3. Malaysian Indian 6. Profession 1. Farmer 2. Salesman 3. Teacher 4. Businessman 5. Lawyer 6. Doctor 7. Age 1. 30 2. 40 3. 50 4. 60 5. 70 8. Gender 1. Male 2. Female The attributes and corresponding attribute levels are displayed in Table 5.1. This conjoint design generates candidate profiles from eight attributes, each of which has between two and six attribute 135 levels. These eight attributes are: candidate position on foreign investments, party/coalition membership, corruption, position on taxes, race, profession, age, and gender. The main attribute of interest for this analysis is the candidate’s position on foreign investments. The remaining attributes are included because they represent factors that are commonly believed to affect how voters evaluate candidates. This enhances the experiment’s external validity by allowing for a comparison of each attribute’s effects, because they are all simultaneously used in the selection of a common outcome. In addition, some attributes are included in order to mitigate the risk of omitted variable bias. Some candidate attributes may be correlated, at least in voters’ minds, with the candidate’s position on foreign investment. Isolating the effects of these potentially correlated attributes limits the risk that these effects may be misattributed to the candidate position on investments and enhances the study’s internal validity. To generate each candidate profile, one attribute level from each attribute is randomly selected, and they are presented as one complete candidate profile. The first attribute is the candidate’s position on foreign investments. As discussed in Chapter 3, Chinese investments have become a political issue in Malaysia. Political elites have referred to the issue repeatedly as BRI funds grow in Malaysia. The projects have generated controversy in regards to their economic viability, effects on local labor and local firms, lack of transparency linking them to corruption, and unease regarding the heightened level of Chinese influence in the country. While this sets BRI up as a 136 political issue with the potential to affect voting behavior, the fact remains that BRI is just one issue among many, and there is no guarantee that is salient enough as an issue to move the needle at the ballot box. The purpose of including a candidate’s position on foreign investments is to test the effect of this variable on vote choice in comparison to other, perhaps more established, factors. The foreign investments attribute is divided into four levels, each of which signals to the respondent a candidate’s policy toward particular foreign investments. The first level is that the candidate supports increased investments from China. The analysis of this attribute level directly relates to the hypotheses tested in this chapter. The next two levels stipulate that the candidate supports either increased investments from the United States or increased investments from Japan. The inclusion of these two alternatives is necessary to delineate between respondents who support increased investments from a particular foreign source, rather than simply supporting foreign investments in general. Including the United States and Japan allows for a direct comparison of these sources to each other and to China, and can shed light on whether respondents have a particular antipathy to Chinese investments over other foreign investments. The final attribute level describes the candidate as opposed to increased foreign investments. This level can reveal if respondents are opposed to foreign investments in general. The second attribute is the candidate’s party or coalition. Partisanship is often a strong predictor of individual voting behavior, and Malaysia is no exception (Welsh 2013). Accounting for a 137 candidate’s partisan alignment mitigates the risk that respondents will conflate a candidate’s investment position with a particular partisan alignment. This is necessary because parties in Malaysia have taken different positions on BRI. For example, PH candidates, as members of the opposition during the onset of the BRI in Malaysia, were generally more critical of BRI projects than were their rivals in BN. Failing to include an attribute for candidate partisanship could cause respondents to infer the candidate’s partisan identity from their issue stance on investments. In that case, respondent preferences for a party or coalition could be misattributed to investment stance. Including the attribute for partisan identity allows respondents to differentiate between the two. Including candidate partisan identity also allows for a comparison of the effects of partisanship with those of investment position and the other candidate attributes. I have simplified the Malaysian party system for the purposes of this analysis. Attribute levels are constructed based on the three major party coalitions at the time the survey was run. Arranging the parties according to their coalition also aids in the creation of plausible candidate profiles, since candidates of any ethnic group can fit into the coalitions. These electoral coalitions are discussed in greater depth in Chapter 3. The first attribute level is Pakatan Harapan (DAP/PKR/AMANAH). Pakatan Harapan is the name of the coalition that won power in the 2018 election, and DAP, PKR, and AMANAH are three of the biggest component parties in the coalition. Grouping parties in this way allows respondents to easily identify the 138 candidate’s general partisan alignment, while maintaining a streamlined conjoint design. The second level is Barisan Nasional (UMNO/MCA/MIC). BN is historically the most successful party coalition, and UMNO, MCA, and MIC are its most steadfast members. The final level is Perikatan Nasional (BERSATU/PAS). This is a relatively new coalition, formed after the 2018 election. The timing of the survey, which was launched in early 2021, necessitated the inclusion of PN, which by that time led the majority in parliament. Its inclusion provides additional insights into the effects of partisan identity on voter choice in contemporary Malaysia. The third attribute is the candidate’s connection to corruption. Corruption is a major issue in Malaysian elections, in regards to BRI projects and otherwise (Gomez 2012, Edwards 2018). Most notable in the lead up to the 2018 election was the scandal surrounding UMNO and Prime Minister Najib Razak’s handling of funds in the Malaysian development fund 1MDB. The scandal had international fallout and ultimately contributed to the BN’s first loss of its parliamentary majority in its history (Wright and Hope 2019, Bisserbe et al. 2020). As noted in Chapter 3, BRI projects are frequently criticized for their opacity, from the bidding process to the implementation of the projects, and have been the target for allegations of corruption. Due to the possibility that respondents could associate a candidate’s stated support for increased investments from China with an increased propensity for corruption, it is necessary to include this attribute to isolate its effects. For these reasons, I include a candidate’s association with corruption as a conjoint attribute. The attribute 139 has four levels, with varying levels of severity. The lowest level is that the candidate has never been accused of corruption. From this baseline, the next three levels are that the candidate has been accused of corruption, has been prosecuted for corruption, and has been convicted of corruption. The benefit of these levels is that they can measure the extent to which voters punish candidates for corruption, and the difference between rather “light” associations with corruption, such as an accusation, and deeper associations such as convictions. Assuming that respondents have an aversion to corrupt officials, respondent preferences for this attribute should be monotonic and negative as the candidate’s association with corruption increases. The fourth attribute is the candidate’s position on taxes. During the 2018 election, a major issue was the cost of living and the level of the Sales and Services Tax (SST) (Kumar and Jaipragas 2018). Currently, the sales tax on most common goods is 5%, and the tax rate for services is 6%. Since tax rates were a major issue in recent elections, I have included them in this analysis. The two levels are simple: the first level specifies that the candidate supports higher SST, while the second level specifies that the candidate supports lower SST. This attribute has been included because the issue of taxation was repeatedly cited as a major campaign issue in my field work and research into the 2018 Malaysian general election. The inclusion of taxation as an attribute increases the realism and external validity of the experiment, and provides a useful point of comparison for how the attribute of primary interest, the candidate’s 140 stance on investments, compares to other issues that are seen as influential on voting choice. The fifth attribute is the candidate’s ethnicity. This attribute is divided into three levels, to represent the three main ethnic groups in Malaysia. The first level is Malay/Bumiputera, grouped together since the two groups are treated similarly in official policies. The second level is Malaysian Chinese, and the third is Malaysian Indian. Including an attribute based on the candidate’s own ethnicity allows me to see the degree to which respondents prefer members of particular ethnic