Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Volatile nationalism: nationalism and its influence on maritime disputes
(USC Thesis Other)
Volatile nationalism: nationalism and its influence on maritime disputes
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Volatile Nationalism | Xinru Ma Volatile Nationalism: Nationalism and Its Influence on Maritime Disputes Xinru Ma Dissertation Committee: David C. Kang (Chair), Gerardo L. Munck, Pablo Barber´ a, Andrew J. Coe, Stephan Haggard Doctor of Philosophy (POLITICAL SCIENCE AND INTERNATIONAL RELATIONS) UNIVERSITY OF SOUTHERN CALIFORNIA FACULTY OF THE USC GRADUATE SCHOOL August 2019 Volatile Nationalism | Xinru Ma Contents 1 Introduction 1 1.1 The Puzzle and The Argument ......................... 1 1.2 Context Within Literature and Limitations................... 22 1.3 Plan of the Dissertation ............................. 24 1.4 Appendix ..................................... 26 2 Long-term Trends versus Short-term Noises: Measure Nationalism with Social Media Data and Evidences from Viet- nam 31 2.1 The Myth of Rising Nationalism......................... 31 2.2 A Two-dimensional Conceptualization of Nationalism ............. 36 2.3 Measuring Nationalism with Facebook Data in Vietnam............ 45 2.4 Low Baseline but High Volatility Levels of Nationalism in Vietnam ..... 56 2.5 Discussion ..................................... 67 2.6 Appendix ..................................... 69 3 The Calm Philippines: Nationalism with Low Baseline and Low Volatility 73 3.1 The July 2016 Tribunal Ruling on the South China Sea............ 76 3.2 Case Selection and Data Collection ....................... 77 3.3 Training Machine Learning Classifiers...................... 78 3.4 Low Baseline and Low Volatility Levels of Nationalism in the Philippines .. 81 3.5 Discussion ..................................... 83 4 Playing with Firecrackers: When and Why Nationalist Protests at Home A↵ect Crisis Bargaining Abroad 87 4.1 Crisis Bargaining, Signaling, and The Management of Nationalist Protests as Information .................................... 94 4.2 A Unified Theory of Managing Protests as A Signal .............. 100 4.3 A Formal Model ................................. 108 4.4 Discussion and Future Extensions ........................ 117 4.5 Empirical Illustration............................... 121 4.6 Proofs of Propositions .............................. 129 5 Conclusions 137 5.1 Appendix ..................................... 144 Volatile Nationalism | Xinru Ma 1 Introduction “...an upwelling of nationalism...showed how fragile reconciliation is, even among nominal friends, decades after the fighting ended with Japan’s sur- render to US and Allied forces...ultra-nationalism among the leading local contestants in north-east Asia - China, Japan and South Korea - is gaining strength, raising the risk of armed conflict.” -The Sydney Morning Herald 1 “Vietnam’sworstflare-upofanti-Chinesesentimentinyearscouldstrainal- ready troubled relations with the country’s powerful neighbour.” - The South China Morning Post 2 1.1 The Puzzle and The Argument In East and Southeast Asia, presumably due to countries’ contested maritime claims, a textbook story is that nationalism is upwelling in the region - China, Japan, South Ko- rea, Vietnam, and the Philippines – and is increasingly raising the risk of armed conflicts (Hollihan, 2014; Reilly, 2011; Glosny et al., 2010; Gries, 2004; Christensen, 1999). Indeed, a cursory glance at newspaper coverage seems to underpin the story very well. Headlines such as “Japan, the land of the rising nationalism,” 3 “Caught between empires: rising nationalism in Taiwan, Okinawa and Hong Kong,” 4 and “Vietnamese nationalism 1 “Nationalism and Political Tensions Rising in North-east Asia.” The Sydney Morning Herald. August 25, 2012. https://www.smh.com.au/politics/federal/ nationalism-and-political-tensions-rising-in-north-east-asia-20120824-24rpl.html 2 “Anti-China protests in Vietnam set to aggravate tensions with Beijing.” The South China Morn- ing Post. June 13, 2018. https://www.scmp.com/news/china/diplomacy-defence/article/2150653/ anti-china-protests-vietnam-set-aggravate-tensions 3 “Japan: The land of the rising nationalism.” The Independent. November 5, 2010. https://www. independent.co.uk/news/world/asia/japan-the-land-of-the-rising-nationalism-2125690.html 4 “Caughtbetweenempires: RisingnationalisminTaiwan,OkinawaandHongKong.”UCLAAsiaPacific Center. March 22, 2018. http://www.international.ucla.edu/apc/article/190028 1 Volatile Nationalism | Xinru Ma surgesoverSouthChinaSearuling” 5 permeatenewsonEastAsia. Thisimpressionisfurther reinforced by the media-fueled images of stochastic outbursts of nationalistic protests, with protesters chanting slogans and waving banners on the street. Alas, newspaper headlines are not the only venue for the meme of rising nationalism and its dangerous implications. A Brookings report on U.S.-China-Japan trilateral relations points out that “China’s domestic structural change is characterized by renewed national- ism among its leaders and the general public...leads China to challenge America’s status and role in Asia...[while] rising nationalism has taken hold in one of America’s closest al- lies [Japan].” 6 AmorerecentcongressionaltestimonyemphasizesthatproblemsinChina’s economic development and internal instability could combine with “rising nationalism to motivate Chinese leaders to adopt more confrontational military approaches, particularly concerning unresolved claims in the Yellow, East, and South China Seas.” 7 As Alastair Iain Johnston, Harvard Professor of China in World A↵airs, summarizes, “[the term] ‘rising’ is one of the most common qualifiers used by analysts to modify the term ‘Chinese national- ism.’...It appears in think tank analyses, in statements from the U.S. Deparment of State, in U.S. Congress members’ commentary, and in other government analysis.”(Johnston, 2017, p.7) However, is nationalism really rising? How do we know? How, if at all, can this supposed nationalism drive risky state policies, some of which might even lead to unintentional wars? These questions are vital to scholars and policy-makers, frequently asked, but critically under-researched, in part because they require rigorous conceptualization and transforma- 5 “Vietnamese Nationalism Surges Over South China Sea Ruling.” VOA News. July 19, 2016. https: //www.voanews.com/a/vietnamese-nationalism-surges-south-china-sea-ruling/3426191.html 6 “A mechanism to stabilize U.S.-China-Japan trilateral relations in Asia.” The Brookings Institution. January 2008. https://www.brookings.edu/wp-content/uploads/2016/06/01_east_asia_chu.pdf p. 3, 8, 9. 7 “Testimony before the U.S.-China Economic and Security Review Commission Hearing on China’s Military Modernization and its Implications for the United States.” U.S.-China Economic and Security Review Commission. United States Congress. January 30, 2014. https://www.uscc.gov/sites/default/ files/Andrew%20Erickson_testimony1.30.14.pdf p. 4 2 Volatile Nationalism | Xinru Ma tive measurement tools. Yet more importantly, the conceptualization of nationalism, the measurement, and the e↵ect of nationalism have been examined by the extant literature in- dividually, in isolation. 8 In turn, we lack a unified theoretical framework that identifies key facets that determine the enormous variation in the levels of nationalism cross countries such as China, Vietnam and the Philippines, the conditions under which nationalism escalates the risks of armed conflicts, and the associated mechanisms. In other words, scholarship has so far generated only a fragmented understanding of nationalism and its e↵ect, while a general conceptual heuristic is needed - at both theoretical and empirical level - to facilitate our cross-country comparisons of nationalism and its varying e↵ects on militarized conflicts. This dissertation attempts to fill this void. In a nutshell, this dissertation seeks to o↵er full-spectrum comparative studies of nation- alism by carefully measuring nationalism with latest methods in big data, combined with a rigorous analysis of the e↵ect of nationalism on international crisis with formal modeling. Yet, its contribution is not only methodological, it is conceptual. Thecentralclaimofthisdissertationisthis: Thequestionofwhethernationalismisrising or not is ill-defined. Without an ex ante measure of nationalism, any examination of ex post nationalistic demonstrations biases the results in favor of the “rising-nationalism” argument, which in turn exaggerates the e↵ect of nationalism on militarized conflicts. By contrast, I propose to conceptualize nationalism in two dimensions: 1) “baseline” nationalism, namely, how nationalistic a nation is when there’re no triggers, and 2) “volatility” of nationalism, namely, how a nation responds to nationalist triggers. Some countries have very volatile nationalism: they experience surges of nationalism in response to very small stimuli. Other countries have very involatile nationalism: even large triggers don’t cause a response. Using Vietnam and the Philippines as key cases, I found a low and stable baseline nationalism in both countries, signaling the overestimation of nationalism in extant literature. The formal 8 See subsequent chapters for a more detailed discussion of these separate bodies of literature. 3 Volatile Nationalism | Xinru Ma model, on the other hand, shows that even nationalist protests - one of the most frequently observed and dangerous platforms for expressing nationalism - can only induce conflicts under stringent conditions. More specifically, the e↵ect of nationalism depends largely on one key factor underexamined in extant literature: the uncertainty that nationalist protests evolving spontaneously into anti-regime mobilizations. Rather than increase linearly the risk of armed conflicts, nationalism can surprisingly decrease ex ante probability of militarized conflicts, if and only if this uncertainty is at middle level. Thisdissertationthusprovidesawealthofnewdataandanalysisofthesecuritydynamics of East Asia informed by local knowledge and regional perceptions. Substantively, it chal- lenges the pervasive assumption of “rising nationalism,” a key assumption that has inflated the perception of China threat and the perceived risk of militarized conflicts in East Asia. For instance, Aaron Friedberg sees East Asia as “ripe for rivalry,” (Friedberg, 1993)while Adam Li↵ and John Ikenberry view the region is in the shadow of “racing towards tragedy.” (Li↵ and Ikenberry , 2014) However, if nationalism is not rising in Asia, and if states are not forced by nationalism to contest their claims militarily (e.g. in the South China Sea), then the rhetoric about the risk of armed conflicts and the U.S. obligation to intervene in the region are overstated. Indeed, the region may be more stable than popularly assumed, and that the perception that East Asia is “racing towards tragedy” is wrong. Methodologically, this dissertation employs cutting-edge computational methods in social media research, in- corporatesmicro-levelanalysisofnationalismwiththerichtextdatao↵eredbysocialmedia, and allows for explicit comparison of nationalism in Vietnam and the Philippines. Comparative studies of nationalism are rare, especially those based on careful method- ologies that allow for explicit comparison. To date, extensive studies on nationalism have beenexclusivelydevotedtoChina, whiletheproposedmeasurementsandtheoriesarerooted merely in China’s history or culture (e.g. Downs and Saunders, 1999; Gries, 2004; Weath- erley and Zhang, 2017), thus lacking cross-regional perspectives on nationalism, as well as 4 Volatile Nationalism | Xinru Ma external validity for most, if not all, concerned actors in the region. By comparing national- ism in Vietnam and the Philippines, my research seeks to advance the literature by moving beyond the traditional single-country focus (especially on China). Devising an ex ante measure of nationalism, as well as clearly-specified scope conditions for when and why we would expect to see nationalism drive dangerous policies, also has enduring empirical implications. Empirically, tensions are ratcheting up in the South China Sea, and have sparked a wave of protests in major claimant states, with some turning into brutal violence against foreign nationals and property. Thus, a better understanding of the levels and the e↵ect of nationalism can inform pundits, media, and policy-makers the intended and unintended consequences of nationalism, so as to better devise policies that can mitigate the risk of embroiling states into unintended costly militarized conflicts. In the remaining of this chapter and the chapters that follow, I develop theoretical argu- ments that elaborate on and qualify these claims, explain the connections between chapters thatusebigdataandformalmodelingmethods,andpresentempiricalevidencesthatsupport them. 1.1.1 Nationalism, Foreign Policies, and the Importance of Baseline Measure- ment A typical account of “rising-nationalism” invokes the image of a spontaneously assembled crowd taking to the streets and alleys, throngs of people chanting “Down with China!” 9 -or any foreign country involved in the dispute - while the leaders engage in a desperate attempt to either appease or disperse the assembled masses. Some of these incidents end with the 9 For example, see “Rare rallies in Vietnam say ’hands o↵’ to China over sea row.” Reuters. May 11, 2014. https://www.reuters.com/article/us-vietnam-china/ rare-rallies-in-vietnam-say-hands-off-to-china-over-sea-row-idUSBREA4A02B20140511 For examples of protests not targeting China, see “‘Dump Trump’ protest erupts ahead of Philip- pines summit.” Reuters. November 12, 2017. https://www.reuters.com/article/asean-summit/ dump-trump-protest-erupts-ahead-of-philippines-summit-idINKBN1DC081 In this case, hundreds of protesters marched towrads the U.S. embassy in Manila, chanting slogans such as “Dump Trump” and “Down with U.S. Imperialism.” 5 Volatile Nationalism | Xinru Ma stando↵ between two countries, potentially opening the way for a militarized conflict. Such typical portrayal of nationalism epitomizes the first problem that I identify with the studiesonnationalism: Extantstudiesoftenmeasurenationalismintheimmediateaftermath of certain triggers, thus failing to provide a baseline of how “nationalistic” a nation is when there’re no triggers. Such single-snapshot view of nationalism fails to clarify what exactly one means by “rising-nationalism,” leaving the question of whether nationalism is rising or not ill-defined. The contrast between Ukraine’s and China’s reactions to a similar nationalist trigger is a case in point. In September 2010, the Japanese coast guard detained a Chinese fishing vessel operating in disputed waters near the Diaoyu Islands (Senkaku by the Japanese). Following the detention, massive protests broke out in major Chinese cities, with protesters shouting slogans urging their government to take tougher actions: “Set the captain free! Chinese should stand up and fight! How can we get the Diaoyu Islands back if we continue to act like cowards?” 10 By contrast, in May 2018, Russian border guards detained a Ukrainian fishing vessel and its five crew members in the Black Sea near the Russian-occupied Crimea. 11 In August 2018, Russian border guards detained another Ukrainian fishing boat and its crew of four in the annexed Crimea. 12 However, both detentions received little attention from the Ukrainian public except for one diplomatic note of protest to Russia from the Foreign Ministry of Ukraine. Observing massive protests in China but only one diplomatic note in Ukraine in response to a similar trigger, can we infer that Chinese are more nationalistic than Ukrainian? Put more broadly, without knowing the baseline level of nationalism, how 10 “Boat collision sparks anger, breakdown in China-Japan talks.” The Washington Post. September 20, 2010. http://www.washingtonpost.com/wp-dyn/content/article/2010/09/20/AR2010092000130. html?noredirect=on 11 “Ukraine demands release of fishing vessel detained by Russia near Crimea.” UNIAN. May 4, 2018. https://www.unian.info/politics/ 10105676-ukraine-demands-release-of-fishing-vessel-detained-by-russia-near-crimea.html 12 “Russia detains Ukrainian vessel in the Black Sea.” September 4, 2018. UAWIRE. https://www. uawire.org/russia-detains-a-ukrainian-vessel-in-the-black-sea 6 Volatile Nationalism | Xinru Ma do we know whether countries are becoming more nationalistic over time (within-country comparison)? Why do we say some countries are more nationalistic than others (cross- country comparison)? Arguably, influenced by the stochastic coverage in media, scholars have been using public opinion polls conducted immediately following some nationalistic triggers to gauge the level of nationalism. For instance, to characterize “the clash of nationalisms” between China and Japan, Chan and Bridges (2006)usedopinionpollsthatimmediatelyfollowedKoizumi’svis- iting to Yasukuni shrine in 2005, and those that immediately followed Koizumi’s campaign forJapan’spermanentmembershipintheUnitedNationsSecurityCouncil(UNSC).Inthese cases, nationalism was gauged by the share of respondents that supported the shrine visit (or the level of opposition for measuring nationalism in China), and the share of support (or opposition in the case of China) for Japan’s campaign for the permanent membership in the UNSC. 13 However, by characterizing all pro-shrine-visit arguments as nationalist ar- guments, and lumping all supporters of Japan’s campaign for permanent membership into the nationalist camp, it is impossible to determine the relative importance of nationalism as acausalfactorindependentlyofothergeopoliticalfactorsornationalsecurityreasons. In addition, by focusing on the immediate aftermath of certain triggers, researchers might focus only on “dogs that bark” and ignore “dogs that don’t bark.” The single-event snapshots not only tend to bias the influence of nationalism on foreign policy when such nationalism is expressed, but also underestimate the influence of nationalism in scenarios in which nation- alistic triggers are not followed by observable expression of nationalism (e.g. either through public opinion polls or protests). Finally, while such data is useful for unpacking short-term dynamics of nationalism, the fact that we don’t have opinion polls following each trigger and/or the lack of continuity in questionnaires when countries are faced with similar triggers 13 For more studies that used these opinion polls immediately after certain nationalist triggers, see Grønning (2014), Self (2002), and Mochizuki and Porter (2013). 7 Volatile Nationalism | Xinru Ma across time has limited our understanding of the long-term trend of nationalism. The first goal of this dissertation is to assess carefully the discourse about rising nation- alism by establishing a baseline measurement with new methods in big data. Theoretically, it disaggregates nationalism into two conceptually distinct dimensions: 1) “baseline” nation- alism, namely, how nationalistic a nation is when there’re no triggers, and 2) “volatility” of nationalism, namely, howanationrespondstonationalisttriggers. Somecountrieshavevery volatile nationalism: they experience surges of nationalism in response to very small stimuli. Other countries have very involatile nationalism: even large triggers don’t cause a response. I maintain that it is the varied levels of baseline and volatility that jointly di↵erentiate the natureandlevelsofnationalismindi↵erentcountries, whichfurthercategorizecountriesinto a 2*2 metrics (Figure 1.1.1). Figure 1.1.1: A Two-Dimensional Conceptualization of Nationalism and Its Application Chapter 2 thus uses Vietnam as the key case to elaborate this two-dimensional concep- tualization of nationalism and the methods of measuring them in details. Nevertheless, the definition of nationalism has been hunted, even till today, by “the stultifying aura of concep- tual ambiguity.” (Geertz, 1963, p. 107) By no means of being exhaustive, Appendix 1.4.1 surveys major approaches in the literature to conceptualize nationalism. To be clear, this dissertation doesn’t intend to propose an uncontroversial definition of nationalism, nor does it aim to conduct a meta-analysis of the di↵erent definitions of 8 Volatile Nationalism | Xinru Ma nationalism. Nor does it seek to add another detailed account of the origins, construction, continuity and discontinuity in the components of national identity in any regional country. Instead, it seeks to break the link between nationalism and national history, and o↵ers an alternative way of conceptualizing nationalism that enables us to (1) conduct within- country and cross-country comparison, and (2) specify the scope conditions under which nationalism can influence foreign policies in a context of international crisis. To make the task manageable, definitional relationships between the state,the nation,and nationalism are set aside. In turn, to organize the discussion, the following consequential choices are made. Following Hechter (2000,p. 11),nationisdefinedas“relativelylargegroupofgeneti- cally unrelated people with high solidarity.” Nationalism is defined - following a minimalist approach with a focus on the micro-individual-level - as “the identification that an indi- vidual has with the core components of national identity, of which he is aware, and the willingness to express and protect the core components of national identity over all other in- terests and values at any costs.” This definition follows Charney (2003,p. 303)byfocusing on subjective conscious identification rather than objectively observable characteristics (i.e. “speakingVietnamese”, “havinganVietnamesecomplexion”). ItalsofollowsHechter(2000) and Smith (1983), by emphasizing that nationalism has to be an “action designed to render the boundaries of the nation congruent with those of its governance unit,” (Hechter, 2000, p. 7), and that it “prescribes certain kinds of action.” (Smith, 1983,p. 9,97)Following Johnston (2017)and Vu (2007, 2014), “core components of national identity” in the specific cases of Vietnam are defined as (1) pride in the nation-state; (2) blind or naive support for the nation-state; (3) defense of national autonomy and territory unity; and (4) negative evaluation of relevant out-groups. Chapter 2 explains why these choices are made in more details. The definition leads to two dimensions of nationalism at the micro-individual level: 1) 9 Volatile Nationalism | Xinru Ma baseline nationalism, a dimension that identifies a baseline level of national identification when there are no triggers, and 2) volatility, a dimension that identifies how citizens react to nationalist triggers. Put di↵erently, the baseline dimension denotes how nationalistic an individual is when there’s no trigger, and can be perceived as a measurement of attitude;By contrast, the volatility dimension denotes the actions an individual may take when there’s a trigger, a dimension that denotes whether that nationalistic attitude is strong enough for an individual to endure some costly actions. Moreover, these two dimensions, while plausibly interconnected at first glance, should be studiedasindependent. Tobetterunderstandtherelationshipbetweenthesetwodimensions, one can think about the securities markets. In the securities markets, volatility refers to the amount of uncertainty related to the size of changes in a security’s value. A higher volatility meansthatasecurity’svaluecanpotentiallybespreadoutoveralargerrangeofvalues. This meansthatthepriceofthesecuritycanchangedramaticallyoverashorttimeperiodineither direction. A lower volatility means that a security’s value does not fluctuate dramatically, and tends to be more steady. Specifically, when the security’s value rises and falls more than one percent over a sustained period of time, it is called a “volatile” security. (Summa, 2018; Kuang, 2018) The volatility of the security is not necessarily linked with the price of the security, as a volatile security can have low trading price, while an involatile security may have consistent high trading price. The baseline and the volatility levels of nationalism are distinct in the similar manner, and I maintain that it is the ex-ante baseline level of nationalism and the varied degrees of volatility that jointly di↵erentiate each nation in their levels of nationalism. Chapter 2 also elaborates on a novel measure of nationalism by applying automated text analysis methods to social media data in Vietnam. Specifically, it establishes both the base- line level and the volatility level of nationalism by computing a daily index of nationalism, which is proxied by the share of nationalistic comments on posts published by key public 10 Volatile Nationalism | Xinru Ma pages on Facebook, the most popular social media platform in Vietnam. Empirically, it focuses on two cases - the anti-China protest in May 2014 and a recent anti-China protest in June 2018 - that represented two of the largest anti-China protests in the bilateral history. By tracing the baseline nationalism precedent to the protests, the peak level of nationalism during the protests, and the new baseline after the protests, Chapter 2 seeks to provide within-country comparison of how nationalism evolves in Vietnam, and to reveal both the short-term and long-term e↵ects of nationalism in various scenarios with and without na- tionalistic triggers. My focus shifts from Vietnam to the Philippines in Chapter 3 by applying the same con- ceptualization and measurement method of nationalism. Like Vietnam, the Philippines has been involved in multiple competiting claims to the South China Seas against China. 14 The confrontation between China and the Philippines can be dated back to February 1995, when the Philippines released photographs showing four octagonal structures on steel pylons fea- turing guard posts with satellite antennas constructed by China on the Philippines-claimed Mischief Reef, along with five naval support ships in the zone claimed by the Philippines. SuchtensionflaredupagaininApril2012, whenChinaandthePhilippinesfoundthemselves lockedinadangerousnavalstand-o↵overScarboroughShoal. In2013,thePhilippines-China dispute once again caught the world’s attention as the Philippines filed a Notification and Statement of Claim with the Hague Permanent Court of Arbitration, challenging China’s ambiguously-defined nine-dash line. Indeed, conventional wisdom has believed that “Chi- nese aggression...has done more to galvanize Philippine nationalism than anything in the past few years.”(Esmaquel II, 2012) As Dor (2015) argues, “the Aquino administration has taken action to help further strengthen nationalistic tendencies in the Philippines over the disputes with China. Embodying the saying ‘never waste a good crisis,’ President Benigno 14 For a detailed timeline of the Philippines-China maritime dispute, see “Timeline: The Philippines- China maritime dispute.” Rappler.com. https://www.rappler.com/world/regions/asia-pacific/ 139392-timeline-west-philippine-sea-dispute 11 Volatile Nationalism | Xinru Ma Aquino III has latched on to the outcry of nationalists, whipping up nationalism through vocal decrees and executive orders.” However, is nationalism in the Philippines really rising? Has China’s behavior on the maritime issues really aggravated Philippine nationalism? If this is the case, why did the public in the Philippines elect a President who even joked to China that “if you want, just make us a province?”(Heydarian, 2018) Thus, the Philippine case itself is empirically puz- zling. Additionally,thePhilippinecaseconstitutesacross-countrycomparisonofnationalism with that of Vietnam, so as to shed light on how maritime disputes with China has shaped di↵erently, if ever, the rise and fall of nationalism in its neighboring countries, as well as how countries, despite both being involved in maritime disputes with China, react di↵erently to China-related nationalist triggers. Including the case of the Philippines is also important for methodological reasons. One potential push-back in measuring the volatility dimension of nationalism is the regime type and its associated repressive capacity. 15 Literature on repression and political mobilization pointsoutthatmanyregimes,especiallyauthoritarianones,useactsofrepressiontosuppress dissent and to exclude large portions of their populations from mobilizing for regime or policy change. (Davenport, 2014, 2007; Carey, 2006; Lichbach, 1987)Inthecontextof this dissertation, a potential push-back is that if the government threatens to repress (or has repressed in the past) the citizens who are vocal in expressing nationalism either via words on social media or via deeds by taking to the street, then it biases the result by underestimating the volatility dimension of nationalism, as citizens may not be willing to respond to nationalist triggers due to the fear of potential punishment. However, this push- back relies on a key assumption that repression negatively a↵ects participation and silences 15 Repression is defined as - following Davenport (2005, p. 122) - “actions taken by authorities against individuals and/or groups within their territorial jurisdiction that either restrict the behavior and/or beliefs of citizens through the imposition of negative sanctions (e.g., applying curfews, conducting mass arrests, and banning political organizations) or that physically damage or eliminate citizens through the violation of personal integrity (e.g., using torture, disappearances, and mass killing).” 12 Volatile Nationalism | Xinru Ma social movements, an assumption that is heatedly debated within the repression literature. Davenport (2007,p. 8)referstothemixedfindingsonthee↵ectofrepressionondissentas the “punishment puzzle,” as such e↵ect can be positive ( Francisco, 1996; Lichbach and Gurr, 1981), curvilinear (Muller, 1985; Brockett, 2005), or nonexistent (Gurr and Moore, 1997). Evenifwesetthisdebateasidebyassumingrepressionindeeddecreasesdissent,wewould expect to observe a higher volatility level of nationalism in the Philippines compared to the case of Vietnam, as democracies are least likely to deploy repressive behavior (Carey, 2006). However, thefactthatnationalisminthePhilippineshaslowervolatilitylevelhighlightsthat the observed low level of volatility is not necessarily due to the fear of potential punishment from the regime. This, again, reinforces the fact that individuals, even though embedded within similar social networks and structural contexts, perceive and react to nationalist triggers di↵erently. Aggregating these individual preferences to the macro-level, we can and should di↵erentiate countries by how they react to nationalist triggers. Chapter 2 and 3 discuss in more details the user demographics, the technology and policies of censorship on social media in Vietnam and the Philippines, the repressive capacity of the two regimes, as well as additional validation methods to address the potential bias in the results. In summary, automated text analysis of rich social media data shows that there exists a low and stable baseline level of nationalism in both Vietnam and the Philippines, signaling an over-estimation of nationalism in extant literature. On the other hand, the two countries di↵er greatly on their levels of volatility. While the Philippines demonstrate an involatile nationalism with few responses to nationalistic triggers, Vietnam has a relatively highly volatile nationalism. Moreover, rather than being responsive only to China-related triggers, publics in Vietnam and the Philippines are equally, if not more, responsive to domestic issues and US-related foreign issues, highlighting the understudied social contexts in which a“rising-nationalism”argumentismade. Treatingnationalismandthestochasticoutbursts of demonstrations by di↵erent societal groups for diverging and often contradictory reasons 13 Volatile Nationalism | Xinru Ma as an exemption is missing the trees for the forest. The literature may infer that nationalism is rising due to increased observed occurrence of protests, while it could simply be the case that citizens in certain countries are becoming more and more vocal on all issues. 1.1.2 TheTwoUncertaintiesofNationalismasCredibleSignalinInternational Crisis After conceptualizing and measuring nationalism, this dissertation proceeds to examine the mechanisms through which nationalism escalates the risks of armed conflicts. With a formal model that emulates the strategic bargaining between nationalist protesters and the gov- ernments, this formal chapter explores 1) the scope conditions under which the government is willing and able to use nationalism – exemplified in the instances of nationalist protests - as leverages in international bargaining, and 2) the conditions under which the foreign government is willing and able to view that as a credible signal. The goal of the formal chapter is to complement the previous chapters that measure nationalism with social media data. After all, nationalism is manifested in words - rather than deeds - on social media. It is true that nationalists may be active online by casting fierce and nasty claims, but it is more important to specify what actions nationalism could cause the government to take in times of crisis. Without crystalizing the scope conditions for nationalism to influence foreign policies, debating on whether “nationalism is rising” informs us little about nationalism that barks from nationalism that bites. Similar to the term “nationalism,” the term “nationalist protests” may invoke di↵erent images in di↵erent readers’ mind. For those who follow news closely in East Asia, a typ- ical account of nationalist protests invokes the images of anti-foreign government protests grounded in historical problems, in particular the controversies surrounding apologies due to the World War II, history textbooks, Koizumi’s visits to the Yasukuni shrine, which are further complicated by the over-lapping maritime claims in the South and East China Sea. 14 Volatile Nationalism | Xinru Ma (Friedberg, 1993; Chan and Bridges, 2006; Cho and Park, 2011; Mochizuki and Porter, 2013; Dudden, 2014)ForthosewhoarefamiliarwiththehistoryinEuropeandtheBalkans,na- tionalist protests often attest to the conflicts resulting from advocacy of or support for the interests of a particular ethnic group, especially with regard to its national independence or self-determination (e.g. protests precedent to the “velvet divorce” that separated Czechoslo- vakia into Slovakia and the Czech Republic, the Catalan independence movement in Spain, and the pro-Brexit marches). When it comes to the U.S., nationalist protests have been increasingly associated with “white nationalism,” referring to the protests by “a group of militant whites who espouse white supremacy and advocate enforced racial segregation”. (Perlman, 2017; Swain, 2002; Smith, 1981) To keep this dissertation focused, I define nationalist protest - following Weiss (2014)- as “a public manifestation by a group of people, containing hostile feeling toward a foreign governmentorpeople,androotedinadvocacyorsupportforthenation’sinterests,especially to the exclusion or detriment of other nations.” Protest may take covert forms including acts of non-compliance. Given this dissertation’s interest in analyzing the challenges posed by nationalist protests that a↵ect outsiders’ utilities and that can be observed by outsiders, protests studied here only encompass overt mass mobilization. Appendix 1.4.2 o↵ers a more detailed analysis of the similarities and di↵erences be- tween anti-foreign government protests and other types of nationalist protests such as ethno- nationalist protests (e.g., Quebec or pan-Arab nationalist protests), and other forms of do- mestic collective actions directed towards issues without explicit connections with a foreign government (e.g., labor or land-use protests). It also provides descriptive statistics for when and where anti-foreign government protests happen by synthesizing extant event datasets such as the Armed Conflict Location and Event Data project (ACLED), the Social Conflict Analysis Database (SCAD), the Integrated Crisis Early Warning System (ICEWS), and the Intranational Political Interactions (IPI) project. With that I hope to clarify the generaliz- 15 Volatile Nationalism | Xinru Ma ability and the limitation in applying the model developed in this dissertation to protests that are not anti-foreign government. Apart from complementing the social media chapters, formally analyzing the e↵ect of na- tionalismonmilitarizedconflictsisnecessaryitself. Despitethemassivevolume, thee↵ectof nationalism hasn’t been formally analyzed yet. Due to unjustified assumptions and unclear scope conditions, anecdotal empirical evidences have been provided for nationalism’s deter- minant or indeterminate influence on militarized conflicts, and the same protests have been used by di↵erent scholars as evidences for completely opposite theories. Specifically, though concurring that nationalism in East Asia has spiked repeatedly over the last decade, the literature diverges on whether such nationalism could drive governments “prone to muscle- flexing” with assertive and belligerent foreign policies (Christensen, 1999; Ross and Feng, 2008; Ross, 2009; Gries, 2004), whether it can only constrain rather than determine policies bysettinglimitsonpossibleoptions(Fewsmith and Rosen,2001;Hughes,2005; Fravel,2008), or whether it is only a tool in international bargaining as the Chinese government becomes arobustandflexibleregimeofgreatadaptabilityandremarkablee↵ectiveness( Reilly, 2011; Han, 2013; Hollihan, 2014). Suchdivergenceisrootedintwoopposingperspectivesonthemechanismsthroughwhich nationalism influences foreign policy in a broader scholarly and historical context. One perspective, known as the elite-focused “top-down” mechanism, holds that nationalism is instrumental for elites in their pursuit of self-interest (Anderson, 1983; Breuilly, 1993), a “doctrine of rule in the name of the people but not necessarily by the people” (Snyder, 2000,p. 36). Specifically,twochannelsemergefornationalismtoexertsuch“top-down” influence: Firstly, nationalism could be an avenue and a tool for intra-elite competition. Gagnon (1994) contends that ethnic nationalism provides means for endangered elites to deter domestic challengers in a shifting political and economic structure. Snyder (2000), and Schrock-Jacobson (2012)contendthatnationalismcanleadtobiasedstrategicassumptions 16 Volatile Nationalism | Xinru Ma and permit the suppression of domestic opposition groups, thus creating the conditions for “nationalist bidding wars” among elites and/or between elites and the masses. Reilly (2011, p. 74-75) argues that public mobilization is most likely when elites are divided, and such divisionisreflectedbytheocialtoleranceofthenationalistprotests. Secondly, nationalism could be a diversionary tactic to provoke the wrath of “national enemies” and their foreign allies, thus directing domestic pent-up grievances in society. (Smith, 1998; Tarar, 2006) As Christensen (1999)argues,sincetheChineseCommunistPartyisnolongercommunist,it must be more Chinese. The alternative perspective, known as the mass-focused “bottom- up” approach, holds that nationalism generates prohibitively high cost in the government’s legitimacy, thus even if the government has sucient resources to stifle nationalism, the government may be forced to cater to nationalism by taking belligerent actions and hawkish foreign policies. (Gries, 2004; Gries et al., 2011) Recently, few works emerged to strike a middle ground between the two mechanisms. Weiss(2013,2014)arguesthatthetwoopposingapproachesprovidegreateranalyticleverage than their individual parts once woven together. The extent to which nationalism influences foreign policies depends on the interaction of government objectives and societal demands: a top-down decision by an authoritarian government confronted with bottom-up mobilization. Inbargainingterms,allowingorrepressingnationalistprotestshelpstheChinesegovernment signal its diplomatic intentions and reveal the degree to which its hands are tied by domestic sentiments. Allowing nationalist protest signals government’s resolve to stand firm, while repressing protest signals its willingness to cooperate. By examining nationalism in a two- level game, Weiss’ work advances the field beyond the impasse between two competing perspectives,andlinksnationalismwiththerichliteratureonbargainingtheoryanddomestic audience costs, especially the audience costs in autocratic regimes in IR literature (Weeks, 2012; Levento˘ glu and Tarar, 2008). Despite this important progress, the scope conditions of nationalist protests to be a 17 Volatile Nationalism | Xinru Ma crediblesignal,andtheoveralle↵ectofnationalistprotestsasasignaloncrisisbargainingstill remain unspecified, particularly due to two sets of uncertainties: 1) the home government, whoisuncertainwhethertheprotestwillevolveintoananti-governmentmobilization,and2) the foreign government, who is uncertain whether nationalism constitutes a genuine credible and costly constraint of the home government. From an authoritarian government’s perspective, a nationalist protest is particularly challenging as it can easily turn into an anti-regime mobilization, if the protest is not a disguised collective action against the regime in the first place. When nationalist protests can spontaneously evolve into anti-regime mobilizations, allowing the protests is like holding afirecrackerinthehand. Thenatureoftheprotestbecomesamovingtarget,andtheprotest is often infused with heterogeneous and often contradictory interests. The anti-Japan protest spread across China in nearly a dozen cities for three weeks in 2012 is a case in point. Initially in response to Japan’s purchase of the disputed Diaoyu (named Senkaku in Japan) Islands, it quickly went beyond the level of expressing anti- Japanese sentiment and became an opportunity for divergent societal groups to express their criticism towards the regime within nationalistic slogans. For instance, in a widely spreadphotoonSinaWeibo, theTwitterofChina, anelderlyshopkeeperwrotethefollowing poem on a blackboard in front of his shop by the protest route:“No medical care, no social welfare, [we] have the Diaoyu Islands in our heart; no property rights, no human rights, on the Diaoyu Islands [we] fight for sovereignty rights; cannot a↵ord houses, cannot a↵ord graves, [we] won’t yield an inch of our land to the Japanese.” 