The Dictator's Modernity Dilemma: Theory and Evidence from South Korea

The Dictator's Modernity Dilemma: Theory and Evidence From South Korea

The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters

Citation Cho, Joan E. 2016. The Dictator's Modernity Dilemma: Theory and Evidence From South Korea. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.

Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:33493342

Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA The Dictator’s Modernity Dilemma: Theory and Evidence from South Korea

A dissertation presented

Joan Eun Cho

The Department of Government

in partial fulﬁllment of the requirements for the degree of Doctor of Philosophy in the subject of Political Science

Harvard University Cambridge, Massachusetts

The Dictator’s Modernity Dilemma: Theory and Evidence from South Korea

Abstract

Under what conditions are political institutions more or less effective at neutralizing opposition forces? To date, political scientists have generated conflicting evidence about the effects of economic development and political institutions on democratization. I argue that the co-opting effect of institutions depends on the strength of opposition groups, which in turn hinges on socioeconomic changes that develop nonlinearly over time. This dissertation identifies the threshold at which the cumulative changes of modernization translate into mass mobilization against the incumbent authoritarian regime, and demonstrates that the timing of democratization is conditional on generational turnover in key groups in civil society. The empirical chapters draw on a wide range of new qualitative and quantitative data from South Korea, a case for which existing scholarship generates contradictory findings that suggest political institutions and socioeconomic modernization explain both authoritarian resilience and democratization. I use archival materials to construct an original events dataset and employ subnational analysis to empirically demonstrate the importance of distinguishing the short- versus long-term effects of socioeconomic development on regime change. The first empirical chapter demonstrates the differences between the short- versus long- term effect of industrialization on citizens’ voting behavior and collective action against the iii authoritarian regimes of Park Chung Hee (1961-1979) and Chun Doo Hwan (1980-1988). The analysis estimates the controlled direct effect of industrial complexes on number of labor protests during the 1987 “Great Workers’ Struggle” in a given county and finds that a sustained presence of industrial complexes increased labor protests during the democratic transition period. The second empirical chapter examines whether and how access to mass media—a key feature of modernization—impacted regime support during the earlier phases of economic development under Park Chung Hee’s pre-Yushin regime (1961-1972). I find that greater access to mass media was correlated with more opposition to the authoritarian incumbent but only when the government’s control of the media was weaker. The third and final empirical chapter directly examines the interaction between co-opting effects of institutions and social changes resulting from modernization that empower civil society. I empirically test whether voting for the opposition party in a multiparty election had a positive or negative effect on participation in mass anti-regime protests during the 1987 “June Democratic Uprising.” I find that voting for the opposition party overall had a dampening effect on protest. However, the significant presence of social activists mitigated the dampening effect of election. These empirical chapters collectively demonstrate that while socioeconomic change and democratization do hang together over time, they do so in a non-liner fashion.

iv | Contents

Abstract iii

Acknowledgements x

Note on Romanization xii

1 Modernization and Generational Turnover Under Authoritarianism 1 1.1 Introduction ...... 1 1.2 Authoritarianism, Development, and Democracy in South Korea ...... 3 1.3 Toward a Theory of Modernization, Generational Change, and Democracy .6 1.4 The 386 Generation and Democracy in South Korea ...... 11 1.5 Overview of the Three Empirical Chapters ...... 21 1.6 Concluding Remarks and Future Research ...... 25

2 Long-term Effect of Industrialization on Labor Activism 28 2.1 Introduction ...... 28 2.2 Literature ...... 31 2.3 Background ...... 34 2.4 Data & Methods ...... 48 2.5 Results ...... 51 2.6 Conclusion ...... 57

3 Media Exposure and Regime Support 59 3.1 Introduction ...... 59 3.2 Existing Literature and the Argument ...... 61 3.3 Context ...... 64 3.4 Research Design ...... 71 3.5 Results ...... 78 3.6 Conclusion ...... 86

4 Authoritarian Elections and Anti-regime Protests 88 4.1 Introduction ...... 88 4.2 Theorizing Elections, Mobilizing Structures, and Protest ...... 90

v 4.3 Context ...... 95 4.4 Data & Methods ...... 99 4.5 Findings ...... 106 4.6 Conclusion ...... 111

A Appendix to Chapter 1: Protest Data and Coding Manual 113 A.1 Data ...... 113 A.2 Concept and Measure of Events ...... 114 A.3 Coding Variables ...... 115

B Appendix to Chapter 1: Additional Tables 123

C Appendix to Chapter 2: Additional Tables 126

D Appendix to Chapter 3: Additional Tables 130

E Appendix to Chapter 4: Additional Tables 133

F Appendix to Chapter 4: Additional Figures 137

Bibliography 139

vi | List of Figures

1.1 Number of Public Versus Private College-Level Institutions, 1962-1987 . . 16 1.2 Expansion of Tertiary Education and Student Enrollment,1962-1987 . . . . 17 1.3 Turnout for National Assembly and Presidential Elections, 1948-2012 . . . 20

2.1 Number of Labor Disputes and Labor Unions, 1963-1991 ...... 42 2.2 Number of Labor Protest in 1987 and Location of Industrial Complexes . . 43 2.3 ACDE of ICs as a Function of the Fixed Level of Manufacturing Firms in 1987...... 53 2.4 Coefﬁcient on Industrial Complex (Presence of IC) ...... 56 2.5 Quantile-Quantile (QQ) Plots of Each Covariate Used to Match ...... 56

3.1 Variation in Radio Signal Strength ...... 76 3.2 Radio Signal and Radio Sets in 1967 ...... 77 3.3 Newspaper Coverage of Presidential Elections in 1963 and 1967 ...... 79 3.4 Newspaper Circulation and Vote for Presidential Party, 1963 and 1967 . . . 79

4.1 Protest Sites During the June Democratic Uprising ...... 102 4.2 Relationship Between NKDP Vote Share and Protest Intensity at Different Degrees of Access to Mobilizing Structures with 95% Conﬁdence Intervals 109

F.1 Bivariate Correlation Between Student Population (1985) and Number of Student Associations (1983) ...... 137 F.2 Number of Mass Events With More Than 10,000 Participants ...... 138

vii | List of Tables

1.1 South Korea’s Political Generations ...... 12 1.2 Spread of Mass Media in Selective Years (Ratio of News Media to Total Households) ...... 15

2.1 Number and Size of Industrial Complexes by Administrative Region, 1963- 1987...... 36 2.2 Estimated Effects of Industrial Complexes on Labor Protest Fixing Number of Manufacturing Firms in 1987 ...... 52 2.3 Estimated Effects of Industrial Complexes on Labor Protest Fixing Popu- lation in 1985 ...... 54 2.4 Matching Results ...... 56

3.1 Newspaper Circulation and Vote for Ruling Party ...... 81 3.2 Newspaper Circulation and Vote for Ruling Party ...... 82 3.3 Determinants of Radio Signal Strength, 1967 and 1971 ...... 84 3.4 Radio Signal Strength and Vote for Ruling Party, 1967 and 1971 ...... 85

4.1 Hypotheses on the Relationship between Elections and Protest ...... 100 4.2 Number of Districts and Events (Per 100,000 People) for Each Province . . 102 4.3 Predictors of Protest Intensity during the 1987 June Democratic Struggle . . 108

B.1 Number of Political Protest by Organizing Groups, 1945-1972 ...... 124 B.2 Main Social Groups Participating in Protest ...... 125

C.1 Summary Statistics ...... 126 C.2 The Great Workers’ Struggle Protests and Electoral Support for the Oppo- sition Candidates in the First Democratic Presidential Election in 1987 . . . 127 C.3 First Stage Estimates from the Sequential g-estimation Model of Table 2.2 128 C.4 First Stage Estimates from the Sequential g-estimation Model of Table 2.3 129

D.1 Summary Statistics: Newspaper ...... 130 D.2 Summary Statistics: Radio ...... 130 D.3 Robustness Check I: Newspaper Circulation and Vote for Ruling Party . . . 131 D.4 Robustness Check II: Newspaper Circulation and Vote for Ruling Party . . 132

viii E.1 1985 NKDP Vote Share and Combined Vote Shares of the Opposition Can- didates (Kim Young Sam and Kim Dae Jung) in the First Democratic Pres- idential Elections in 1987 ...... 133 E.2 Results Using Different Measures of NKDP Vote Share ...... 134 E.3 Summary Statistics ...... 135 E.4 Robustness Checks: Predictors of Protest Intensity ...... 136

ix | Acknowledgements

I dedicate this dissertation to my parents—Hee Sang Cho and Kwang Ja Park—and grandparents who have inspired me to study the authoritarian period and democratization process in South Korea. Their experiences and stories not only informed and reminded me of my Korean roots, but instilled in me a curiosity to study the political history of my “motherland.” I am greatly indebted to my parents for the sacrifices they made to provide me with the best life experiences and education in the United Kingdom, South Korea, and the United States to freely explore and pursue my dreams. I am very fortunate to have worked with brilliant scholars during my undergraduate and graduate education. I express my sincere gratitude to my teachers and mentors from the University of Rochester (Bonnie Meguid, Mark Kayser, Richard Neimi, and Bingham Powell) and Harvard University (Daniel Ziblatt, Steve Levitsky, and Paul Chang), who have influenced and shaped me to become a scholar of political science. My dissertation advi- sor, Daniel Ziblatt, provided me with constant intellectual guidance and support since my first year of graduate school. My meetings with him has always been uplifting, enriching, and inspirational. Steve Levitsky has exemplified what it entails to be a comparativist by challenging me to develop generalizable theories and to situate Korea in a broader theoretical framework. Paul Chang, with his enthusiasm and deep knowledge of modern Korea, played a critical role in shaping me as a Korean scholar. I want to express my gratitude to many institutions and individuals who have helped

x and vii supported my dissertation research. I thank the Harvard University Asia Center, Asiatic Re- search Institute of Korea University, and Taiwan Foundation for Democracy for supporting my dissertation research. I thank Ms. Munjin Park and Dr. Ho-ryong Lee at the Korea Democracy Foundation and Ms. Mikyung Kang at the Harvard-Yenching Library for their valuable help in locating and accessing archival materials and primary sources. I also thank Professor Nae-Young Lee (Korea University) and Professor Young-kwan Yoon (Seoul Na- tional University) for hosting me at their respective institutes and helping me navigate my field research in South Korea. At Harvard, I was surrounded by incredibly supportive and knowledgeable colleagues. I received insightful comments and suggestions on numerous drafts of my papers from Cosette Creamer and Mai Hassan. Emily Clough supported me and help me go through my growing pains as a teacher-scholar. Patrick Lam, Anton Strezhnev, and Matthew Blackwell patiently taught and helped me with the statistical components of my research. I also thank Jeehye Kim, Kyle Jaros, Ruxandra Paul, Sparsha Saha, Kris-Stella Trump, and Hye Young You for their valuable feedback, encouragement, and support. Last but not least, I am truly grateful to my supporting family, friends, counselor (Ariel Phillips), pastors, and Highrock Church community to get through my ups and downs of graduate school and helping me discern and confirm my passion for research and teach- ing. I especially thank my brother, John Cho, who helped me take care of various family- related matters while I was in and out of the country for fieldwork. I also thank my dearest friends—Angelena Shim, Carol Yu, Elizabeth Kang, Eugenia Jeong, Eunice Lim, Irene Chung, and Patricia Kim—and Michael Kang for their continued encouragement, prayers, and moral support. Without all these people, I would not have been able to make it this far. And certainly I would not be able to continue my academic pursuits without an amazing support network of family, mentors, colleagues, and friends. Soli Deo gloria. xi | Note on Romanization

Korean names, words, and phrases have been Romanized using the McCune-Reischauer system. Exceptions are made for authors who have published in English using a different spelling, other well-known names, geographical places, and organizations with standard or ofﬁcial English spellings (such as Park Chung Hee, Syngman Rhee, Seoul, and Daewoo).

xii 1| Modernization and Generational Turnover Under Authoritarianism

1.1 Introduction

When are opposition groups no longer susceptible to co-optation and repression by authoritarian regimes? Under what conditions are political institutions more or less effective at neutralizing opposition forces? To date, political scientists have generated conﬂicting evidence about the effects of economic development and political institutions on democratization. Existing accounts on the impact of economic development differ over whether socioeconomic modernization increases or decreases the legitimacy of authoritarian regimes. Studies on political institutions disagree over whether elections, parties, and legislatures tend to stabilize or destabilize authoritarian regimes. My dissertation provides one solution to this debate by identifying the nonlinear relationship that exists between socioeconomic development and democracy. It argues that the effectiveness of political institutions in co-opting the opposition depends on the cumulative social changes from economic development that gradually strengthens the opposition vis-à-vis the regime. My dissertation advances the scholarship on comparative democratization by providing a micro-level analysis of modernization theory that incorporates the role of political institutions. It clariﬁes the existing debates by demonstrating that political institutions such as elections do not have a uniform effect on the probability of democratization in modernizing

1 countries across all contexts and at all times. Because societal changes resulting from modernization are often gradual and cumulative, an authoritarian regime can use sociopolitical institutions in the short run to sufﬁciently co-opt or repress opposition forces. In the long run, however, these same political institutions can have the opposite effect: economic development empowers opposition forces with resources and capabilities to mobilize against a regime, transforming institutions that previously bolstered an authoritarian regime into weapons for the opposition. In other words, political institutions have a non-uniform ef-

fect over time on authoritarian survival. At T1, institutions such as elections or legislatures may aid the survival of an authoritarian regime (as the literature on authoritarian durability suggests). At T2, these same institutions no longer have this effect and may even contribute to the regime’s destabilization. By examining the co-optive effect of institutions in con- junction with modernization processes, my dissertation contributes to our understanding of how and when modernization theory unfolds on the ground at the micro-level. The dissertation also makes a critical contribution to our understanding of democratization by finding that the gradual effect of modernization operates through a mechanism of generational turnover in key groups in civil society. The full and transformative effects of socioeconomic change on civil society thus appear long after narrow thresholds of socioeconomic development have been passed, contingent on the amount of time it takes for a new generation to become the numerically dominant group. Development-induced social changes that strengthen democratic opposition do not appear immediately, are cumulative, and have a modest or negligible impact until they reach a threshold. An important conclusion of my work follows from this: while socioeconomic change and democratization do hang together over time, they often do so in a non-linear fashion. I draw on a wide range of new qualitative and quantitative data from South Korea, a case for which existing scholarship counter-intuitively finds that political institutions and socioeconomic modernization explain both authoritarian durability and democratization. I 2 examine spatial and temporal variations in regime support and anti-regime protest under the authoritarian tenures of Park Chung Hee (1961-1979) and Chun Doo Hwan (1980-1988). In addition, I conduct subnational analysis, for which I created an original events dataset of nearly 2,500 events during two nationwide protests that were seminal in the history of South Korea’s democracy movement—the 1987 “June Democratic Struggle” and the 1987 “Great Workers’ Struggle.” I drew from archival materials obtained during my year of field research in South Korea and then collected fine-grained data on elections, industrialization, and various social-economic characteristics of each county for statistical and Geographic Information System (GIS) analyses.

1.2 Authoritarianism, Development, and Democracy in South

Korea

South Korea—a celebrated case of democratization during the “third wave of democracy” (Huntington 1991)—developed economically under authoritarian rule, became wealthy, and democratized. Korea’s economy began to rapidly grow starting in the 1960s under Park Chung Hee’s authoritarian rule, and maintained an annual growth rate of eight percent through the 1980s under his military authoritarian successor, Chun Doo Hwan. In creating what has become the “Miracle on the Han River,” these autocrats suppressed all forms of dissidence. Yet in this “dark age of democracy” (Lee 2010b, 2007; Yi 2011), a sustained democracy movement emerged (Chang 2015b). In the late 1980s, “economic development had proceeded to the point where pressures for expanded political participation compelled [the South Korean government] to begin process of democratization” (Huntington 1991, 72). Mass demonstrations by civil society groups during the 1987 June Democratic Up- rising (6-wolˇ Minju Hangjaeng) forced the ruling government to hold direct presidential

3 elections and institute other democratic reforms. As a result, South Korea transitioned into an electoral democracy in 1987. Since 1987, South Korea has stabilized into a liberal democracy. Political contestation has become increasingly fair—passing Samuel Huntington’s true test for democratic consolidation by having twice made peaceful transfer of power from the ruling to the opposition party (Huntington 1991, 266-67). Civil liberties have also substantially expanded: press censorship and restrictive labor laws from the authoritarian period were abolished and overhauled. The Freedom House1 has continually assigned a score of no worse than two for political rights and civil liberties, with one representing the most free and seven the least free. Further, with improved civilian control of the military under Kim Young Sam administration (1993-1998), South Korea has successfully consolidated as a liberal democracy. At ﬁrst blush, South Korea illustrates the basic premise of modernization theory: economic development leads to democracy (Deutsch 1961; Inkeles 1966; Lipset 1959). In- deed, a large literature on South Korea’s economic and political development depicts a relatively “smooth and peaceful capitalist transition toward modernity” (Koo 1999, 55). However, under the authoritarian tenures of Park Chung Hee (1961-1979) and Chun Doo Hwan (1980-1988), the political system became increasingly authoritarian with the growth of national economy. Democracy in South Korea was not a natural outcome of economic growth and expansion of the middle class. Rather, South Korea’s path to modernization and industrialization was a “dialectic process of social change,” characterized by “continuous opposition and conﬂicts between a strong autonomous state and an unruly society” (Koo 1999, 56). My dissertation seeks to reconcile these two seemingly opposing views regarding South

1Freedom House, http://www.freedomhouse.org

4 Korea’s path to modernity and democracy by arguing and providing evidence to show that the relationship between economic development and democracy is non-linear: modernization does have a stabilizing effect on authoritarian rule, but it also provides the social foundation for a successful democracy movement to emerge later on. In explaining South Korea’s democratization in particular, I argue that the timing of democratic transition is better explained by generational effects instead of a class-based analytical framework that emphasizes the role of the middle class. As this dissertation will show, the “real agents” of the political breakthrough in 1987 were students and workers belonging to the “386 generation.” Finally, while South Korea’s process of democratization has been examined temporally or even cross-nationally (usually in comparison to Taiwan and other Third Wave democracies), the relationship between economic development and democracy, to my knowledge, has not been empirically examined at the subnational level. This is quite surprising given the accounts on the uneven regional development under Park Chung Hee that arguably contributed to anti-regime activism in the southwest region of the country. Voting patterns along regional lines (southwest Honam vs. southeast Yongnam)ˇ continue to reﬂect intensive regionalism in contemporary South Korean politics.2 My dissertation explores temporal as well as subnational variations in voting behavior and protest patterns during the authoritarian period to systematically account for the differences within and across different regions in South Korea.

2The southwest, a regional known as Honam, generally supports left-leaning candidates. The southeast, a region known as Yongnam,ˇ is generally more supportive of right-leaning candidates. In 2012 presidential elections, the losing (left-leaning) candidate of the center-left Democratic United Party, Moon Jae-in, gathered over 80 percent of the votes in the southwest and won more than 90 percent in the important south- western city Kwangju. In contrast, the winning candidate of the conservative Saenuri Party, Park Geun-hye, also the daughter of Park Chung Hee, garnered more than 80 percent of the vote in North Kyongsangˇ and Taegu city.

5 1.3 Toward a Theory of Modernization, Generational Change,

and Democracy

The current state of the democratization literature is unclear as to whether democratization is a short-term or a long-term outcome. Scholars of democratization conceptually recognize the gradual process of democratization that occurs over the longue durée, yet their empirical analyses have used measures that are either strictly dichotomous (“democracy” or “authoritarianism”) or that conflate the different dimensions (or episodes) of democratization (Ziblatt 2006). Indeed, many large-N studies have essentially examined the “snapshot” of the democratization process by measuring democracy or authoritarianism based on whether a Polity score or Freedom House Index of a given country at a given year has crossed the “threshold value of democracy.” Although research on democratization has become increasingly sophisticated due to new datasets and advanced statistical methods, the deep mismatch between concept and measures of democratization continues to mark the field. In investigating the causes of democratization, the relationship between economic development and democracy has received a lot of attention from political scientists.3 Building on Lipset (1959)’s seminal piece Some Social Requisites of Democracy: Economic Devel- opment and Political Development, earlier studies have examined the relationship between modernization and democracy using regression analysis. In doing so, these studies have essentially assumed a linear relationship between the two variables and modernization has often been measured narrowly as simply “level of socioeconomic wealth” (i.e., GDP per capita). Many scholars do find empirical support for modernization theory (e.g., Jackman 1973; Bollen 1979; Burkhart and Lewis-Beck 1994; Londregan and Poole 1996; Barro

3See Geddes (2009) for an overview of the literature on democratization.

6 1990; Boix and Stokes 2003; Epstein et al. 2006) while others do not find a causal effect of economic development on democracy (e.g., Przeworski and Limongi 1997; Przeworski et al. 2000; Acemoglu, Robinson, and Yared 2008, 2009). While scholars and observers have paid significant attention to the causal relationship between economic development and democracy, they are ambiguous about the timing of democratization resulting from economic development. Despite the common use of regression analysis, studies on this topic imply a nonlinear relationship between the two variables, involving a threshold effect. For instance, Przeworski and Limongi (1997) argue that once countries reach a certain level of economic development, there is a significant probability that authoritarian governments turn into democracies. Over time, there will be fewer and fewer authoritarian governments above that economic threshold level. Similarly, according to Inglehart and Welzel (2005, 2009)’s revised theory of modernization, the emergence of postindustrial society brings social and cultural changes that are conducive to democratization and “beyond a certain point [emphasis added] ... it becomes difficult to avoid democratization” (Inglehart and Welzel 2009, 38). Many important social and political processes, including modernization and democratization, take a long time to unfold. In explaining these long-term causes and outcomes, it is critical to incorporate the temporal structure of the causes and outcomes into theoretical accounts (Pierson 2004). Societal changes resulting from modernization that are conducive to democratization are often gradual, slow-moving, and cumulative. Because the cumulative social and cultural changes that result from socioeconomic development do not appear immediately and have a modest or negligible impact until they reach a threshold, socioeconomic change and democratization do hang together over time, but they often do so in a non-linear fashion. To explain the timing of democratization, I argue and demonstrate in this dissertation that the gradual effect of modernization on democratization operates through a mechanism 7 of generational turnover (or replacement) in civil society. The full and transformative effects of socioeconomic change on civil society thus appear long after narrow thresholds of socioeconomic development have been passed, given the amount of time it takes for a new generation to become numerically dominant in the adult population to participate in politics. Social changes resulting from autocrat-led modernization such as high levels of industrialization and education give rise to a new “democratizing generation” that gradually replaces the previously co-opted generation of the authoritarian period. Scholars have considered generational effects to explain various social and political outcomes including political participation (e.g., Miller 1992; Miller and Shanks 1996), civic engagement (e.g., Putnam 1995), social movements (e.g., Jennings 1987) and democratization (e.g., Haddad 2012). Generational effects, as described in Karl Mannheim’s classic essay on “The Problem of Generation,” represent the fact that “[i]ndividuals who belong to the same generation, who share the same year of birth, are endowed, to that extent, with a common location in the historical dimension of the social process” (Mannheim 1952, 290). Generational effects differ from life-cycle effects (i.e., the differences attributable to stage of life) and period effects (that affect all people who live through a given era, regardless of their age). There are two major explanations to account for generational effects. The first places great emphasis on the process of political socialization provided by social institutions such as the family, schools, peer groups, and the mass media. The second claims that a cohort is strongly influenced by major historical or political events (usually in their early adulthood), establishing a set of attitudes and world views that guide and shape their political behavior through the rest of their lives (Mannheim 1952). Indeed, “[i]ndividuals coming of age during periods of pronounced stress and drama, epochal events, or rapid socioeconomic change [emphasis added] are often said to be uniquely identified in a political sense” (Jen- nings 1987, 368). For instance, Jennings (1987) finds that the Vietnam War protests and 8 the Civil Rights Movement in the 1960s fostered similar attitudes among a ‘protest generation’ in the U.S. Similarly, Putnam (1995) shows that the cohort born in 1925-1930 in the U.S—who attended grade school during the Great Depression and spent World War II in high school—has been exceptionally civic: voting more, joining more, reading newspaper more, and trusting more. To be clear, political generation has dynamics that are not biological. As Pyle (2008, 3) notes,

[a] political generation, as opposed to a biological generation, is determined by major events that shape a distinctive outlook or an approach to issues among young people living through these events at a formative age. Political generations may not be of common mind. Formative experiences need not lead to a consensus of views. There may be a diversity of viewpoints. There may be debates within a generation. But there is common concern, a shared sense of problem consciousness, that is shaped by the formative experience that creates a generation.

According to Haddad (2012)’s tipping point model of political change, a “turning point” in the political history creates a new generation and a “tipping point” is reached when that “new” generation becomes the majority of the country’s politically active population and elite. There are two ways in which the new generation can bring about sociopolitical change (Ryu 2013). First is through social turnover. The members of one salient generation are replaced by the members of another salient and younger generation. The second this through political turnover. When certain members of a new generation occupy key positions of power, they can affect policy changes. Hence, the full effects of generational developments generally appear several decades after their onset, because it given the amount of time it takes for a generation to become numerically dominant in the adult population. Building on these studies, I argue that economic development facilitates democratization through the social mechanisms described in modernization theory, but there are differing effects on democratization in the short run versus the long run. This is due to the 9 fact that different generations experience (or are exposed to) different stages and aspects of economic development. Further, the effects of rapid economic growth on society—whether positive or negative—are manifested at different moments in time. To examine the full effects of economic development on the probability of democratization, we need to examine how the process of economic development maps onto the differences in the ways in which a given generation (1) perceives authoritarian rule to be legitimate and (2) has the resources to challenge an authoritarian regime. Economic development promotes vast changes in values among younger generations with implications for political development. Improvement in societal well-being tends to promote secular-rational values and self-expression values among the younger generations (Inglehart and Welzel 2005). “Industrialization–especially rapid catch-up industrialization– brings with it social dislocations caused by the movement of people from the country to the city, by the psychological strain caused by the undermining of old values and the distur- bance of vested interests by economic change, and by widening differences between generations” (Pyle 2008, 6). Hence, experiencing rapid industrialization and economic security at a formative age (rather than earlier or than later) is more likely to impart self-expressive values that give high priority to free choice and motivate political action (Inglehart and Welzel 2009, 38). Economic development also empowers opposition forces with resources and capabilities to mobilize against an authoritarian regime. Resource mobilization theory of social movements (e.g., McCarthy and Zald 1977; Tilly 1978; McAdam 1982) argues that grievances are not enough for collective action. Access to and control over resources is a crucial factor for the emergence and survival of social movements. Rueschemeyer, Rueschemeyer, and Stephens (1992) demonstrate the importance of (capitalist) development to the empowerment of subordinate classes (i.e., the working and middle classes). More recent studies also show that media and advancement in communication technolo- 10 gies facilitated political mobilization in repressed societies (Adena et al. 2013; Enikolopov, Petrova, and Zhuravskaya 2011; Hassanpour 2011; Meier 2011; Kern and Hainmueller 2009). In the following section, I argue and provide evidence to show that economic development that began under Park Chung Hee’s authoritarian rule in South Korea contributed to the creation and development of a new political generation—the so-called 386 generation— that played a signiﬁcant role during the democratization movement in the 1980s.

1.4 The 386 Generation and Democracy in South Korea

In the case of South Korea’s democratization, those who were at the forefront of pro- democracy protests in the late 1980s belong to the the “386 generation” (sampallyuk sedae). The term was coined in the early 1990s, in reference to what was then the latest computer model–Intel’s 386, and referring to people in their 30s (hence “3”), having attended university in the 1980s4 (hence “8”), and born in the 1960s (hence “6”). Arguably, the 386 generation is “the most vocal, active, and self-conscious generation in contemporary Asia” (Pyle 2008, 10). South Korean society is currently comprised of generations exposed and shaped by different political contexts. More importantly, all of these generations experienced South Korea’s economic progress and democratization at different stages in life, resulting in re- markably different views of and approaches to the opportunities and challenges faced by South Korea (Flake 2008). Table 1.1 lists South Korea’s political generations as of 2013. The youngest generation, the candlelight generation, were born after the democratization in 1987 and experienced the 2008 candlelight protest against U.S. beef5 in their late

4Korean university classes (hakbon) are determined by entrance year rather than by graduation year. 5The protest began after the South Korean government reversed a ban on U.S beef imports in 2008. The

11 Table 1.1: South Korea’s Political Generations

Birth Years Political Generations 1988 – 1993 Candlelight Generation 1979 – 1987 World Cup Generation 1970 – 1978 IMF Generation 1960 – 1969 386 Generation 1952 – 1959 Yushin Generation 1942 – 1951 Post-War Industrialization Generation – 1942 Korean War Generation Source: Noh, Song, and Kang (2013, 125) teens and early twenties. They become eligible to vote at a time of rapid changes in the political environment such as activation of online fan clubs for politicians and the spread of unconventional forms of political participation. The members of the World Cup generation were born during Chun Doo Hwan’s authoritarian rule. They experienced the 2002 FIFA World Cup6 in the late teens and/or early- to mid-twenties. They were able to ﬁrst exercise their voting rights during Kim Dae Jung (1998-2003) and Roh Moo-hyun (2003-2008) ad- ministrations. Both Kim and Roh were active in South Korea’s democracy movement and Kim’s administration was the ﬁrst non-conservative administration since democratic transition. The IMF generation, born in the 1970s, experienced the 1997 Asian Financial Crisis (commonly referred to as IMF in South Korea). The above-mentioned 386 generation was born in the 1960s, the time industrialization began and accelerated. The Yushin generation, the baby boomers of South Korea, include those who were in their twenties during Park’s Yushin regime (1972-1979). The post-war industrialization generation experienced political turmoils such as the 1960 April Revolution and the 1961 May 16 Coup. Lastly, those

ban had been placed since December 2003, when ‘mad cow disease’ was detected in U.S beef cattle. At its height, the protest involved tens of thousands of people. 6The 2002 FIFA World Cup was the ﬁrst Asian World Cup jointly hosted by Japan and South Korea. During the World Cup, millions of Koreans wearing red T-shirts cheered collectively on the streets for the South Korean national soccer team.

