Communist and Post-Communist Studies xxx (2014) 1e12

Contents lists available at ScienceDirect

Communist and Post-Communist Studies

journal homepage: www.elsevier .com/locate/postcomstud

North Korea and diversion: A quantitative analysis (1997e2011)

Robert Daniel WallaceCenter for Strategic Intelligence Research, National Intelligence University, 200 MacDill Blvd Southeast, Washington, DC, USA

a r t i c l e i n f o

Article history:Available online xxx

Keywords:North KoreaConflictDiversionary theoryEvent dataQuantitative analysis

1 Diversionary theory contends that leaders, in tdomestic to international issues. This concept has aforce, or diversionary foreign policy (Levy, 1989; Mi

http://dx.doi.org/10.1016/j.postcomstud.2014.04.0040967-067X/Published by Elsevier Ltd on behalf of T

Please cite this article in press as:Wallace, RPost-Communist Studies (2014), http://dx.

a b s t r a c t

In this research I propose that the concept of diversionary theory provides at least a partialexplanation for North Korea’s conflict activities. I examine and analyze the country’s dataon diplomatic and military activities from 1997 to 2011 and argue that North Korea’sdomestic conditions influence its willingness to engage in external conflict. I also examinethe impact of such external influences as UN sanctions, leadership changes in the region,national capacities of the US, South Korea and Japan, and strategic military exercises onDPRK-initiated conflicts. This study provides insight into the activities of this reclusivestate and also demonstrates useful techniques that can be applied to analyze other simi-larly closed nations. The findings suggest that there are identified links between internalconditions and the Kim regime’s aggressive actions between 1997 and 2011 in support ofthe diversionary argument. Concurrently, there is less evidence that North Korea’s hostilediplomatic and military activities are based on external pressures.

Published by Elsevier Ltd on behalf of The Regents of the University of California.

1. Introduction

North Korea presents a unique dilemma for the security of East Asia (and the international community) as both a poverty-stricken nation and a nuclear-armed state. The North Korean government routinely uses both diplomatic threats and armedforce to advance its agenda, but the Kim regime’s closed nature makes empirical research on its actions difficult. Scholarsrarely attempt to analyze the actions of North Korea using systematic methods, and most available research on this stateconsists of qualitative or policy-driven studies. Yet data is available that potentially helps understand why the ruling Kimregime routinely uses conflict actions to achieve national and foreign policy goals. By analyzing data between 1997 and 2011, Ifound that North Korea’s use of hostile foreign policy was influenced by internal conditions faced by the Kim regime whichsupported diversionary theory.1 Concurrently, I argue that external conditions are less important factors for Kim regime’sdecisions to engage in conflict. Most international relations scholars readily admit that domestic factors influence foreignpolicy, yet the linkages between internal conditions and international actions by authoritarian states are often difficult toidentify. This research seeks to address this gap in knowledge about the world’s most secluded state, the Democratic People’sRepublic of Korea (DPRK).

This paper is organized as follows. In the first section, I discuss diversionary theory and derive two hypotheses. These arebased on the relationships between DPRK-initiated conflicts and the conditions faced by the North Korean regime (internal

imes of crisis will commit their nations to external use of force to alleviate the national focus fromnumber of names such as the diversionary theory of war, the diversionary hypothesis, diversionarytchell and Thyne, 2010; Kisangani and Pickering, 2007; Oakes, 2006).

he Regents of the University of California.

.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist anddoi.org/10.1016/j.postcomstud.2014.04.004

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e122

and external pressures). The next portion identifies and defines both the dependent variable (Kim regime’s hostile foreignpolicy activities) and independent variables (internal and external conditions). The third part explains the conflict dataanalysis of North Korean hostile foreign policy; and the fourth portion provides the overall statistical analysis of the data. Inthe final section, I discuss both the relationship between the analytical results and diversionary theory and the overallcontribution of this research project.

2. Diversionary concepts and derived hypotheses

There is an abundance of historical anecdotes surrounding the use of diversion by national leaders to solve domestictensions. Notable figures such as Machiavelli, Shakespeare, and Jean Bodin have commented on the unifying potential ofdiversionary foreign policy activity.2 Additionally, one of themost famous interstate conflicts, the Russo-JapaneseWar (1904e1905), began purportedly because Russia needed a “short victorious war to stem the tide of revolution” (Walder, 1974). Morerecent conflicts associated with diversionary force include the 1982 Falklands War, US invasions of Iraq in 1991 and 2003, US1998 missile attacks against terrorist targets in Afghanistan and Sudan, and the conflict between Russia and Georgia in 2008(Oakes, 2006; Kisangani and Pickering, 2009; Oreskes, 1990; Milbank, 2002; Purdum, 1998; Filippov, 2009).3

For North Korea, diversionary behavior offers the potential advantages of unifying the public against a commonfoedgenerally the United States or South Koreadand a means for the government to direct public attention from domesticdifficulties.4 While some may argue that authoritarian governments have little use for diversion, since these types of regimestend to limit the activities of the public (Peceny and Beer, 2003), other scholars suggest that diversionary activities arecommon among many states and North Korea is no exception (Mitchell and Prins, 2004; Reilly, 2004).

North Korea’s foreign policy activities, as with any other state, range from cooperation to conflict relationships with othernations.5 Yet the international community is most concerned with Pyongyang’s provocative foreign policy actions, whichseem to demonstrate the Kim regime’s willingness to risk war to achieve foreign policy goals. This “hostile foreign policy”(HFP) includes activities such as clandestine infiltrations into South Korea, nuclear weapons testing, and diplomatic threats.To analyze these potential diversionary activities, I formed the two hypotheses. The first contends that domestic factors causethe Kim regime to engage in hostile foreign policies.6 The second supports a systemic view of the influences on state’sbehavior and argues that the international community is more influential in North Korea’s foreign policy choices.7 Theseconcepts are expressed as follows:

Hypothesis 1. (H1): Domestic difficulties in North Korea cause an increase in hostile foreign policy initiated by this country.

Hypothesis 2. (H2): External actors and increased international tensions cause increases in hostile foreign policy initiated byNorth Korea.

3. The variables

North Korea’s hostile foreign policy constitutes the dependent variable for this research, while its domestic conditions andinternational influences are the independent variables. To address alternative explanations for North Korean conflict activitiesI include two control variables: the type of South Korean administration and trade levels between North Korea and China.

3.1. The dependent variable: hostile foreign policy actions

The dependent variable used in this analysis is hostile foreign policy (HFP). I define HFP as domestic or international actionsby governments or government-sanctioned entities intended to negatively influence or detrimentally affect a target state

2 Machiavelli (1882) was referring to diversionary behavior when he observes “. the present king of Spain . attacked Granada. [and] kept the noblesof the Castile occupied with this enterprise, and, their minds being thus engaged by war, they gave no attention to the innovations introduced by the king,who thereby acquired a reputation and an influence over the nobles without their being aware of it.” In his play King Henry IV, Shakespeare (1823)comments, “Be it thy course to busy giddy minds with foreign quarrels; that action, hence borne out, may waste the memory of former days.” Addi-tionally, Bodin (1955) stated, “The best way of preserving a state, and guaranteeing it against sedition, rebellion and civil war is to keep the subjects in amityone with another, and to this end, to find an enemy against whom they can make common cause.”

3 Many other conflicts have been associated with diversionary war such as the Franco-Prussian Wars, World War I, the United States’ war with Vietnam,and the 1996 conflict in Rwanda (Mayer, 1969; Lenin, 1930; Fordham, 1998; Pickering and Kisangani, 2005).

4 North Korea remains one of the poorest nations in the region ($1800 per capita GDP) and has been historically reliant on international aid to feed 10e15% of its people (CIA, 2013; Manyin and Nikitin, 2013; UN FAIS, 2012). At the same time, the North Korea’s communist leaders exercise expansive controlover the social, political, and economic aspects of North Korean society (Gause, 2012). North Korea is arguably the world’s most repressive police state, yetremains reliant on the international community to survive.

5 For example, economic cooperation does occur and North Korea has gone to considerable effort to establish and maintain special economic zones withboth China and South Korea (Lankov, 2011).