groups, for example if they prefer co- ethnic candidates or if they display a dislike for candidates of rival ethnic groups. Existing studies are aware of the risk of social desirability bias when asking respondents to use sensitive topics like ethnicity to rank candidates, and go to lengths to mitigate risk of this bias (Nederhof 1985). In this case, the conjoint design functions in a similar manner, because the candidate’s ethnicity is presented as only one attribute of a complete candidate profile, thereby allowing respondents to “disguise” any ethnic biases, if they hold any. As discussed above, it is also necessary to include an attribute for candidate ethnicity to avoid masking the effects of ethnicity with the candidate’s investment stance. This prevents respondents from inferring candidate ethnicity from the candidate’s investment stance and isolates each effect. The next three attributes describe features of the candidate that are commonly included in conjoint studies that ask respondents to select their preferred candidate (Doherty et al. 2019, Ono and Yamada 141 2020). The sixth attribute is the candidate’s profession, which may give voters a rough signal as to the candidate’s background, class interests, or position on economic issues. I have included six attribute levels for candidate profession, representing a range of common professions. These are: farmer, salesman, teacher, businessman, lawyer, and doctor. The seventh attribute is the candidate’s age. It has five levels, increasing by intervals of ten years, from 30 to 70. The eighth attribute is the candidate’s gender, denoted by male or female. Including these attributes allows for a comparison of the size of the effect of the candidate’s investment stance with the effect of these descriptive candidate attributes on vote choice. The attribute levels were carefully selected so that they would not create implausible combinations when arranged together into a full candidate profile. This allows the conjoint design to randomly select one level from each attribute. The attribute combinations are plausible due to the heterogeneity within the party coalitions used here on candidate ethnicity, investment stances, and tax positions. My strategy of identifying only the candidate’s coalition instead of party gives additional flexibility to create plausible candidate profiles. Each coalition is made up of multiple parties and has multiethnic elements. For example, the BN is composed of UMNO (which is primarily ethnic Malay) as well as the MCA (Malaysian Chinese) and the MIC (Malaysian Indian). Similarly, the PH coalition is composed of parties that are commonly associated with different ethnic groups. DAP membership is primarily Malaysian Chinese, while PKR is mainly 142 Malay. For the third coalition, PN, I listed the constituent parties of BERSATU and PAS. While these two parties are generally thought of as Malay parties, in 2021 PN added the primarily Malaysian Chinese GERAKAN to the coalition. The presence of ethnic diversity within each coalition ensures that candidate profiles are not made implausible by the candidate ethnicities and coalition memberships used in this design. Within each coalition, there is also heterogeneity of investment stances. To be sure, the coalitions have generally taken identifiable stances on investments. For example, the BN, having been the party in power during the acceleration of investments from China, has often touted the economic benefits of the BRI to Malaysia, while the opposition in PH has called attention to the negative aspects of the BRI, such as its high debts. For the purposes of plausibility of the candidate profiles, what is important is that candidates from within each coalition can take each of the investment stances available in the conjoint design. This is not an issue with the investment stance options presented here. For example, although BN has generally emphasized the economic benefits of the BRI, BN support of more investment from the US or Japan is not precluded by this stance. In a similar vein, although PH candidates more frequently campaigned by calling attention to the BRI’s negative aspects, they recognized the benefits of foreign investment to Malaysia, and were generally wary to frame their messaging as wanting the right kind of investment from China (Ong 2020). In designing the attribute levels for the candidate’s investment stance, I attempted to create levels that would 143 ensure plausibility of the stance regardless of the candidate’s coalition. It also must be plausible for a candidate from each coalition to support the two tax policies in the design. The two policy levels are relatively simple: the candidate either supports higher or lower sales and services tax. After PH won the 2018 election, they fulfilled a campaign promise to repeal Malaysia’s previous tax scheme (known as the Goods and Services Tax) and replace it with the Sales and Services Tax. As this experiment was run in 2021, I elected to include a straightforward candidate tax position in regards to the existing SST. It is plausible for candidates from all included coalitions to support either a higher or lower rate of taxation. The attribute has been kept intentionally general as a way of maintaining plausibility. Using the attributes and attribute levels described above, the execution of the conjoint experiment is identical to that carried out in the previous chapter. Two full profiles are displayed next to one another, and the respondents are asked to select the profile of the candidate they would prefer to vote for. The outcome of the task for each candidate profile is binary, depending on whether the profile was selected or not. An example of a single conjoint task for this study can be seen in Figure 5.1.24 24 Sample task generated using Sawtooth Software. Although the display shows a total of 13 tasks, only 12 are used in the analysis. The 13th task is a fixed task (not randomized) which should not be used in estimation of AMCEs. 144 Figure 5.1: Sample conjoint task Respondents repeated the choice task 12 times, yielding 24 observations per respondent. The main benefit of repeated tasks per respondent is to increase the statistical power of the model. Power analysis with 1,308 respondents, 12 tasks, and the specified attribute levels in this design yields a predicted statistical power over 98%, indicating that I have sufficient observations to detect statistically significant results.25 Additional benefits of asking respondents to complete multiple tasks are that it allows respondents to learn from 25 Power analysis was conducted using the tool created by Martin Lukac and Alberto Stefanelli. https://mblukac.shinyapps.io/conjoints-power- shiny/ 145 one task to the next, and is a cost-friendly way to increase the number of observations while retaining the integrity of the model (Johnson and Orme 1996). Reasonable concerns may be raised that asking respondents to complete so many tasks can be onerous and induce fatigue that would compromise results. Such risks of “survey satisficing” among respondents can include loss of focus, rushing through tasks, or uniform answers due to fatigue or boredom (Galesic and Bosnjak 2009). In this case, the risks may be even higher, because this particular conjoint was conducted after respondents had already completed the conjoint tasks detailed in Chapter 4 (regarding investment projects). This means that the final conjoint task completed by each respondent in this study is actually the 26th task (13 investment project tasks and 13 candidate tasks). Research into the acceptable number of conjoint tasks has examined this issue specifically, finding that respondents can perform 30 conjoint tasks without showing signs of satisficing that impact the stability of results (Bansak et al. 2018). Similar to this design, Bansak et al.’s test employed a choice-based conjoint with two profiles per task. This particular design places relatively low demands on the respondent due to the straightforward decision structure. It is unlikely that the results of my conjoint task are adversely affected by problems of respondent fatigue. The order of the attributes in the candidate profile was randomized between respondents. This is to prevent biasing responses based on the position of the attribute in the profile. In tasks for a single respondent, the order of the attributes remained constant for 146 each of the tasks that respondent saw in order to reduce the cognitive demands on the respondent. Data and Results Data from the conjoint experiment are analyzed by estimating the average marginal component effect (AMCE) described in Chapter 4 (Hainmueller, Hopkins and Yamamoto 2014). In the context of this experiment, the AMCE is the average change in the probability that a candidate profile is chosen when it includes a particular attribute level, compared to a specified baseline level. For example, the baseline category in the “foreign investments” attribute is set as “opposed to increased foreign investments.” The AMCE is the change in the probability that a profile containing another level within that same attribute, such as “supports increased investments from China,” is selected, averaged across respondents and accounting for the possible values of the other attributes. Calculating the AMCE for each attribute level gives the causal effect of the presence of each attribute level on the preference respondents have for