12 who were born prior to 1942, the Korean War generation, experienced Japanese colonial rule (1910-1945), national independence, and the Korean War (1950-1953). The 386 generation serves as the dividing line between the “authoritarian generations” and “post-transition generations.” The post-transition generations, including members of the IMF generation, World Cup generation, and candlelight generation either did not experience authoritarian rule and pro-democracy protests in the 1980s or they were too young to be affected by them. Among the authoritarian generations, the 386 generation is the youngest and the first generation to grow up free from poverty. Some 386ers (born in the early 1960s) experienced poverty in their childhood, but they did not know the real destitu- tion that their parents experienced during parts of Japanese colonial rule (1910-1945) and the Korean War (1950-1953). Per capita income was less than $100—about what it is today in the poorest south Asian and African countries. The Korean War devastated the country, bringing massive damage to its economy and infrastructure. As a result, the generation of the 1930s share the profound sense of economic and national insecurity from the Korean War. In contrast, to the 386ers, the war is only a distant memory recalled largely through their parents’ stories and official anti-communist propaganda disseminated by the autocratic military regime. Their parents, born and raised in poverty, were willing to accept a dictatorship as long as it delivered economic growth. “The older generation was proud to move from bicycles to motorbikes, and then to cars, and they were happy when it became affordable to vacation in Jeju Island and then even sometimes overseas” (Lankov 2008). The 386 generation took all these for granted. “They were not ready to accept the tacit deal military dictators made with [the previous generations]: authoritarianism in exchange for economic growth. The 386ers wanted complete democracy” (Ibid). In fact, the 386 generation is the largest generation involved in social movements in South Korea (Min and Kim 2008). Although there had been active student and pro- 13 democracy movements throughout the 1960s and the 1970s, student protest became more radicalized and politicized in the 1980s (Lee 2007; Shin et al. 2007). According to Shin (1987), student movement has a long history dating back to the independence movement during Japanese colonial rule, but it was not until Park Chung Hee’s authoritarian rule that students became the driving force of political protest. Students indeed were at the forefront of the 1960 April 19 Revolution that ended Syngman Rhee’s autocratic rule in 1960. How- ever, the actual number (and percentage) of political protests organized by students was not very high. The percentage of protests by students increases significantly from 21.1 percent during the military government rule under Park Chung Hee (1961-1963) to 76.2 percent under Park’s Third Republic (1963-1972) (see Appendix Table B.17). In the subsequent Fourth and Fifth Republics under Park Chung Hee and Chun Doo Hwan, students became the sole organizing force of political protest (Shin 1987, 2). Data from the Stanford Korea Democracy Project (Shin et al. 2007), covering the Fourth and Fifth Republics, confirm students’ dominant role in social movements in the 1970s and 1980s (see Appendix Table B.2).8 The rapid industrialization that began under the rule of Park Chung Hee created more opportunities for socialization through urbanization and the expansion of education and the mass media. Between 1960 and 1990, the urban population of South Korea grew from 28.3 percent to 76.9 percent of the total population. The mass media (radios, newspapers, and televisions) also developed rapidly, reaching almost every single household by 1979 (see

7South Korea’s civilian governments are numbered from the First Republic of Syngman Rhee (1948– 1960) to the contemporary Sixth Republic (1987–present). The First Republic, arguably democratic at its inception, became increasingly autocratic until its collapse in 1960. The Second Republic was established as a parliamentary democracy, but was overthrown in less than a year and replaced by an autocratic military regime. The Third, Fourth, and Fifth Republics were a continuation of military rule. 8While Shin (1982) reports the number and percentage of protests by organizing groups, Shin et al. (2007) reports percentage of protests by groups participating in protests.

14 Table 1.2).9

Table 1.2: Spread of Mass Media in Selective Years (Ratio of News Media to Total House- holds)

1963 1974 1979 Total number of households (1,000s) 4378 6761 7256 Radios 3.4:1 1:1 1:1 Newspaper circulation 5.5:1 1.9:1 1.2:1 Television sets 138:1 5.3:1 1.1:1 Sources: The EPB, Korea Statistical Yearbook, 1964; Hapdong yongam 1975; Korea University, The Journal of Communication Sci- ence, 2 (June, 1980): 40-55.

Along with urbanization and spread of mass media, education was also expanded at all levels. The intended role of education was to foster “the basic attitude of compliance with a strong central government ... [and] education did play a critical role in the modernization of Korea [up until the 1970s]—primarily by assisting a strong government with ‘modernizing’ policies to impose its will upon the nation” (Mason et al. 1980, 378). However, under Park’s state-led industrialization, numbers of private schools as well as students enrolled in private institutions continued to increase (see Figure 1.110). This temporal trend (from 1962 to 1973)11 suggests that as the country was undergoing economic development the population had more access to private education, providing more opportunities for independent/critical thinking and socialization among college students. More importantly, between 1970s and 1980s, college enrollment increased more than threefold. According to the ofﬁcial statistics of the Ministry of Education, 2,933,683 students entered colleges in the 1980s. Approximately 1,870,000 were enrolled in four-year

9Original table is found in Choi (1989, 71). 10Teachers’ colleges were all national colleges. 11From 1962 to 1973 the Korea Statistical Yearbook publications contain information on the breakdown of state-provided (national/public) versus privately established college-level institutions. Such information is unavailable in the Yearbooks published in the 1980s.

15 colleges and the rest were enrolled in vocational or two-year junior colleges. In comparison, in the 1970s, student enrollment in college was only 891,328 for all types of colleges— four-year colleges and the rest.

Figure 1.1: Number of Public Versus Private College-Level Institutions, 1962-1987

Source: Korea Statistical Yearbook [Han’guk tong’gye yongam], 1963-1988

While the 386ers did have easier access to college-level education, rising enrollment rates in the 1980s is not merely a reﬂection of the (increased) number of schools. Figure 1.212 shows that while the number of college-level institutions has been increasing steadily over time, student enrollment has increased more dramatically, especially in the 1980s. The difference in the slopes (i.e., rates of change) between number of schools and student enrollment is even greater when we look at colleges and universities (shaded in black). Student enrollment in these four-year colleges and universities increased by 12,638 students per year from 1961 to 1979 while the rate at which student enrollment increased was 82,395 students per year just from 1980 to 1987. College students played a signiﬁcant role during the democratization movement in the

12In 1978, junior colleges and junior technical colleges became junior vocational colleges. School types found in the Korea Statistical Yearbook publications are college/university, junior college, junior technical college, junior vocational college, and teachers’ college.

16 Figure 1.2: Expansion of Tertiary Education and Student Enrollment,1962-1987

Source: Korea Statistical Yearbook [Han’guk tong’gye yongam], 1963-1988

1980s. Students led the 1980 Kwangju Uprising (or Kwangju massacre)13 that “radicalized” the democracy movement (Lee 2007, 26). When Chun Doo Hwan relaxed his repressive policies in 1984,14 students at Seoul National, Yonsei, and Korea Universities organized the Committee for Autonomy on Campus (Hagwonˇ Chayulhwa Ch’ujin Wiwonhoeˇ ) to revive student associations and achieve democratization of college campuses. In 1985, more than 10,000 students participated in nationwide demonstrations demanding the government’s apology for the Kwangju massacre. In 1987, students were at the forefront of the famous June Democratic Uprising that forced the ruling government to hold direct presidential elections and institute other democratic reforms which marked the beginning of South Korea’s democratic transition. Student associations such as Sodaehyˇ opˇ (the As- sociation of Student Representatives in Seoul), Puchongyˇ opˇ (the Association of Student

13From May 18 to 27, 1980, students of Chonnamˇ University and citizens of Kwangju city staged violent protests against Chun’s military regime. According to the May 18 Memorial Foundation, 154 were killed, 74 were missing, and 4,141 were wounded (including those who died from their wounds) and placed under arrest by the martial law force. 14Chun’s decision to liberalize was driven primary by two factors exogenous to opposition demands. First, Chun was (over)conﬁdent in the stability and popularity of his regime largely due to the country’s strong economic performance. Second, two international athletic games were to be hosted in South Korea—the Asian Games in 1986 and the Olympics in 1988 (Kim 2000, 82).

17 Representatives in Pusan), and Daedaehyopˇ (the Association of Student Representatives of Taegu) were the main organizational force behind the national rallies held during the June Democratic Uprising. Furthermore, during the 1980s when labor was severely repressed by the Chun regime,15 college students dropped out of college and entered factories as “disguised workers” to promote class consciousness and aid union organization. Unlike in the 1970s when students were individually involved in the labor movement, student in the 1980s were organized and they participated in the movement on a large-scale basis (Oh 2010). Their ultimate goal was to lead the labor struggles toward the larger political goals of ending military rule. Inside factory compounds, students formed small groups with workers to instill class consciousness among workers through group studies, discussions, and recreational activities such as hiking and traditional mask dancing. The estimates of students-turned-workers (hakchul) range from at least 1,000 (Hwang 1985, 14) to 3,000 by the mid-1980s (Ogle 1990, 99). As a result, with the help from students (in the 1980s) and Christian groups (in the 1970s), laborers were able to lead the Great Workers’ Struggle (Nodongja Daet’ujaeng)16 from July to August in 1987 that provided the ﬁnal impetus for full democratization in South Korea. Starting in the 1980s, the 386 generation was also ﬁnally old enough to participate in institutional politics. The oldest among the 386 generation (i.e., those born in 1961) turned 20 in 1980. At the age of 20, they entered college as freshmen in 1980 and were eligible to vote.17 The 386ers were not old enough to vote in any of the legislative elections

15According to Hart-Landsberg (1993, 219), labor politics under Chun’s rule were “the most restrictive and oppressive in South Korean history”. 16About 1.2 million workers (about a third of the regularly employed workforce) from most major industries (mining, manufacturing, shipbuilding, transportation and service sectors) staged industrial protests and strikes that focused on economic issues (such as labor rights, wage issues, improvement of labor conditions) as well as democratization at workplace (liquidation of the existing state-corporatist unions and establishment of democratic unions). 17From 1950 until 2005, voting age in South Korea was 20.

18 held in the 1970s under Park Chung Hee’s rule. The first election in which the 386ers (but only those born in 1961 and 1962) were eligible to vote was the 11th National As- sembly elections in 1981, but no genuine opposition party was able to contest since they were banned from engaging in politics. In the following 12th National Assembly elections in 1985, majority of the 386ers were able to vote (i.e., those who were born in 1961 to 1966). Compared to the legislative elections in 1981, the 1985 legislative elections experienced an unprecedented voter turnout rate of 85 percent. During the 1985 elections, a genuine opposition party—the New Korea Democratic Party (NKDP; Shinhan Minju- dang)—contested in elections and emerged as the leading opposition—winning 29 percent of the votes compared to Chun’s ruling Democratic Justice Party (DJP; Minjongdangˇ ). Im- portant to the electoral success of the NKDP was the support of student groups such as the Youth Coalition for Democracy Movement (Minch’ongnyˇ onˇ ) publicly announced that they would back the NKDP and actively campaigned for the NKDP as volunteers. Notably, this active participation by student groups represented the first time since the early 1960s that university students publicly supported a particular political party (Kim 2000, 85). The alliance formed between social movement groups and the opposition NKDP during the 1985 National Assembly elections outlived the elections and subsequently developed into a grand pro-democracy coalition against the authoritarian regime (Kim 2000; Chung 2011). Finally, following the June Democratic Uprising in 1987, first direct presidential elections were held since 1971 and almost all of the 368ers (i.e., those born in 1961 to 1968) were able to vote. The turnout in this historical elections was 89.2 percent, the highest turnout since the beginning of authoritarian rule in 1961 (see Figure 1.3). The liberal ideals held among the 386 generation—formed under authoritarianism and exhibited during the transition period—continue to persist in the post-transition (democratic) period. The members of the 386 generation are in their prime and now comprise much of the elite of South Korean society. Many participants in the democracy movement 19 Figure 1.3: Turnout for National Assembly and Presidential Elections, 1948-2012

Source: National Elections Commission of South Korea in the 1980s were elected into the National Assembly and this entry of 386ers was inten- siﬁed with the presidencies of Kim Dae Jung and Roh Moo Hyun, both former leaders in the democracy movement (Kim and Chang 2011). Over 30 percent of both the 16th and 17th National Assemblies consisted of 386ers who were once members of democracy movement organizations including the famous National Council of Student Associa- tions (Chondaehyˇ opˇ ). The 386ers are far more left leaning than that of their parents and their children. Studies show that the 386 generation—as both electorates and politicians— continues to maintain its progressive stance and political ideologies from the democracy movement in the 1980s such as anti-Americanism and minjung democracy18 (Noh, Song, and Kang 2013; Ryu 2013; Kim and Chang 2011; Min and Kim 2008; Oh and Yi 2005).

18The “minjung” is a term that is translated as the masses or the “common people.” This minjung sentiment was a “hallmark of South Korea’s democracy movement” in the 1980s that was “manifest not only in political rhetoric but also in music, art, literature, philosophy, and theology” (Chang 2015a)

20 1.5 Overview of the Three Empirical Chapters

The three empirical chapters of this dissertation examines how different aspects of modernization affected regime support for Park Chung Hee and Chun Doo Hwan by examining voting behavior and patterns of anti-regime protest. Chapter 2 shows the differing effects of industrialization in the short versus long run. Chapter 3 explores the short-term effect of mass media on regime support. Chapter 4 demonstrates the long-term effect of the expansion of college education. Chapter 2 distinguishes short versus long-term effects of industrialization policy on regimes support and labor activism in authoritarian South Korea. At first blush, South Korea seems to have taken a “linear path” from industrialization to democracy. How- ever, at least in the short term, industrialization had a stabilizing effect on the authoritarian regime. South Korea’s successful export-led industrialization relied on the construction of massive industrial complexes. The authoritarian regimes achieved economic growth and increased popular support for the ruling party through the development of industrial complexes, which provided material benefits and helped modernize the country. Concurrently, there was a gradual growth of workers’ consciousness and capacity to organize and engage in collective action against the state and capital holders. This cumulated in a nationwide labor strike that provided the final impetus for democratization in the late 1980s. To empirically investigate the effect of government’s investment in industrial complexes on labor activism—mediated by the socioeconomic changes resulting from industrialization— I follow Acharya, Blackwell, and Sen (2016b,a) and use a two-stage estimator, sequential g-estimator (Vansteelandt 2009; Joffe and Hsu 2009), to estimate the controlled direct effect of industrial complexes built by the autocrats starting in the 1960s. I find that both the presence and the duration of an industrial complex increased labor protests during the democratic transition period in the late 1980s. I also empirically demonstrate that the ge- 21 ographic concentration of factories was a crucial mechanism that mediated the effect of industrial complexes on labor protest by facilitating solidarity among workers and shaping mobilization strategies of labor activists. The result from this chapter complements a recent study by Hong and Park (2015) which finds a strong positive effect of industrial complexes on regime support under Park Chung Hee (1961-1979), but a weaker effect during the subsequent regime under Chun Doo Hwan (1980-1988). My finding demonstrates that industrialization, in the long run, had a destabilizing effect on the regime by facilitating labor mobilization. The construction of industrial complexes that brought workers together for the production of the national economy during the authoritarian period, inadvertently facilitated labor mobilization in the long run. The autocrats’ construction of industrial complexes created “pockets of resistance,” where workers and labor activists were conveniently brought together to mobilize and go against the very regime that created those industrial complexes in the first place. Overall, the chapter demonstrates the overarching argument of this dissertation that economic development has different effects on authoritarian durability in the short- versus long-term. It makes a critical contribution to our understanding of democratization by demonstrating that development-induced social changes that strengthen labor movements in the long run do not appear immediately, are cumulative, and have a modest or negligible in the short run. Chapter 319 examines how access to mass media, one key feature of modernization, impacted regime support in the short term—i.e., during the first decade of Park Chung Hee’s competitive authoritarian rule from 1961 to 1972. This period marks the beginning of South Korea’s rapid industrialization through the implementation of the Five-Year Plans of South Korea. It is also the period in which Park’s competitive authoritarian regime was

19This chapter is coauthored with Jae Seung Lee (Hongik University) and B.K. Song (Hanyang University).

22 transitioning towards “full authoritarianism” by installing the 1972 Yushin Constitution, which imposed authoritarian government and repression of all civil liberties. During this period, a series of acts intended to repress news agencies and broadcasting took place that helped to silence the critical media and laid the groundwork for Park’s absolute control over society. The electorate for the 1963, 1967, and 1971 elections was comprised of members of the Korean War generation and the post-war industrialization generation. Having spent their formative years during the Cold War prior to the “take-off” stage of economic development (Rustow 1970), these two generations are most exposed to physical economic insecurity among all political generations in South Korea since the establishment of the Republic of Korea in 1948. These generations were more vulnerable to regime repression or co- optation given their desire for physical and economic security tends to take higher priority than democracy. Using geographical and temporal variation in newspaper circulation and radio signal strength at the county level, we find that greater access to media was correlated with more opposition to the authoritarian incumbent but only when the government’s control of the media was weaker in the early 1960s. When state control of the media was stronger in the late 1960s, the relationship between media exposure and regime support disappears. Through a content analysis of newspaper articles, we demonstrate that pro-incumbency bias in news coverage resulting from tighter media control explains this distinctive pattern in the relationship between media exposure and electoral support. Findings from this chapter suggest that more access to mass media at an early stage of economic development do not automatically translate into more opposition to authoritarian rule, especially given the regime’s (varying) capability of exercising media control. Further, they provide support to my argument that in the early stages of socioeconomic development—when the politically active population has already been formed prior 23 to the onset of economic development—increasing levels of modernization is associated with regime legitimacy (derived from good economic performance) rather than demand for democracy. This is because the transformative effects of modernization is not exhibited by those who are old enough to participate in politics through voting or protesting. Lastly, Chapter 4 analyzes the conditions under which multiparty elections are more or less effective at co-opting opposition groups in a modernized authoritarian regime under Chun Doo Hwan (1980-1988). Specifically, using an original subnational events dataset,20 I empirically test whether voting for the new opposition party (New Korea Democratic Party) in the 1985 National Assembly elections had a positive or negative effect on participation in mass anti-regime protests during the famous June Democratic Uprising in 1987 that resulted in democratic concessions by the incumbent regime. I find that voting for the opposition party overall had a dampening effect on protest. However, the significant presence of social activists mitigated the dampening effect of election. This dampening effect is strongest in regions with a low proportion of college students but the magnitude of the effect decreases in areas with high percentage of college students. In short, voting for the opposition party had a heterogenous effect on anti-regime protest and the effect is moderated by the presence of college students in a given area. In areas with low proportion of college students, voting for the opposition party sub- stituted for participating in anti-regime protest because voting functioned as a safer means to directly express one’s opposition. However, in areas with high percentage of college students, students and citizens channeled their electoral mobilization (from campaigns and rallies) into street demonstrations. College students of the 1980s were the only group with the resource (time and knowledge) and pre-existent mobilizing structures (student councils and societies) to challenge the powerholders (Park 2008). Indeed, college students played a

20Description and coding manual for the events dataset are found in Appendix A.

24 significant role during both the 1985 National Assembly elections as well as the 1987 June Democratic Uprising by actively campaigning for the NKDP and organizing national rallies during the June Democratic Uprising. Hence, while multiparty elections seemed to have served as a safety valve by allowing citizens to “let off their steam,” elections may have also functioned as a focal point for mobilization for those—i.e., college students—who re- garded street campaigns and rallies (for opposition party candidates) as an opportunity to further engage in street demonstrations against the authoritarian regime. This chapter shows that there were differing impacts of the authoritarian government’s co-optation strategy (i.e., holding multiparty elections) in areas with more or less college students, the subpopulation belonging to the 386 generation. This finding is consistent with my argument that the full and transformative effects of socioeconomic changes (conducive to the emergence of democracy) appears through the rise of a “democratizing generation” that fully benefitted from economic development under authoritarianism.

1.6 Concluding Remarks and Future Research

This dissertation focuses on South Korea, a successful third-wave democracy. While the importance of economic development to South Korea’s democratization is not new, this dissertation reveals the differing short versus long term effects of development on authoritarian durability. In the short term, economic growth allowed South Korean autocrats to justify authoritarianism at the expense of political freedom. However, in the long run, the autocrats’ investment in economic growth gave rise to a new political generation— possessing democratic values and resources to engage in anti-regime mobilization—that replaced the previously co-opted generation that favored economic security over democracy. The South Korean case demonstrates that economic development does not quickly lead to democracy, as anticipated by some variants of modernization theory. Instead, the

25 positive effect of development on democracy manifests later in time as it takes a sufficient amount of time for the new generation to become the dominant political actors. The main findings of the dissertation has several implications for research and policy on democratization. Given the positive long-run effects of economic development on democracy, entities aimed at fostering democracy should welcome and encourage economic development in autocracies. However, when evaluating the relationship between the two, scholars and observers need to pay closer attention to how the two concepts are measured. Furthermore, rather than examining the impact of economic development on the society as a whole, we need to differentiate the effects on different subpopulations within a society, in particular the different generations that get exposed to different stages and aspects of socioeconomic development. Future research will expand the scope of the dissertation by exploring whether people’s economic and political experiences from the authoritarian period have an effect on their views on democracy and the ways in which they participate in politics. Building directly on my finding that generation turnover in civil society tilted the balance of power between the authoritarian regime and opposition forces in society, I will further explore this generational effect in the democratic context. I will first focus on the 386 generation that “fought for democracy” using extra-institutional tactics during the authoritarian period. Are they more or less likely to be satisfied with democracy? Are they more likely to trust and comply with the democratic governments? Do they still resort to extra-institutional tactics (e.g., protest) over institutional tactics (e.g., petitions and contacting legislative representatives) to hold the governments accountable? Secondly, I will compare and seek to explain the differences in the political behavior of the 386 generation and the subsequent generation born and raised under a democratic regime and thus more likely to have taken democracy for granted. In order to carry out this research, I will first utilize existing longitudinal survey data 26 such as the General Social Survey and track individual cohorts as they age to gather data on the generational differences in levels of political participation. I will also collect new data by conducting surveys and in-depth interviews in South Korea. In addition, using survey experiments, I will examine how an individual’s view regarding the authoritarian past affects one’s current view on democracy and level of political participation. Specifically, I will explore whether voters reward or punish undongkwonˇ 21 candidates—who belong to the 386 generation—for their past involvement in the democracy movement and how such background shapes their perception of candidates’ ideological position, competence, and electability in the democratic era. This research will contribute to our understanding of the process of democratic consolidation. While democratic consolidation is a crucial concept in comparative politics, it has been difficult to measure or empirically address the issues associated with the process in which a new democracy matures. By examining the generational effect of authoritarian legacies on political behavior, this research will identify the individual and cohort-level factors that dampen or facilitate democratic consolidation. This research will also theoretically and empirically advance the nascent scholarships on authoritarian nostalgia and authoritarian successor parties by demonstrating the generational effect that exists on views on democracy, levels of political participation, and voting behavior.

21Undongkwonˇ literally means the “movement sphere” and refers to “an individual activist or the minjung movement itself as a whole, or both” (Lee 2007, 16). 27 2| Long-term Effect of Industrialization on La- bor Activism

“The development of Modern Industry, therefore, cuts from under its feet the very foundation on which the bourgeoisie produces and appropriates products. What the bourgeoisie therefore produces, above all, are its own grave-diggers.” — Karl Marx (Communist Manifesto)

2.1 Introduction

What is the effect of industrialization on democratization? The democratization literature depicts labor as a pro-democratic force that can destabilize an authoritarian regime given its size and potential to disrupt economic and political stability.1 This threat to regime stability is even greater for industrializing countries when workers constitute a large seg- ment of the population and an important input to production. However, despite the crucial role that labor plays in the process of regime change, little is known about where the demand for democracy comes from or how and when the demand is formed in the ﬁrst place. Under what conditions can organized labor emerge as a pro-democracy force within an autocracy? For autocrats, industrialization is a mixed blessing. Economic growth can stabilize au-

1Some examples of the extant literature on labor and regime change include Therborn (1977, 1979); Valenzuela (1989); Rueschemeyer, Rueschemeyer, and Stephens (1992); Collier (1999); Collier and Collier (2002). 28 thoritarian regimes by enhancing political legitimacy and popular support of the regime. However, modernization theory predicts that high levels of economic development increases the probability of democratization. Recent developments in the democratization literature suggest that economic development has a conditional effect on democratization: while economic growth is stabilizing for the dictator, it is destabilizing for the dictatorship. This study develops this logic and demonstrates that while industrialization helps stabilize authoritarian regimes in the short-term, it facilitates the gradual development of pro- democratic labor mobilization which destabilizes the regime in the long-term. I argue that the contradictory effects of industrialization in the short- versus long-term is explained by the time lags between the implementation of industrial policies and socioeconomic changes resulting from economic development. Economic development does lead to democracy, but not in the smooth, incremental way anticipated by modernization theory. The empirical analysis relies on data from South Korea, a critical case for which existing scholarship counter-intuitively finds that industrialization contributed to both authoritarian durability and democratization. At first blush, Korea seems to have taken a “linear path” from industrialization to democracy. However, at least in the short term, industrialization had a stabilizing effect on the authoritarian regime. South Korea’s successful export- led industrialization relied on the construction of massive industrial complexes. Successful authoritarian regimes achieved economic growth and increased popular support for the ruling party through the development of industrial complexes which provided material benefits (employment, welfare, public infrastructure, etc.) and helped modernize the country. Concurrently, although labor was largely submissive throughout the authoritarian period, there was a gradual growth of workers’ consciousness and capacity to organize and engage in collective action against the state and capital holders. This culminated in a nationwide labor strike that provided the final impetus for democratization in the late 1980s. To empirically investigate the effect of government’s investment in industrial complexes 29 on labor activism—mediated by the socioeconomic changes resulting from industrialization— I follow Acharya, Blackwell, and Sen (2016a,b) and use a two-stage estimator, sequential g-estimator (Vansteelandt 2009; Joffe and Hsu 2009), to estimate the controlled direct effect of industrial complexes built by autocrats starting in the 1960s. I find that both the presence and the duration of an industrial complex increased labor protests during the democratic transition period in the late 1980s. I also empirically demonstrate that the geographic concentration of factories was a crucial mechanism that mediated the effect of industrial complexes on labor protest by facilitating solidarity among workers and shaping mobilization strategies of labor activists. This study has important implications for both scholars and policy makers. It provides empirical evidence to the emerging literature that highlights the conditional effect of economic development on democracy. By examining the differing effects of industrial policy on regime support in the short- versus long-term, this study enhances our understanding of the overall relationship between economic development and democratization. The finding demonstrates the importance of distinguishing the short- versus long-run effects of economic development on regime change. This suggests that, in the long run, encouraging economic development can increase the probability of democratization, the core tenant of modernization theory that many have discounted.2 The remainder of the paper is organized as follows. Section 2.2 presents the relevant literature. Section 2.3 consists of two parts. The first part provides background information on South Korea’s rapid industrialization and the gradual development of the labor movement under authoritarianism. The second part discusses how industrialization shaped labor

2For instance, Przeworski and Limongi (1997) and Przeworski et al. (2000) argue that development makes democracies endure, but it does not make them more likely to emerge. Studies that do not ﬁnd a positive relationship between economic development and democracy include Arat (1988); Sirowy and Inkeles (1990); Hadenius (1992). Acemoglu, Robinson, and Yared (2008, 2009) also do not ﬁnd a causal relationship between the two.

30 activism over time. Section 2.4 describes the data and methods and Section 2.5 presents the results. Section 2.6 concludes.

2.2 Literature

The causes of democratization have been explored theoretically and empirically by many political scientists since Lipset (1959)’s seminal piece, “Some Social Requisites of Democracy: Economic Development and Political Development.” Modernization theorists argue that various aspects of economic development—industrialization, urbanization, wealth, and education—constitute the preconditions for democracy (Lerner 1958; Lipset 1959, 1960; Deutsch 1961; Inkeles 1966).3 Earlier studies have investigated the relationship between democracy and economic development using regression analysis—assuming a linear relationship between the two variables. Many scholars do ﬁnd empirical support for modernization theory (e.g., Jackman 1973; Bollen 1979; Burkhart and Lewis-Beck 1994; Londregan and Poole 1996; Barro 1990; Boix and Stokes 2003; Epstein et al. 2006) while others do not ﬁnd a causal effect of economic development on democracy (e.g., Przeworski and Limongi 1997; Przeworski et al. 2000; Acemoglu, Robinson, and Yared 2008, 2009). More recent studies suggest that economic development has a conditional effect on democratization (Kennedy 2010; Miller 2012; Treisman 2014). While economic growth and increase in national income enhance support for the current regime, high levels of economic development bring about socioeconomic changes that—eventually—promote democratization. The “income effect” on democratization is activated by major institutional changes (Kennedy 2010) and leadership turnover (Miller 2012; Treisman 2014). These studies

3Modernization theory has been criticized by subsequent studies on democratization for being Western- centric, positing a single linear path from economic modernization to democracy, and overlooking the role of political institutions (Moore 1966; Huntington 1968; Rustow 1970; Rueschemeyer, Rueschemeyer, and Stephens 1992).

31 help reconcile apparently contradictory empirical results by demonstrating that economic growth is stabilizing for the dictator, but destabilizing for the dictatorship. However, these scholars take pro-democratic citizens as given and overlook the role of collective action undertaken by civil society actors that may have shaped these institutional changes and leadership turnover in the ﬁrst place. Social movements are increasingly considered relevant actors in democratization literature. One of the civil society actors frequently mentioned and examined in the democratization literature is organized labor. Labor movements have played an important role in the process of democratic transition (Therborn 1977, 1979; Valenzuela 1989; Rueschemeyer, Rueschemeyer, and Stephens 1992; Collier and Mahoney 1997; Collier 1999; Collier and Collier 2002). Labor is one of the major forces in the political opposition that explicitly demands for democratic regimes (Collier and Mahoney 1997). Rueschemeyer, Rueschemeyer, and Stephens (1992) even argue that labor is the most consistent pro-democratic force. La- bor movements generally have a greater capacity to engage in extensive and effective mass mobilization than other forces of civil society. Through more or less permanently established unions, labor movements have a network providing an underlying structure for organizing demonstrations and protests. Their mass base normally has speciﬁc and common interests, as well as a collective identity rooted in history. Most importantly, unlike other social forces in civil society such as student and church-related groups, labor movements can disrupt the economy by carrying general strikes, which enables them to put tremendous pressure on an authoritarian regime.4 Authoritarian industrialization can inadvertently produce strong labor movements. Struc- tural changes during the course of rapid industrialization not only alter the state-employer

4Studies in labor economics show that strikes have a negative impact on productivity (McHugh 1991), product quality (Krueger and Mas 2004), industry value (Becker and Olson 1986; DeFusco and Scott Fuess, Jr. 1991; Neumann 1980), and stock valuations (Dinardo and Hallock 2002).

32 relationship, but also affect working-class life in ways that help form worker organization and workers’ consciousness (Seidman 1994). Because workers’ capacity and policy preferences are determined by their structural position in the national economy vis-à-vis employers and the state, workers in the manufacturing sector and in competitive industries are expected to be better organized and exert greater leverage on their employers than workers in service sector or in declining industries (Korpi and Shalev 1979; Rogowski 1989; Pontusson 1993; Pontusson and Swenson 1996). Especially with the development of heavy-industry, workers in large, heavy-industry companies concentrated in industrial zones have better capabilities to organize unions and to engage in confrontational strategies than workers who are dispersed throughout numerous small ﬁrms (Kim 1993; Shin 1994; Orru, Woosely Biggart, Nicole, and Hamilton 1997; Huang 1999; Koo 2001). Moreover, severe repression or co-optation of labor that accompanies authoritarian industrialization strategies leads to politicization (and radicalization) of labor movement outside the legal labor relations framework, as well as alliance formation with other pro-democratic forces in society such as intellectuals, social activists, and community groups (Seidman 1994; Koo 2001; Lee 2007, 2011; Koo 2011). Industrialization does lead to democracy through the empowerment of subordinate classes, as modernization theory predicts. However, the effect of industrialization does not translate linearly into incremental increases in democracy. Social variables that modernization theorists identify as contributory to the development of a democratic society— industrialization, urbanization, education, and the development of the middle class—are all social conditions that change dramatically over an extended period time, but these changes usually happen slowly and at varying speeds. Above all, working-class formation is a “social and cultural transformation” that does not automatically emerge from its structural position in the national economy (Thompson 1966, 357). Rather, it is “the emergence of a relatively cohesive working class, self-conscious of its position in the social structure and 33 willing and capable of acting to affect it” (Katznelson 1986, 11). Because of the time lags between between the implementation of industrial policies and socioeconomic changes resulting from economic development, the cumulative strength of labor is manifested later in time. While socioeconomic change and democratization do hang together over time, they often do so in a non-linear fashion.

2.3 Background

Industrialization & Industrial Complexes in South Korea

Korea was a largely agrarian society prior to Japanese colonialism (1910-1945). Indus- trial development initially began under colonial rule mainly to aid the Japanese war effort. There were no industrial complexes or even large-scale industrial estates. Factories existed and operated individually, mostly in the southern part of Korea (present-day South Korea). They were concentrated sectorally in “light” industry such as textile, printing, and food manufacturing.5 When colonial rule ended, separation from the Japanese economy and social unrest brought about a sharp 40-75 percent decline in manufacturing from its height in the 1930s (Bank of Korea 1985; Shim and Lee 2008, 74). Further, the Korean War (1950- 1953) destroyed much of the physical infrastructure and industrial facilities built by the Japanese. It was not until the 1960s when President Park Chung Hee (1961-1979) carried out a series of economic reforms—including the development of industrial complexes in the 1960s and 1970s—that Korea’s industrialization took off again. During Park’s regime, South Korea grew rapidly—tripling in GDP over the course of three decades. After unsuccessful attempts at import-substitution industrialization from

5On the other hand, metal and chemical industries in the northern part of Korea (present-day North Korea), made up 90 percent and 82 percent of the total industrial production (Chosonˇ Bank Research Department 1948, I-101).