6 A number of scholars argue that domestic interests are related to external conflict activities (Fearon, 1994; Morgan and Bickers, 1992; Putnam, 1988;Bueno de Mesquita and Lalman, 1992; Bueno de Mesquita, 1985).

7 This supports the “traditional theorist” view of interstate activity (Waltz, 1954, 1979; Wendt, 1992) and focuses on external causes of conflict such as theinternational system or a state’s perception of the international environment.

Please cite this article in press as: Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e12 3

through diplomatic, social, economic, or military activities ranging from provocative statements to hostile acts in support ofnational or regime goals.

This definition of HFP focuses on foreign policy activities that are intended to have detrimental effects on the target state.These include actions that have direct impacts, such asmilitary conflict, or lesser activities that are intended to influence otherstates, including propaganda statements or an unwillingness to cooperate. Additionally, these activities span the full range ofleadership actions and include diplomatic actions by the government or its representatives, economic measures with local orinternational effects, and military activities such as exercises or actual hostilities. Finally, these actions are linked to thenational leadership and efforts to retain both sovereignty and power: these are deliberate activities in support of national orregime objectives.

In the case of North Korea, this definition provides for the inclusion of a wide range of hostile foreign policy, such as NorthKorea’s routine use of propaganda statements and aggressive policy announcements, which often spur international concern.An example of a diplomatic HFP would be North Korea’s threat to turn Seoul into a “sea of fire”, which was a warning of DPRKintentions to use nuclear weapons against the Republic of Korea (ROK) (Financial Times,1994).8 Other provocative statementsfrom the Kim leadership on the security situation (such as North Korea’s annual “New Year’s Statement”)9 often includethreats from the civilian leaders of the DPRK against both the US and South Korea.

Economic events that fit the category of “hostile foreign policy” often include North Korean actions involving joint eco-nomic ventures between North and South Korea, such as the Kaesong Industrial Complex and the Mount Kumgang TouristArea, both located in North Korea. In August 2011, North Korea seized all South Korean assets at the Mount Kumgang andended its joint venturewith the ROK at that location (Chosun Ilbo, 2011a; Chosun Ilbo, 2011b).10 Finally, North Korea’smilitaryactivities are the most visible of its HFP events and range from DMZ incidents and naval clashes to nuclear tests. Yet there arelimits to North Korea’s hostile actions, as the Kim regime does not intend to start a second Koreanwar.11 Instead, North Koreanleaders are most likely focused on shaping the overall security environment to retain the DPRK’s national sovereignty and toensure the Kim regime maintains control over its government and society.

3.2. The independent variables: conditions faced by the regime

The independent variables I use to analyze these events are divided into domestic difficulties and external influences. SouthKorean defense officials note that both are influential in North Korean aggressive foreign policy choices (KNDU, 2011).12

3.2.1. Domestic difficulties (Internal Conditions)I define domestic difficulties as the actual or perceived political, social, or economic conditions that negatively affect a given

society, including conditions that interfere with the regime’s ability to govern and maintain control over its population. Thisdefinition assumes that certain internal conditions, which threaten the Kim regime’s sovereignty or ability to govern, couldpotentially spur domestic unrest and internal instability. This research proposes that North Korea’s domestic conditions arepotentially the influential triggers for hostile actions. If these internal conditions become unstable or deteriorate, the resultanteffects will negatively influence North Korean society and conflict actions will increase.

The first set of independent variables is related to domestic difficulties and support Hypothesis 1. These include indicators ofdomestic instability (political, economic, and social) and their relationships to hostile foreign policy actions. These variableshave been measured per capita using North Korean population statistics (USG, 2013) and discussed in the following section.

3.2.1.1. Political instability. For North Korea, political conditions that fall into the category of domestic difficulties include regimeinstability or change and elite or mass unrest. Claude Ake (1975) defined political stability as “the extent that members ofsociety restrict themselves to the behavior patterns that fall in the limits imposed by political roles expectations. Any act thatdeviates from these limits is an instance of political instability” (Ake, 1975). Gates et al. (2006) add that it requires “insti-tutional consistency” and that both autocracies and democracies have inherent characteristics that “self-enforce” politicalstability.

The Kim regime hasweathered two generations of familial succession (Kim Il-sung to Kim Jong-il and now to Kim Jong-un)and conducted a number of purges to consolidate power over the years, often associated with political or military conflict.

8 In reaction to ROK military exercises in 2011, Pyongyang threatened direct attacks against the South Korean president (DPRK Radio, 2011).9 This annual message, broadcast in North Korea and internationally via national television and print media, has been compared to the United States’

yearly “State of the Union” address. It is intended for both domestic and international audiences and outlines key areas of progress and concern. Forexample, the 2011 message “warned the South’s [ROK] government to stop what it called ‘north-targeted moves’ and a ‘smear campaign’ against it [DPRK]”(Ramstad, 2011).10 In 2008, North Korean guards killed a South Korean tourist for “trespassing in a military-controlled area”; the Kim regime has linked its seizure of SouthKorean assets to events surrounding that incident (Chosun Ilbo, 2011a). In 2011, North Korea began soliciting businesses in other nations (including China,Japan, and the US) to provide assistance in restarting tours to the area (Chosun Ilbo, 2011b). While North Korea’s actions are partially inspired by its owneconomic objectives, the killing of the South Korean tourist became an international incident that triggered the closure of this resort and both negativeactions and proclamations by North Korea. This incident demonstrates the complexities of NortheSouth relations.11 War on the Korean peninsula would yield catastrophic effects and most analysts agree that it would result in the demise of the Kim regime.12 In this study, I use quantitative methods similar to other research on authoritarian governments (Weeks, 2011; Pickering and Kisangani, 2010; Lai andSlater, 2006; Peceny and Butler, 2004).

Please cite this article in press as:Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e124

Another influential event for the DPRK would be a leadership crisis, which can potentially be associated with increases inconflict activities. These include regime leadership changes (in 1994 and 2011), purges by the Kim regime (in the 1960s and1990s), and other leadership efforts to consolidate power.13 The analysis of these events and the presence of heightened (ordecreased) levels of internal or external conflict might provide important insight to relationship between regime change orconsolidation and the hostile foreign policy activities by the DPRK.

The first measure for political instability is a component of Kaufmann’s (2009)World Governance Indicators (WGI) project.The WGI includes an index that measures political stability and the level of violence (designated “PV”) and includes data onNorth Korea.14 Kaufmann (2009) defines PV as “perceptions of the likelihood that the government will be destabilized oroverthrown by unconstitutional or violent means, including politically-motivated violence and terrorism.” This definitionprovides a measure of the stability of the North Korean government which has had a reported history of political violence and(to a lesser extent) political instability (Hawk, 2003; Kang, 2002; Hassig and Oh, 2009). The second measure of politicalinstability is Ishiyama’s (2012) elite volatility scores, which measures how the “political elite” or key leaders that surroundedKim Jong-il and his successor Kim Jong-un have changed over time. Ishiyama (2012) measures this by analyzing the atten-dance of senior North Korean leaders on inspection visits conducted by Kim Jong-il. Changes to these attendees potentiallyindicate the overall composition of the Kim regime’s inner circle and provide a proxy measure for political stability.