16 Concerned about being imprisoned for protesting without proper legal proceedings, a protester waved a banner saying “Abolish the re-education through labor system 17 first, in case [we] get interned 16 The original slogan reads, “° ; › ° > › √ -Å ì | õ 1 ó ? ú { _ Å b ì | õ ° i C ° ∫ C ì | õ ⌦â; C p w ? Ó w _ ¯ ©  , ∫ ⇥ ”Iused Zhang (2015)’s translation of this post. 17 Thereeducationthroughlaborsystem, institutedunderMaoZedongtoreform“opponentsofsocialism” atthewhimsoflocalpoliceauthoritieswithoutlegalproceedings,wasabolishedin2013inChina. Theslogan 18 Volatile Nationalism | Xinru Ma for protesting for the Diaoyu Islands.” Other protesters embedded their criticism towards chengguan,thewidelyloathedviolentcorruptedmunicipalpolice,innationalisticbanners saying“Send3000chengguan toJapan, weensurethattheywillreclaimallthelostterritory; Send 500 corrupt ocials to Japan, we ensure that they will eat Japan to bankruptcy.” 18 From the foreign government’s perspective, when a nationalist protest can easily spin out of control into an anti-regime mobilization, repressing the protest can hardly reveal any information regarding the home government’s preference over the disputed issue. Instead, the action might be perceived merely as an indicator of the home government’s domestic preference on regime survival, which o↵ers little help to solve the problem that induces bargainingfailureattheinternationallevel. Thisthencallsintothequestionwhyallowingor repressingtheprotestcanbeasignal,andasignalofwhat. Forinstance,uponobservinganti- US sentiment in China few years following the Boxer Rebellion, the largest and most violent anti-foreign uprising with the wholesale massacre of foreigners in 1900, then US ambassador reportedthat“seriousanti-foreigndemonstrationsareinevitableduringthecomingsummer, if not before...There is danger that foreigners will be drawn into the vortex and massacred if they do not leave the country...[however,] primarily it will be a domestic trouble...a revolt of the people against Manchu rule.” (Shaw, 1906,p. 278)Inotherwords,whenaforeign government enters a bargaining being highly confident that the protest is mostly a problem for the home government, it is not at all clear what underlies the foreign government’s interestsincapitulatingtothisprotestabroad,regardlesshowthehomegovernmentmanages it. Consequently, allowing or repressing the protest becomes a noisy signal for the foreign is rhythmed in Chinese, which reads “H ü ≥ Y ç › ì  2 › å ´ ≥ Y ”. 18 Chengguan, ocially known as Urban Administrative and Law Enforcement Bureau, are local law en- forcementagenciesinChinaresponsibleformanagingmorequotidianaspectsofurbanlife,suchasregulating streetvendorsandunlicensedconstructionsites. Separatefromconventionalpoliceforces, chengguan appear to operate with little oversight. As a result, stories of chengguan brutality are common. On the other hand, a common form of corruption in China is ocials attending lavish banquets in private clubs and restaurants with issuing only IOUs (an informal document acknowledging debt) for food and drink consumed. The original banner reads “Ÿ ⌘ C Œ ° u ö 6 fi ì | õ Ÿ ⌘ î ~ * P ò › ¡ ⇤ Æ ✏  , ”. For more examples of such slogans and banners, see http://www.wenxuecity.com/news/2012/09/18/1987373.html 19 Volatile Nationalism | Xinru Ma observers, and often, it is particularly dicult for the foreign observes to decide what should be interpreted from this signal. Thepurposeoftheformalchapteristotakeseriouslytheuncertaintythatanynationalist protest, no matter how sincere it appears in the first place, can evolve spontaneously into an anti-regime mobilization, and clarify that given this uncertainty, under what conditions it can serve as a credible signal, and signal of what, in international crisis bargaining. With a signaling game, it seeks to address three sub-questions: (1) What assumptions are needed on the nature of the dispute and the bargaining environment to make nationalist protests a potential signal in international bargaining? (2) What is the exact problem that induces bargainingfailure, whichultimatelyleadstomilitarizedconflicts? And(3)whyisnationalist protest a signal? And why is it updating? In a nutshell, I argue that the uncertainty that any nationalist protest can evolve sponta- neously into an anti-regime mobilization is a liability but also an asset, even a necessity for nationalist protest to be a credible signal in international bargaining. It is this uncertainty thatincursdi↵erentcostsfordi↵erenttypesofthehomegovernment,whichfurtherstabilizes information transmission and honest signaling. Specifically, not all authoritarian governments are equal in state capacity. Some are “strong,”andsomeare“weak.”Intimesofconflict,thekeyobstaclethatpreventstheforeign government from reaching ecient peaceful settlements in an international bargaining is the asymmetricinformationproblem,inthesensethattheforeigngovernmentisuncertainabout the home government’s type. That is, the foreign government’s utility from war or peaceful settlement is dependent on whether the home government is strong or weak: a “strong” government has high capacity to fight and win in war, while a “weak” government has weak fighting capability in external wars. Nationalist protests can be a credible signal as the uncertainty of nationalist protests turning into anti-regime mobilizations incurs di↵erential costs for di↵erent types of home government: A “strong” government that has high capacity 20 Volatile Nationalism | Xinru Ma to fight in external wars can also e↵ectively contain domestic mobilizations (more money, moretroops, morelegitimacy, morecentralizedpower, etc.), whilea“weak”governmentthat has weak fighting capability in external wars has high cost to contain domestic mobilizations internally. Thus, allowing or repressing a protest - an action that can be endogenously chosen by the home government - can be informative and updating as this action enters into the home government’s utility by incurring di↵erent costs for di↵erent types, which induces astrongandweakhomegovernmenttoatleastpartiallyseparatetheirresponsestothe protests. Apart from pointing out the necessity of this uncertainty on the e↵ect of nationalist protestsasasignalininternationalcrisis,themodelfurthershowcasesanuancedrelationship between the two. Specifically, there exists a U-shaped relationship between the probability of nationalist protests transitioning to anti-regime mobilizations and the probability of bar- gaining failure and the ensuing conflicts (Figure 1.1.2). When the probability of transition is low, nationalist protests are not costly enough, thus failing to credibly reveal information; the high probability of transition, on the other hand, is likely to result in repression of all protests, forestalling them being observed as a signal. Thus, nationalist protests have the best chance to be a credible signal with mid-level transition probability. Here, nationalist protests are costly enough to allow for credible information transmission, while the home government is not so averse that they resort to repression. In extreme cases, when the protests have no uncertainty, allowing nationalist protests is costless, and there’s no mecha- nism to prevent the weak type of home government from mimicking a strong type. Blung thus happens, and the probability of conflicts is determined by the foreign government’s a prior belief of to what extent the home government is strong. On the other hand, when the protests for sure turn into anti-regime mobilizations, protests will be repressed at some point, and signaling with protests is either not attempted or irrelevant in the bargaining. This formal chapter thus advances the literature by clarifying the scope conditions under 21 Volatile Nationalism | Xinru Ma Figure 1.1.2: Relationship between the Uncertainty of Nationalist Protests Transitioning to Anti-regime Mobilizations and Bargaining Failure which nationalist protests - despite the potential pointed out by the latest literature such as Weiss (2013, 2014)-canindeedbecomeacrediblesignalininternationalbargaining,and more importantly, why it is so. Substantively, it challenges the conventional wisdom that assumesa linearrelationshipbetweennationalismandthe riskofarmedconflicts. Thequan- titative studies point out that in general nationalism significantly increases the probability of interstate war initiation (Schrock-Jacobson, 2012; Cederman et al., 2009), or the severity of violence (Carment and James, 1995; Fox, 2004). By contrast, the model shows that not all forms of nationalism is dangerous. When it comes to nationalist protests - one of the most frequently observed and dangerous platforms for expressing nationalism - rather than increase linearly the risk of armed conflicts, they can surprisingly decrease ex ante proba- bility of militarized conflicts, when the uncertainty of nationalist protests transitioning to anti-regime mobilizations is at middle level. 1.2 Context Within Literature and Limitations Myresearchonmeasuringnationalismanditse↵ectismotivatedbytwobroaderpuzzles: Are there any new methods in measuring public opinion? How do we address the methodological 22 Volatile Nationalism | Xinru Ma challenges for causal inference that stem from strategic conflict behavior? To date, nationalism is measured at best with “proxy” questions in annually aggregated surveys (e.g. World Values Survey, Asian Barometer), or opinion polls conducted immedi- atelyfollowingnationalistictriggers(e.g. telephonepollsbynewspapersonJapan’sdetention ofChinesefishermen). Appendix1.4.3providesalistofquestionsonnationalismavailablein popular time-series surveys. Similar to the recent well-publicized “polling misses” (e.g. the Brexitelection andthe U.S.presidential election), researchers working withpolling dataface severechallengesrangingfromsamplingframe, nonresponsebias, continuityinquestionnaire design, to the interpretation of results. (Atkeson and Alvarez, 2018)Forthoseworkingwith surveydatainauthoritariancontexts, gaugingpublicopinionisevenmorechallengingdueto issues such as “preference falsification” (Kuran, 1997), “spiral of silence” (Noelle-Neumann, 1974),anddatareliabilityinstate-sponsoredsurveysforpropagandapurposes(Sieger,1990). This dissertation thus joins a growing body of scholarship that recognizes the importance of conceptualizing, measuring, and testing the e↵ect of public opinion with new methods. (King and Lowe, 2003; Nagler et al., 2016; Steinert-Threlkeld, 2017)Bycombiningstudies on nationalism with latest methods in big data and formal modeling, it aims to facilitate interdisciplinary conversation between data science and International Relations, incorporate micro-dynamics of nationalism enabled by the rigorous content analysis of the rich details in social media data, and strengthen the clear advantage of two-country comparative analysis of nationalism. Nevertheless, a comprehensive examination of the e↵ect of nationalism on all types of international disputes is beyond the scope of this dissertation. To crystalize assumptions and take-aways, I’ve confined myself to the analysis of territorial disputes, defined as a con- flicting claim by two or more states over the sovereignty of the same pieces of land borders or maritime boundaries (Fravel, 2008). As a field, territorial disputes have been documented as the most important single cause of war between states in the last two or three centuries. 23 Volatile Nationalism | Xinru Ma (Vasquez, 1993; Hensel, 1996; Huth and Allee, 2002; Gibler, 2012) Historically, territorial disputes are associated with between 80 and 90 percent of all wars (Vasquez, 1993), as well as enduring or strategic rivalries among states (Thompson, 2001). Recently, it has been per- ceived as the most dangerous destabilizing factor and the most likely flashpoint especially in East and Southeast Asia. However, although a growing literature posits the importance of nationalist sentiment on territorial claims, there’ve been few specific hypotheses as to why this is the case (Johnston, 2012,p. 68). Ontheotherhand,existingexplanationsforterri- torial disputes – the strategic values, the relative capacities, the existence of ally, and regime types – still rely on the structural di↵erences at the systemic level, thus remaining limited in their informative powers. Consequently, though being all conscious of the weakness, the focus on territorial disputes and the omission of other types of policies such as trade policies during economic disputes is deliberately made, while the inclusion of these cases would also muddy the focus and extend significantly the length and scope of this dissertation. 1.3 Plan of the Dissertation The rest of the dissertation is structured as follows. Chapter 2 assesses the extant literature on the nexus between nationalism and foreign policies, and identifies how the absence of the concept of baseline nationalism has limited our understanding of the e↵ects of nationalism, if at all,onforeignpolicies. Italsoproposesatheoryofvolatilenationalismbyintroducing two dimensions of nationalism - the baseline level and the volatility level of nationalism - and details the methodology of using social media data to measure the two dimensions, including the case selection, data collection, process of automatic classification, potential caveats and measurestakentoaddressthesecaveats. Empirically, itappliesthismeasuretothe2014and 2018 anti-China protests in Vietnam, in order to identify and constitute a within-country comparison of the baseline and volatility levels of nationalism in Vietnam. Chapter 3 applies thismethodto thePhilippines, soasto constitutea cross-country comparison ofnationalism 24 Volatile Nationalism | Xinru Ma in Vietnam and the Philippines. The focus moves from measuring nationalism to formally studying the e↵ect of nation- alism in Chapter 4. It first proposes a baseline bargaining model, in which the nationalist protest has no uncertainty and remains as anti-foreign government throughout. In this set- up, I clarify the specific information problem investigated in this model, the key parameters that shape the preferences of each actor, as well as how and why nationalist protests fail to become a credible signal in the international bargaining when the protests have no un- certainty. I then modify the model by incorporating the uncertainty regarding the protest. Here, allowing or repressing nationalist protests not only generates bargaining leverage but also allows for information transmission if the two types of the home government at least partially separate their responses to the protests. With this, I explore under what conditions allowing or repressing nationalist protests can separate a strong home government from a weak type, how it induces the foreign government to make concessions or fight, and the conditions on the prior belief of the foreign government for such equilibrium to exist. With this two-step set-up, I hope to clarify how this uncertainty of transition changes the calculus of both the home and the foreign governments, and the significantly distinct conditions on the prior beliefs of the foreign government to make concessions in international bargaining. The chapter ends with an illustration of the (in)e↵ective signaling with the 2014 anti-China protest in Vietnam. As the discussion in this introduction implies, the formal and informal chapters address a di↵erent facet of nationalism at di↵erent geographical locations in di↵erent contexts. Each is, therefore, suciently theoretically and empirically self-contained to be read a la carte. The dissertation ends with Chapter 5, which discusses the implications of my theoretical arguments for several promising policy questions, and proposes the way forward. 25 Volatile Nationalism | Xinru Ma 1.4 Appendix 1.4.1 Various Ways of Conceptualizing Nationalism in Extant Literature AsBrubaker and Laitin (1998, p. 427) famously point out, “When studying nationalism, the problem is not that there is no agreement on how things are to be explained; it is that there is no agreement on what is to be explained, or whether there is a single set of phenomena to be explained. Rather than confronting competing theories or explanations, we confront alternative ways of posing questions, alternative approaches to or ‘takes’ on nationalism, alternative ways of conceptualizing the phenomenon and of situating it in the context of wider theoretical debates.” By no means of being exhaustive, Table 1.4.1 lists major approaches of conceptualizing nationalism in extant literature. For a more detailed review of definitions of nationalism, see Bonikowski (2016). Unit of Analysis Core of Nationalism Pillars of Nationalism Exemplar Definition Macro-state level Ideology: Nationalism as an elite ideology that facilitated modern nation-state formation and the subsequent di↵usion of the nation-state form. Imagined and constructed ideology “a political principle, which holds that the political and the national unit should be congruent” (Gellner, 1983, p.1) “a heterogeneous set of ‘nation’-oriented idioms, practices, and possibilities that are continuously available or ‘endemic’ in modern cultural and political life” (Brubaker et al., 1996,p.10) Macro-group level Ideology: Nationalism as a doctrine that a group of people of a sucient size has a prima facie right to govern itself and decide its future membership, if the members of the group so wish. Self-rule and self- determination “the doctrine that a people who themselves as distinct in their origins, culture, history, institutions, or principles should rule themselves in a political system that expresses and protects those distinctive characteristics.” (Snyder, 2000,p.23) Micro- individual level Dispositions: Nationalism as a set of attitudes that shape the perceptions and behaviors of ordinary people as they come into contact with political institutions (e.g., by voting) and engage in social interaction (e.g., with immigrants or ethnic minorities). In-group identification “any form of in-group identification for a group defined in part by reference to a geographical area along with some form of sovereign government over that area” (Hogan, 2009,p. 4) Out-group objection “a perception of national superiority and an orientation toward national dominance” (Kosterman and Feshbach, 1989,p. 271) In-group identification + out-group objection “an inclination to define one’s own group by criteria of descent, race, or cultural aliation; and derogation of groups not considered to be part of the nation” (Blank and Schmidt, 2003,p. 292) Table 1.4.1: Various Ways of Conceptualizing Nationalism in Extant Literature 26 Volatile Nationalism | Xinru Ma 1.4.2 When and Where Do Nationalist Protests Occur Following Weiss (2014), I define nationalist protest as “a public manifestation by a group of people, containing hostile feeling toward a foreign government or people, and rooted in advocacy or support for the nation’s interests, especially to the exclusion or detriment of other nations.” Anti-foreign government protests are similar to other forms of domestic collective actions in the following aspects: in terms of goal, both of them represent means premised on sig- nificant mass mobilization by which a disgruntled populace with the intention of instigating policychangefrombelow; intermsoforganization, bothofthemencompasscollectiveaction problems such as free-riding issues, and the protestors face potential punishment from the government. However, the di↵erence lies at: in terms of the target, the civic protests is an act of provocation by a mobilized sector of society to the ruling regime, while anti-foreign gov- ernment protests often intend to promote policy changes of foreign government, whether directly (e.g. by directly challenging their positions) or indirectly (e.g. by demanding do- mestic government to take a tougher stance). This leads to the second di↵erence: in terms of bargaining process, either ethno-nationalist protests or civic protests mostly only involve domestic bargaining (e.g., home state versus separatists without the direct engagement of third-party foreign country). The following map (Figure 1.4.2) shows the incidence of anti-foreign government protests around the world. With that I hope to clarify the generalizability and the limitation in applying the model developed in this paper to protests that are not anti-foreign government. 27 Volatile Nationalism | Xinru Ma Figure 1.4.2: Incidence of Anti-Foreign Government Protests (Beta Version) Note: The complete dataset that synthesizes extant event datasets such as the Armed Conflict Location and Event Data project (ACLED), the Social Conflict Analysis Database (SCAD), the Integrated Crisis Early Warning System (ICEWS), and the Intranational Political Interactions (IPI) project is still in the process of collection. The map above was based on SCAD (Africa and Latin America) that includes social conflicts from 1990-2017. For more details of SCAD, see Salehyan et al. (2012). 28 Volatile Nationalism | Xinru Ma 1.4.3 Potential Measures for Nationalism in Annually Aggregated Surveys Table 1.4.3 lists questions in three time-series surveys - World Values Survey, ISSP, and Asian Barometer - that can be used potentially by scholars to gauge nationalism. Survey Operationalization of Nationalism Survey Questions Scale Year and East Asian Countries Included in the Survey World Values Surveys (Wave 2/3/4/5/6) Pride in citizenship How proud are you to be [**]? 1=veryproud 2=somewhatproud 3=notveryproud 4=notproudatall 1990: Japan, South Korea, China 1995: China, Japan, Taiwan 1996: Philippines, South Korea 2000: Japan 2001: China, Philippines, South Korea, Vietnam 2002: Singapore 2005: Hong Kong, Japan, South Korea 2006: India, Malaysia, Taiwan, Vietnam 2007: Chin 2010: Japan, South Korea 2011: Malaysia 2012: China, India, Pakistan, Philippines, Singapore, Taiwan 2013: Hong Kong, Palestine, Thailand 2014: India Support for the nation-state Of course, we all hope that there will not be another war, but if it were to come to that, would you be willing to fight for your country? 1=yes 2=no ISSP (1995, 2003, 2013) Superiority of one’s own country over other countries The world would be abetterplaceif people from other countries were more like the [**]. 1= Agree strongly 2 = Agree 3 = Neither agree nor disagree 4=Disagree 5=Disagreestrongly 1995: Japan, Philippines 2003: Japan, Philippines, Taiwan, South Korea 2013: India, Japan, Philippines, South Korea, Taiwan Generally speaking, [**] is a better country than most other countries. Pride in citizenship How proud are you of (R’s country) in each of the following: political influence/economic achievements/armed forces/history 1=veryproud 2=somewhatproud 3=notveryproud 4=notproudatall Blind or naive support for the nation-state People should support their country even if in the wrong. 1= Agree strongly 2 = Agree 3 = Neither agree nor disagree 4=Disagree 5=Disagreestrongly 29 Volatile Nationalism | Xinru Ma Survey Operationalization of Nationalism Survey Questions Scale Year and East Asian Countries Included in the Survey Asian Barometer (wave 2/3/4) Pride in citizenship How proud are you to be [**]? 1=veryproud 2=somewhatproud 3=notveryproud 4=notproudatall 2005/11-2005/12: Vietnam 2005/11-2005/12: Philippines 2006/04-2006/09: Thailand 2006/05-2006/06: Mongolia 2006/09: South Korea 2006/07-2006/11: Singapore 2006/11: Indonesia 2007/01-2007/03: Japan 2007/11-2008/12: China 2007/09-2007/11: Hong Kong 2007/07-2007/10: Malaysia 2008/04-2008/05: Cambodia 2010/01-2010/02: Taiwan 2010/03: Philippines 2010/04-2010/05: Mongolia 2010/04-2010/08: Singapore 2010/08-2010/12: Thailand 2010/09-2010/10: Vietnam 2010/10-2011/11: Malaysia 2011/05: South Korea 2011/07-2011/10: China 2011/11: Japan 2012/02-2012/03: Cambodia 2012/09-2012/11: Hong Kong 2013/09-2014/08: Philippines Mongolia 2014/01-2014/11: Taiwan, Singapore 2014/08-2015/07: Malaysia 2014/12-2015/12: China 2015/01-2015/06: Myanmar 2015/06-2016/05: Indonesia, Vietnam 2015/08-2016/07: Cambodia Blind or naive support for the nation-state For the sake of the national community/society, the individual should be prepared to sacrifice his/her personal interest. 1= Agree strongly 2 = Agree 3 = Neither agree nor disagree 4=Disagree 5=Disagreestrongly Table 1.4.3: Potential Measures for Nationalism in Annually Aggregated Surveys 30 Volatile Nationalism | Xinru Ma 2 Long-term Trends versus Short-term Noises: Measure Nationalism with Social Media Data and Evidences from Vietnam “If a vocal minority, however fervent its cause, prevails over reason and the will of the majority, this Nation has no future as a free society.” -Richard Nixon, “The Silent Majority” speech 2.1 The Myth of Rising Nationalism Nationalism has been a both central and marginalized concept in the literature of interna- tionalrelations. Thecentralityofnationalismderivesfromtheveritablefloodofpublications on nationalism in the past decade, especially those on nationalism in East and Southeast Asia. Factors such as the rise of China and the consequent power transition in the re- gion, the increased confrontations at the sea, and the media-fueled stochastic outbursts of demonstrations have spurred an unprecedented increase in the number of investigations of nationalism in countries such as China, Taiwan, Japan, and South Korea, with a prolif- eration of typologies of nationalism ranging from naval nationalism (Ross, 2009), cultural nationalism (Yoshino, 2005; Befu, 1993; Hutchinson, 1994), economic nationalism (Sugino- hara, 2008; Helleiner and Pickel, 2005; Helleiner, 2002), liberal nationalism (Moore, 2001), aggrieved nationalism (Pei, 2003; Gries et al., 2011), state and popular nationalism (Chan andBridges,2006;Ko,2018), assertivenationalism(Whiting,1983), xenophobicnationalism (Shambaugh, 1999), and so on. In the past two years, 39 articles have been published on nationalism in the top five IR journals, 19 while an additional 49 articles have been published 19 Journal ranking is based on InCites Journal Citation Report provided by Web of Science (retrieved on October 16, 2017). The top five journals are: American Political Science Review, American Journal of 31 Volatile Nationalism | Xinru Ma in the leading China-focused journal The China Quarterly. Despite the impressive volume, many authors expressed frustration with the poor mea- sures of nationalism that prevailed to date. As Fravel (2008,p. 311)observes,since1990s, scholars and observers of Asia have noted the increasing prominence of popular nationalism, although it is unclear if nationalism is in fact increasing. Johnston (2017, 2012,p. 68)points out that a growing literature posits nationalist sentiment cannot be ignored on hot-button issues in the region such as China’s relations with Japan on territorial claims, although there are few specific hypotheses as to why this is the case. Conceptually, the term “rising” in extant literature has seldom been defined, and its usage “seems to mean anything from more intensely felt, to more extreme, to more pervasive within society, to more salient for leaders, to more influential on policy.” (Johnston, 2017, p. 8) As Carlson (2009,p. 25)pointsout more bluntly: “[D]espite (or, perhaps, because of) the pervasiveness of the resulting conven- tional wisdom about a surge in Chinese nationalism, to date the literature lacks any rigorous empirical testing of such an observation.” The lack of empirical testing is in part rooted in how nationalism has been studied so far. To date, most studies on nationalism, especially those in the context of East Asia, have been devoted to explain various facets of the origins, construction,and manifestations of nationalism with rich empirical and historical materials (Gries, 2004; Gries et al., 2011; Han, 2013; Downs and Saunders, 1999; Wang, 2008; Vu, 2007). This does not mean that this approach of studying nationalism in East Asia is of poor quality or of no value. However, nationalism has been absorbed into detailed and well-researched national historiography rather than a subject in its own right. The operationalization of nationalism has been much less well served, constituting a turning away from attempts to measure and provide theoretical accounts of the e↵ects of nationalism. Apart from the traditional focus on the origins, construction, and manifestations of na- Political Science, Journal of Politics, International Organization, and Journal of Conflict Resolution. 32 Volatile Nationalism | Xinru Ma tionalism, whether nationalism is rising or not is far from obvious for at least two other reasons. First, influenced by the stochastic coverage in media, scholars have been gauging the level of nationalism by using measures - either through public opinion polls or protests - immediately following some nationalistic triggers, thus biasing the estimation of national- ism in favor of the “rising-nationalism” argument. For instance, to characterize “the clash of nationalisms” between China and Japan, Chan and Bridges (2006)usedopinionpolls that immediately followed Koizumi’s visiting to Yasukuni shrine in 2005, and those that immediately followed Koizumi’s campaign for Japan’s permanent membership in the United Nations Security Council (UNSC). In these cases, nationalism was gauged by the share of respondents that supported the shrine visit (or the level of opposition for measuring nation- alism in China), and the share of support (or opposition in the case of China) for Japan’s campaign for the permanent membership in the UNSC. 20 However, by characterizing all pro-shrine-visit arguments as nationalist arguments, and lumping all supporters of Japan’s campaign for permanent membership into the nationalist camp, it is impossibleto determine the relative importance of nationalism as a causal factor independently of other geopolitical factors or national security reasons. In addition, while such data is useful for unpacking short-term dynamics of nationalism, the fact that we don’t have opinion polls following each trigger across time has limited our understanding of the long-term trend of nationalism. Consequently, it is often unclear how nationalistic a country when there’s no trigger, and whether the same nation always reacts to similar triggers in the same way. Second, for those scholars who seek to measure the long-term trend of nationalism with cross-year cross-sectional surveys, the lack of continuity in questionnaires and the lack of background information on whether the survey responses are collected in times of national- istic evens or not leaves the direction of bias in the estimation unknown. 20 For more studies that used these opinion polls immediately after certain nationalist triggers, see Grønning (2014), Self (2002), and Mochizuki and Porter (2013). 