candidates. The dependent variable in the model is determined by whether the respondent chooses the candidate profile or not. It is a binary measure that takes a value of 1 if the respondent chooses the profile over its side-by-side comparison, and 0 if not. This variable is then regressed against a series of dummy variables that specify if the attribute level they correspond to was present in the profile, omitting reference categories. Standard errors are clustered at the 147 respondent level to account for nonindependence in tasks completed by the same individual. The results from the pooled model are presented in Figure 5.2. AMCE estimates are plotted with 95% confidence intervals. Points that are to the left of the dotted line mean that candidate profiles that contained the specified attribute level were less likely to be selected in comparison to profiles that contained the baseline level. Points to the right correspond to an increased probability that the inclusion of the attribute level led to preference for the profile compared to the baseline level. And estimates that cross the dotted line at 0 show no statistical difference in respondent preference between profiles that contained the specified attribute level compared to the baseline. The pooled model estimates are based on 31,560 observations generated by 1,308 respondents.26 26 Additional details on the sample and sampling procedures can be found in Chapter 1. 148 Figure 5.2: AMCE estimates for the fully pooled model The results from the model demonstrate that a candidate’s position on investments does have a significant effect on vote choice. Respondents were about 3% more likely to select a candidate who supports Japanese investments over a candidate who opposes further foreign investments. Respondents also show a preference for candidates who support Japanese investments over those who support Chinese investments. This shows that Malaysian respondents are not opposed to foreign investments in general, but differentiate between the effects of investments from various origins. However, the effects of candidate investment stance appear to be limited otherwise. The results also show that respondents are indifferent between candidates 149 who oppose increased foreign investments and those who support Chinese investments. There is also no statistical distinction between candidates who support Chinese investments and those who support investments from the United States. While this suggests that candidate support for Chinese investments does not lead voters to punish candidates (compared to a pro-American or an anti-investment stance), if candidates support a more attractive alternative, such as Japanese investments, citizens prefer those candidates. A possible reason for this is that Malaysian citizens hold a higher opinion of Japan in general compared to both China and the United States. A 2019 survey conducted in Malaysia by the Asian Barometer asked respondents which country should be a model for Malaysia. A plurality of respondents selected Japan (37%), while far fewer selected China (23%) or the United States (6%). Malaysian respondents in my conjoint experiment may be responding to the tendency of Chinese or American investments to come with “strings,” such as Chinese influence in a country with a sensitive history to it, or in the American case conditionalities for democratic practices. Both countries also have geopolitical goals, which may cause unease in a country with a history of nonalignment. However, the data at hand are not equipped to answer these questions and can only justifiably support the claim that candidates who support Chinese investments are preferred less than an identical candidate who supports Japanese investments. Among the other attributes tested in this model, the effect of corruption is notable. Respondents have a very clear preference for candidates who have never been accused of corruption. Simply having 150 been accused of corruption is sufficient for a candidate to be 21% less likely to be selected compared to a candidate who has never been accused of corruption. As might be expected, more severe associations with corruption warrant harsher punishment by voters. Candidates who were prosecuted for corruption were 28% less likely to be voted for compared to candidates with no past corruption accusations, while corruption convictions made candidates 35% less likely to be selected by voters. The strength of the findings for corruption reveals the necessity of including the corruption attribute in the experimental design to protect against its effects being masked by another attribute such as candidate investment position. Isolating the effects of corruption as a separate attribute gives more confidence that the AMCE estimates for candidate investment position are not masking a significant alternative causal factor in corruption. While the negative effect of corruption on respondent preference may be unsurprising, the magnitude of the effect shows that corruption is a major determinant of how Malaysian citizens vote. This corroborates perspectives on the 2018 election among the media and in interviews I conducted that claim that corruption was a major factor in PH’s success over BN (Ong 2020, Bisserbe et al. 2020). It also suggests that if BRI projects continue to be closely associated with corruption, especially in comparison to other sources of foreign investment, voters will likely turn against the initiative. Another clear preference respondents have is for candidates who support lower taxes. Against the baseline candidate who supports higher taxes, a candidate who supports lower taxes is approximately 151 11% more likely to win a respondent’s vote. As mentioned in Chapter 3, cost of living was a major issue in the 2018 election, and the advantage a candidate has by supporting lower taxes is significant in this analysis. The pooled model shows that respondents have no significant preference based on the candidate’s coalition. The baseline candidate (a member of Pakatan Harapan), is statistically indistinguishable from candidates from Barisan Nasional or Perikatan Nasional. At first glance, this result is in contrast to literature that holds partisan identity as one of the strongest predictors of vote choice, and belies the highly partisan nature of Malaysian politics. However, segmenting respondents by their own ethnicity reveals patterns that are obscured by the pooled data, and shows that Malaysian respondents do in fact use partisan identity as a voting cue. These results will be discussed in greater detail below. Another candidate attribute that I expect to affect vote choice is the candidate’s ethnicity. Chapter 3 details how ethnicity is engrained into the Malaysian political system, and the predominance of ethnic parties in Malaysia leads me to expect that candidate ethnicity would play a part in determining vote choice. Indeed that is the case, as Bumiputera candidates are favored over Malaysian Indian candidates. Malaysian Chinese candidates are also favored over Malaysian Indian candidates, though the effect is reduced. As is the case with the attribute for candidate partisan identity, the effects for candidate ethnicity are conditional in part on respondent ethnicity. The section below will discuss this relationship further. 152 The two attributes that describe the candidate’s profession and age also have some effect on respondent preference. Respondents seem to favor candidates with white collar professions. Compared to the baseline category of a farmer, respondents preferred candidates with a background in business, law, or medicine. Respondents also monotonically favored younger candidates. A 70-year-old candidate, for example, was about 7% less likely to be preferred by respondents in comparison to a 30-year-old candidate. The final attribute, candidate gender, has no effect on respondent preferences. Respondents were equally likely to vote for a male candidate as a female candidate. Ethnic group differences Due to the strength of ethnic identity in Malaysian politics and the close ties of BRI investments to Chinese ethnicity, it is reasonable to expect that the two major ethnic groups in Malaysia, the Bumiputera and the Malaysian Chinese, exhibit divergent preferences when evaluating candidate positions on Chinese investments. The following models interact the respondents’ ethnic identity with the conjoint attributes to estimate AMCEs. Figure 5.3 is a plot of the AMCE estimates for Malay and Bumiputera respondents, who are grouped together for this analysis. This is a sample of 831 respondents, generating 20,112 observations. Figure 5.4 plots the AMCE estimates for the Malaysian Chinese respondents, derived from 8,160 observations from 340 respondents. These segmented samples are identical to those in Chapter 4. Power 153 analyses for the plots in Figure 5.3 and Figure 5.4 show that the models are sufficiently powered with predicted statistical power of 95% and 85%, respectively. Figure 5.3: AMCE estimates for Bumiputera respondents only 154 Figure 5.4: AMCE estimates for Malaysian Chinese respondents only Segmenting the respondents on ethnicity does not yield starkly different results when considering candidate investment positions. Both Bumiputera and Malaysian Chinese respondents have a preference for candidates who support increased investments from Japan compared to the baseline value of a candidate who is