Full Text

1 Introduction South Korea’s foreign relations with Japan are frequently impeded by confrontations over the content of Japan’s history books, as well as the legacy of the comfort women/sex slaves issue during World War II, and the continued territorial dispute over Takeshima/Dokdo. Recently, in August 2012, President Lee Myung-bak landed on Takeshima/Dokdo, in the first official visit by a South Korean president (Asahi Shimbun, 10 August 2012; Korea Times, 13 August 2012; Washington Post, 11 August 2012). This was followed two months later, in October 2012, by Japan and South Korea failing to renew a temporal bilateral currency swap facility to protect their economies against financial crisis (International Herald Tribune, 12 October 2012; Reuters, 9 October 2012). Such disputes have become more frequent in recent years since South Korean democratization. To understand these disputes, we explored the political incentives motivating foreign relations with Japan, because, while a visit to Takeshima/Dokdo may demand a response, the larger political issues are rooted in historical events that occurred generations ago. The fact that these issues are not new suggests that their escalation is subject to political posturing. This article focuses on foreign policy making in a new democracy and examines the economic and political diversionary incentives of South Korea’s leaders. Our time series analysis of the period 1988–2011 provides three distinctive findings. First, economic distress has driven South Korea to avoid its responsibility and divert public attention through international disputes against Japan. Second, a drop in public approval did not stimulate the leaders’ diversionary incentives, which we contend is primarily due to having a one-term presidential tenure limit. Third, national elections, whether presidential or legislative, have not encouraged leaders to resort to diversionary foreign policy, because candidates distinguish themselves from others by presenting different policy positions on domestic problems, not on foreign policy. These findings shed light on the diversionary 34 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from mechanism in a new democracy and add important context to the literature on democratization and international conflict. Scholars of international relations have debated the impact of democratization on international conflict. Mansfield and Snyder (1995a,b) argued that during periods of transition to democracy, new leaders might seek to break the political impasse by stoking nationalist sentiment in foreign relations against a known adversary, sometimes even resorting to war to unite mass support. Over a series of articles, they provided empirical evidence that democratizing states are more dangerous than autocracies or mature democracies (Mansfield and Snyder, 1995a,b, 2002a,b, 2005, 2009). However, some authors have found no correlation between democratization and the likelihood of war (Thompson and Tucker, 1997; Narang and Nelson, 2009). In contrast, other scholars have found a negative correlation between democratization and the likelihood of war (Enterline, 1998a,b;Maoz, 1998;Ward and Gleditsch, 1998). At the same time,Ward and Gleditsch (1998) point out that if democratizing states are war-prone, then this trait should be dominant in the evolutionary process and melded into the social norms and institutional settings that determine the subsequent patterns of international conflict. That expectation would be further at odds with the democratic peace theory and the overwhelming empirical support for the pacifying effect of democracy. Though the democratization-conflict literature has generated mixed support, the most common empirical strategy utilizes panel (crosssectional time series) analyses that cannot sufficiently capture temporal political dynamics within a country. In one of the few studies that examined a single case over time, Lind (2011) used a qualitative analysis of South Korea to examine the war-proneness of a democratizing state, but similarly, found no evidence of the dangerous effect of democratization on war. In contrast, our findings show that, similar to mature democracies, South Korean leaders have incentives to divert public attention away from economic distress to low-intensity disputes against Japan. However, unlike mature democracies, the institutional context in South Korea removes the influence of the presidential ratings and national elections from foreign policy making. In the following section, we begin by examining the diversionary incentives in South Korea. Section 3 discusses our measurement of foreign policy behavior and describes trends during the period 1988–2011. Sections 4 and 5 present the statistical model and findings. In Section 6, we South Korean diversionary foreign policy 35 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from briefly present two illustrative cases of economic diversion, Presidents Roh Tae-woo (1988–1993) and Kim Young-sam (1993–1998). Finally, we discuss the implications of our theory and analysis for the management of future disputes between Japan and South Korea. 