An additional proxy measure of political stability is North Korea’s national capacity based on the Composite Indicator ofNational Capabilities (CINC) index from the Correlates of War Project (COW, 2012). The CINC database uses a number of mea-sures to determine overall capabilities including military personnel, defense expenditures, population levels, iron, steel, andenergy consumption (CINC, 2012). This measure has not been used previously to gauge North Korea’s internal stability, yet theCINC researchers note that the components of this measure (demography, industrial capacity, and military characteristics)“reflect the breadth and depth of the resources that a nation could bring to bear in instances of militarized disputes”(NMC,2005).15 These samecapabilities allowtheKimregime to remain inpowerand limit thepotential forpoliticaldissent, thusincreased CINC scores for North Korea equates to enhanced power for the regime and less possibility of political instability.16

3.2.1.2. Economic instability. Economic conditions are also useful indicators of internal stability. For North Korea, these aremeasured using proxy data to include the rise and fall of the DPRK’s gross domestic product (GDP) and the balance of tradewith foreign countries (how much the DPRK imports and exports). These two indicators help determine the health of NorthKorea’s economy and relative dependence on external nations to support its domestic material needs. For this measure I usegross domestic product growth and total trade. GDP growth (calculated year to year) provides a relatively simple measure ofthe wealth and economic health of the DPRK.17 The Bank of Korea (BOK, 2012) routinely monitors North Korea’s economy andthe BOK’s historic estimates of GDP growth are included for this analysis. The other indicator for the performance of the NorthKorean economy is the level of imports and exports, which has been used by other scholars to similarly estimate DPRKeconomic performance (Davies, 2002; Eberstadt, 1997).18

3.2.1.3. Social instability. I measure social instability by analyzing the level of dissatisfaction that the DPRK citizens feel to-wards the ruling government. While information on civic unrest in North Korea is difficult to obtain, the numbers of refugeesthat flee North Korea provides a useful proxy for public dissatisfaction. Data for other social indicators, such as infant mor-tality and food availability, also provide insight into the stability of DPRK society from a citizen’s perspective and are used forthis study.

The first measure of social instability of North Korean society is measured bywillingness of individuals to fleeNorth Korea.Choosing to leave North Korea for China or the ROK is a “drastic act” since the “defectorwould knowall toowell that his familymembers remaining behind quite likely would be sent off to political prison camps, perhaps for the rest of their lives.”(Martin, 2006). Yet on a yearly basis, hundreds (and sometimes thousands) of North Koreans have chosen to leave theDPRK for South Korea (ROK MOU, 2012) and I use this data as a proxy measure of social stability.19

13 Michishita (2010) noted that many instances of efforts to consolidate power by the Kim regime from 1960 to 2008, were associated with conflictactivity.14 The World Governance Index (WGI) (2012) attempts to measure the level of “good” governance using “six dimensions of governance: Voice andAccountability, Political Stability and Absence of Violence/Terrorism, Government Effectiveness, Regulatory Quality, Rule of Law, and Control of Corruption”(Kaufmann, 2009). This report includes data beginning in 1996 through 2008 for 212 territories and countries. Although the timeframe of the WGI datalimits its explanatory capability, this index has the advantage of being one of the only structured measures of political stability applied to North Korea.15 While this measure is not generally used to measure political stability, it does provide, at least in North Korea’s case, a measure of the Kim regime’sability to remain stable and in power.16 A number of authors use mass and elite unrest in a particular society as additional measures of political instability. This technique proposes that theinfluencers of social unrest include the level of dissatisfaction that both the elites (privileged members) of society as well as the masses (ordinary citizen)feel towards the ruling government. Kisangani and Pickering (2009) defined elite unrest in terms of “government crises and purges” and mass unrest ascharacterized by “general strikes, riots, and anti-government demonstrations” in their analysis of the effects of diversionary military activities.17 For this research, I assume that yearly increases and decreases in GDP per person indicate whether North Korea’s national economy is growing orshrinking over time.18 For this data, I use the Correlates of War database information on trade (COW Trade, 2012) and the total trade was computed as the yearly sum of allexports and imports. This information is supplemented by CIA data for the years 2010 and 2011 (CIA, 2013).19 North Korean refugee information is available on a yearly basis from the South Korean government (ROK MOU, 2012).

Please cite this article in press as: Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e12 5

Infant mortality rates (IMR) provide a second proxy for the overall health of North Korean society. For example, Abouhardand Kimball (2007) note that IMR can be used to “measure development and indicate the extent to which governmentsprovide for the economic and social welfare of their citizens, both of which are correlates of conflict.” North Korea experi-enced declining rates of infant deaths between the 1960s and late 1980s, but after the end of the Cold War and during itsfamine period in the 1990s, the rates began to fluctuate (World Data Bank, 2012; UN Population, 2012).20 I rely on the WorldBank’s World Development Indicators (World Data Bank, 2012) to measure IMR in North Korea.

The final measure of social stability involves food availability in North Korea with the assumption that if adequate sus-tenance is available, the public is more content with the DPRK leaders and their government. This became an acute issueduring the 1990s, when North Korea experienced a severe famine in which up to a million deaths occurred as a result of foodshortages (Haggard and Noland, 2007; Hassig and Oh, 2009). Thus, North Korea’s ability to provide food, either throughdomestic production or international aid, is a stabilizing factor in DPRK society and used as a proxy measure for social sta-bility. This data is available from the Food and Agriculture Organization of the United Nations (FAOSTAT, 2012).21

3.2.2. External conditionsThe second category of independent variables examines the pressure exerted on North Korea by the international com-

munity in attempts to change the Kim regime’s behavior. Indicators of these external conditions include international re-actions to North Korean provocative activities such as United Nations resolutions, national capabilities of other regionalactors, election cycles of externally influential states (US, ROK, and Japan), and the incidence of ROK-US alliance militaryexercises. These conditions often result in political rhetoric and occasional military responses from the DPRK.

3.2.2.1. UN Security Council Resolutions. North Korea’s actions have often resulted in international responses, the most sig-nificant being the UN reaction to the North Korean invasion of South Korea in 1950. North Korea’s decisions to engage inaggressive foreign policy andmilitary activitymight be influenced by these types of international responses to its own activityand UN Security Council resolutions.22

3.2.2.2. National capacity scores. The CINC (2012) scores from South Korea, the US, and Japan are additional measures ofpossible external “influencers” on DPRK conflict activities. During the 1990s, North Korea’s efforts to ensure its sovereigntyresulted in a number of measures, including reposturing of DPRK military forces, ballistic missile and nuclear technologydevelopment, and the initiation of “Songun” (North Korea’s “military-first” policy). Given this emphasis by North Korea andthe strong regional military alliances (US-ROK and US-Japan) faced by the Kim regime, the national capacity and defensecapabilities of the US, South Korea, and Japan might be influential in DPRK decisions to pursue hostile foreign policies. Thus,changes in the CINC scores might be correlated with DPRK conflict activities.

3.2.2.3. Leadership changes. I also examine another indicator, external state leadership changes for South Korea, the US andJapan, to determine if there is a correlation between these events and North Korean hostile foreign policies. While there areseveral studies on the relationship between national elections and leadership decisions to use force, much of the research hasprimarily focused on the US and presidential decisions to use force in relation to a US election cycle (Ostrom and Job, 1986;James and Oneal, 1991; Meernick, 1994).23

3.2.2.4. Military exercises. Finally, I consider the impact of ROK-US strategic-level military exercises as a measure of externalconditions as these types of military exercises often cause intense reactions from North Korea. In fact, one of the negotiatingdemands from North Korea during the 1993 nuclear crisis was the cancellation of the US-ROK “Team Spirit” military trainingexercises (Sigal, 1998). For example, the DPRK stated the following in reaction to joint ROK-US exercises in August 2011: “TheKorean peninsula is faced with the worst crisis ever. An all-out war can be triggered by any accidents” (The Telegraph, 2011).Limited publically available research exists on the effects of ROK/US military exercises on North Korean behavior and noquantitative studies of North Korea (with the exception of D’Orazio, 2012)24 incorporate this variable.25

20 More information on the causes of the increase in infant mortality during the 1990s is available in case studies on North Korea’s famine (Natsios, 2001;Haggard and Noland, 2007).21 Food availability is calculated per capita using US census data (US Census Bureau, 2012).22 Chapman and Reiter (2004) use this indicator to measure international involvement in crises. This concept proposes that the increased internationalattention given to North Korea because of a new resolution might spur DPRK choices to use diversionary force. Ongoing resolutions (and economicsanctions) are not considered to have similar impacts on the Kim regime. These resolutions are coded as binary variables: quarters that include theenactment of a new resolution are coded as “1,” and all others are coded as “0.”23 Leadership changes are coded “1” for the quarter in which they occur (based on the election date) and periods that include no change are coded “0.”24 D’Orazio (2012) examines joint US-ROK military exercises between 1998 and 2010, and found that they “do not trigger a systematic escalation inconflictual rhetoric or behavior” by North Korea.25 Military exercises at the strategic level are those which employ “instruments of national power in a synchronized and integrated fashion to achievetheater, national, and/or multinational objectives” and operational exercises are those which link the “tactical [combat] employment of forces to nationaland military strategic objectives” (US Department of Defense, 2011). Military war games such as the now defunct “Team Spirit” and ongoing Ulchi FreedomGuardian exercise fit this category of strategic events. Strategic military exercises jointly conducted by the ROK and US in a given quarter are coded as “1” forexercise periods (and “0” for non-exercise periods).