opposed to increased foreign investment. On average, Bumiputera respondents were approximately 3% more likely to select a candidate in support of Japanese investments compared to a candidate who is opposed to increased foreign investments, whereas Malaysian Chinese respondents were about 5% more likely in the same scenario. Both sets of respondents were statistically indifferent between candidates who 155 opposed increased foreign investments, supported more investment from the United States, or supported more investment from China. While the AMCE estimates for Malaysian Chinese respondents suggest that they may be more favorable to candidates who support investments from China or the United States than their Bumiputera counterparts, the results do not achieve statistical significance at the 95% level.27 These results are surprising, given the ethnicized nature of investment projects in Malaysia and results of the conjoint experiment in Chapter 4. This may be due to the diluted strength of the ethnic signal that is given by this attribute. When voters consider candidate support for a particular investor identity, it is just one part of a candidate’s appeal, and its effect may be mitigated by the presence of other factors. For example, one factor that may weaken the effect of the candidate’s investment stance is the candidate’s ethnic identity. The ethnic aspect of investment projects, which I argue in Chapter 4 is a central driver of how many Malaysian respondents formulate opinions on investments, is presented alongside several other attributes that send ethnic signals to voters and may be overshadowed by the straightforward ethnic cue given by the candidate’s ethnic identity. Candidate ethnicity sends a more direct signal to voters as to the candidate’s possible behavior and consequently may account for much of the effect of ethnicity, thereby weakening, while not eliminating, the effect of the candidate’s 27 Malaysian Chinese respondents may prefer candidates who support American investments compared to the baseline level. They were about 3% more likely to select these candidates. However, this result only achieves statistical significance at the 90% level. 156 investment stance. It could also be the case that while respondents may interpret investments through ethnic pathways, they do not interpret candidate positions on investments in the same way. Unlike a candidate’s ethnicity, his policy positions lack the ascriptive or descent-based nature that scholars commonly use to conceptualize ethnicity and to explain its mobilizational capacities. As a result, candidate positions on investments may have a diminished role in affecting vote choice. This does not preclude the ability of investment stance to affect how citizens vote, but likely makes the effectiveness of candidate policy positions more situational. The results of these two models show that ethnicity does play a strong role in vote choice, and its effects are conditional on respondent ethnicity. Respondents from both ethnic groups showed strong preference for coethnic candidates. Bumiputera respondents were 8% more likely to select Bumiputera candidates than to select the baseline Malaysian Indian candidate. Malaysian Chinese respondents were almost 7% more likely to prefer their coethnics over a Malaysian Indian candidate. Also of note is that both sets of respondents were indifferent between the two non-coethnic options tested, suggesting that each group on average did not necessarily punish candidates from a specific outgroup, but primarily expressed support for their own coethnics. These results are in line with expectations that respondents depend on ethnicity to provide voting cues in low information environments, and that respondents generally favor their coethnics when doing so. 157 Differences among the respondent ethnic groups are also apparent in the candidate coalition attribute. Bumiputera respondents preferred BN and PN candidates over PH candidates and selected candidates from these coalitions approximately 5% and 6%, respectively, more often than candidates from PH. As discussed in Chapter 3, BN represents the traditional power structure with the Malay nationalist UMNO party at its core. PH, the traditional opposition, draws more multiethnic support and won power in the 2018 election, only to lose its majority following parliamentary maneuvers in 2020. PN, which formed in 2020 with members of BERSATU and PAS at its core and draws its support primarily from ethnic Malays, briefly took power as a result of this transition. The support seen in this analysis from Bumiputera respondents for BN and PN in comparison to PH is consistent with these coalitions’ reputations as protectors of Bumiputera rights and privileges. On the other hand, Malaysian Chinese respondents strongly favored PH over the other two coalitions. They were 16% more likely to select a PH candidate over a candidate from BN, and 12% more likely to prefer PH over PN. Again, these results show that voters predictably favor coalitions that have reputations and track records for promoting their own ethnic group’s interests. The remaining conjoint attributes tested in these models do not show significant differences between the Bumiputera and Malaysian Chinese groups. Both groups showed strong preference for candidates who supported lower tax rates, suggesting that cost-of-living concerns remain an important determinant of vote choice among Malaysians across 158 the ethnic divide. Regarding candidate corruption, both groups again showed symmetrical preferences for less corrupt candidates, with preferences decreasing as a candidate’s association with corruption became more concrete. In both respondent groups, candidates convicted of corruption were over 30% less likely to be selected by respondents compared to the baseline candidate who had not been accused of corruption. The effects are slightly more pronounced among Bumiputera respondents, but very strong in both groups. These two ethnic groups both display a strong aversion to candidates associated with corruption. Candidate profession does not seem to be a strong determinant of vote choice, at least when presented alongside the other attributes in this design. The attribute levels represent a diverse group of professions, but among Bumiputera respondents the only statistically significant preference over the baseline candidate of farmer was for a businessman. Malaysian Chinese only expressed a preference for a doctor over the baseline category. In the candidate age attribute, respondents from both ethnic groups showed a preference for progressively younger candidates. Finally, for the candidate gender attribute, respondents from both ethnic groups were statistically indifferent between male and female candidates. As a whole, these results show that a candidate’s investment stance can affect voter preference. Voters showed a consistent preference for candidates who supported increased investments from Japan, compared to candidates who supported either increased investments from China or the United States or were opposed to 159 increased foreign investments in general. This relationship is discernable in the model run with pooled respondent data as well as when the sample is segmented by respondent ethnicity. Even after controlling for a host of established determinants of vote choice, a candidate’s investment position still has a significant effect on vote choice. However, the effect of candidate investment stance on respondent preference contends with a number of competing factors on vote choice. As noted above, respondents were approximately 3-5% more likely to select a candidate profile that supported greater investment from Japan compared to a profile that was opposed to increased foreign investment. This was the largest effect detected, and even it is surpassed by the effects of some other attributes when tested against their baseline attribute levels. A possible interpretation of these results is that while a candidate’s investment position can impact vote choice, the effect is limited to only certain types of investments. And when investment position is arrayed against several other common determinants of vote choice, other factors simultaneously play a role in voters’ decisions. Other attributes that may be considered more “classical” determinants of vote choice such as partisan identity, ethnic identity, history of corruption, and tax policy also have a strong effect on respondent vote choice. A candidate’s investment position can affect vote choice, but campaigns will likely use it as part of a suite of appeals to voters. 160 Conclusion The goal of this chapter was to establish the degree to which voters consider a candidate’s position on foreign investments when casting their votes. This topic sheds light on how foreign investments such as the BRI fit into the grand scheme of host country politics. Importantly, the conjoint design allowed for a comparison of this effect with multiple other common drivers of vote choice. Evidence from this chapter’s conjoint experiment establishes that a candidate’s position on foreign investments has the capacity to affect vote choice. On average, respondents expressed a moderate, but statistically significant, preference for candidates in support of Japanese investments compared to investments from the United States or China. This relationship is present among respondents from both the Bumiputera and Malaysian Chinese ethnic groups. Even when accounting for other candidate attributes that are often influential on vote choice, such as partisan identity, candidate ethnicity, and corruption, a candidate’s investment stance has a discernable effect on voter preferences for candidates. While Chapter 4 showed that ethnicity of both the investor and respondent had a strong effect on how individuals formed preferences on investments, the analysis in this chapter suggests that voters also take into account candidate position on investments when voting, though the effect is exerted alongside other determinants of vote choice. 