33 Volatile Nationalism | Xinru Ma Johnston (2017) represents one of the recent e↵orts that assesses this empirical claim on “rising” nationalism with cross-year surveys, and is worth explaining in some detail. Defining nationalism as a three-pillar concept - “pride in and valuation of the nation, blind or naive support for the nation-state, and the in-group’s valuation of relevant out-groups” (p. 14) - Johnston (2017)testedfiveclaimsaboutChina’srisingnationalismwithtime- series survey data in Beijing area from 2002 to 2015. Specifically, the first two components of nationalism were gauged with the following questions: (1) even if I could choose any other country in the world, I would prefer to be a citizen of China than any other country (measuring “pride in citizenship”), and (2) everyone should support their government even when it is wrong (measuring “blind or naive support for the nation-state”). (p. 19) The third component of nationalism - the in-group’s valuation of relevant out-groups - were operationalized and measured in three ways: (1) a survey question that measures “the beliefsinthesuperiorityofChinaoverothercountries,”(2)respondents’perceiveddi↵erence between Chinese, Japanese and American people (e.g., Chinese were ranked at the positive endandJapaneseandAmericansatthenegativeendona7-pointscale),and(3)respondents’ amity toward Japan and the U.S. on a feeling thermometer. Altogether, Johnston argued that on a number of measures, levels of Chinese nationalism have stagnated or dropped since around 2009, and that rising popular nationalism may not be a critically important variable constraining Chinese foreign policy. (Johnston, 2017,p. 9) While Johnston’s work has provided initial intellectual stimuli for a more systematic assessmentoftheclaimaboutrisingnationalism, italsoshowshowscholars’handshavebeen tied by the limited availability of public opinion polls when measuring nationalism. After all, researchers working with polling data have long been acknowledging severe challenges ranging from sampling frame, nonresponse bias, continuity in questionnaire design, to the interpretation of results. (Atkeson and Alvarez, 2018)Forthoseworkingwithsurveydata in authoritarian contexts, gauging public opinion is even more challenging due to issues such 34 Volatile Nationalism | Xinru Ma as “preference falsification” (Kuran, 1997), “spiral of silence” (Noelle-Neumann, 1974), and data reliability in state-sponsored surveys for propaganda purposes (Sieger, 1990). This chapter seeks to fill the gap in existing studies by o↵ering an alternative way of conceptualizing and measuring nationalism that depicts both the long-term trend and the short-term dynamics of nationalism in a given country. The next section proposes a two- dimensional conceptualization of nationalism: 1) “baseline” nationalism, namely, how na- tionalistic a nation is when there’re no triggers, and 2) “volatility” of nationalism, namely, how a nation responds to nationalist triggers. It then details the methodology of using su- pervised machine learning methods to measure these two dimensions. Empirically, it applies this method to the 2014 and 2018 anti-China protests in Vietnam, in order to identify and constitute a within-country comparison of how nationalism has evolved in Vietnam. In summary, this chapter generates three key findings that all challenge the “rising- nationalism” assumption: First, the baseline of nationalism in Vietnam has remained stable and low from 2014 to 2018, signaling an over-estimation of nationalism in extant literature. Second, even at the peak of the 2014 anti-China protest, the largest and most violent nation- alist protest in recent bilateral history, nationalism was not as high as we expected, which further implies that nationalism is overestimated if we only focus on the protesters immedi- atelyfollowingthenationalistictriggerswithoutgauginghowrepresentativetheyare. Lastly, though China’s behavior triggered surges in nationalistic comments during the protests in comparison to times without triggers in 2014 (low baseline but high volatility), the absolute amount as well as the share of nationalistic comments both decreased when the public was faced with a similar trigger in 2018, implying again that over time, nationalism in Vietnam has been decreasing - rather than increasing - since 2014. 35 Volatile Nationalism | Xinru Ma 2.2 A Two-dimensional Conceptualization of Nationalism 2.2.1 A Micro-Level Definition of Nationalism I define nationalism in the following minimalist approach as: “The identification that an individual has with the core components of national identity, of which he is aware, and the willingness to express and protect the core components of national identity over all other interests and values at any costs.” 21 This definition has two key distinct features. First, it purposefully leaves the core com- ponents of national identity undefined, but emphasizes the level of identification with the core components of national identity as opposed to what the components are for specific national identity, thus shifting the focus from distinct national identity possessed by people to the identification of which people are aware. Tworeasonsdrivethischoice. First,theoretically,subjectiveconsciousidentificationthat an individual is aware of itself matters. People’s identities are doubtless shaped by innu- merable factors, including language, socioeconomic, normative, and institutional features of their nation. However, such formative influences don’t necessarily mean that people identify with these features. For instance, both Anderson (1983)and Gellner (1983) emphasize the importance of language in constituting one’s national identity, however, as (Charney, 2003, 21 This definition is revised based on Snyder (2000, p. 23), which defines nationalism as “the doctrine that a people who themselves as distinct in their origins, culture, history, institutions, or principles should rule themselves in a political system that expresses and protects those distinctive characteristics.” Snyder’s definition, by emphasizing the political legitimacy of self-rule, serves well to study both domestic and in- ternational nationalist conflicts with the excluded groups inside the country and with any of these groups’ purported foreign allies, as well as a nation’s capture of foreign territory where its ethnic brethren reside (e.g. the nationalist strife in the Post-Communist states such as Yugoslavia, Caucasus, the Chechen region inside the Russian Federation, Serbia, or the Quebeois nationalists who seek autonomy within the Canadian federal state). However, defining the aim of nationalism as achieving a sovereign state or the seeking of political rights short of sovereign statehood by ethnic or cultural groups limits its investigation of conflicts to those whose goal is to establish or protect self-rule by the ethnic group, and excludes potential conflicts in which countries compete over inhabited and uninhabited islands, as is more commonly seen in East and Southeast Asia. Thus, Snyder’s definition, though serves well as a starting point, needs to be broadened. 36 Volatile Nationalism | Xinru Ma p. 303) points out, it is the nationalists, not native speakers, that view language as a source of pride and distinctiveness. Thus, one cannot assume the existence of nationalism from the bare fact that certain features of the nation have exerted a formative influence upon its people. In other words, nationalism is not a natural product if one has certain national identity, but can only exist when one identifies to a sucient extent with the nation. Thisdefinitionalsohighlightsthewillingnesstotakeactionsbytheindividualnationalist. National identification might provide the motivation for either national leaders or citizens to reacttocertainnationalistictriggers,however,identificationalonedoesn’tleadtheindividual to invest in the course of action. It is usually one thing for an individual to claim certain island as an integral piece of the nation’s territory, and quite another when we actually push him and see how invested he is in defending such claim. After all, as Anthony Smith argues, nationalism “prescribes certain kinds of action.” (Smith, 1983,p. 9,97)Thus,in additiontothestaticdimensionofnationalidentification, thedynamicdimension-basically, how individual reacts to nationalist triggers, or in other words, how a nationalist trigger transforms the latent national identification into manifested collective actions - should also be an important dimension to assess and di↵erentiate levels and types of nationalism. This dynamic dimension is particularly important if we are interested in exploring the e↵ect of nationalism on foreign policies and its associated mechanisms. According to Bunge (1996)and Hedstr¨ om et al. (1998), the specification of mechanisms for any theory must incorporate multiple levels of analysis by addressing how individuals construct ideas within their minds (the micro level), and how ideas di↵use across a population through social inter- action (the macro level). Following this, a complete theory of nationalism on foreign policies requires three parts (Figure 2.2.1.1). First, the situational mechanism, which explains how an individual actor is exposed to a specific social situation (nationalist trigger). Second, the action-formation mechanism, which explains how a specific combination of individual beliefs and action opportunities generate specific actions. Lastly, the transformation mechanism, 37 Volatile Nationalism | Xinru Ma which explains how these collective actions are transformed into certain macro-outcomes (elites’ collective decision on foreign policies). Thus, if we only study the static level of national identification without specifying how individuals react to nationalist triggers, the theory of nationalism on foreign policies is then incomplete as it misses the second step: the action-formation mechanism. Figure 2.2.1.1: A Complete Theory of Nationalism on Foreign Policies To better understand this second dimension, it is also useful to think about how the preference of a nationalist would di↵er from that of a non-nationalist, and how nationalists di↵er among themselves in a given country. Assume a crisis scenario in which the home government and the foreign government bargains over a disputed good. The society in the home government is divided into a unitary government and a large number of citizens. There are four types of citizens: volatile nationalists, involatile nationalists, volatile non- nationalists, and involatile non-nationalist. Each individual citizen, after learning the crisis, can choose to express his preferences on the disputed goods (e.g. protest on the street). Expressing his preference (e.g. through protest) incurs a cost (e.g. the individual needs to ask for leave from work to protest on the street, or any resources and transaction costs that are needed to overcome the coordination problem for organizing a protest). 38 Volatile Nationalism | Xinru Ma Suppose national identity in this scenario only has one core component: the autonomy and the territory unity of the nation-state, a core component of nationalism proposed by Smith (1983)and Gries et al. (2011). An individual citizen derives utility from three parts: (1) the non-material value that he attaches to the disputed goods based on the core compo- nents of nationalism (that is, national autonomy and territory unity), (2) material benefits from the share of disputed goods gained from international bargaining, if the disputed good is divided peacefully, and (3) the costs that an individual needs to pay to express his pref- erences, with the hope to inform and influence the government to adopt this individual’s preferences during the bargaining. The conceptualization proposed above thus captures two dimensions that di↵erentiate nationalists from non-nationalists: 1) baseline nationalism, a dimension that identifies to what extent an individual values core components of nationalism when there are no triggers, and2)volatility ofnationalism,adimensionthatidentifieshowanindividualreactstocertain nationalist trigger to express and protect the core components of nationalism. Nationalists value more the non-material intangible value to the disputed good due to their identification with the core component - national autonomy and territory unity - and the consequent feelings of attachment to the territory (Hensel, 2001; Hensel and Mitchell, 2005). They di↵er from non-nationalists, who value more the material benefits from a peaceful division of the disputed goods. 22 On the other dimension, nationalists also di↵er among themselves based on their levels of willingness to express and protect the core components of nationalism. Assuming a given 22 Anempiricalexampleisinthe2014China’santi-JapanprotestoverthedisputedDiaoyuIsland,Chinese nationalist protesters held banners with slogans such as “The Diaoyu island is forever Chinese territory! Diaoyu Island is sacred indivisible part of China!” In the following days, Chinese Ministry of Foreign A↵airs also issued a statement reiterating China’s “indisputable” claim to the “indivisible” Diaoyu islands. Thus, it is reasonable to assume that the nationalists value more the non-material benefit from demanding the entire disputed good, though this claim renders the dispute unresolved (thus no material benefit gained from the disputed good), while non-nationalists, who value less national autonomy and territory unity, are happy to divide the disputed good so as to obtain the material benefits generated by it. 39 Volatile Nationalism | Xinru Ma cost in the same institutional context for all individuals (that is, each individual, no matter whether you are nationalist or not, su↵ers the same punishment from the regime), a volatile nationalist is characterized by an individual that discounts the cost to a greater degree, thus being more willing to express and protect the core components of nationalism over all other interests and values at any costs. In other words, a volatile nationalist is more willing to take actions despite the cost of action, while an involatile nationalist is less likely to take actions, though both attach high intangible values to the disputed goods. Conceptualizing nationalism in a two-dimensional framework - the baseline and the volatility - generates two key insights that sharpen our understanding of nationalism. First, for a long time, existing literature has been relying on public opinion surveys by examining nationalism as a self-expressed attitude. However, studying nationalism as an attitude only captures half of the story that happens especially in times of crisis. Instead, we need both the attitude and action dimensions to correctly assess nationalism. While the baseline di- mension in the proposed conceptualization can be conceived as a measurement of attitude, the volatility dimension denotes the degree to which an individual is willing to take actions when there’s a trigger. In other words, it denotes whether that baseline attitude is strong enough to motivate an individual to endure some costly actions. The value of taking consideration the dimension of action into studies of nationalism has been backed up by decades of studies on other social phenomena, which have long identified the gap between self-reported attitude and observed action. For instance, the 2014 British Social Attitudes Survey reveals that among those people who define themselves as belonging to the Church of England, 51.9% never attend services. In fact, only 10.7% of people who identify with the Church of England report attending church at least weekly. This doesn’t even take into account that self-reported Church attendance is invariably higher than actual recorded attendance. (UK, 2019)Thisgapbetweenattitudeandactioncanbecausedby factors such as di↵ering contexts and mindsets, as well as social desirability bias, defined 40 Volatile Nationalism | Xinru Ma as the tendency “to skirt the truth when the question we’re asked has a societally accepted ‘right answer’” (Matfield, 2015). Specifically, Van de Mortel et al. (2008)reviewaseriesof studies that explored how the social desirability bias a↵ects behavior across a wide range of areas, including whether participants over reported how often they exercise, and factors influencingtheminimization,blameanddenialamongdomesticviolenceo↵enders. Similarly, Granberg and Holmberg (1990), Manski (1990), and Sheeran (2002)haveaddressedwhether peopleaccuratelyreporttheirvotingintentions, howbigthe“gap”isbetweenintentionsand behavior,andwhatpsychologicalvariablesmightbeabletobridgetheattitude-behaviorgap. Thetwo-dimensionalconceptualizationalsoshowshowill-definedthequestioniswhether nationalism is rising or not. At a minimum, we need to clarify: (1) Are there more and more citizens identifying with the nation (e.g. increased proportion of nationalists among the population holding the share of nationalists that would actually take actions constant, say, vertical dotted line in the left-panel shifts left)? Or (2) are citizens becoming more and more reactive to similar nationalist triggers (e.g. more likely to protest holding the total amount of nationalists constant, say, horizontal dotted line shifts downward)? Conceptualizingnationalisminatwo-dimensionalframeworkalsohascausalimplications, as it forces us to crystalize whose preferences the leaders care about when making foreign policies, and why this is the case. When conventional wisdom claims that nationalism drives aggressive foreign policies, an underlying assumption is that leaders care about the preferences of one particular group of citizens - the volatile nationalists - and are willing to cater to their demands. One way for the leaders to discern such preference is aggregate attitude expressed in the public opinion polls, by assuming that the group who expressed certain attitude would indeed transform their attitude into actions. However, expressed attitude can lead to the changes in policy only when the expressed attitude is strong enough to cause changes in behaviors. Just like in autocratic countries, whether the citizens detest autocracyposeslittlethreattotheregime, however, suchpreferencemattersfortheautocrat 41 Volatile Nationalism | Xinru Ma when the citizens detest the autocrat enough to take actions against the autocrat. Moreover, if the citizens’ preferences are now distributed in four categories (volatile na- tionalists and non-nationalists as well as involatile nationalists and non-nationalists), it is far from obvious which group of citizens the leaders care more about. If the majority of the society is comprised of involatile nationalists who won’t respond to nationalist triggers, would the leaders still adopt aggressive nationalist foreign policies? Or if the majority of the society is comprised of volatile non-nationalists, would the leaders be forced to seek peaceful division of disputed territory? Thus, without considering the dimension of volatility, it is dicult for researchers to specify the scope conditions for nationalism to influence foreign policies, as well as to distinguish nationalism that barks from nationalism that bites. Figure2.2.1.2illustrateshowthistwo-dimensionalconceptualizationhelpsusdi↵erentiate the types of individuals as well as nations at micro- and macro- levels, respectively. Figure 2.2.1.2: A Two-Dimensional Conceptualization of Nationalism and Its Application 42 Volatile Nationalism | Xinru Ma 2.2.2 Aggregating Individual Preferences to a Macro-Level Theory of Nation- alism on Foreign Policies So far, I’ve proposed a two-dimensional conceptualization of nationalism, as well as how these two dimensions - the baseline and the volatility dimensions of nationalism - jointly di↵erentiate individuals’ preferences when there’s no trigger, and the dynamic behaviors when there’s a trigger. I now aggregate these two dimensions to the macro-level, and give empiricalexamplesofhowthebaselinelevelofnationalismandthevarieddegreesofvolatility jointly di↵erentiate each nation in their levels of nationalism. To define the volatility level of nationalism at the marco-level, I build on the insights from economics. In the securities markets, volatility refers to the amount of uncertainty related to the size of changes in a security’s value. A higher volatility means that a security’s value can potentially be spread out over a larger range of values. This means that the price of the security can change dramatically over a short time period in either direction. A lower volatility means that a security’s value does not fluctuate dramatically, and tends to be more steady. Specifically, when the security’s value rises and falls more than one percent over a sustained period of time, it is called a “volatile” security. In terms of the time-span, avolatilityindexistypicallygaugedata 30-day period. (Summa, 2018; Kuang, 2018) Following this, I define the baseline level of nationalism in country i as the mean of the static level of national identification of the entire population when there’s no trigger, and the volatility of nationalism in countryi as the magnitude of the change (both up and down) of the baseline level of nationalism when there’s a nationalist trigger. The varying aggregated baseline and volatility levels of nationalism help categorize coun- tries into a 2*2 metrics (right panel in Figure 2.2.1.2). Perhaps the most studied, or em- pirically assumed, quadrant, is the upper right quadrant, which emphasizes a high level of baseline nationalism, and a high level of volatility when faced with potential triggers. For countrieslocatedinthisquadrant, thehighbaselinelevelofnationalismmotivatesthepublic 43 Volatile Nationalism | Xinru Ma to add high intangible value to the disputed goods based on lost autonomy or feelings of attachment to the territory, while the high volatility level of nationalism makes the public highly responsive to potential nationalist triggers. By contrast, the lower-right quadrant emphasizes a high baseline level of nationalism but a low level of volatility. Empirically, this probably is best represented by Japan. A recent major nationalist trigger between China and Japan is the fishing trawler collision in 2010. According to joint polls by Genron NPO and China Daily, 72% of the Japanese respondents hold unfavorable impression of China in 2010. By comparison, 55.9% of Chinese respondents had unfavorable impression of Japan in 2010. 23 However, at the time when the fishingtrawlerincidentunveiled, massiveprotestsbrokeoutinmultiplecitiesinChina, while only two small-scale protests were documented in Japan. This points to the scenario that despite a potential high baseline level of nationalism, the low volatility indicates that the publicislessresponsivetonationalisttriggers,atleastintheimmediatetime-framefollowing the trigger. The upper-left quadrant captures a low level of baseline nationalism, but a high level of volatility when faced with potential triggers. For countries located in this quadrant, the low level of baseline nationalism not only makes the citizens less incentivized to push the government for nationalistic policies, but also makes any demonstration, if any, more likely to evolve into demonstration against the incumbent government due to domestic grievances. Finally, the lower-left quadrant captures countries with a low level of baseline nationalism, as well as a low level of volatility when faced with potential triggers. Altogether, I maintain that it is the baseline level of nationalism and the varied degrees of volatility that jointly di↵erentiate each nation in their levels of nationalism. 23 For more details, see The Genron NPO. “On the Results of the 7th Japan-China Joint Opinion Survey.” Retrieved from http://www.genron-npo.net/en/opinion_polls/archives/5265.html. 44 Volatile Nationalism | Xinru Ma 2.3 Measuring Nationalism with Facebook Data in Vietnam As discussed earlier, nationalism is measured at best with annually aggregated survey data, thus failing to illustrate the short-term dynamics of nationalistic sentiment, especially how thissentiment,ifatall,drivesgovernmenttoescalatecertainconflictatthatmoment. Onthe other hand, scholars who seek to measure nationalism with public opinion polls immediately followingcertaintriggersarelimitedbythelackofcomparabledatafollowingsimilartriggers across di↵erent geographic units and over time. Also, examining nationalism immediately followingnationalisttriggerswillmakeresearchersfocusonlyon“dogsthatbark”andignore “dogs that don’t bark.” Inrecentyears,progressindataminingandmachinelearninghasdemonstrateditsability to enhance reliable, timely, and comprehensive data collection that used to require intense human labor (King and Lowe, 2003). On the other, social media have desirable properties that traditional public opinion surveys cannot match, especially its ability to provide rich textdatacomposedofpostsandcommentsmadebyindividualsocialmediausersinresponse to an event in real time (Steinert-Threlkeld, 2017), as well as its ability to capture the micro- dynamics of conflict (Zeitzo↵ , 2011). Thus, to complement extant measures, I propose an alternative measure of nationalism with social media data, with the goal to portrait both the short-term dynamics and the long-term trend of nationalism both within-country over time and across-countries. Specif- ically, nationalism in a given country is measured with a daily index aggregated from data drawn from the most viable social media in the interested country, proxied by the share of nationalistic comments of all comments by citizens of that country in response to posts published by key information sources. The reason to use the share of nationalistic comments as a metric is two-fold. The abso- luteamountofnationalisticcommentscanindeedprovideasnapshotofselectednationalists, however,theshareofnationalisticcommentscanprovidethereaderswithinformationontwo 45 Volatile Nationalism | Xinru Ma broader contexts: First, if we look at the share of nationalistic comments of all comments in responsetoallpostsonagivenday, wecangaugewhichtrigger-nationalistic/foreignpolicy- related or domestic topics - the public is more likely to react to. At the micro-individual level, it is fair to assume that an individual’s total time that he spends on social media is limited, and that commenting on a post requires more careful reading of the post content, as well as more time to think over and actually post a comment. Thus, examining the share of nationalistic comments of all comments can reveal the topics to which an individual chooses to allocate his limited resource (time): If a post on a domestic social issue such as a 14-year- old singer became top star in Vietnam received far more comments than a China-related foreign policy post, and that both posts were published by the same public page, it then indicates well the extent to which the followers of this page is willing to react to nationalistic triggers. Moreover, if we look at the the share of nationalistic comments of all comments in response to China-related posts, we can then gauge how representative the nationalistic sentiment is when the public is faced with China-related trigger. Nevertheless, the absolute amount of comments in response to each post will also be provided for readers’ reference. Tobetterillustratethemethodology, InowusethecomputingofVietnamesenationalism index in the following section as an example. 2.3.1 Case Selection and Data Collection The choice of Vietnam as my case is driven by three reasons. First, the importance of social media as a source of information. In Vietnam, Facebook, as the most popular social media, has 34.7 million monthly active users in June 2015. 24 In July 2017, Vietnam has around 64 millionFacebookusers, whileHoChiMinhCitybecameoneofthetop10citiesgloballywith 24 Tuoi Tre News. “Facebook now has 30 million monthly active users in Vietnam.” June 17, 2015. http://tuoitrenews.vn/society/28733/ facebook-now-has-30-million-monthly-active-users-in-vietnam. 46 Volatile Nationalism | Xinru Ma its 14 million Facebook users. 25 Given a total population of 93 million in 2015, active users on Facebook constitute a fairly sizable representation of the entire population. Moreover, though it’s been long recognized that social media users tend to skew towards young and urban, the latest census statics from Vietnam shows that over 85.34% of Vietnamese popu- lation is younger than 54 years old with 34.9% of its total population being urban (Figure 2.3.1.1). Latest data also shows that the gap between the numbers of Facebook users in Vietnamese rural areas and cities has diminished significantly. 26 Moreover, Facebook has replaced newspaper and TV and became the most important source of daily news among Vietnamese. 27 According to a Pew survey, as many as 81 percent of Vietnamese respondents aged 18-29, or 62 percent of high income earners, use social media daily to access news. Relatedly, Vietnamese government has gradually embraced this social media tool to inform and engage the public. For instance, the Vietnamese government has launched an ocial Facebook page “Thong tin Chinh phu” (Government information), while its former Prime Minister Nguyen Tan Dung urged government ocials to embrace social media. As Vu Tu Thanh, chief Vietnam representative of the US-ASEAN Business Council, identified, “A new and younger generation of ocials have been elected to various provincial positions who are familiar with modern tools like Facebook.” 28 Second, the degree of censorship on social media. The reliability and accuracy of mea- suring nationalism with social media data is subject greatly to the degree of censorship in 25 VNExpress. “Vietnam climbs to seventh worldwide for number of Facebook users: report.” July 16, 2017. https://e.vnexpress.net/news/business/data-speaks/ vietnam-climbs-to-seventh-worldwide-for-number-of-facebook-users-report-3614034.html 26 VNExpress. “Facebook users on the rise in rural Vietnam: survey.” November 28, 2017. https://e.vnexpress.net/news/business/data-speaks/ facebook-users-on-the-rise-in-rural-vietnam-survey-3677019.html 27 VNExpress. “Half of Vietnamese get news from social media, survey finds.” January 13, 2018. https://e.vnexpress.net/news/business/data-speaks/ half-of-vietnamese-get-news-from-social-media-survey-finds-3698212.html 28 Boudreau, J. 2015. “Why Vietnam’s Communists Are Learning to Like Facebook.” Bloomberg. October 21, 2015. https://www.bloomberg.com/news/articles/2015-10-22/ vietnam-communists-turn-to-once-blocked-facebook-to-hone-message?cmpid=yhoo.headline 47 Volatile Nationalism | Xinru Ma Figure 2.3.1.1: Facebook User Profile and Population Structure in Vietnam (2017) Data source: Nguyen (2017) di↵erent countries. In general, a severer degree of censorship will bias the result by under- estimating the level of nationalism, as nationalism will not be observed if social media users don’t even succeed in making their posts online. This is exactly why using social media such as Sina Weibo in China to measure nationalism can be extremely challenging. While Viet- nam also has a long history of censorship, there are two reasons that make using Facebook data in Vietnam less susceptible to this concern, or at least makes the degree of censorship and the resulting direction of bias more tractable. From the technical perspective, unlike China, which blocks websites at the internet ser- vice providers (ISPs) level, censorship in Vietnam is done at the domain name system (DNS) level, thus being more of “a smoldering bamboo fence - an inconvenience more than an out- rightprohibition.” 29 Thismeansthatthegovernmentsimplytellsserviceproviderstoredirect their servers away from sites, which makes it easier to circumnavigate than the censorship in China, where censors that work inside of Chinese social media firms and government read individual posts and decide which ones to publish or take down (King et al., 2014). From the procedural perspective, except for content that violates Facebook’s “Community Stan- dards”, 30 government-initiated censorship to remove content on Facebook are done ex post, 29 Public Radio International. “Facebook in Vietnam: Why the block doesn’t work.” October 4, 2010. https://www.pri.org/stories/2010-10-04/facebook-vietnam-why-block-doesnt-work 30 For details, see https://www.facebook.com/communitystandards 48 Volatile Nationalism | Xinru Ma with governments sending requests to Facebook for review. These requests are published twice a year by Facebook. For the period investigated in this chapter, Facebook reported there were no Vietnamese government requests for content restrictions. 31 ThethirdreasontouseVietnamasamaincaseistheempiricaldeficiencyintheliterature. Empirically, with few exceptions (Vu, 2007, 2010, 2014), studies on nationalism in East and Southeast Asia almost exclusively focus on China, thus lacking cross-regional perspectives onnationalism, aswellasexternalvalidityformost, ifnotall, concernedactorsintheregion. By measuring how strong nationalism is in Vietnam and how the elites in Vietnam react to and manage nationalism, this dissertation hopes to shed light on the more generalizable e↵ects of nationalism on foreign policies in the region. Nevertheless, the major component of Vietnamese nationalism index is the share of na- tionalistic comments of all comments in response to posts published by four Facebook public pages in its Vietnamese platform: Bao Thanh Nien (Youth), and Tuoi Tre (Youth), two of the most circulated daily newspapers in Vietnam; Quan Doi Nhan Dan Viet Nam (VietNam People’s Army), the ocial public page of Vietnam People’s Army, and BBC Tieng Viet (BBC Vietnamese), BBC’s branch in Vietnam that publishes news in Vietnamese. These public pages are chosen due to 1) their popularity as sources of information among Viet- namese, and 2) their respective ideological bias in the selection and coverage of events and stories. Representing reformist, commercial, ocial (party/military), and Western sources of information, these pages capture a broader readership than any of them by its own, and should exhibit di↵erent patterns in levels of nationalism if we assume that users prefer to follow public pages they perceive to be ideologically close to their own position and are exposed to the information selectively. Using these public pages as firehose, I use “Netvizz,” 32 atoolthatextractsdatafrom 31 Facebook. 2017. Government Requests Report January 2014 – June 2014. Available at https:// transparency.facebook.com/government/. (Accessed on May 16, 2017) 32 Ever since Facebook changed its “Page Public Content Access” permission, this app may stop working 49 Volatile Nationalism | Xinru Ma Figure 2.3.1.2: Summary Statistics of Four Public Pages the Facebook platform, to retrieve all posts published by each public page, no matter which topic that post is about. I then collected all the comments in response to these posts, resulting in a dataset containing every post, every comment, every comment to comment and key statistics regarding each post and comment. The data collection procedure covers from April 1 to June 30 in 2014 and 2018, respectively, roughly one month prior to and over one month following two largest anti-China protests in recent bilateral history (May 2014 and June 2018 protests). The final outcome is a database of 931 posts and 95,555 comments from four public pages. 33 Figure 2.3.1.2 summarizes the key statistics from these four public pages. Tomeasurebothdimensionsofnationalism,itisnecessarytoclassifyeachcommentbased on whether it is nationalistic and how nationalistic it is. To achieve this, I randomly selected asampleof10%oftotalcomments(9,555 comments), which I used to train a machine at any time. For more details, see https://apps.facebook.com/107036545989762/ 33 The initial dataset retrieved via “Netvizz” contains 99,889 comments (observations). In pre-processing step, observations that have no comment content after removing emoji in translated comments (4,310 com- ments)aswellascommentsthatarenotinVietnamese(24comments)wereremovedfromtheinitialdataset. 50 Volatile Nationalism | Xinru Ma learning classifier that predicts the category to which all comments in the final dataset correspond. Specifically, I first used Google Translate API to translate all the posts and comments from Vietnamese into English. I then adopted a dichotomous rule of classification by coding each comment as nationalistic (1) or not (0). 34 Following Johnston (2017), Vu (2014)and Vu (2007), a comment is hand-coded as na- tionalistic if it falls into at least one of the following four pillars: • Pride in the nation-state; • Blind or naive support for the nation-state; • Negative evaluation of relevant out-groups; • Defense of national autonomy and territory unity; Specifically, the first three pillars are the conner-stone components of nationalism used by Johnston (2017)inhisseminalpieceonthatmeasureswhethernationalismisrisingor not in China. The fourth pillar is added due to the importance of national autonomy and territorial unity - the continuous struggle against foreign invasion and colonization, as well as the lasting maritime and land-border disputes - in the formation of nationalism in mod- ern Vietnam. 35 Substantively, this coding scheme biases in favor of the ”rising-nationalism” argument as more comments would be coded as nationalistic with this broader definition of nationalism than otherwise. However, the fact that the observed share of nationalis- tic comments remains low further supports my argument that nationalism in Vietnam is overestimated, an argument that I will elaborate more in the following section. Table 2.3.1.3 surveys some examples of the typical nationalistic comments based on this coding. Table 2.3.1.4 surveys some examples of the more nuanced nationalistic comments. 34 The alternative is to code the comment along the “nationalistic-unnationalistic” spectrum, but it is dicult to implement as coding instruction for coders to put a value along a spectrum can be obscure. 35 Vu (2014) provides a detailed extraordinary account of what Vietnamese national identity is and how it has evolved. For more details on the discontinuity between traditional nationalism (prior early twentieth century) and modern nationalism, see Vu (2007). 51 Volatile Nationalism | Xinru Ma Table 2.3.1.3: Examples of Typical Nationalistic Comments Table 2.3.1.4: Examples of Nuanced Nationalistic Comments 52 Volatile Nationalism | Xinru Ma Appendix 2.6.1 reports further details of the coding scheme and estimates of inter-coder reliability. 