34 1961 to 1962, the Park government officially adopted export-oriented industrialization in 1963. As a result, exports increased from $87 million in 1963 to $835 million in 1970, with GNP increasing approximately 10 percent each year. In 1973, Park announced an industrial upgrading plan to promote heavy and chemical industrialization and from 1971 to 1980 the South Korean economy achieved a growth rate of 7.8 percent each year and the manufacturing sector grew at an annual rate of 14.8 percent. Important to the successful implementation of Park’s export-oriented strategy was the abundant supply of labor and the state-led development of manufacturing industries (John- son 1982; Amsden 1989; Deyo 1989; Jones and Sakong 1990; Woo-Cummings 1999). Starting in 1963, the government planned and built specialized industrial complexes (kongopˇ tanji) to “maximize synergic effects among related industries and to enhance competition among firms in the same industry” (Hong and Park 2015, 6). Industrial complexes were expected to generate economies of scale among firms and bring in external economy and overhead facilities to a given area (Bank of Korea 1970, 11). Although either the minister or local government was formally in charge of the official administration of industrial complexes, de facto decision-making regarding the industrial complexes were made by the president and the central government since all ministers, mayors, and governors were ap- pointed by him. The infrastructure for South Korea’s industrial complexes was built with public resources, but private investments were also induced through various incentive measures such as tax exceptions, subsidies, and controlled prices of land below the market price (Suh 1978, 18). The export-oriented industrial strategy (especially in labor-intensive light manufacturing such as textiles and electronics) in the 1960s accompanied development of industrial complexes in major cities in inland areas,6 where there were abundant supply of labor.

6For example, Seoul, Inch’on,ˇ Songnam,ˇ Pusan, Taegu, Taejon,ˇ Chonju,ˇ Chongju,ˇ Ch’unch’on.ˇ

35 Table 2.1: Number and Size of Industrial Complexes by Administrative Region, 1963-1987

Administrative Region Number of ICs Size Seoul 3 2,185 Inch’onˇ & Kyonggiˇ 11 50,138 Kangwonˇ 5 5,786 Ch’ungbuk 2 4,032 Ch’ungnam & Daejeon 3 2,115 Chonbukˇ 7 12,434 Chonnamˇ & Kwangju 11 26,575 Kyongbukˇ & Taegu 13 42,968 Kyongnamˇ & Pusan 12 124,466 Jeju – – Note: Size is measured in 1,000 m2. Agricultural industrial complexes are excluded. Source: Ministry of Commerce and Industry . [Sanggongbu] (1989)

With the heavy-chemical industry drive in the 1970s, large-scale industrial complexes were built in the coastal areas7 where factories had easier access to ports and water. In the 1980s industrial complexes were built in areas lacking industrial complexes to reduce regional economic disparity caused by the uneven patterns of industrialization in the preceding two decades. Table 2.1 provides information on the number and size of industrial complexes in each administrative region.

Short-Term Effects of Industrial Complexes & State Repression

These industrial complexes were not only the key engine of South Korea’s economic growth but also simultaneously allowed the authoritarian government to “diffuse political discontent and buy political legitimacy with economic performance” (Koo 2001, 30). Con- struction of industrial complexes provided targeted economic beneﬁts to the surrounding

7These cities are P’ohang, Ulsan, Koje,ˇ and Ch’angwonˇ and they are all located in Kyongnamˇ province.

36 areas by (1) increasing local employment, (2) enhancing the local economy by increasing local population and tax revenue, (3) bringing in large investments in infrastructure such as transportation facilities, sewage systems, electricity, and housing clusters, and (4) providing welfare benefits to workers in manufacturing sector including medical injury insurance and medical insurance (Bank of Korea 1970; Hong and Park 2015, 6). Because of the increased economic activities in these areas surrounding industrial complexes, residents became more supportive of the ruling party. According to a study by Hong and Park (2015), constituents living near an industrial complex became more supportive of the ruling party after the allocation of the industrial complex. For instance, the estimate shows that in 1973 voters cast almost 22.5 percent more votes for Park’s ruling party. How- ever, the political effect of industrial complexes disappears as the construction of industrial complexes approaches completion during the subsequent authoritarian regime under Chun Doo Hwan (1980-1987). The authors find that constituents increased their support for the incumbent party in the election immediately following the appointment of an industrial complex, but they do not find an additional positive effect on the beginning or completion of industrial complexes on ruling party vote share. They even observe stagnation of support during the construction period. While the construction of industrial complexes did help the ruling party in garnering electoral support to maintain its control of the National Assembly, the political effects of industrial complexes on regime support were smaller and weaker in the long term. Aside from garnering popular support through the construction of industrial complexes, the authoritarian governments of both Park Chung Hee (1961-1979) and Chun Doo Hwan (1980-1988) relied heavily on repression to demobilize labor. According to Deyo (1989, 3-5),

organized labor [in East Asian newly-industrialized countries] played a politically marginal and insigniﬁcant role in national affairs ... Rapid, sustained 37 industrialization has not altered the weak political position of labor ... [and] despite the creation of a vast factory work force over a period of three decades, labor movements in general remain controlled and inconsequential.

It is generally understood that South Korean workers largely remained submissive, unorganized, and politically quiescent under authoritarianism (Koo 2001).8 There were very few labor disputes in the 1960s and 1970s and these disputes were localized and focused on economic and welfare issues, not on political (or democracy-related) issues (Kim 2000; Koo 2001). The dominant explanations for labor weakness during the authoritarian period attribute colonial legacy (Kohli 2004), strong anti-Communist states that emerged in the post-war wiping out pre-existing leftist labor movements, and the authoritarian government’s use of security-oriented approach of labor exclusion and repression (Koo 2001, 2011). The union structure that came into existence under Park resembled the state-corporatist model, which is characterized by state sanctioning of unions, exclusive representation rights of ofﬁcial unions, and industrial union structure. However, in practice, ofﬁcial unions were neither encouraged, nor allowed, to represent workers. Industrial unions were not provided with much authority or resources to represent the interests formed at the local union level. Furthermore, unlike corporatist systems in other countries, Korea’s state-corporatist system did not employ any inducements to keep labor under control. The labor policies that accompanied export-led industrial development attempted to mobilize workers economically as an element of production and demobilize them politically as a possible threat to security and national development (Koo 2011, 125). The government used rhetoric of

8Labor organization did emerge in the 1920s under Japanese colonialism as well as brieﬂy in the post- liberation period from 1945 to 1948, but the leftist unions were completely destroyed by right-wing forces and the U.S. military forces during and after the Korean War (1950-1953). Labor disputes reemerged in 1960 after the fall of the Rhee regime (1948-1960), but was immediately repressed by Park Chung Hee’s new regime established through a military coup of May 1961. An exception to this general portrayal of weak labor in the 1960s and 1970s is Nam (2009)’s work on unionization of the Korea Shipbuilding and Engineering Corporation workers in Pusan under the Park regime.

38 nationalism and developmentalism to shape the motivation and self-identity of the workers.9 To control labor activism, the state primarily relied on security forces (such as the police, the Korea Central Intelligence Agency, and military security forces) rather than through legal or bureaucratic institutions, thus raising the costs of protest for any individual protestor. And when there were any attempts by workers to express their grievances in a more organized fashion, they were either co-opted, intimidated, ﬁred, or blacklisted.

Gradual Development of Labor Activism

Despite the state’s repressive and manipulative control of labor, there was a gradual growth of workers’ consciousness and organizational capacity that facilitated the labor movement—i.e., the process of creating new independent unions and capturing control of company unions (oyongˇ chohap) to subsequently transform them into genuine representative unions. In fact, South Korea developed a stronger labor movement than their coun- terparts in the other East Asian newly-industrialized economies (Deyo 1989; Huang 1999; Lee 2011). While labor conflicts (strikes and protests) were spontaneous, unorganized, short-lived, and confined to individual firms in the late 1960s to early 1970s, starting in the mid-1970s workers began to realize that importance of organizing representative unions in improving their working conditions and labor relations (Koo 2001, 72-73). Independent union movement began to emerge in the late 1970s and early 1980s first among female workers in small, light manufacturing industries, but it only occurred in a small number of factories located in the highly-urbanized Seoul and Inch’onˇ areas.10 In the development of the independent union movement, alliances were formed with other opposition groups—

9Factory workers were called sanopˇ uiˇ chonsaˇ “industrial warriors”), sanopˇ uiˇ yokkunˇ (“builders of industry"), and such’ul uiˇ yokkunˇ (“chief producers of exports”). 10These factories include Dongil Textiles, Wonpoong Textiles, Bando Trading Company, Pangrim Textiles, Y.H. Trading Company, Control Data, Crown Electronics, Signetics, and Tongkwang Textiles.

39 first between workers and church organizations in the 1970s and later between workers and students in the 1980s. Although the Protestant and Catholic churches in South Korea remained largely conservative and pro-government (Lee 2007, 222), two church organizations played a critical role during the 1970s: the Catholic organization Jeunesse Ouvnere Chretienne (JOC or Young Catholic Workers) and the Protestant group Urban Industrial Mission (UIM).11 As “factory pastors,” ministers of these organizations helped factory workers by visiting unions and management offices, calling families of workers, and visiting and praying for workers injured on the job at work. They also offered night courses to factory workers and formed small groups to teach workers about their legal and human rights as well as how to organize a union. Between 1972 and 1974, the UIM organized about eighty small groups in twenty companies. By 1975, one hundred small groups were active. These small groups were instrumental in sustaining the labor movement during the 1970s and 1980s (Lee 2007, 225). The nascent labor movement that developed under harsh political conditions in the 1970s was completely neutralized under the new military leader, Chun Doo Hwan. Chun’s new military took power through a coup in December 1979, a civil uprising against the new government known as the 5.18 Democratization Movement (or Kwangju Massacre) took place from May 18 to 27, 1980. More than 200 citizens, students, and workers were killed by the military. In the aftermath of the Kwangju Uprising the regime severely cracked down on labor, abolishing the new independent unions, expelling and blacklisting union leaders, and even sending them to prisons or “purification camps.” The Chun regime’s labor policies were “the most restrictive and oppressive in South Korean history” (Hart-Landsberg

11Both JOC and UIM were organized in the late 1950s, under the auspices of their international organizations, and began labor-oriented mission work in the early 1960s (Han’guk Kidokkyo Kyohoe Hyobˇ uihoeˇ [Christian Institute for the Study of Justice and Development] 1984, 255; Ogle 1990, 86-9).

40 1993, 219), which resulted in a drastic drop in the number of unions and disputes in the early to mid-1980s. Ironically, despite the increased severeness of state repression of labor, the labor movement grew stronger during the 1980s (Koo 2001; Lee 2007). The Labor Law of 1980 prohibited “third-party intervention,” making bargaining possible only between a company and a plant union. This prohibition of third-party intervention in disputes inadvertently contributed to the migration of opposition groups into factories to directly get involved with the labor movement by becoming factory workers. College students dropped out of college and entered factories as “disguised workers” to promote class consciousness and aid union organization. Their ultimate goal was to lead the labor struggles toward the larger political goals of ending military rule. Students formed small groups with workers to instill class consciousness among workers through group studies, discussions, and recreational activities. The estimates of students-turned-workers (hakchul) range from at least 1,000 (Hwang 1985, 14) to approximately 3,000 by the mid-1980s (Ogle 1990, 99). The product of this cumulative growth in the workers’ capacity to systematically organize and engage in collective action was manifested in the “Great Workers’ Struggle.” In the wake of democratization in 1987 (indicated by the dashed vertical line in Figure 2.1), there was an explosion of industrial protest and strikes on a scale unseen before in South Korea involving about 1.2 million workers (about a third of the regularly employed workforce) from most major industries (mining, manufacturing, shipbuilding, transportation and service sectors). These protests, now referred to as the Great Workers’ Struggle, erupted during the democratic transition period from July to August 1987 and represented the largest labor protests since the beginning of industrialization. It was the ﬁrst nationwide labor movement and the demands raised during the struggle were not limited to economic issues but also focused on democratization of workplace (Kim 2000). The issues raised at these sit-ins and strikes included labor rights, wage issues (guarantee of a minimum 41 wage, paid vacation, overtime pay), improvement of labor conditions, and democratization at workplace (liquidation of the existing state-corporatist unions and establishment of democratic unions).

Figure 2.1: Number of Labor Disputes and Labor Unions, 1963-1991

Source: Punki Pyolˇ Nodong Tonghyang Punsokˇ [Quarterly Labor Review] (Seoul: Korea Labor Institute, 1991, 1992); Figures produced using Table 7.1 in Koo (2001, 159).

Note: The ﬁrst peak in Figure 2.1 circa 1979-1980 corresponds to “Seoul Spring,” a very short period of political opening following the assassination of President Park Chung Hee on October 26, 1979 until General Chun Doo Hwan’s military coup on December 12, 1979.

The intensity of the Great Workers’ Struggle demonstrated that over time industrial complexes have become sites of labor activism that provided the ﬁnal impetus for full democratization in South Korea (see Figure 2.2). In the words of Koo (2001, 186),

the militancy and class solidarity demonstrated in the 1987 Great Workers Struggle did not occur accidentally, but was the outcome of accumulated past struggles, a few victories and many defeats, in which workers’ class aware- ness and political consciousness grew continuously. This gradual change in workers’ consciousness, largely hidden beneath the surface [emphasis added], prepared the fertile ground for the volcanic eruption in 1987.

42 Figure 2.2: Number of Labor Protest in 1987 and Location of Industrial Complexes

Source: Author’s dataset for labor protest and Ministry of Commerce and Industry [Sanggongbu] (1989) for data on industrial complexes.

43 How Industrial Complexes Facilitated Labor Activism

Previous research shows that workers in large, heavy-industry companies concentrated in industrial complexes have better capabilities to organize unions and to engage in confrontational strategies than workers who are dispersed throughout numerous small ﬁrms (Kim 1993; Shin 1994; Orru, Woosely Biggart, Nicole, and Hamilton 1997; Huang 1999; Koo 2001). The construction of industrial complexes during the authoritarian period in South Korea produced spatial concentration of factories that facilitated organization of independent unions by industrial workers as well as the entry of student and labor activists who assisted these factory workers in forming independent unions and organizing protests. The same industrial complexes that had previously bolstered regime popularity, now had the opposite effect. Drawing on the social movement literature that identiﬁes the importance of social ecology (e.g., living and working environments) for movement mobilization (e.g., Zhao 1998, 2001),12 I use secondary literature and primary sources to explain how spatial concentration of factories inside and outside these industrial complexes shaped the mobilization strategies of industrial workers and social activists. The South Korean government created an industrial structure that concentrated the largest numbers of workers in the smallest amount of space. The majority of the manufacturing industries were located in large urban areas around the major axis connecting Seoul and Pusan. By 1984, approximately 50 percent of all manufacturing workers were found in the Seoul-Kyoginˇ area (Seoul, Inch’on,ˇ and surrounding areas in Kyonggiˇ province), and

12“Ecological” impacts on movement mobilization have appeared in many studies on student movements in the U.S. in the 1960s (Berk 1974; Heirich 1971; Loﬂand 1970), China (Chow 1967; Wasserstrom 1991; Zhao 1998, 2001), and Russia (Kassow 1989). Chartism in England and Paris Commune in Paris (Gould 1995; Mann 1993), community movements in contemporary West (Delgado 1968; Perry, Gillespie, and Parker 1976), and black riots in the U.S. (Feagin and Hahn 1973; Fogelson 1971) exhibited ecology-dependent process of mobilization. Studies on the labor movement also demonstrate that tight-knit working-class communities provided important resources for movement (Calhoun 1981; Hanagan 1989; Gutman 1977; Perry 1993).

44 another 40 percent were in the Yongnamˇ (southeastern) region, which includes two major cities, Pusan and Taegu. The development of heavy chemical industry in the 1970s further created several new industrial cities in Yongnamˇ such as Ulsan, Masan, Ch’angwon,ˇ Kumi, and Okp’o (Koo 2001, 42). Industrial complexes were not only clustered in these areas (see clusters of red dots in Figure 2.2), but also the factories housed in these industrial complexes. For example, there were 265,273 workers in 2,493 factories in 15 industrial complexes in the Seoul-Kyonginˇ area in 1987.13 Inside the walls of these industrial complexes, the structure of dormitories and asso- ciational life inside these dormitories aided the process of organizing independent labor unions. The majority of workers lived together in tiny rooming houses in the surrounding areas or in dormitories inside factory compounds. In the case of Y.H. Trading Company, a major wig exporter to the U.S. in the 1970s, about 80 percent of their workers (1,200 workers) lived in three dormitory buildings with 10 to 12 workers per one large room and seven to eight workers per one small room (Chongˇ 1993, 102-3). These dormitories were built to monitor and increase productivity of the manufacturing workers to meet the government- set export goals. There were strict curfews and head residents (sagam) of each dormitory building watching over the workers during their limited amount of leisure time in between of after their shifts. However, these dormitories also became places of “everyday forms of resistance” where workers formed class consciousness an devised strategies to oppose their employers (Scott 1985). The physical layout of the dormitories facilitated the spread of information and provided a starting point for workers to create a network as well as an organizational structure for their movement. At Wonpoong Textile—one of the two companies that experienced increased labor activism among female workers in the early 1970s—workers were assigned

13Numbers calculated using information found in Ministry of Commerce and Industry [Sanggongbu] (1989).

45 to their dormitory rooms based on the order in which they entered the factory. This made it easier for workers from different departments within the factory to interact with one another and share their grievances (Chongˇ 1993, 107). During the process of organizing their own “democratic” independent labor unions, Wonpoong and Y.H. workers used the physical structure of their dormitories as a template for an organizational structure of the union. They replaced the existing head residents of each dormitory building with their own representatives and these representatives formed a management committee that would independently and democratically run the dormitories (Kim 2004, 121-122). Workers also formed social small groups (e.g., mutual aid society, hiking, singing, mask dance, Chinese character clubs) by themselves or through outside help from Chris- tian groups. They organized group activities such as outings and retreats. Organizing small groups and clubs was the easiest way to congregate factory workers in an organized fashion. During these small group meetings, workers exchanged their experiences and were ﬁrst exposed to the concept of organization (Kim 2004, 131-133). These groups also helped foster solidarity, recruit members, and produce a “discourse of resistance” for the independent union movement in the 1970s (Pang 1993; Chongˇ 1993; Kim 2004; Yu 2010).14 The close proximity of factories within each industrial complex also promoted social ties and communication among workers from different companies, a feature that contributed to inter-ﬁrm solidarity struggles in the 1980s. The leaders of independent unions inside the Kuro Industrial Complex, located in Seoul, collaborated closely with one another before and during the establishment of their unions by exchanging information and devising common strategies. When three union leaders of a textile company named Daewoo Apparel were arrested in 1985, workers from nine other companies in the Kuro Industrial Complex

14The Acacia group organized outings (yayuhoe) for Chunggye garment workers as a way to congregate the workers to organize their Chunggye Textile Union. The mask dance club members at Control Data were also the most active and committed workers in their struggle to form an independent union (Yu 2010, 66, 74).

46 organized a solidarity strike for six days that involved approximately 2,500 workers. Finally, activists (church groups, students, and intellectuals) considered geographic proximity of factories as a crucial factor when devising their mobilizing strategies to effec- tively assist the labor movement. First, Christian activists in the 1960-70s set up churches and ofﬁces in areas containing industrial complexes to increase their interactions with factory workers. These buildings located near industrial complexes provided physical spaces for some of the above-mentioned small groups to take place. Second, activists lived and worked as factory workers inside these complexes where many factories were located prox- imate to each other. For instance, Reverend Cho Hwa-sun, a Methodist minister assigned to Urban Industrial Mission, became a factory worker at a textile company named Dongil located inside the Korea Export Industrial Complex in Inch’on.ˇ She not only ministered to the young female workers at Dongil but also played a critical role in helping these workers during their lengthy struggle in the 1970s to capture control of the existing union and transform it into a representative one. Similarly, in the 1980s, when college students entered factories as disguised workers, they deliberately considered ecological factors such as density of factories in a given area and distance to night schools when deciding which factories to work at. According to the the “Guidelines for Factory Activism” (Kongjang hwaldong annaesoˇ) published by the Korean Student Christian Federation (Han’guk Ki- dok Haksaeng Ch’ong Yonmaengˇ ) in 1984, industrial complexes and areas surrounding the industrial complexes were the most ideal places to enter to efﬁciently mobilize factory workers.15

15Korea Democracy Foundation Archives No. 0063840.

47 2.4 Data & Methods

In this section, I describe the data and empirical strategy employed to examine the causal effect of industrial complex on labor activism in South Korea. Given historical accounts on the gradual growth in labor activism over time facilitated by the spatial concentration of factories and workers in these industrial complexes, I test whether the counties that had industrial complexes exhibited more labor protests during the Great Workers’ Struggle. To empirically explore the short versus long run effect of industrial complexes on labor activism, I also examine the effect of the duration of industrial complexes. Lastly, I statistically address whether and to what extent the casual effect of industrial complexes operated through the concentration of manufacturing ﬁrms. The outcome variable is measured as the total number of labor protests observed in each county during the 1987 Great Workers’ Struggle.16 These include strikes, sit-ins, street demonstrations, and rallies including union formation rallies (nojo kyolsˇ ongˇ dae- hoe). Using archival materials and publications from the Korea Democracy Foundation (KDF) and the National Council of Trade Unions (NCTU),17 I create an original dataset of 1,285 events that took place during the Great Workers’ Struggle.18 I also use Hoesa

16To ensure that these labor protests were pro-democratic, I regress this measure on the opposition candidates’ vote shares in the first free direct presidential election in December 1987 and find the coefficient of Great Workers’ Struggle to be statistically significant and positive. The results are reported in Table E.1 in the Appendix. 17Data is taken from a series of reports compiled by the KDF—“the Reports on the History of South Korea’s Regional Democracy Movement” and “the KDF Dictionary of Events Related to the Democracy Movement.” Reference numbers of individual KDF archival documents are 00368511, 0063998, 00839291. 18According to Han’guk Kidokkyo Kyohoe Hyobˇ uihoeˇ [Christian Institute for the Study of Justice and Development] (1987a)’s publication on the Great Workers’ Struggle, there were 1,796 events in the manufacturing sector, 1,247 among transportation workers (buses and taxis), and 127 in mining. I do not include events in the transportation industry as my outcome of interest is mobilization among factory workers. I consider and code protest events occurring over an extended period of time (on consecutive days) as one event and record duration of that event. Hence, the exact number of events may differ from the recorded number in the CISJD publication (which also includes events from September).

48 Yon’gamˇ [Company Yearbook] published by the Maeil Kyongjeˇ Sinmunsa (1987) to obtain additional information on the exact location of each factory when such information was unavailable in the KDF and NCTU materials. Two measures are constructed to examine the effect of an industrial complex: (1)

Presence of IC is a dummy variable indicating whether a county contains at least one industrial complex and (2) Duration of IC is the number of years since a given county has had at least one industrial complex. Given that factories started to operate as early as in the beginning year of construction, the duration measure is calculated as years since the beginning date of construction of the first industrial complex in a given county.19 While the first measure gets at the treatment effect of industrial complexes, the second measure captures the temporal dimension of the treatment effect. These measures are constructed using data on the 62 industrial complexes that were existing and operating in 1987 found in An Overview of Industrial Complexes published by the Ministry of Commerce and Industry [Sanggongbu] (1989). I include a number of pre-treatment confounders for the outcome (protests) and treatment (receipt of industrial complexes) along with post-treatment confounders for the relationship between the mediator of interest (concentration of factories) and the outcome. The pre-treatment variables are distance to the nearest port,20 proportion of rural households, proportion of employed population in manufacturing, population in 1960. These are the measures of the geographic conditions and locational efficiency that the government considered in allocating industrial complexes, especially during the early phase of the development of industrial complexes. The data is found in the 1960 population and housing census report produced by the Economic Planning Board.

19There were no industrial complexes closing down from their respective construction years until 1987. 20Distance was calculated using GIS ArcMap using the near tool. Data on the location of ports comes from Ministy of Construction [Konsˇ olbu]ˇ (1969).

49 The causal mechanism that I identify in Section 2.3—concentration of factories—is measured by the total number of manufacturing firms in each county found in the Report on Mining and Manufacturing Survey (Economic Planning Board 1987). The post-treatment control variables collected as controls for the mediator-outcome relationship include unemployment, population density, and population in 1985. The data for these covariates come from the 1985 population and housing census report. In addition to these controls, I include a dummy variable for counties that experienced protest during a nationwide democracy movement, the“June Democratic Uprising,” that occurred 1-2 months before the Great Workers’ Struggle in 1987. Approximately one million people, including opposition politicians, students, labor, and middle-class citizens, participated nationwide in anti-regime demonstrations during the June Democratic Uprising. The data comes from an events dataset that I created using archival materials from the Korea Democracy Foundation Archives. Appendix Table C.1 reports summary statistics for these data. Estimating the treatment effect of industrial complex on labor activism in 1987 while controlling for post-treatment covariates possibly introduces post-treatment bias into the analysis. To address this concern, following Acharya, Blackwell, and Sen (2016b,a), I use a two-stage estimator called the sequential g-estimator (Vansteelandt 2009) to estimate the controlled direct effect of industrial complex, which is the causal effect of industrial complex when the mediator is fixed at the same value for all units. In my case, a potential mechanisms driving the effect of industrial complexes on labor protest in 1987 is the geographic concentration of manufacturing firms. The exact procedure for implementing the sequential g-estimation is as follows. First, I estimate the effect of the mediator on labor protest in 1987 controlling for all of my pre-treatment and post-treatment covariates. I then transform the dependent variable by subtracting this effect. Lastly, I estimate the effect of industrial complex on the transformed variable, which is the controlled direct effect of 50 industrial complex.

2.5 Results

Table 2.2 reports the total effect of industrial complex in columns (1) and (3) for

Presence of IC and Duration of IC. Columns (2) and (4) report the controlled direct effect (ACDE) of industrial complex setting Manufacturing Firms (1987) to 0. The estimated ACDEs for both Presence of IC and Duration of IC are statistically significant and positive, suggesting that there exist a strong direct effect of industrial complex on labor protest that is not completely driven by the mediator. There are approximately 8 more protests during the Great Workers’ Struggle in counties with industrial complexes compared to those without an industrial complex, holding everything else at constant and fixing number of manufacturing firms at zero. And, ceteris paribus, a 5-year increase in the duration of industrial complexes is associated increase in 3 protest events, with the number of manufacturing firms fixed at zero. The results demonstrate that industrial complexes did increase labor activism and counties that had industrial complexes for a longer period of time had more protests during the Great Workers’ Struggle. To assess the strength of the causal mechanism, I examine the difference between the total effect (or average treatment effect; ATE) and the average controlled direct effect (ACDE). The difference in the estimated ATE and the estimated ACDE (“Portion Elim- inated”) is the extent to which a mediator participates in a mechanism either through (1) indirect effects or (2) the (causal) interaction between the treatment and the mediator at the individual level (VanderWeele 2014; Acharya, Blackwell, and Sen 2016a). Assuming that there is no interaction between the treatment and the mediators at the individual level, the estimated ACDE in columns (2) and (4) are smaller than the estimated ATEs in columns (1) and (3) suggesting that the causal mechanism—concentration of manufacturing firms—

51 Table 2.2: Estimated Effects of Industrial Complexes on Labor Protest Fixing Number of Manufacturing Firms in 1987

Labor Protest, 1987 (1) (2) (3) (4) Presence of IC 11.924∗∗∗ 7.636∗∗∗ (1.770) (2.666)

Duration of IC 0.885∗∗∗ 0.560∗∗∗ (0.112) (0.198)

Province Fixed Effects XXXX 1960 Covariates XXXX 1987 Covariates XX Bootstrapped SEs XX

Model OLS Seq. g-est. OLS Seq. g-est.

N 214 214 214 214 R2 0.823 0.822 0.834 0.828 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Baseline estimates are reported in columns (1) and (3). Columns (2) and (4) report the ACDE under Manufacturing firms, 1987 = 0. Columns (2)and (4) use sequential g-estimator of Vansteelandt (2009) to estimate the controlled direct effect (ACDE) of industrial complexes. The standard errors are bootstrapped using the R code provided in Appendix C of Acharya, Blackwell, and Sen (2016a). First stage estimates from the sequential g-estimation model are reported in Appendix Table C.3.

52 did play an important role. In fact, fixing the mediator eliminates about 36 percent of the average treatment effect.21 Figure 2.322 further shows that a part of the effect is driven by the interaction between industrial complex and concentration of manufacturing firms at the individual level. The effects of both Presence of IC and Duration of IC on labor protests are stronger (more positive) in counties with high concentration of manufacturing firms.

Figure 2.3: ACDE of ICs as a Function of the Fixed Level of Manufacturing Firms in 1987

I now examine whether and to what extent the above results are alternatively driven by changes in local population resulting from the construction of industrial complexes. I use sequential-g estimation to estimate the ACDE of industrial complex ﬁxing (logged) population at zero and check its proximity to zero. Showing that there is a non-zero ACDE

21 ATE−ACDE It is calculated as the following using the estimates from columns (1) and (2) in Table 2.2: ATE = 11.924−7.636 11.924 = .3596. 22Vertical lines are 95% conﬁdence intervals from 1,000 bootstrapped replications.

53 would suggest that the effect of industrial complex is not exclusively due to population change. Table 2.3 reports the results. The ACDE for Presence of IC and Duration of IC, reported in columns (2) and (4), are both statistically signiﬁcant, positive, and greater than zero, suggesting that there is an effect of industrial complex on labor protest other than through levels of population.

Table 2.3: Estimated Effects of Industrial Complexes on Labor Protest Fixing Population in 1985

Labor Protest, 1987 (1) (2) (3) (4) Presence of IC 11.924∗∗∗ 9.599∗∗∗ (1.770) (3.959)

Duration of IC 0.885∗∗∗ 0.739∗∗∗ (0.112) (0.281)

Province Fixed Effects XXXX 1960 Covariates XXXX 1987 Covariates XX Bootstrapped SEs XX

Model OLS Seq. g-est. OLS Seq. g-est.

N 214 214 214 214 R2 0.823 0.818 0.834 0.828 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Baseline estimates are reported in columns (1) and (3). Columns (2) and (4) report the ACDE under Log Population, 1985 = 0. Columns (2) and (4) use sequential g-estimator of Vansteelandt (2009) to estimate the controlled direct effect (ACDE) of industrial complexes. The standard errors are bootstrapped using the R code provided in Appendix C of Acharya, Blackwell, and Sen (2016a). First stage estimates from the sequential g-estimation model are reported in Appendix Table C.4.