161 Chapter 6: Conclusion Broadly speaking, this dissertation’s goal is to examine the effects of international investments on host country domestic politics. Significant motivation for pursuing this topic comes from the arrival of the BRI, and with it, a massive infusion of Chinese capital and influence in Southeast Asia. Countries in this region manage complex domestic ethnic relationships, and the injection of Chinese influence via BRI into these delicate systems has the potential to alter the conduct of domestic politics. In many ways, the effects of international factors on domestic politics remains an understudied area of political science. Admittedly, the scale of such effects and the many forms they can take often make studies of this topic unwieldy. As such, I restricted this project’s scope to an examination of this process at work in a single country, Malaysia, and with a single type of international factor in international investments. Obviously, this kind of research strategy carries both analytical benefits and costs. A main benefit of this approach is that it narrows the scope of possible effects into a country and issue context that are more tractable. My goal is largely to identify the existence of the effect on public opinion, its direction, and its magnitude. In this pursuit, focusing on a single country case and on a single issue is helpful. Even so, tracing the contours of the theoretical mechanisms and justifying my argument in the context of Malaysia and the BRI is an intensive task. A drawback of this type of approach is in its limited ability to speak to the generalizability of the proposed theoretical 162 relationships to travel to contexts outside of Malaysia and beyond the specific effects of the BRI. Further discussions of the implications of the research design are made below. What can this dissertation say about the effects of the BRI on host country politics? The BRI serves many purposes for China, from providing new economic outlets for its industrial capacity, to improving Beijing’s geopolitical leverage with local partners. But as Rolland states, “BRI is likely to provoke counteractions and will undoubtedly create unexpected consequences too” (Rolland 2017, 178). Although China has anticipated potential pushback against the BRI and has consciously tried to present the BRI as a win-win situation with its partners, China likely did not foresee the potential for the BRI to be an issue that aggravated host country politics along ethnic lines. This study shows that the BRI can indeed have a strong effect on host country ethnic politics. Even though on its face the BRI is an economic issue, it increases the presence and the influence of China in its host countries (Lew and Roughead 2021). The evidence presented in Chapter 3 shows that this aspect of the BRI has been seized upon by host country ethnic political entrepreneurs to transform the BRI from an economic issue into an ethnicized one. The results from the investment projects conjoint experiment in Chapter 4 reveal that voters are receptive to ethnic appeals surrounding the BRI. Voters’ preferences for projects are clearly driven in part by the ethnic and national identity of the investor. The manner of their preferences is 163 also determined by the respondents’ own ethnic identity. Respondent bias in favor of coethnic investors is strong for members of both the Bumiputera and Malaysian Chinese ethnic groups. These findings establish an ethnic effect of BRI investments on host country politics that reflects and reinforces the existing structure of ethnic competition, as the investments are manipulated into an ethnic issue and affect the preferences of voters along these lines. When the outcome of interest is turned to the respondent’s vote choice, and the candidate’s position on foreign investments is made known to respondents, experimental results presented in Chapter 5 show that candidate investment position significantly affects vote choice. Respondents showed a preference for candidates who supported Japanese investments compared to investments from other sources, supporting the idea that a candidate’s investment position matters in the voting booth. This result was observed even after controlling for multiple factors that are commonly thought to affect voters’ preferences for candidates. However, whether the effect of the candidate’s investment position is conditioned by respondent ethnicity seems to be less clear. Respondents from both the Bumiputera and Malaysian Chinese groups expressed similar preferences for the candidate’s investment position. Taken with the results from Chapters 3 and 4, this suggests that while BRI may be a political issue that resonates with voters along ethnic pathways, in the full constellation of Malaysian politics a candidate’s investment position is just one issue among many, and its effects on vote choice compete with other indicators of candidate performance, such as the candidate’s own 164 ethnic or partisan identities. BRI may yet stand as a powerful tool for ethnic mobilization, but its ultimate effect on vote choice may be limited or situational. Nonetheless, this study provides strong evidence for the interaction of international factors and domestic ethnic politics and provides valuable insights for both the second image reversed and comparative ethnic politics literatures. International factors such as investment programs should be considered alongside other established drivers of ethnic salience, particularly in their use as tools of strategic politicians in mobilizing ethnic support. BRI investments can be used as an issue to stir up ethnic grievances or to capitalize on ethnic ties to the investors to support the investments, thereby reinforcing the role of ethnic politics in host countries. Limitations of the study At the beginning of this chapter, I mentioned that several analytical tradeoffs were necessary in the implementation of this research design. While the research strategy that was employed allowed me to identify the effect of ethnicity on public opinion of investment projects and voting patterns, it is subject to several important limitations as to the extent that it can justify its causal claims. The first limitation is in its ability to speak to cases outside of Malaysia. All of the respondents in my survey were Malaysian citizens, and the majority of the analysis focused on Malaysian politics and investment projects. Naturally, this raises 165 questions on the generalizability of my results to cases outside of Malaysia. In what country-issue contexts can I expect to see the results found in the Malaysia-BRI case? Some guidance to answer this question may be found by referring to the scope conditions outlined in Chapter 2. Countries that have high levels of ethnic salience in politics and international issues that can credibly be tied to the host country’s ethnic competition create situations where the ethnicization of the international issue is most likely. The BRI has been turned into an ethnic issue in Malaysia because its growth has increased Chinese influence in Malaysia, where issues of power between the ethnic Bumiputera and Malaysian Chinese define ethnic competition. This type of situation could feasibly be found in other BRI recipient countries. Several Southeast Asian countries are home to a Chinese diaspora and are also BRI host countries. Further exploration into whether the relationships identified in this study generalize to the experiences of countries such as Indonesia or Thailand would provide an interesting area for future research. Political entrepreneurs in these countries could see the opportunity provided by the BRI to pursue mobilization of ethnic support groups, and the Chinese origins of the BRI could provide the basis for credible connections to these countries’ domestic politics. Another limitation of this study is that it only examines ethnicization around investments in the BRI. While the BRI’s potential impacts on Chinese development, regional development, and geopolitical balance of power make it a worthy topic to focus on, 166 valid questions can be asked whether similar effects would be plausible for international investments from other sources. While the Malaysia-BRI context describes Chinese influence having an effect on a domestic context with a Chinese diaspora, there is little reason not to expect similar relationships to hold in other contexts. For example, further research could explore if investments from Russia are similarly ethnicized in former Soviet republics that have a significant Russian diaspora. The source of investment does not necessarily have to be a global power, but could be a regional center. Turkish investments into some of its neighbors could represent an additional context where investments