2 Theory Diversionary use of force has been attracting scholars of international relations since the seminal work of Ostrom and Job (1986). They found that a decline in public approval and a worsening economy (higher misery index) largely accounted for the United States to use military force rather than traditional international security variables. In the wake of this finding, diversionary theory assumes that leaders are primarily motivated to seek and retain political office, suggesting that leaders act first and foremost to retain their own political support. According to the social-psychological explanation, leaders facing domestic difficulties might evoke a ‘rally ’round the flag’ atmosphere by initiating an international dispute. More precisely, leaders may unite their political base by instigating hostilities abroad, because an international crisis draws the public’s attention toward a foreign adversary and in the process replaces the domestic problem with a foreign crisis (Levy, 1989). A number of studies have assessed the rally effect in the context of US foreign policy by examining the relationship between US use of military force and presidential approval ratings (Ostrom and Job, 1986; James and Oneal, 1991; Morgan and Bickers, 1992). With a few notable exceptions (Morgan and Bickers, 1992; DeRouen, 1995), empirical studies have found that the use of force is likely to lead to a decline in public approval (Ostrom and Job, 1986; James and Oneal, 1991) or only a short-term surge in public approval (Lian and Oneal, 1993; Meernik and Waterman, 1996).1 Therefore, when leaders become enmeshed in international disputes, they must also be concerned with hurting their public approval ratings. Others have investigated whether diversionary incentives are influenced by policy availability and policy substitution (Miller, 1995; Gelpi, 1997; Brulé, 2006, 2008). According to these studies, leaders only employ a diversionary use of force when other direct policy options for the domestic problems are unavailable (Morgan and Bickers, 1992; Miller, 1995; Gelpi, 1 Lai and Reiter (2005) detected the rally effect in the UK. 36 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from 1997). Marra et al. (1990) called it the perpetual election, in which the current level of public approval indicates the public’s evaluation of policy outcomes that reflect leaders’ policy availability to respond to domestic turmoil. Therefore, other things being equal, the low level of public approval should lead to more aggressive foreign policy behavior. In short, whether a rally ‘round the flag effect is enough to generate diversionary incentives, or whether this relationship is conditioned policy availability, leaders are generally more likely to respond to domestic unrest with hawkish foreign policy behavior. 2.1 Economic diversion The literature on the diversionary use of force attributes adventurous foreign policy to economic distress. An aggressive foreign policy can provide a policy alternative when economic policies are unable to address the underlying economic problems. When tensions escalate between countries, the general public might pay more attention to the foreign disputes rather than to an economic downturn (Ostrom and Job, 1986; DeRouen, 1995; Fordham, 1998, 2002). Therefore, economic turmoil can be a driving force of diversionary foreign policy. Consider economic diversion from the viewpoint of policy availability. Leaders are limited in the policies they can pursue to address an economic downturn, because macroeconomic policy alternatives such as financial and fiscal policies can stimulate growth only in the short term. If leaders increase public works projects to ameliorate unemployment or reduce taxes to stimulate the economy, they need to tackle fiscal deficit reduction through budget cuts or tax increases sooner or later. If leaders increase the money supply to stimulate the market, they eventually need to suppress the money supply in order to tame inflation in the near future. If these macroeconomic policies cannot serve as solutions for serious recession, leaders need another policy alternative to divert public attention from economic distress. Thus, as long as economic growth is the primary goal, economic turmoil is a driving force behind diversionary foreign policy. Since the 1960s, South Korean leaders