54 To further explore the role of population change in the casual effect as a mediator versus the one of my proposed mechanism—concentration of manufacturing firms—I separately compare the ATE and ACDE for each mediator. The difference between the ATE and ACDE summarizes the role of the mediator in a causal mechanism for the effect of industrial complex. Comparing the differences between ATE and ACDE for each mediator will allow me to estimate support for a preferred mechanism. Figure 2.4 displays the coefficient on Presence of IC for the total effect, ACDE fixing population at zero, and ACDE fixing number of manufacturing firms at 0. Assuming that there is no interaction between the treatment (Presence of IC) and mediators at the individual level, the difference between ATE and ACDE are greater for Manufacturing Firms (1987) compared to the one of Log Population (1985). While fixing the concentration of manufacturing firms eliminates about 36 percent of the ATE, fixing the population eliminates about 19 percent of the ATE.23 This suggests that the concentration of manufacturing firms appears to play a greater role in a causal mechanism. Lastly, to demonstrate that the results are not driven by modelling assumptions in sequential g-estimation, I perform nearest-neighbor Mahalanobis distance matching. I match by pre-treatment covariates from 1960—population, distance to the closest port, proportion of rural households, proportion of workers in manufacturing industry. The balance achieved in the resulting matched data is shown in Figure 2.5 and Table 2.4 gives the results. The results show that there is a statistically significant and positive effect of the presence of industrial complexes on labor protest. The average treatment effects of the treated are approximately 15-16 protests during the Great Workers’ Struggle without or with pre-treatment controls.

23 ATE−ACDE It is calculated as the following using the estimates from columns (1) and (2) in Table 2.3: ATE = 11.924−9.599 11.924 = .1950

55 Figure 2.4: Coefﬁcient on Industrial Complex (Presence of IC)

Figure 2.5: Quantile-Quantile (QQ) Plots of Each Covariate Used to Match

Table 2.4: Matching Results

Average Treatment Effect 95% Conﬁdence Interval (Without Pre-treatment Controls) 16.341 (3.747, 28.802) (With Pre-treatment Controls) 15.465 (3.258, 28.719)

56 To summarize, the results show that counties that had at least one industrial complex had more labor protest during the 1987 Great Labor Struggle. And the counties that had at least one industrial complex for a longer period of time exhibited more protest events in 1987. These ﬁndings empirically demonstrate that the autocrats’ construction of industrial complexes (inadvertently) facilitated labor mobilization that challenged the authoritarian regime. The result also shows that the geographic concentration of factories resulting from the construction of industrial complexes played a mediating role in the effect of industrial complexes on labor mobilization.

2.6 Conclusion

This paper has examined the differing effects of industrial policy on regime support in the short- versus long-term. While scholars ﬁnd that economic growth contributes to regime stability, modernization theory predicts that industrialization will lead to democratization. This study shows that economic development does lead to democracy, but not in the smooth, incremental way anticipated by modernization theory. The apparently contradictory ﬁndings on the effects of economic development can be explained by the time lags between the implementation of industrial policies and socioeconomic changes resulting from economic development. To show case this argument, I relied on data from South Korea, a critical case in which industrialization contributed to both authoritarian durability and democratization. Although South Korea seems to have taken a linear path from poor/autocracy to rich/democracy, economic growth stabilized the authoritarian regime in the short run, but facilitated democratization in the long run. While labor was largely submissive during the authoritarian era, over time there was a gradual growth of grievances among workers that prompted labor mobilization in alliance with church organizations, students, and intellectuals. By estimating

57 the controlled direct effect of industrialization on labor activism, this study finds that a sustained presence of industrial complexes—that produced high concentration of factories— increased labor protests during the democratic transition period in the late 1980s. The empirical result of this study complements a recent study by Hong and Park (2015) which finds a strong positive effect of industrial complexes on regime support under Park Chung Hee (1961-1979), but a weaker effect during the subsequent regime under Chun Doo Hwan (1980-1988). My finding demonstrates that industrialization, in the long run, had a destabilizing effect on the regime by facilitating labor mobilization. The construction of industrial complexes that brought workers together for the production of the national economy during the authoritarian period, inadvertently facilitated labor mobilization in the long run. The autocrats’ construction of industrial complexes created “pockets of resistance,” where workers and activists were conveniently brought together to mobilize and go against the very regime that created those industrial complexes in the first place. Through an examination of the effect of economic development on a societal outcome associated with regime stability at the subnational level, this study enhances our understanding of the relationship between economic development and democracy by demonstrating that economic development has different effects in the short- versus long-term. Furthermore, it makes a critical contribution to our understanding of democratization by demonstrating that development-induced social changes that strengthen labor movements in the long run do not appear immediately, are cumulative, and have a modest or negligible in the short run. While socioeconomic development and democratization do hang together over time, they often do so in a non-linear fashion.

58 3| Media Exposure and Regime Support

3.1 Introduction

South Korea faced severe state censorship and control throughout the twentieth century, especially under the authoritarian regimes of Park Chung Hee (1961-1979) and Chun Doo Hwan (1980-1987). During the period in which Park’s competitive authoritarian regime was transitioning towards “full authoritarianism” (1961-1972), a series of acts intended to repress news agencies and broadcasting stations took place that helped to silence the critical media and laid the groundwork for Park’s absolute control over society. However, South Korea’s free press movement emerged precisely during this dark age of state censorship in the 1970s. The impact of Park’s suppression of the media on the expression of citizens’ views and support for the regime thus remain unclear. With the lack of empirical research on the role and effects of media in South Korea, these mixed outcomes call into question the relationship between media access and regime support in authoritarian regimes. Studies on media development emphasize that the media can have a liberalizing effect on political systems by serving as a watchdog, but that state-controlled media props up authoritarian rule by functioning as a government mouthpiece. This study examines the relationship between media exposure and regime support in authoritarian regimes where mass media is not continuously or entirely controlled but where the regime maintains the capability of exercising control. Variation in media control in competitive authoritarian

59 regimes allows us to identify the conditions under which media exposure has a subversive or bolstering effect on authoritarian rule. To empirically explore the impact of media access on regime support in competitive authoritarian regimes, we look at Park Chung Hee’s dictatorship in South Korea. Despite a plethora of anecdotal evidence on the Park regime’s media control, this study is the first attempt to our knowledge to empirically explore when and how media exposure affected public support for the authoritarian incumbent. We specifically focus on Park’s pre-Yushin regime, the competitive authoritarian period in which Park contested in competitive elections until the promulgation of the Yushin Constitution in 1972 that abolished direct presidential elections. During this period, media control of both newspapers and radio became more severe beginning in the mid-1960s. We leverage this temporal variation in media control to examine the differences in regime support across Park’s pre-Yushin regime using fine-grained data on newspaper circulation and election results at the county level. We also conduct a content analysis of newspapers for more than 400 articles published in the election years (1963 and 1967) to investigate the mechanism linking media control and regime support. Finally, we collect data on radio transmitters and spatial boundaries of election districts to estimate radio signal strength using geographic information system (GIS) and the Irregular Terrain Model (Hufford 2002). The content analysis of newspapers reveals that the government’s tighter control of the media in the late 1960s is associated with media coverage that favors the authoritarian incumbent, Park Chung Hee. With greater state control of news agencies, there was less coverage on the opposition candidates and fewer articles containing negative information or criticism of Park. Corroborating with our findings from the content analysis, the results from the regression analyses show that when government control of the media was weaker in the early 1960s, greater access to media was correlated with more opposition to the regime. However, we also find that when the media was severely repressed in the late 60 1960s, the correlation between media exposure and opposition disappears. More generally, our work advances the growing literature on competitive authoritarianism by demonstrating that the effect of media access on regime support is ambiguous in competitive authoritarian regimes. We also add to the discussion of repression and mobilization in authoritarian regimes by showing that media control, as a form of repression, can silence opposition. This paper proceeds as following: we first discuss the existing literature on the political effects of media development and then provide background information on Park’s pre-Yushin regime and temporal patterns of media control. Subsequently we lay out our empirical hypotheses and describe the data and empirical strategy used to test the hypotheses. Lastly, we report our results and conclude.

3.2 Existing Literature and the Argument

The conventional wisdom from modernization theory holds that the development of mass media leads to positive social and political change (Lerner 1958). Media development is often linked to increase in freedom of expression, which is one of the key components of a liberal democracy. Scholars have used freedom of the press (from Freedom House and Polity Project) as one indicator of the extent to which a political system is lib- eralized. Indeed, free media—domestic and international—played a critical role in the diffusion of democratic ideas and social movements (Howard et al. 2011; Snow, Vliegenthart, and Corrigall-Brown 2007; Strang and Soule 1998). Television reporting had a signaling and accelerating effect on public protest in Eastern Europe in the late 1980s (Hunting- ton 1996) and foreign media played a key role in the 1989 East German revolution (Kern 2011). Western broadcast’s coverage of domestic politics of democratic nations, in the long run, installed pro-democratic values and undermined public support for communism

61 and authoritarianism (Rustow 1990; Diamond 1993a,b; Dalton 1994; Rohrschneider 1994, 1996; Whitehead 1996; Bennett 1998; Sükösd 2000) and criticisms voiced by mass media during regime change contributed to the erosion of regime legitimacy and propelled democratization in many third wave democracies (Lawson 2002; Olukotun 2002; Rawnsley and Rawnsley 1998). In advanced and consolidating democracies, the development of mass media has a liberalizing, if not democratizing, effect on political system because of the media’s watchdog role in monitoring the conduct of government officials. Mass media provide information to voters to electorally punish politicians for “bad” (e.g., being suspected of corruption or being captured by special interests) behavior. For instance, Chang, Golden, and Hill (2010) find that Italian legislators suspected of wrongdoing are punished by voters only when the press begins to report on political corruption. Similarly, Costas-Pérez, Solé-Ollé, and Sorribas-Navarro (2012) show that incumbent mayors in Spain are punished more for involvement in a scandal when the press devotes a large amount of coverage to the scandal. Fergusson (2014) finds that concentration of special interest contributions to incumbent senators in the United States is punished by voters living in areas where candidates receive more coverage from their local television stations. Larreguy, Marshall, and Snyder (2014) also show that voters in Mexico punish the party of corrupt mayors only in areas covered by local media stations. In contrast, access to media does not cleanly translate into political behavior in authoritarian regimes because authoritarian governments vary in the degree and ways in which they exercise control over media. Some authoritarian leaders attempt to attenuate the potentially destabilizing effects of the media by spending significant financial and administrative resources to quell their influences. For instance, the Fujimori regime paid the largest amounts of bribe to the news media (among politicians, judges, and the news media) in maintaining its rule (McMillan and Zoido 2004). Others such as the Chinese Communist 62 Party (CCP) regime used the press media in propagandizing the legal system as overly positive (Stockmann and Gallagher 2011) and exercised its bureaucratic capacity to censor information on the Internet that may spur collective action (King, Pan, and Roberts 2013). This impetus to control the media is especially strong in competitive authoritarian regimes where there are constitutional channels through which opposition groups can compete for executive power in a meaningful way (Levitsky and Way 2002, 2010). As noted by Lev- itsky and Way (2010), leaders in competitive authoritarian regimes—whose authoritarian survival depends on votes—often control state-owned as well as private-owned media to create an uneven playing field and sway elections in favor of the ruling party. While more access to free media can have subversive effects in authoritarian regimes, government-controlled media can bolster authoritarian rule. Recent studies suggest that media exposure can increase popular support for authoritarian regimes when the media is highly controlled. Enikolopov, Petrova, and Zhuravskaya (2011) find that Kremlin’s media control of state-owned television played a crucial role in Putin’s electoral victory in 2000 despite his low popularity rating (below two percent) the previous year. Similarly, Adena et al. (2013) show that while access to radio was negatively correlated with the Nazi vote before the Nazis controlled the media, it helped the Nazis electorally when they used radio as their propaganda tool. Building on these studies, we explore the impact of mass media on authoritarian incumbent support by distinguishing between media exposure and media control. While exposure to different views and information within media can impact voters’ political knowledge, authoritarian governments can alter this effect through media control. Governments can exercise selective censorship, limit citizen access to mass media, and/or actively shape media content. To better understand media development in authoritarian regimes, it is imperative to examine the effect of media exposure on political behavior conditional on the timing and extent of media control. 63 Furthermore, given that authoritarian governments differ in the extent and nature of media control, this article argues that the effect of access to authoritarian media is ambiguous, especially in competitive authoritarian regimes. In fully authoritarian regimes, the state exercises extensive control over mass media. Autocrats exploit mass media to propagate their ideology and to enhance popular support for their regimes. Simultaneously, they monitor and control mass media to preempt it from becoming a focal point for political mobilization that could undermine the regime’s legitimacy. However, in competitive authoritarian regimes, where formal democratic institutions—such as elections, political parties, and legislature—are the primary means of gaining power, independent media exist and civic and opposition groups operate above ground (Levitsky and Way 2010). The media is not entirely controlled at all times although the regime maintains the capability of exercising control, which generates variance in media freedom across time and space. This variance in media exposure under competitive authoritarianism, in turn, can have divergent effects on regime support. Mass media in hybrid regimes could theoretically have a subversive effect by serving as a watchdog, as media does in other democratic regimes that hold regular, competitive elections. However, mass media could also bolster hybrid regimes by functioning as a mouthpiece of the government as they do, as in fully authoritarian regimes. Media exposure may have these contradictory effects in different competitive authoritarian regimes. Whether the development of media is harmful or beneficial for the regime is thus heavily shaped by the ways in which the government exercises its control over the media.

3.3 Context

We examine the relationships between media exposure, media control, and regime support in South Korea under Park Chung Hee (1961-1979). Speciﬁcally, we study the ﬁrst

64 decade of Park’s rule (pre-Yushin regime; 1961-1972), which is characterized as “competi- tively authoritarian” (Levitsky and Way 2010) Three direct presidential elections were held in 1963, 1967, and 1971. Park won the 1963 and 1967 presidential elections by narrow margins and amended the constitution in 1969 to allow three terms in ofﬁce. Multiparty elections in these years were competitive and opposition candidates were able to formally contest with no legal controls that prevented them from running public campaigns. On the other hand, civil liberties were violated in the sense that media, opposition politicians, and activists were subject to harassment, arrest, and violent attacks.1 In 1972 Park consolidated his authoritarian grip on power and abolished presidential elections through the inaugu- ration of the Yushin (Revitalization) Constitution, which endowed him with near-absolute power as a “legal dictatorship.”2 We thus restrict our period of study to Park’s Pre-Yushin regime (1961-1972). The media industry was relatively young and underdeveloped when Park Chung Hee came to power in 1961. Initially the Park government stipulated laws and legal restrictions to control the media industry. State censorship increased in the late 1960s as a preemptive strategy in preparation for the promulgation of the Yushin Constitution. Media control strategy also became more blatant and coercive over time. This temporal variation in the extent of media control under Park during the competitive authoritarian period provides a suitable environment to explore the effect of access to mass media on regime support in competitive authoritarian regimes. We now provide descriptive information on when and how Park’s government controlled the press and broadcast media leading up to the installment of the Yushin Constitution as well as how the media industry responded to such

1The Article 18 of the Constitution of the Third Republic (1962-1972) guarantees freedom of speech and press, and freedom of assembly and association, but these freedoms were qualiﬁed in varying degrees (Youm 1996, 52). 2According to the Polity IV data series, the Polity Score, which ranges from +10 (strongly democratic) to −10 (strongly autocratic), changes from 3 in 1971 to −9 in 1972.

65 changes.

3.3.1 Press Media

Immediately preceding Park’s coup-born regime was a short-lived parliamentary government led by a ﬁgurehead president (Yun Bo Son)ˇ and a prime minister (Chang Myon)ˇ from August 1960 to May 1961. Under Chang’s weak government—marred by poverty, political instability, and social stagnation—the Korean media was left uncontrolled and number of periodicals proliferated at an unprecedented rate.3 Upon taking power in 1961, Park believed that the “free and chaotic press” was at least partially to blame the “social chaos”of the Chang regime (Youm 1996, 54). As such, under his initial period of martial law (1961-1962) he introduced censorship measures that aimed to weaken overtly incen- diary periodicals and news outlets. The Supreme Council for National Reconstruction (SCNR), a temporary governing body of the coup-born regime, announced Decrees No. 1 and No. 4, prohibiting the press and media from “agitating, distorting, exaggerating, or criticizing” the “revolution” (i.e., the coup). Subsequently, the SCNR announced Decree No. 11, which intended to “purify” the media by purging “pseudo-journalists and pseudo- media agencies” from the previous regime. Through Decree No. 11, the SCNR imposed strict facility standards of newspapers and news services requiring that newspapers be published only by those with complete printing facilities for newspaper production and that news services be limited to those with complete wire service facilities for transmission and reception. This resulted in a signiﬁcant reduction in the number of newspapers and news agencies.4

3Compared to those in 1959, the number of dailies increased 51 percent, weeklies 96 percent, other publications 31 percent, and news agencies 300 percent. By the end of the Chang government, the number of dailies tripled and that of news agencies increased about 20 times (Watanabe 1964). 4There were 960 “pseudo” journalists who were either arrested or put on trial from May 16, 1961 to June 22, 1962 (Joo, Kim, and Park 1997, 83) and 141 of them were arrested for press coverage issues or violation

66 Nevertheless, with the lifting of martial law in December 1962, the press recovered from two years of stagnation and established their critical stance towards the regime. For instance, in 1962 Park initially took back his promise to transfer power to a democratically- elected civilian government and that military rule would be extended for another four years. Major newspapers including Chosonˇ , Kyonghyangˇ , and Tonga immediately displayed their opposition to Park’s statement by not including any editorial in their editions. In response, as originally promised, Park ran as a civilian candidate of the Democratic Republican Party (DRP) in the 1963 presidential elections and won by a narrow margin of 1.5 percent. Alongside Park’s tenuous popularity, his perception of threat shifted to popular unrest.5 Mass student protests erupted in 1964 against Park’s diplomatic talks to normalize relations with Japan (which had colonized Korea from 1910 to 1945).6 Park’s normalization talks with Japan met vociferous domestic opposition from opposition politicians, college students, and the media who criticized the talks as a “humiliating diplomacy.” The media provided extensive coverage of student demonstrations on Korea-Japan Treaty talks, including the Park regime’s crackdown on these protests. Due to the severity of the protests, Park declared martial law in June 1964 and imposed new restrictions on the media with the intention to limit the ability of the press to incite or prolong popular unrest. In August 1964 the National Assembly passed the Media Ethics Committee Law without the presence of any opposition party members. The law aimed at “enhancing the effectiveness of self-regulation by the press and broadcasting,” further limited the autonomy of the press and was strongly criticized and opposed by the

of the decrees or the Anti-Communist Law (Kim 2009, 141). 5Park’s primary domestic security was with popular unrests as elite threat was relatively low. See Chapters 5 and 8 in Greitens (n.d.) on Park’s changing threat perceptions and its impact on the internal security apparatus in the 1960s and 1970s. 6Park pursued normalization talks with Japan starting in 1964 primarily for the security of his regime (including obtaining funds for economic development) as well as in response to the pressures from the U.S.

67 press and the public.7 Five newspapers with substantial readership—Chosonˇ , Kyonghyangˇ , Maeil, Taehan, and Tonga—explicitly or implicitly defied this. The government retaliated by canceling the subscriptions of these five newspapers in all government organizations and households of public servants as well as excluding them from receiving any special benefits (e.g., tax breaks and free loans). State censorship further increased in the late 1960s as Park was preparing to amend the constitution to allow a third presidential limit. After having experienced the press- government confrontation in 1964, Park looked for an alternative way to control the media without having to enact media-related laws and he issued a special order to create a media-control unit within the Korean Central Intelligence Agency (KCIA) (Kim 2009, 180). Although the KCIA was only allowed to investigate cases involving the violation of the Anti-Communist Law, they arrested and investigated any journalists or editors responsible for articles that criticized the government (National Intelligence Service [Kukka Chongbowˇ on]ˇ 2007, 43). The KCIA agents were also stationed at central offices of the major newspapers to directly monitor and control their activities. They analyzed newspaper articles in detail, categorizing them according to the targets or issues that were being criticized, as well as where the article was placed in the newspaper, suggesting that the KCIA paid keen attention to the content of these newspapers and their stances towards the regime.8 The most well-known media repression event during this period of increased state cen-

7See Chapter 5 of Chang (2015b) for a review of media repression and the free press movement under Park Chung Hee. 8A report produced by the KCIA in 1966 entitled, “An Analysis on Chosonˇ Daily’s Primary Tone from 1965.1.11 to 1966.6.30,” includes the following categories: articles criticizing the government or the DRP, articles agitating the public, articles advocating the opposition party’s views, articles criticizing state em- ployee’s corruption, articles praising left-wing ideology, articles containing anti-American sentiments, articles on uniﬁcation, etc. (National Intelligence Service [Kukka Chongbowˇ on]ˇ 2007, 39).

68 sorship is the enforced sale of the Kyonghyangˇ Daily,9 which has been known to be critical and resistant of the government ever since its founding in 1966.10 The main reason for the sale was its failure to repay a debt of 46 million won. However, given the fact that the majority of newspapers at that time were operating under great deﬁcits, it was evident that the government wanted to control the management of Kyonghyangˇ that had been critical of the regime and its policies. The owner of Kyonghyangˇ was arrested in 1965 for violating the Anti-Communist Law and the paper was sold to Kia Industry, the sole government- supported bidder in the auction in 1966. With the management change, Kyonghyangˇ immediately became a pro-government newspaper. The Korean media—in contrast to their collective resistance in 1964—remained silent regarding the enforced sale of Kyonghyangˇ . The media’s subordination to government control became more apparent in the run-up to the 1967 election when the media refused to support the opposition party’s effort in advocating for press freedom. The major opposition party, the New Democratic Party (NDP), led a campaign for a free press, which included sending letters to international organizations such as the International Press Institute and the UN Commission for the Uniﬁcation and Rehabilitation of Korea. When the NDP asked the press for their support, the newspapers not only turned down the request, but also criticized the opposition by questioning the political motivation of the campaign (Song et al. 2000, 294-295). By the end of the 1960s, the papers that unequivocally resisted the government in 1964 were completely subordinate to government control. Kyonghyangˇ was sold to a

9Kyunghyang Daily was founded in 1947 by the Catholic Church. It was temporarily closed down in April 1959 by the Rhee administration for having printed “false editorials,” but revived its publication after the April 19 Movement. Later it was privately owned by Jun-Gu Lee in 1963 until he was imprisoned for violating the Anti-Communist Law in 1965. 10For instance, Kyonghyangˇ exposed a political scandal (“Three-Minutes Profiteering Incident”) in 1963 in which the ruling Democratic Republican Party assisted big business monopolies in tax evasion and ex- cessing profiteering in return for political funds. It also criticized Park’s unpopular Korea-Japan Treaty talks. Furthermore, during the 1963 presidential election, Kyonghyangˇ provided the opposition candidate, Yun Bo Son,ˇ with materials revealing Park’s past involvement in a Communist party named South Korean Workers’ Party, which defied Park’s heavy emphasis on anti-Communism. 69 government-supported company in 1966. Chosonˇ Daily was co-opted by the government in 1967 upon receiving four million U.S. dollars of foreign loans from Japan to build a tourist hotel in Seoul. Tonga, the last remaining resistant newspaper agency, finally gave in when the government arrested its several senior journalists for violating the Anti-Communist Law in 1968. The Korean media again took no action against government’s attack on Tonga as well as press freedom issues in general11 and this marked the beginning of the ‘dark age’ of the Korean press, which lasted until democratization in 1987.12

3.3.2 Broadcast Media

The Park regime also used broadcast media to propagandize and legitimize the coup- born regime. Unlike the press media, the introduction and development of commercial broadcast media were considered to be a part of the economic development process.13 The commercial broadcasters were not exempt from government control, but as long as they shared the economic and political interests of the government, Park was committed to maintain them in operation (Kwak 2012, 14). The Korean Broadcasting System (KBS), a state-owned broadcaster, lacked autonomy from the state, presenting few views that were different from those of the government. MBC (Munhwa Broadcasting Company later renamed as Munhwa Broadcasting Corpo- ration), although a private broadcaster, was essentially owned and managed by the ruling

11While Maeil did remain as an anti-government newspaper, as a local newspaper in Taegu city with Catholic background, its audience and inﬂuence were limited. 12Although the media industry was severely repressed by the authoritarian governments, it is important to note that during this ‘dark age’ of state censorship the free press movement emerged and journalists joined the larger democracy movement in the 1970s (Chang 2015b). 13Throughout the paper, we restrict our attention to radio. According to the Ministry of Culture and Information [Munhwa Kongbobu] (1968)’s publication, there were signiﬁcantly more broadcasting studios and transmitters for radio than for television in the 1960s—3 out of 43 broadcasting studios for TV and 10 out of 49 transmitters were for TV.

70 elite. Since 1962 MBC was operated by the 5.16 Foundation,14 which was established in 1962 by incorporating Kim Ji-Tae’s Puil Scholarship Foundation. Kim Chit’ae, a journalist as well as the president of Pusan MBC and Pusan Daily, was arrested in 1961 for “illegally amassing wealth” and released the following year on the condition that he “donate” his scholarship foundation to the state. By the early 1970s both KBS and MBC essentially functioned as a mouthpiece of the government by explicitly supporting Park during the 1971 presidential elections and pro- moting the Yushin system in 1972. All broadcasting companies each set up a separate di- vision to provide reports and information related to the New Village Movement (Saemaulˇ Undong)—Park’s key political initiative to modernize rural South Korean economy during the Yushin period. KBS was at the forefront with having thirty radio programs and eigh- teen television programs (approximately three to ﬁve programs per day) broadcasted on the movement (Kim 2007, 476).

3.4 Research Design

We ﬁrst conduct a content analysis of newspaper articles in the run-up to the presidential elections in 1963 and 1967 to systematically examine the changes in newspaper coverage and stances towards the authoritarian incumbent, Park Chung Hee. A content analysis of newspapers allows us to observe the changes in the coverage and information that voters receive regarding the presidential candidates and their campaigns, which would in turn impact their voting decisions. We expect to observe more biased media coverage on Park versus opposition candidates in 1967. More speciﬁcally, we present the following two

14The number 5.16 refers to the May 16, 1961 coup by Park Chung Hee. The foundation was later renamed as Jeongsu Foundation, after Park Chung Hee and his wife Yuk Young Su. In July 2005, the National Intelligence Service’s truth committee announced, “The Park Chung Hee regime forced the donation [of the foundation] through the Central Intelligence Agency.” The Jeongsu Scholarship Foundation still has 100 percent stake in the Pusan Daily and 30 percent stake in MBC as of today. 71 hypotheses:

1. The proportion of newspaper articles covering the opposition candidates decreases in the election year following tighter media control.

2. The proportion of newspaper articles criticizing the authoritarian incumbent decreases in the election year following tighter media control.

Subsequently, we explore whether exposure to newspaper affected Park’s vote share before and after the government’s tightened control of the press in the mid- to late-1960s. Given that the four national newspapers—Choson,ˇ Kyonghyangˇ Taehan, and Tonga—opposed Park’s plan to extend military rule in 1963 and the Media Ethics Law in 1964, we expect districts with higher circulation of these four newspapers to be less supportive of Park in the 1963 election. However, we expect this negative correlation between newspaper circulation and regime support to no longer hold in 1967 due to the fact that (1) the government directly repressed or co-opted these newspaper agencies starting with Kyonghyangˇ in 1966 and (2) the press media did not join, but rather criticized, the opposition party’s campaign for free press during the 1967 election. With this in mind, we hypothesize that the relationship between newspaper circulation and incumbent vote share depends on the government’s control over the media. We speciﬁcally develop the following three hypotheses:

3. When the government control of the media is limited, the incumbent vote share is lower in counties with higher newspaper circulation.

4.1 When the government exercises greater control over the media, the incumbent vote share is higher in counties with higher newspaper circulation.

4.2 When the government exercises greater control over the media, the incumbent vote share is not associated with newspaper circulation. 72 Lastly, we investigate the effect of access to radio on support for Park. Given accounts that (1) the broadcast media, unlike the press media, was new and inherently tied to economic development since its inception in the 1960s and (2) the broadcast media served as a mouthpiece for the government increasingly toward the installment of the Yushin Consti- tution in 1972, we expect districts with more access to radio to be more supportive of Park in the elections following government’s tightened control of the media (i.e., 1967 and 1971 presidential elections). We, therefore, present the following hypothesis:

5. The incumbent vote share is higher in counties with more access to radio.

In our regression analyses we consider various county-level characteristics as control variables that may affect the relationship between media exposure and voting behavior. These include population size, land size, population density, percent male, percent population aged 60 and older, percent eligible voters, and percent married.

3.4.1 Data

Newspaper Content.Using the Naver Digital News Archive,15 we examine all newspaper articles from Kyonghyangˇ and Tonga from two weeks prior to the Election Day until a day before the Election Day for each election. For both elections, the date of the polls was announced only two weeks prior to the election itself and print coverage of the elections was heaviest during this time period. The articles provide information on the elections, candidates, campaign rallies, and summaries of campaign speeches made by the candidates. The resulting sample is 412 articles. For each newspaper article, we coded (1) whether the content of the article has a negative or positive implication for Park Chung Hee and (2) the source of that information (i.e., the incumbent or the opposition party). There were several topics covered in the newspapers that had negative implication for Park. For example,

15Naver Digital News Archive, URL:www.newslibrary.naver.com 73 in 1963 the opposition candidates demanded Park to resign from the race by questioning the legality of Park joining the Democratic Republican Party and running as a presidential candidate while being the Chief Head of the Supreme Council for National Reconstruction. They also raised doubts about Park’s anti-communist ideology based on rumors regarding his past involvements with the South Korea Workers Party (a communist party in South Ko- rea that existed from 1946 to 1949) and the Yosu-Sun’chˇ onˇ Rebellion, a leftist movement against the South Korean government in 1948. Furthermore, the newspapers reported the opposition party’s claim that Park’s Democratic Republican Party was created with funds provided by a North Korean spy. Similarly, in 1967, the opposition attacked Park for an alleged intervention in election by involving government employees in campaigning and vote buying.

Newspaper Circulation. The data on newspaper circulation in 1963 and 1969 come from two volumes of the National Survey of Newspaper Circulation [Chon’gukˇ sinmun bogupˇ silt’ae] published by the Ministry of Communication. Using these publications, we construct the circulation data of four major newspapers—Choson,ˇ Kyonghyangˇ Taehan, and Tonga—at the county level. We linearly interpolate the circulation ﬁgures between 1963 and 1969 to get an estimate for the circulation in 1967.16

Radio Signal. To estimate radio signal strength, we collect data on radio transmitters in South Korea using publications by the Korea Broadcasting Business Association [Hanguk Bangsong Saophoe] (1969, 1973) containing information on location, frequency, call letter, power, and construction date of all radio transmitters in South Korea. We ﬁrst identify the locations of radio transmitters in 1969 and 1971. We then calculate the signal strengths

16The results remain similar when we use the 1969 circulation ﬁgures as a proxy for the year 1967.

74 from all radio transmitters treating the geographic center of the election district as the receiving location and take the maximum of the calculated signal strengths as our measure of radio access of the district. The spatial boundaries of the election districts are created based on the geographic information system (GIS) data that contain administrative districts of South Korea in 2012. We compared administrative districts in the GIS files and the election district maps for the 1967 and 1971 elections. The administrative districts tend to be finer than the election districts—i.e., many election districts include several administrative districts. Therefore, we are able to combine administrative districts in each election district to generate the boundaries of the election districts. To calculate the predicted radio signal strengths, we use the Irregular Terrain Model (Hufford 2002), which has been used by previous scholars (e.g., Olken 2009; Enikolopov, Petrova, and Zhuravskaya 2011; DellaVigna et al. 2013; Adena et al. 2013). Figure 3.1 displays the variation in radio signal strength in 1967 and 1971. To check whether our measure of radio accessibility is correlated with radio ownership, we collect data on the number of radio sets in South Korea at the county level in 1969 using the National Distribution of Radio Receivers [Chongukˇ sushin’gi bogupˇ shilt’ae] published by the Ministry of Culture and Information [Munhwa Gongbobu] (1969). The correlation between radio signal strength and the number of radio sets per thousand in 1969 is .67.17 Figure 3.2 shows the relationship nonparametrically. The figure indicates that radio listenership, measured as the number of radio sets per thousand, is positively correlated with the predicted radio signal strengths, especially when the signal is above a certain threshold.