could become ethnicized in host countries. If these countries commonly experience ethnic political competition, increasing influence via investments could see a similar process unfold as has been documented in Malaysia with the BRI. Finally, the claims of this study should be tempered by limitations in sampling and data collection. This study derives its conclusions from a sample of 1,308 Malaysian citizens, and although multiple steps were taken to try to recruit as high-quality and representative of a sample as possible (described in detail in Chapter 1), sampling was constrained by funding and the social-distancing necessities of COVID. As a result, I elected to use a cost-friendly online sampling approach. While this type of sampling is inferior to the gold standard of random sampling, I attempted to anticipate the main ways that my online sample would systematically differ from the Malaysian population, and took steps to mitigate sources of bias. Nonetheless, there could be additional and unaccounted for sources of 167 sampling bias, and any conclusions drawn from this study remain subject to this caveat. What is the outlook for the BRI in the future? Almost a decade removed from Xi Jinping’s announcement of the BRI, the Initiative looks set to continue to be a centerpiece of Chinese foreign policy (Rolland 2019). The underlying economic and geopolitical factors that led to its launch remain, and China will likely continue to be a major source of international investment in Southeast Asia and the BRI’s other regions of focus (Lew and Roughead 2021). In its short period of existence, however, the BRI has encountered significant headwinds (Minter 2018, Hurley et al. 2018, Hilton 2019), and this study suggests that in some contexts the BRI will contend with the forces of ethnic politics in host countries. This is a potentially serious obstacle, because the ability of ethnicity to arouse powerful emotional responses, in addition to its ability to mobilize support around the material stakes involved, presents risks to the continued growth and implementation of BRI projects. The Malaysian example has shown how the BRI has already been ethnicized as a way of mobilizing the ethnic Malay vote. Mahathir Mohamad pursued this strategy, and later cancelled or renegotiated several major BRI projects (Grassi 2020). The BRI has been ethnicized in Malaysia, and in turn domestic factors in Malaysia have affected its foreign policy (Waltz 1959). The BRI is at risk of becoming the foil to ethnic entrepreneurs in the future in Malaysia or 168 in other host country contexts, which threatens its ability to fulfill the goals set out for it by the Chinese leadership. However, Beijing can take steps to counter the risk of ethnic framing. As Rolland (2019) notes, Beijing has anticipated criticism of the BRI from “reluctant recipient countries,” and possesses some tools to counter pushback to the BRI. Possible steps could include reorienting propaganda around the BRI to explicitly note its benefits for ethnic groups that have opposed the Initiative or changing some of its practices to counter perceptions of ethnically based grievances. For example, construction sites could increase their use of local labor and materials, real estate developments could increase marketing or offer incentives to previously excluded groups, or the BRI could increase the presence of its capacity-building projects like power plants or transportation infrastructure in areas where aggrieved ethnic groups are concentrated. The BRI remains a much-needed source of funding in its partner countries, and Chinese officials are likely to find partners moving forward despite the headwinds it has faced. After Mahathir won power, for example, he showed a willingness to renegotiate the terms of controversial BRI projects, a softer line than he took in his campaigning (Grassi 2020). In sum, the relationship between BRI projects and ethnic politics highlighted in this study present a potentially significant obstacle for BRI implementation in the future, but likely not an insurmountable one. 169 Contributions and areas for further research This study makes contributions to several bodies of literature. First, it expands the scope of the second image reversed literature to include noneconomic effects on domestic politics. There are multiple pathways through which international factors can affect domestic politics, and its impact on ethnic politics has been understudied. Comparative politics research sometimes fails to properly account for how international factors condition domestic behaviors and policies. The results of this study suggest that researchers would do well to be mindful of international effects, particularly those beyond the economic effects that usually are the subject of this literature. The ethnic politics literature can also benefit from this study’s findings. I have identified a clear effect of both respondent ethnicity and investor origin on citizens’ preferences for investment projects. This shows that although investment projects are at their core an economic issue, under the right conditions they can be transformed by rational and strategic politicians into ethnic issues, and that voters are receptive to the ethnicization of investment projects. Voters are also sensitive to the origin of investments and to candidate ethnicity when they vote. This shows a flexibility of ethnic political entrepreneurs to present an international economic issue into an ethnic framework and suggests that issues such as the BRI can become reinforcing agents of ethnic politics. Additionally, the BRI literature stands to benefit from the collection of original data and from the experiments performed in this study. This study represents a more methodologically rigorous 170 examination of public opinion of investment projects such as the BRI, and its findings stand on firmer ground than most previous studies. Substantively, it calls attention to the ethnicization of BRI projects in Malaysia, which is not a topic that has received attention to date. As seen in this study, however, the effects of ethnicity on respondents’ preferences regarding projects and on their votes can be significant and warrant further consideration. The results and analysis of this study also suggest several fruitful avenues of future research. Beyond exploring additional contexts in which international issues can be ethnicized, two main areas of future research should explore changes over time in ethnic salience and delve deeper into candidate and voter behavior. This study only uses data from a single point in time and restricts its analysis to establishing the presence of ethnicized preferences of investment projects and candidates. It is not equipped to make strong claims as to whether the use of ethnic appeals around investments has increased or decreased the respondents’ willingness to resort to ethnic cues in forming their preferences. Future studies can be designed to test specific mechanisms of the process of ethnicization of investment projects in the minds of voters and changes in ethnic salience as a result of ethnic priming regarding investments. The BRI is still a new factor in Malaysian politics, and over time, public opinion on the investments could shift as citizens become more knowledgeable about the effects of BRI projects on their lives. This may decrease their reliance on ethnicity as a cue in the face of low information. On the other hand, if the BRI continues to be used as an 171 ethnic tool by politicians, these ideas could crystallize in voters’ minds. Alternatively, as the issue becomes less novel over time, it is possible that it could fade from voters’ minds. Further research is necessary in order to track these trends over time. Finally, this study is primarily focused on public opinion regarding investment projects, and it offers evidence for how the public views these projects in an ethnic light and the extent to which the issue affects votes. In this process, candidates play a major role as a primary source of information to the public as to the projects’ effects and as strategic actors looking to mobilize winning coalitions for office. There is clearly diversity among candidates in how likely they are to ethnicize investment projects and the manner in which they do so. Not all candidates pursue a strategy of ethnic mobilization when discussing investment projects. Among those who do, some candidates have attacked the BRI, while others have attempted to promote its benefits to their ethnic constituencies. At least two factors likely influence candidate behavior in this regard. One is candidates’ ethnic identities, which influence the ethnic constituencies that they appeal to and the types of ethnic arguments that are likely to resonate with voters. For example, a candidate who is appealing to a Malay base may be more likely to take a more critical stance of BRI investments, while a candidate appealing to a Malaysian Chinese base would not face the same set of incentives. This effect is bound to be conditioned by other factors, however, such as partisan identity. Electoral coalitions generally have representatives from multiple ethnic groups, meaning that members of 172 each ethnic group are part of the government or opposition at any time. This gives competing incentives on how candidates will craft their positions on the BRI. While a closer examination of candidate behavior was beyond the scope of this study, parsing out the mechanisms that contribute to public opinion of BRI investments presents a promising area of study for future work. Research into these questions can build upon the foundations set by this study and the evidence it has provided in support of the ethnicization of the BRI in Malaysia. 