have pursued a governmentled export-driven strategy (Haggard, 1990, Chapter 3). While the government plays a large role in promoting economic growth, a South Korean newspaper survey shows that 48.5% of respondents believed that reforms of politicians and bureaucrats were needed for a full recovery from financial crisis because they are responsible for the management of South Korean diversionary foreign policy 37 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from the economy (Donga-Ilbo, 2 November 1998). A sharp decline in economic growth can have devastating effects on political support for the government and even trigger demonstrations and social unrest, which in turn induces leaders to divert the public attention away from economic distress (Kong, 2000, p.243). Therefore, economic growth is a prominent issue and economic downturns can provoke diversionary incentives in South Korea. H1: If the state of the economy is in a downturn, then South Korea’s leaders are more likely to resort to a hostile foreign policy against Japan. 2.2 Political diversion Similar to economic diversion is the argument that a decline in public approval and/or impending national elections drives diversionary foreign policy. Leaders resort to a diversionary use of force only when other direct policy options to address the domestic problems are unavailable (Morgan and Bickers, 1992; Miller, 1995; Gelpi, 1997). The current level of public approval represents the public’s evaluation of policy outcomes, which reflect the leaders’ policy availability to respond to domestic turmoil. This is called the perpetual election (Marra et al., 1990). Therefore, a low level of public approval reflects that leaders are facing political difficulties and provokes diversionary incentives. In South Korea, the president retains the power of initiative in the process of cabinet member appointments, even with congressional involvement, and can pursue policy outcomes due to the centralization of authority (Shugart and Carey, 1992, pp. 106–111). Public opinion is sensitive to failed public policies and scandals involving the president and administrative members, because the president must be responsible for them, due to the centralization of authority. Public approval is determined by much broader criteria, although we do not exclude the possibility that economic conditions have an impact on public approval. Figure 1 shows that public approval ratings fall before economic downturns occur; these variables display little correlation.2 Thus, a lower public approval level does not always reflect a lack of availability of direct policy options. On the other hand, the public’s loyalty to a national leader is eroded by a five-year, one-term presidential tenure limit, because the public becomes more concerned with the potential candidates for the next term (Im, 2004, 2 The correlation coefficient is 0.12. 38 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from pp. 190–191). Since the lame-duck leader is aware of the erosion of the public’s loyalty, he/she does not respond to popularity ratings after the third or fourth year of his/her tenure. For these reasons, the public approval ratings do not always reflect the public’s evaluation of the president’s policy performance, and leaders do not attempt to adopt a diversionary foreign policy in response to the low public approval level. H2: The presidential approval rate is not associated with South Korea’s hostile foreign policy toward Japan. In the literature, national elections are also thought to provoke aggressive foreign policy. Scholars have examined the diversionary incentives of office-seeking politicians by focusing on voter uncertainty about the incumbent’s competency. They have assumed that voters attempt to judge the incumbent’s competency based on policy outcomes and cannot completely attribute policy outcomes to the incumbent’s skills. When direct policy instruments for domestic problems are not available, only competent incumbent politicians look for opportunities to distinguish themselves from incompetent politicians. Therefore, they have an incentive to divert the public’s attention away from domestic problems and toward foreign crisis (Richards et al., 1993; Downs and Rocke, 1995; Smith, 1996).3 That Figure 1 GDP growth rate and public approval, 1988–2011. 