17We aggregate the number of radio sets at the district level to calculate the correlation.

75 (a) 1967

(b) 1971

Figure 3.1: Variation in Radio Signal Strength

76 200 150 100 Radio Sets per Radio Thousand 50 0 -2 -1 0 1 2 Radio Signal (All) bandwidth = .8

Figure 3.2: Radio Signal and Radio Sets in 1967

Election and Census Data. The election data is from an online elections data archive compiled by the National Election Commission (NEC).18 We construct county level presidential elections data for 1963, 1967, and 1971. The data on demographic variables are from Statistics Korea.19 We construct the following variables at the county level: population, percent male, percent population aged 60 and older, percent eligible voters, and percent married. Using GIS data, we also measure the land size (km2) as well as population per square kilometer of each election district.20 For the analysis of the effect of radio, we aggregate the data at the election district level.

18The data is collected from the following URL: http://info.nec.go.kr/. 19The data is available at the following URL: http://kostat.go.kr/portal/korea/index.action. 20County-level geographic information is not available for the time period of this study.

77 3.5 Results

3.5.1 Newspaper Content

Through a content analysis of 412 articles from the 1963 and 1967 presidential elections, we ﬁnd that the government’s tighter control of the media in the late 1960s is associated with media coverage favoring Park. Both newspapers covered accusations made by the opposition in 1963 prior to the election, but in 1967 the number of such articles decreased signiﬁcantly along with the number of articles with negative implications for Park. Figure 3.3 shows that for Kyonghyangˇ the number of negative articles decreased from 54 (47.8 percent of the total articles) in 1963 to 15 (15.4 percent) in 1967 and the articles citing the opposition party decreased from 31 (27.4 percent) to 11 (10.6 percent) as well. As for Tonga, negative articles decreased from 81 (60.9 percent) in 1963 to 28 (45.2 percent) in 1967. The articles citing the opposition party also decreased from 46 (34.6 percent) to 14 (22.6 percent). These results show that there was a noticeable change in the coverage and content of newspaper articles between 1963 and 1967 as state censorship increased in the late 1960s.

3.5.2 Newspaper Circulation

Figure 3.4 shows the relationship between circulations (per thousand) of the four major newspapers—Kyonghyang,ˇ Choson,ˇ Tonga, and Taehan—and the incumbent vote share. We ﬁnd that newspaper circulations are negatively correlated with the incumbent vote share in 1963, when state control of the press was weaker. This relationship, however, disappears in 1967 as the government tightened its control over the media. Consistent with the results in the previous section, Figure 3.4 suggests that government control over the media undermined the watchdog role of newspapers. 78 Figure 3.3: Newspaper Coverage of Presidential Elections in 1963 and 1967

Figure 3.4: Newspaper Circulation and Vote for Presidential Party, 1963 and 1967

79 To test this idea more speciﬁcally, we estimate regressions of the following form:

0 IncVoteit = α0 + β1Newspaperit + β2Y1967t + β3Newspaperit ×Y1967t + δ Xit + εit,

where i and t index county and year, respectively. IncVote is the incumbent vote share, Newspaper is newspaper circulation for thousand, Y1967 is a dummy variable for the

year 1967, X is a set of control variables, and ε is an error term. Negative β1 suggests that the incumbent vote share is smaller in areas with high newspaper circulation. More importantly, positive β3 implies that this relationship between newspaper circulation and the vote share becomes weaker in 1967. Table 3.1 provides the results.21 The coefficients of newspaper circulations are all negative and statistically significant, which implies that the incumbent, Park Chung Hee, received less votes in counties with higher newspaper circulations. The estimates suggest that one additional copy per thousand of Kyonghyang,ˇ Choson,ˇ Tonga, and Taehan is associated with .69, .38, .15, and .37 percent fewer votes for the incumbent, respectively. As predicted, the coefficient for the interaction term is negative and statistically significant, which suggests that the negative relationship between newspaper circulations and incumbent vote share disappears in 1967.

21For summary statistics, see Appendix Table D.1. 80 Table 3.1: Newspaper Circulation and Vote for Ruling Party

Dependent Var. = Incumbent Vote Share Newspaper: Kyong-ˇ Chosonˇ Tonga Taehan hyang (1) (2) (3) (4) Circulation per Thousand -0.685∗∗∗ -0.383∗∗∗ -0.149∗∗ -0.365∗∗∗ (0.119) (0.078) (0.047) (0.064)

Circulation per Thousand × Year 1967 0.896∗∗∗ 0.409∗∗∗ 0.251∗∗∗ 0.408∗∗∗ (0.116) (0.074) (0.035) (0.057) Observations 367 367 367 367

Robust standard errors in parentheses. A dummy variable for the year 1967 is included. Control variables are: population, population per km2, percent male, percent aged 60 and older, percent married, percent eligible voters, and the land size (km2). All the variables except population per square km and the land size are measured at the county level. Population per square km and the land size are measured at the district level. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001.

81 Table 3.2: Newspaper Circulation and Vote for Ruling Party

Dependent Var. = Incumbent Vote Share Newspaper: Kyonghyangˇ Chosonˇ Tonga Taehan 1963 1967 1963 1967 1963 1967 1963 1967 (1) (2) (3) (4) (5) (6) (7) (8) Circulation per Thousand -0.497∗∗∗ 0.039 -0.283∗∗∗ -0.077 -0.126∗ 0.090∗ -0.297∗∗∗ -0.019 (0.130) (0.144) (0.058) (0.073) (0.054) (0.042) (0.067) (0.084) Observations 178 189 178 189 178 189 178 189

Robust standard errors in parentheses. Control variables are: population, population per km2, percent male, percent aged 60 and older, percent married, percent eligible voters, and the land size (km2). All the variables except population per square

82 km and the land size are measured at the county level. Population per square km and the land size are measured at the district level. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. In Table 3.2, we estimate the relationship between newspaper circulations and the incumbent vote share separately for the years 1963 and 1967. The results confirm our previous findings: the negative relationship between newspapers circulation and the incumbent vote share disappears in 1967. The Park regime’s tighter control of the media in 1967 can cause a change in consump- tion patterns of newspapers among voters, which can bias our results. For instance, if the newspaper circulation increased in areas where the regime was successful in silencing the opposition, our estimates could be upward biased. To empirically address this concern, we repeat the analyses fixing the circulation at the 1963 level, before the government changed its tactics toward the press. The results, reported in Appendix Table D.3 and Appendix Table D.4, remain similar.

3.5.3 Radio

We now show the relationship between regime support and exposure to radio, a media outlet that was more vulnerable than the press to state control. We regress the incumbent vote share on radio signal strength22 with the following form:

0 IncVoteit = α0 + θt + β1Radioit + β2IncVotet−1 + δ Xit + εit,

where i and t index district and year, respectively. Inc Vote is the incumbent vote share, θt is a year ﬁxed effect, which control for time-varying nation-wide shocks. Radio is radio

signal strength, X is a set of control variables, and ε is an error term. We include a lagged vote share to account for the possibility that radio signal strength is positively correlated

with the support for the authoritarian regime. Positive β1 would support our hypothesis.

22Due to data limitation, we restrict our analysis to the years 1967 and 1971. Summary statistics are provided in Appendix Table D.2.

83 Table 3.3: Determinants of Radio Signal Strength, 1967 and 1971

Dependent Var. = Radio Signal (1) (2) Lagged Vote 0.009 (0.005)

Population (Thousand) 0.000 0.000 (0.001) (0.001)

Population (Thousand) per km2 0.022 0.023 (0.006) (0.007)

Percent Male 0.205 0.244 (0.098) (0.103)

Percent Aged 60 and Older -0.288 -0.299 (0.050) (0.049)

Percent Married -0.047 -0.054 (0.043) (0.043)

Percent Eligible Voters 0.010 0.011 (0.007) (0.007)

Area (km2) -0.104 -0.100 (0.045) (0.046) Observations 258 258 R2 0.341 0.349

Bold = Signiﬁcant at .05 level. Robust standard errors in parentheses. Year ﬁxed effect included. Radio signal variables are standardized. All the variables are measured at the district level.

84 Table 3.3 shows the determinants of radio signal strengths. The results in columns (1) and (2) show that radio signal is positively correlated with population density and percent male and negatively correlated with percent aged 60 and older and the land size, suggesting that radio reception was higher in urban and industrialized areas. More importantly, radio signal strength is not correlated with lagged incumbent vote share, as shown in column (2). This result is assuring because it suggests that radio reception was not likely to be affected by previous electoral support for the incumbent. The results of the main analyses are shown in Table 3.4. The estimate in column (1) suggests that one-standard deviation increase in radio signal strength is associated with 2.6 percent increase in the incumbent vote share. The results remain similar when we include the lagged incumbent vote share, as column (2) shows. Overall, our results are consistent with our expectation that having more access to government-controlled media is positively associated with electoral support for the authoritarian incumbent.

Table 3.4: Radio Signal Strength and Vote for Ruling Party, 1967 and 1971

(1) (2) Lagged Vote 0.743∗∗∗ (0.055)

Radio Signal (KBS)

Radio Signal (All) 2.615∗ 1.582∗ (1.036) (0.785)

Controls Yes Yes Observations 258 258 Robust standard errors in parentheses. Signal strength measures are standardized. Year ﬁxed effect included.

85 3.6 Conclusion

In this paper, we explored whether and how mass media affects regime support in competitive authoritarian regimes. Modernization theory and the democratization literature posit a positive relationship between development in mass media and political liberalization. Indeed, studies on media and elections in advanced and consolidating democracies show that media has a liberalizing effect. Media plays a watchdog role in elections by providing information on issues and events related to politicians. However, despite the theoretical expectation in the democratization literature that media can undermine authoritarian rule, scholars and policymakers have also been pessimistic as to whether media can truly be independent from state control and play a watchdog role in authoritarian regimes. This study examined the effect of media exposure on regime support in South Korea during Park Chung Hee’s pre-Yushin regime (1961-1972). This period is classified as competitive authoritarian and the government exercised varying degrees of control over media. Using subnational level data on media exposure and regime support, we find greater access to media was correlated with more opposition to the authoritarian incumbent but only when the government’s control of the media was weaker. When state control of the media was stronger, the relationship between media exposure and electoral support disappears. Through a content analysis of newspaper articles, we demonstrate that pro-incumbency bias in news coverage resulting from tighter media control explains this distinctive pattern in the relationship between media exposure and electoral support. More generally, our work advances the growing literature on competitive authoritarianism by demonstrating that government-controlled media has ambiguous effects in competitive authoritarian regimes. Unlike the predictions of modernization theory, mass media can have divergent effects that may or may not promote political liberalization because media is neither completely free nor fully controlled in hybrid regimes. Our findings suggest that 86 in order to better understand the political effects of mass media in competitive authoritarian regimes, more attention should be paid to the conditional effects of media exposure depending on the timing and extent of media control. Lastly, we also add to the discussion of repression and mobilization in authoritarian regimes. We have a much better understanding about the effectiveness of mobilizing support through patronage and political institutions than about the effect of state repression (e.g., Brownlee 2007; Magaloni 2006; Gandhi 2008; Blaydes 2011; Schedler 2002). Dic- tators can use repression to eliminate threat to their rule, but violent repression can further decrease the political legitimacy of their regimes (e.g., Gurr 1970; Opp and Roehl 1990). Our results suggest that media control—as a form of repression—can also silence opposition, but the subversive power of mass media cannot be ignored either. Future research can potentially address the conditions under which media control creates dissent among citizens and even potentially facilitates mobilization against authoritarian rule.

87 4| Authoritarian Elections and Anti-regime Protests

4.1 Introduction

One of the most enduring debates within the comparative politics literature on regime durability concerns whether multiparty elections stabilize authoritarian regimes or contribute to their democratization. Some scholars argue that authoritarian rulers bolster their regimes by holding elections, because they can buy off supporters and co-opt opposition groups. Indeed, autocrats often hold multiparty elections while still engaging in serious democratic abuses, which creates an uneven playing ﬁeld and helps to consolidate their power. On the other hand, other scholars posit that authoritarian elections, regardless of their freeness and fairness, can still facilitate democratization by instigating intra-regime divisions, coalescing opposition groups, and fostering democratic values and practices. Why do elections seem to have a democratizing impact in some authoritarian regimes and the opposite effect in others? This study argues that these divergent effects stem largely from the structure of opposition forces. In contexts that lack mobilizing structures, elections dampen opposition activities, reducing the number of anti-regime protests. Here, elections appear to work as a “safety valve” reducing pressure on the regime. By contrast, elections provide a focal point for street mobilization and increase the likelihood of observing anti-regime protest in contexts with pre-existing mobilizing structures. In short, the impact of multiparty elections on

88 mass protest is moderated by the mobilizing structure of the anti-regime opposition groups. To evaluate this argument, I analyze within-case variation in South Korea’s last authoritarian regime (1980-1988). Existing studies tend to characterize democratization in South Korea as a consequence of active social movements led by a strong civil society,1 largely overlooking the role of elections in deactivating or spurring these mass demonstrations in the ﬁrst place. As in many hybrid regimes, authoritarian leaders in South Korea held un- fair executive and legislative elections,2 engaged in electoral fraud,3 and controlled and funded many “opposition” parties. Yet direct legislative elections, held fairly regularly, were seemingly competitive in the sense that the incumbent party never won an outright majority. Not only were these elections relatively competitive, but they also entailed substantial public participation and formed a key element of opposition’s strategy to challenge the authoritarian regime in the late 1980s. For these reasons, the role of elections in South Korea’s democratization merits closer examination. Using an original dataset of almost 1,200 protest events during South Korea’s 1987 “June Democratic Uprising,” this study examines whether having a “genuine” opposition party (one not created or funded by the regime) on the ballot in the 1985 legislative elections impacted subsequent local participation in mass anti-regime protests. It shows that voting for the opposition party had a dampening effect on anti-government protest. However, the magnitude of the effect decreases in areas with a higher density of activists. This robust ﬁnding demonstrates that elections provided both a safety valve for citizens to let off steam and a focal point for activists to further mobilize, conditional on the underlying movement

1E.g., see Choi (1993, 2002); Kim (2000, 2003). 2Direct presidential elections were held under South Korean presidents Rhee Syngman (1953-1960) and Park Chung Hee (1961-1971), but they were halted from 1972 until democratization in 1987. National legislative elections were held every three to four years, but only two-thirds of the seats in the National Assembly were up for contestation from 1972 to 1979. 3The fourth presidential election in 1960 was a fraudulent election, which sparked the April 19 student revolution that brought down the Rhee Syngman regime.

89 structure of the opposition in a given district. This study contributes to the ﬂourishing literature on authoritarianism by providing a new perspective on the relationship between authoritarian elections and regime durability at the subnational level. By moving away from investigating the “average effect” of holding elections on authoritarian durability or democratization at the national level, this more ﬁne-grained analysis at the subnational level shows that elections may differentially impact geographic areas or social segments within autocratic societies. Rather than view- ing elections as a single “silver bullet” for authoritarianism, this study implies that future studies should pay more attention to their conditional effects and the structural basis for anti-regime mobilization. The article proceeds as follows. Section 4.2 situates this article within ongoing debates in the electoral authoritarianism and social movements literatures and introduces a theory of elections, mobilizing structures, and protest to explain the relationship between authoritarian elections and protest at the subnational level. Section 4.3 provides contextual information on Chun Doo Hwan’s regime (1980-1988) and Section 4.4 describes the empirical strategy and data used to test the theoretical argument and evaluate competing explanations. Section 4.5 reports and discusses the results and Section 4.6 concludes.

4.2 Theorizing Elections, Mobilizing Structures, and Protest

Two central debates within the literatures on electoral authoritarianism and social movements concern the effects of elections. The ﬁrst debates whether political elections have a stabilizing or destabilizing effect on authoritarian regimes. The second relates to whether the political opportunities arising from elections have a mobilizing or deactivating effect on social protest. Studies of electoral authoritarian regimes provide mixed evidence on whether elections

90 help uphold authoritarian regimes or facilitate democratization.4 Many scholars demonstrate that authoritarian leaders strategically use political institutions to co-opt the opposition to ensure regime survival (Bueno de Mesquita et al. 2004; Simpser 2005; Magaloni 2006, 2008; Brownlee 2007; Greene 2007; Gandhi and Przeworski 2007; Gandhi 2008; Slater 2010; Blaydes 2011; Boix and Svolik 2013). Autocrats can use elections to prevent or solve intra-elite conflict (by spreading the spoils of office to domestic political elite) and as a focal point for patronage distribution to the citizenry (through vote buying) (Ged- des 2005; Lust-Okar 2006; Magaloni 2006; Blaydes 2011). Additionally, by holding and winning elections by huge margins, autocrats signal regime stability while simultaneously obtaining information about supporters and opponents of the regime (Simpser 2005). In contrast, other scholars find that elections can undermine authoritarian regimes instead (Bunce and Wolchik 2006, 2011; Hadenius and Teorell 2007; Lindberg 2006a,b; Kuntz and Thompson 2009; Tucker 2007). Opposition leaders and citizens often use elections and protest surrounding them to challenge authoritarian incumbents (Beissinger 2002; Bunce and Wolchik 2011). Regular elections, regardless of how free or fair, can also facilitate democratization (Hadenius and Teorell 2007; Pei 1995; Roessler and Howard 2009) by instilling democratic values (Lindberg 2006a,b). Moreover, electoral fraud committed by incumbents can backfire by sparking mass anti-regime protests (Eisenstadt 2007; Hafner- Burton, Hyde, and Jablonski 2014; Kuntz and Thompson 2009; Lehoucq and Molina 2002; Schedler 2006; Tucker 2007). The literature is similarly divided over the role of elections, but in relation to their impact on social movements. According to McAdam and Tarrow (2010), elections can serve as a political opportunity that can, in turn, empower social movements. Elections increase the costs of repression as well as international audience costs for incumbents (Blaydes

4For a review of the literature on elections under authoritarianism, see Gandhi and Lust-Okar (2009).

91 2011; Wright 2011). By increasing the amount of domestic and international media attention, election cycles allow social movements to publicize their grievances and demands (Bunce and Wolchik 2011). They also provide opportunities for social movement groups to partner with new institutional allies (such as opposition parties) in order to advance their claims (Trejo 2014). In turn, opposition parties—collaborating with civil society groups during campaigns—can portray themselves as a mass movement, signaling their capacity to engage in post-election mobilization when necessary (Bunce and Wolchik 2011). However, other studies suggest that the political opportunities provided by elections actually demobilize social movement through the organization-building process (Piven and Cloward 1979) or by reducing citizens’ popular involvement in politics (Blee and Currier 2006; Hirschman 1979). Elections serve as a “safety-valve” that regulates social discontent by allowing citizens to privately and safely address their grievances with the regime at the ballot box (Brownlee 2012; Buehler 2013; Lust-Okar and Jamal 2002). By providing nonviolent alternative channels for opposition groups to release their political energy, autocratic regimes use elections as a vehicle to direct, contain, and weaken discontent (Buehler 2013). While these debates consider the effects of elections on two different outcomes, together the literature suggests that one way in which elections can impact authoritarian durability is through mass social protest. Yet these studies collectively offer conflicting evidence and empirical implications regarding the relationship between elections and protest due to their focus on investigating the “average effect” of elections. In doing so, they overlook the possibility of elections having different effects within autocratic societies across social segments or geographic areas. In this study, I provide one explanation for why studies on the relationship between elections and protest in authoritarian regimes generate such conflicting claims and evidence. I argue that the tradeoff between voting and protesting posed by the introduction of 92 multiparty elections in authoritarian regimes is moderated by the mobilizing structures of the opposition forces. The argument proceeds in three steps. First, the introduction of multiparty elections creates an opening in the political opportunity structure within an authoritarian regime. Participation in a protest is incredibly costly for citizens in authoritarian regimes given low chances of success and severe state repression. Multiparty elections provide citizens with a state-sanctioned method to address their grievances by casting votes for opposition candidates. Rallies and campaigns during election cycles also provide opposition party candidates with opportunities to mobilize voters. The mere availability of an alternative option, in the form of voting for an opposition candidate and exposure to the opposition’s mobilization efforts, alters an individual’s calculus regarding whether to engage in protest. Although an individual can participate in elections in addition to subsequent anti-regime protests, voting for an opposition candidate can also function as a safer means to voice discontent that substitutes for participating in anti-regime protest. Second, the extent to which voting for the opposition can substitute for protest depends on people’s access to pre-existing mobilizing structures of the anti-regime opposition forces.5 Mobilizing structures, defined as “collective vehicles, informal as well as formal, through which people mobilize and engage in collective action,” help activists to get the movement off the group (McAdam, McCarthy, and Zald 1996).6 They serve as a building block for dense social relations to develop and provide solidarity, leadership, communication network, movement recruitment, and collective action frames (Clemens 1996; McAdam 1982; Snow, Louis A. Zurcher, and Ekland-Olson 1980). The informal and grass-

5On the link between electoral opportunities and mobilizing capacities of opposition forces, see Eisenstadt (2007); Robertson (2011); Trejo (2014). 6Examples of mobilizing structures include kinship and friendship networks, informal networks among activists, movement communities as well as more formal organizations (e.g., colleges and churches) that are not aimed primarily at movement mobilization, but where mobilization may be generated.

93 roots aspects of mobilizing structures are particularly crucial for movement emergence in authoritarian societies where people lack the ability and freedom to establish and utilize formal organizations to advance their interests. Lastly, mobilizing structures and framing processes help actualize objective opportunities (arising from elections) as de facto opportunities for mobilization. According to the social movements literature, collective action frames, derived from perceived opportunities, serve as action-oriented sets of beliefs and meanings that inspire and legitimate the activities of social movement groups (Benford and Snow 2000; Gamson and Meyer 1996; Klan- dermans 1997; Kurzman 1996; Voss 1993). Political openings remain as potential, rather than actual, opportunities unless and until they are perceived and defined as such by a group of actors sufficiently well organized to leverage these openings (McAdam, McCarthy, and Zald 1996; Suh 2001). Elections provide opportunities for opposition candidates and social movement forces to mobilize for a common goal—electoral victory. Further, they generate an outcome—electoral performance—that shapes the “cognitive opportunity” of movement participants and the general public by providing information on the likelihood of successful collective action (Choe and Kim 2012). In sum, the mobilizing effects of elections on protest are moderated by the mobilizing structures in place, especially under authoritarian regimes where formal (non-corporatist) organizations are either nonexistent or difficult to observe. The empirical implication is that the proactive electoral mobilization generated during election cycles is linked to further street mobilization only when the mobilizing structure is there to be leveraged by anti- regime activists. In the absence of, or with limited access to, mobilizing structures, citizens are likely to be demobilized because there is the “safer” electoral option to address their grievances toward the regime.

94 4.3 Context

I examine this theory of elections, mobilizing structures, and protest in the context of South Korea under Chun Doo Hwan’s authoritarian rule (1980-1988). In this section, I provide contextual information on the introduction of multiparty elections in 1985 and subsequent activities of the opposition forces leading to the June Democratic Uprising in 1987. Chun Doo Hwan came to power through a military coup after the assassination of his predecessor Park Chung Hee (1961-1979). Chun and his neo-military force formed the Democratic Justice Party (DJP; Minjongdangˇ ) as the ruling party and created the Demo- cratic Korea Party (DKP; Minhandang) and the Korean Nationalist Party (KNP; Kungmin- dang) as “opposition parties” that were directly controlled by the government. The newly installed constitution stipulated a single seven-year presidential term. However, instead of holding a direct election, an electoral college (T’ongil chuch’e gungmin hoeuiˇ )—essentially hand-picked by the president himself—selected the president. Severe state repression of political and civil society characterized the ﬁrst few years of Chun’s rule. Following a violent suppression of a pro-democracy movement in Kwangju city in May 1980,7 the regime carried out a series of large-scale coercive campaigns to “cleanse” both political and civil society, including the dissolution of the National Assem- bly. In its place, Chun created the Legislative Council for National Security (LCNS), a legislature pro tempore. The LCNS banned anyone “responsible for causing political and social corruption of fomenting confusion” from formal political activities and passed sev-

7From May 18 to 27, 1980, Chonnamˇ University students and citizens of Kwangju staged violent anti- regime protests, which were severely repressed by the Chun regime. According to The May 18 Memorial Foundation (https://518.org), 154 were killed, seventy-four were missing, and 4,141 were wounded (including those who died from their wounds) and placed under arrest by the martial law force.

95 eral bills the same year to regulate assembly, demonstration, and the press.8 The LCNS further prohibited third party intervention in local labor disputes, to prevent non-labor or professional activists such as student groups from mobilizing the working class. During this period of severe political repression, the 11th National Assembly elections were held on March 25, 1981 but no genuine opposition party was able to contest since they were banned from participating in politics. In 1983 Chun began to loosen gradually his tight control over society. Chun’s decision to liberalize was driven primarily by two factors exogenous to opposition demands. First, Chun was (over)conﬁdent in the stability and popularity of his regime and the ruling DJP due largely to the country’s strong economic performance and the LCNS’s previous regu- lations that succeeded in maintaining political and social order (Kim 2000, 82).9 Second, two international athletic games were to be hosted in South Korea—the Asian Games in 1986 and the Olympics in 1988—which incentivized Chun to introduce some degree of political contestation and participation in order to enhance the international standing of his regime. During this period of liberalization, Chun decided to introduce multiparty elections for the 12th National Assembly. This move represented a deliberate effort to create divisions among the opposition groups by allowing them to form their own parties (Chung, Kim, and Chung 2004, 56). In the wake of this political opening, formerly ousted opposition politicians collectively formed the New Korea Democratic Party (NKDP; Shinhan minju- dang) and participated in the 1985 legislative elections for the ﬁrst time under Chun’s rule.

8The Political Climate Renovation Law, the bills controlling assembly and demonstration, and the Basic Press Law (press censorship) were adopted on November 3rd, November 29th, and December 26th respectively. 9According to Han’guk kyongjeˇ uiˇ chuyo chip’yo [Major Statistics of Korean Economy] published by the Korea International Trade Association in 1990, the Gross National Product growth increased from -3.7 percent in 1980 to 12.6 percent in 1983 and the unemployment rate declined from 5.2 percent in 1980 to 4.1 percent in 1983.

96 Citizens also expressed great interest in multiparty elections. Compared to the legislative elections in 1981, the 1985 legislative elections experienced a 36 percent increase in rally attendance and an unprecedented voter turnout rate of almost 85 percent, the highest turnout since the beginning of authoritarian rule in 1961 (Chung, Kim, and Chung 2004, 58). Citizens were finally able to express their “frustrations” and “rebellious sentiments” through the act of voting (Yi 1987, 170-203, 185-186). Their electoral participation sym- bolized a “peaceful revolution” through the act of voting—instead of protesting—against the authoritarian regime (Choi 1989, 216). Despite DJP candidates’ considerable access to political machines and sizeable election campaign funds, the NKDP emerged as the leading opposition—winning 29 percent of the votes compared to 35 percent for the ruling DJP. Important to the electoral success of the NKDP was the support of civil society and activist groups. Student groups such as the Youth Coalition for Democracy Movement (Minch’ongnyˇ onˇ ) publicly announced that they would back the NKDP and actively campaigned for the NKDP as volunteers. Notably, this active participation by student groups represented the first time since the early 1960s that university students publicly supported a particular political party (Kim 2000, 85). The alliance formed between social movement groups and the opposition NKDP during the 1985 National Assembly elections endured and subsequently developed into a grand pro- democracy coalition against the authoritarian regime (Kim 2000, 86; Chung 2011, 157). Following the NKDP’s electoral success, students, civil society groups, and the NKDP members continued to engage in mass demonstrations against the regime. Inside the legislature, the NKDP actively demanded constitutional reform to include direct presidential elections. They also partnered with civil society groups to launch a signature collection campaign, make public statements and declarations, and organize mass rallies reprimand- ing the authoritarian regime and demanding constitutional reform. In response, Chun declared on April 13, 1987 that he would no longer tolerate “waste- 97 ful” discussion on constitutional reform, which sparked strong mobilization against the regime. A series of protests—known as the June Democratic Uprising of 1987—were staged starting April 13th and reached their pinnacle in June. Approximately one million people, including NKDP politicians, students, labor, and middle-class citizens, participated nationwide in anti-regime demonstrations during the June Democratic Uprising. On June 10th, approximately 400,000 people in twenty-two cities staged mass anti-regime demonstrations across the country, demanding an end to dictatorship and constitutional reform. Mass rallies aimed at repression and democratization were also organized on June 18th and 26th. Student associations such as Sodaehyˇ opˇ (the Association of Student Represen- tatives in Seoul), Puch’ongyopˇ (the Association of Student Representative in Pusan), and Daedaehyopˇ (the Association of Student Representatives in Taegu) were the main organizational force behind the national rallies on June 10th and 26th. Students comprised the major portion of the protestors (Chung 2011, 174) and staged street demonstrations almost every single day from June 10th to 26th (Soˇ 2011, 596). This intense struggle ended with the June 29 declaration, whereby the incumbent party announced formal reforms to the constitution, including adoption of a direct presidential election system as a key element of its eight-point democratization package, marking the beginning of South Korea’s democratic transition.10 The active involvement of students during the June Democratic Uprising had its roots in the historical tradition of student activism. Students were the key actors during critical moments in South Korea’s democracy movement, including the 1960 April Revolution that demonstrated “South Korea’s first massive and bottom-up expression of the desire and

10The incumbent party promised to (1) amend the constitution to provide for the direct election of the president, (2) revise the Presidential Election Law to ensure free and competitive elections, (3) grant amnesty to political prisoners including Kim Dae Jung, (4) respect human dignity and extend the right of habeas corpus, (5) abolish the Basic Press Law and promote the freedom of the press, (6) strengthen local and educational autonomy, (7) create a political climate conducive to dialogue and compromise, and (8) carry out social reforms to build a clean and honest society.

98 willingness to ﬁght for democracy” and the 1980 Kwangju Uprising that “radicalized” the democracy movement (Lee 2007, 26). In fact, they were “the only social group with the resources (time and knowledge) and pre-existent mobilizing structures (student councils and societies) that enabled them to challenge the power holders” (Park 2008, 140). The “nexus of senior and junior (sonbae-hubaeˇ ) ties,”11 circulation of texts (e.g., banned literature, pamphlets, underground newsletters, diaries of symbolic ﬁgures, and statement of appeal written by imprisoned students),12 and circles (ssok’ˇ ulˇ ; extracurricular clubs from the English word “circle”) and seminars (or reading groups) were vital in organizationally sustaining the student movement throughout the authoritarian period.13

4.4 Data & Methods

This section empirically examines the effect of expanding political opportunities in the 1985 National Assembly elections on people’s participation in anti-regime protests during the June Democratic Uprising. Table 4.1 introduces four hypotheses derived from the empirical implications of existing theories and the theoretical argument introduced in Section

4.2, as applied to the South Korean case. The ﬁrst (H1) and second (H2) hypotheses are derived from the electoral authoritarianism and social movements literatures and the third

11Students became members of underground circles through the introduction of their seniors (sonbaeˇ ). The underclassmen (hubae) also inherited organizational strategies from their seniors. See Lee 2007, 160 and Chang 2015b, 54. 12The circulation of underground texts became the main mode of communication for the student movement when open and public channels of communication became unavailable. The circulation of texts contributed to the ideological development and homogeneity within the movement (Lee 2007, 162-163). 13Circles were student-run extracurricular clubs that functioned as “a nursery for movement activities.” Circle activities such as Membership Training (MT) not only functioned as recruitment venues for student groups to identify potential recruits and train them into core activists but also politicized the issues of workers, farmers, and the urban poor through factory activities (gonghwal), countryside activity (nonghwal), night- study activity (yahak), and urban poor activity (binhwal). Circles also organized seminars (or reading groups) covering wide-range of topics in philosophy, political economy and social movements. See Lee (2007, 165- 166) and Park (2008, 123-126).