173 BIBLIOGRAPHY Abdullah, Suhaila. 2013. “Effect of Malay-China Trade Relations During the Malacca Sultanate on the Emergence of Chinese Peranakan Community.” World Journal of Islamic History and Civilization 3(4): 143-149. Adida, Claire, Jessica Gottlieb, Eric Kramon, and Gwyneth McClendon. 2017. “Overcoming or Reinforcing Coethnic Preferences? An Experiment on Information and Ethnic Voting”. Quarterly Journal of Political Science 12(4): 437-477. Akerloff, George A. and Rachel E. Kranton. 2000. “Economics and Identity.” The Quarterly Journal of Economics 3: 715-753. Albertson, Bethany L. 2015. “Dog-Whistle Politics: Multivocal Communication and Religious Appeals.” Political Behavior 37(3): 3- 26. Alesina, Alberto, Reza Baqir and William Easterly. 1999. “Public Goods and Ethnic Divisions.” The Quarterly Journal of Economics 114(4): 1243-1284. Alon, Ilan, Wenxian Zheng, and Christoph Latteman. 2018. “Introduction”. In China’s Belt and Road Initiative: Changing the Rules of Globalization. Palgrave Macmillan: Cham. Amnesty International. 2020. “Attack on students’ freedom of expression must end.” Press Statement. https://www.amnesty.my/2020/11/07/attack-on-students-freedom-of- expression-must-end/. Accessed 12 June 2022. Asian Barometer. 2014. Survey Wave 4 Malaysia. Retrieved from http://www.asianbarometer.org/survey/wave4-malaysia. Asian Development Bank (ADB). 2017. Meeting Asia’s Infrastructure Needs. Manila. © ADB. Baginda, Abdul Razak. 2016. China-Malaysia Relations and Foreign Policy. New York: Routledge. Baker, Jim. 2008. Crossroads: A Popular History of Malaysia and Singapore. Marshall Cavendish International (Asia): Singapore. Balding, Christopher. 2018. “Why Democracies are Turning Against Belt and Road: Corruption, Debt, and Backlash.” Foreign Affairs. Bansak, Kirk, Jens Hainmueller, Daniel J. Hopkins, and Teppei Yamamoto. 2018. “The Number of Choice Tasks and Survey Satisficing in Conjoint Experiments”. Political Analysis 26: 112-119. 174 Barr, Michael D. 1997. “Lee Kuan Yew in Malaysia: A Reappraisal of Lee Kuan Yew’s Role in the Separation of Singapore from Malaysia.” Asian Studies Review 21(1): 1-17. Bassi, Anna, Rebecca B. Morton, and Kenneth C. Williams. 2011. “The Effects of Identities, Incentives, and Information on Voting.” The Journal of Politics 73(2): 558-571. Bates, Robert. 1974. “Ethnic Competition and Modernization in Contemporary Africa.” Comparative Political Studies 6(4): 457-484. Bates, Robert H. 1983. “Modernization, Ethnic Competition, and the Rationality of Politics in Contemporary Africa.” In State versus Ethnic Claims: African Policy Dilemmas, ed. Donald Rothchild and Victor A. Olunsorola. Boulder, CO: Westview Press. Beech, Hannah. “’We Cannot Afford This’: Malaysia Pushes Back Against China’s Vision.” The New York Times, 20 Aug 2018, www.nytimes.com/2018/08/20/world/asia/china-malaysia.html. Accessed 8 Jan 2019. Berinsky, Adam J., Gregory A. Juber, and Gabriel S. Lenz. “Evaluating Online Labor Markets for Experimental Research: Amazon.com’s Mechanical Turk”. Political Analysis 20: 351-368. Birnir, Johanna K. 2006. Ethnicity and Electoral Politics. Cambridge University Press. Bisserbe, Noemie, Aruna Viswanatha and Bradley Hope. 2020. “1MDB’s Suspected Mastermind Found New Ways to Move Money in Kuwait.” The Wall Street Journal. Brader, Ted. 2005. “Striking a Responsive Chord: How Political Ads Motivate and Persuade Voters by Appealing to Emotions.” American Journal of Political Science 49(2): 388-405. Brady, Henry E. and Paul M. Sniderman. 1985. “Attitude Attribution: A Group Basis for Political Reasoning.” The American Political Science Review 79(4): 1061-1078. Chandra, Kanchan. 2004. Why Ethnic Parties Succeed: Patronage and Head Counts in India. Cambridge: Cambridge University Press. Chandra, Kanchan. 2006. “What is Ethnicity and Does it Matter?” Annual Review of Political Science 9: 397-424. Chin, Yee-Mun. 2020. “Assimilation and Chinese Ethnic Consciousness in Malaysia: An Overview.” Malaysian Journal of Chinese Studies 9(1): 15-22. 175 Conroy-Krutz, Jeffrey. 2012. “Information and Ethnic Politics in Africa”. British Journal of Political Science 43: 345-373. Coppock, Alexander and Oliver A. McClellan. 2019. “Validating the Demographic, Political, Psychological, and Experimental Results Obtained from a New Source of Online Survey Respondents”. Research and Politics January-March: 1-14. Coval, Joshua and Tobias J. Moskowitz. 1999. “Home Bias at Home: Local Equity Preference in Domestic Portfolios.” The Journal of Finance 54(6): 2045-2073. Daniels, Timothy P. Building Cultural Nationalism in Malaysia: Identity, Representation and Citizenship. Routledge. deFiguerido, Rul J.P. and Barry R. Weingast. 1999. “Rationality of Fear: Political Opportunism and Ethnic Conflict.” In Civil Wars, Insecurity, and Intervention, eds. Barbara F. Walter and Jack Snyder. Columbia University Press: New York. Dickson, Eric S. and Kenneth Scheve. 2006. “Social Identity, Political Speech, and Electoral Competition.” Journal of Theoretical Politics 18(1): 5-39. Doherty, David, Conor M. Dowling, and Michael G. Miller. 2019. “Do Local Party Chairs Think Women and Minority Candidates Can Win? Evidence from a Conjoint Experiment”. The Journal of Politics 81(4). Doig, Will. 2019. “The Belt and Road Initiative is a Corruption Bonanza.” Foreign Policy, 15 January 2019. https://foreignpolicy.com/2019/01/15/the-belt-and-road-initiative- is-a-corruption-bonanza/. Accessed 1 August 2022. Edwards, Scott. 2018. “Malaysia’s Elections: Corruption, Foreign Money, and Burying-the-Hatchet Politics.” Al Jazeera Centre for Studies. Fallon, Theresa. 2015. “The New Silk Road: Xi Jinping’s Grand Strategy for Eurasia.” American Foreign Policy Interests 37: 140-147. Fearon, James D. 1996. “Explaining Interethnic Cooperation.” American Political Science Review 90: 715-735. Fearon, James D. 1998. “Commitment Problems and the Spread of Ethnic Conflict.” In The International Spread of Ethnic Conflict, ed. David A. Lake and Donald Rothchild. Princeton, NJ: Princeton University Press pp. 107-126. Fearon, James D. 2006. “Ethnic Mobilization and Ethnic Violence”. In Oxford Handbook of Political Economy, Barry R. Weingast and Donald Wittman, eds. Oxford: Oxford University Press. 176 Feinstein, Yuval. 2016. “Rallying Around the President: When and Why do Americans Close Ranks Behind Their Presidents During International Crisis and War?” Social Science History 40(2): 305- 338. Ferree, Karen E. 2006. “Explaining South Africa’s Racial Census.” The Journal of Politics 68(4): 803-815. Fitzgerald, C.P. 1965. “The Expulsion of Singapore.” Nation 201(11): 208-212. Galesic, Mirta and Michael Bosnjak. 2009. “Effects of Questionnaire Length on Participation and Indicators of Response Quality in a Web Survey.” Public Opinion Quarterly 73(2): 349-360. Garrett, Geoffrey and Peter Lange. 1995. “Internationalization, Institutions, and Political Change. International Organization 49(4): 627-655. Gomez, Edmund Terence. 2006. “Chinese Networks and Enterprise Development: Malaysian Investments in China.” Journal of Contemporary Asia 36(3): 350-363. Gomez, Edmund Terence. 2012. “Monetizing Politics: Financing Parties and Elections in Malaysia.” Modern Asian Studies 46(5): 1370-1397. Gomez, Edmund Terence. Interview conducted by Alexander Mardon, 20 January, 2020. Gourevitch, Peter. 1978. “The Second Image Reversed: The International Sources of Domestic Politics.” International Organization 32(4): 881-912. Grassi, Sergio. 2020. The Belt and Road Initiative in Malaysia: China’s Geopolitics and Geoeconomics Challenged by Democratic Transformation. Friedrich-Ebert-Stiftung. Gutierrez-Romero, Roxana and Adrienne LeBas. 2020. “Does Electoral Violence Affect Vote Choice and Willingness to Vote? Conjoint Analysis of a Vignette Experiment”. Journal of Peace Research 57(1): 77-92. Habyarimana, James, Macartan Humphreys, Daniel N. Posner, and Jeremy M. Weinstein. 2007. “Why Does Ethnic Diversity Undermine Public Goods Provision?” American Political Science Review 101(4): 709-725. Hainmueller, Jens, Daniel J. Hopkins and Teppei Yamamoto. 2014. “Causal Inference in Conjoint Analysis: Understanding Multidimensional Choices via Stated Preference Experiments”. Political Analysis: 1-30. 177 Hallerberg, Mark and Scott Basinger. 1998. “Internationalization and Changes in Tax Policy in OECD Countries: The Importance of Domestic Veto Players.” Comparative Political Studies 31(3): 321-352. Hamzah, Hanim. 2018. “Legal Issues and Implications of the BRI”. In China’s Belt and Road Initiative (BRI) and Southeast Asia. CIMB ASEAN Research Institute. Hilton, Isabel. 2019. “How China’s Big Overseas Initiative Threatens Global Climate Progress.” Yale Environment 360. Holst, Frederik. 2012. Ethnicization and Identity Construction in Malaysia. Routledge: New York. Huff, Connor and Joshua D. Kertzer. 2017. “How the Public Defines Terrorism”. American Journal of Political Science 62(1): 55-71. Hurley, John, Scott Morris, and Gailyn Portelance. 