3 According to Tarar (2006), only when the benefits of holding office are great enough are incumbents more likely to escalate a dispute using diversionary force. South Korean diversionary foreign policy 39 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from is, only competent incumbents provoke an aggressive foreign policy in the shadow of an election and can further their political gain by handling the international crisis well. Thus, national elections can stimulate diversionary incentives. The incumbent president in South Korea has no incentive to take a hard line against foreign adversaries in the shadow of elections, because his/her term is limited to one five-year tenure as president. Furthermore, the incumbent legislators have no diversionary incentive because they cannot increase their electoral gains by adopting an aggressive foreign policy. Although South Korea has experienced five presidential elections and six legislative elections since its democratization, party alignment is not consolidated and no party survived for more than twenty years.4 No party has been able to pursue its platform consistently enough to appeal to the electorate (Im, 2004, p.189). Thus, instability of the party system, with short-lived parties, has prevented the formation of a clear partisan alignment regarding foreign policy. Moreover, given the strong anti-Japan sentiment in South Korean society, all parties share similar policy stances toward Japan and they cannot distinguish themselves from each other, which depoliticizes the Japan issue during electoral campaigns.5 Parties cannot improve their electoral competitiveness by resorting to adventurous foreign policy projects. Therefore, unlike other established democracies, South Korean leaders have no incentive to formulate a diversionary foreign policy in the shadow of an election. H3a: Presidential elections are not associated with South Korea’s hostile foreign policy toward Japan. H3b: Legislative elections are not associated with South Korea’s hostile foreign policy toward Japan. 4 With the data from the Comparative Study of Electoral Systems, the comparative analysis of East Asian countries shows that clear party polarization exists in the consolidated democracies of Japan and New Zealand, but not in the new democracies of Taiwan, the Philippines, and South Korea (Dalton and Tanaka, 2007). Moreover, in South Korea, political parties appeal to regionalism by offering regional benefits, and they accuse competing parties of discriminating against their regions, which impedes ideological voting decisions (Moon, 2005). 5 In the legislature, there is clear consensus on foreign policy toward Japan. Whenever the historical issue appears as a diplomatic issue, both the ruling and opposition parties strongly criticize Japan (Shirai, 2009). 40 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from 3 Event data We begin by explaining the use of the directed-dyad data to describe foreign policy behavior. The directed-dyad data are the record of State A’s foreign policy actions against State B. Scholars often use the Correlates of War Militarized Interstate Dispute (MID) data to analyze interstate conflict behavior. The MID data measure hostility across a six-point scale, where 0 = no hostility, 1 = no militarized action, 2 = threats to use force, 3 = display of force, 4 = use of force, and 5 = war (Ghosn et al., 2004). However, we are interested in the full range of South Korea’s actions toward Japan since World War II, not only those that were militarized actions, such as a seizure of a fishing boat or an occupation of territory. Specifically, we are most interested in low-intensity conflict behavior, and especially, South Korea’s criticisms regarding the Takeshima/Dokdo issue, the history textbook issue, and the comfort women/sex slaves issue, which the MID data do not capture completely. For this reason, we use event data, which reflect a more detailed record of the actions and events that comprise interstate con- flict and cooperative behavior. Scholars developed event data such as the Conflict and Peace Data Bank and World Events Interaction Survey in the 1980s (Azar, 1980; McClelland, 1983) and assigned differently weighted points to each event in order to measure the degree of conflict and cooperation (Goldstein, 1992, pp. 376–377). In the early 2000s, King and Lowe (2003a) automated these procedures using machine coding of newspaper articles from the Reuters Business Briefing newswire and published the Integrated Data for Event Analysis (IDEA), which covers 10 million international dyadic events in the post-Cold War period, 1990–2004.6 Leetaru and Schrodt (2013) recently published the Global Data on Events, Location and Tone (GDELT) data. Like the IDEA data, the TABARI content analysis program examines >200 million separate events across the entire world during the period 1979–2012 from the broadest collection of international and regional news sources and turns them into the directed dyadic data of interstate conflict and cooperation, called the GDELT data.7 The GDELT 6 See King and Lowe (2003b)for accuracy of machine coding. 