2.3.2 Training Machine Learning Classifiers Using the dataset of labeled comments, I then constructed machine learning classifiers that allow us to estimate the probability that each comment is nationalistic or not. The analysis is done in three steps: text pre-processing, training and validation of the classifiers, and applicationtotheentirecorpus. UsingXGBoostasthekeyalgorithm,theaccuracy,precision and recall of this method were 0.94, 0.41 and 0.44, respectively. As an additional validation for the automated classification method, the terms that the model identified as being most associated with each category of comment are also examined, and they correspond to the ex ante expectations based on human coding. For instance, the classifierpredictsasnationalisticthosecommentsthatcontaininsultsandexpletivesdirected at China, with words such as “rob”, “fatherland”, “patriot”, “dog”, “i love”, “proud”, “pro- tect”, “uncl ho”, “invad[e]”, “blood”, “sovereignti”, “territori”, and “sea”. Figure 2.3.2 lists top key features associated with nationalistic comments. In Appendix 2.6.2,Iusednon-translatedcommentstotrainthemodel,whichyieldsto similartopfeatures. Onenewfeaturethatwasn’tcapturedinthemodelthatusedtranslated comments is “khua,” an inferior word used by Vietnamese to refer to Chinese. Figure 2.3.2: Top Features for Nationalistic Comments 53 Volatile Nationalism | Xinru Ma This model is then applied to predict the labels for all the comments in the dataset. The share of nationalistic comments of all comments is computed for each day of the investigated time frame. The baseline level of nationalism is measured as the aggregated share of nation- alistic comments, while the volatility of nationalism is calculated as the di↵erence between the peak and the baseline level of nationalism. 2.3.3 Limitations of Social Media Data and Validation Nevertheless, using Facebook data to measure baseline and volatility of nationalism still poses particular challenges. The first challenge is the reliability of historical data retrieved on social media. Though the Facebook transparency report can help us track censorship requestsmadebytheVietnamesegovernment, westillneedtobecautiousoftheroleofusers in content censorship. Users of social media can edit or even delete their posts, especially when the topic is sensitive. Thus, any data that is not retrieved in real time might be biased by underestimating the nationalism at the moment when certain event unveils. To address this challenge, I chose to examine comments to public pages rather than individual posts, relying on the assumption that the former has a lower chance of being edited or deleted afterwards. If this assumption doesn’t hold, that is, if the users indeed deleted their nationalistic comments afterwards, then the result presented in this chapter is biased in terms of underestimating the level of nationalism. Second, citizens of other countries, especially foreigners in Vietnam, can also follow and comment on these public pages as sources of information. Thus, including all comments without scrutiny will increase the total amount of comments, thus biasing the results by un- derestimatingtheshareofnationalisticcomments. Toaddressthischallenge, onlycomments in Vietnamese are used to compute the final index. Thoughthesejustificationsarenotentirelysatisfying,itisimportanttoassessthevalidity of measuring nationalism with social media data in context with measures that we have to 54 Volatile Nationalism | Xinru Ma date. So far, there’re two major methods to gauge the level of nationalism: 1) public opinion surveys, and 2) qualitative description of nationalistic protests along with interviews with protesters. As discussed earlier, results from surveys may be biased due to preference falsification (Kuran, 1997), social desirability bias (Van de Mortel et al., 2008), di↵erent mind-setsandcontexts(Matfield,2015),whilethegapbetweenexpressedattitudeandactual behavior is often unknown. In the context of measuring nationalism, it can be one thing for survey respondents to express their attitude on whether “they should be prepared to sacrifice his/her personal interest for the sake of the national community/society,” and quite another thing for them to join a protest or fight for the country when crisis emerges. (See Appendix 2.6.3 for results on measuring nationalism in Vietnam with traditional cross- sectional time-series surveys.) Survey respondents can also be primed with the “political correct answer” that one should sacrifice personal interests for the sake of country. On the other hand, qualitative examination of the protest as well as interviews with protesters, though quite informative, can su↵er from self-selection and self-reporting bias: It is often hard to researchers to gauge how representative the protesters are, rendering the direction of bias unknown. By contrast, commenting on a social media post can incur some real-life costs - though still cheap - to a given social media user in times of real crises rather than a hypothetical scenario in a survey, without priming the respondents “the politically correct answer.” In addition, the availability of data on social media users’ demographics allows us to compare that of the social media users with population demographics from the census data, thus allowing to us better interpret the representativeness of the result, as well as the direction of bias. Apart from adopting these ex ante measures, I also adopted one ex post method to validate the result. Specifically, I examined the share of nationalistic comments on posts published by these pages on Vietnamese National Day on September 2, 2014, in order to validate the peak level of nationalism. Results of this validation method is presented in the 55 Volatile Nationalism | Xinru Ma following section. In short, despite all the limitations, data gleaned from Facebook in Vietnam provides a reasonableapproximationofthelevelsofnationalismonadailybasis, whichhelpsusidentify the baseline level of nationalism when there is no trigger, and the volatility of nationalism that captures how much nationalism could diverge from the baseline in light of nationalistic triggers. 2.4 Low Baseline but High Volatility Levels of Nationalism in Vietnam This section presents the empirical results of nationalism in Vietnam in 2014 and 2018, re- spectively, identifythetypeofnationalisminVietnamineachperiod, andprovidevalidation for the results, so as to constitute a within-country comparison of nationalism in Vietnam. It’s important to note that I do not maintain that the levels and the type of nationalism in Vietnam identified in this chapter can be extended beyond the time frame investigated here. 2.4.1 Anti-China Protest in May 2014 I first identify several critical junctures of the 2014 anti-China protest in Vietnam. On May 2, 2014, China placed an oil-drilling rig - Haiyang Shiyou 981 - into a disputed area in the South China Sea. As a response, the Vietnamese government immediately sent frogmen and other underwater agents to the area, and dropped obstacles such as fishing nets and floating objects into the waters, with the aim to disrupt the rig’s placement and operations. However, this incident received little attention from the public until May 11, when hundreds of people gathered and protested against China in central Hanoi and Ho Chi Minh city. By May 13, the crowd swelled to about 20,000, some of whom looted and set fire to more than 15 factories owned by China, Taiwan and Hong Kong in three neighboring provinces of Ho 56 Volatile Nationalism | Xinru Ma Cho Minh city. This riot was soon cracked down within about one and a half day. By May 18, a planned second protest in Hanoi and Ho Cho Minh city was subdued, with more than 500 people detained for looting and arson. Observing Vietnamese government’s response, the Haiyang Shiyou 981 kept its operation in the disputed waters, and was only moved back to Hainan island on July 16, 2014, two months after the demonstrations. Figure 2.4.1.1 plots the results from April 1 to June 30, 2014. The black solid line in the upper graph denotes the share of nationalistic comments of all comments in response to posts published on a given day. The red bar in the bottom graph denotes the total amount of nationalistic comments in comparison to total comments (grey bars). The blue dashed lines indicate the linear trend of the share and the amount of nationalistic comments during this period. Three observations stand out: First, the daily share of nationalistic comments on Face- book exhibits a stable trend with a mean of 2.7% over time, indicating a low and stable baseline level of nationalism in Vietnam. Second, Vietnamese social media users are highly volatiletonationalistictriggers, specifically, theanti-Chinaprotest, asboththetotalamount ofcommentsandtheamountofnationalisticcommentspeakedimmediatelywhentheprotest broke out. The amount of total comments peaked on May 14 (6,744 comments), over 32 times of average daily total comments in times without triggers (208 comments per day). The amount of total nationalistic comments also reached its peak on May 14 (185 com- ments), over 15 times of the average daily nationalistic comments in times without triggers (12 comments per day). 36 Third, despite the high volatility, even at the peak of the protest, nationalistic sentiment is not as high as we expected, as only 185 comments out of 6,744 (or 2.7%) are nationalistic. By contrast, the highest share of nationalistic comments (11.9%) documented in this 36 The average daily amount of total comments and average daily amount of nationalist comments are calculated as the mean of respective amounts excluding dates from May 11 to May 18, 2014. 57 Volatile Nationalism | Xinru Ma Figure 2.4.1.1: Total Amount and Share of Nationalistic Comments in Vietnam (2014) period was on May 28, 2014, driven by responses to two posts, one of which may potentially incur nationalistic responses. The first post was about the most popular bank chung rice cake in Saigon, which received 19 comments in total (0 is nationalistic). The second post 58 Volatile Nationalism | Xinru Ma covered the press conference in which the US State Department confirmed China as the provocative party in the South China Sea in terms of putting drilling rigs into Vietnam’s waters. This post received 23 comments in total, while 5 of them were nationalistic. The fact that the share of nationalistic comments - as well as the absolute amount of nationalistic comments - is low also holds even if we examine responses to China-related posts only. Figure 2.4.1.2 denotes this result. From the graphs, we can see that even the post itself is about China, majority of the followers of these pages don’t necessarily link the issue with nationalism. More specifically, the peak of the absolute amount of total comments - as well as the absolute amount of total nationalistic comments - occurred on May 7, 2014, when a series of 7 posts were published on the same day regarding how Chinese vessel actively rammed into Vietnamese ships during the confrontation regarding the HD 981 rig. These 7 posts in total received 2,172 comments, with 127 comments being nationalistic (5.8%). By contrast, the post on a 14-year-old singer with two German-Thai bloodlines that became top star in Vietnam, published on April 15, 2014, received a total of 3,381 comments alone. This comparison shows further that the infrequent nationalistic comments and total comments in response to China-related topics are not because people don’t comment on Facebook. These findings contradict the conventional wisdom of rising nationalism in Vietnam. If we examine nationalism only by looking at the scale of protests in the immediate after- math of certain trigger, one probably would concur that nationalism in Vietnam is rising to an unprecedented peak in May 2014. Indeed, “surging” nationalism is often examined and evidenced in extant literature in the immediate aftermath of certain triggers, which is further reinforced by the stories of people taking to the streets, with images of protestors marching, chanting, and waving banners. This approach, though vividly captures the spirit of nationalism, makes researchers focus only on the vocal minority, rather than the silent majorities, thus biasing the estimation of nationalism in favoring the “rising-nationalism” 59 Volatile Nationalism | Xinru Ma Figure 2.4.1.2: Responses to China-related Posts in Vietnam (2014) argument. By contrast, the low baseline levels of nationalism prior to and after the protest, along with the relatively low level of nationalism even at the peak of the protest, signals an over-estimation of nationalism in East and Southeast Asia. 60 Volatile Nationalism | Xinru Ma The over reporting of the protests in foreign media might also contribute to this overes- timation of nationalism. As one Japanese citizen that works in Ho Cho Min city tweets, “I cannot watch Japanese news, so I do not know how severe the Vietnamese riots are being reported. In reality, for now, Ho Chi Minh City seems to only have had a bit extra police force.” Another twitter user states that “it is not so noisy in Vietnam. [The protest] is being reported too much in Japanese media.” Other tweets also noted that despite the alert of potential demonstration in the vicinity of the Opera, Ho Chi Minh city is just as usual, and that people in Ho Chi Min are pretty calm. A qualitative examination of the comments further shows how Vietnamese public quickly connected the protest with domestic grievances. On May 14, the next day when the protest turnedviolent,theamountofcommentsweredrivenbythefollowingthreeposts: aphotoand avideopostedby BBC Vietnamese showing protesters in Binh Duong chanting “Vietnam, Vietnam, Vietnam” and “objections, objections, objections” to China in the South China Sea,andapostfromQuan Doi Nhan Dan Viet Nam (VietNamPeople’sArmy)claimingthat there were sacrifices after the rally in Binh Duong due to improper extremists. These three posts boosted the total amount of comments as well as the total nationalistic comments, with a lot referring to the 1979 Sino-Vietnamese war, or the “Chinese invasion.” However, despite the increased amount of nationalist comments, a large portion of the comments were crowdsourcing, as they were questions and descriptions of the protests, with no specific sentiment expressed in the comments or demands on government policies. Other comments focused on the need to stop violence as the protests went violent. For instance, Tuoi Tre (Youth) reacted to the video by stating that “property destruction is not nation- alistic.” Some posted images of stop signs with the words “stop violence” on it. Many advocated tougher actions by the police to punish the violent protesters, whom they called “uncivilized” and “na¨ ıve.” For instance, one comment reads, “Patriotism + foolishness + sabotage = sabotage,” while another states that “ignorance + cretinism = great vandalism.” 61 Volatile Nationalism | Xinru Ma The following comment shows further such critical attitude towards the protesters: “The people in Binh Duong are not doing any good demonstrations...they are all short-minded people taking advantage of this stressful time to cause disorder on the roads a↵ecting the lives of Vietnamese soldiers...You guys are wild, uncultured, uninformed, you don’t love VN...You are destroying the beautiful image of VN.” Another comment reads, “Be a patriot in an educated and cultural way. Don’t be ignorant + enthu- siastic = destructive. Most afraid of stupid people.” Other comments expressed upset about the fact that Vietnamese people are willing to protestoverdisputedwaterratherthantheirworkingconditions,thedroughtintheprovince, and domestic social issues such as unemployment or income inequality. For instance, one commentsreads“emotionlesssociety,peoplearelessinterestedinthecountry’spoliticalecon- omy.” Some suspected the demonstrations were driven by workers’ greed or their grievances againstforeignfirms. OthersurgedVietnamesegovernmenttochangeitspracticesonhuman rights for greater international support. For instance, one comment states: “The current measures of Vietnamese people such as protesting against China to set up HD981 rig on the waters of Vietnam are not e↵ective. It is even counterproductive....In my opinion: To ask China to move the HD981 platform out of the Vietnamese waters, there must be help from the international community. To get that, Vietnam needs to change some issues in political institutions that international opinion is concerned and not satisfied. That is: Vietnam needs to expand democracy, add to the law of freedom of the press, freedom of speech...Especially, it is necessary to adjust to the observance of human rights law.” Others expressed anti-regime sentiment more explicitly. For instance, one comments reads, “I love the country but do not believe in the Party,” while another says, “It is better 62 Volatile Nationalism | Xinru Ma to die for the fatherland, not to die for blood bureaucrats!” One comment quotes Mark Twain to express the attitude towards the regime, “Always loyal to the Fatherland. Only loyal to the government when it deserves it. Mark Twain.” Another comment states, “If you ask a million Vietnamese people that whether you love the disputed waters, one million people will answer: Yes. If you ask them whether you are satisfied with the leadership, the Party, or the government of Vietnam, then nine hundred thousand people will answer: No.” 2.4.2 Anti-China Protest in June 2018 Nevertheless, with Vietnam during May 2014 being the only case, it’s still arbitrary to label Vietnam without being able to compare it to other countries or Vietnam at other times. A baseline of 2.7% nationalistic comments might actually be really high, cross-nationally, and apeakof12%(5timesintermsofshareor15timesintermsofabsoluteamount)mightbe really low compared to other places or times. Recently in June 2018, another wave of anti-China protest broke out with dozens being arrested and hurt across Vietnam in violent demonstrations. 37 What led to these protests were the Special Zone Act, a law that would create “special economic zones” (SEZs) with the goal of sparking investment and economic reform. However, the unclear prospect that this act allegedly would hand land over to Chinese investors provoked a flood of angry demonstrations, with protesters holding placards that read “No Special Zone - No leasing land to China - Even for one day!”(Fawthrop, 2018) The protests started in Ho Chi Minh City and Hanoi but soon spread to towns in six provinces in Vietnam. The following section applies the same method to this period, with the hope to trace the baseline and the volatility of nationalism in Vietnam in four years after the last wave of 37 “Anti-Chinese protests in Vietnam fuelled by fears at home carry risk.” South China Morn- ing Post. June 26, 2018. https://www.scmp.com/comment/insight-opinion/article/2152442/ anti-chinese-protests-vietnam-fuelled-fears-home-carry-risk 63 Volatile Nationalism | Xinru Ma anti-China demonstrations. Figure 2.4.2.1 demonstrates the result. Figure 2.4.2.1: Total Amount and Share of Nationalistic Comments in Vietnam (2018) Three observations: First, in 2018, Vietnamese social media users are still highly volatile 64 Volatile Nationalism | Xinru Ma to nationalistic triggers, as both the amounts of total comments and nationalistic comments peaked the second day when the protests went violent. The absolute amount of total com- ments (6,502 comments) is again over 19 times of average daily total comments in times without triggers (335 comments per day). The absolute amount of nationalistic comments also reached its peak on the same day (127 comments), over 8 times of the average daily nationalistic comments in times without triggers (5 comments). 38 Third, despite the high volatility, even at the peak of the protest, nationalistic sentiment is not as high as we ex- pected, as only 127 comments out of 6,502 (or 2%) are nationalistic. These findings further support that nationalism in Vietnam in 2018 continues to have a low baseline and high volatility. They also show that over time, nationalism in Vietnam has been decreasing - rather than increasing - since 2014. Though China’s behavior continues to trigger surges of nationalistic responses during the protests in comparison to times without triggers, the absolute amount of nationalistic comments (185 in 2014 versus 127 in 2018) as well as the share of nationalistic comments (highest as 12% in 2014 versus 6% in 2018) both decreased when the public was faced with a similar trigger in 2018. Comparedtothe2014anti-Chinaprotest,commentsinresponsetotheanti-Chinaprotest revealed stronger anti-Vietnamese government sentiment in 2018. Many comments accused the Vietnamese Communist Party as deliberately selling water to the Chinese Communist Party, and expressed disappointment towards the regime. For instance, one comment reads, “...I love patriotism, but I can’t love the party, a bunch of corruption that sells water.” Another comment reads, 38 The amounts of average daily total comments and average daily nationalist comments are calculated as the mean of comments amount excluding dates from May 9 to May 12, 2018. 65 Volatile Nationalism | Xinru Ma “Why high-ranking Party cadres are corrupt (eat up people’s land)...people speak out strongly against social networks, so the State Party seeks to lock people’s mouths as national policies.” 2.4.3 Validation The above analysis identifies that despite the surges in times of nationalistic triggers, the highest documented share of nationalistic comments is less than 12%, which is far lower than the extant literature would have assumed. To validate this, I computed the share of nationalistic comments on Vietnamese national day on September 2, 2014, on the same public pages. If the model is correct, we should see asimilarshareofnationalisticcomments,withtheassumptionthatnationalistsentiment expressed on national independence days constitutes a peak level of nationalism. Surprisingly, only the public page of Quan Doi Nhan Dan Viet Nam (VietNam People’s Army) had a post about Vietnamese National Day, which received 3 comments in total. Of the three comments, only 1 expressed the pride of being Vietnamese, signaling the share of nationalistic comments as 1/3 at best. Given the amount of comments on this particular post is so little, I further examined all the posts made by all public pages from September 2 to September 4, 2014. The post received most comments was about a rare color video made by Japanese filmmakers on President Ho Chi Minh’s National Funeral. Of the 28 comments it received, 11 (39%) of them expressed their love for “Uncle Ho” and their pride of being Vietnamese, while others were about the rarity and the value of the film. Given the followers ofthispublicpage-mostlysoldiers,veteranandcitizensinterestedinnationalsecurityissues -canbemorenationalisticthantheaveragepublic,itisreasonabletoassumethatthepeak level of nationalism of the entire population is lower than the result from this page. This provides support for the model fit. 66 Volatile Nationalism | Xinru Ma 2.5 Discussion To date, existing literature tends to focus on describing the origins and elements of certain national identity, thus rendering the essence of nationalism as moving target. Moreover, nationalism is often measured in the immediate aftermath of certain triggers, thus failing to provide a baseline for how “nationalistic” the nation is when there are no triggers. Such single-event or single-country snapshots tend to bias the influence of nationalism on foreign policy. Without knowing the baseline level of nationalism, how do we know whether coun- tries are becoming more nationalistic over time? Why do we say some countries are more nationalistic than others at a given time? This chapter seeks to fill the gap by o↵ering an alternative way of conceptualizing and measuringnationalismthatdepictsboththelong-termtrendandtheshort-termdynamicsof nationalism in a given country. Yet, its contribution is not only methodological, it is concep- tual. Theoretically, it disaggregates nationalism into two conceptually distinct dimensions: “baseline” nationalism, namely, how nationalistic a nation is when there’re no triggers, and “volatility” of nationalism, namely, how a nation responds to nationalist triggers. Some individuals (countries) have very volatile nationalism: they experience surges in observable demonstrations of nationalism in response to very small stimuli. Other individuals (coun- tries) have very involatile nationalism: even large triggers don’t cause a response. I maintain that it is the baseline level of nationalism and the varied degrees of volatility that jointly di↵erentiate individuals and nations in their levels of nationalism. This conceptualization thus demonstrates how the question whether nationalism is rising is ill-defined. It also forces usto crystalizewhosepreferencesthe leaders careabout when making foreign policies. Ifthe majorityofthesocietyiscomprisedofinvolatilenationalistswhodon’trespondtonationalist triggers, would the leaders still adopt aggressive nationalist foreign policies? Empirically,thischapterappliesthesupervisedmachinelearningmethodstosocialmedia data during the 2014 and 2018 anti-China protests in Vietnam, with the aim to identify and 67 Volatile Nationalism | Xinru Ma constitute a within-country comparison of the baseline and volatility levels of nationalism. In summary, this chapter demonstrates three key findings that all challenge the “rising- nationalism” assumption: First, the baseline of nationalism in Vietnam has remained stable and low from 2014 to 2018, signaling an over-estimation of nationalism in extant literature. Second, even at the peak of the 2014 and 2018 anti-China protests, the largest and most violent nationalist protests in recent bilateral history, nationalism was not as high as we expected, which further implies that nationalism is overestimated. Lastly, though China’s behavior triggered surges of nationalistic responses in both 2014 and 2018 (low baseline but high volatility), the absolute amount as well as the share of nationalistic comments both decreased when the public was faced with a similar trigger in 2018, implying again that over time, nationalism in Vietnam has been decreasing - rather than increasing - since 2014. These findings are counter-intuitive especially if we measure nationalism in Vietnam with traditional cross-sectional time-series surveys (Appendix 2.6.3). This contrast itself is an interesting and important finding. It’s important to note that I do not maintain that the contrast alone falsifies extant studies entirely. Instead, when di↵erent sources and analytical methods lead to di↵erent conclusions about a political phenomenon, the next challenge is to determine why, rather than to claim victory for one approach over another. Lastly, the analysis in this chapter focuses on only one country - Vietnam. An ideal research design would find four countries that fit the varying levels of both dimensions of nationalism, and test all possible outcomes. However, Vietnam is widely considered a lynchpin in East Asian security, and has often been used as a key indicator of how the region is responding to China. Thus, the single case of Vietnam has both empirical and theoretical implications. Nevertheless, to address this caveat, the following chapter applies the two-dimensional conceptualization of nationalism as well as the measurement method to the Philippines. 68 Volatile Nationalism | Xinru Ma 2.6 Appendix 2.6.1 Inter-coder Reliability Due to the elusive nature of the construct studied in this dissertation, I recruited a master student from a private university in the U.S. to examine the inter-coder reliability in the initial stage of human coding. The importance of rater reliability lies in the fact that it represents the extent to which the data collected in the study are correct representations of the variables measured. Whiletherehavebeenavarietyofmethodstomeasureinterraterreliability,IuseCohen’s kappa as the main statics. Similar to correlation coecients, Cohen’s kappa can range from 1 to +1, where 0 represents the amount of agreement that can be expected from random chance, and 1 represents perfect agreement between the raters. Table2.6.1providesMcHugh (2012)’s guidance in interpreting Cohen’s kappa. After providing the coding instruction with an initial training of coding 10 comments, the second coder coded a random sample of 270 comments from the entire corpus. The Cohen’sKappafor2coders(unweighted)is0.77(z-score12.6,p-value0),indicatingadequate agreement among the raters. Table 2.6.1: Interpretation of Cohen’s Kappa 69 Volatile Nationalism | Xinru Ma 2.6.2 Training Classifiers with Original Comments in Vietnamese Given Vietnamese is not the first language of both coders, the results reported in the main textarederivedfromusingtranslatedpostsandcommentsbyusingGoogleTranslateAPI.In thissection, Iuseoriginalnon-translatedcommentsinVietnamesetotrainthemodel. Given Vietnamese is not available in “snowball,” the stemming algorithms used by “quanteda” to preprocessing corpus, words in the original Vietnamese comments are not stemmed, and stopwords are maintained. Nevertheless, results from using original Vietnamese comments are similar from those using translated ones. The accuracy, precision, and recall of this model are 0.85, 0.47, and 0.44. Table 2.6.2 reports the top features from the model using Vietnamese comments and their translation. Table 2.6.2: Top Features for Nationalistic (1) and Unnationalistic (0) Comments 70 Volatile Nationalism | Xinru Ma 2.6.3 Measuring Nationalism in Vietnam with Traditional Survey Data In the main text, I argued that nationalism measured with social media data contradicts results from using traditional measures of nationalism. Below are the results from two waves of Asian Barometer, by using questions that were similarly used in Johnston (2017). The results point to an extremely high level of nationalism in Vietnam, and a slightly decreaseinnationalism-thoughstillhigh-since2005. Thenextstep,insteadofadjudicating victory for one approach over another, is to determine why such discrepancy happens. 71 Volatile Nationalism | Xinru Ma Figure 2.6.3: Measuring Nationalism in Vietnam with Asian Barometer Data 72 Volatile Nationalism | Xinru Ma 3 The Calm Philippines: Nationalism with Low Base- line and Low Volatility “Chinese aggression is very good for the Philippines...and has done more to galvanize Philippine nationalism than anything in the past few years.” - The Rappler 39 “If you want, just make us a province, like Fujian?” - The Rappler 40 The case of the Philippines is particularly interesting for three reasons. First, like Viet- nam,thePhilippineshasbeeninvolvedinmultiplecompetingclaimstotheSouthChinaSeas against China. The confrontation peaked in April 2012, when China and the Philippines foundthemselveslockedinadangerousnavalstand-o↵overScarboroughShoal. In2013, this tension once again flared up as the Philippines filed a Notification and Statement of Claim with the Hague Permanent Court of Arbitration, challenging China’s ambiguously-defined nine-dashline. Indeed,conventionalwisdomhasbelievedthat“Chineseaggression...hasdone more to galvanize Philippine nationalism than anything in the past few years.”(Esmaquel II, 2012) As Dor (2015) argues, “the Aquino administration has taken action to help further strengthen nationalistic tendencies in the Philippines over the disputes with China. Em- bodying the saying ‘never waste a good crisis,’ President Benigno Aquino III has latched on to the outcry of nationalists, whipping up nationalism through vocal decrees and executive orders.” 39 “China’s aggression ’very good’ for PH.” The Rappler. July 20, 2012. https://www.rappler.com/ nation/8872-china-s-aggression-very-good-for-ph 40 “Duterte jokes: Why not make Philippines a province of China?” The Rappler. June 28, 2018. https: //www.rappler.com/nation/196426-duterte-philippines-province-china 73 Volatile Nationalism | Xinru Ma However, is nationalism in the Philippines really rising? Has China’s behavior on the maritime issues really aggravated Philippine nationalism? If this is the case, why did the public in the Philippines elect a President who even joked to China that “if you want, just make us a province?”(Heydarian, 2018)Thus,whethernationalismisrisingornotandwhy it is so in the case of the Philippines is empirically puzzling in itself. In addition, Facebook enjoys remarkable popularity among the public in the Philippines. A recent survey indicates that the Philippines rank No. 1 globally in terms of time spent on socialmedia. 41 DatafromFacebookcouldthenprovideamorebalancedrepresentationofthe populationinthePhilippines,thushelpingusgaugemoreaccuratelythelevelsofnationalism in the country. In terms of cross-country comparison, the Philippine case constitutes a nice cross-country comparison of nationalism with that of Vietnam, thus sheding light on how maritime disputes with China has shaped di↵erently, if ever, the rise and fall of nationalism initsneighboringcountries,aswellashowcountries,despitebothbeinginvolvedinmaritime disputes with China, react di↵erently to nationalist triggers. Lastly,thecaseofthePhilippinesisimportantfordelineatingtheconfoundingassociation between volatility and the citizens’ fear of potential punishment. One potential push-back in measuring the volatility dimension of nationalism is the regime type and its associated repressive capacity. Literature on repression and political mobilization points out that many regimes, especially authoritarian ones, use acts of repression to suppress dissent and to exclude large portions of their populations from mobilizing for regime or policy change. (Davenport, 2014, 2007; Carey, 2006; Lichbach, 1987) In the context of this dissertation, if the government threatens to repress (or has repressed in the past) the citizens who are vocal in expressing nationalism either via words on social media or via deeds by taking to the street, thenitmaybiastheresultbyunderestimatingthevolatilitydimensionofnationalism, 41 Inquirer.net. “PH world’s No. 1 in terms of time spent on social media.” January 24, 2017. http: //technology.inquirer.net/58090/ph-worlds-no-1-terms-time-spent-social-media 74 Volatile Nationalism | Xinru Ma as citizens may not be willing to respond to nationalist triggers due to the fear of potential punishment. However, this push-back relies on a key assumption that repression negatively a↵ects participation and silences social movements, an assumption that is still heatedly debated in the repression literature itself. Davenport (2007,p. 8)referstothemixedfindingson the e↵ect of repression on dissent as the “punishment puzzle,” as such e↵ect can be positive (Francisco, 1996; Lichbach and Gurr, 1981), curvilinear (Muller, 1985; Brockett, 2005), or nonexistent(GurrandMoore,1997). Evenifwesetthisdebateasidebyassumingrepression indeed decreases dissent, we would expect to observe a higher volatility level of nationalism in the Philippines compared to the case of Vietnam, as democracies are least likely to deploy repressive behavior. (Carey, 2006) However, the fact that nationalism in the Philippines has lower volatility level highlights that the observed low level of volatility is not necessarily due to the fear of potential punishment from the regime. The following sections explain the case selection and data collection process, the model, and the empirical findings. To put the cards on the table, the case of the Philippines reveals three key insights: First, similar to Vietnam, nationalism in the Philippines exhibits a low and stable baseline, which again indicates the overestimation of nationalism in extant literature without considering a baseline measure; Second, di↵erent from nationalism in Vietnam that has high volatility, the Philippine case demonstrates a low volatility: even largenationalistictriggersdon’tcausemuchresponse. Lastly, whiletheamountofcomments peaked due to China-related issues in Vietnam, peaks of comments in the Philippines are mostly driven by domestic issues such as LGBT and same-sex marriage, President Duterte’s policies on illegal drugs, corruption and domestic terrorism, and surprisingly, trivial issues such as a local cafe o↵ered napping and showers services to its customers. 75 Volatile Nationalism | Xinru Ma 3.1 The July 2016 Tribunal Ruling on the South China Sea Despite the sustained tension in the South China Sea, overlapping claims and incidents have been mostly dealt within a bi-lateral setting rather than a legal approach, until the Philippines initiated arbitration proceedings against China under Annex VII to the United Nations Convention on the Law of the Sea on January 22, 2013. Specifically, the Philippines stated that it seeks an Award that: • “declares that the Parties’ respective rights and obligations in regard to the waters, seabed and maritime features of the South China Sea are governed by UNCLOS, and that China’s claims based on its ‘nine dash line’ are inconsistent with the Convention and therefore invalid; • determineswhether,underArticle121ofUNCLOS,certainofthemar- itime features claimed by both China and the Philippines are islands, low tide elevations or submerged banks, and whether they are capable of generating entitlement to maritime zones greater than 12 M; • enables the Philippines to exercise and enjoy the rights within and beyond its exclusive economic zone and continental shelf that are es- tablished in the Convention.” 42 ThePhilippineswelcomeditsfirstsuccessonOctober29,2015,whenthearbitraltribunal ruled that it has jurisdiction over the case, taking up seven of the 15 submissions made by the Philippines. This decision “firmly rejected one country’s non-appearance and refusal to participate...