Please cite this article in press as:Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e126

3.2.3. Controlling for ROK administration type and China tradeAn alternative explanation for the initiation of diplomatic or military conflict by the Kim regime might be the influence of

the political characteristics of the ruling party in South Korea. Generally, conservative ROK governments have traditionallytaken a more “hardline” approach towards the DPRK while liberal administrations have been more open to engagement withthe North. North Korean conflict might have been more prevalent during conservative ROK administrations, possibly becauseduring these times, South Koreawas less engaging and more hostile towards the Kim regime. This might have caused conflictwith the DPRK because of North Korea reacting to perceived threats from the ROK.26

To address the economic influence the PRC exerts upon the DPRK, I add a control variable to examine the relationshipbetween ChinaeNorth Korean trade and conflict levels. For example, in 2011 North Korea’s trade with China increased from$3.5 billion to $5.6 billion and accounted for over 70 percent of all DPRK trade that year, up from 57 percent in 2010 (Yoon,2013). This trade relationship might influence North Korea’s proclivity to engage in hostile foreign policy activities and wasincluded as the second control variable.27

4. Research design and methods

This research uses time-series event data analysis of incidents of DPRK hostile foreign policy to examine the relationshipsbetween North Korean HFP and conditions faced by the Kim regime between 1997 and 2011. These events provide the data forexamining North Korea’s foreign policy activities using the concept of “longitudinal history” (Allison, 1984).28 Additionally, Iuse regression analysis to examine the relationship between internal and external conditions faced by the Kim regime and itsconduct of hostile foreign policy activities.

4.1. Event data analysis and scoring

Event data is typically taken from public sources (newspapers, journals, and broadcasts) and “constitute an imperfectsummaryof the events. coverage is not uniform, and it varies according to the needs of the reporters rather than the scholarlyneed for representativeness” (King and Lowe, 2003). Unfortunately, no databasewas available to support this research,29 thus Iconstructed a newdatabase using event data fromUS, ROK, andNorthKorean government sources. Rather than code all eventsin a similar manner, I sought to grade each event by “intensity” to account for variations in conflict events. For example, afirefight at the DMZ is not given the sameweight as the sinking of an ROK navy ship or the threat of nuclearwarfare. To addressthis issue, I categorized and scored North Korea’s hostile foreign policy events using Azar’s (1993) Conflict and Peace Databank(COPDAB) definitions and scalingmethodology,which scores hostile events between 0 (neutral) and 102 (full-scalewarfare).30

While many other event data research projects rely upon machine-coded31 information (King and Lowe, 2003; Bond et al.,2000), this research uses hand coding of events between 1997 and 2011.32 For the analysis of North Korean events, handcoding provides advantages such as ability to integrate data frommultiple languages and formats as well as judgments madebased on my own analysis (rather than automated retrieval) to determine the applicability of specific events.33

4.1.1. Database constructionResearch on a closed state can be a difficult task and North Korea is no exception. Data from the North Korean gov-

ernment is often either not available or highly suspect and any study of the DPRK must incorporate reporting from

26 South Korean administrations, such as Kim Dae-jung (1998e2003) and Roh Myoo-hung (2003e2008), are coded “0,” representing more liberal gov-ernments. The Lee Myun-bak administration began in 2008 and is coded as “1” indicating a more conservative South Korean government in power.27 China-DPRK trade data is available from the UN Comtrade database (2013) and calculated per DPRK citizen based on US Census (2012) data.28 Longitudinal research involves the collection of data across time, using the same methods to collect and analyze data, and comparisons within or acrossthe timeframe under consideration (Menard, 1991).29 For example, Arthur Banks’ (2004) Cross-National Time-Series (CNTS) Data Archive has reporting on internal instability (including purges, strikes andriots) for many nations and has been used by a number of scholars (Pickering and Kisangani, 2005; Bell, 2009; Tir, 2010) to examine diversionary cases.Additionally, the Correlates of War (COW MID, 2012) dataset is another conflict database often used by scholars to analyze similar data (Leeds and Davis,1997; Crescenzi et al., 2008). Finally, Bond et al., 2000 conduct automatic coding of DPRK event data between 1990 and 2000. Unfortunately, the CNTS,COW, and Bond databases do not contain adequate information on DPRK conflict and instability required to support event data analysis from 1997 to 2011.30 This method is based on Nincic (1975), Azar (1982), McClelland (1999), and Goldstein (1992).31 This method of research involves automated scanning of media for specific keywords and recording events based on that information. King and Lowe(2003b) have used this technique to construct a database of over 10 million dyadic interstate activities from 1990 to 2004, coded into 200 event types(using a typology based on the WEIS methodology).32 Hand coding refers to the selection of events by an individual researcher or group of researchers without the aid of automated scanning processes.Although keyword searches were used to support this research, I analyzed each event individually and coded accordingly.33 King and Lowe (2003a) ran an experiment comparing the effectiveness of a computerized data collection program to manual efforts using Reuter’sreports and found that the automated system “was able to extract information from Reuter’s news reports on a level equal to trained Harvard un-dergraduates.” Yet the diverse nature of the data sources on North Korea, redundancies due to multiple reports on the same event, and analysis requiredprior to categorization all required hand coding for accuracy.

Please cite this article in press as: Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

Fig. 1. DPRK hostile foreign policy event levels (1997e2011).Sources include: USFK (2012), UNCMAC (2012), KINU (2011), Fischer (2007), KCNA (2012), and New York Times Historical Archives 1997e2011 via Lexis-NexisAcademic Search (2012).

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e12 7

external sources.34 Fortunately, the dependent variable data (HFP events) is available from US and South Korean sources,and from official government pronouncements from North Korea. Much of the other data was available from “mirrorstatistics” and relied upon information on DPRK from other states. While data limitations are a concern, one of theoverarching goals of this research is to demonstrate the usefulness of a systematic method to examine closed states.Despite these shortcomings, the data used in this project is the best public information available on North Koreanactivities.

I use a number of primary sources of information to construct the event database used in this analysis. This includeshistoric records and reports from a USFK historian (USFK, 2012), the United Nations Military Armistice Command (UNCMAC,2012), South Korea’s Korean Institute for National Unification (KINU, 2011), and Fischer’s (2007) CRS Report. Additionally, I usemedia reports from several sources such as the KCNA website (2012), New York Times Historical Archives via Lexis-NexisAcademic Search. This resulted in a dataset with diverse and overlapping sources, which helped ensure the fidelity of theevent data. Most of the reports were available in both English and Korean, but 41% were available only in Korean and requiredtranslation.35 The resulting database, after cross-checking for duplication, includes 1108 events.36

4.1.2. DPRK conflict activitiesFig. 1 provides a graphic depiction of the intensity scores for the hostile foreign policy events that occurred between 1997

and 2011.Fig. 1 indicates the scores (based on Azar’s scale) for each year for the HFP events, aggregated by quarter from 1997 to 2011.

Peaks and valleys in this chart do sometimes correspond to key events, such as the 2000 ROK-North Korea summit (whichwasaccompanied by a significant decrease in hostilities). Additionally, both the US invasion of Iraq (2003) and emergence of KimJong-un as the future leader of the DPRK (2009e2011) occurred during periods of heightened conflict. Yet this chart onlydemonstrates the incidence of DPRK conflict activities; this research seeks to identify linkages between levels of hostileevents (as shown in Fig. 1) and conditions faced by the Kim regime.

4.2. Time-series regression models

I constructed two models (representing internal and external influences) using time series analysis of quarterly obser-vations between 1997 and 2011 of the variables described above (dependent, independent and control). The first modelfocuses on internal effects (Hypothesis 1) and the second examines the effects of external influences (Hypothesis 2) that arepotentially correlated with DPRK conflict.