7 The information source includes AfricaNews, Agence France Presse, Associated Press Online, Associated Press Worldstream, BBC Monitoring, Christian Science Monitor, Facts on File, Foreign Broadcast Information Service, United Press International, the Washington Post, and all national and international news coverage from the New York Times, all South Korean diversionary foreign policy 41 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from data are the most comprehensive event data that cover the postdemocratization period in South Korea. Therefore, we adopt the GDELT data to describe South Korea’s foreign policy behavior. In using the GDELT data, we need to decide whether to use the event count or to assign different weighted points to those events. We found the possibility of systematic measurement bias in the data. Consider the following Agence France Presse report on 15 August 2001: South Korea’s President Kim Dae-Jung on Wednesday launched his own attackon Japanese Prime Minister Junichiro Koizumi’s visit to a controversial war shrine as “comfort women” staged a protest. In this example, South Korea’s President Kim criticized Japanese Prime Minister (PM) Koizumi, but the TABARI software mistakenly coded this case as a military attack. Therefore, when we use the weighted event score to measure the degree of hostility, we are likely to make a wrong inference. Instead, we use the event count to measure South Korean hostile foreign behavior, assuming that many hostile actions reflect more hostility against Japan and the existence of significant disputes. We also assume that few hostile actions represent less hostility and no significant dispute. Figure 2 shows the trend of South Korea’s hostile foreign actions during the period 1988–2011. The GDELT data provide information about verbal and material actions. The top graph represents the total count of South Korea’s hostile actions, whereas the middle and bottom graphs show South Korea’s verbal and material hostile actions, respectively. We find two distinct characteristics of South Korean hostile behavior toward Japan. First, verbal actions are a large share of all hostile actions. Second, the number of hostile actions suddenly increases and its movement becomes volatile after 2000. The mean count of verbal actions increases from 1.67 to 5.90, whereas its standard deviation also increases from 1.80 to 7.61. The mean count of material actions increases from 0.40 to 1.73, whereas its standard deviation also increases from 0.79 to 2.07. Thus, this trend exists regardless of the types of hostile behavior. Below, we examine the volatility of South Korean foreign behavior by focusing on diversionary incentives. international and major US national stories from the Associated Press, and all national and international news from Google News, with the exception of sports, entertainment, and strictly economic news. 42 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from 4 The statistical model We need the model to explain the event counts to test the diversionary theory. As such, we use the count model in which the number of South Korea’s hostile behavior toward Japan is the function of the state of the economy, political approval, and national elections. The Poisson regression model is used to explain the expected counts of the events E(Yt );λt , given a certain period of time (unit of time) t. Supposing that xt β represents all explanatory variables for the analysis, the expected counts are described as λt = exp(xt β).8 The dependent variable is the quarterly count of South Korea’s hostile actions against Japan during the period 1988–2011. We focus only on the actions by the government and elites. As explained in the previous section, South Korea’s hostile behavior consists of verbal and material actions. We begin by estimating the model with hostile actions regardless of types of Figure 2 South Korea’s hostile foreign policy behavior against Japan, 1988–2011. 8 See Kennedy (2003) and King (1998)for details of the Poisson regression model. South Korean diversionary foreign policy 43 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from actions (Models 1 and 2), and then running the model with verbal and material actions separately (Models 3–6). The independent variables consist of the state of the South Korean economy, the presidential approval rate, and national elections. First, we use GDP per capita growth rate and misery index separately to capture the state of the South Korean economy. We calculate GDP per capita growth rate as follows. Since the quarterly data of real GDP per capita are affected by seasonal fluctuations and errors, we purge those effects by calculating the moving average of four-quarter (one-year) GDP per capita, which we define as the quarterly GDP per