[and] disposed of China’s complaint that the Philippines’ unilateral initiation of arbitration proceedings is an ‘abuse of legal process’ in the absence of Chinese consent and participation.” (Batongbacal, 2015) Almost a year after the Hague accepted the case, a five-judge tribunal released its long- awaited award on July 12, 2016. The tribunal invalidated Beijing’s claims to historic rights 42 PCA Case No. 2013-19 Award In The Matter of The South China Sea Attribution Before An Arbitral Tribunal Constituted Under Annex VII To 1982United Nations Convention On The Law of the SeaBetween The Republic of The Philippines And The People’s Republic of China. July 12, 2016. p. 11. https: //pcacases.com/web/sendAttach/2086 76 Volatile Nationalism | Xinru Ma throughout the nine-dash line, and surprised many observers by ruling on the legal status of everyfeatureintheSpratlyIslandsraisedbythePhilippines. OfthesevenSpratlysoccupied by China, the court ruled that Johnson Reef, Cuarteron Reef, Fiery Cross Reef, and Gaven Reef are rocks, while Hughes Reef and Mischief Reef are below water at high-tide, thus generating no maritime entitlements of their own (CSIS, 2016). The award constitutes “a nearly across-the-board win for the Philippines, and a searing verdict on the lawfulness of China’s artificial island construction and other actions in the South China Sea.”(Williams, 2016) In addition to the award, multiple events concerning China and the Philippines also attracted the media’s attention around the same time. June that year marked the 41st An- niversary of China-Philippines Diplomatic Relations and the 15th Chinese-Filipino Friend- ship Day. The South China Sea disputes also became the central topic discussed at the Asia Security Summit held in the same month that year. Meanwhile, a joint Philippine-China task force against illegal drugs was proposed, with China reportedly o↵ering help in the construction of four mega drug rehabilitation and treatment centers. (Mirasol, 2016) 3.2 Case Selection and Data Collection This period thus constitutes a nice time frame for gauging the levels of nationalism in the Philippines. To make the results comparable to those in the previous chapter, I retrieved all comments in response to posts published by four Facebook public pages in the Philippines (Figure 3.2): Philippine Star,and The Daily Inquirer,twoofthemostcirculateddaily newspapers in the Philippines; Philippine Navy,theocialFacebookpublicpageofthe Philippine navy, and CNN Philippines, CNN’s branch in the Philippines that publishes news in English and Tagalog. Again, as in the case of Vietnam, these public pages are chosen due to 1) their popularity as sources of information among the Philippine social media users, and 2) their respective 77 Volatile Nationalism | Xinru Ma ideologicalbiasintheselectionandcoverageofeventsandstories. Thedatacollectionlasted from June 1 to August 31, 2016, covering roughly one month prior to and after the tribunal’s award. The final dataset contains 421 posts and 58,970 comments. Figure 3.2: Summary Statistics of Four Public Pages 3.3 Training Machine Learning Classifiers To train the machine learning classifiers, I randomly selected a sample of 576 comments, and coded each comment as nationalistic (1) or not (0), following the same set of coding schemes based on the following four pillars: (1) Pride in the nation-state; (2) Blind or naive support for the nation-state; (3) Defense of national autonomy and territory unity; and (4) Negative evaluation of relevant out-groups. All comments and posts are translated from Tagalog into English with Google Translate API. Table 3.3.1 and Table 3.3.2 survey some examples of the nationalistic comments. Using the dataset of labeled comments, I then constructed machine learning classifiers that allow us to estimate the probability that each comment is nationalistic or not. The analysis is done in three steps: text pre-processing, training and validation of the classifiers, 78 Volatile Nationalism | Xinru Ma Table 3.3.1: Examples of Typical Nationalistic Comments in the Philippines and application to the entire corpus. Using XGBoost as the key algorithm, the accuracy, precision and recall of this method were 0.74, 0.39 and 0.33, respectively. 43 As additional validation for the automated classification method, the terms that the model identified as being most associated with each category of comment are also examined, and they correspond to the ex ante expectations based on human coding. For instance, the classifier predicts as nationalistic those comments that contain words closely associated with maritime disputes and navy capability, such as “afp” (Armed Forces of the Philippines), “navi” and “philippin navi” (navy and Philippine navy), “modern” (referring to AFP mod- 43 The low precision and recall stats are due to the small sample size of the hand-coded dataset, as well as its unbalanced structure. In future revision, it is important to increase this sample size for training and testing purposes. 79 Volatile Nationalism | Xinru Ma Table 3.3.2: Examples of Nuanced Nationalistic Comments in the Philippines ernization), “scarborough”, “island”, “sea”, “fight”, “china” and “proud”. Table 3.3.3 lists top key features associated with nationalistic comments. Table 3.3.3: Top Features for Nationalistic Comments 80 Volatile Nationalism | Xinru Ma 3.4 Low Baseline and Low Volatility Levels of Nationalism in the Philippines Figure 3.4.1 presents the empirical results on the daily volume and share of nationalistic comments of all comments in the case of the Philippines. One observation immediately stands out: compared to the massive amount of total comments, the absolute amount, as well as the share of nationalistic comments is so low, indicating a low baseline and low volatility of nationalism in the Philippines. Indeed, the highest ever daily total amount of nationalisticcommentsislessthan90,whilethehighestdailyshareofnationalisticcomments is less than 1.5%. The low baseline and low volatility can be further demonstrated if we examine how the Philippines respond to China-related posts. From June 1 to August 31, 2016, there were in total 16 China-related posts with a total of 329 comments. 44 Out of them, 11 posts were directly on the South China Sea disputes, the Hague ruling, as well as in-depth feature storiesthatinterviewedfishermenonhowthedisputesandChina’sseizureoftheislandshave impacted their life. Three were posted by the Philippine Navy, explaining the navy’s mission of “Protecting the Seas, Securing our Future!” and its capacities on territorial defense in the West Philippine Sea. The other posts were about the Filipino-Chinese Friendship Day, and the joint Philippines-China task force versus illegal drugs. Unfortunately, each post received veryfewcommentsfromthefollowersofthesepages, and theshareofnationalisticcomments is unprecedentedly low (Figure 3.4.2). The lack of response received among the Pilipino regarding China-related triggers is more striking if we compare it to how they react to domestic issues. Figure 3.4.3 shows the average comments received in the Philippines regarding non-China-related issues. While 44 These posts were coded if they contain one of the following key words in the post text: China, china, chinese, tribunal, tribune, ruling, waters, beijing, wps (denoting West Philippine Sea, the name used by the Philippine government to refer to the South China Sea), Hague. 81 Volatile Nationalism | Xinru Ma Figure3.4.1: DailyShareandAbsoluteAmountofNationalisticCommentsinthePhilippines (2016) the maximum amount of comments received by a post on China was less than 100 in the examined period, post on domestic issues easily surpassed that number, with the highest being as many as 11,395. 82 Volatile Nationalism | Xinru Ma Figure 3.4.2: Responses to China-Related Posts in the Philippines (2016) A closer scrutiny of the news topics during the peaks of total amount of comments can further reveal this pattern. Table3.4.4 shows the posts the received most comments on these days. Clearly,comparedtoissuesonChina,thefollowersofthesepagesinthePhilippinesare more volatile to domestic news, such as LGBT and same-sex marriage, URCC Fight Night (a professional mixed martial arts), President Duterte’s policies on illegal drugs, corruption and domestic terrorism, and surprisingly, trivial issues such as a local cafe o↵ered napping and showers services to its customers. 3.5 Discussion In summary, this chapter applies the two-dimensional conceptualization of nationalism and its measurement with social media data to the case of the Philippines. Similarly to na- tionalism in Vietnam, nationalism in the Philippines exhibits a low and stable baseline, 83 Volatile Nationalism | Xinru Ma Figure 3.4.3: Responses to Non-China-Related Posts in the Philippines (2016) which challenges again the assumption of “rising-nationalism” in the extant literature. On the other hand, Vietnam and the Philippines di↵er greatly on the dimension of volatility. While Vietnam has a relatively highly volatile nationalism, the Philippines demonstrate an involatile nationalism with few responses to nationalistic, especially China-related triggers. By contrast, the Philippine public is more reactive to domestic issues such as LGBT and same-sex marriage, President Duterte’s policies on illegal drugs, corruption and domestic terrorism, and issues that are closely related with their daily life. The Philippine and the Vietnamese cases thus constitute an interesting comparison. Al- though disputes in the South China Sea do cause bilateral tension periodically, a crisis similar to the scale of the 2014 anti-China protest in Vietnam has not occurred since the 1990s. However, rather than use the protest as a bargaining tool to gain better terms in its negotiation with China, the Communist Party of Vietnam quickly suppressed the protests, fearing that the protests might debunk its national myth and jeopardize its legitimacy. In 84 Volatile Nationalism | Xinru Ma Table 3.4.4: Top Posts that Received Most Comments in the Philippines (2016) the Philippines, by contrast, due to the lack of public protests against China, senators and military personnel have admitted candidly the internal debate of mobilizing the anti-Chinese sentiment, with the aim to rebuild domestic society, and to consolidate the national identity of the Philippines as a way to end the Muslim insurgencies in the South. The discussion so far enables us to trace the within-country variation in the levels of nationalism in each country, and how it di↵ers cross-nationally. However, it is limited in illustrating the interaction between two governments at the international level, especially 85 Volatile Nationalism | Xinru Ma when and how nationalism can indeed impact crisis bargaining at the international level. After all, the continued interest from International Relations scholars on nationalism is due to its ambiguous e↵ect during the times of crisis. Often, the negotiations and settlements of territorialdisputesarehappeningasa“two-levelgame,”inwhichthenegotiatorsitsbetween his negotiating counterpart from another state, and his domestic constituents who favor or disfavor the agreement being negotiated. Ideally, one should address the intertwining e↵ects of domestic forces, the size of win-sets, the role of institutions, the strategies of negotia- tors, international law, and even systemic changes in the world when assessing the e↵ect of nationalism on crisis bargaining, as is specified in Putnam’s two-level game framework. However, a comprehensive analysis of the interaction between a government and its domes- tic pressure groups apart from nationalists, and the relations between that government and other governments is beyond the scope of this dissertation. With this caveat in mind, I now shift my focus in Chapter 4 from conceptualizing and measuring nationalism to formally studying the e↵ect of nationalism on international bar- gaining,andthemechanismsthroughwhichnationalismescalatestherisksofarmedconflicts. With a formal model that emulates the strategic bargaining between nationalist protesters andthegovernments, Iexplore1)thescopeconditionsunderwhichthegovernmentiswilling and able to use nationalism – exemplified in the instances of nationalist protests - as lever- ages in international bargaining, and 2) the conditions under which the foreign government is willing and able to view that as a credible signal. 86 Volatile Nationalism | Xinru Ma 4 Playing with Firecrackers: WhenandWhyNationalistProtestsatHomeA↵ect Crisis Bargaining Abroad “This movement has multiple purposes. Nationalism is only part of the protests...More people are angry about the current government, and they wanted to use this opportunity to vent.” -Zheng Yongnian 45 May 4, 1919. China. ThousandsofstudentsfromthirteenBeijinguniversitiesmarchedonthestreet,protesting in response to how China was treated in the Treaty of Versailles and the cession of Shandong totheJapanese. Intheirindignation,theyburnedZhaojialou,thehouseofthenpro-Japanese Vice-Minister of Foreign A↵airs, and beat up Lu Zhongxiang, then Chinese envoy to Japan. “We had no political theory to guide us at that time, only our strong patriotic enthusiasm.” Deng Yingchao, a former high-ranking Communist Party ocial and the widow of Prime Minister Zhou Enlai, recalled many years later, “We demanded, ‘Abrogate the Twenty-One Demands!’ ‘Boycott Japanese Goods!’ and ‘Buy Chinese-made goods!’” (quoted in Ebrey, 2009,p. 360)ProtestssoonspreadacrossChina. InShanghaialone,itisestimatedthat around 60,000 workers joined the strikes (Spence, 1990,p. 312). Despite the mass-scale demonstrations, the international negotiation at the Paris Peace Conference continued against the interests of China. Of the four goals 46 Chinese delegation 45 “In protests, Mao holds subtle messages for Beijing.” Asian Correspondent. https:// asiancorrespondent.com/2012/09/in-protests-mao-holds-subtle-messages-for-beijing/ 46 Theseinclude: First,regainingcontrolovertheterritoriesthatwereinGermanpossession,andespecially of Shandong. Second, the abrogation of the treaties and agreements that resulted from the Twenty-One Demands. Third, the cancellation of all economic and political interests of Germany and Austria-Hungary 87 Volatile Nationalism | Xinru Ma outlinedduringthenegotiationgoals,noneofthemwasachieved. Article156oftheTreatyof Versailles transferred German concessions in Shandong to Japan, not to China. The Chinese delegation refused to sign the peace treaty, but according to Abbass (2010), it was “a largely symbolic gesture as Japan still retained control of Shantung province.” The discussion above does not intend to disparage the May 4th Movement. After all, despitethelimitedimpactontheconcurrentnegotiationattheinternationallevel,theimpact of this movement on domestic politics is far-reaching. Paul Reinsch, then minister of the UnitedStatestoChina,described“themobilizationofanactivepublicopinion,definiteinits aims, was a new development in Chinese.” (Cohen, 2010,p. 93)Themovementconstitutes “a realization that help from Western Nations will not be forthcoming until real reforms have been introduced by the Chinese.” (Cohen, 1966, p. 88) Many social and political leaders demanded to punish the traitors at home along with radical political reforms, which ultimately led to the demise of the Beiyang government and the founding of the Republic of China. Nevertheless, the protest’s long-term influence on domestic regime and the lack of short- term impact on the treaty negotiation at the international level epitomizes the key puzzle that I address in this chapter: How, if at all, can nationalist protests at home a↵ect crisis bargaining at the international level? Iarguethatthescopeconditionsandtheoveralle↵ectofnationalistprotestsonin- ternational bargaining in times of crisis still remain unspecified, particularly due to the interweaving of two uncertainties: 1) the home government, who is uncertain whether the protest will spontaneously evolve into an anti-government mobilization, and 2) the foreign government, who is uncertain whether nationalism constitutes a genuine credible and costly constraint of the home government. 47 in China. And fourth, the revision and possible abolition of the unequal treaties. Quoted in Teon (2017). For original source, see Pollard (1933, p. 56) 47 I use “the home government” for country in which the nationalist protest occurs, and “the foreign 88 Volatile Nationalism | Xinru Ma From an authoritarian government’s perspective, a nationalist protest is particularly challenging as it can easily turn into an anti-regime mobilization, if the protest is not a disguised collective action against the regime in the first place. When nationalist protests can spontaneously evolve into anti-regime mobilizations, allowing the protests is like holding afirecrackerinthehand: thenatureoftheprotestbecomesamovingtarget,andtheprotest is infused with heterogeneous and often contradictory interests. The impact of nationalist protests on the fate of the domestic regime doesn’t necessarily need long to take e↵ect, as in the case of the May 4th movement. The anti-Japan protest spread across China in nearly a dozen cities for three weeks in 2012 is a case in point. Initially in response to Japan’s purchase of the disputed Diaoyu (named Senkaku in Japan) Islands, it quickly went beyond the level of expressing anti-Japanese sentiment and became an opportunity for divergent societal groups to express their criticism towards the regime within nationalistic slogans. For instance, in a widely spread photo on Sina Weibo, the Twitter of China, an elderly shopkeeper wrote the following poem on a blackboard in front of his shop by the protest route:“no medical care, no social welfare, [we] have the Diaoyu Islandsinourheart; nopropertyrights, nohumanrights, ontheDiaoyuIslands[we]fightfor sovereignty rights; cannot a↵ord houses, cannot a↵ord graves, [we] won’t yield an inch of our land to the Japanese.” 48 Concerned about being imprisoned for protesting without proper legal proceedings, a protester waved a banner saying “Abolish the re-education through labor system 49 first, in case [we] get interned for protesting for the Diaoyu Islands.” Other protesters embedded their criticism towards chengguan,thewidelyloathedviolentcorrupted government” for country against which the nationalist protest is. In the case of the 2012 anti-Japan protest in China, the Chinese government is the home government, and the Japanese government is the foreign government. 48 The original slogan reads, “° ; › ° > › √ -Å ì | õ 1 ó ? ú { _ Å b ì | õ ° i C ° ∫ C ì | õ ⌦â; C p w ? Ó w _ ¯ ©  , ∫ ⇥ ”Iused Zhang (2015)’s translation of this post. 49 Thereeducationthroughlaborsystem, institutedunderMaoZedongtoreform“opponentsofsocialism” atthewhimsoflocalpoliceauthoritieswithoutlegalproceedings,wasabolishedin2013inChina. Theslogan is rhythmed in Chinese, which reads “H ü ≥ Y ç › ì  2 › å ´ ≥ Y ”. 89 Volatile Nationalism | Xinru Ma municipal police, in nationalistic banners saying “Send 3000 chengguan to Japan, we ensure that they will reclaim all the lost territory; Send 500 corrupt ocials to Japan, we ensure that they will eat Japan to bankruptcy.” 50 From the foreign government’s perspective, when a nationalist protest can easily spin out of control into an anti-regime mobilization, repressing the protest can hardly reveal any information regarding the home government’s preference over the disputed issue. Instead, the action might be perceived merely as an indicator of the home government’s domestic preferenceonregimesurvival,whicho↵erslittlehelptosolvetheproblem-beitcommitment or information - that induces bargaining failure at the international level. This then calls intothequestionwhyallowingorrepressingtheprotestcanbeasignal, andasignalofwhat. For instance, upon observing anti-US sentiment in China few years following the Boxer Rebellion, the largest and most violent anti-foreign uprising with the wholesale massacre of foreigners in 1900, then US ambassador reported that “serious anti-foreign demonstrations are inevitable during the coming summer, if not before...There is danger that foreigners will bedrawnintothevortexandmassacrediftheydonotleavethecountry...[however,]primarily it will be a domestic trouble...a revolt of the people against Manchu rule.” (Shaw, 1906,p. 278) In other words, when a foreign government enters a bargaining being highly confident that the protest is mostly a problem for the home government, it is not at all clear what underlies the foreign government’s interests in capitulating to this protest abroad, regardless how the home government manages it. Consequently, allowing or repressing the protest becomes a noisy signal for the foreign observers, and often, it is particularly dicult for the 50 Chengguan, ocially known as Urban Administrative and Law Enforcement Bureau, are local law en- forcementagenciesinChinaresponsibleformanagingmorequotidianaspectsofurbanlife,suchasregulating streetvendorsandunlicensedconstructionsites. Separatefromconventionalpoliceforces, chengguan appear to operate with little oversight. As a result, stories of chengguan brutality are common. On the other hand, a common form of corruption in China is ocials attending lavish banquets in private clubs and restaurants with issuing only IOUs (an informal document acknowledging debt) for food and drink consumed. The original banner reads “Ÿ ⌘ C Œ ° u ö 6 fi ì | õ Ÿ ⌘ î ~ * P ò › ¡ ⇤ Æ ✏  , ”. For more examples of such slogans and banners, see http://www.wenxuecity.com/news/2012/09/18/1987373.html 90 Volatile Nationalism | Xinru Ma foreign observes to decide what should be interpreted from this signal. Thepurposeofthisformalchapteristotakeseriouslytheuncertaintythatanynationalist protest, no matter how sincere it appears in the first place, can evolve spontaneously into an anti-regime mobilization, and clarify that given this uncertainty, under what conditions it can serve as a credible signal, and signal of what, in international crisis bargaining. With a signaling game, it seeks to address three sub-questions: (1) What assumptions are needed on the nature of the dispute and the bargaining environment to make a nationalist protest a potential signal in international bargaining? (2) What is the exact problem that induces bargainingfailure, whichultimatelyleadstomilitarizedconflicts? And(3)whyisnationalist protest a signal? And why is it updating? The theoretical model integrates the answers to these questions and yields two central insights: First, not all nationalist protests can be a credible signal. Nationalism detected from a genuine nationalist protest that causes no threat to the regime can be sincere, but it doesn’t alter the foreign government’s calculus at the international bargaining, as it fails to transmit information that di↵erentiates di↵erent types of home governments, nor does it re-arrange the incentives of both governments in the bargaining through the mechanism of domestic audience costs. Instead, the uncertainty that any nationalist protest can evolve spontaneouslyintoananti-regimemobilizationisaliabilitybutalsoanasset,evenanecessity fornationalistprotesttobeacrediblesignalininternationalbargaining. Itisthisuncertainty thatincursdi↵erentcostsfordi↵erenttypesofthehomegovernment,whichfurtherstabilizes information transmission and honest signaling. Second, not all nationalist protests increase linearly the risk of armed conflicts. Instead, nationalist protests can surprisingly decrease ex ante probability of militarized conflicts, when the uncertainty of nationalist protests transitioning to anti-regime mobilizations is at middle level. The analysis in this chapter implies that the empirical association between the severity of threats to the home government from nationalist protests - represented by the probability of 91 Volatile Nationalism | Xinru Ma nationalist protests transitioning into anti-regime mobilizations - and the likelihood of con- flicts resulting from bargaining failure will be non-monotonic - that is, first decreasing and then increasing in the magnitude of threat (U-shaped). When the probability is low, nation- alist protests are not costly enough, thus failing to credibly reveal information regarding the home government; the high probability, on the other hand, is likely to result in repression of all protests, forestalling them being observed as a signal. At both ends, the management of the protests has no information value, and whether the foreign government makes concession or not is largely determined by its prior belief that to what extent the home government is strong. By contrast, nationalist protests have the best chance to be a credible signal with middle levels of uncertainty. Here, nationalist protests are costly so as to allow for credible information transmission, while the home country elites are not so averse that they resort to repression. Altogether, this non-monotonic relationship explains why such tactic helps authoritarian governments gain leverage in some cases but not in other cases. In what follows, I first present existing theories on how nationalist protests influence international bargaining, and their limits. I then propose a game-theoretic model of bar- gaining, in which the nationalist protest has no uncertainty and remains as anti-foreign government throughout. In this set-up, I clarify the specific information problem investi- gated in this model, the key parameters that shape the preferences of each actor, how and why nationalist protests fail to be a credible signal in the international bargaining when the protests have no uncertainty. I then modify the model by incorporating the uncertainty regarding the protest. Here, allowing or repressing nationalist protests allows for informa- tion transmission if the two types of the home government at least partially separate their responses to the protests. With this, I explore under what conditions allowing or repressing nationalistprotestscanseparateastronghomegovernmentfromaweaktype, howitinduces the foreign government to make concessions, and the conditions on the prior belief of the foreign government for such equilibrium to exist. With this two-step set-up, I establish the 92 Volatile Nationalism | Xinru Ma necessity of this uncertainty as a condition for nationalist protests to be a credible signal in international bargaining, how it alters the strategic calculation of both the home and the foreign governments, and the significantly distinct conditions on the prior beliefs of the foreign government to make concessions in international bargaining. The chapter ends with an illustration of the (in)e↵ective signaling with the 2014 anti-China protest in Vietnam. The findings from this chapter di↵er from the extant research on nationalist protests in three aspects: First, the exercise of formalization crystalizes assumptions that we need to believe in and the scope conditions that make nationalist protests an indeed credible signal. It thus generates empirical predictions that are not all inherent or coherent in a qualitative description of the case. With this, it hopes to clarify not only the potential but also the limits to use nationalist protests as a signal in international crisis bargaining. Second, it points out that the domestic audience costs are not the only mechanism to explain the variation in how authoritarian governments handled nationalist protests. Often, the costs for using nationalist protests as a signal don’t come from backing down in international bargaining: they come from allowing protests that are dangerous to regime stability. Third, the paper considers uncertainty about both the type of government and the nature of the information problem that induces bargaining failure, and o↵ers a unified theory that links the management of protests with the type of the home government, and how this, if at all, incentivizes the foreign government to capitulate to this protest abroad, regardless how the home government manages it. Clarifying the boundary and scope conditions of nationalist protests as a signal also has enduring implications for contemporary East Asian regional security dynamics. Empirically, tensions are ratcheting up in the South China Sea, and have sparked a wave of protests in major claimant states including China, Japan, Vietnam, and the Philippines, with some turning into brutal violence against foreign nationals and property. On the other hand, authoritarian countries such as China and Vietnam have been managing the nationalist 93 Volatile Nationalism | Xinru Ma protests un-transparently, which gives rise to unique information problems that increases the risk of militarized conflicts as a result of bargaining failure. Thus, understanding when nationalist protests can be a credible signal and when such a conclusion doesn’t hold can inform policy-makers the intended and unintended consequences of nationalist protests, and better devise policies that can mitigate the risk of embroiling states into unintended costly militarized conflicts as a result of bargaining failure. 4.1 CrisisBargaining,Signaling,andTheManagementofNation- alist Protests as Information To date, the management of protests and international crisis bargaining has mostly been addressed by two separate bodies of the literature: the literature on contentious politics, and the literature on diversionary war theory. The literature on contentious politics have provided insightful explanations on the strate- gicinteractionsbetweenthehomegovernmentsandtheprotesters. Withafocusonthee↵ect of repression on participation and/or the reverse e↵ect of participation on government’s re- pression, this literature highlights the importance of individuals’ dissatisfaction with the provision of collective goods (Finkel and Muller, 1998), the cost of collective action (Boix andSvolik,2013;Edmond,2013), thenumberofdissidentswillingtorevolt(Lohmann,1994; ShadmehrandBernhardt,2011), thestructureofthesocialnetwork(Siegel,2011), thesever- ity of state repression (Francisco, 1993, 1995), and the consistence of the policies employed by the government (Lichbach, 1987)asimportantdriversofprotestsparticipationandgov- ernment’s repression. 51 More recently, Lorentzen (2013)breaksdownthehomegovernment into the central and local ones, and points out how the central government could strategi- cally allow or repress protests to address conflicting interests between the central and local 51 For a more detailed review of this literature, see Carey (2006). For discussion on the endogeneity issue and the observational challenges of dissent and repression, see Ritter and Conrad (2016). 94 Volatile Nationalism | Xinru Ma government. Specifically, by tolerating regular small-scale narrowly economic protests in an authoritarian regime, the central government identifies discontented communities before they turn to more extreme counter-regime activities. Nevertheless, works on contentious politics are limited in shedding light on the calculus of government when the target of the protest is moving over time with the interests involved in being so heterogeneous. After all, the protests studied in this literature often have pre- defined targets - e.g. local government or a small set of elites - with the demand to change aspecificpolicy(Robertson, 2007; Carey, 2006; Hollyer et al., 2015). In Lorentzen (2013)’s case, theprotesters’demandsalmostalwayshavetodoonlywithmaterialinterestsandlocal grievances typically stemmed from corruption, and the rhetoric is explicitly loyal towards the central regime. In addition, despite the insights on the domestic interaction between the protesters and the government, this literature is limited in helping us understand how this interaction impacts the crisis outcome with which the home government engages at the international level. On the other hand, the literature on crisis bargaining and militarized conflicts lumps nationalist protests under the diversionary war theory, assuming that nationalist protests are the same as other kinds of civic protests that cause the leaders’ reputation or legitimacy costs. When a state leader is faced with domestic dissension and feeling vulnerable and constrained, he or she might be willing to engage in certain types of conflictual international behavior in the hope that these actions will improve his or her domestic political standing (Leeds and Davis, 1997,p. 817). Richards et al. (1993)employprincipal-agentmodeland show that democratic executives demonstrate their competence through assertive handling of foreign policy crises when they are politically vulnerable at home. A substantial literature on the “Rally’round the flag” e↵ect has established that the president enjoys a boost in popularity following the use of force internationally. However, the documentation of rally e↵ect has not been replicated in authoritarian countries ( Gelpi, 1997, p. 257). As Johnston 95 Volatile Nationalism | Xinru Ma (2013,p. 37)illustratesinthecaseofChina: “Inapoliticalsystemwherethereareno electoral costs to ignoring public opinion, it is unclear why China’s authoritarian leaders would care much about public views.” In her seminal work, Weiss (2014, 2013)providedinitialintellectualimpetusforlinking nationalist protests with signaling in crisis bargaining. According to Weiss, allowing protests enables autocrats to signal resolve and credibly commit to a tough diplomatic stance. By contrast, repressing the protests constitutes a costly signal of reassurance that the govern- ment places high enough value on international cooperation (Weiss, 2013,p. 3). Weiss’ theory relies on several mechanisms and assumptions that are often implicit in her initial works, thus it is worth teasing out each mechanism and their connections carefully. The main mechanism that enables allowing protests to signal resolve is the domestic audi- ence costs mechanism. Specifically allowing protests generates “domestic audience costs” that “the authoritarian leaders will su↵er ex post if they do not follow through on their commitment” (Fearon, 1997). Allowing the protests visible on the street creates “a means of showcasing domestic pressure as leverage in diplomatic negotiations, a form of brinkmanship that conveys resolve and commitment to an unwavering stance...These ‘audience costs’ make it harder for the government to o↵er concessions, increasing the risk that the government will be locked into a position it cannot yield. The decision to go public signals resolve; the ensuing threat of domestic punishment ties the government’s hands.” (Weiss, 2014,p. 5,p. 17) Here, the home government uses nationalist protests as the tactic of burning bridges, and seeks to change the opponent’s behavior by limiting the future options. “These costs are not paid if the leadership stands firm, creating incentives to take a hawkish stance. If nationalist protesters see progress toward their objectives, they will more readily disperse without blaming the government.”(Weiss, 2013,p. 9) However, as the literature on domestic audience costs points out, this mechanism works only when the domestic audience costs are large enough to reverse one’s preference between 96 Volatile Nationalism | Xinru Ma war and living without the disputed good, especially when the disputed good is indivisible. For example, Slantchev (2005, p. 544) writes that the audience cost mechanism “requires the demanding assumption that leaders incur suciently high audience costs; so high, in fact, that peace becomes worse than war.” Trager and Vavreck (2011,pp. 526-527)state that “if the political costs of conflict are higher than the costs of backing down after a threat, the credibility of threats to use force will be much less than otherwise.” Apart from the substantial level needed for the costs, Snyder and Borghard (2011,p. 439-441)further summarize three additional assumptions that must hold true for domestic audience costs to have a significant e↵ect on crisis outcomes: (1) leaders seek lock-in, not flexibility, (2) publics care a great deal about consistency between threats and deeds, independent of their preferences on policy substance, and (3) targets of threats understand domestic audience costs mechanisms. Only when the disputed good is divisible, limited audience cost can a↵ect bargaining outcome, but through a noninformative, “bargaining leverage” mechanism (Tarar and Levento˘ glu,2013). Thus, whenauthoritariangovernmentscane↵ectivelycontain or state-manage nationalist protests, the cost of backing down after a threat to stand firm is much less than that su↵ered by governments that don’t have the capacity to contain the protests, thus significantly undermining the credibility of allowing protests as a signal of resolve. In addition, when nationalist protests can spontaneously evolve into anti-regime mobi- lizations, allowing the protests is like holding a firecracker in the hand, makes the nature of the protest a moving target, and infuses the protest with heterogeneous and often contradic- tory interests. Consequently, using nationalist protests as a signal creates ex ante costs that are independent of the hand-tying domestic audience costs that the authoritarian leaders might su↵er ex post,andsuchcostscan endogenously change the authoritarian government’s preferences before the outcome of the bargaining is observed/realized. For instance, follow- ing NATO’s bombing of Chinese Embassy in Belgrade in 1999, “the Chinese government 97 Volatile Nationalism | Xinru Ma was very concerned about the possibility of losing control if it allowed a large-scale anti-U.S. demonstration.”