4.2.1. Model 1 (Internal conditions) and model 2 (External conditions)According to the H1, the dependent variable (Hostile Foreign Policy or HFP) is North Korea’s hostile foreign policy activity

and the independent variables is the “deterioration of domestic conditions.” The deterioration of domestic conditions is

34 Internal surveys and research in North Korea are rare, although some important data was collected by aid organizations during the famine period on thesocial characteristics of North Korea (Woo-Cumings, 2002). Most of the survey data on North Korea social and political conditions comes from refugees,especially those in China (Haggard and Noland, 2006).35 These reports were translated initially using automated methods (Google Translate, 2013) and then were reviewed by a Korean translator for accuracy.36 For the statistical analysis, the events were aggregated and analyzed by quarter.

Please cite this article in press as:Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

Table 1Quarterly hostile intensity scoring.

Hostile event (Ei) in a given quarter: Azar score ðSE1 ÞE1: DMZ firefight 50E2: Kidnapping of ROK citizen 44E3: Mild Diplomatic protest at UN 6E4: Expulsion of Foreign Press 29HFPqtr 129

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e128

measured by social, political, and economic conditions (the independent variables show above). Model 1 is conceptuallyshown below.

Model 1 (Internal Conditions)

HFPt ¼ b0 þ b1Elite Volt�1 þ b2WGIt�1 þ b3DPRK CINCt�1 þ b4GDPt�1 þ B5Tradet�1 þ b6Refugeest�1

þ b7Infant Mortt�1 þ b8Foodt�1 þ b9ROK Administrationst�1 þ B10DPRK and China Tradet�1 þ mt

I use the model above for the regression analysis to test Hypothesis 1. The second proposition (H2) suggests that externalconditions are more influential in North Korea’s choices to engage in hostile foreign policy activities and follows below.

Model 2 (External Conditions)

HFPt ¼ b0 þ b1UN Resolutionst�1 þ b2Mil Exercisest�1 þ b3US Leadershipt�1 þ b4ROK Leadershipt�1

þ b5Japan Leadershipt�1 þ b6US CINCt�1 þ b7ROK CINCt�1 þ b8Japan CINCt�1 þ b9ROK Admint�1

þ b10DPRK and China Tradet�1 þ mt

4.3. Data and statistical diagnostics

Because the dependent variable data, that is, HFPt in both models, was found initially in daily increments and all of theindependent variables were reported either yearly or quarterly, all of the data required other adjustments to allow for sta-tistical analysis. For example, after running models using monthly, quarterly and yearly aggregation, the models usingquarterly data were found to provide the best representation (and best statistical fit) of the data to the research question.Monthly data, although available for the hostility (dependent variable) data, required too much manipulation of the othervariables while models using only yearly data provided too few observations. Thus, using quarterly data provided a suitable“middle solution” that afforded an acceptable number of observations for analysis without causing significant problems withcollapsing yearly data into reportable units.37

For each quarter, the total event intensities are summed and annotated. Scaled quarterly hostile foreign policy events werecalculated using the following equation:

HFPqtr ¼Xn

n¼1

SE1

where HFPqtr is the quarterly total intensity of hostile foreign policy events, SE1 is the Azar score for the ith event, and n is thetotal number of events. A simplified example and calculation follows (Table 1):

This example shows that for this particular quarter, the hostility intensity score is the product of all of the hostile eventsthat occurred during that quarter (a score of 129). For each quarter, I record the dependent variable (intensity level) score andinclude that in the statistical analysis.38 The continuous independent variables (that were estimated on a yearly basis) werecollapsed into quarterly data reports for analysis. This was done using one of two methods: (1) interpolation was used for“rate” data, such as GDP growth per person, CINC, WGI, elite volatility, infant mortality, and food availability and (2) datawhich reflected “hard numbers,” such as refugees and total trade, were divided into quarterly reports based on yearly data.

To correct for correlation between successive time variables (autocorrelation), the independent variables were lagged.Tests for specification bias indicated that while it was possibly present in the models, solutions (such as finding additional

37 King et al. (1994) discuss problems with research involving small numbers of observations and using quarterly (rather than yearly) data helped mitigatethis issue.38 The use of quarterly data did require adjustments due to missing data, as there were entire years in which data was not reported. To fill gaps in annualreporting, single imputation using arithmetic mean or the last known values were used. While this is not optimal (original data is always preferred) andpossibly induces some bias, this is more desirable than dropping the years with incomplete data (an alternate solution) which would result in the omission ofcritical observations. Baraldi and Enders (2010), and Howell (1983) discuss the merits and risks of this approach. While there are more sophisticated methods(such as multiple imputation) available (see King et al., 2001), after examining imputation options, the simplest solution and best fit for this data was tosimply input data based on arithmetic mean or the last known value. This estimated data was included along with the reported information in the analysis.

Please cite this article in press as: Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

Table 2Statistical analysis results.

Model 1 (internal) Model 2 (external)

World Gov Indicators(t-1) 130.8264 (172.8839)Elite Volatility(t�1) 883.0903 (529.5215)*DPRK CINC(t�1) 123040.3 (83316.05)GDP Growth(t�1) �5.732308 (11.61382)Total Trade (t�1) �.1402127 (.1280578)Refugees(t�1) .1704198 (.306227)Infant Mortality(t�1) 10.15777 (11.51928)Food Availability(t�1) �20025.59 (9971.864)**UN Resolutions(t�1) �24.3103 (51.3671)Military Exercises(t�1) 30.35326 (23.69077)US Leadership Change(t�1) 15.89118 (59.30885)ROK Leadership Change(t�1) 79.93548 (60.56048)Japan Leadership Change(t�1) �63.02935 (36.27616)*US CINC(t�1) �13828.96 (16027.38)ROK CINC(t�1) �165922 (88745.22)*Japan CINC(t�1) 2365.23 (14998.03)ROK Administration(t�1) 82.8621 (56.2484) 85.17583 (74.08891)China and DPRK Trade(t�1) �2.002619 (1.441219) �.9032349 (.801781)Adjusted R2 0.1358 0.1405Number 58 58

*p < 0.10, **p < 0.05, ***p < 0.01.Note: Coefficients are listed first followed by standard errors in parentheses.Sources: Statistical analysis of events from USFK (2012), UNCMAC (2012), KINU (2011), Fischer (2007), KCNA (2012), and New York TimesHistorical Archives 1997e2011 via Lexis-Nexis Academic Search (2012).

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e12 9

variables) are problematic, considering the lack of information on North Korea. Thus, while there might be specification biasin the models, this is an assumed risk for this analysis. To account for potential heteroscedasticity issues, I use robust standarderrors in both models. Finally, diagnostic tests indicated that autocorrelation was possibly present in both models and toaddress this issue, I use Cochrane-Orcutt AR(1) regression.39 Finally, I test for correlations (“multicollinearity”) and find highlevels of correlation did exist between some of the variables.40 After attempting a number of standardized remedies toproblems associated with multicollinearity (that is, transforming variables), none were effective. Multicollinearity is anassumed risk in this analysis.41

5. Statistical analysis of DPRK hostile foreign policy activity

This study examines time-series event data surrounding the Kim regime’s hostile foreign policy events between 1997 and201142 using longitudinal research. The data has been obtained through repeated observation of DPRK-related activities(designated as the dependent variable) during a given historical period (14 years). This method of focusing solely on the long-term activity of North Korea allows for a more structured and detailed examination of the DPRK and its conflict activities.