capita. Then, we calculate the GDP per capita growth rate by comparing the current quarter to the previous quarter. This quarterly GDP per capita reflects both trend and cyclical components of time series and is helpful for capturing the short-time swings. This allows us to examine how leaders respond to short-term economic volatilities. GDP per capita growth rate represents interests of the middle-income cohort and a higher growth rate is expected to reduce the number of hostile actions. In contrast, misery index is the sum of the inflation and unemployment rates, which indicates economic pressures on the lower income cohort. This is based on the assumption that both a higher inflation rate and a higher unemployment rate reflect socio-economic costs. Thus, a high misery index describes the situation wherein a leader confronts social anxiety. We predict that a higher misery index increases the hostile action count.9 Second, the political approval rate represents the level of political support reflecting the president’s availability of policy options. We use the presidential approval data based on the Gallup quarterly survey.10We expect that higher approval ratings reduce the number of hostile behavior. Third, neither the presidential nor legislative elections encourages leaders to engage in foreign adventurous projects in order to appeal to the public. We use the binary variables to describe electoral incentives. The presidential election variable equals 1 if the quarter is one 9 We used the data of real GDP and inflation rate from International Financial Statistics of the International Monetary Fund (elibrary-data.imf.org) and population data from World Development Indicators of the World Bank (data.worldbank.org/country/korea-republic). We also calculated the quarterly unemployment rate as the average value of the monthly unemployment rate data from LABORSTA website of the International Labor Organization (laborsta.ilo.org). These data were retrieved on 19 May 2012. 10 We used the value of the previous term to substitute for the missing value. We retrieved the data from Gallup on 17 May 2012 (www.gallup.co.kr). 44 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from quarter before the presidential election, and 0 otherwise. The legislative election variable equals 1 if the quarter is one quarter before the legislative election, and 0 otherwise. We also include the other possible explanatory variables for South Korea’s hostile behavior in the model. First, we include the change in Japanese prime ministers because Japan’s new leaders have less experience in handling foreign challenges and are more likely to be risk-averse in response to South Korea’s diversionary challenges (Kagotani and lida, 2012). The binary variable is equal to 1 if Japan experiences a change in prime minister, and 0 otherwise. The expected effect of a Japanese prime minister change is positive. Second, Japan’s firm reaction may escalate Japan–South Korea disputes and increase the number of events. We include the counts of Japan’s hostile actions against South Korea, which might have a positive correlation with the dependent variable. Third, realists argue that the most primary goal of states is to secure their survival and that the international system forces states to prepare for potential external threats. Therefore, we include the total count of hostile actions by other rivals, such as China, North Korea, Russia, and Taiwan, to describe the security environment surrounding South Korea. Since external threats do not allow South Korea to focus on Japan–South Korea disputes, heightened tensions with other adversaries might reduce the number of hostile behavior against Japan. External threats might have a negative correlation with the dependent variable. Finally, we address the autocorrelation in the count of South Korea’s hostile actions. The results of the augmented Dickey–Fuller test detected stationary processes in the counts of all hostile actions, verbal actions, and material actions. Correlograms found autocorrelation in the counts of all hostile actions and verbal actions. To control the autoregressive correlation, we follow the modeling approach of Brandt and Williams (2001) and redefine the Poisson model with autoregressive processes of order p, called the PAR(p). In this model, the mean count of hostile action mt is the conditional of the linear AR process of E[yt |Yt−1 ]. Since correlograms show the existence of autocorrelation in the total count of South Korea’s hostile actions and the count of its verbal hostile actions, we use the PAR (p) to examine the effect of the independent variables. After we estimate the Poisson regression, looking at the PAR(1), and the PAR(2) models, we report the results with