99 (H3a) and fourth (H3b) hypotheses test my argument on the conditional effect of elections on protest.

Table 4.1: Hypotheses on the Relationship between Elections and Protest

Less Access to High Access to Mobilizing Structures Mobilizing Structures

H3a: The negative effect of Elections as a H1: NKDP vote share is NKDP vote share on protest is “Safety Valve” negatively associated with weaker in regions with higher anti-regime protest access to mobilizing structures

Elections as a H3b: The positive effect of NKDP H2: NKDP vote share is “Focal Point” vote share on protest is weaker positively associated with for Mobilization in regions with less access to anti-regime protest mobilizing structures

I construct an original events dataset of 1,194 events that occurred during the 1987 June Democratic Uprising between the April 13 Statement (April 13, 1987) and the June 29 Declaration (June 29, 1987).14 Data on the June Democratic Uprising is drawn from a series of reports compiled by the Korea Democracy Foundation (KDF)—“the Reports on the History of South Korea’s Regional Democracy Movement” and “the KDF Dictionary of Events Related to the Democracy Movement.”15 I also use The Great June Uprising for Democratization published by the Christian Institute for the Study of Justice and Develop- ment (CISJD) to supplement data on the June Democratic Uprising. The dataset includes

14Coding manuals of other independent datasets on protest events were consulted, including the ones of Olzak and West (1995), Chang (2015b), and Lee (2010a). 15The Korea Democracy Foundation (KDF: Minjuhwa Undong Kinyomˇ Saophoeˇ ) was established in 2001 by the South Korean government in accordance with the Korea Democracy Foundation Act (Act 6495) to memorialize South Korea’s democracy movement. One of the primary objectives of the foundation is to collect, digitize, and manage various original documents related to the democracy movement including local/school newspapers, pamphlets, government documents, SMO organizational histories, personal memoirs and other primary documents. 100 all protest events during the June Democratic Uprising recorded as protest events in the CISJD and KDF publications. It also quantiﬁes various characteristics of each event.16 By using the CISJD and KDF publications, which are each compiled from a variety of primary sources, I address the data limitations of relying solely on protest events reported in the national newspapers given the high degree of press censorship under Chun’s rule.17 This dataset supplements existing datasets on South Korea’s democracy movement that are heavily biased towards events that occurred in the capital city, Seoul. To my knowledge, no other dataset contains comprehensive information on events that have occurred in other regions of the country. My outcome of interest is the protest “intensity” of the 1987 June Democratic Uprising in a given county. Using the events dataset, I construct my dependent variable—protest intensity—as the total number of protest events observed in each county from April 13, 1987 to June 29, 1987.18 Figure 4.1 displays all protest sites during the June Democratic Uprising and Table 4.2 reports the regional variation in the number of protest events at the provincial level, one administrative unit higher than my unit of analysis. Although this measure provides the most appropriate and available proxy of intensity given the nationwide prevalence of protest during this period, I also consider an alternative measure of protest intensity—number of days during the June Democratic Uprising in which protests

16Characteristics coded for each event in the dataset include date, location, issues raised by protestors, tactics used by protestors, main target(s), groups participating in the event (e.g., students, labor, opposition politicians, journalists, citizens (shimin), etc.), number of protestors, and type of repression used if the event was repressed. 17However, as noted by Shin and Chang (2011, 22), there are also several limitations to using the sourcebooks. First, there is a potential problem of omission. Events that were not found in any source were not included because there was no way for the foundation to collect information on them and report them in their sourcebooks. Second is the potential problem of data reliability. Given that the KDF and CISJD used any or all sources of information available, the quality and amount of information on protest events may vary systematically by region or book. 18There are three administrative divisions in South Korea: provincial-level (-do, tukpyˇ olˇ chach’i-do, kwangyok-siˇ , tukpyˇ ol-siˇ ), municipal-level (shi, gun, gu), and submunicipal-level (eup, myeon, dong) divisions. This study uses municipal-level divisions (i.e., counties) as the unit of analysis.

101 have occurred in a given county—as a robustness check in Section 4.5.

Figure 4.1: Protest Sites During the June Democratic Uprising

Source: Korea National Elections Commission, Statistics Korea, and author’s dataset

Table 4.2: Number of Districts and Events (Per 100,000 People) for Each Province

Province Number of Districts Events (Per 100,000 People) Seoul 17 1.74 Pusan 10 3.98 Taegu 6 5.71 Inch’onˇ 4 5.77 Kyonggiˇ 27 3.25 Kangwonˇ 21 3.52 Ch’ungbuk 13 2.01 Ch’ungnam 18 3.57 Chonbukˇ 18 4.90 Chonnamˇ 29 4.91 Kyongbukˇ 31 1.59 Kyongnamˇ 27 1.25 Cheju 4 2.87 Source: Author’s dataset

102 To empirically investigate how elections impacted protest, I separately consider the main effect of the NKDP Vote Share19 on protest intensity and the effect of the NKDP vote share conditional on the degree of access to mobilizing structures by including an interaction term—Proportion of College Students × NKDP Vote Share. To ease interpretation of the marginal effect of vote share in a nonlinear model, the NKDP Vote Share is coded as a dummy variable to distinguish counties with low versus high levels of electoral support for the opposition NKDP. Speciﬁcally, I code districts with values higher than the third quartile (32 percent) as high and those with values lower than 32 percent as low.20 I operationalize a district’s degree of access to mobilizing structures as the density of

college students within each district (Proportion of College Students). College student density represents a strong proxy given that students have greater access to mobilizing structures and are more likely to participate in protest events than other social groups. Similar to social movements in many parts of the world, college students in South Korea have been particularly active—and in fact the dominant social group—within the country’s democracy movement. Historically, their access to mobilizing structures has been sustained by senior-junior ties, text circulation, circles and seminars. Student associations such as Sodaehyˇ opˇ , Puchonghyˇ opˇ , and Daedaehyopˇ helped organize the two crucial national rallies and mobilized students and—even more critically—other social groups (including workers, urban poor, and the middle class) for street demonstrations during the June Democratic

19The election data is compiled by the Korea National Election Commission available at http://info. nec.go.kr/. To ensure that this measure is actually capturing electoral support for the opposition, I regress the 1985 NKDP vote share on the vote shares of the opposition candidates in the first democratic presidential elections in 1987. The results (reported in Table E.1 in the Appendix) show that there is a statistically significant and positive association between the two. 20I use the third quartile because it is the median of the upper half of the data. I also use a continuous measure and another dummy variable with 62.5th percentile cut-off point (the median of the first and third quartiles of the distribution; 24 percent) and find similar results. The results are found in Table E.2 in the Appendix.

103 Uprising. Arguably, students also represent the population most likely to protest given that, unlike other social groups, they are free from personal constraints imposed by full-time employment and family commitments that may otherwise deter one from participating in protest events (McAdam 1988; Snow, Louis A. Zurcher, and Ekland-Olson 1980). Given these strong theoretical, historical, and demographic expectations of observing

mobilizing structures in areas populated by college students, Proportion of College Students represents ﬁnest data available at the subnational level. Severe state control of dissident activities on and off-campus led many student activities to operate primarily underground until the mid-1980s. For this reason, more direct data on student group activity does not exist. While the Han’guk Taehak Yon’gamˇ [Korean University Yearbook] series does provide information on student associations at junior colleges and universities, there are no publications containing data for the critical years under examination: 1985-1987. Moreover, the available albeit limited and incomplete data on the number of student associations available for 198321 is strongly and positively correlated (r = 0.93) with the proxy used—student population (see Figure F.1 in the Appendix).22 This suggests that the use of density of college students represents a reliable proxy for mobilizing structures in a given county, as there did indeed exist greater availability of and access to mobilizing structures in areas with more college students. In the empirical analysis, I also control for the previous levels of support for the NKDP and electoral competitiveness23 in each electoral district. Unfortunately there is no direct

21Found in the 1984 Yearbook for 1983, this represents the latest data available prior to the 1987 June Democratic Uprising. There is variation in the quantity and quality of information for each college in the Yearbook and there is no way to determine the exhaustiveness of the data. 22An outlier (Korea Correspondence University in Seoul Chongno) was removed from the bivariate correlation. Even with the inclusion of the outlier, a strong positive correlation between the two variables remains (r = 0.87). 23I use total number of candidates instead of effective number of parties in each districts because more than one candidate from a party was able to contest in a given electoral district.

104 measure for previous levels of support because there was no genuine opposition party con- testing the 1981 legislative elections. Instead, I use a district’s abstention rate as a proxy for previous levels of support given that those who abstained in 1981 are more likely to be those dissatisﬁed with the incumbent party.24 I also include measures of industrialization, urbanization, and unemployment to control for the structural factors highlighted by the modernization theory (Lerner 1958; Lipset 1959; Rustow 1970) and grievance-based theories of social movement (Davies 1962; Gurr 1970; Muller 1985; Buechler 2004; Korn- hauser 1959) that may be associated with protest intensity. Other county-level characteris-

tics such as population size and prior levels of mobilization (Pusan & Masan, Kwangju, Inch’ˇon)25 are also included as control variables. I report summary statistics of all variables in Table E.3 in the Appendix. I use a zero-inﬂated negative binomial model given that my dependent variable is a non-negative count of protest events with excessive zeroes (i.e., counties with no protest events).26 Due to the nature of protest being a rare event, aggregation of events is often “inﬂated” with structural zeroes in datasets on subnational protest events.27 The structural zeroes represent cases that have zero probability of ever experiencing protest. Treating these cases as protest-free observations within a model can lead to biased inferences, but

24On voter turnout see Birch (2010) and Lijphart (1997). 25Dummy variables are included for counties that have experienced pro-democracy protests prior to the June Democratic Uprising. These region-specific pro-democracy cycles of protest are 1979 Pusan-Masan Uprising, 1980 Kwangju Uprising, and 1986 Inch’onˇ May 3rd Uprising. 26There were 76 nonzero districts among the 211 counties that were contested by the NKDP candidates. The negative binomial model is more appropriate than the poisson model since the mean and variance of my dependent variable are not equal to each other (σ = 12.83 > µ = 5.08). I also perform the Vuong test of a zero-inflated negative binomial model and a standard negative binomial model and find the zero-inflated negative binomial model to be an improvement over a standard negative binomial model (test statistic = 3.086 and p-value = 0.001). 27Datasets on inter- and intra-state conflicts and wars (e.g., Correlates of War datasets) are also “inflated” with structural zeroes and the zero-inflated count (negative binomial or poisson) models are often used in studies utilizing such datasets.

105 eliminating these potential structural zeroes from the sample not only excludes a signiﬁcant portion of relevant protest-observations but also produces selection bias (Lemke and Reed 2001; Clark and Regan 2003; Xiang 2010). In my case, the districts in which we observe zero counts could be due to the following: (1) there were attempts to organize protest events but they were either insigniﬁcant or unsuccessful to be recorded or (2) protests just never took place. Hence the expected count is expressed as a combination of these two processes:

E(nevents = k) = P(no mobilization) ∗ 0 + P(mobilization) ∗ E(y = k | mobilization)

4.5 Findings

The regression results from the zero-inflated negative binomial model are presented in Table 4.3. The lower half of the table reports the inflate equation of the model (i.e., the extent to which there are more zeroes in the data than implied by the negative binomial distribution).28 A negative estimate means that an increase in the explanatory variable increases the probability of observing at least one event in a location. The upper half reports the coefficients corresponding to the count (protest events) equation of the model. Given the difficulty of directly interpreting the results from a nonlinear regression model, I transform the coefficients obtained from the count portion of the model into incidence rate ratios (IRR) to facilitate substantive interpretation. Both Columns 1 and 2 include all control variables and province fixed effects. Col- umn 2 presents the results of a model specification that includes the interaction term—

Proportion of College Students × NKDP Vote Share. The inﬂate equation in Col- umn 2 shows that districts with more electoral competitiveness, population density, and

28Inch’ˇon,Kwangju, and Pusan & Masan are excluded in the inflate equation given that these regions have experienced region-specific movements in the past. These variables would not be useful in distinguishing non-zero districts from “always zero” districts. 106 industrialization had a higher probability of seeing at least one protest event. As expected, the protest events equation in Column 2 demonstrates that the main effect of mobilizing structures (College Students Per Capita) on number of protest events is statistically significant and positive. The IRR suggests that one unit increase in a district’s density of

college students increased protest by 74.1 percent. The main effect of NKDP Vote Share on protest intensity is also statistically signiﬁcant and negative. Holding all other variables constant, districts with high opposition party vote share (higher than 32 percent) exhibited fewer protests—in fact, protest was 96.3 percent lower in these districts. These ﬁndings

provide support for the ﬁrst hypothesis (H1), that availability of the option to vote for gen-

uine opposition candidates serves as a safety valve for citizens to let off steam (H1) rather than a focal point generating proactive electoral mobilization (H2). The coefficient of the interaction term, Proportion of College Students × NKDP Vote Share, is statistically significant, suggesting that there is a difference in the effect of NKDP Vote Share across different values of the moderating variable, Proportion of College Students. The coefficient is positive since the effect is increasing as density of college students increases, but the main effect is negative (indicated by the negative coefficient of NKDP Vote Share). To substantively interpret the interaction between NKDP vote share and density of college students (Proportion of College Students × NKDP Vote Share), I calculate the marginal effects of NKDP Vote Share on protest intensity (IRR) at different levels of mobilizing structures (Proportion of College Students).29

29I follow Hilbe (2011) to calculate the IRR and the standard errors of IRR and create the marginal effects plot visualizing the conditional effect of NKDP Vote Share on Protest at different levels of Proportion of College Students. IRRNKDPVoteShare×Proportiono fCollegeStudents = exp[-1.07670 + 0.54340*Proportion of College Students] where IRRNKDPVoteShare×Proportiono fCollegeStudents is the interaction of binary (NKDP Vote Share) and continuous (Proportion of College Students) predictors. The coefﬁcients of the NKDP Vote Share and the interaction term come from Table 2 column (2). The IRR standard errors for the interactions (at each level of Proportion of College Students) are determined by ﬁrst calculating the variance using the variance- 2 covariance matrix: VNKDPVoteShare×Proportiono fCollegeStudents = 0.7670987+ Proportion of College Students ∗ 0.03649927 + 2∗ Proportion of College Students ∗ − 0.1516218. Subsequently, I take the square root of the variance to obtain the standard error.

107 Table 4.3: Predictors of Protest Intensity during the 1987 June Democratic Struggle

(1) (2) Protest Events Equation (negative binomial with log link) Coefﬁcients IRR Coefﬁcients IRR

NKDP Vote Share × College Students Per Capita 0.543∗∗∗ 1.722 (0.191) NKDP Vote Share −1.077∗∗∗ 0.341 −3.301∗∗∗ 0.037 (0.384)(0.876) College Students Per Capita 0.667∗∗∗ 1.949 0.555∗∗∗ 1.741 (0.117)(0.105) Abstention (1981) 0.016 1.017 0.013 1.013 (0.024)(0.023) Electoral Competitiveness 0.028 1.028 0.019 1.019 (0.129)(0.125) (Log) Population 0.181 1.198 0.239 1.270 (0.266)(0.261) Population Density 0.764∗∗∗ 2.147 0.790∗∗∗ 2.203 (0.239)(0.243) Industrial 0.043∗∗ 1.044 0.046∗∗∗ 1.048 (0.018)(0.016) Unemployment 0.069∗∗ 1.071 0.072∗∗ 1.074 (0.030)(0.028) Inch’onˇ 2.074 7.957 2.623∗∗ 13.770 (1.262)(1.231) Kwangju −1.379 0.252 −2.538∗∗∗ 0.079 (0.921)(0.966) Pusan & Masan 0.173 1.189 0.246 1.279 (1.118)(1.072) Count Constant −8.233∗∗ −8.638∗∗∗ (3.311)(3.240) Province FE Yes Yes

Inflate Equation (binomial with log link) Coefficients Coefficients

NKDP Vote Share × College Students Per Capita 0.789 (1.316) NKDP Vote Share 4.569∗ 2.336 (2.620)(3.785) College Students Per Capita −0.956 −1.267 (0.841)(0.956) Abstention (1981) 0.162 0.156∗ (0.107)(0.093) Electoral Competitiveness −2.534∗ −2.449∗∗ (1.299)(1.205) (Log) Population −1.220 −0.994 (1.713)(1.723) Population Density −72.990∗∗ −72.668∗∗ (34.267)(31.621) Industrial −0.310∗∗ −0.331∗∗ (0.156)(0.160) Unemployment 0.948∗ 0.968∗∗ (0.513)(0.446) Zero Constant −5.089 −7.550 (23.927)(24.099) AIC 646.763 642.271 Log Likelihood -290.382 -286.135 Num. obs. 210 210 Nonzero obs. 76 76 ∗∗∗ p < 0.01, ∗∗ p < 0.05, ∗ p < 0.1. Standard errors in parentheses. IRR = incidence rate ratio. 108 Figure 4.2 plots the log-risk ratio (logged IRR) of NKDP Vote Share for ease of interpretation and visualization of the relationship. As the ﬁgure shows, the negative effect of NKDP vote share on protest intensity is signiﬁcant and stronger in districts with fewer mobilizing structures. The effect of NKDP Vote Share is negative for regions where there are smaller percentage of college students, but that effect dissipates for areas with large college student concentrations. This lends further support to H1 and provides evidence congruent with

H3a that the dampening effect of elections is reduced by the mobilizing structure of the opposition.

Figure 4.2: Relationship Between NKDP Vote Share and Protest Intensity at Different Degrees of Access to Mobilizing Structures with 95% Conﬁdence Intervals

As robustness checks, I address two potential concerns that one may have regarding the analysis and results above. The first concern (as already addressed in Section 4.4) relates to the conceptualization and measurement of protest intensity. The second derives from uncertainty about whether protest participants are protesting in their respective electoral districts. To address the first concern, I use total number of days in which protests are 109 observed in a given district as an alternative to the total number of events to assess whether a different measure of protest intensity affects my results. To address the second concern, I investigate the effect of elections on protest on a subset of observations that are more likely to include mass participation by citizens in their own electoral districts. This subset is limited to protest events on days that an umbrella organization–National Movement Headquarters for Democratic Constitution (NMHDC)30– organized national rallies because protest events on these days were more sporadic and involved large-scale citizen participation than events on days without NMHDC involvement.31 More importantly, citizens comprising the majority of the electorate in a given district are more likely to participate in mass national rallies in their respective districts given that they are more risk-averse and are less likely to travel and participate in protest events in distant counties. Conversely, smaller isolated events attracted activists who are more likely to have traveled outside of their districts to participate and represent a smaller proportion of the electorate. Appendix Table E.4 reports the results from the additional analyses conducted to test the robustness of my original results. Column 1 gives the original results as a baseline comparison. Column 2 shows the results with number of days as the dependent variable. Lastly, Column 3 provides the results from using number of events from a subset of observations on days that the NMHDC was involved. The statistical significance and direction of the coefficients in Columns 2 and 3 do not substantively differ from the original results in Column

30The National Movement Headquarters for Democratic Constitution (Kungmin Undong Bonbu) was an umbrella organization established on May 26, 1987. This organization consisted of 2,191 inaugural members from the People’s Movement Coalition for Democracy and Reunification (Mint’ongnyonˇ ) and twenty- five other social movement groups, covering all major sectoral groups (Protestant pastors, Buddhist monks, opposition politicians, women’s movement leaders, peasant activists, labor activists, urban poor activists, publishers and journalists, authors and writers, artists, educators, youth movement leaders, and lawyers) and geographical areas in South Korea. 31Four days in particular—June 10th, June 18th, June 21st, and June 26th—exhibited the highest number of mass events (i.e., events with more than 10,000 participants) in the dataset (see Figure F.2 in the Appendix), with the NMHDC involved in three of these four days (June 10th, June 18th, and June 26th). 110 1, suggesting that my findings still hold after addressing the two potential concerns.

4.6 Conclusion

Existing studies of authoritarian elections offer conﬂicting ﬁndings and suggest contradictory empirical expectations regarding the impact of elections on social movements and democratization. Elections may sometimes be a tool for authoritarian leaders to co- opt opposition groups while at other times they provide opportunities for the opposition to challenge the incumbent regime. Similarly, elections may encourage or deter protest participation by altering an individual’s calculus regarding whether or not to participate in a protest against the regime. Using new subnational-level data from South Korea, this study argues that the effect of elections is conditional on the mobilizing structure of opposition groups within a given geographic region. It suggests that elections can have both positive and negative effects on authoritarian regimes by simultaneously decreasing and increasing anti-regime protest in different geographic regions or social segments. As illustrated by the South Korean case, elections dampen protest in areas lacking mobilizing structures, but elections can also serve as a focal point for coordinated opposition against the regime in areas with high access to mobilizing structures. This study recasts the traditional understanding of South Korea’s democratization by highlighting the critical enabling factor—preexistent mobilizing structures—that negated the dampening effect of elections on protest during the democratic transition period. While previous studies largely overlook the role of elections in dictatorial South Korea by char- acterizing them as unstable political institutions corrupted by irregularities and fraud (e.g., Im 2010), this study shows that in the absence of mobilizing structures, elections may have further stabilized the authoritarian regime.

111 The theoretical argument and findings from this study expand our current understanding of elections and protest in autocracies in two ways. First, unlike explanations that only consider the main (or average) effects of holding elections, my argument accounts for the conditional effects of elections on protest. I find that the reciprocal relationship between elections and social protest does not hold when there are no mobilizing structures to help actors channel electoral mobilization into coordinated opposition against the regime. Sec- ond, unlike studies that emphasize election fraud as a catalyst for opposition mobilization, I show that post-electoral mobilization can emerge even in the absence of fraud. Indeed, exogenous factors such as major electoral fraud can help solve collection action problem in autocracies by altering individuals’ calculus regarding whether to participate in a protest. However, my evidence suggests that this calculus is shaped by individuals’ access to pre- existing mobilizing structures regardless of whether the election was stolen. In conclusion, this study suggests that future research on electoral authoritarianism and regime durability would benefit from examining the non-uniform impact that elections have within each regime. In addition, more attention should be paid to the conditional effects of elections and the structural basis for mobilization in examining and understanding the effects of authoritarian elections on social protest.

112 A| Appendix to Chapter 1: Protest Data and Coding Manual

A.1 Data

The data for this dataset comes from a series of sourcebooks titled, “the KDF Reports on the History of South Korea’s Regional Democracy Movement” published by the Korea Democracy Foundation (KDF) and 6-wol Minjuhwa Daetujaeng [The Great June Uprising for Democratization] published by the Christian Institute for the Study of Justice and Devel- opment (CISJD) (both in Korean). There are 11 volumes of the KDF sourcebooks for each subregion of the country: Ch’ungbuk, Taejonˇ and Ch’ungnam, Wonjuˇ and Ch’unch’on,ˇ T’aebaek and Ch’ongsˇ on,ˇ Inch’on,ˇ Kyonggi,ˇ Cheju, Chonbuk,ˇ Kwangju and Chonnam,ˇ Taegu and Kyongbuk,ˇ Pusan and Kyongnam.ˇ Due to KDF’s efforts to collect and docu- ment information related to South Korea’s democracy outside of the capital city (chibang), this particular set of sourcebooks do not include a separate events that occurred in the capital city, Seoul. The CISJD book as well as an earlier series of KDF sourcebooks titled, “the KDF Dictionary of Events Related to the Democracy Movement” provide data on events in Seoul. Coding manuals of other independent datasets on protest events were consulted, including the ones of Olzak and West (1995), Chang (2015b), and Lee (2010a).

113 A.2 Concept and Measure of Events

Events are commonly defined as nonroutine, collective, and public acts that involve claims on behalf of a larger collective (Paige 1975; Tilly 1978). A minimal definition of collective action is that it (a) involves more than one person, and (b) makes claims of agency (or corporate) status (Olzak 1989). Tilly (1978)’s list of claims includes petitioning, memorializing, and opposition to or support of an enemy of the government. Tilly (1978, 275) defines agency status as involving acts that “make a visible claim which, if realized, would affect the interests of some specific person(s) or group(s) outside their own number.” I code all reported protest events mentioned in the KDF and CISJD books. Below is a specific instructions regarding how to or how not to code an event:

• The unit of analysis is not a movement (i.e., not a protest cycle) but the individual event and it is possible that multiple events happen each day. These (multiple) individual events occurring each day are coded as separate events that may or may not be linked to other events.

• When the book reports several different forms of protest happening simultaneously in a locality, these events are coded as separate events because they involve separate groups.

• I do not code reports on future protest events that actually did not take place or failed to happen (due to repression or other reasons speciﬁed in the book).

• I do not code reports on people “gathered” (unjip) without any additional information is not considered an event on its own.

• If the book mentions a related event that happens in a different locality, I create a new row in the Excel spreadsheet for that related event (which gets a unique ID) and 114 ﬁll out as much information as I can about that event. Then I make sure to indicate that they are linked using the link variables in my events dataset.

A.3 Coding Variables

ID: a unique ID number for each event

TIMING VARIABLES

• STMT: month when the event began (e.g., June = 6)

• STDT: day in the month when the event began

• EDMT: month when event ended

• EDDT: day in the month when the event ended

• TOTDUR: an estimate of the duration of an event (if available). Most of the time it will just be “1.”

LOCATION VARIABLES

• SD_HCODE: a unique code assigned to each province (shi, do) from Statistics Ko- rea1

• SGG_HCODE: a unique code assigned to each county (shi, gun, gu) from Statistics Korea

PROTEST VARIABLES

1URL: http://kostat.go.kr

115 ISSUE1-5: issue(s) raised by the protesting group(s). There could be more than one (up to five issues) being raised. For example, if there are two issues raised, only ISSUE1 and ISSUE2 columns need to be filled out and the remaining columns (ISSUE3-ISSUE5) will be left blank. Code Issue 0 Not specified or unknown 1 Anti-government (panjongbuˇ ) 2 4.13 Statement (hohonch’ˇ olp’yeˇ ) 3 Constitutional revision, democratic constitution 4 Democratization (in general), struggle for democratization 5 Direct presidential election 6 (Military) dictatorship (tokchaet’ado, kunbudokchae, changgijipkwonˇ ) 7 Tear gas (ch’oerut’anch’ubang) 8 Democratization on campus (hagwonminjuhwaˇ ) 9 Repression (in general) and repression events 10 Memorial events (e.g., 4.19, Kwangju (5.18), funerals) 11 Dismissal of university presidents and/or professors 12 Release of those arrested/taken into custody 13 Anti-American 14 Anti-Japanese 15 Reunification with North Korea 16 Save the nation, salvation, justice, and peace (in Korea) 17 Media freedom/media-related (including boycotting KBS) 18 Labor rights 19 Wage-related issues 20 Movement (for solidarity, evaluation) 21 Current political situation (siguk) 22 6.26 Declaration 23 Freedom of association 24 Information collecting/fact finding (chinsanggyumyongˇ ) 25 Political prisoners (yangsimsu) 26 Housing issues (city planning) 27 Anti-poverty/inequality 28 Other economic/labor-related issues (reinstatement, vacations, company unions)

TACTICS1-6: tactic(s) used by the protesting group(s). There could be more than one (up to six tactics) being used. 116 Code Tactic Additional Notes/in Korean 0 Not specified or unknown 1 Rally (taehoe, chiphoe, ch’ulchongsikˇ ) stationary 2 Meeting not targeting the public; mostly internal 3 Demonstration (siwi) moves around 4 March 5 Hunger strike 6 Placards, pickets, flags, and posters including taejabo, pyokpoˇ 7 Disseminating leaflets, pamphlets, propaganda 8 Making declaration reading a report, statement, siguksonˇ onˇ 9 Candlelight vigilant 10 Sit-in (nongsong,ˇ yonjwanongsˇ ongˇ ) 11 Occupation (chomgˇ oˇ) 12 Throwing stones and/or firebombs hwayombyˇ ong,ˇ t’usokchˇ onˇ 13 Damaging properties/facilities 14 Setting building/objects on fire chonso,ˇ pangwa 15 Cultural event madangnori, p’ungmullori 16 Hurting or killing oneself suicide, self-immolation 17 Singing songs 18 Shaving one’s head 19 Writing with one’s blood 20 Honking, using cars 21 Memorial event 22 Prayer meeting 23 Religious service 24 Refusal/boycott class, work (paopˇ ) 25 Talk lecture, seminar, classes 26 Debates t’oronhoe 27 Signature collecting 28 Symbolic actions wearing ribbons, destroying statues, etc. 29 Silent protest 30 Publicizing events including broadcasting 31 Organization founding joining an organization 32 Hold someone as hostage 33 Gathering 34 Mock elections 35 Fire torch 36 daechee 37 Press conference 38 Submitting a petition

117 TARGET1-4: main target(s) of event (if known). There could be more than one (up to four targets) mentioned in the event. Code Tactic 0 No target (usually for soldiery events ) 1 Government/regime (in general) 2 Government speciﬁc (e.g., branch, institution) 3 Park Chung Hee, Chun Doo Hwan, Roh Tae Woo 4 Democratic Justice Party 5 Police 6 KBS/media 7 Employer/company 8 University/university bureaucracy/professors 9 Independent parties and/or organizations 10 Foreign (U.S., Japan)

GROUPS1-5: groups participating in the event. There could be more than one (up to give groups) participating.

Code Participants 0 Not speciﬁed or unknown 1 Student and youth 2 Labor/workers/merchants 3 Chaeya individuals and organizations 4 Opposition politicians 5 Intellectuals (includes professors) 6 Citizens (simin) 7 Lawyer 8 Journalists 9 Writers 10 Artists 11 Medical professionals (doctors, dentists, pharmacists) 12 Farmers and miners 13 Christian 14 Buddhists 15 Teachers (including lecturers) 16 Women’s SMO 17 Blue-collar workers 18 Family of prisoners SMO 19 Human rights SMO 20 Political prisoners (yangsimsu) 21 Family of 5.18 SMO 22 Economically poor (binmin) 23 Other

118 NUMB_PART: number of protestors (excluding the repressive forces or bystanders)

• If the book states around/approximately 500 people (500 yomyˇ ongˇ ) participated, code it as 500 (to be conservative/safe)

• If the book mentions, for example, 500 to 600 people, put down the average of the two numbers (550).

• Record the highest number of participants for a given event if the size of the event increases or decreases over time.

• If the book mentions two events that happened separately but provide only one number for the total of the two events, record one-half of the number of total participants for each event.

• Leave blank if unknown.

SIZE: size of event

Code Size Number of participants 0 Unknown 1 Small 2-9 2 Medium 10-49 3 Large 50-499 4 Very large 500-999 5 Mass > 1000

MOV: movement type

Code Type Description 0 Unknown 1 Single protest One group in an industry 2 Inter-movement Same industries (e.g., students from different schools) 3 Intra-movement Different industries (e.g., students and Christians)

119 REPRESS: whether or not the event was repressed (i.e., the repressive force(s) was/were present at the event) 0 = Repressive force not present (or not mentioned) 1 = Repressive force present

REPGROUP1-4: individual(s) or group(s) repressing

Code Group 0 Unknown 1 Riot police 2 Normal police 3 Special police 4 Hired thugs 5 Local authority (including universities) 6 Employers 7 Government (via repressive law) 8 Other

REPTYPE1-3: type of repression used by repressive force(s)

Code Type of repression Additional notes 0 Unknown 1 Disruption e.g., barricade, containment, daechee 2 Physical violence e.g., beating, throwing stones 3 Use of tear gas 4 Interrogation, torture 5 Conviction, custody/arrests 6 Organizational repression e.g., investigation, search 7 Killing (by torture, death sentence) 8 Censorship 9 Surveillance/spying 10 Firing, suspending individuals 11 Disbanding, closing down organizations, parties, schools, workplaces 12 Co-optation e.g., bribing, forced to sign papers

120 LINK1-3: Links between events already coded. Events may be linked in two ways: (1) subsequent events may be direct responses to an initial event or (2) two or more events may be related to the same issue or incident (e.g., death of a protestor, a well-known repression event).