5.1. Hostile Foreign Policy Statistical analysis

The outputs of the statistical analysis using Cochrane-Orcutt AR(1) regression for the two models are shown in Table 2.Model 1 tests the overall hypothesis that internal conditions influence external conflict activities (H1) and examines the

relationship between HFP and internal political, economic, and social conditions. In this first model, statistically significantrelationships were found between hostile foreign policy activities and political instability (elite volatility) at the .10 level andfood availability (at the .05 level). This provides support to H1 and that there is a potential correlation between internal(political and social) instability and increases in hostile foreign policy activities. This model demonstrated statistical signif-icance and accounts for 13.58 percent of the variance in the hostility score (based on the R2 value). These results provide

39 In both models, tests for autocorrelation (Durbin Watson and Breusch-Godfrey LM tests) indicate that autocorrelation was possibly present. Theadvantage of using the Cochrane-Orcutt AR(1) regression technique is that it initially provides an estimate of the autocorrelation error, then includes thaterror while estimating new regression coefficients, and finally provides a regression output that confirms that the “fitted residuals are independent”(Maggin et al., 2011, 308).40 The mean variable inflation factor (VIF) scores both models were at 15.90 (Model 1) and 3.95 (Model 2). After conducting diagnostic tests of jointsignificance or the “joint f test” (Blackwell, 2008), I find that the variables are influential in relation to the dependent variable, regardless of theircorrelation.41 Gujarati and Porter (2009, 342) note that one method to deal with multicollinearity is to “do nothing” and given the characteristics of this data, this isthe best option for this research.42 Statistical analysis, with the help of the computer program Stata, is used to examine the data and help support conclusions on relationships betweenthe variables. Stata is a commercially available statistical software package often used by social scientists to analyze the relationships between measurablecharacteristics (variables) of a specific phenomenon (Stata, 2012).

Please cite this article in press as:Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e1210

partial support to the idea that internal conditions do indeed influence DPRK conflict activities, especially the conditions facedby the political leaders. Additionally, the relationships not found (economic conditions) are also important and demonstratethat arguments surrounding North Korea’s conflict activities in response to economic challenges may have only limited value.While these results tell only part of the story, they do support the idea that when the North Korean regime is under politicalpressure or lacks adequate food supplies, it may choose conflict activities as a response.

Model 2 tests the second proposition (H2) that North Korea’s foreign policy choices are more influenced by externalfactors. This model includes the regression of a number of external factors (UN resolutions, leadership changes, nationalcapabilities, military exercises and the two control variables) against the dependent variable (hostile foreign policy). Sta-tistically significant relationships found in this model include Japanese leadership changes and ROK CINC.43 This modeldemonstrated statistical significance, but accounts for only 14 percent of the variance in the hostility score. These resultsmight indicate that North Korea actively works to limit its conflict activities during leadership transition periods in Japan andwhen ROK national capabilities are increasing. Yet the relationships not found in Model 2 are also important. These include alack of correlation between conflict and UN resolutions, military exercises, and US and ROK election cycles. Arguments thatNorth Korea responds to external threats from the international community find less support because of this analysis. Finally,neither of the control variables (ROK Administration type or China-DPRK Trade) were found to be significant.

6. The determinants of North Korea’s external conflict

This research demonstrates the complicated nature of unruly data on a closed state and yields a number of useful con-clusions. The first overarching observation is that Model 1 provides partial support to contention that between 1997 and2011internal political factors and food availability were potentially influential in North Korea’s external conflict activities.While this observation supports the concept of diversion, determining the true intentions of the Kim regime when it choosesto embark on conflict activities remains problematic. In any case, these statistically significant findings are consistent with theassumption that there is a relationship between internal conditions and external conflict.

Many of the external variables (such as UN resolutions, military exercises, and threats posed by the US and South Korea)are often assumed to cause increases in the Kim regime’s hostile foreign policy activities. This research provides only limitedsupport for these types of “conventional wisdom” assumptions about North Korean behavior. Thus, based on these findings,efforts by the international community to change North Korea’s foreign policy behavior (for example, UN Security CouncilResolutions) potentially have little effect in either increasing or decreasing Pyongyang’s conflict activities. This research al-ludes to other factors that might be influential in the Kim regime’s efforts to use conflict as a foreign policy alternative. Thesemight include internal dynamics, such as efforts to bolster the position of the DPRK’s relatively new ruler, Kim Jong-un, or inconjunction with negotiations to secure additional international support and aid for the regime.

This study demonstrates that useful analysis can be done with publically available conflict data to examine the DPRK’spolitical and military actions. It also provides an example of how to gather and analyze data on closed societies since NorthKorea is arguably the most isolated garrison state in existence. At the very least, studying the DPRK’s foreign policy behaviortells us something important about North Korea’s activities in the late 1990s through 2011: the Kim regime is more influencedby internal conditions than international community actions aimed at changing the Kim regime’s behavior.

Acknowledgment

The Center for Strategic Intelligence Research (CSIR) staff provided tremendous support for this research during myyearlong fellowship at the National Intelligence University (NIU) in Washington, DC.

This paper and its conclusions are those of the author and do not represent the views of CSIR, NIU, or the US government.

References

Abouharb, M., Kimball, Anessa, 2007. A new dataset on infant mortality rates, 1816e2002. J. Peace Res. 44 (6), 743e754.Ake, Claude, 1975. A definition of political stability. Comp. Polit. 7 (2), 271e283.Allison, Paul David, 1984. Event History Analysis: Regression for Longitudinal Event Data. Sage Publications, Beverly Hills, CA.Azar, Edward E., 1982. The Codebook of the Conflict and Peace Data Bank (COPDAB): a Computer-Assisted Approach to Monitoring and Analyzing Inter-

national and Domestic Events. The Center for International Development and Conflict Management, University of Maryland, College Park, MD.Azar, Edward E., 1993. Conflict and Peace Data Bank (COPDAB), 1948e1978 Codebook. ICPSR07767-v4. Inter-university Consortium for Political and Social

Research, Ann Arbor, MI.Banks, Arthur S., 2004. Cross-National Time-Series Data Archive. Center for Social Analysis, Binghamton, NY.Baraldi, Amanda N., Enders, Craig K., 2010. An introduction to modern missing data analyses. J. Sch. Psychol. 48 (1), 5e37.Bell, Sam R., 2009. What You Don’t Know Can Hurt You: Information, Transparency, Domestic Challenges, and Interstate Conflict. Dissertation. State

University of New York, New York.Blackwell, Matthew, 2008. Multiple Hypothesis Testing: The F-test. Available at: http://www.mattblackwell.org/files/teaching/ftests.pdf (accessed 13.05.13.).Bodin, Jean, 1955 [1606]. Six Books of the Commonwealth. Translated by M J. Tooley. Basil Blackwell, Oxford.BOK (Bank of Korea), 2012. The Bank of Korea Online. Available at: www.bok.or.kr/index.jsp. keyword search e North Korea GDP (accessed 21.06.12.).

43 Both relationships between these variables and HFP were negative.

Please cite this article in press as: Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e12 11

Bond, Doug, Bond, Joe, Oh, Churl, Jenkins, J.C., Taylor, Charles Lewis, 2003. Integrated data for events analysis (IDEA): an event typology for automatedevents data development. J. Peace Res. 40 (6), 733e745.

Bueno de Mesquita, Bruce, 1985. Toward a scientific understanding of international conflict: a personal view. Int. Stud. Q. 29 (2), 121e136.Bueno de Mesquita, Bruce, Lalman, David, 1992. War and Reason. Yale University Press, New Haven, CT.Chapman, Terrence L., Reiter, Dan, 2004. The United Nations Security Council and the Rally Round the Flag effect. J. Confl. Resolut. 48 (6), 886e909.CIA, 2013. CIA World Factbook. Central Intelligence Agency, Washington, D.C.. Available at: https://www.cia.gov/library/publications/ (accessed 23.02.13.).CINC (Composite Indicator of National Capabilities Index), 2012. Composite Indicator of National Capabilities Database. Correlates of War Project. Available

at: http://www.correlatesofwar.org/COW2%20Data/Capabilities/nmc3-02.htm#cinc (accessed 23.08.12.).COW MID (Correlates of War Militarized Interstate Dispute) Data, 2012. Correlates of War Website. Available at: http://www.correlatesofwar.org (accessed:

15.10.12.).COW Trade Data, 2012. COW International Trade 1870e2009. Available at: http://www.correlatesofwar.org/COW2%20Data/Trade/Trade.html (accessed 22.

10.12.).Crescenzi, Mark, Enterline, Andrew, Long, Stephen, 2008. Bringing cooperation back in: a dynamic model of interstate interaction. Confl. Manag. Peace Sci.