the highest likelihood and the lowest Akaike South Korean diversionary foreign policy 45 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from Information Criteria (AIC).11 The details of the PAR( p) model are presented in Appendix. In contrast, we use the standard count models to analyze the material action counts because a correlogram did not find an autocorrelation. The Poisson regression model assumes that the expected counts are equal to the variance of the counts: E(Yt ) = Var(Yt ) =λt . However, this assumption does not always hold. The variance of the counts may show overdispersion E(Yt ) < Var(Yt ). In such a case, we have to use the binomial regression model, which takes overdispersion into account. After estimating the Poisson regression, the constant-dispersion negative binomial regression (NB1), and the mean-dispersion negative binomial regression (NB2), we report the results with the highest likelihood and the lowest AIC.12 Table 1 shows the summary statistics of the variables used in the count models of South Korean Diversion during the period 1988–2011. 5 Results The data cover the period 1988–2011; Table 2 shows the estimated results. First, the effect of economic variables is consistent with our hypothesis and, except for Model 3, is statistically significant regardless of the types of hostile behavior. Second, we found no effect of the presidential approval ratings on South Korea’s hostile behavior because a lack of a presidential re-election negates the diversionary incentive. Third, we also did not detect any evidence across the various models of the presidential elections shaping South Korea’s behavior due to the prohibition on seeking reelection. However, legislative elections significantly reduce the counts of South Korea’s verbal hostile actions, which is contrary to our prediction (Models 3 and 4). In South Korea, political parties have similar foreign policy stances toward Japan and they cannot appeal to the electorate by criticizing Japan. Parties focus more on the state of the economy and on domestic issues, making them less likely to take on adventurous foreign policies during the legislative election campaigns. Fourth, the effect of a change in Japan’s prime minister is consistent with our expectation and shows strong significance over the models. Fifth, hostile actions on the part of Japan generate the predicted effect and strong significance over the 11 We used R version 2.15.3 and the source code of Brandt and Williams (2001) to estimate the PAR( p) models. 12 See Long (2006) for the detail of the negative binomial regression models. We used STATA version 11.2 to estimate the Poisson regression and the negative binomial regression models. 46 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from models. On the other hand, external threats by South Korea’s rivals show the marginal predicted effect on all hostile actions and on verbal hostile actions (Models 1–4). The autoregressive parameter ρ shows the existence of marginal autocorrelation. The dispersion parameter δ shows the difference between the expected counts and the variance of counts. We focus on the effects of economic variables, change of prime ministers in Japan, and hostile actions on the part of Japan at quarter t on all hostile behavior in the models (Models 1 and 2) because the models with only verbal hostile behavior show similar results. Note that the variables have both short-term (instantaneous) and long-term (dynamic) effects on the conditional mean count of events mt (Brandt and Williams 2001, pp.169–170). They are measured by a one-unit increase in the variables of interest. First, the short-term effect of the GDP per capita growth rate is −0.13 counts, whereas its long-term effect is −0.14 counts. The short-term effect of the misery rate is 0.17 counts, whereas its long-term effect is 0.19 counts. These demonstrate that economic turmoil is a driving force Table 1 Summary statistics Variable Mean SD Min Max N Dependent variable South Korea’s hostile action against Japan (all) 5.70 8.01 0 45 96 South Korea’s hostile action against Japan (verbal) 3.78 5.90 0 34 96 South Korea’s hostile action against Japan (material) 1.06 1.70 0 9 96 Independent variable GDP growth rate 1.19 0.93 −2.22 2.93 90 Misery index 4.45 1.38 1.91 10.86 96 Public approval 34.57 16.50 6 83 96 Presidential election 0.04 0.20 0 1 96 Legislative election 0.06 0.24 0 1 96 Japan’s PM change 0.16 0.37 0 1 96 Japan’s hostile action against South Korea (all) 4.33 4.89 0 21 96 Japan’s hostile action against South Korea (verbal) 2.58 3.38 0 17 96 Japan’s hostile action against South Korea (material) 1.14 1.54 0 6 96 Rivals’ hostile action against South Korea (all) 10.37 11.92 0 92 96 Rivals’ hostile action against South