• If you know the unique ID number of the previous event(s) related to the current event you are coding, input ID numbers of all events that are related to the current event.

• If you do not know the ID number of the previous event(s), describe the event as much as you can in each cell for each event is related to the current event.

REFERENCES VARIABLES

BOOK1-2: books used for coding

Code Book 1 KDF Taejonˇ & Ch’ungnam 2 KDF Chonnamˇ 3 KDF T’aebaek & Ch’ongsˇ onˇ 4 KDF Wonjuˇ & Ch’unch’onˇ 5 KDF Inch’onˇ 6 KDF Cheju 7 KDF Kyonggiˇ 8 KDF Ch’vonbuk 9 KDF Taegu & Kyongbukˇ 10 KDF Pusan & Kyongnamˇ 11 KDF Ch’ungbuk 12 CISJD (Great Workers’ Struggle) 13 KDF Timeline 14 CISJD (June Democratic Struggle) 15 KDF00017076 16 KDF 1980s

PAGE1-2: page numbers from respective books

121 SOURCE: type of original source 1 = Newspaper (national) 2 = Newspaper (local, school) 3 = Book 4 = KDF 5 = Archival material

COMMENTS: short description of each event

122 B| Appendix to Chapter 1: Additional Tables

123 Table B.1: Number of Political Protest by Organizing Groups, 1945-1972

U.S. Military Rule First Republic Interim Government Second Republic Military Government Third Republic (1945-1948) (1948-1960) (1960) (1960-1961) (1961-1963) (1963-1972) Students 14 (10.0) 106 (19.3) 60 (29.9) 29 (22.3) 8 (21.1) 922 (76.2) Politicians 17 (12.1) 91 (16.8) 15 (7.5) 11 (8.5) 9 (23.7) 85 (7.0) Citizens 22 (15.7) 89 (16.4) 40 (19.9) 13 (10.0) 3 (7.9) 65 (5.3) Professionals 11 (7.9) 53 (9.9) 23 (11.4) 40 (30.8) 7 (18.4) 71 (5.9) Laborers 16 (11.4) 29 (5.4) 14 (7.0) 18 (13.9) 1 (2.6) 38 (3.2) Communists 51 (36.4) 121 (22.5) 0 0 0 0 Others 9 (6.5) 50 (9.3) 49 (24.3) 19 (14.5) 10 (26.3) 29 (2.4) Total 140 (100) 539 (100) 201 (100) 130 (100) 38 (100) 1,210 (100) Source: Shin 1982, 29 124 Table B.2: Main Social Groups Participating in Protest

1970s Percent 1980–1987 Percent Students/Youth 31.71 Students/Youth 48.41 Laborers 17.00 Laborers 23.65 Christians 16.50 Christians 4.92 Journalists 6.39 Intellectuals 3.65 Christian Students 6.30 General Activists 3.65 Other 22.10 Other 15.72 Total 100 Total 100 Source: Shin et al. 2007, 22

125 C| Appendix to Chapter 2: Additional Tables

Table C.1: Summary Statistics

Statistic N Mean St. Dev. Min Max Outcome Variable Labor Protest Events, 1987 214 5.178 20.449 0 253

Industrial Complex Variables Presence of Industrial Complex 214 0.196 0.398 0 1 Number of Industrial Complexes 214 0.276 0.694 0 5 Duration of Industrial Complexes 214 2.752 6.164 0 26

Pre-treatment Covariates Distance to Nearest Port (m) 214 52,419.410 33,669.910 971.486 148,125.500 Prop. Rural Households, 1960 214 55.818 27.449 0.012 86.460 Prop. Employed in Manufacturing, 1960 214 7.052 7.672 0.858 55.203 Population, 1960 214 117,026.300 76,318.300 8,866 702,863

1987 Covariates Manufacturing Firms, 1987 214 243.379 492.305 2 3,394 Unemployment, 1985 214 0.395 0.105 0.161 0.723 Population Density, 1985 214 0.138 0.427 0.003 4.102 Population, 1985 214 185,726.300 233,925.800 17,281 2,029,853 June Democratic Uprising, 1987 214 0.318 0.467 0 1

126 Table C.2: The Great Workers’ Struggle Protests and Electoral Support for the Opposition Candidates in the First Democratic Presidential Election in 1987

Opposition Vote Share (1987)

Great Workers’ Struggle Protests 0.135∗∗∗ (0.045)

NKDP Vote Share (1985 legislative) 11.777∗ (6.029)

Urbanization 7.620∗∗∗ (1.526)

(Log) College Student Population 1.439∗∗ (0.628)

Honam Region 48.113∗∗∗ (1.836)

Yongnamˇ Region 0.249 (1.748)

N 211 R2 0.805 Adjusted R2 0.799 Residual Std. Error 10.082 (df = 204) F Statistic 140.265∗∗∗ (df = 6; 204) Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Standard errors in parentheses. Re- gional dummies—Honam Region and Yongnamˇ Region—are included to control for regionalism that affected the electoral results in the ﬁrst democratic presidential elections in 1987. Honam Region includes Chonbukˇ and Chonnamˇ and Yongnamˇ Region includes Kyongbukˇ and Kyongnam.ˇ

127 Table C.3: First Stage Estimates from the Sequential g-estimation Model of Table 2.2

Labor Protest, 1987 (1) (2) Presence of IC 5.081∗∗∗ (1.672)

Duration of IC 0.399∗∗∗ (0.114)

Manufacturing Firms, 1987 0.017∗∗∗ 0.016∗∗∗ (0.002) (0.002)

Unemployment, 1985 −8.579 −8.660 (8.547) (8.451)

Population Density, 1985 1.217 1.231 (1.965) (1.911)

Log Population, 1985 3.875∗∗∗ 3.520∗∗ (1.408) (1.419)

June Democratic Uprising, 1987 2.910∗ 2.965∗∗ (1.493) (1.478)

Province Fixed Effects XX 1960 Covariates XX

N 214 214 R2 0.890 0.892 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

128 Table C.4: First Stage Estimates from the Sequential g-estimation Model of Table 2.3

Labor Protest, 1987 (1) (2) Presence of IC 5.081∗∗∗ (1.672)

Duration of IC 0.395∗∗∗ (0.114)

Manufacturing Firms, 1987 0.017∗∗∗ 0.016∗∗∗ (0.002) (0.002)

Unemployment, 1985 −8.579 −8.914 (8.547) (8.459)

Population Density, 1985 1.217 0.825 (1.965) (1.960)

Log Population, 1985 3.875∗∗∗ 3.376∗∗ (1.408) (1.428)

June Democratic Uprising, 1987 2.910∗ 2.940∗∗ (1.493) (1.479)

Province Fixed Effects XX 1960 Covariates XX

Observations 214 214 R2 0.890 0.892 Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

129 D| Appendix to Chapter 3: Additional Tables

Table D.1: Summary Statistics: Newspaper

Panel A: 1963 Mean Std. Dev. Min Max Obs. Incumbent Vote Share 47.396 12.580 24.051 77.841 178 Circ. per Thousand (Kyonghyangˇ ) 5.962 6.311 0.000 46.976 178 Circ. per Thousand (Chosonˇ ) 9.741 13.987 0.000 114.998 178 Circ. per Thousand (Tonga) 14.076 21.032 0.000 107.369 178 Circ. per Thousand (Taehan) 11.660 14.980 0.000 71.517 178 Panel A: 1967 Mean Std. Dev. Min Max Obs. Incumbent Vote Share 51.931 11.636 32.185 82.116 189 Circ. per Thousand (Kyonghyangˇ ) 8.621 9.287 0.809 54.279 189 Circ. per Thousand (Chosonˇ ) 14.131 18.259 1.107 100.434 189 Circ. per Thousand (Tonga) 20.732 31.530 0.947 167.807 189 Circ. per Thousand (Taehan) 15.590 17.931 0.473 103.700 189

Table D.2: Summary Statistics: Radio

Mean Std. Dev. Min Max Obs. Incumbent Vote Share 50.357 13.746 19.977 85.226 258 Lagged Incumbent Vote Share 45.937 11.572 22.674 74.522 258 Radio Signal 0.000 1.000 -2.281 1.934 258

Radio signal variables are standardized.

130 Table D.3: Robustness Check I: Newspaper Circulation and Vote for Ruling Party

Dependent Var. = Incumbent Vote Share Newspaper: Kyong-ˇ Chosonˇ Tonga Taehan hyang (1) (2) (3) (4) Circulation per Thousand -0.716∗∗∗ -0.421∗∗∗ -0.150∗∗∗ -0.402∗∗∗ (0.124) (0.081) (0.042) (0.064)

Circulation per Thousand × Year 1967 0.922∗∗∗ 0.331∗∗∗ 0.272∗∗∗ 0.379∗∗∗ (0.152) (0.078) (0.034) (0.059) Observations 367 367 367 367

This table replicates Table 3.1 using the 1963 circulation ﬁgures. Robust standard errors in parentheses. A dummy variable for the year 1967 is included. Control variables are: population, population per km2, percent male, percent aged 60 and older, percent married, percent eligible voters, and the land size (km2). All the variables except population per square km and the land size are measured at the county level. Population per square km and the land size are measured at the district level. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001.

131 Table D.4: Robustness Check II: Newspaper Circulation and Vote for Ruling Party

Dependent Var. = Incumbent Vote Share Newspaper: Kyonghyangˇ Chosonˇ Tonga Taehan 1963 1967 1963 1967 1963 1967 1963 1967 (1) (2) (3) (4) (5) (6) (7) (8) Circulation per Thousand (1963) -0.497∗∗∗ -0.001 -0.283∗∗∗ -0.228∗∗∗ -0.126∗ 0.097∗∗ -0.297∗∗∗ -0.149 (0.130) (0.178) (0.058) (0.056) (0.054) (0.036) (0.067) (0.086) Observations 178 189 178 189 178 189 178 189

This table replicates Table 3.2 using the 1963 circulation ﬁgures. Robust standard errors in parentheses. Control variables are: population, population per km2, percent male, percent aged 60 and older, percent married, percent eligible voters, and the land size 132 (km2). All the variables except population per square km and the land size are measured at the county level. Population per square km and the land size are measured at the district level. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001. E| Appendix to Chapter 4: Additional Tables

Table E.1: 1985 NKDP Vote Share and Combined Vote Shares of the Opposition Candi- dates (Kim Young Sam and Kim Dae Jung) in the First Democratic Presidential Elections in 1987

Dependent Variable: Opposition Vote Share (1987) 1985 NKDP Vote Share 14.111∗∗ (6.090)

Urbanization 7.059∗∗∗ (1.542)

(Log) College Student Population 1.832∗∗∗ (0.625)

Honam region 47.466∗∗∗ (1.857)

Yongnamˇ region −0.376 (1.768)

Constant 22.041∗∗∗ (4.116)

Observations 211 R2 0.796 Adjusted R2 0.792 Residual Std. Error 10.272 (df = 205) F Statistic 160.447∗∗∗ (df = 5; 205) Note: ∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01. Standard errors in parentheses. Re- gional dummies—Honam Region and Yongnamˇ Region—are included to control for regionalism that affected the electoral results in the ﬁrst democratic presidential elections in 1987. Honam Region includes Chonbukˇ and Chonnamˇ and Yongnamˇ Region includes Kyongbukˇ and Kyongnam.ˇ

133 Table E.2: Results Using Different Measures of NKDP Vote Share

Continuous Dummy (24%) Protest Events Equation (negative binomial with log link)

NKDP Vote Share × College Students Per Capita 1.099∗ 0.477∗∗ (0.647)(0.189) NKDP Vote Share −10.829∗∗∗ −2.640∗∗∗ (2.689)(0.777) College Students Per Capita 0.423∗∗ 0.602∗∗∗ (0.177)(0.115) Abstention (1981) 5.346∗∗ 2.074 (2.528)(2.424) Electoral Competitiveness −0.043 0.172 (0.128)(0.120) (Log) Population 0.241 0.061 (0.254)(0.256) Population Density 0.860∗∗∗ 0.791∗∗∗ (0.247)(0.268) Industrial 4.414∗∗∗ 6.017∗∗∗ (1.571)(1.534) Unemployment 7.237∗∗ 6.422∗∗ (3.080)(3.015) Inch’onˇ 2.188∗ 2.931∗∗ (1.227)(1.334) Kwangju −2.012∗∗ −3.331∗∗∗ (0.963)(1.007) Pusan & Masan 0.590 0.787 (1.080)(1.173) Count Constant −7.696∗∗ −8.268∗∗ (3.302)(3.278) Province FE Yes Yes

Inﬂate Equation (binomial with log link)

NKDP Vote Share × College Students Per Capita 6.215 10.657 (5.668)(50.597) NKDP Vote Share −31.509∗ −76.725 (16.912)(149.393) College Students Per Capita −1.605 3.392 (1.088)(32.333) Abstention (1981) 17.902∗∗ 266.423 (8.772)(271.660) Electoral Competitiveness −1.392∗∗ −6.019 (0.680)(10.705) (Log) Population −0.564 −11.130 (1.409)(20.019) Population Density −43.472∗∗ −710.700 (19.777)(1384.216) Unemployment 48.104∗∗ 1056.815 (20.382)(1426.277) Industrial −20.549∗∗ −319.064 (9.298)(337.911) Zero Constant 2.930 −247.292 (18.463)(522.421) AIC 646.661 645.601 Log Likelihood -288.331 -287.801 Num. obs. 210 210 Nonzero obs. 76 76 ∗∗∗ p < 0.01, ∗∗ p < 0.05, ∗ p < 0.1. Standard errors in parentheses. 134 Table E.3: Summary Statistics

Variable Obs. Mean St. Dev. Min Max Events 211 5.374 13.282 0 89 Days 211 2.403 4.822 0 23 NMHDC Events 211 2.076 6.099 0 64 NKDP Vote Share 211 0.222 0.146 0.022 0.583 NKDP Vote Share (Dummy) 211 0.275 0.448 0 1 Abstention (1981) 210 18.683 7.292 −7.093 34.996 Electoral Competitiveness 211 4.720 1.152 3 9 College Students Per Capita 211 2.112 1.772 0.155 8.418 Population Density 211 0.319 0.630 0.003 3.045 Unemployment 211 40.083 10.240 16.108 55.564 Industrial 211 22.043 15.919 2.676 63.537 Population 211 182,842.600 176,833.900 17,281 942,617

135 Table E.4: Robustness Checks: Predictors of Protest Intensity

(1) (2) (3) Protest Events Equation (negative binomial with log link)

NKDP Vote Share × College Students Per Capita 0.543∗∗∗ 0.554∗∗∗ 0.672∗∗∗ (0.191)(0.157)(0.215) NKDP Vote Share −3.302∗∗∗ −2.500∗∗∗ −3.960∗∗∗ (0.876)(0.723)(1.009) College Students Per Capita 0.555∗∗∗ 0.485∗∗∗ 0.476∗∗∗ (0.105)(0.079)(0.129) (Log) Population 0.239 −0.025 −0.106 (0.261)(0.206)(0.297) Population Density 0.790∗∗∗ 0.275 0.937∗∗∗ (0.243)(0.201)(0.257) Abstention (1981) 1.320 −0.289 2.636 (2.308)(1.966)(2.641) Industrial 4.647∗∗∗ 5.888∗∗∗ 4.569∗∗ (1.590)(1.303)(1.789) Unemployment 7.173∗∗ 4.808∗∗ 9.876∗∗∗ (2.814)(2.323)(3.458) Electoral Competitiveness 0.019 0.026 0.154 (0.125)(0.105)(0.138) Inch’onˇ 2.621∗∗ 2.312∗∗ 2.934∗∗ (1.231)(0.994)(1.349) Kwangju −2.540∗∗∗ −2.697∗∗∗ −4.329∗∗∗ (0.967)(0.781)(1.192) Pusan & Masan 0.245 0.507 −0.391 (1.072)(0.887)(1.115) Count Constant −8.642∗∗∗ −4.773∗ −6.705∗ (3.240)(2.572)(3.536) Province FE Yes Yes Yes Inﬂate Equation (binomial with log link)

NKDP Vote Share × College Students Per Capita 0.787 −2.805 −5.741∗ (1.318)(20.935)(3.334) NKDP Vote Share 2.348 29.484 6.902 (3.788)(47.908)(6.195) College Students Per Capita −1.269 −5.186 −3.913∗∗ (0.957)(4.471)(1.728) (Log) Population −0.994 −9.320 −2.178 (1.725)(6.193)(1.913) Population Density −72.829∗∗ −436.487 5.440∗∗ (31.712)(337.329)(2.562) Abstention (1981) 15.649∗ 78.795 26.763 (9.340)(49.621)(16.374) Industrial −33.198∗∗ −140.258 −28.019∗∗ (16.015)(111.448)(14.004) Unemployment 97.015∗∗ 657.772 86.439∗∗ (44.738)(454.047)(39.184) Electoral Competitiveness −2.455∗∗ −15.103∗ −2.837∗∗ (1.208)(9.121)(1.339) Zero Constant −7.609 −63.290 9.095 (24.169)(63.399)(23.014) AIC 642.271 525.673 488.589 Log Likelihood -286.135 -227.837 -209.294 Num. obs. 210 210 210 Nonzero obs. 76 76 76 ∗∗∗ ∗∗ ∗ p < 0.01, p < 0.05, p < 0.1. Standard errors in parentheses. 136 F| Appendix to Chapter 4: Additional Figures

Figure F.1: Bivariate Correlation Between Student Population (1985) and Number of Stu- dent Associations (1983)

Source: Chon’gukˇ Taehak Yon’gamˇ P’yonch’anˇ Wiwonhoeˇ (1984)

137 Figure F.2: Number of Mass Events With More Than 10,000 Participants

Source: Author’s dataset

138 | Bibliography

Acemoglu, Daron, James Robinson, and Pierre Yared. 2008. “Income and Democracy.” American Economic Review 98 (3): 308-342.

Acemoglu, Daron, James Robinson, and Pierre Yared. 2009. “Revisiting the Modernization Hypothesis.” Journal of Monetary Economics 56: 1043-1058.

Acharya, Avidit, Matthew Blackwell, and Maya Sen. 2016a. “Explaining Causal Effects Without Bias: Detecting and Assessing Direct Effects.” American Political Science Re- view.

Acharya, Avidit, Matthew Blackwell, and Maya Sen. 2016b. “The Political Legacy of American Slavery.” Journal of Politics.

Adena, Maja, Ruben Enikolopov, Maria Petrova, Veronica Santarosa, and Ekaterina Zhu- ravskaya. 2013. “Radio and the Rise of Nazis in Pre-War Germany.” Working Paper.

Amsden, Alice. 1989. Asia’s Next Giant: South Korea and Late Industrialisation. Oxford: Oxford University Press.

Arat, Zehra F. 1988. “Democracy and Economic Development: Modernization Theory Re- visited.” Comparative Politics 21: 21-36.

Bank of Korea. 1970. “Economic Signiﬁcance of Industrial Complex and South Korea’s Composition of Industrial Complexes (in Korean).” Bank of Korea Weekly Domestic and Foreign Economy.

Bank of Korea. 1985. Economics Statistics Yearbook. Bank of Korea.

Barro, Robert J. 1990. “Determinants of Democracy.” Journal of Political Economy 107 (6): 157-183.

Becker, Brian, and Craig Olson. 1986. “The Impact of Strikes on Shareholder Equity.” Industrial and Labor Relations Review 39 (3): 425-438.

Beissinger, Mark. 2002. Nationalist Mobilization and the Collapase of the Soviet state. Cambridge: Cambridge University Press.

139 Benford, Robert D., and David A. Snow. 2000. “Framing Processes and Social Movements: An Overview and Assessment.” Annual Review of Sociology 26: 611-639.

Bennett, W. Lance. 1998. “The Media and Democratic Development: The Social Basis of Political Communication.” In Communicating Democracy: The Media and Political Transitions, ed. Patrick O’Neil. Boulder, CO: Lynne Rienner.

Berk, Richard A. 1974. “A Gaming Approach to Crowd Behavior.” American Sociological Review 39 (355-373).

Birch, Sarah. 2010. “Perceptions of Electoral Fairness and Voter Turnout.” Comparative Political Studies 43 (10).

Blaydes, Lisa. 2011. Elections and Distributive Politics in Mubarak’s Egypt. New York: Cambridge University Press.

Blee, Kathleen M., and Ashley Currier. 2006. “How Local Social Movement Groups Han- dle a Presidential Election.” Qualitative Sociology 29 (3): 261-280.

Boix, Carles, and Milan Svolik. 2013. “The Foundation of Limited Authoritarian Gov- ernment: Institutions, Commitment, and Power-Sharing in Dictatorships.” Journal of Politics 75 (April): 300-316.

Boix, Carles, and Susan Stokes. 2003. “Endogenous Democratization.” World Politics 53 (4): 517-549.

Bollen, Kenneth. 1979. “Political Democracy and the Timing of Development.” American Sociological Review 44: 572-585.

Brownlee, Jason. 2007. Authoritarianism in an Age of Democratization. Cambridge: Cam- bridge University Press.

Brownlee, Jason. 2012. “Executive Elections in the Arab World: When and How do they Matter?” Comparative Political Studies 44 (7): 807-828.

Buechler, Steven M. 2004. “The Strange Career of Strain and Breakdown Theories of Col- lective Action.” In The Blackwell Companion to Social Movements, ed. David A. Snow, Sarah A. Soule, and Hanspeter Kriesi. Malden, MA: Blackwell Publishing Ltd.

Buehler, Matt. 2013. “Safety-Valve Elections and the Arab Spring: The Weakening (and Resurgence) of Morocco’s Islamist Opposition Party.” Terrorism and Political Violence 25: 137-156.

Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Siverson, and James D. Marrow. 2004. The Logic of Political Survival. The MIT Press.

140 Bunce, Valerie J., and Sharon L. Wolchik. 2006. “International Diffusion and Postcommu- nist Electoral Revolutions.” Communist and Post-Communist Studies 39 (3): 283-304. Bunce, Valerie J., and Sharon L. Wolchik. 2011. Defeating Authoritarian Leaders in Post- communist Countries. New York: Cambridge University Press. Burkhart, Ross E., and Michael S. Lewis-Beck. 1994. “Compartive Democracy: The Eco- nomic Development Thesis.” American Political Science Review 88: 903-910. Calhoun, Craig J. 1981. The Question of Class Struggle: The Social Foundations of Popular Radicalism during the Industrial Revolution. Chicago, IL: University of Chicago Press. Chang, Eric CC, Miriam A. Golden, and Seth J. Hill. 2010. “Legislative Malfeasance and Political Accountability.” World Politics 62 (02): 177–220. Chang, Paul. 2015a. “The Polarization of South Korea: The Legacy of Authoritarianism and People’s Movements.” Stanford University Press Blog. Chang, Paul Y. 2015b. Protest Dialectics: State Repression and South Korea’s Democracy Movement, 1970-1979. Stanford, California: Stanford University Press. Choe, Hyun, and Jiyoung Kim. 2012. “South Korea’s Democratization Movements, 1980- 1987: Political Structure, Political Opportunity, and Framing.” Inter-Asia Cultural Stud- ies (1): 55-68. Choi, Jang Jip. 1989. Han’guk hyondaeˇ chongchi’iˇ uiˇ kujowa pyonhwaˇ [Contemporary Ko- rean politics: structure and change]. Seoul: Kkach’i. Choi, Jang Jip. 1993. Han’guk minjujuuiui iron [Theory of Korean Democracy]. Seoul: Han’gilsa. Choi, Jang Jip. 2002. Minjuhwa ihuui minjujuui [Democracy after Democratization]. Seoul: Humanitas. Chosonˇ Bank Research Department. 1948. Chosonˇ Economic Statistics Yearbook. Chosonˇ Bank. Chow, Tse-tsung. 1967. The May Fourth Movement: Intellectual Revolution in Modern China. Stanford, Calif.: Stanford University Press. Chung, Chulhee. 2011. “Mesomobilization and the June Uprising and Cultural Integration in Pro-democracy Movements in South Korea.” In East Asian Social Movements: Power, Protest, and Change in a Dynamic Region, ed. Jeffrey Broadbent and Vicky Brockman. Springer. Chung, Hae-Gu, Hye-Jin Kim, and Sang-Ho Chung. 2004. 6-wolˇ hangjaeng kwa han’guk uiˇ minjujuuiˇ [June Democratic Struggle and Korea’s Democracy]. Korea Democracy Foundation. 141 Chong,ˇ Mi-suk. 1993. 70yondaeˇ yosˇ ongnodongundongˇ uiˇ hwalsongwaeˇ kwanhan kyongˇ omˇ segyejokˇ yon’gu:ˇ somyuˇ obˇ ulˇ chungsimuroˇ [A Study on the Development of the 70s Women’s Labor Movement in Textile Industry]. Master’s thesis Ihwa Women’s Univer- sity. Chon’gukˇ Taehak Yon’gamˇ P’yonch’anˇ Wiwonhoe.ˇ 1984. Han’guk taehak yon’gamˇ [Ko- rean University Yearbook]. Seoul: Aeduˇ Yong.ˇ Clark, David H., and Patrick M. Regan. 2003. “Opportunities to Fight: A Statistical Technique for Modeling Unobservable Phenomena.” The Journal of Conﬂict Resolution 47 (1): 94-115. Clemens, Elisabeth S. 1996. “Organizational Form as Frame: Collective Identity and Polit- ical Strategy in the American Labor Movement, 1880-1920.” In Comparative Perspec- tives on Social Movements: Political Opportunity Structures, Mobilizing Structures, and Cultural Framings, ed. Doug McAdam, John D. McCarthy, and Mayer N. Zald. Cam- bridge Studies in Comparative Politics Cambridge University Press. Collier, Ruth Berins. 1999. Paths toward Democracy: The Working Class and Elites in Western Europe and South America. Cambridge: Cambridge University Press. Collier, Ruth Berins, and David Collier. 2002. Shaping the Political Arena. Notre Dame: University of Notre Dame Press. Collier, Ruth Berins, and James Mahoney. 1997. “Adding Collective Actors to Collec- tive Outcomes: Labor Recent Democratization in South America and Southern Europe.” Comparative Politics 29 (3): 285-303. Costas-Pérez, Elena, Albert Solé-Ollé, and Pilar Sorribas-Navarro. 2012. “Corruption Scandals, Voter Information, and Accountability.” European Journal of Political Econ- omy 28 (4): 469–484. Dalton, Russell J. 1994. “Communists and Democrats: Democratic Attitudes in the two Germanies.” British Journal of Political Science 24: 469-493. Davies, James C. 1962. “Toward a Theory of Revolution.” American Sociology Review 26 (7): 5-19. DeFusco, Richard, and Scott Fuess, Jr. 1991. “The Effects of Airline Strikes and Struck on Nonstruck Carriers.” Industrial and Labor Relations Review 44 (2): 324-333. Delgado, Gary. 1968. Organizing the Movement: The Roots and Growth of Acorn. Philadel- phia: Temple University Press. DellaVigna, Stefano, Ruben Enikolopov, Vera Mironova, Maria Petrova, and Ekaterina Zhuravskaya. 2013. “Cross-Border Effects of Foreign Media: Serbian Radio and Na- tionalism in Croatia.” American Economic Journal: Applied Economics. Forthcoming. 142 Deutsch, Karl. 1961. “Social Mobilization and Political Development.” American Political Science Review 55 (September).

Deyo, Frederic. 1989. Beneath the Miracle: Labor Subordination in the New Asian Indus- trialism. Berkeley: University of California Press.

Diamond, Larry. 1993a. “The Globalization of Democracy.” In Global Transformation and the Third World, ed. Steven R. Dorr Robert O. Slater, Barry M. Schutz. Boulder, CO: Lynne Rienner.

Diamond, Larry. 1993b. Political Culture and Democracy in Developing Countries. Boul- der, CO: Lynne Rienner.

Dinardo, John, and Kevin Hallock. 2002. “When Unions ‘Mattered’: The Impact of Strikes on Financial Markets, 1925-1937.” Industrial and Labor Relations Review 55 (2): 219- 233.

Economic Planning Board. 1987. Kwanggongopˇ t’onggye chosa pogosoˇ [Report on Mining and Manufacturing]. Seoul: Kyongjeˇ Kihoegwonˇ [Economic Planning Board].

Eisenstadt, Todd. 2007. Courting Democracy in Mexico: Party Strategy and Electoral In- stitutions. New York, N.Y.: Cambridge University Press.

Enikolopov, Ruben, Maria Petrova, and Ekaterina Zhuravskaya. 2011. “Media and Political Persuasion: Evidence from Russia.” American Economic Review 101 (7): 3253–3285.

Epstein, David, Robert Bates, Jack Goldstone, Ida Kristensen, and Sharyn O’Halloran. 2006. “Democratic Transitions.” American Journal of Political Science 50: 551-569.

Feagin, Joe R., and Harlan Hahn. 1973. Ghetto Revolts: The Politics of Violence in Ameri- can Cities. New York: Macmillan.

Fergusson, Leopoldo. 2014. “Media Markets, Special Interests, and Voters.” Journal of Public Economics 109: 13–26.

Flake, L. Gordon. 2008. “The Rise, Fall, and Transformation of the “386”: Generational Change in Korea.” In Emerging Leaders in East Asia: The Next Generation of Political Leadership in China, Japan, South Korea, and Taiwan, ed. Cheng Li. National Bureau of Asian Research.

Fogelson, Robert M. 1971. Violence as Protest: A Study of Riots and Ghettos. Garden City, N.Y.: Anchor Book.

Gamson, William A., and David S. Meyer. 1996. “Framing Political Opportunity.” In Com- parative Perspectives on Social Movements: Political Opportunity Structures, Mobi- lizing Structures, and Cultural Framings, ed. Doug McAdam, John D. McCarthy, and Mayer Zald. Cambridge University Press. 143 Gandhi, Jennifer. 2008. Political Institutions Under Dictatorship. New York: Cambridge University Press. Gandhi, Jennifer, and Adam Przeworski. 2007. “Authoritrian Institutions and Survival of Autocrats.” Comparative Political Studies 40 (11): 1279-1301. Gandhi, Jennifer, and Ellen Lust-Okar. 2009. “Elections Under Authoritarianism.” Annual Review of Political Science 12: 403-422. Geddes, Barbara. 2005. “Why parties and elections in authoritarian regimes?” Paper presented at the Annual Meeting of the American Political Science Association. Geddes, Barbara. 2009. “What Causes Democratization?” In The Oxford Handbook of Comparative Politics, ed. Carles Boix and Susan Stokes. Oxford University Press. Gould, Roger V. 1995. Insurgent Identities: Class, Community, and Protest in Paris from 1848 to the Commune. Chicago, Ill.: University of Chicago Press. Greene, Kenneth F. 2007. Why Dominant Parties Lose: Mexico’s Democratization in Com- parative Perspective. Cambridge: Cambridge University Press. Greitens, Sheena Chestnut. N.d. “Dictators & Their Secret Police: Coercive Institutions & State Violence Under Authoritarianism.” Draft manuscript. Gurr, Ted R. 1970. Why Men Rebel? Princeton, NJ: Princeton University Press. Gutman, Herbert. 1977. Work, Culture and Society in Industrializing America. New York: Vintage. Haddad, Mary Alice. 2012. Building Democracy in Japan. New York: Cambridge Univer- sity Press. Hadenius, Axel. 1992. Democracy and Development. New York: Cambridge University Press. Hadenius, Axel, and Jan Teorell. 2007. “Pathways from Authoritarianism.” Journal of Democracy 18 (1): 143-157. Hafner-Burton, Emilie M., Susan D. Hyde, and Ryan S. Jablonski. 2014. “When Do Gov- ernments Resort to Election Violence?” British Journal of Political Science 44 (January): 149-179. Hanagan, Michael. 1989. Nascent Proletarians: Class Formation in Post-Revolutionary France. Cambridge, MA: Basil Blackwell. Han’guk Kidokkyo Kyohoe Hyobˇ uihoeˇ [Christian Institute for the Study of Justice and De- velopment]. 1984. Nodong hyonjanggwaˇ chungˇ onˇ [Labor Sites and Testimonies]. Chris- tian Institute for the Study of Justice and Development. 144 Han’guk Kidokkyo Kyohoe Hyobˇ uihoeˇ [Christian Institute for the Study of Justice and De- velopment]. 1987a. 7-8wol nodongja daetujaeng [July-August Great Workers’ Struggle]. Seoul: Christian Institute for the Study of Justice and Development.