(Zhao, 2003, p. 9) Here, the costs that Chinese leaders consider don’t come from backing down in international bargaining: they come from allowing protests that are dangerous to regime stability. In other words, using the protests as a signal can backfire even if the home government follows through the tough stance that nationalists demand at the international bargaining. Thiscostofinstabilityanticipatesthesecondmechanismatworkforallowingtheprotests to signal resolve in Weiss’ theory. “At worst, protests could turn against the government, growtoolargeforstatesecuritytodisperse,orgeneratesuchsupportthatstateinsiders...may defect and disobey orders to suppress the protests. The decision to run this risk, however small, enables the government to signal resolve, di↵erentiating it from a government that cares less about the international dispute.” (Weiss, 2013,p. 8) Though plausible, the scope conditions for instability costs to credibly separate a re- solved government from an unresolved one, and why it can di↵erentiate di↵erent types of governments need to be specified, rather than assumed. At a minimum, when the risk of instability is low, an unresolved government should also seek to allow the protests, as long as the international gain outweighs the domestic costs (blung thus happens). After all, the managementoftheprotestsshouldresultfromfactoringinbothdomesticcostsandpotential gain from the international bargaining. This thus creates a moral hazard that undermines the credibility of allowing protests as a signal of resolve. Thus, despite the potential, the conditions under which allowing the protest indeed signals a resolved government, rather than a blung technique employed by an unresolved one, remains unspecified. On the other hand, when the risk of instability is high, repressing the protests can hardly reveal any information regarding the home government’s preference over the disputed issue. Even if an unresolved government intends to signal reassurance, the action might be perceived by outside observers merely as an indicator of the home government’s domestic 98 Volatile Nationalism | Xinru Ma preference on regime survival, which o↵ers little help to solve the asymmetric information problemthatinducesbargainingfailureattheinternationallevel. Similarly, evenifthehome government is indeed resolved with a high value on the disputed goods, the fear of regime collapse may force it to repress the protests, an action that sends out the wrong message at theinternationallevel. Thisincentivizestheforeigngovernmenttorefuseanyconcession,ifit mistakenly assumed that the home government is weak. In this way, the action of repression increases - rather than decreases via signaling reassurance - the probability of militarized conflicts. Apart from adding noises to the signal, the instability risk further compounds the cred- ibility of signaling by generating additional incentives for the foreign government to delay the settlement of the bargaining, even when the protest is indeed a sincerely self-motivated anti-foreign government protest at the beginning. Observing the outbreak of a nationalist protest,theforeigngovernmentmighttakeadvantageofthisuncertainty,lettheprotestburn for a while, and see if the home government would put the firecracker out before and/or as the anti-regime voices emerge. If yes, the willingness and the ability of the home government to control the protest indicate that the domestic costs are no longer large enough to credibly di↵erentiate between di↵erent types of signaling actors, hence the mechanism of signaling resolve via costly protests breaks down. In other words, in the shadow of this uncertainty, success of perceiving the nationalist protest as a credible signal depends largely on (1) how the foreign government conjectures the nature and the evolvement of the protest, (2) its belief about to what extent the home government is willing and able to curtail the protest before and/or as it becomes destabilizing, and (3) to what extent the foreign government is willingtolinkthemanagementoftheprotestwiththespecificproblemthatinducesbargain- ing failure in times of conflict. Put di↵erently, faced with a nationalist protest, the foreign government not only has to update (1) its belief about the type of the home government (e.g., strong versus weak government), (2) its belief about the target of the protest, but also 99 Volatile Nationalism | Xinru Ma Figure 4.1: Anticipated Outcomes of Protest Management in Weiss’ Theory and Its Limita- tions ‡: The left panel is reproduced from Weiss (2013, p. 7). (3) its belief about the home government’s belief of the target of the protest. Figure 4.1 reproduces a stylized universe of possible protests in Weiss’ work, and its limitations. Insummary, thelatestscholarshiphassofaroutlinedthepotential ofnationalist protests as a signal in international bargaining, but realizing such potential depends on countless assumptions made for either practical or substantive reasons, some of them largely implicit in extant works. 4.2 A Unified Theory of Managing Protests as A Signal The above discussion calls into the question why allowing or repressing the protest can be a signal, asignalofwhat, andwhyitisupdating. Ataminimum, acompletetheoryofhowthe management of nationalist protests as a signal impacts crisis bargaining outcomes requires clear answers to the following questions: First, what is the nature of the disputed issue in times of conflict? Second, what is the nature of the problem that induces bargaining failure? 100 Volatile Nationalism | Xinru Ma Third, why is the management of nationalist protests signal? And why is it updating? Only after clarifying these questions can one specify the scope and boundary conditions for the management of protests to be a credible signal, and the conditions under which blung and failed signaling are prevented. I now answer these questions in order, so as to clarify the assumptions of the ensuing model developed in the next section, and to what extent the findings in this chapter depend on a particular formalization of the bargaining environment. First, the nature of the disputed good. One important concept that has been investigated extensively in the bargaining literature but not in the nationalism literature is whether the disputed good is divisible or not. Fearon (1995)contendsthatallgoodsare essentially divisible or can be made so through political bargaining schemes. Hensel and Mitchell (2005), however, claims that sacred religious spaces are indivisible due to their centrality and vulnerability, and his work has inspired scholars to ponder whether secular territories are indivisible too. Since disputes examined in the nationalism literature varies so widely from territory issues such as China’s behavior in territorial disputes (Fravel, 2008), to policy issues such as military expenditure and maritime strategies (Ross, 2009), China’s tradenegotiationsandpolicies(Weiss,2013), andregionalmultilateralism(Hughes,2005), it is important to specify the nature of such disputes, which may alter the e↵ect of nationalist protests on bargaining dramatically. To keep this paper focused and tractable, I examine crisis bargaining over indivisible good in non-cooperative context (Table 4.2). Theoretically, this is a common assumption that underlies much of the literature on bargaining theory and territorial disputes. (Schultz, 1998; Toft, 2002) Empirically, claimants such as China, Vietnam, and the Philippines insist rhetorically that the disputed islands and waters are “integral parts” of their territories. On the other hand, by focusing on the non-cooperative context, this paper doesn’t assume nor investigate actors’ “willingness to cooperate” implied by any action taken by the players. 52 52 Weiss’ seminal works didn’t specify whether the context of signaling is zero-sum or positive-sum. One 101 Volatile Nationalism | Xinru Ma Thepayo↵structureofeachplayersfollowsazero-sumissuesetup,andtheonlyinformational value of allowing or repressing a protest investigated in this chapter is to separate the home government that has high value for war (strong-type), from that with low value (weak-type) for war. 53 Table 4.2: Uncertainties in Crisis Bargaining and the Focus of the Paper Second, what is the problem that induces bargaining failure? Inecient conflict results from commitment or informational problems, with either being the primary cause or acomplementtotheotherproblemunderlying. (Powell, 2006; Fearon, 1995)Informational problems arise when 1) the bargainers have private information and 2) the bargainers have incentives to misrepresent their private information (Powell, 2006,p. 170). Morerecently, Fey and Ramsay (2011)examinedtwodi↵erentsourcesforsuchprivateinformation-un- certainty about relative cost of fighting versus uncertainty about the opponent’s military capabilities - and show that though both can be interpreted as a country’s “resolve” (Smith, canstillinvestigatetheconditionsforthemanagementofprotestsasacrediblesignalinpositive-sumsettings with divisible goods, however, this requires di↵erent set-up and payo↵ structures of the model. 53 From here and for the rest of this chapter, I use “strong” and “weak” as opposed to“resolved” and “unresolved” to refer to di↵erent types of home governments. This choice is made to avoid confusion regarding the word “resolve.” Though in formal literature, a resolved type is one that has higher reservation value for the disputed goods, either due to high winning probability or lower cost of war. However, in literature such as political psychology, a resolved type may have high cost of war, but still choose to fight as a result of “heuristics.” (Petersen, 2015) 102 Volatile Nationalism | Xinru Ma 1998;Schultz,1998, p. 832), thelinkbetweenuncertaintyandwarinthebargainingproblem depends in important ways on the kind of private information states have. So what is the problem that may induce bargaining failure in crises that involved nation- alistprotestshistorically? Ofthe476crisesdocumentedbytheInternationalCrisisBehavior (ICB) dataset from 1918 to 2015, 10 crises had concurrent mass mobilizations targeting the foreign governments as the crises unfolded, while many of them involved uncertainty regard- ing the opponent’s capability of fighting. 54 For instance, in the 1956 Suez nationalization crisis, there was little doubt that Egypt would defend itself resolutely if attacked. But even knowing this fact, the Israel-UK-France force was uncertain about how e↵ectively Egypt would perform in battlefields, due to two confounders: First, though the Egyptian military was well equipped with weapons from the Soviet Union such as T-34 and IS-3 tanks, MiG- 15 fighters, SU-100 self-propelled guns and assault rifles, it was also well known that there existed mutual “mistrust and contempt” between Egyptian military commanders and the soldiers who served under them, as most commanders were purely political appointees who owed their positions to the close friendship with President Nasser. (Varble, 2008)Second, the resistance from a popular mobilization led by Communist activists, who had played a central role in the mass movement against the British occupation during the 1940s. They organized civil defense, smuggled arms to keep guerrilla resistance going in occupied Port Said and mobilized thousands of volunteers for military training. (Alexander, 2006) Thus, this paper focuses on the information asymmetry that the foreign government is uncertainaboutthehomegovernment’scapabilityoffightingandwinning. Thatis,countries are assumed to be informed about their opponent’s costs of fighting, but the utility from 54 These crises were retrieved by searching the key word “protest” in the ICB dataset (Version 12). Each crisis that contained this key word in the description is then read to discern whether the protest is mass mobilization targeting foreign country or not. These crises are: the 1934 assassination of King Alexander crisis, the 1939 crisis among Albania, France, Greece, and the U.K, the 1944 Iran-oil concessions crisis, the 1956 Suez nationalization war, the 1956 crisis over Poland’s liberalization, the 1969 USSURI river crisis, the 1970 invasion of Cambodia crisis, the 1987 Syria-Lebanon crisis, the 1990 India and Pakistan near-nuclear crisis, and the 1991 Armenia and Azerbaijan crisis over Nagornyy-Karabakh. 103 Volatile Nationalism | Xinru Ma war is dependent on the type of its opponent, reflected in the probability of victory. Third, whyisthemanagementoftheprotestsignal? Andwhyisitupdating? Inotherwords,whatunderliestheforeigngovernment’sinterestincapitulatingtonationalist protests abroad, and why does the foreign government care about how they are managed? Clarifying this question is critical, as this is the a prior condition for nationalist protests to o↵er any bargaining leverage. Research that explores explicitly why a foreign government is concerned about o↵shore nationalist protests is scarce. The most empirically documented concern is that nationalist protests can escalate into violence against foreign nationals and diplomatic property. For instance, Weiss (2014)contendsthattheincentiveforforeigngovernmentstoshowleniency is the threat of domestic and diplomatic instability that nationalist protests pose. “The real specter of extreme nationalism and instability is a shared risk that outside actors have an incentive to mitigate,” (p. 247) and such vested interest in stability of these outside actors holds even if regime change could bring a friendlier government to power. “The risk of a failed state...may be prohibitively high...Even the potential disruption of trade, investment,andsupplyofnaturalresourcesmaybesucientincentivetomaintainstability.” (p. 245) Consequently, “once the government has tied its hands and demonstrated resolve by allowing antiforeign protests, the burden of conciliation falls to the foreign government.” (Weiss, 2014,p. 28)Insum,thisapproachviewsthatnationalistprotestsmatterforthe foreign government as they impact their invested interests in the country, independent of the disputed good featured in the international bargaining. However, to alter the foreign government’s preference between conceding and standing firm, the damage to foreign life and property has to be so large that it o↵sets the utility the foreign government could gain from the disputed goods. This is less likely especially given the number of casualties and economic damages in contemporary anti-foreign protests is much less devastating. From de-classified documents and correspondence between home 104 Volatile Nationalism | Xinru Ma and foreign ocers, there’s also little evidence that such concern ever altered the foreign government’s preferences. For instance, upon observing massive anti-Japan protests in 1927 before the Second Sino-Japanese War, Japanese Prime Minister Baron Tanaka Giichi con- curred, “At present foreigners and especially Japanese are su↵ering direct damage to life and property at their hands. This is something that Japan cannot ignore...[However,] the sending and withdrawing of troops should be determined by the thinking of Japan herself. It is quite unwise to do so if we are only afraid of boycotts.” (Nish, 2001,pp. 292-293) In particular, game-theoretic analyses find that a costly signal can a↵ect international crises if at least some types of the informed actor - usually referred to as the “strong” types - aregeneratinglargeenoughcoststhatdiscouragethe“weaker”typesfrommakingthesignal. Thus, for nationalist protests to be a costly signal that is relevant for the crisis bargaining, it has to satisfy two conditions: first, the signal is an action that can be endogenously chosen by the informed home government, and second, the cost incurred by this action can di↵erentiate a strong home government from a weaker home government who prefers not to incur this signaling cost. Iarguethatwhetherthemanagementoftheprotestsisasignalornotdependsonthe type of the uncertainty that impedes negotiated settlement at the international level. 55 Not allauthoritariangovernmentsareequalinstatecapacity: somearestrong,someareweak. In timesofconflict,thekeyobstaclethatpreventstheforeigngovernmentfromreachingecient 55 An alternative mechanism through which nationalist protests could potentially induce the foreign gov- ernment to make concessions is the additive e↵ect of the protests on the cost of war. A genuine large-scale anti-foreign protest underlies heightened nationalism in the home government, which generates additional cost of war for the foreign government and reduces her war payo↵. This doesn’t mean that nationalism immediately makes war inevitable, rather, it primes citizens with expanded sensationalist coverage in pop- ular media, unrestrained aggressiveness against the foreign government, and convinces citizens their nation is being threatened by the scheming rivals, thereby influencing the cost of war if and only if war does arise later. In a similar vein, a small-scaled anti-foreign protest - or an anti-regime protest disguised as a na- tionalist protest - underlies citizens’ lack of support for the home government, which decreases the foreign government’sestimationofheractualcostofwar(e.g. easierfortheforeigngovernmenttorecruitdefectors). This can be true, however, modeling this mechanism is not interesting, as bargaining in this scenario is not influenced by the actions endogenously chosen by the strategic players, but solely by an external factor - the type and the scale of the protest - that is not part of the bargaining. 105 Volatile Nationalism | Xinru Ma peacefulsettlementsinaninternationalbargainingistheasymmetricinformationproblem,in the sense that the foreign government is uncertain about the home government’s type. That is, the foreign government’s utility from war or peaceful settlement is dependent on whether the home government is strong or weak: a “strong” government has high capacity to fight and win in war, while a “weak” government has weak fighting capability in external wars. Plausibly, a“strong”governmentthathashighcapacitytofightandwininwarwithanother country can also e↵ectively contain domestic mobilizations (more money, more troops, more legitimacy, more centralized power, etc.), and that a “weak” government that has weak fighting capability in external wars also has high cost to contain domestic mobilizations internally. Thus, allowing or repressing a protest - an action that can be endogenously chosen by the home government - can be informative and updating as this action enters into the home government’s utility by incurring di↵erent costs for di↵erent types, which might induce a strong and weak home government to at least partially separate their responses to the protests. Figure 4.2 illustrates this key mechanism that my theory relies on. The upper panel il- lustrates the problems faced by both governments in times of crisis, and the key mechanism that links the management of protests with the type of the home government through the intertwined external military and domestic repressive capacity. The lower panel illustrates the associated costs for each type of home government when the protest can turn sponta- neouslyintoananti-regimemobilization, whichinducesastrongandweakhomegovernment to at least partially separate their responses to the protests. In the next section, I present aformalmodelthatclarifiesthescopeconditionsforhowthesedi↵erentialcostsincurred by nationalist protests induce information transmission and honest signaling in international bargaining. 106 Volatile Nationalism | Xinru Ma Figure 4.2: Signals and Their Associated Costs for Each Type of HG When Protests Can Turn Into Anti-regime Mobilizations 107 Volatile Nationalism | Xinru Ma 4.3 A Formal Model 4.3.1 The Baseline Model When Nationalist Protests Have No Uncertainty The game involves two states, the home government (HG,pronoun“he”)andtheforeign government (FG, pronoun “she”) bargaining over an indivisible good whose value is nor- malized to 1. War is costly, with HG andFG’s costs of war being c H ,c F > 0, respectively. Consider a case of one-sided uncertainty, where HG’s fighting capability can take two possible values, corresponding to two types of HG:a“strong”typewithhighcapacityto fight and win in war with another country and e↵ectively repress internal dissent ( p), or a “weak” type with both low internal and external fighting capacity (p). Nature chooses HG to be strong with a probability ⇡ 2 (0,1). The probability of FG winning the war is thus 1 p and 1 p,whenfightingwithastrongandweaktypeof HG,respectively. Learning the crisis at the international level, citizens in the home country decide to protest and protest successfully. 56 HG,whoisinformedofhisowntype,canchoosetoallow the protest on the street or repress it. Given the protest in this setting has no uncertainty of transitioning into an anti-regime protest, the management of the protest incurs only one type of cost, g2{ g,g},whichdenotesthelegitimacycost HG has to pay for stifling genuine nationalist protests from the committed regime supporters. Empirically, a textbook story is that this legitimacy cost is particularly high for countries such as China, where the government has been using nationalism as a “legitimating glue to hold the society together,” (Friedman, 2008)whichfurthermakesthecountrybecome“increasinglydependentupon its nationalist credentials to rule.” (Gries, 2004; Gries et al., 2011)Astrong HG that has strong external fighting capacities also has stronger repressive capacity, and can repress the protest more e↵ectively and quickly before the protests spreading too fast. Thus, a strong HG su↵ers a smaller legitimacy cost ( g)comparedtoaweakHG (g). Upon observingHG’s 56 This assumes that HG’s management of protests as a tactic for signaling is an ex post re-action to protest that is already on the street, an assumption that will be relaxed in future extensions. 108 Volatile Nationalism | Xinru Ma Figure 1: Signaling When Nationalist Protest Has No Uncertainty response to the protest,FG chooses to concede or resist. The game ends whenFG concedes. Otherwise, observingFG’s resistance, HG can choose to concede or further resist by going to war. Outcome and payo↵s. The three outcomes this section intends to explain are: 1) when FG concedes (HG gets all the indivisible good), 2) when HG concedes (FG gets all the indivisible good), and 3) when war occurs (HG andFG both resist). The payo↵s are as follows. If HG allows the protest andFG concedes, then HG’s payo↵ is 1 and HG gets 0. However, ifHG represses the protest andFG concedes,HG’s payo↵ is 1 g ifHG is strong and 1 g if HG is weak. The legitimacy cost from repressing the protest not only applies to successful bargaining (peaceful negotiations) but also war payo↵s. That is, if FG resists and war occurs, HG has to pay the legitimacy cost for repressing the protest earlier. The expected utilities from war whenHG represses the protests are EU strong H (war)=p c H g andEU weak H (war)=p c H g. FG’swarpayo↵s, whichareonlya↵ectedby HG’stype, are: EU F (war)=1 p c F and 1 p c F ,forfightingwithastrongandweakHG,respectively. Figure 4.3.1 illustrates the sequence and the payo↵s associated with each outcome. 109 Volatile Nationalism | Xinru Ma Incentive to misrepresent. There are two possible motivations for HG to misrepre- sent himself. First, when HG is strong, he may seek to misrepresent the capacity to make war more likely. Second, when HG is weak, he may seek to misrepresent himself as the strong type to induce FG to concede. The paper focuses on the second scenario, that is, the weak type may blu↵. To better illustrate this incentive, the model makes the following assumption on the relationship of parameters to clarify both actors’ initial preferences over war and peace. Assumption 1. [Di↵erential Incentives to Fight]. In the setting when nationalist protests have no uncertainty, assume that the strong type of HG begins the interaction with commit- ment to fight, but a weak type doesn’t. That is, p c H < 0 < p c H . Also assume that FG prefers to fight with a weak HG, but is better o↵ by conceding to a strong HG. That is, 1 p c F < 0< 1 p c F . Solving the model leads to the following comparative statics. Claim 1. When nationalist protests have no uncertainty of turning into anti-regime mo- bilizations, there exists no separating equilibrium in which a strong HG always allows the nationalist protest to signal resolve, and a weakFG fully reveals himself by always repressing the nationalist protests. Claim 2. When nationalist protests have no uncertainty of turning into anti-regime mo- bilizations, the weak HG will always allow nationalist protests to mimic the strong type. Allowing nationalist protests to signal resolve is “cheap talk,” and whether FG concedes or not is determined by the belief that she assigns to HG being the strong type when observing a protest. These two claims are derived from the following two formal propositions. Let = Pr(Strong|Repress)denotethebeliefFG assigns toHGbeingthestrongtypeafterobserv- ing repression. Let =Pr(Strong|Allow)denotethebeliefthat FG assigns to HG being 110 Volatile Nationalism | Xinru Ma the strong type conditional on observing a protest. Appendix 4.6.1 shows the proof for these two propositions, as well as the lack of separating equilibrium described in Claim 1. Proposition 1. [Cheap Talk Pooling Equilibrium - Peace Outcome]. In incomplete infor- mation setting when the protest has no uncertainty, when ⇡ 1 p+c F , the following pooling equilibrium exists: HG allows protest regardless of his type. Upon observing the protests, FG concedes and war is avoided with certainty. On o↵-equilibrium path, FG concedes after observing the repression of the protest, when ⇤ = 1 p+c F . Proposition 2. [Cheap Talk Pooling Equilibrium - Probabilistic War Outcome]. In incom- plete information setting when the protest has no uncertainty, when⇡< 1 p+c F , the following pooling equilibrium exists: HG allows protest regardless of his type. Upon observing the protests, FG resists. Following FG’s resistance, a strong HG resists (thus war happens), and a weakHG concedes (blu↵ed and failed). Here, war occurs with positive probability, and this probability is not modified by the signaling. On o↵-equilibrium path, FG resists after observing the repression of the protest, when the belief FG assigns to HG being the strong type after observing repression is smaller threshold< ⇤ = 1 p+c F . The comparative statics from this proposition are intuitive but worth walking through. The only incentive for HG to use this signaling tactic is when a weak type trying to mimic the strong type, so as to induce FG to concede in the international bargaining. However, since the protest has no uncertainty, there’s no cost for a weak HG to mimic the strong type, while he has to pay a cost when he represses the protest. Given this legitimacy cost carries over even when war happens, repressing the protests makes his payo↵ lower than his war payo↵ from allowing the protests. Hence, HG is better o↵ to allow the protests regardless of his type. Consequently, the management of protests is uninformative, and the signal transmitted by the management of protests is irrelevant. FG’responsesarefixedby her initial beliefs: FG who has a suciently strong belief that HG is strong will always 111 Volatile Nationalism | Xinru Ma concede;FG who has a suciently strong belief that HG is weak will always resist. In other words, when the nationalist protest has no uncertainty, allowing the protest might induce FG to concede than she would otherwise, but it is the high prior that FG holds doing all the hard work. Altogether, these findings call in to question the conventional wisdom that “playing the ‘nationalist card’ is diplomatically advantageous,” especially when the protests seem to be rootedinsincereangeragainstforeignactsofperceivedaggressionorhumiliation(He,2009). Instead, as long as there’s no uncertainty associated with the protest, allowing the protest cannot help the home government gain any leverage, as the bargaining outcomes are solely determined by FG’s prior belief of HG’ being the strong type and his actions given he is strong. This logic helps explain the stagnation in the bargaining between the U.S. and China in the 2001 EP-3 crisis, in which China detained 24 crew members of the US Navy EP- 3E intelligence aircraft for 11 days after it collided with a People’s Liberation Army Navy (PLAN) fighter jet. Following this incident, China repressed all anti-U.S. protests and made three demands in the negotiation: first, the United States should accept full responsibility and issue a formal apology for the collision; second, the United States should put an end to all the surveillance over the South China Sea; and third, the United States should o↵er a $1 million compensation payment “to cover China’s cost.” (Wu, 2007) However, observing China’s demands, the U.S. o↵ered only a formal letter expressing “regret for the loss of life,” and a compensation of US $34,576. According to Phillip T. Reeker, State Department spokesman, “We made quite clear that we are only going to pay reasonable, tangible costs associatedwiththerecoveryoftheaircraft, andwebelievethatourcalculationwasareason- able calculation...We have presented the o↵er to the Chinese; it remains on the table, and it is up to the Chinese as to whether they choose to accept it.” 57 Rear Admiral Craig Quigley, 57 Daily Press Briefing. Phillip T. Reeker, Deputy Spokesman. Washington, DC. August 15, 2001. 112 Volatile Nationalism | Xinru Ma a Pentagon spokesman, added that the payment, which would be made by the US embassy in Beijing in the next few days, was “non-negotiable.” 58 In other words, the legitimacy cost from repressing the protests alone cannot change the bargaining at the international level. That is, no matter how costly the repression is, FG is indi↵erent about this cost and only responds based on her war payo↵s, which are determined by her prior beliefs. 4.3.2 Incorporating Nationalist Protests with Transition Uncertainty Inowintroducetheuncertaintyregardingtheprotestintothemodel. Thegamebeginswith nature N drawing the type of the home government. Informed of his type, HG chooses to allow or repress the protests. Upon observingHG’s management of the protest,FG chooses to concede or resist. The game ends if FG concedes. Otherwise, the protest may take two types if it is allowed and FG resists: with probability ✓ the protest evolves into an anti- regime mobilization. For instance, protests, once begun, can trigger the sudden realization that protest is acceptable; They may also spread the protest techniques, provide a protective umbrella for domestic dissent, and equip citizens experience with political mobilization. (Weiss, 2013)Duetothesepossibilities,allowingananti-regimemobilizationundermines authoritarian stability, thus incurring a low instability cost (in relative terms)k for a strong HG,andahighinstabilitycost k for a weak HG. With probability 1 ✓ ,theproteststays as nationalist and incurs no cost. Learning the nature of the protest,HG chooses to concede or resist, and the game ends. Table 4.3.2.1 denotes the signals - HG’s responses to the protest - and their associated costs for each type of HG. Figure 4.3.2.1 denotes the new sequence of the game and respective payo↵s. Table 4.3.2.2 summarizes all the parameters considered in the final model. https://2001-2009.state.gov/r/pa/prs/dpb/2001/4551.htm 58 “Pentagon Spokesman Quigley Addresses Reporters About Spy Plane Row With China.” http://transcripts.cnn.com/TRANSCRIPTS/0104/03/se.02.html 113 Volatile Nationalism | Xinru Ma Table 4.3.2.1: Signals and Their Associated Costs for Each Type of HG When Protests Can Turn Into Anti-regime Mobilizations Figure 4.3.2.1: Sequence and Payo↵s When Protests May Turn Into Anti-regime Mobiliza- tions The incorporation of uncertainty now leads to a separating equilibrium that doesn’t exist in the baseline model. That is, when the protest has uncertainty, honest signaling happens as a strong HG can separate himself from the weak HG by always allowing the protests, whenthefollowingconditionsaremet. Here, themanagementofprotestsservesasacredible signal, and war is avoided with certainty (see Appendix 4.6.2 for proof). Proposition 3 (Credible Signaling Equilibrium). . In incomplete information setting when the protest has uncertainty, there exists a separating equilibrium in which a strongHG always 114 Volatile Nationalism | Xinru Ma Table 4.3.2.2: Actors, Signals, and Parameters Considered in the Full Model allows the protests, and a weakHG always represses the protests, if the following condition is met: 1+g k <✓< 1 p+c H +g k . Upon observing the protest,FG always concedes. Upon observing the repression of the protest, FG always resists. Observing FG’s resistance, the weak HG concedes. Two implications follow from this proposition. First, the uncertainty that the protest mightturnintoananti-regimemobilizationisaliability,butalsoanasset,evenanecessityfor the management of protests to impact crisis bargaining outcomes. Recall in previous model when the protest has no uncertainty. In that setting, there exists no separating equilibrium 115 Volatile Nationalism | Xinru Ma that allows for honest signaling and reliable information transmission, as allowing a protest that has no risk is costless for the weak HG. However, once the uncertainty is considered, the risk of instability incurred by a nationalist protest that may evolve into an anti-regime mobilization generates a considerable di↵erential costs for di↵erent types of HG.Theweak HG, who has weak domestic repressive and external fighting capacity, is unwilling to incur that cost as the international gain from blung is mitigated by the domestic costs, while astrong HG, who has both strong external fighting and domestic repressive capacities, is willing to bear that cost after factoring in both domestic loss and international gain. Thus, the uncertainty creates di↵erent costs for di↵erent types of HG, which induce them to separate their responses to the protests. Second, this proposition shows the limits of using nationalist protests as a signal even when the protest has uncertainty. The uncertainty associated with nationalist protests can help stabilize crisis bargaining, but not for all values of ✓.Theseparatingequilibriumis sustained only when this transition probability is within the sweet middle range. When the probability is low, nationalist protests are not costly enough, thus failing to credibly reveal information regarding the home government (weak type blu↵s); the high probability, on the other hand, is likely to result in repression of all protests, even when the government is indeed strong. At both ends, whether FG concedes or resists again depends solely on the beliefFG assigns to HG being the strong type and the associated actions. The blung equilibrium is an easy extension from the Cheap Talk Pooling Equilibria found in previous section, and is formally characterized by the following Proposition 4. The failed signaling - when the probability of transition is high and both types repress - is captured by Proposition 5. Proposition 4 (Blung Pooling Equilibrium) . . In incomplete information setting when the uncertainty of a nationalist protests is smaller than threshold✓<✓ ⇤ = 1+g k , the following pooling equilibrium exists: HG allows protest regardless of his type. Upon observing the 116 Volatile Nationalism | Xinru Ma protest, FG concedes, when ⇡ 1 p+c F . On o↵-equilibrium path, FG resists after observing the repression of the protest, when the belief that FG assigns to HG being the strong type when observing protests is smaller than< ⇤ = 1 p+c F . Proposition 5 (Failed Signaling Pooling Equilibrium). . In incomplete information set- ting when the uncertainty of a nationalist protests exceeds threshold✓>✓ ⇤ = 1 p+c H +g k , the following pooling equilibrium exists: HG represses protest regardless of his type. Upon observing the repression of the protest,FG resists, when⇡< 1 p+c F . On o↵-equilibrium path, FG concedes after observing the protest, when the belief that FG assigns to HG being the strong type when observing the protest exceeds ⇤ = 1 p+c F . 