25 (3), 264e280.Davies, Graeme A.M., 2002. Domestic strife and the initiation of international conflicts: a directed dyad analysis, 1950e1982. J. Confl. Resolut. 465, 672e692.D’Orazio, Vito, 2012. War games: North Korea’s reaction to US and South Korean military exercises. J. East Asian Stud. 12 (2), 275e294.DPRK Radio Central Broadcasting Station, 2011. North Korean Army Command Slams South’s Military Drills. BBC Summary of World Broadcasts, Pyongyang

(accessed via LexisNexis, 07.11.11.).Eberstadt, Nicholas, 1997. North Korea as an economy under multiple severe stresses: analogies and lessons from past and recent historical experience.

Communist Econ. Econ. Transform. 9 (2), 233e255.FAOSTAT (Food and Agriculture Organization of the United Nations), 2012. Online Database. Available at: http://faostat.fao.org/site/616/DesktopDefault.

aspx?PageID¼616#ancor (accessed 20.05.12.).Fearon, James D., 1994. Domestic political audiences and the escalation of international disputes. Am. Polit. Sci. Rev. 88 (3), 577e592.Filippov, Mikhail, 2009. Diversionary role of the GeorgiaeRussia Conflict: international constraints and domestic appeal. Europe-Asia Stud. 61 (10), 1825e

1847.Fischer, Hannah, 2007. North Korean Provocative Actions, 1950e2007. CRS Report. Congressional Research Service, Washington, DC.Fordham, Benjamin, 1998. The politics of threat perception and the use of force: a political economy model of US uses of force, 1949e1994. Int. Stud. Q. 42

(3), 567e590.Gause, Ken E., 2012. Coercion, Control, Surveillance and Punishment: Examination of the North Korean Police State. The Committee for Human Rights in

North Korea, Washington, DC.Gates, Scott, Hegre, Håvard, Jones, Mark P., Strand, Håvard, 2006. Institutional inconsistency and political instability: polity duration, 1800e2000. Am. J.

Polit. Sci. 50 (4), 893e908.Goldstein, Joshua S., 1992. A conflict-cooperation scale for WEIS events data. J. Confl. Resolut. 36 (2), 369e385.Google Translate, 2013. Google Corporation. Available at: http://translate.google.com/ (accessed 10.05.12.).Gujarati, Damodar N., Porter, Dawn C., 2009. Basic Econometrics. McGraw-Hill, Boston.Haggard, Stephan, Noland, Marcus, 2007. Famine in North Korea: Markets, Aid, and Reform. Columbia University Press, New York.Haggard, Stephen, Noland, Marcus, 2006. The North Korean Refugee Crisis: Human Rights and International Response. U.S. Committee for Human Rights in

North Korea, Washington, DC.Hassig, Ralph, Oh, Kongdan, 2009. The Hidden People of North Korea: Everyday Life in the Hermit Kingdom. Rowman and Littlefield Publishers, Lanham,

MD.Hawk, David, 2003. The Hidden Gulag: Exposing North Korea’s Prison Camps. US Committee for Human Rights in North Korea, Washington, DC.Howell, Llewellyn D., 1983. A comparative study of the WEIS and COPDAB data sets. Int. Stud. Q. 27 (2), 149e159.Ilbo, Chosun, 2011a. Kaesong Firms Worry as N.Korea Seizes Mt. Kumgang Assets, 11 August 2011. Available at: http://english.chosun.com/site/data/html_

dir/2011/08/24/2011082401039.html (accessed 12.02.12.).Ilbo, Chosun, 2011b. N.Korea Refuses to Budge Over Mt. Kumgang, 27 July 2011. Available at: http://english.chosun.com/site/data/html_dir/2011/07/27/

2011072700683.html (accessed 12.02.12.).Ishiyama, John, 2012. Assessing the Leadership Transition in North Korea: Using Network Analysis of Field Inspections, 1997e2012. Unpublished paper

presented at the International Studies Association (Midwest) Conference, 6 November 2012, St. Louis, MO.James, Patrick, Oneal, John R., 1991. The influence of domestic and international politics on the president’s use of force. J. Confl. Resolut. 35 (2), 307e332.Kang, Chol-Hwan, 2002. The Aquariums of Pyongyang: Ten Years in the North Korean Gulag. Basic Books, New York.Kaufmann, Daniel, Kraay, Aart, Mastruzzi, Massimo, 2009. Governance Matters VIII: Aggregate and Individual Governance Indicators, 1996e2008 (June 29,

2009). World Bank Policy Research Working Paper No. 4978. Available at: http://ssrn.com/abstract¼1424591 (accessed 02.09.11.).KCNA, 2012. Korea Central News Agency Online. Available at: www.kcna.co.jp/index-e.htm (accessed 07.12.12.).King, Gary, et al., 2001. Analyzing incomplete political science data: an alternative algorithm for multiple imputation. Am. Polit. Sci. Assoc. 95 (1), 49e69.King, Gary, Lowe, Will, 2003a. An automated information extraction tool for international conflict data with performance as good as Human Coders: a rare

events evaluation design. Int. Organ. 57 (3), 617e642.King, Gary, Lowe, Will, 2003b. 10 Million International Dyadic Events. Available at: http://hdl.handle.net/1902.1/FYXLAWZRIAUNF:3:

dSE0bsQK2o6xXlxeaDEhcg¼¼. IQSS Dataverse Network, Version 3 (accessed 10.05.12.).King, Gary, Keohane, Robert O., Verba, Sidney, 1994. Designing Social Inquiry: Scientific Inference in Qualitative Research. Princeton University Press,

Princeton, NJ.KINU (Korea Institute for National Unification), 2011. Inter-Korean Relations Chronology, 1948e2011. Available at: http://www.uniedu.go.kr/uniedu/

PdsDataroomHome.do?cmd¼readArticle&dataroomArticleDTO.atclSn¼1382 (accessed 22.10.11.).Kisangani, Emizet F., Pickering, Jeffrey, 2007. Diverting with benevolent military force: reducing risks and rising above strategic behavior. Int. Stud. Q. 51,

277e299.Kisangani, Emizet F., Pickering, Jeffrey, 2009. The dividends of diversion: mature democracies’ proclivity to use diversionary force and the rewards they reap

from it. Br. J. Polit. Sci. 39 (3), 483e515.KNDU (Korea National Defense University) Official, 2011. Personal Communication [email]. 8 June 2011.Lai, Brian, Slater, Dan, 2006. Institutions on the offensive: domestic sources of dispute initiation in authoritarian regimes, 1950e1992. Am. J. Polit. Sci. 50,

113e126.Lankov, Andrei N., 2011. North KoreaeChina special economic zones. East Asia Forum, 14 July 2011. Available at: http://www.eastasiaforum.org/2011/07/14/

north-korea-china-special-economiczones/ (accessed 05.02.12.).Leeds, Brett Ashley, Davis, David R., 1997. Domestic political vulnerability and international disputes. J. Confl. Resolut. 41 (6), 814e834.Lenin, V.I., 1930. The Imperialist War: the Struggle Against Social Chauvinism and Social Pacifism. International Publishers, New York.Levy, Jack S., 1989. The diversionary theory of war: a critique. In: Midlarsky, Manus I. (Ed.), Handbook of War Studies. Uwin Hyman, Boston, pp. 259e288.Machiavelli, Niccolò, 1882 [1513]. Translated by Christian E. Detmold. The Historical, Political, and Diplomatic Writings of Niccolo Machiavelli, vol. II. James

R. Osgood and Company, Boston.Maggin, D., Swaminathan, H., Rogers, H.J., O’Keefe, B., Sugai, G., Horner, R., 2011. A generalized least squares regression approach for computing effect sizes

in single-case research: application examples. J. Sch. Psychol. 49, 301e321.Manyin, Mark E., Nikitin, Mary Beth, 2013. Foreign Assistance to North Korea. CRS Report. Congressional Research Service, Washington, DC.