2018. “Examining the Debt Implications of the Belt and Road Initiative from a Policy Perspective.” CGD Policy Paper. Washington, DC: Center for Global Development. https://www.cgdev.org/publication/examining-debt- implication-belt-and-road-initiative-policy-perspective. Husain, Wan Ahmad Fauzi Wan, Od. M. Anwar, Zulyati Zakaria, Intan Maizura Abd Rashid, Najuwa Mohd Nasir, and Irza Hanie Abu Samah. 2021. “The Malay Sultanates as the Impetus for the Formation of Malaysia.” Psychology and Education 58(2): 5820-5827. Jaipragas, Bhavan. “Mahathir versus the Sultan: How Chinese Investment Could Sway Malaysian Election.” South China Morning Post, 18 January 2017. Johnson, Richard M. and Bryan K. Orme. 1996. “How Many Questions Should You Ask in Choice-Based Conjoint Studies?” Sawtooth Software Research Paper Series. Jomo K.S. 2004. “The New Economic Policy and Interethnic Relations in Malaysia.” Identities, Conflict and Cohesion 7. United Nations Research Institute for Social Development. Kalin, Michael and Nicholas Sambanis. 2018. “How to Think About Social Identity.” Annual Review of Political Science 21: 239-257. Kasara, Kimuli. 2007. “Tax Me if You Can: Ethnic Geography, Democracy, and the Taxation of Agriculture in Africa”. American Journal of Political Science 101(1): 159-172. Kaynak, Erdener and Ali Kara. 2002. “Consumer Perceptions of Foreign Products: An Analysis of Product-Country Images and Ethnocentrism.” European Journal of Marketing 36(7/8): 928-949. 178 Keck, Margaret E. and Kathryn Sikkink. 1998. Activists Beyond Borders: Advocacy Networks in International Politics. Cornell University Press. Kelley, Judith. 2004. “International Actors on the Domestic Scene: Membership Conditionality and Socialization by International Institutions.” International Organization 58: 425-457. Khalid, Muhammad Abdul and Li Yang. 2019. “Income Inequality and Ethnic Cleavages in Malaysia: Evidence from Distributional National Accounts (1984-2014). World Inequality Lab. Khoo Boo Teik. 2004. “Managing Ethnic Relations in Post-Crisis Malaysia and Indonesia: Lessons from the New Economic Policy?” United Nations Research Institute for Social Development. Kong Tuan Yuen. 2016. “China’s 21st Century Maritime Silk Road: Malaysian Perspectives.” In China’s One Belt One Road Initiative, eds. Lim Tai Wei, Chan Hing Lee, Tseng Hui Yi, and Lim Wen Xin. Imperial College Press. Kumar, Kamles and Bhavan Jaipragas. 2018. “The Malaysian election issue no one can afford to ignore.” South China Morning Post. Lee Hwok Aun. 2017. “Fault Lines – and Common Ground – in Malaysia’s Ethnic Relations and Policies”. ISEAS Yusof Ishak Institute 63. Lee Kuan Yew. 2000. The Singapore Story: Memoirs of Lee Kuan Yew. Marshall Cavendish Editions: Singapore. Lemiere, Sophie. 2018. “The Downfall of Malaysia’s Ruling Party.” Journal of Democracy 29(4): 114-128. Lew, Jacob J. and Gary Roughead. 2021. “China’s Belt and Road: Implications for the United States”. Council on Foreign Relations Independent Task Force Report No. 79. Lijphart, Arend. 1969. “Consociational Democracy.” World Politics 21 (2). 207-225. Lim Tai Wei. 2016. “Narratives Related to Zheng He: Explaining the Emergence of Ethnic Chinese Communities Overseas and the Rise of a Regional Trading Network.” In China’s One Belt One Road Initiative, eds. Lim Tai Wei, Chan Hing Lee, Tseng Hui Yi, and Lim Wen Xin. Massa, Massimo and Andrei Simonov. 2006. “Hedging, Familiarity and Portfolio Choice.” The Review of Financial Studies 19(2): 633-685. Mauzy, Diane K. 1978. “Consociationalism and Coalition Politics in Malaysia.” PhD Thesis, University of British Columbia. 179 McDermott, Monika L. 1998. “Race and Gender Cues in Low-Information Elections”. Political Research Quarterly 51(4): 895-918. Milner, Anthony. 2011. The Malays. Wiley-Blackwell: Oxford. Milner, Helen V. and Robert O. Keohane. 1996. “Internationalization and Domestic Politics: An Introduction.” In Internationalization and Domestic Politics, eds. Robert O. Keohane and Helen V. Milner. Cambridge: Cambridge University Press. Minter, Adam. “Malaysia is Fed Up with Chinese Cash.” The Korea Herald, 9 May 2018. Nair, Sheila. 1999. “Colonial ‘Others’ and Nationalist Politics in Malaysia”. Akademika 54: 55-79. Nederhof, Anton J. 1985. “Methods of Coping with Social Desirability Bias: A Review.” European Journal of Social Psychology 15(3): 263- 280. Ngeow Chow-Bing and Tan Chee Beng. 2018. “Cultural ties and states’ interests: Malaysian Chinese and China’s rise”. In China’s Rise and the Chinese Overseas. Routledge: New York. Noh, Abdillah. 2014. “Malaysia 13th General Election: A Short Note on Malaysia’s Continuing Battle with Ethnic Politics.” Electoral Studies 34: 266-269. Noor, Noraini M. 2009. “The Future of Malay-Chinese Relations in Malaysia.” Peace Psychology in Asia. Springer. OECD. 2018. “China’s Belt and Road Initiative in the Global Trade, Investment and Finance Landscape.” Ong Kian Ming. Interview. Conducted by Alexander Mardon, 13 January 2020. Ono, Yoshikuni and Masahiro Yamada. 2020. “Do voters prefer gender stereotypic candidates? Evidence from a conjoint survey experiment in Japan.” Political Science Research and Methods 8: 477-492. Osborne, Milton. 2010. Southeast Asia: An Introductory History. Australia: Allen & Unwin. Page, Jeremy and Saeed Shah. 2018. “China’s Global Building Spree Runs Into Trouble in Pakistan.” The Wall Street Journal, 22 July 2018. https://www.wsj.com/articles/chinas-global-building-spree-runs-into- trouble-in-pakistan-1532280460. Accessed 1 August 2022. Posner, Daniel. 2005. Institutions and Ethnic Politics in Africa. Cambridge University Press: New York. 180 Rao, Vithala R. 2014. Applied Conjoint Analysis. Springer Heidelberg: New York. Rapanyane, Makhura B. and Kgothatso B. Shai. 2020. “An Afrocentric Dissection of the Key Drivers of China’s Zambia Policy.” Journal of African Foreign Affairs 7(1): 7-22. Rodrik, Dani. 1999. “The Asian Financial Crisis and the Virtues of Democracy.” Challenge 42(4). Rolland, Nadege. 2017. China’s Eurasian Century? Political and Strategic Implications of the Belt and Road Initiative. The National Bureau of Asian Research. Rolland, Nadege. 2022. “China’s Southern Strategy: Beijing is Using the Global South to Constrain America.” Foreign Affairs. Sadka, Emma. 1962. “Singapore and the Federation: Problems of Merger.” Asian Survey 1(11). 17-25. Scissors, Derek. 2018. “Private Data, Not Private Firms: The Real Issues in Chinese Investment.” American Enterprise Institute. SERC. China’s Investment in Malaysia: Perceptions, Issues and Prescriptions. Socio-Economic Research Centre (SERC), October 2017. Shayo, Moses. 2009. “A Model of Social Identity with an Application to Political Economy: Nation, Class, and Redistribution.” American Political Science Review 103: 147-174. Shieh, Shawn, Lowell Chow, Zhong Huang, and Jinfei Yue. 2021. “Understanding and Mitigating Social Risks to Sustainable Development in China’s BRI: Evidence from Nepal and Zambia.” ODI Report. London: ODI. Shullman, David O. 2019. “Protect the Party: China’s growing influence in the developing world”. Brookings. Sin Yee Koh. 2015. “How and Why Race Matters: Malaysian-Chinese Transnational Migrants Interpreting and Practising Bumiputera- differentiated Citizenship.” Journal of Ethnic and Migration Studies 41(3): 531-550. Solingen, Etel. 2001. “Mapping Internationalization: Domestic and Regional Impacts.” International Studies Quarterly: 45: 517-555. Stokes, Jacob. 2015. “China’s Road Rules: Beijing Looks West Toward Eurasian Integration”. Foreign Affairs. Tan Chee_Beng. 1997. “Chinese Identities in Malaysia.” Southeast Asian Journal of Social Science 25(2): 103-116. 181 Time Magazine. 27 July 1962. “Malaysia: Merger is a Must.” https://content.time.com/time/subscriber/article/0,33009,827471,00.h tml. Accessed 12 June 2022. Todd, Laurence and Meghan Slattery. 2018. “Impacts of Investment from China in Malaysia on the Local Economy.” Policy Ideas 54. Toh Boon Kwan. 1999. “’Same Bed, Different Dreams’: Singapore’s Merger with the Federation of Malaya and the Malaysia Agreement, November 1962-September 1963.” The History Journal 1999. 37-50. Waltz, Kenneth. 1959. Man, the State, and War: A Theoretical Analysis. Columbia University press: New York. Waltz, Kenneth. 1979. Theory of International Politics. Reading, MA: Addison-Wesley. Welsh, Bridget. 2013. “Malaysia’s Elections: A Step Backward.” Journal of Democracy 24(4): 136-150. World Wildlife Fund. 2017. “The Belt and Road Initiative: WWF Recommendations and Spatial Analysis.” Wright, Tom and Bradley Hope. “WSJ Investigation: China Offered to Bail Out Troubled Malaysian Fund in Return for Deals.” The Wall Street Journal, 7 January 2019, www.wsj.com/articles/how-china- flexes-its-politcal-muscle-to-expand-power-overseas- 11546890449?mod=hp_lead_pos5. Accessed 8 January 2019. Yean, Tham Siew. 2018. “Chinese Investment in Malaysia: Five Years into the BRI.” ISEAS Yusof Ishak Institute 11. Yeoh, Emile K.K. 2008. “Beyond Reductionism: State, Ethnicity and Public Policy in Plural Societies.” In The State, Development and Identity in Multi-Ethnic Societies: Ethnicity, Equity and the Nation, eds. Nicholas Tarling and Edmund Terence Gomez. Routledge: Oxford, London and New York: 57-96. Yu, Hong. 2017. “China’s Belt and Road Initiative and its Implications for Southeast Asia.” Asia Policy 24: 117-122. Zack, Justin. “Cops record statements of UM duo over FB post on King”. The Star., 5 Nov. 2020, https://www.thestar.com.my/news/nation/2020/11/05/cops-record- statements-of-um-duo-over-fb-post-on-king. Accessed 13 June 2022. 182