4.4 Discussion and Future Extensions So far, I’ve established not only the necessity of uncertainty that any nationalist protest can evolve spontaneously into an anti-regime mobilization for the management of protests to be asignal,butalsothescopeconditionsonhowlargethatuncertaintyneedstobeforcredible information transformation (Figure 4.4). Thesescopeconditionsarepremisedontheassumptionthatecientpeacefulsettlements in an international bargaining over indivisible goods is jeopardized by the asymmetric in- formation problem, in the sense that the foreign government is uncertain about the home government’s capability of fighting, while the cost of fight is common knowledge. Thus, the key mechanism for the management of the protest to be updating is the linkage between domestic and external military capabilities: a “strong” government that has high capacity to fight and win in war with another country can also e↵ectively contain domestic mobi- lizations (more money, more troops, more legitimacy, more centralized power, etc.), while a “weak” government that has weak fighting capability in external wars also has high cost to contain domestic mobilizations internally. Allowing or repressing a protest - an action that 117 Volatile Nationalism | Xinru Ma Figure 2: Relationship between the Uncertainty of Nationalist Protests Transitioning to Anti-regime Mobilizations and Bargaining Failure canbeendogenouslychosenbythehomegovernment-thusincursdi↵erentcostsfordi↵erent types, which induces a strong and weak home government to separate their responses to the protests. These results thus generate two key insights that advance the literature. First, not all nationalistprotestscanbeacrediblesignal. Nationalismdetectedfromagenuinenationalist protest that causes no threat to the regime can be sincere, but it doesn’t alter the foreign government’s calculus at the international bargaining, as it fails to transmit any information on the type of the home government. Instead, the uncertainty that any nationalist protest can evolve spontaneously into an anti-regime mobilization incurs di↵erent costs for di↵erent types of the home government, which further stabilizes information transmission and honest signaling. Itisnotsurprisingtoseethatthereexistsnoseparatingequilibriumthatinvolvesallowing the protest when the protest has no uncertainty. Intuitively, a PBE based on information 118 Volatile Nationalism | Xinru Ma revelation depends on there being some risk that deters the weak type from fully mimicking the behavior of the strong type. It might be intuitive to assume that a strong HG would allow the protest to signal his resolve, while a weak HG might fail to do so. However, when the protest has no uncertainty, allowing the protest is costless, so the weak type can still achieve the same payo↵ by conceding if mimicking the strong type doesn’t achieve the desired end of getting FG to concede. Thus, mimicking is risk-free, and this prevents the existence of separating equilibrium based on information revelation. However, when the protest can turn into anti-regime mobilization, mimicking the strong type now carries a risk. If mimicking does not get FG to concede, the weak HG is worse than repress the protests and concede immediately. Hence, a separating equilibrium based on information revelation exists in which the strong type chooses to allow the protest, while the weak type finds risky to mimic. These are two aspects where the findings from this model di↵er from Weiss (2013, 2014). First, Weiss argues that “the decision to run this (instability) risk, however small, 59 enables the government to signal resolve, di↵erentiating it from a government that cares less about the international dispute.” (Weiss, 2013,p. 8)Bycontrast,themodelshowsthatcredible signaling (separating equilibrium) only exists if and only if the instability cost generated by the uncertainty is large enough but not too large (at middle levels). In addition, the key information the foreign government needs to gather is not whether the home government cares about the dispute or not, but the fighting capability in case of war, so as to render the foreign government being better o↵ to concede than go to war. Second,inWeiss’conceptualization,thebiggestobstaclethatforestallsnationalistprotests from being a credible signal is that the foreign government is not able to distinguish “sincere protests” in which “participants are self-motivated, self-chosen, and largely self-organized,” from “manufactured protests” defined as “rent-a-crowd” controlled mobs by the state as 59 Emphasis added by me. 119 Volatile Nationalism | Xinru Ma “safety valves” for popular anger. The argument implies that as long as the protest is sin- cere, the management of the protest could solve the information problem at the bargaining. By contrast, the model here shows: Whether the protest is manufactured or self-motivated is one but not the fundamental obstacle that forestalls credible signaling; The key condition is whether the protests generate high enough (but not too high) costs that induce di↵erent types of signalers to separate their responses. In other words, it’s not the nature or the origin of the protests that determine their signaling e↵ects, rather, it is how costly the protest is for the home government. As long as they generate enough instability cost, manufactured protests can still serve as a credible signal. By contrast, even when the protest is sincere nationalist and self-mobilized, if it fails to pose any risk to the regime, the management of protests has no information value to impact international bargaining. This might help explain the puzzling empirical cases when genuine nationalist protests fail to induce foreign concession (The May 4th Protest), and the cases in which foreign government makes con- cession even when she observes the home government’s stage-management of the protests (anti-US protest against the US bombing of the Chinese embassy in Belgrade). Second, not all nationalist protests increase linearly the risk of armed conflicts. Instead, nationalist protests can surprisingly decrease ex ante probability of militarized conflicts, when the uncertainty of nationalist protests transitioning to anti-regime mobilizations is at middle level. In this scenario, two types of governments fully separate themselves with di↵erent ways of managing the protests, and war is avoided with certainty. Nevertheless, the analysis so far hinges on few restrictive assumptions that are worth being relaxed in future extensions. First, it assumes that the probability of a nationalist protest transitioning into anti-regime mobilization is a common knowledge observed by both the home and the foreign government. Empirically, this assumption is highly plausible if one has in mind the size and sophistication of the monitoring and censorship programs and the institutional resources available to monitor collective mobilizations in authoritarian regimes 120 Volatile Nationalism | Xinru Ma like China and Russia. (King et al., 2014; Reuter and Robertson, 2014) However, in future extensions, it is important to extend the model into an incomplete information game in the sense that the transition probability of the protest is unknown to both governments, or at least unknown to the foreign government. Second, this paper only examines stationary sub-game perfect equilibrium. The propo- sitions presented clearly imply that how the home government manages protests in the past should have lingering e↵ects on the foreign government’s belief over time. Thus, one way forward is to model the game in repeated forms. Relatedly, it will be worth exploring what factors influence the foreign government’s priors, especially in the pooling equilibrium. Another important research agenda is to examine alternative explanations, especially the elite divisions. So far, the model assumes the government and the citizens as unitary actors, however, as the literature review shows, hawkish critics within the government may foment nationalist protests to strengthen their position in internal power struggles, providing an opening for public opinion to influence foreign policy. (O’donnell et al., 2013)Thus,to further examine the conditions for authoritarian governments to use nationalist protests as credible signals, alternative explanations, especially domestic factors, need to be examined. 4.5 Empirical Illustration In this section, I use the 2014 anti-China protest in Vietnam as an example, introduce my empirical strategy to measure the key parameter - the transition probability ✓ -andtrace Vietnamese and Chinese governments’ reactions to the protest and their consequent behav- iors at the international crisis. With this, I show that with a high probability of evolving into an anti-regime mobilization, the Vietnamese government swiftly repressed the protest. Upon observing the repression along with a high prior that the Vietnamese government is weak, the Chinese government refused to back down at the bargaining. As described in Chapter 2, there were the following critical junctures of the 2014 anti- 121 Volatile Nationalism | Xinru Ma China protest in Vietnam. On May 2, 2014, China placed an oil-drilling rig - Haiyang Shiyou 981 - into a disputed area in the South China Sea. As a response, the Vietnamese governmentimmediatelysentfrogmenandotherunderwateragentstothearea, anddropped obstacles such as fishing nets and floating objects into the waters, with the aim to disrupt the rig’s placement and operations. However, this incident received little attention from the public until May 11, when hundreds of people gathered and protested against China in central Hanoi and Ho Chi Minh city. By May 13, the crowd swelled to about 20,000, some of whom looted and set fire to more than 15 factories owned by China, Taiwan and Hong Kong in three neighboring provinces of Ho Cho Minh city. This riot was soon cracked down within about one and a half day. By May 18, a planned second protest in Hanoi and Ho Cho Minh city was subdued, with more than 500 people detained for looting and arson. Observing Vietnamese government’s response, the Haiyang Shiyou 981 kept its operation in the disputed waters, and was only moved back to Hainan island on July 16, 2014, two months after the demonstrations. 4.5.1 Measuring ✓ with Social Media Data Systematic data on to what extent a protest would evolve into an anti-regime mobilization is rare. In the introduction, I provided ad hoc anecdotes about how protesters in China imbedded their criticisms towards the regime in their nationalistic slogans, however, it is hard to assess how representative these protesters are. After all, a nationalist protest turns into an anti-regime mobilization that is costly enough for the authoritarian government, only when the anti-regime sentiment is strong and representative enough. To get a better sense of how likely the protest may evolve into an anti-regime mobilization, I examine the comments Vietnamese citizens respond to posts published by key news sources on Facebook in Vietnam, and compute the share of the comments that are anti-China and anti-Vietnam, respectively. 122 Volatile Nationalism | Xinru Ma Similar to the method in Chapter 2, the four Facebook public pages that I examined are: BaoThanhNien(Youth), andTuoiTre(Youth),twoofthemostcirculateddailynewspapers in Vietnam; Quan Doi Nhan Dan Viet Nam (VietNam People’s Army), the ocial public page of Vietnam People’s Army, and BBC Tieng Viet (BBC Vietnamese), BBC’s branch in Vietnam that publishes news in Vietnamese. These public pages are chosen due to 1) their popularity as sources of information among Vietnamese, and 2) their respective ideological bias in the selection and coverage of events and stories. Representing reformist, commercial, ocial (party/military), and Western sources of information, these pages capture a broader readership than any of them by its own. However, there’re some important di↵erences between the specific coding in this chapter and that in Chapter 2. First, rather than retrieve all posts regardless of the topic, here I retrieved only posts that contain the key word “China” in the post text. I then collected all the comments in response to these posts. The data collection procedure covers from May 2, 2014 to May 18, 2014, roughly from the time when China deployed the oil rig to the time when the government subdued all protests. The final outcome is a database of 39 posts and 12,006 comments from four public pages. Figure 4.5.1.1 shows the summary statistics of these four pages. Second, rather than coding whether a comment is nationalistic or not, the focus here is on whether the comment is anti-China or anti-Vietnamese government. What this indicates is that comments that express direct support and love of the nation and nation-leaders are excluded from the coding (coded as “irrelevant”). For instance, in Chapter 2, comments such as “I love my beloved Uncle Ho, Uncle is the idol in my heart forever” or “On this earth there will be no second person like our beloved Uncle Ho” are coded as nationalistic, as they expressed direct support and love of the nation and nation-leaders. By contrast, given the interest here is whether the public criticizes more China or its own regime when faced with aChina-relatedtrigger,thiscommentiscodedintoneithercategory. 123 Volatile Nationalism | Xinru Ma Figure 4.5.1.1: Summary Statistics of Four Public Pages To measure the shares of anti-China and anti-Vietnamese government comments respec- tively, I randomly select 10% of the total comments (1,200 comments), which I used to train a machine learning classifier that predicts the category to which all comments in the final dataset correspond. Again, all the posts and comments are translated with Google Translate API from Vietnamese into English. I then coded each comment as anti-China or anti-Vietnamese government. Using the dataset of labeled comments, machine learning classifiers that allow us to estimate the probability that each comment is anti-China or anti-Vietnamese government are constructed. The analysis is done in three steps: text pre-processing, training and validation of the classifiers, and application to the entire corpus. Using XGBoost as the key algorithm, theaccuracy, precisionandrecallofthismethodwere0.88, 0.33, and0.32foranti- Vietnamesegovernmentcomment,and0.86,0.2,0.18foranti-Chinacomment,respectively. 60 60 The low precision and recall stats of these two models are partly due to the unbalanced structure of this dataset. Of the hand-coded 1,200 comments, 93 comments are anti-Vietnamese government, while 91 comments are anti-China. By comparison, 1,016 comments (84.7%) are coded as “irrelevant” as these comments don’t have explicit sentiment that is either anti-China or anti-Vietnamese government. This finding itself is interesting, as it shows majority of the comments in response to China-related posts is not 124 Volatile Nationalism | Xinru Ma Figure 4.5.1.2 lists top key features associated with each category of comments. Figure 4.5.1.2: Top Features for Anti-China and Anti-Vietnamese Government Comments This model is then applied to predict the labels for all the comments in the dataset. The final result shows that 6.1% of the total comments are anti-Vietnamese government. Though the absolute amount of this share seems low, it is important to note that only 3.2% of the comments are against China. The following comment demonstrates well this attitude: “Certainly the authorities are afraid of the air coming down the road of protesters, which is considered more dangerous for the regime than China’s 981 rig. Evidence is on Sunday...Saigon ordered the soldiers working in Saigon cannot leave the locality for any reason (this is internal news, 100% real). Onlyafter11amthatdaywillthisorderbetakenaway. Deeplyafraid of losing the throne, the people-watching party is the No. 1 threat. China is still more trusted by the party...except for military clashes between the two countries.” directly commenting on both regimes. In future revision, it is important to increase the sample size of hand- coded data. Also, it’s worth coding the “irrelevant” comments and see if there’re common themes among these comments. 125 Volatile Nationalism | Xinru Ma 4.5.2 Reactions of Vietnamese Government In short, despite China’s controversial deployment of oil rig in the disputed waters, Viet- namese public quickly connect the issue and the protest with domestic grievances, while the anti-Vietnamese government sentiment discerned in the comments in response to China- related posts surpassed the intensity of anti-China sentiment. In the next section, I will examine the protest from the elites’ perspective, in order to illustrate whether Vietnamese and the Chinese governments acted in manners consistent with the proposed mechanism from the model. When China first deployed the oil rig on May 2, Vietnamese government ocials reacted byissuinganumberofstatementsandpositionpapersthroughdiplomaticchannelsandpress conferences, labeling China’s behavior as “illegal” and “invalid.” 61 However, such statements remained largely vague about what exactly Vietnamese government would do. For instance, at various times, Le Hai Binh, Vietnamese Foreign Ministry spokesman said that Vietnam “resolutely opposed” China’s action, and would “take appropriate measures” in response. 62 However, it is unclear what measures exactly would be taken, nor a clear answer was given when asked whether Vietnam would file a case against China at the international court. When the protest first broke out on May 11, Vietnamese government allowed reporters and media to cover the protests. However, the initial tolerance backpedalled as the protests turned violent on May 13. With rioters targeting factories in industrial parks that have links with China, the government first issued a media blackout, and allowed no coverage on the violence and vandalism in domestic media. The Prime Minster also sent out a text messagetocitizens. Thoughtheviolenceintheprotestswasnotcondemneddirectly,thetext message urged citizens to defend the sovereignty with actions in line with the law, and that 61 “Position paper of Viet Nam on China’s illegal placement of Haiyang Shiyou 981 oil rig in the exclusive economic zone and continental shelf of viet nam.” Consulate of the Socialist Republic of Viet Nam in New York. July 7, 2014. 62 “Vietnam says Chinese O↵shore Rig is Illegal; China Disagrees.” VOA News. May 5, 2014. http: //www.reuters.com/article/uk-vietnam-china-oil-idUKKBN0DL0DQ20140505 126 Volatile Nationalism | Xinru Ma “bad elements should not be allowed to instigate extremist actions that harm the interests and image of the country.” 63 In the following day on May 17, the Prime Minister called an end to all “illegal protests” in the country, and ordered the Ministry of Public Security and provincial governments to “conduct concerted and determined measures not to allow illegal protests that cause security and social order disturbances.” 64 The protests soon fizzled out across the country on May 18, as major cities such as Hanoi andHoChiMinhcitysaturatedwithpoliceandsecuritypersonnelbothinuniformandplain clothes. Demonstrators got dragged away, and major routes to Chinese embassy and the consulatewereguardedheavilywithpolice. TherhetoricinVietnamesemediaontheprotests also changed. According to Hanoi Moi, Vietnamese Communist Party’s mouthpiece, the anti-China protests were incited by “anti-state forces,” and “their conspiracy and intention has been to disrupt the great national unity, instigating national hatred and separating relations between Vietnam and China.” 65 As one protestor commented, “PM Nguyen Tan Dung told people that they have rights to show their patriotism on Thursday, but in reality, the government has banned us.” 66 4.5.3 Reactions of Chinese Government The above analysis demonstrates that leaders of Vietnam were highly cautious about the nationalistic protests, and were ready to curtail the protest as soon as it began to jeopardize domestic stability. On the other hand, observing the willingness and ability of Vietnamese government to repress the demonstration, China refused to make any compromises in this 63 “Vietnam PM Nguyen Tan Dung urges citizens to defend sovereignty against China.” CBC News. May 16, 2014. http://www.cbc.ca/news/world/ vietnam-pm-nguyen-tan-dung-urges-citizens-to-defend-sovereignty-against-china-1.2645334 64 “Vietnam’s PM calls for end to ’illegal’ anti-China protests.” Associated Press. May 17, 2014. http: //www.taipeitimes.com/News/front/archives/2014/05/18/2003590615 65 “Hanoi warns anti-China protesters to stop.” RNW Media. https://www.rnw.org/archive/ hanoi-warns-anti-china-protesters-stop 66 “Anti-China protests in Vietnam cities fizzle out as government deploys huge secu- rity.” The Straits Times. May 18, 2014. http://www.straitstimes.com/asia/se-asia/ anti-china-protests-in-vietnam-cities-fizzle-out-as-government-deploys-huge-security 127 Volatile Nationalism | Xinru Ma crisis. For instance, China Daily,China’sstate-runmedia,attributedtheproteststoacouple of extremists, and questioned whether the local authorities acquiesced in the escalation into violent crimes. It urged Vietnam to “shoulder full responsibility for whatever serious consequences the crimes will have on China-Vietnam ties.” It also accused that “the lethal riots are proof that China’s calls for dialogue over territorial claims have fallen on deaf ears inVietnam.” 67 CriticizingHanoiauthoritiesforforcingpublicopiniontoturnagainstChina, an editorial article on Global Times said that “Vietnam has ended up embarrassing itself in front of people across the globe.” It added, “the truth is that the Vietnam government does not know the boundaries of nationalism and neither do they have ability to control the violence.” 68 AccordingtothedirectorgeneraloftheForeignMinistry’sDepartmentofOcean and Boundary A↵airs, the whole incident was “a media hype and a show put up in front of the international audience by the Vietnamese side.” 69 The spokesperson of Chinese Foreign Ministry also warned that the Vietnamese government “has an inescapable responsibility for the beating, smashing, looting and burning targeted at China and other countries,” and demanded Vietnam to take swift measures to punish the perpetrators. 70 In short, from the initial outburst to the final crushing of the protest, there exists little evidence of China being willing to make compromise in its foreign policies. The Haiyang Shiyou 981 kept its operation in the disputed waters, and was only moved back to Hainan island on July 16, 2014, two months after the demonstrations. 67 “Vietnam must end violence.” China Daily. May 16, 2014. http://www.chinadaily.com.cn/cndy/ 2014-05/16/content_17511251.htm 68 > ƒ ⇢ ä W S 8 ¢ Á ç L b M "∫ ∞ < . May 16, 2015. http://opinion.huanqiu.com/editorial/ 2014-05/4996625.html 69 “Vietnam PM Nguyen Tan Dung urges citizens to defend sovereignty against China.” CBC News. May 16, 2014. http://www.cbc.ca/news/world/ vietnam-pm-nguyen-tan-dung-urges-citizens-to-defend-sovereignty-against-china-1.2645334 70 China lodges solemn protest with Vietnam over deadly violence. Xinhua News. May 16, 2014. http: //news.xinhuanet.com/english/china/2014-05/16/c_133337186.htm 128 Volatile Nationalism | Xinru Ma 4.6 Proofs of Propositions 4.6.1 Proofs of Propositions When Nationalist Protests Have No Uncertainty Claim 1. When nationalist protests have no uncertainty of turning into anti-regime mo- bilizations, there exists no separating equilibrium in which a strong HG always allows the nationalist protest to signal resolve, and a weakFG fully reveals himself by always repressing the nationalist protests. Proof. SupposethereexistsaseparatingequilibriuminwhichastrongHGalwaysallowsthe protest, while a weak HG always represses. When FG observes a protest (in other words, in the right-hand information set), she will assign probability 1 to HG being strong, thus, = 1. WhenFG observes the repression of the protest, she will assign probability 0 toHG being strong, thus, =0. FG’sexpectedutilityfromconcedingandresistingwhenobservingaprotest,respectively, is EU F (Concede|Allow)=1and EU F (Resist|Allow)=1 p c F < 0(byAssumption1). Thus,FG will always concede when observing a protest. Similarly,FG’s expected utility from conceding and resisting when observing the repres- sionoftheprotest, respectively, isEU F (Concede|Repress)=0andEU F (Resist|Repress)= 1. Thus,FG will always resist when observing a protest. Since FG’beliefsareBayesianbyconstruction,and FG’s strategy is a best response given those beliefs, this is an equilibrium if and only if neither type of HG has an incentive to deviate. When HG is strong, he gets p c H g if he deviates, which is less than his current payo↵ 1. However, when HG is weak, he gets 1 from deviation, which is greater than his current payo↵ g. Thus, this separating equilibrium doesn’t hold, as the weak type always seeks deviation by allowing the protests. 129 Volatile Nationalism | Xinru Ma Proposition 1. [Cheap Talk Pooling Equilibrium - Peace Outcome]. In incomplete infor- mation setting when the protest has no uncertainty, when ⇡ 1 p+c F , the following pooling equilibrium exists: HG allows protest regardless of his type. Upon observing the protests, FG concedes and war is avoided with certainty. On o↵-equilibrium path, FG concedes after observing the repression of the protest, when ⇤ = 1 p+c F . Proof. Let denote FG’s updated belief that HG is strong when she observes the protest is repressed, and denoteFG’s updated belief that HG is strong when protest is allowed. InapoolingPBE,bothtypesofHGdeploythesamemessage,andthere’snoinformation updatingalongtheequilibriumpath. Thatis, = Pr(Allow|Strong)Pr(Strong) Pr(Allow|Strong)Pr(Strong)+Pr(Allow|Weak)Pr(Weak) = 1⇤ ⇡ 1⇤ ⇡ +1⇤ (1 ⇡ ) = ⇡ . FG’s expected utility from conceding after observing the protest is: EU F (Concede|Allow)= ⇡ ⇤ 0+(1 ⇡ )⇤ 0=0 FG’s expected utility from resisting after observing the protest is: EU F (Resist|Allow)= ⇡ ⇤ (1 p c F )+(1 ⇡ )⇤ 1 =1 (p+c F )⇡ Therefore, when EU F (Concede|Allow) EU F (Resist|Allow), that is, when ⇡ 1 p+c F , FG will always concede after observing the protest. O↵theequilibriumpath, FG’sexpectedutilityfromconcedingafterobservingrepressing protest is: EU F (Concede|Repress)= ⇤ 0+(1 )⇤ 0=0 130 Volatile Nationalism | Xinru Ma FG’s expected utility from resisting after observing repressing protest is: EU F (Resist|Repress)= ⇤ (1 p c F )+(1 )⇤ 1 =1 (p+c F ) Thus,FG’s best response on o↵-the-equilibrium path depends on her belief : BR F (Repress)= 8 >< >: Concede if 1 p+c F Resist if < 1 p+c F Since FG’s beliefs are Bayesian by construction, and FG’s strategy is a best response given those beliefs, this is an equilibrium if and only if neither type of HG has an incentive to deviate. When 1 p+c F ,strong HG obtains 1 g if he deviates, which is less than current payo↵ 1. Weak HG gets 1 g if he deviates, which is also less than current payo↵ 1. Thus, when 1 p+c F ,neithertypedeviates,andthisequilibriumholds. Proposition1 summarizes the equilibrium strategies. Nevertheless, there exists a second pooling equilibrium, which has the same along- equilibrium-path strategy as the one indicated in Proposition 1. The only di↵erence is that on o↵-equilibrium path, FG resists if her belief< ⇤ = 1 p+c F .Inthisscenario,observ- ing FG’s resistance, a strong HG resist and a weak HG concedes. Strong HG thus obtains p c H g if he deviates, which is smaller than current payo↵ 1. Weak HG obtains g if he deviates, which is also less than current payo↵ 1. This equilibrium holds as neither type of HG deviates. However, this equilibrium doesn’t generate new insights as it ultimately shows that FG’s responses on the o↵-equilibrium path is determined by FG’s belief that HG is strong when protest is repressed, which is the same intuition from Proposition 1. Thus, this equilibrium is proved here but not presented in the main text. 131 Volatile Nationalism | Xinru Ma Proposition 2. [Cheap Talk Pooling Equilibrium - Probabilistic War Outcome]. In incom- plete information setting when the protest has no uncertainty, when⇡< 1 p+c F , the following pooling equilibrium exists: HG allows protest regardless of his type. Upon observing the protests, FG resists. Following FG’s resistance, a strong HG resists (thus war happens), and a weakHG concedes (blu↵ed and failed). Here, war occurs with positive probability, and this probability is not modified by the signaling. On o↵-equilibrium path, FG resists after observing the repression of the protest, when the belief FG assigns to HG being the strong type after observing repression is smaller threshold< ⇤ = 1 p+c F . Proof. As demonstrated in above proof, both types of HG deploy the same message, and there’s no information updating along the equilibrium path, thus = ⇡ .Therefore,when EU F (Concede|Allow)= ⇡ ⇤ 0+(1 ⇡ )⇤ 0=0<EU F (Resist|Allow)= ⇡ ⇤ (1 p c F )+ (1 ⇡ )⇤ 1, that is, when⇡< 1 p+c F ,FG will always resist after observing the protest. O↵ the equilibrium path, FG resists if EU F (Concede|Repress)= ⇤ 0+(1 )⇤ 0 < EU F (Resist|Repress)=1 (p+c F ) ,thatis,when< 1 p+c F . Now check HG’s incentive for deviation. When< 1 p+c F ,FG resists on o↵-equilibrium path. Observing FG’s resistance, a strong HG resist and a weak HG concedes. Strong HG thus obtains p c H g if he deviates, which is smaller than current payo↵ p c H .Weak HG obtains g if he deviates, which is also less than current payo↵ 0. This equilibrium thus holds as neither type of HG deviates. 4.6.2 Proofs of Propositions When Nationalist Protests Have Uncertainty Proposition 3 (Credible Signaling Equilibrium). . In incomplete information setting when the protest has uncertainty, there exists a separating equilibrium in which a strongHG always allows the protests, and a weakHG always represses the protests, if the following condition is met: 1+g k <✓< 1 p+c H +g k . Upon observing the protest,FG always concedes. Upon observing the repression of the protest, FG always resists. Observing FG’s resistance, the weak HG 132 Volatile Nationalism | Xinru Ma concedes. Proof. WhenFG observes a protest (in other words, in the right-hand information set), she will assign probability 1 toHG being strong, thus, = 1. WhenFG observes the repression of the protest, she will assign probability 0 to HG being strong, thus, =0. FG’sexpectedutilityfromconcedingandresistingwhenobservingaprotest,respectively, is EU F (Concede|Allow)=0and EU F (Resist|Allow)=1 p c F < 0 (by Assumption 1). Thus,FG will always concede when observing a protest. FG’s expected utility from conceding and resisting when observing the repression of the protest, respectively, is EU F (Concede|Repress)=0and EU F (Resist|Repress)=1. Thus, FG will always resist when observing a protest. Since FG’beliefsareBayesianbyconstruction,and FG’s strategy is a best response given those beliefs, this is an equilibrium if and only if neither type of HG has an incentive to deviate. When HG is strong, he gets p c H g if he deviates, while his current payo↵ is 1 ✓k.Thus,astrong HG doesn’t deviate if and only if 1 ✓k > p c H g,thatis, ✓< 1 p+c H +g k .Similarly,when HG is weak, he gets 1 ✓ k if he deviates, while his current payo↵ is g.Thus,aweakHG doesn’t deviate if and only if 1 ✓ k< g,thatis,✓> 1+g k . Summarizing, this separating equilibrium is sustained when 1+g k <✓< 1 p+c H +g k . Proposition 4 (Blung Pooling Equilibrium) . . In incomplete information setting when the uncertainty of a nationalist protests is smaller than threshold✓<✓ ⇤ = 1+g k , the following pooling equilibrium exists: HG allows protest regardless of his type. Upon observing the protest, FG concedes, when ⇡ 1 p+c F . On o↵-equilibrium path, FG resists after observing the repression of the protest, when the belief that FG assigns to HG being the strong type when observing protests is smaller than< ⇤ = 1 p+c F . Proof. In pooling PBE, both types of HG deploy the same message, and there’s no in- formation updating along the equilibrium path, thus = ⇡.Therefore, FG will always 133 Volatile Nationalism | Xinru Ma concede after observing the protest, whenEU F (Concede|Allow)= ⇡ ⇤ 0+(1 ⇡ )⇤ 0=0 EU F (Resist|Allow)= ⇡ ⇤ (1 p c F )+(1 ⇡ )⇤ 1, that is, ⇡ 1 p+c F (1) O↵theequilibriumpath, FGalwaysresistsifEU F (Concede|Repress)= ⇤ 0+(1 )⇤ 0< EU F (Resist|Repress)=1 (p+c F ) ,thatis: < 1 p+c F (2) Now check HG’s incentive for deviation. When< 1 p+c F ,FG resists on o↵-equilibrium path. Observing FG’s resistance, a strong HG resist and a weak HG concedes. A weak HG obtains g if he deviates, while his current payo↵ is 1 ✓ k.Thus,aweak HG doesn’t deviate if and only if 1 ✓ k> g,thatis: ✓<✓ ⇤ = 1+g k (3) Similarly, a strongHG thus obtains p c H g if he deviates, while his current payo↵ is 1 ✓k .Thus,astrong HG doesn’t deviate if and only if p c H g < 1 ✓k ,thatis: ✓<✓ ⇤⇤ = 1 p+c H +g k (4) Compare the values of ✓ ⇤ and ✓ ⇤⇤ .Since ✓ ⇤ ✓ ⇤⇤ < 0, we use the more restrict condition as the final condition (the smaller ✓ ⇤ ). The pooling equilibrium specified in Proposition 4 thus holds when conditions (1), (2), and (3) are hold. Proposition 5 (Failed Signaling Pooling Equilibrium). . In incomplete information set- ting when the uncertainty of a nationalist protests exceeds threshold✓>✓ ⇤ = 1 p+c H +g k , 134 Volatile Nationalism | Xinru Ma the following pooling equilibrium exists: HG represses protest regardless of his type. Upon observing the repression of the protest,FG resists, when⇡< 1 p+c F . On o↵-equilibrium path, FG concedes after observing the protest, when the belief that FG assigns to HG being the strong type when observing the protest exceeds ⇤ = 1 p+c F . Proof. InthispoolingPBE,bothtypesofHGrepresstheprotest,andthere’snoinformation updating along the equilibrium path, thus = ⇡.Therefore, FG will always resist after observing the repression of the protest, whenEU F (Concede|Repress)= ⇡ ⇤ 0+(1 ⇡ )⇤ 0= 0<EU F (Resist|Repress)= ⇡ ⇤ (1 p c F )+(1 ⇡ )⇤ 1, that is, ⇡< 1 p+c F (5) O↵ the equilibrium path, FG always concedes if EU F (Concede|Allow)= ⇤ 0+(1 )⇤ 0 EU F (Resist|Allow)=1 (p+c F ) ,thatis: 1 p+c F (6) NowcheckHG’sincentivefordeviation. When 1 p+c F ,FGconcedesono↵-equilibrium path. A weak HG obtains 1 ✓ k if he deviates, while his current payo↵ is g.Thus,a weak HG doesn’t deviate if and only if g> 1 ✓ k,thatis: ✓>✓ ⇤ = 1+g k (7) Similarly, a strongHG obtains 1 ✓k if he deviates, while his current payo↵ is p c H g. Thus, a strong HG doesn’t deviate if and only if p c H g > 1 ✓k ,thatis: ✓>✓ ⇤⇤ = 1 p+c H +g k (8) 135 Volatile Nationalism | Xinru Ma Comparethevaluesof✓ ⇤ and✓ ⇤⇤ ,weusethemorerestrictconditionasthefinalcondition (thebigger✓ ⇤⇤ )ThepoolingequilibriumspecifiedinProposition5thusholdswhenconditions (5), (6), and (8) are hold. 136 Volatile Nationalism | Xinru Ma 5 Conclusions In EastandSoutheast Asia, atextbookstoryis thatnationalism isrising, and isincreasingly raising the risk of armed conflict in the region. This dissertation challenges this conventional wisdom on two fronts: Is nationalism really rising? How do we know? And second, even if we take this “rising” nationalism as granted, underwhatconditionswillnationalismdriveaggressivebehavior,especiallyintimesofcrisis? In a nutshell, the central take-away of this dissertation is that nationalism is not rising in the region, and that even if it is, nationalism is not necessarily as dangerous as the conventional wisdom perceives. We need to understand both the baseline and volatility dimensions of nationalism to correctly assess this phenomenon, and the scope conditions for nationalism to actually induce conflicts need to be specified, not assumed. The dissertation thus has two key pieces that are interconnected and complementary to each other. In the first part, it argues that rather than focus on the immediate aftermath of certain triggers, there’re two dimensions that we need to consider when conceptualizing nationalism. First, a baseline dimension, which denotes how nationalistic an individual is when there’s no trigger. This dimension can be perceived as a measurement of attitude. The second dimension - which I term as volatility - denotes the actions an individual may take when there’s a trigger, which further assesses whether the nationalistic attitude is strong enough for a person to endure some costly actions. At the macro-nation level, some countries have very volatile nationalism: they experience surges of nationalism in response to very small stimuli. Other countries have very involatile nationalism: even large triggers don’t cause a response. Methodologically, by applying automated text analysis methods to social media data in 137 Volatile Nationalism | Xinru Ma Vietnam and the Philippines, this dissertation generates two key insights that challenge and add more nuance to the conventional wisdom of “rising” nationalism: First, the baseline nationalism is low and stable in both countries, signaling an overestimation of nationalistic attitudes in extant literature. Second, Vietnam and the Philippines di↵er tremendously in terms of how they react to triggers. Vietnam experiences surges of nationalism in response to China-related stimuli, however, such triggers merely cause responses in the Philippines. These, altogether, show that the two dimensions proposed in this dissertation could help categorize countries into a 2*2 metrics, which further jointly di↵erentiate the nature and levels of nationalism in di↵erent countries. If the first part of this dissertation is descriptive, the second part is more causal. Specifi- cally, it argues that even if we take the “rising-nationalism” as granted, nationalism doesn’t necessarily drive aggressive behavior. With a formal model, the second part of this disserta- tion seeks to clarify the mechanisms through which nationalism escalates the risks of armed conflicts. The formal model emulates the strategic bargaining between nationalist protesters and the governments, and explores the scope conditions under which the government is will- ing and able to use nationalism – exemplified in the instances of nationalist protests - as leverages in international bargaining, and the conditions under which the foreign govern- ment is willing and able to view that as a credible signal. This e↵ort joins the recent works that link nationalist protests with international bargaining (e.g. Weiss (2013)and Weiss (2014)), however, it advances this literature by clarifying the specific conditions for how and why this is the case. Specifically, it answers three questions that are keen to the e↵ect of nationalist protests on crisis bargaining: First, what assumptions are needed on the nature of the dispute and the bargaining environment to make nationalist protests a potential sig- nal in international bargaining? Second, what is the exact problem that induces bargaining failure, which ultimately leads to militarized conflicts? And third, why is nationalist protest a signal? And why is it updating? 