Please cite this article in press as:Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004

R.D. Wallace / Communist and Post-Communist Studies xxx (2014) 1e1212

Martin, Bradley K., 2006. Under the Loving Care of the Fatherly Leader: North Korea and the Kim Dynasty. Griffin, New York.Mayer, Arno J., 1969. Internal causes and purpose of war in Europe, 1870e1956. J. Mod. Hist. 41 (3), 291e303.McClelland, Charles, 1999. World Event/Interaction Survey (WEIS) Project Codebook, 1966e1978. ICPSR05211-v3. Inter-university Consortium for Political

and Social Research, Ann Arbor, MI.Meernick, John J., 1994. Presidential decision-making and the political use of force. Int. Stud. Q. 38, 121e138.Menard, Scott W., 1991. Longitudinal Research. Sage Publications, Newbury Park, CA.Michishita, Narushige, 2010. North Korea’s Military-Diplomatic Campaigns, 1966e2008. Routledge, New York, NY.Milbank, Dan, 2002. Democrats Question Iraq Timing; Talk of War Distracts From Election Issues. Washington Post, 16 September 2002, A1.Mitchell, Sara McLaughlin, Prins, Brandon C., 2004. Rivalry and diversionary uses of force. J. Confl. Resolut. 48 (6), 937e961.Mitchell, Sara McLaughlin, Thyne, Clayton L., 2010. Contentious issues as opportunities for diversionary behavior. Confl. Manag. Peace Sci. 27 (5), 461e485.Morgan, T. Clifton, Bickers, Kenneth N., 1992. Domestic discontent and the external use of force. J. Confl. Resolut. 36 (1), 25e52.Natsios, Andrew S., 2001. The Great North Korean Famine: Famine, Politics, and Foreign Policy. United Stated Institute of Peace, Washington, DC.Nincic, Miroslav, 1975. Determinants of third world hostility toward the United States: an exploratory analysis. J. Confl. Resolut. 19 (4), 620e642.NMC (National Material Capabilities) Documentation, 2005. Correlates of War Website. Available at: http://www.correlatesofwar.org (accessed 15.10.12.).Oakes, Amy, 2006. Diversionary war and Argentina’s invasion of the Falkland Islands. Secur. Stud. 15 (3), 431e463.Oreskes, Michael, 1990. Bush Trying a New Topic; Is Talk of War Driven By Policy or Politics? New York Times, 31 October 1990, A21.Ostrom Jr., Charles W., Job, Brian L., 1986. The president and the political use of force. Am. Polit. Sci. Rev. 80 (2), 541e566.Peceny, Mark, Beer, Caroline C., 2003. Peaceful parties and puzzling personalists. Am. Polit. Sci. Rev. 97, 339e342.Peceny, Mark, Butler, Christopher K., 2004. The conflict behavior of authoritarian regimes. Int. Polit. 41, 565e581.Pickering, Jeffrey, Kisangani, Emizet F., 2005. Democracy and diversionary military intervention: reassessing regime type and the diversionary hypothesis.

Int. Stud. Q. 49 (1), 23e43.Pickering, Jeffrey, Kisangani, Emizet F., 2010. Diversionary despots? Comparing autocracies’ propensities to use and to benefit from military force. Am. J.

Polit. Sci. 54 (2), 477e493.Purdum, Todd S., 1998. U.S. Fury on 2 Continents: Congress; Critics of Clinton Support Attacks. New York Times, 21 August 1998, A1.Putnam, Robert D., 1988. Diplomacy and domestic politics: the Logic of two-level Games. Int. Organ. 42 (3), 427e460.Ramstad, Evan, 2011. North Korea warns south in annual message. Wall Str. J. Online, 1 January 2011. Available at: http://online.wsj.com/article/

SB10001424052748704543604576054890727297846.html (accessed 15.10.11.).Reilly, David, 2004. The growing importance of the failing state: sovereignty, security, and the return to power politics. J. Confl. Stud. 24 (June). Available at:

http://journals.hil.unb.ca/index.php/JCS/article/view/289/459 (accessed 27.10.11.).ROK MOU (Republic of Korea Ministry of Unification), 2012. Major Statistics in Inter-Korean Relations. Available at: http://eng.unikorea.go.kr/CmsWeb/

viewPage.req?idx¼PG0000000541 (accessed 23.10.12.).Shakespeare, William, 1823. King Henry IV, Act 4. The Dramatic Works of William Shakespeare. Chiswick, London.Sigal, Leon V., 1998. Disarming Strangers: Nuclear Diplomacy with North Korea. Princeton University Press, Princeton, NJ.Stata (Stata Data Analysis and Statistical Software), 2012. Available at: http://stata.com/ (accessed 11.09.12.).The Telegraph (UK), 2011. North Korea says Korean Peninsula Faces ‘Worst Crisis’, 16 August 2011. Available at: http://www.telegraph.co.uk/news/

worldnews/asia/northkorea/8703786/North-Korea-says-Korean-peninsula-faces-worst-crisis.html (accessed 19.09.12.).Tir, Jaroslav, 2010. Territorial diversion: diversionary theory of war and territorial conflict. J. Polit. 72 (2), 413e425.UN FAIS (United Nations Food Aid Information System) World Food Program, 2012. World Food Program Website. Available at: http://www.wfp.org/fais/

quantity-reporting (accessed 10.08.12.).UNCMAC (United Nations Military Armistice Command), 2012. Author interview and data provided by a United Nations Command Military Armistice

Commission official. Seoul, South Korea. 20 September 2012.UN (United Nations) Population Division, 2012. World Population Prospects: the 2010 Revision and World Urbanization Prospects: the 2011 Revision.

Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat. Available at: http://esa.un.org/unpd/wup/unup/p2k0data.asp (accessed 25.05.12.).

UN Comtrade, 2013. United Nations Commodity Trade Statistics Database. Available at: http://comtrade.un.org/db/ce/ceSearch.aspx (accessed 20.04.13.).US Census Bureau, 2012. International Programs. Available at: http://www.census.gov/population/international/data/idb/informationGateway.php

(accessed 20.05.13.).US Department of Defense, 2011. Dictionary of Military and Associated Terms. Joint Publication 1-02. US Government Printing Office, Washington, DC.USG (US Government), 2013. United States Population Division e International Census Data. Available at: http://www.census.gov/population/international/

data/idb/region.php?N¼%20Results%20&T¼13&A¼separate&RT¼0&Y¼1997,1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,2008,2009,2010,2011&R¼-1&C¼KN (accessed 20.04.13.).

USFK, 2012. Author Interview with USFK Historian. Seoul, South Korea. 25 September 2012.Walder, David, 1974. The Short Victorious War: the Russo-Japanese Conflict, 1904e5. Harper & Row, New York.Waltz, Kenneth N., 1954. Man, the State, and War: a Theoretical Analysis. Columbia University Press, New York.Waltz, Kenneth N., 1979. Theory of International Politics. Addison-Wesley, Reading, MA.Weeks, Jessica, 2011. Strongmen and Straw Men: Authoritarian Regimes and the Initiation of International Conflict. Unpublished Working Paper. Cornell

University. Available at: http://falcon.arts.cornell.edu.er.lib.k-state.edu/jlw338/index_files/Weeks-Strongmen.pdf (accessed 15.08.12.).Wendt, Alexander, 1992. Anarchy is what states make of it: the social construction of power politics. Int. Organ. 46 (2), 391e425.Woo-Cumings, Meredith, 2002. The Political Ecology of Famine: the North Korean Catastrophe and its Lessons. Asian Development Bank Institute, Tokyo.World Data Bank, 2012. World Bank World Development Indicators and Global Development Finance. Available at: http://data.worldbank.org/indicator/SH.

DYN.MORT?page¼1 (accessed 09.08.12.).Yoon, Sangwon, 2013. N. Korea’s 2011 China Trade Grew More Than 60 Percent. Bloomberg News, 2 January 2013. Available at: http://www.businessweek.

com/news/2012-12-26/north-korea-s-china-trade-expanded-more-than-60-percent-in-2011 (accessed 13.02.13.).

Please cite this article in press as: Wallace, R.D., North Korea and diversion: A quantitative analysis (1997e2011), Communist andPost-Communist Studies (2014), http://dx.doi.org/10.1016/j.postcomstud.2014.04.004