Han’guk Kidokkyo Kyohoe Hyobˇ uihoeˇ [Christian Institute for the Study of Justice and Development]. 1987b. 6-wolˇ minjuhwa daet’ujaeng [The Great June Uprising for democratization]. Seoul: .

Hart-Landsberg, Martin. 1993. The Rush to Development: Economic Change and Political Struggle in South Korea. Monthly Review Press.

Hassanpour, Navid. 2011. “Media Disruption Exacerbates Revolutionary Unrest: Evidence from Mubarak’s Natural Experiment.” Paper prepared for the American Political Science Association Annual Meeting, Seattle, WA.

Heirich, Max. 1971. The Spiral of Conﬂict: Berkeley 1964. New York: Columbia Univer- sity Press.

Hilbe, Joseph M. 2011. Negative Binomial Regression. New York: Cambridge University Press.

Hirschman, Albert O. 1979. Shifting Involvements: Private Interest and Public Action. Princeton, NJ: Princeton University Press.

Hong, Ji Yeon, and Sunkyoung Park. 2015. “Factories for Votes? How Dictators Gain Popular Support Using Targeted Industrial Policy.” British Journal of Political Science.

Howard, Philip N., Aiden Duffy, Deen Freelon, Muzammil Hussain, Will Mari, and Marwa Mazaid. 2011. “Opening Closed Regimes: What Was the Role of Social Media during the Arab Spring?” Project on Information Technology and Political Islam.

Huang, Chang-Ling. 1999. “Labor Militancy and the Neo-Mercantilist Development Ex- perience: South Korea and Taiwan in Comparison.” Ph.D. diss. University of Chicago.

Hufford, George A. 2002. The ITS Irregular Terrain Model, Version 1.2.2: The Algorithm. Institute for Telecommunication Sciences, National Telecom- munications and Information Administration, US Department of Commerce. http://ﬂattop.its.bldrdoc.gov/itm.html.

Huntington, Samuel. 1996. “Democracy’s Third Wave.” In The Global Resurgence of Democracy, ed. Larry Diamond and Marc F. Plattner. Johns Hopkins University Press.

Huntington, Samuel P. 1968. Political Order in Changing Societies. New Haven: Yale University Press.

Huntington, Samuel P. 1991. The Third Wave: Democratization in the Late Twentieth Cen- tury. Norman: University of Oklahoma Press. 145 Hwang, Eui-bong. 1985. “Nodongyonjangˇ uiˇ chisigindulˇ [Labor’s Intellectuals].” Nupposidaeˇ 2.

Im, Seong-ho. 2010. “Gukhoiui yeoksajeok byeoncheon [Historical Transformation of the National Assembly].” In Hanguk gukhoiwa jeongchi gwajeong [Korean National Assem- bly and Its Political Process], ed. National Assembly Research Institute. Orum.

Inglehart, Ronald, and Christian Welzel. 2005. Modernization, Cultural Change, and Democracy: The Human Development Sequence. New York: Cambridge University Press.

Inglehart, Ronald, and Christian Welzel. 2009. “How Does Development Lead to Democ- racy? What We Know About Modenrization.” Foreign Affairs 88 (2): 33-48.

Inkeles, Alex. 1966. Modernization: The Dynamics of Growth. New York: Basic Books.

Jackman, Robert W. 1973. “On the Relations of Economic Development to Democratic Performance.” American Journal of Political Science 17: 611-621.

Jennings, M. Kent. 1987. “Residue of a Movement: The Aging of the American Protest Generation.” American Political Science Review 81 (2): 367-382.

Joffe, Marshall M., and Chi-Yuan Hsu. 2009. “Deﬁning and Estimating Intervention Effects for Groups that will Develop an Auxiliary Outcome.” Statistical Science 22 (1): 74-97.

Johnson, Chalmers. 1982. MITI and the Japanese Miracle: The Growth of Industrial Policy, 1925-1975. Stanford: Stanford University Press.

Jones, Leroy, and Il Sakong. 1990. Government, Business, and Entrepreneurship in Eco- nomic Development: The Korean Case. Cambridge: Harvard University Press.

Joo, Dong Hwang, Hye Shik Kim, and Young Kyu Park. 1997. Han’guk ollonsaˇ uiˇ ihae [Understanding the History of Korean Media]. Seoul: Chon’gukˇ Ollonˇ Nodong Chohap Yonmaeng.ˇ

Kassow, Samuel D. 1989. Students, Professors, and the State in Tsarist Russia. Berkeley and Los Angeles: University of California Press.

Katznelson, Ira. 1986. “Working-Class Formation: Constructing Cases and Comparison.” In Working-Class Formation: Nineteenth-Century Patterns in Western Europe and the United States, ed. Ira Katznelson and Aristide R. Zolberg. Princeton: Princeton Univer- sity Press.

Kennedy, Ryan. 2010. “The Contradictions of Modernization: A Conditional Model of Endogenous Democratization.” Journal of Politics 72 (3): 785-798.

146 Kern, Holger Lutz. 2011. “Foreign Media and Protest Diffusion in Authoritarian Regimes: The Case of the 1989 East German Revolution.” Comparative Political Studies 44 (September): 1179-1205.

Kern, Holger Lutz, and Jens Hainmueller. 2009. “Opium for the Masses: How Foreign Media Can Stabilize Authoritarian Regimes.” Political Analysis 17: 377–399.

Kim, Hak Chon. 2007. Han’guk uiˇ songch’alˇ kwa kaehyokˇ [Insights and Innovation of Broadcasting in South Korea]. Paju: Han’guk Haksul Chongbo.ˇ

Kim, Ju Eon. 2009. Han’guk uiˇ ollonˇ t’ongje : ollonˇ t’ongje e taehan majimak kirok igil yomwˇ onˇ hamyoˇ [Media Control in Korea]. Seoul: Ribuk.

Kim, Jun. 1993. “Asia gwonwijuui gukkaui nodong jeongchiwa nodongundong [Labor Pol- itics and Labor Movements in Asian Authoritarian States].” Ph.D. diss. Seoul National University.

Kim, Sookyung, and Paul Y. Chang. 2011. “The Entry of Past Activists into the National Assembly and South Korea’s Participation in the Iraq War.” In South Korean Social Movements: From Democracy to Civil Society, ed. Gi-Wook Shin and Paul Y. Chang. London: Routledge.

Kim, Sunhyuk. 2000. The Politics of Democratization in Korea: The Role of Civil Society. Pittsburgh, PA: University of Pittsburgh Press.

Kim, Sunhyuk. 2003. “Civil Society in Democratizing Korea.” In Korea’s Democratization, ed. Samuel S. Kim. New York: Cambridge University Press.

Kim, Won. 2004. “1970yvondae ‘yogong’ˇ uiˇ munhwa: minjunojo saopchangˇ uiˇ kisuksawa somoim munhwarulˇ chungsimuroˇ [The Culture of ‘Factory Girls’ in 1970s].” Feminism yon’guˇ [Feminism Studies] 4 (1): 101-148.

King, Gary, Jennifer Pan, and Margaret E. Roberts. 2013. “How Censorship in China Al- lows Government Criticisms but Silences Collective Expression.” American Political Science Review 107 (2): 1-18.

Klandermans, Bert. 1997. The Social Psychology of Protest. Cambridge: Blackwell Pub- lishing Ltd.

Kohli, Atul. 2004. State-Directed Development: Political Power and Industrialization in the Global Periphery. Cambridge and New York: Cambridge University Press.

Koo, Hagen. 1999. “Modernity in South Korea: An Alternative Narrative.” Thesis Eleven 57: 53-64.

Koo, Hagen. 2001. Korean Workers: The Culture and Politics of Class Formation. Ithaca, NY: Cornell University Press. 147 Koo, Hagen. 2011. “Labor Policy and Labor Relations during the Park Chung Hee Era.” In Reassessing the Park Chung Hee Era, 1961-1979: Development, Political Thought, Democracy, and Cultural Inﬂuence, ed. Hyung-A Kim and Clark W. Sorensen. Univer- sity of Washington Press.

Korea Broadcasting Business Association [Hanguk Bangsong Saophoe]. 1969. Bangsong yongamˇ 1969 [Broadcasting Yearbook 1969].

Korea Broadcasting Business Association [Hanguk Bangsong Saophoe]. 1973. Bangsong yongamˇ 1973 [Broadcasting Yearbook 1973].

Kornhauser, William. 1959. The Politics of Mass Society. Glencoe: The Free Press.

Korpi, Walter, and Michael Shalev. 1979. “Strikes, Industrial Relations and Class Conﬂict in Capitalist Societies.” British Journal of Sociology 30 (2): 164-187.

Krueger, Alan, and Alexandre Mas. 2004. “Strikes, Scabs, and Tread Separations: Labor Strife and the Production of Defective Bridgestone/Firestone Tires.” Journal of Political Economy 112: 253-289.

Kuntz, Philipp, and Mark Thompson. 2009. “More than Just the fInal Straw: Stolen Elec- tions as Revolutionary Triggers.” Comparative Politics 41 (3): 253-272.

Kurzman, Charles. 1996. “Structural Opportunity and Perceived Opportunity in Social Movement Theory: The Iranian Revolution of 1979.” American Sociological Review 61: 153-170.

Kwak, Ki-Sung. 2012. Media and Democratic Transition in South Korea. New York: Rout- ledge.

Lankov, Andrei. 2008. “Fiasco of 386 Generation.” Korea Times.

Larreguy, Horacio A., John Marshall, and James M. Snyder, Jr. 2014. “Revealing Malfea- sance: How Local Media Facilitates Electoral Sanctioning of Mayors in Mexico.” Work- ing Paper.

Lawson, J. Chappell H. 2002. Building the Fourth Estate: Democratization and the Rise of a Free Press in Mexico. Berkeley, CA: University of California Press.

Lee, Jung-eun. 2010a. “Dynamics of Interorganizational Collaboration: Social Movements During Korea’s Transition to Democracy.” Ph.D. diss. Department of Sociology, Stan- ford University.

Lee, Myung-sik. 2010b. The History of the Democratization Movement in Korea. Seoul: Korea Democracy Foundation.

148 Lee, Namhee. 2007. The Making of Minjung: Democracy and the Politics of Representation in South Korea. Ithaca and London: Cornell University Press. Lee, Yoonkyung. 2011. Militants or Partisans: Labor Unions and Democratic Politics in Korea and Taiwan. Stanford University Press. Lehoucq, Fabrice, and Ivan Molina. 2002. Stufﬁng the Ballot Box. Cambridge: Cambridge University Press.

Leifeld, Philip. 2013. “texreg: Conversion of Statistical Model Output in R to LATEX and HTML Tables.” Journal of Statistical Software 55 (8): 1-24. Lemke, Douglas, and William Reed. 2001. “The Relevance of Politically Relevant Dyads.” Journal of Conflict Resolution 45 (1): 125-144. Lerner, Daniel. 1958. The Passing of Traditional Society: Modernizing the Middle East. Glencoe, New York: The Free Press. Levitsky, Steven, and Lucan Way. 2002. “Elections Without Democracy: The Rise of Com- petitive Authoritarianism.” Journal of Democracy 13 (April): 51-65. Levitsky, Steven, and Lucan Way. 2010. Competitive Authoritarianism: Hybrid Regimes After the Cold War. New York, NY: Cambridge Univ Press. Lijphart, Arend. 1997. “Unequal Participation: Democracy’s Unresolved Dilemma.” Amer- ican Political Science Review 91 (1): 1-14. Lindberg, Staffan I. 2006a. Democracy and Elections in Africa. Baltimore, Maryland: The Johns Hopkins University Press. Lindberg, Staffan I. 2006b. “The Surprising Significance of African Elections.” Journal of Democracy 17 (1): 139-151. Lipset, Seymour Martin. 1959. “Some Social Prerequisites of Democracy: Economic Development and Political Legitimacy.” The American Political Science Review 53 (March): 69-105. Lipset, Seymour Martin. 1960. Political Man: The Social Bases of Politics. Garden City, N.Y.: Doubleday. Lofland, John. 1970. “The Youth Ghetto.” In The Logic of Social Hierarchies, ed. Edward Laumann, Paul M. Siegel, and Robert W. Hodge. Chicago: Marrham. Londregan, John B., and Keith T. Poole. 1996. “Does High Income Promote Democracy?” World Politics 49: 1-30. Lust-Okar, Ellen. 2006. “Elections Under Authoritarianism: Preliminary Lessons from Jor- dan.” Democratization 13 (3): 456-471. 149 Lust-Okar, Ellen, and Amaney Jamal. 2002. “Rulers and Rules: Re-assessing the Influence of Regime Type on Electoral Law Formations.” Comparative Political Studies 35 (April): 337-366.

Maeil Kyongjeˇ Sinmunsa. 1987. Hoesa yon’gamˇ [Company Yearbook]. Seoul: Maeil Kyongjeˇ Sinmunsa.

Magaloni, Beatriz. 2006. Voting for Autocracy: Hegemonic Party Survival and its Demise in Mexico. Cambridge University Press.

Magaloni, Beatriz. 2008. “Credible Power-Sharing and the Longevity of Authoritarian Rule.” Comparative Political Studies 41 (4-5): 715-741.

Mann, Michael. 1993. The Sources of Social Power, Vol II: The Rise of Classes and Nation- States, 1760 -1914. Cambridge: Cambridge University Press.

Mannheim, Karl. 1952. Essays on the Sociology of Knowledge. Routledge and Kegan Paul.

Marshall, Monty G., Ted Robert Gurr, and Keith Jaggers. N.d. “Polity IV Project: Political Regime Characteristics and Transitions, 1800-2013.” URL: http://www.systemicpeace.org/inscrdata.html.

Marx, Karl, Friedrich Engels, Samuel Moore, and David McLellan. 1992. The Communist Manifesto. Oxford: Oxford University Press.

Mason, Edward S., Mahn Je Kim, Dwight H. Perkins, Kwang Suk Kim, and David C. Cole. 1980. The Economic and Social Modernization of the Republic of Korea. Cambridge: Harvard University Press.

McAdam, Doug. 1982. Political Process and Development of Black Insurgency. Chicago, IL: The University of Chicago Press.

McAdam, Doug. 1988. Freedom Summer. Chicago, Ill.: University of Chicago Press.

McAdam, Doug, John D. McCarthy, and Mayer N. Zald, eds. 1996. Comparative Perspec- tives on Social Movements: Political Opportunity Structures, Mobilizing Structures, and Cultural Framings. Cambridge Studies in Comparative Politics Cambridge University Press.

McAdam, Doug, and Sidney Tarrow. 2010. “Ballots and Barricades: On the Reciprocal Re- lationship between Elections and Social Movements.” Perspectives on Politics 8 (June): 529-542.

McCarthy, John D., and Mayer Zald. 1977. “Resource Mobilization and Social Move- ments.” American Journal of Sociology 82: 1212-1241.

150 McHugh, Richard. 1991. “Productivity Effects of Strikes in Struck and Nonstruck Indus- tries.” Industrial and Labor Relations Review 44 (4): 722-732.

McMillan, John, and Pablo Zoido. 2004. “How to Subvert Democracy: Montesinos in Peru.” Journal of Economic Perspectives 18 (4): 69–92.

Meier, Patrick. 2011. “Do ‘Liberation Technologies’ Change the Balance of Power Be- tween Repressive States and Civil Society?” Ph.D. diss. Tufts University.

Miller, Michael K. 2012. “Economic Development, Violent Leader Removal, and Democ- ratization.” American Journal of Political Science.

Miller, Warren E. 1992. “The Puzzle Transformed: Explaining Declining Turnout.” Politi- cal Behavior 14 (1): 1-40.

Miller, Warren E., and J. Merrill Shanks. 1996. The New American Voter. Cambridge: Harvard University Press.

Min, Hee, and Chang Nam Kim. 2008. “A Study on the 386 Generation’s Political Atti- tude.” Journal of KSSSS 17: 23-44.

Ministry of Commerce and Industry [Sanggongbu]. 1989. Kongopˇ tanji hyonhwangˇ [An Overview of Industrial Complexes].

Ministry of Culture and Information [Munhwa Gongbobu]. 1969. Jeonguk Susingi Bogeop Siltae [National Distribution of Radio Receivers].

Ministry of Culture and Information [Munhwa Kongbobu]. 1968. Han’guk Pangsong Hyonhwangˇ [The Current Situations of Korean Broadcasting].

Ministy of Construction [Konsˇ olbu].ˇ 1969. Korea’s Ports [Han’gukuiˇ hangman].

Moore, Barrington. 1966. The Social Origins of Dictatorship and Democracy. Boston: Beacon Press.

Muller, Edward N. 1985. “Income Inequality, Regime Repressiveness, and Political Vio- lence.” American Sociology Review 50 (1): 47-61.

Nam, Hwasook. 2009. Building Ships, Building a Nation: Korea’s Democratic Unionism under Park Chung Hee. Seattle and London: University of Washington Press.

National Intelligence Service [Kukka Chongbowˇ on].ˇ 2007. Kwagoˇ wa taehwa mirae uiˇ songch’alˇ [Communication with the Past]. Vol. V: Media and Labor.

Neumann, George. 1980. “The Predictability of Strikes: Evidence from the Stock Market.” Industrial and Labor Relations Review 33 (4): 525-532.

151 Noh, Hwanhee, Jungmin Song, and Won-Taek Kang. 2013. “Generation Effects Remain Salient? An Analysis of the South Korean Presidential Elections between 1997 and 2012.” Korean Party Studies Review 12 (1): 113-140.

Ogle, George E. 1990. South Korea: Dissent within the Economic Miracle. London: Zed Books, Ltd.

Oh, Ha-na. 2010. Hakch’ul : 80-yondae,ˇ kongjang uroˇ kan taehaksaengdulˇ [Student Work- ers: College Students Who Went to Factories in the 80s]. Seoul: Imaejin.

Oh, Se-je, and Hyonuˇ Yi. 2005. “The Conditional Effect of the Generation Effect of the 386 Generation: A Study on Ideological Tendency and Voting in Presidential Elections.” Uijˇ ongyˇ on’guˇ [Legislative Research] 20 (1): 201-230.

Olken, Benjamin A. 2009. “Do Television and Radio Destroy Social Capital? Evidence from Indonesian Villages.” American Economic Journal: Applied Economics 1 (4): 1– 33.

Olukotun, Ayo. 2002. “Authoritarian State, Crisis of Democratization and the Underground Media in Nigeria.” African Affairs 101: 317-342.

Olzak, Susan. 1989. “Labor Unrest, Immigration, and Ethnic Conﬂict in Urban America, 1880-1914.” The American Journal of Sociology 94: 1303-33.

Olzak, Susan, and Elizabeth West. 1995. “Ethnic Collective Action in Contemporary Urban U.S.”

Opp, Karl-Dieter, and Wolfgang Roehl. 1990. “Repression, Micromobilization, and Politi- cal Protest.” Social Forces 69 (2): 521-547.

Orru, Marco, Woosely Biggart, Nicole, and Gary G. Hamilton. 1997. The Economic Orga- nization of East Asian Capitalism. Thousand Oaks: Sage.

Paige, Jeffery M. 1975. Agrarian Revolution: Social Movements and Export Agriculture in the Underdeveloped World. New York: Free Press.

Pang, Hye-sin. 1993. 70yondaeˇ yosˇ ongnodongundongesˇ oˇ yosˇ ongˇ t’uksugwajeˇ uiˇ sirhyonjogˇ oneˇ kwanhan yon’guˇ [A Study on the Special Topics in the Women’s Labor Movement in the 1970s]. Master’s thesis Sogang University.

Park, Mi. 2008. Democracy and Social Change: A History of South Korean Student Move- ments, 1980-2000. Peter Lang Ltd.

Pei, Minxin. 1995. “Creeping Democratization in China.” Journal of Democracy 6 (4): 65-79.

152 Perry, Elizabeth. 1993. Shanghai on Strike: The Politics of Chinese Labor. Stanford, Cali- fornia: Stanford University Press.

Perry, Ronald, David F. Gillespie, and Howard A. Parker. 1976. Social Movements and the Local Community. Beverly Hills, Calif.: Sage.

Pierson, Paul. 2004. Politis in Time: History, Institutions, and Social Analysis. Princeton: Princeton University Press.

Piven, Frances Fox, and Richard A. Cloward. 1979. Poor People’s Movements: Why They Succeed and How They Fail. New York: Vintage Books.

Pontusson, Jonas. 1993. “The Comparative Politics of Labor-Initiated Reforms.” Compar- ative Political Studies 25 (4): 548-578.

Pontusson, Jonas, and Peter Swenson. 1996. “Labor Markets, Production Strategies, and Wage Bargaining Institutions.” Comparative Political Studies 29 (2): 223-250.

Przeworski, Adam, and Fernando Limongi. 1997. “Modernization Theories and Facts.” World Politics 49 (January): 155-183.

Przeworski, Adam, Michael E. Alvarez, José Antonio Cheibub, and Fernando. 2000. De- velopment and Democracy. New York: Cambridge University Press.

Putnam, Robert D. 1995. “Tuning In, Tuning Out: The Strange Disappearance of Social Capital in America.” PS: Political Science and Politics 28 (December): 664-683.

Pyle, Kenneth B. 2008. “Political Generations in East Asia: The Policy Signiﬁcance.” In Emerging Leaders in East Asia: The Next Generation of Political Leadership in China, Japan, South Korea, and Taiwan, ed. Li Cheng. National Bureau of Asian Research.

Rawnsley, Gary D., and Ming-Yeh T. Rawnsley. 1998. “Regime Transition and the Media in Taiwan.” Democratization 5 (2): 106-124.

Robertson, Graeme B. 2011. The Politics of Protest in Hybrid Regimes: Managing Dissent in Post-Communist Russia. New York, N.Y.: Cambridge University Press.

Roessler, Philip G., and Marc M. Howard. 2009. “Post-Cold War political regimes: When Do elections matter?” In Democratization by Elections, ed. Staffan I. Lindberg. Balti- more, Maryland: Johns Hopkins University Press.

Rogowski, Ronald. 1989. Commerce and Coalitions: How Trade Affects Domestic Political Alignments. Princeton: Princeton University Press.

Rohrschneider, Robert. 1994. “Report from the Laboratory: The Inﬂuence of Institutions on Political Elites’ Democratic Values in Germany.” American Political Science Review 89: 927-941. 153 Rohrschneider, Robert. 1996. “Institutional Learning Versus Value Diffusion: The Evolu- tion of Democratic Values Among Parliamentarians in Eastern and Western Germany.” Journal of Politics 58: 442-446. Rueschemeyer, Dietrich, Evelyne Rueschemeyer, and John D. Stephens. 1992. Capitalist Development and Democracy. Chicago: University of Chicago Press. Rustow, Dankwart. 1970. “Transition to Democracy: Toward a Dynamic Model.” Compar- ative Politics 2 (April): 337-364. Rustow, Dankwart A. 1990. “Democracy: A Global Revolution?” Foreign Affairs 69 (4): 75-91. Ryu, Yongwook. 2013. “Does Poltiical Generation Matter for Foreign Policy? South Ko- rea’s Foreign Policy Changes, 2003 to 2007.” The Korean Journal of International Stud- ies 11 (1): 113-141. Schedler, Andreas. 2002. “The Menu of Manipulation.” Journal of Democracy 13 (2): 36- 50. Schedler, Andreas. 2006. “The Logic of Electoral Authoritarianism.” In Electoral Authori- tarianism, ed. Andreas Schedler. Boulder, CO: Lynne Rienner. Scott, James C. 1985. Weapons of the Weak. New Haven: Yale University Press. Seidman, Gay. 1994. Manufacturing Militance: Workers’ Movements in Brazil and South Africa, 1970-1985. Berkeley: University of California Press. Shim, Jae-Seung, and Moosung Lee. 2008. The Korean Economic System: Governments, Big Business, and Financial Institutions. Aldershot: Ashgate Publishing. Shin, Gi-Wook, and Paul Y. Chang. 2011. South Korean Social Movements: From Democ- racy to Civil Society. London: Routledge. Shin, Gi-Wook, Paul Y. Chang, Jung-eun Lee, and Sookyung Kim. 2007. “South Korea’s Democracy Movement (1970-1992): Stanford Korea Democracy Project Report.” Ko- rea Democracy Foundation and the Shorenstein Asia Paciﬁc Research Center, Stanford University, Research Report: 1-119. Shin, Gwang-Yeong. 1994. Gegeupgwa nodongundongui sahoehak [The Sociology of Class and Labor Movements]. Seoul: Nanam. Shin, Myungsoon. 1982. “Effectiveness of Political Protest in Korean Politics.” Korean Political Science Review 16: 25-44. Shin, Myungsoon. 1987. “Analysis of the Nature of the Student Movement in the 1980s: A Study on College Students’ Attitudes Toward Student Movement.” The Journal of Asiatic Studies 30 (1): 109-149. 154 Simpser, Alberto. 2005. “Making Votes Not Count: Strategic Incentives for Electoral Cor- ruption.” Ph.D. diss. Department of Political Science, Stanford University.

Sirowy, Larry, and Alex Inkeles. 1990. “The Effects of Democracy on Economic Growth and Inequality: A Review.” Comparative International Development 25: 126-157.

Slater, Dan. 2010. Ordering Power: Contentious Politics and Authoritarian Leviathans in Southeast Asia. New York: Cambridge University Press.

Snow, David A., Jr. Louis A. Zurcher, and Sheldon Ekland-Olson. 1980. “Social Networks and Social Movements: A Microstructural Approach to Differential Recruitment.” Amer- ican Sociological Review 45: 787-801.

Snow, David A., Rens Vliegenthart, and Catherine Corrigall-Brown. 2007. “Framing the French Riots: A Comparative Study of Frame Variations.” Social Forces 86 (2): 385-415.

Song, Kon-Ho, Min-Ji Choi, Ji-Dong Park, Deok-Han Yoon, Seok-Chun Son, and Myung- Gu Kang. 2000. Understanding 100 Years of Korean Media [Han’guk ollonˇ paro pogi 100-yon]ˇ . Seoul: Tasotˇ Sure.

Stockmann, Daniela, and Mary E. Gallagher. 2011. “Remote Control: How the Media Sustain Authoritarian Rule in China.” Comparative Political Studies 44 (4): 436-467.

Strang, David, and Sarah A. Soule. 1998. “Diffusion in Organizations and Social Move- ments: From Hybrid Corn to Poison Pills.” Annual Review of Sociology 24: 265-290.

Suh, Doowon. 2001. “How do Political Opportunities Matter for Social Movements? Politi- cal Opportunity, Misframing, Pseudosucess, and Pseudofailure.” The Sociological Quar- terly 42 (3): 437-460.

Suh, Sang Chul. 1978. “Policies for Industrialization and Regional Development.” The Journal of Asiatic Studies 21 (2): 1-32.

Sükösd, Miklos. 2000. “Democratic Transformation and the Mass Media in Hungary: From Stalinism to Democratic Consolidation.” In Democracy and the Media, ed. Richard Gun- ther and Anthony Mughan. Cambridge and New York: Cambridge Univ Press.

So,ˇ Chung-sok.ˇ 2011. 6-wolˇ hangjaeng: 1987-yonˇ minjung undong uiˇ changomˇ han p’anorama [The June Democratic Struggle: 1987 Minjung Movement’s Grand Panorama ]. Paju: Tolbegae.

Therborn, Göran. 1977. “The Rule of Capital and the Rise of Democracy.” New Left Review 103 (May-June).

Therborn, Göran. 1979. “The Travail of Latin American Democracy.” New Left Review 113 (January-April): 71-109.

155 Thompson, E. P. 1966. “The Pecularities of the English.” In Socialist Register, 1965, ed. Ralph Miliband and John Saville. London: Merlin Press. Tilly, Charles. 1978. From Mobilization to Revolution. Addison-Wesley Publishing Com- pany. Treisman, Daniel. 2014. “Income, Democracy, and Leader Turnover.” American Journal of Political Science. Trejo, Guillermo. 2014. “The Ballot and the Street: An Electoral Theory of Social Protest in Autocracies.” Perspectives on Politics 12 (June): 322-352. Tucker, Joshua A. 2007. “Enough! Electoral Fraud, Collective Action Problems, and Post- Communist Colored Revolutions.” Perspectives on Politics 5 (3): 535-551. Valenzuela, J. Samuel. 1989. “Labor Movements in Transitions to Democracy: A Frame- work for Analysis.” Comparative Politics 21 (July): 445-472. VanderWeele, Tyler J. 2014. “A Uniﬁcation of Mediaton and Interaction: A 4-Way Decom- position.” Epidemiology 25 (5): 749-761. Vansteelandt, Sijin. 2009. “Estimating Direct Effects in Cohort and Case-Control Studies.” Epidemiology 20 (6): 851-860. Voss, Kim. 1993. The Making of American Exceptionalism: The Knights of Labor and Class Formation in the Nineteenth Century. Ithaca, NY: Cornell University Press. Wasserstrom, Jeffrey N. 1991. Student Protests in Twentieth-Century China: The View from Shanghai. Stanford, Calif.: Stanford University Press. Watanabe, Haruko. 1964. “South Korean Press, 1945–.” Freedom of Information Center Publication 119. Whitehead, Laurence. 1996. “Three International Dimensions of Democratization.” In The International Dimensions of Democratization, ed. Laurence Whitehead. Oxford and New York: Oxford University Press. Woo-Cummings, Meredith. 1999. The Developmental State. Ithaca, NY: Cornell University Press. Wright, Joseph. 2011. “Electoral Spending Cycles in Dictatorships.” Unpublished manuscript, Pennsylvania State University. Xiang, Jun. 2010. “Relevance as a Latent Variable in Dyadic Analysis of Conﬂict.” Journal of Politics 72 (2): 484-498. Yi, Chaeo. 2011. Han’guk haksaeng undongsa: 1945-1979-yonˇ [History of Korean Student Movements: 1945-1979]. Seoul: P’ara Puksu.ˇ 156 Yi, Gyong-Jae.ˇ 1987. “Minjungui sungni: 5.17eso 6.29kkaji [People’s Victory: from 5.17 to 6.29].” Shin Tonga August.

Youm, Kyu Ho. 1996. Press Law in South Korea. Ames, Iowa: Iowa State University Press.

Yu, Je-ch’ol.ˇ 2010. “1970yondaeˇ yosˇ ongnodongjaˇ uiˇ yogasiganˇ ulˇ tullossanˇ t’ujaeng [Women Workers’ Struggle Surrounding Leisure Time in the 1970s].” Sahoewa yoksaˇ [Society and History] 85: 53-82.

Zeileis, Achim, Christian Kleiber, and Simon Jackman. 2008. “Regression Models for Count Data in R.” Journal of Statistical Software 27 (8).

Zhao, Dingxin. 1998. “Ecologies of Social Movements: Student Mobilization during the 1989 Prodemocracy Movement.” American Journal of Sociology 103 (May): 1493-1529.

Zhao, Dingxin. 2001. The Power of Tiananmen: State-Society Relations and the 1989 Bei- jing Student Movement. Chicago: University of Chicago Press.

Ziblatt, Daniel. 2006. “How Did Europe Democratize?” World Politics 58: 311-38.

157