138 Volatile Nationalism | Xinru Ma Themodelfocusesononekeyfeatureofnationalistproteststhatiscentraltothestrategic calculation of the home government when entering international bargaining: the uncertainty that nationalist protests evolving spontaneously into anti-regime mobilizations. Indeed, it is the instability cost generated by this uncertainty that di↵erentiates di↵erent types of home governments. Specifically, there exists a U-shaped relationship between the probability of nationalistproteststransitioningtoanti-regimemobilizationsandtheprobabilityofbargain- ing failure and the ensuing conflicts. When the probability of transition is low, nationalist protests are not costly enough, thus failing to credibly reveal information; the high probabil- ity of transition, on the other hand, is likely to result in repression of all protests, forestalling them being observed as a signal. Thus, nationalist protests have the best chance to be a credible signal with mid-level transition probability. Here, nationalist protests are costly enough to allow for credible information transmission, while the home government is not so averse that it resorts to repression. In a nutshell, this dissertation provides a wealth of new data and analysis of nationalism andthesecuritydynamicsofEastAsiainformedbylocalknowledgeandregionalperceptions. Ito↵ersfull-spectrumcomparativestudiesofnationalismbycarefullymeasuringnationalism withlatestmethodsinbigdata,combinedwitharigorousanalysisofthee↵ectofnationalism oninternationalcrisiswithformalmodeling. Bycombiningstudiesonnationalismwiththese latest methods, this dissertation aims to achieve three broader impact goals: • Facilitating interdisciplinary conversation. Progressindataminingandmachinelearn- ing has demonstrated its ability to enhance reliable, timely, and comprehensive data collection that used to require intense human labor (King and Lowe, 2003). By adopt- ing automated text analysis methods to social media data fine-grained to daily or even hourly basis, this dissertation seeks to push the boundaries of our field by embracing latest developments in both fields of data science and International Relations. 139 Volatile Nationalism | Xinru Ma • Incorporating micro-level analysis. Social media have desirable properties that tradi- tional public opinion surveys cannot match, especially its ability to provide rich text data composed of posts and comments made by individual social media users in re- sponse to an event in real time (Steinert-Threlkeld, 2017). Through rigorous content analysis of the rich details in social media data, this dissertation seeks to move beyond themacro-surveyfocusofpastresearchbymappingthemicro-dynamicsofnationalism with a focus at the individual level. • Enhancing comparative perspectives on nationalism. Comparative studies of nation- alism are rare, especially those based on careful methodologies that allow for explicit comparisonofphenomena. Empirically, massivevolumeofstudieshasbeenexclusively devoted to China, by focusing on how and to what extent Chinese nationalism can in- fluence China’s foreign policies. As a result, proposed measurements and theories are rooted merely in China’s history or culture, thus lacking cross-regional perspectives on nationalism, as well as external validity for most, if not all, concerned actors in the region. By comparing nationalism in Vietnam and the Philippines, this dissertation seeks to put the dyadic countries into a comparative perspective, and advances the literature by moving beyond the traditional single-country focus. Nevertheless, due to the limited scope of this dissertation, there’re still many areas that requirefurtherwork. Hence, Iconcludethisdissertationwithseveralpotentialwaysforward. An immediate extension of this dissertation is to apply this two-dimensional conceptual- ization and measurement method to other countries such as South Korea, Japan and China, so as to depict the great variance in the levels and types of nationalism in East Asia. As an exploratory e↵ort, I used the International Social Survey Program (ISSP) on National Iden- tity, hosted by the GESIS-Leibniz-Institute for the Social Sciences in Germany, to provide a preliminary assessment of how nationalism in other Asian countries looks like with this 140 Volatile Nationalism | Xinru Ma two-dimensional conceptualization. (See Appendix 5.1.1 for specific survey questions used, the caveats of using survey data, as well as the limitation when interpreting the results). Initial results show that while the Japanese public has a high level of baseline identification and a low level of volatility, the Korean public demonstrates a high level of volatility. This indicates that nationalism in South Korea and Japan should exhibit di↵erent e↵ects when faced with the same trigger, thus constituting another pair of interesting cases. Similarly,onecouldalsowonderhowwellthisconceptualizationandmeasurementapplies to non-East Asian countries, so as to test the hypotheses on nationalism on a global scale. Specifically, in the midst of an increasingly aggressive Russia, the terrorist threat posed by aburgeoningIslamicStates,therefugeecrisis,andtheenduringEurozonecrisis,agrowing literature contends that nationalism is sweeping across Europe, and that such an inward- focused, self-absorbed EU has constituted a major threat to both European integration and transatlantic alliance. 71 Thus, extending this measure to non-East Asian countries would be helpful for us to assess whether nationalism is rising globally. A third extension is to investigate the formation and di↵usion of nationalism from on- line to o↵-line nationalist protests, which serves as a bridge between chapters that measure the two dimensions of nationalism with social media data and the chapter that investigate the interaction between o↵-line protesters and governments. The exercise of formalization in this dissertation allows me to clarify the scope conditions and my expectations of the data-generating process. In future research, it is important to test the generalizability and empirical relevance of the theoretical model with data that links nationalist protests, repres- sion behaviors of the home government, and international crises. On one hand, extant tests of nationalism heavily focus on ethno-nationalism with a 71 Forexample,see“FromGreeceToFranceToChina,NationalismIsBackInABigWay.”TheHungton Post. July1,2015. “NationalismWreaksHavocinDividedEurope.”TheWallStreetJournal. Feb. 15,2016. “Europe’s Populist Insurgents: Turning Right.” The Economist. January. 14, 2014 “Europe Rediscovers Nationalism.” Stratfor. January. 11, 2015. “UK Election: Nationalism on March Across Europe.” CNN. May 6, 2015. 141 Volatile Nationalism | Xinru Ma geopolitical focus on Europe. For instance, to date, there are only five articles in total that quantitatively test the e↵ect of nationalism on militarized conflicts, with four of them focusing on irredentist crises, separatism, marginalized ethnic minorities, and religious na- tionalism, and their e↵ects on intra-state conflicts. ( Carment and James, 1995; Braumoeller, 1997; Fox, 2004; Cederman et al., 2009)Thesetestsdonotshedmuchlightonthee↵ectof anti-foreign protests, as they often involve only domestic bargaining between the home state and separatists/minorities without the direct engagement of third-party foreign country. On the other hand, event datasets on protests that use automated coding are not rare. Recent datasets such as the Uppsala Conflict Data Program (UCDP) and Integrated Cri- sis Early Warning System (ICEWS) have almost realized near-real-time global event data production, and the scale of their automated coding e↵ort has covered 175 countries and nearly 20-million stories (Schrodt, 2012). However, the trade-o↵ of this impressive scale is the lack of nuance in the coding rule, especially when it comes to actors involved in the event. Usually, actors are broadly grouped into one big category such as “governments and state security services,” with little information on the specific government entity, either at the central or local level, involved in the event. For instance, the only actor related with China in UCDP’s actor list is “Government of China.” Information on the organizers of the protest or protesters themselves is even more rare. Consequently, a dataset with fine-grained details on the size, location, the groups, state and sub-state entities involved in the protests in East Asia is needed, with the goal to enhance our understanding of the formation and di↵usion of the protests. What this requires is detailed pre-defined lists of actors and loca- tions specific to the countries of interest, as well as the words identifying them in natural language, preferably informed by local knowledge. In short, nationalism remains both central to and marginalized in mainstream IR, and the mechanisms for nationalism to influence international crises await more investigation. Nationalism is not just an issue of domestic politics. Given more and more conflicts that 142 Volatile Nationalism | Xinru Ma have been or potential conflicts that might be generated by nationalism, it is high time for IR scholars to re-examine nationalism, starting with a proper conceptualization and measurement of nationalism. 143 Volatile Nationalism | Xinru Ma 5.1 Appendix 5.1.1 Mapping Other Cases with Two-Dimensional Conceptualization of Na- tionalism with ISSP Data (1995-2013) As emphasized throughout this dissertation, one important caveat of using public opinion polls to measure nationalism is that they only measure expressed attitude of the respondents, while the gap between expressed attitude and observed actions is understudied. (UK, 2019; Matfield, 2015; Granberg and Holmberg, 1990) This was also one major motivation for this dissertation to take consideration “volatility” - a dimension that denotes how likely an individual responds to triggers in times of crises - into the conceptualization of nationalism. Using ISSP data to gauge the two dimensions of nationalism is no exception. Thus, the results presented below should be interpreted as very preliminary exploratory e↵orts to see the potential variation of nationalism in other East Asian countries that haven’t been studied in this dissertation. Specifically, results on the baseline and volatility dimensions of nationalism based on survey data might be significantly biased (most likely higher) than results derived from social media data. Moreover, the framing of the survey questions available in ISSP may also be limited in gauging the volatility dimension of nationalism, namely, how likely an individual would be in following through on a course of action despite the risk of conflict. A more appropriate questiontogaugethisdimensionwouldbesomethingaskingifanindividualwouldbewilling to fight on behalf of the state or some other indication of their willingness that captures if theyareactuallywillingtoactuponthetriggers. Itisonethingtosaythestateshouldfollow interests despite the risk of conflicts in the abstract and quite another when we attempt to gauge how likely the individual in such a scenario will do so. This is probably even more the case when asked in the absence of a trigger or in the midst of a crisis. Of course, in an ideal world, one may design and conduct his own survey to capture the 144 Volatile Nationalism | Xinru Ma volatility dimension. For now, as preliminary exploratory e↵ort, the baseline dimension is gaugedasafunctionofnationalpride,whichconsistsfourelements: politicalpride,economic pride, military pride, and pride in the history. 72 Volatility is measured as a function of the willingness to express and protect their perceived distinctive characteristics and interests even if it is wrong or leads to conflicts. 73 Figure 5.1.1 presents the results. The upper panel shows changes in national pride in KoreaandJapanovertheyears. Thelowerpanelshowsthemeanofanswerstothequestions used to gauge volatility. The graphs imply that while the Japanese public has a high level of baseline identification and a low level of volatility, the Korean public demonstrates a high level of volatility (relatively speaking). 72 The specific survey question is: How proud are you of (R’s country) in each of the following: a. Its political influence in the world; b. Economic achievements; c. Armed forces; d. History. Answers: 1. Very proud; 2. Somewhat proud; 3. Not very proud; 4. Not proud at all; 8. Cannot choose, don’t know; 9. NA, refused. 73 The specific survey questions are: 1) People should support their country even if in the wrong. 2) Country should follow own interests, even if it leads to conflicts with other nations. Answers for both questions include: 1. Agree strong; 2. Agree; 3. Neither agree or disagree; 4. Disagree; 5. Disagree strongly; 8. Cannot choose, don’t know; 9. NA, refused. 145 Volatile Nationalism | Xinru Ma Figure5.1.1: MappingOtherCaseswithTwo-DimensionalConceptualizationofNationalism with ISSP Data (1995-2013) 146 Volatile Nationalism | Xinru Ma References Abbass, S.(2010). Chinesestudentsprotestthetreatyofversailles(themayfourthincident), 1919. Alexander, A. (2006). Suez and the high tide of arab nationalism. Anderson, B. (1983). Imagined communities: Reflections on the origin and spread of nation- alism.VersoBooks. Atkeson, L. R. and R. M. Alvarez (2018). The Oxford Handbook of Polling and Survey Methods. Oxford University Press. Batongbacal, J. (2015, 11). Implications of the philippines v. china award on jurisdic- tion. Asia Maritime Transparency Initiative. https://amti.csis.org/implications-of-the- philippines-v-china-award-on-jurisdiction/. Befu, H. (1993). Cultural nationalism in East Asia: Representation and identity,Volume39. RoutledgeCurzon. Blank, T. and P. Schmidt (2003). National identity in a united germany: Nationalism or patriotism? an empirical test with representative data. Political Psychology 24(2), 289– 312. Boix, C. and M. W. Svolik (2013). The foundations of limited authoritarian government: In- stitutions,commitment,andpower-sharingindictatorships. TheJournalofPolitics 75(2), 300–316. Bonikowski,B.(2016). Nationalisminsettledtimes. AnnualReviewofSociology 42,427–449. Braumoeller, B. F. (1997). Deadly doves: Liberal nationalism and the democratic peace in the soviet successor states. International Studies Quarterly 41(3), 375–402. Breuilly, J. (1993). Nationalism and the State. Manchester University Press. Brockett, C. D. (2005). Political movements and violence in Central America.Cambridge University Press. Brubaker, R. et al. (1996). Nationalism reframed: Nationhood and the national question in the new Europe. Cambridge University Press. Brubaker, R. and D. D. Laitin (1998). Ethnic and nationalist violence. Annual Review of Sociology 24(1), 423–452. Bunge, M. (1996). Finding philosophy in social science. Yale University Press. Carey, S. C. (2006). The dynamic relationship between protest and repression. Political Research Quarterly 59(1), 1–11. 147 Volatile Nationalism | Xinru Ma Carlson,A.(2009). Aflawedperspective: thelimitationsinherentwithinthestudyofchinese nationalism 1. Nations and Nationalism 15(1), 20–35. Carment, D.andP.James(1995). Internalconstraintsandinterstateethnicconflict: Toward acrisis-basedassessmentofirredentism. Journal of Conflict Resolution 39(1), 82–109. Cederman, L.-E., H. Buhaug, and J. K. Rød (2009). Ethno-nationalist dyads and civil war: Agis-basedanalysis. Journal of Conflict Resolution 53(4), 496–525. Chan, C.-p. and B. Bridges (2006). China, japan, and the clash of nationalisms. Asian Perspective,127–156. Charney, E. (2003). Identity and liberal nationalism. American Political Science Re- view 97(2), 295–310. Cho, I. H. and S.-H. Park (2011). Anti-chinese and anti-japanese sentiments in east asia: the politics of opinion, distrust, and prejudice. The Chinese Journal of International Politics 4(3), 265–290. Christensen, T. J. (1999). China, the us-japan alliance, and the security dilemma in east asia. International Security 23(4), 49–80. Cohen, W. I. (1966). America and the may fourth movement: The response to chinese nationalism, 1917-1921. Pacific Historical Review 35,83. Cohen, W. I. (2010). America’s response to China: a history of Sino-American relations. Columbia University Press. CSIS (2016). Philippines v. china: Arbitration outcomes. Asia Maritime Transparency Initiative. https://amti.csis.org/arbitration-map/. Davenport,C.(2005). Understandingcovertrepressiveaction: Thecaseoftheusgovernment against the republic of new africa. Journal of Conflict Resolution 49(1), 120–140. Davenport, C. (2007). State repression and political order. Annual Review of Political Science 10,1–23. Davenport, C. (2014). How Social Movements Die: Repression and Demobilization of the Republic of New Africa. Cambridge University Press. Dor, A. (2015). Philippine nationalism and the south china sea. The Diplomat. https://thediplomat.com/2015/08/philippine-nationalism-and-the-south-china-sea/. Downs, E. S. and P. C. Saunders (1999). Legitimacy and the limits of nationalism: China and the diaoyu islands. International Security 23(3), 114–146. Dudden, A. (2014). Troubled Apologies among Japan, Korea, and the United States. Columbia University Press. 148 Volatile Nationalism | Xinru Ma Ebrey, P. B. (2009). Chinese civilization: A sourcebook.SimonandSchuster. Edmond, C. (2013). Information manipulation, coordination, and regime change. Review of Economic Studies 80(4), 1422–1458. Esmaquel II, P. (2012). China’s aggression “very good” for ph. The Rappler. https://www.rappler.com/nation/8872-china-s-aggression-very-good-for-ph. Fawthrop,T.(2018,6).Vietnammassprotestsexposehanoi?schinadilemma.TheDiplomat. https://thediplomat.com/2018/06/vietnam-mass-protests-expose-hanois-china-dilemma/. Fearon, J. D. (1995). Rationalist explanations for war. International Organization 49(3), 379–414. Fearon, J. D. (1997). Signaling foreign policy interests: Tying hands versus sinking costs. Journal of Conflict Resolution 41(1), 68–90. Fewsmith, J. and S. Rosen (2001). The domestic context of chinese foreign policy: Does ‘public opinion’ matter? The making of Chinese foreign and security policy in the era of reform, 1978-2000,151–187. Fey,M.andK.W.Ramsay(2011).Uncertaintyandincentivesincrisisbargaining: Game-free analysis of international conflict. American Journal of Political Science 55(1), 149–169. Finkel, S. E. and E. N. Muller (1998). Rational choice and the dynamics of collective politicalaction: Evaluatingalternativemodelswithpaneldata. AmericanPoliticalScience Review 92(1), 37–49. Fox, J. (2004). The rise of religious nationalism and conflict: Ethnic conflict and revolution- ary wars, 1945-2001. Journal of peace Research 41(6), 715–731. Francisco, R. A. (1993). Theories of protest and the revolutions of 1989. American Journal of Political Science 37(3), 663–680. Francisco, R. A. (1995). The relationship between coercion and protest: An empirical eval- uation in three coercive states. Journal of Conflict Resolution 39(2), 263–282. Francisco, R. A. (1996). Coercion and protest: An empirical test in two democratic states. American Journal of Political Science,1179–1204. Fravel, M. T. (2008). Strong borders, secure nation: cooperation and conflict in China’s territorial disputes. Princeton University Press. Friedberg, A. L. (1993). Ripe for rivalry: prospects for peace in a multipolar asia. Interna- tional security 18(3), 5–33. 149 Volatile Nationalism | Xinru Ma Friedman, E. (2008, 4). Friedman: Chinese believe tibetans, other eth- nic groups should be incorporated into one china. Council on Foreign Re- lations. https://www.cfr.org/interview/friedman-chinese-believe-tibetans-other-ethnic- groups-should-be-incorporated-one-china. Gagnon, V. P. (1994). Ethnic nationalism and international conflict: The case of serbia. International Security 19(3), 130–166. Geertz, C. (1963). The integrative revolution: Primordial sentiments and civil politics in the new states. Old societies and new states,105–157. Gellner, E. (1983). Nations and nationalism. Cornell University Press. Gelpi, C. (1997). Democratic diversions: Governmental structure and the externalization of domestic conflict. Journal of Conflict Resolution 41(2), 255–282. Gibler, D. M. (2012). The territorial peace: Borders, state development, and international conflict. Cambridge University Press. Glosny, M. A., P. C. Saunders, and R. S. Ross (2010). Debating china’s naval nationalism. International Security 35(2), 161–175. Granberg, D. and S. Holmberg (1990). The intention-behavior relationship among us and swedish voters. Social Psychology Quarterly,44–54. Gries, P. H. (2004). China’s new nationalism: Pride, politics, and diplomacy. University of California Press. Gries, P. H., Q. Zhang, H. M. Crowson, and H. Cai (2011). Patriotism, nationalism and china’s us policy: Structures and consequences of chinese national identity. The China Quarterly 205,1–17. Grønning, B. E. M. (2014). Japan?s shifting military priorities: Counterbalancing china?s rise. Asian Security 10(1), 1–21. Gurr, T. R. and W. H. Moore (1997). Ethnopolitical rebellion: A cross-sectional analysis of the1980swithriskassessmentsforthe1990s. American Journal of Political Science 41(4), 1079. Han, E. (2013). Contestation and adaptation: The politics of national identity in China. Oxford University Press. He, Y. (2009). The Search for Reconciliation: Sino-Japanese and German-Polish Relations since World War II. Cambridge University Press. Hechter, M. (2000). Containing nationalism. Oxford University Press. 150 Volatile Nationalism | Xinru Ma Hedstr¨ om, P., R. Swedberg, and G. Hernes (1998). Social mechanisms: An analytical ap- proach to social theory. Cambridge University Press. Helleiner, E. (2002). Economic nationalism as a challenge to economic liberalism? lessons from the 19th century. International Studies Quarterly 46(3), 307–329. Helleiner, E. and A. Pickel (2005). Economic nationalism in a globalizing world.Cornell University Press. Hensel, P. R. (1996). Charting a course to conflict: Territorial issues and interstate conflict, 1816-1992. Conflict Management and Peace Science 15(1), 43–73. Hensel, P. R. (2001). Contentious issues and world politics: The management of territorial claims in the americas, 1816–1992. International Studies Quarterly 45(1), 81–109. Hensel, P. R. and S. M. Mitchell (2005). Issue indivisibility and territorial claims. GeoJour- nal 64(4), 275–285. Heydarian, R. J. (2018). Duterte to china: ‘if you want, just make us a province’. The National Interest. https://nationalinterest.org/blog/the-buzz/duterte-china-‘if-you-want- just-make-us-province’-24599. Hogan, P. C. (2009). Understanding nationalism: On narrative, cognitive science, and iden- tity. The Ohio State University Press. Hollihan, T. (2014). The dispute over the Diaoyu/Senkaku Islands: How media narratives shape public opinion and challenge the global order.Springer. Hollyer, J. R., B. P. Rosendor↵, and J. R. Vreeland (2015). Transparency, protest, and autocratic instability. American Political Science Review 109(4), 764–784. Hughes,C.R.(2005). Nationalismandmultilateralisminchineseforeignpolicy: implications for southeast asia. The Pacific Review 18(1), 119–135. Hutchinson, J. (1994). Cultural nationalism and moral regeneration. Hutchinson and Smith, 123–24. Huth, P. K. and T. L. Allee (2002). The democratic peace and territorial conflict in the twentieth century, Volume 82. Cambridge University Press. Johnston, A.I.(2012). What(ifanything)doeseastasiatellusaboutinternationalrelations theory? Annual Review of Political Science 15,53–78. Johnston, A. I. (2013). How new and assertive is china’s new assertiveness? International Security 37(4), 7–48. Johnston, A. I. (2017). Is chinese nationalism rising? evidence from beijing. International Security 41(3), 7–43. 151 Volatile Nationalism | Xinru Ma King, G. and W. Lowe (2003). An automated information extraction tool for international conflict data with performance as good as human coders: A rare events evaluation design. International Organization 57(3), 617–642. King, G., J. Pan, and M. E. Roberts (2014). Reverse-engineering censorship in china: Ran- domized experimentation and participant observation. Science 345(6199), 1251722. Ko, J. (2018). Popular nationalism and international conflict. Working Paper. Kosterman, R. and S. Feshbach (1989). Toward a measure of patriotic and nationalistic attitudes. Political psychology,257–274. Kuang, C. (2018, 4). Calculating volatility: A simplified approach. Investrope- dia. https://www.investopedia.com/articles/basics/09/simplified-measuring-interpreting- volatility.asp. Kuran, T. (1997). Private truths, public lies: The social consequences of preference falsifi- cation. Harvard University Press. Leeds, B. A. and D. R. Davis (1997). Domestic political vulnerability and international disputes. Journal of Conflict Resolution 41(6), 814–834. Levento˘ glu, B. and A. Tarar (2008). Does private information lead to delay or war in crisis bargaining? International Studies Quarterly 52(3), 533–553. Lichbach,M.I.(1987). Deterrenceorescalation? thepuzzleofaggregatestudiesofrepression and dissent. Journal of Conflict Resolution 31(2), 266–297. Lichbach, M. I. and T. R. Gurr (1981). The conflict process: A formal model. Journal of Conflict Resolution 25(1), 3–29. Li↵, A. P. and G. J. Ikenberry (2014). Racing toward tragedy?: China’s rise, military competition in the asia pacific, and the security dilemma. International Security 39(2), 52–91. Lohmann, S. (1994). The dynamics of informational cascades: The monday demonstrations in leipzig, east germany, 1989–91. World politics 47(1), 42–101. Lorentzen, P. L. (2013). Regularizing rioting: Permitting public protest in an authoritarian regime. Quarterly Journal of Political Science 8(2), 127–158. Manski, C. F. (1990). The use of intentions data to predict behavior: A best-case analysis. Journal of the American Statistical Association 85(412), 934–940. Matfield, K. (2015). Bridging the gap between actual and reported behavior. UX Booth. McHugh, M. L. (2012). Interrater reliability: the kappa statistic. Biochemia medica: Bio- chemia medica 22(3), 276–282. 152 Volatile Nationalism | Xinru Ma Mirasol, J. D. B. (2016). Cooperation with china on the philippines’ war on drugs. Foreign Service Institute of Republic of the Philippines. http://www.fsi.gov.ph/cooperation-with- china-on-the-philippines-war-on-drugs/. Mochizuki, M. M. and S. P. Porter (2013). Japan under abe: toward moderation or nation- alism? The Washington Quarterly 36(4), 25–41. Moore, M. (2001). Normative justifications for liberal nationalism: justice, democracy and national identity. Nations and nationalism 7(1), 1–20. Muller, E. N. (1985). Income inequality, regime repressiveness, and political violence. Amer- ican sociological review,47–61. Nagler, J., J. Tucker, P. Barbera, M. Metzger, D. Penfold-Brown, and R. Bonneau (2016). Big data, social media, and protest: Foundations for a research agenda. In Computational Social Science. Cambridge University Press. Nguyen, X. (2017). Vietnam digitial landscape 2016 - 2017. Nish,I.(2001). Japaneseforeignpolicy, 1869-1942: KasumigasekitoMiyakezaka,Volume11. Psychology Press. Noelle-Neumann, E. (1974). The spiral of silence a theory of public opinion. Journal of communication 24(2), 43–51. O’donnell, G., P. C. Schmitter, and C. J. Arnson (2013). Transitions from authoritarian rule: Tentative conclusions about uncertain democracies. JHU Press. Pei, M. (2003). The paradoxes of american nationalism. Foreign Policy,31–37. Perlman, M. (2017). The key di↵erence between “nationalists” and “supremacists”. Columbia Journalism Review. Petersen, M. B. (2015). Evolutionary political psychology: On the origin and structure of heuristics and biases in politics. Political Psychology 36,45–78. Pollard, R. T. (1933). China’s Foreign Relations, 1917-1931.Macmillan. Powell, R. (2006). War as a commitment problem. International organization 60(1), 169– 203. Reilly, J. (2011). Strong society, smart state: The rise of public opinion in China’s Japan policy. Columbia University Press. Reuter, O. J. and G. B. Robertson (2014). Legislatures, cooptation, and social protest in contemporary authoritarian regimes. The Journal of Politics 77(1), 235–248. 153 Volatile Nationalism | Xinru Ma Richards, D., T. C. Morgan, R. K. Wilson, V. L. Schwebach, and G. D. Young (1993). Good times, bad times, and the diversionary use of force: A tale of some not-so-free agents. Journal of Conflict Resolution 37(3), 504–535. Ritter, E. H. and C. R. Conrad (2016). Preventing and responding to dissent: The ob- servational challenges of explaining strategic repression. American Political Science Re- view 110(1), 85–99. Robertson,G.B.(2007). Strikesandlabororganizationinhybridregimes. AmericanPolitical Science Review 101(4), 781–798. Ross, R. S. (2009). China’s naval nationalism: Sources, prospects, and the us response. International Security 34(2), 46–81. Ross, R. S. and Z. Feng (2008). The rise of china: Theoretical and policy perspectives. China’s Ascent: Power, Security, and the Future of International Politics,303. Salehyan, I., C. S. Hendrix, J. Hamner, C. Case, C. Linebarger, E. Stull, and J. Williams (2012). Social conflict in africa: A new database. International Interactions 38(4), 503– 511. Schrock-Jacobson, G. (2012). The violent consequences of the nation: Nationalism and the initiation of interstate war. Journal of Conflict Resolution 56(5), 825–852. Schrodt, P. A. (2012). Precedents, progress, and prospects in political event data. Interna- tional Interactions 38(4), 546–569. Schultz, K. A. (1998). Domestic opposition and signaling in international crises. American Political Science Review 92(4), 829–844. Self, B. (2002). China and japan: a fa¸ cade of friendship. Washington Quarterly 26(1), 77–88. Shadmehr, M. and D. Bernhardt (2011). Collective action with uncertain payo↵s: coordina- tion,publicsignals,andpunishmentdilemmas. American Political Science Review 105(4), 829–851. Shambaugh, D. (1999). Insecure china is stoking xenophobic nationalism. International Herald Tribune 15. Shaw, A. (1906). The american monthly review of reviews. Volum 33. Sheeran, P. (2002). Intention?behavior relations: a conceptual and empirical review. Euro- pean review of social psychology 12(1), 1–36. Siegel, D. A. (2011). When does repression work? collective action in social networks. The Journal of Politics 73(4), 993–1010. 154 Volatile Nationalism | Xinru Ma Sieger, K. (1990). Opinion research in east germany: A challenge to professional standards. International Journal of Public Opinion Research 2(4), 323–344. Slantchev, B. L. (2005). Military coercion in interstate crises. American Political Science Review 99(4), 533–547. Smith, A. (1998). International crises and domestic politics. American Political Science Review 92(3), 623–638. Smith, A. D. (1983). Theories of Nationalism. Holmes & Meier Pub. Smith, R. C. (1981). Black power and the transformation from protest to policies. Political Science Quarterly 96(3), 431–443. Snyder, J. (2000). From voting to violence: Democratization and nationalist conflict. W. W. Norton & Company. Snyder, J. and E. D. Borghard (2011). The cost of empty threats: A penny, not a pound. American Political Science Review 105(3), 437–456. Spence, J. D. (1990). The search for modern China. WW Norton & Company. Steinert-Threlkeld, Z. C. (2017). Twitter as data. Cambridge University Press. Suginohara, M. (2008). The politics of economic nationalism in japan: Backlash against inward foreign direct investment? Asian Survey 48(5), 839–859. Summa, J. (2018, 3). Option volatility. Investropedia. https://www.investopedia.com/university/optionvolatility/. Swain, C. M. (2002). The new white nationalism in America: Its challenge to integration. Cambridge University Press. Tarar, A. (2006). Diversionary incentives and the bargaining approach to war. International Studies Quarterly 50(1), 169–188. Tarar, A. and B. Levento˘ glu (2013). Limited audience costs in international crises. Journal of Conflict Resolution 57(6), 1065–1089. Teon, A. (2017). China and the may 4th movement. The Greater China Journal. Thompson, W. R. (2001). Identifying rivals and rivalries in world politics. International Studies Quarterly 45(4), 557–586. Toft, M. D. (2002). Indivisible territory, geographic concentration, and ethnic war. Security Studies 12(2), 82–119. Trager, R. F. and L. Vavreck (2011). The political costs of crisis bargaining. American Journal of Political Science 55(3), 526–545. 155 Volatile Nationalism | Xinru Ma UK, H. (2019). Religion and belief: some surveys and statistics. Humanists UK. Van de Mortel, T. F. et al. (2008). Faking it: social desirability response bias in self-report research. Australian Journal of Advanced Nursing 25(4), 40. Varble, D. (2008). The Suez Crisis.TheRosenPublishingGroup. Vasquez, J. A. (1993). The war puzzle. Number 27. Cambridge University Press. Vu, T. (2007). Vietnamese political studies and debates on vietnamese nationalism. Journal of Vietnamese Studies 2(2), 175–230. Vu, T. (2010). Studying the state through state formation. World politics 62(1), 148–175. Vu, T. (2014). The party v. the people: Anti-china nationalism in contemporary vietnam. Journal of Vietnamese Studies 9(4), 33–66. Wang, Z. (2008). National humiliation, history education, and the politics of historical memory: Patriotic education campaign in china. International Studies Quarterly 52(4), 783–806. Weatherley, R. and Q. Zhang (2017). History and Nationalist Legitimacy in Contemporary China: A Double-edged Sword.Springer. Weeks, J. L. (2012). Strongmen and straw men: Authoritarian regimes and the initiation of international conflict. American Political Science Review 106(2), 326–347. Weiss, J. C. (2013). Authoritarian signaling, mass audiences, and nationalist protest in china. International Organization 67(1), 1–35. Weiss,J.C.(2014). Powerfulpatriots: nationalistprotestinChina’sforeignrelations.Oxford University Press. Whiting, A. S. (1983). Assertive nationalism in chinese foreign policy. Asian Survey 23(8), 913–933. Williams, R. D. (2016, 7). Tribunal issues landmark ruling in south china sea arbitration. Lawfare. https://www.lawfareblog.com/tribunal-issues-landmark-ruling-south-china-sea- arbitration. Wu, J. (2007). Waijiao anli (case studies in diplomacy). Beijing: Zhongguo Renmin Daxue Chubanshe. Yoshino, K. (2005). Cultural nationalism in contemporary Japan: A sociological enquiry. Routledge. Zeitzo↵, T. (2011). Using social media to measure conflict dynamics: An application to the 2008–2009 gaza conflict. Journal of Conflict Resolution 55(6), 938–969. 156 Volatile Nationalism | Xinru Ma Zhang, K. (2015). “patriots” with di↵erent characteristics: deconstruct- ing the chinese anti-japan protests in 2012. MIT Newsletter, Spring 2015. https://cis.mit.edu/publications/newsletter/patriots-di↵erent-characteristics- deconstructing-chinese-anti?japan. Zhao, D. (2003). Nationalism and authoritarianism: student-government conflicts during the 1999 beijing student protests. Asian Perspective,5–34. 157
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The origins and evolution of the U.S. alliance network: how military allies transform and transact
PDF
Habits and policy: the socio-cognitive foundations of foreign policy stability
PDF
Perceiving and coping with threat: explaining East Asian perceptions toward China’s rise
PDF
Flinging the boomerang: locating instability and the threat potential of identity-bias in US national security policy
PDF
The changing state of tolerance and tolerance in changing states
PDF
Economic policy-making in a changing environment: the politics of success in Peru vis-à-vis China
PDF
Creating destruction: the political economy of zombie firms
PDF
The role of local actors in international development: how incentive types affect foreign aid effectiveness
PDF
Status seeking in hierarchy: Korea and Vietnam under Chinese hegemony in early modern Asia
PDF
The erosion of democracy in the post–Cold War world: understanding the phenomenon and its causes
PDF
Developing countries in the digital era: state and business interactions for industrial upgrading
PDF
Like father, like son? A succession-based explanation for conflict initiation by authoritarian regimes
PDF
Goldilocks’ signal for security cooperation in East Asia: China’s rise, hedging, and joint military exercises
PDF
The business of nationalism: how commodification sustains bilateral tensions
PDF
Power to the masses? The rise of direct democracy in Latin America
PDF
Identity salience and political violence: centering social identity in the study of intrastate conflict
PDF
Essays on sovereign debt
PDF
Financial crises and trade policy in developing countries
PDF
Political identity and political participation: China's post-80's generation
PDF
Securitizing language: political elites and language policy securitization in South Korea
Asset Metadata
Creator
Ma, Xinru
(author)
Core Title
Volatile nationalism: nationalism and its influence on maritime disputes
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Political Science and International Relations
Publication Date
07/25/2019
Defense Date
05/20/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
East Asian security,maritime disputes,nationalism,nationalist protests,OAI-PMH Harvest
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Kang, David (
committee chair
), Barberá, Pablo (
committee member
), Coe, Andrew (
committee member
), Haggard, Stephan (
committee member
), Munck, Gerardo (
committee member
)
Creator Email
xinrum@gmail.com,xinruma@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-191581
Unique identifier
UC11662797
Identifier
etd-MaXinru-7621.pdf (filename),usctheses-c89-191581 (legacy record id)
Legacy Identifier
etd-MaXinru-7621.pdf
Dmrecord
191581
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Ma, Xinru
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
East Asian security
maritime disputes
nationalism
nationalist protests