Korea (verbal) 5.32 6.19 0 39 96 Rivals’ hostile action against South Korea (material) 2.99 4.22 0 32 96 South Korean diversionary foreign policy 47 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from Table 2 Count model of South Korea’s diversion All Verbal Material Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 PAR(1) PAR(2) PAR(2) PAR(1) NB1 NB1 GDP per capita growth rate −0.12** (0.06) −0.10 (0.07) −0.30** (0.14) Misery index 0.18*** (0.05) 0.20*** (0.06) 0.25** (0.10) Presidential approval rate 0.004 (0.003) −0.0007 (0.004) 0.01** (0.004) 0.002 (0.005) 0.001 (0.01) −0.01 (0.01) Presidential election −0.82* (0.50) −0.76 (0.54) −0.33 (0.52) −0.51 (0.59) −16.51 (1460.23) −15.13 (836.94) Legislative election −0.28* (0.16) −0.14 (0.15) −0.74** (0.29) −0.72*** (0.29) 0.58 (0.51) 0.35 (0.49) Japan’s PM change 0.29** (0.12) 0.52*** (0.12) 0.30* (0.15) 0.63*** (0.14) 0.80** (0.34) 1.01*** (0.30) Japan’s hostile action 0.14*** (0.01) 0.14*** (0.01) 0.22*** (0.02) 0.19*** (0.02) 0.37*** (0.09) 0.42*** (0.08) Rivals’ hostile action −0.02*** (0.01) −0.02*** (0.01) −0.03* (0.02) −0.06*** (0.02) 0.01 (0.04) −0.02 (0.03) Constant term 1.03*** (0.19) 0.18 (0.28) 0.42* (0.25) −0.23 (0.31) −0.36 (0.47) −1.51** (0.52) ρ1 0.07* (0.04) 0.08* (0.05) 0.08* (0.04) 0.08* (0.04) ρ2 0.03 (0.03) −0.05* (0.03) δ 1.08*** (0.45) 0.89*** (0.39) Log likelihood −233.47 −265.38 −182.78 −223.57 −108.01 −117.88 AIC 482.94 548.75 383.56 463.13 235.23 253.76 Wald chi 2 2.52 4.59 4.25 3.37 31.15 35.05 Number of observation 90 96 90 96 90 96 *P< 0.1; **P< 0.05; ***P< 0.01. 48 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from underlying South Korea’s diversionary foreign policy. Next, the short-term effects of a Japanese prime minister change are 0.28 and 0.52 counts in Models 1 and 2, respectively, and its long-term effects are 0.31 and 0.58 counts in Models 1 and 2, respectively. Changing prime ministers in Japan has a huge impact on the mean counts of South Korea’s hostile behavior, which means that a new Japanese prime minister opens a window of opportunity. Finally, the short-term effects of Japan’s hostile actions toward South Korea are 0.15 and 0.13 counts in Models 1 and 2, respectively. Its long-term effects are 0.16 and 0.15 counts in Models 1 and 2, respectively. The results reflect the fact that Japan’s firm responses quickly escalate disputes between Japan and South Korea within quartert. The results of the models with only material hostile actions validate our findings (Models 5 and 6). Since the models have no autocorrelation, the effect of each explanatory variables can be simply described by the incidence-rate ratios. First, a one-point increase in GDP per capita growth rate generates a 26% of reduction in the mean count of South Korea’s material hostile behavior in quarter t (i.e. a 0.12 count reduction). A onepoint increase in misery index generates a 28% of increase in the mean count of South Korea’s material hostile behavior in quarter t (i.e. a 0.11 count increase). These results show that economic distress drives leaders to divert public attention toward disputes with Japan. Second, when a new Japanese prime minister comes into power, the mean count of South Korea’s material hostile behavior increases by 122 and 175% in Models 5 and 6, respectively (i.e. a 0.43 count increase and a 0.66 count increase). The results imply that leaders strategically exploit opportunities to attain foreign policy success by targeting new Japanese prime ministers. Third, a one-count increase in Japan’s material hostile behavior toward South Korea generates a 44 and 52% increase in the mean count of South Korea’s material hostile behavior (i.e. a 0.14 count increase and a 0.18 count increase, respectively). Thus, the results indicate that disputes can quickly escalate due to Japan’s firm reactions. In short, we found three significant driving forces underlying the fluctuations of tensions between Japan and South Korea: (1) economic diversionary incentives, (2) windows of opportunity based on changes in Japanese prime ministers, and (3) Japan’s firm reactions. South Korea’s leaders are concerned with both their own domestic distress and Japan’s domestic turmoil. The escalation of disputes can be explained by strategic interactions between Japan and South Korea. South Korean diversionary foreign policy 49 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from 6 Two cases of economic diversion The presidencies of Roh Tae-woo (1988–1993) and Kim Young-sam (1993–1998) provide two examples where economic distress was a driving force behind South Korean diversionary behavior. In these examples, both presidents started their tenure with moderate attitudes toward Japan and consequently took hawkish foreign policy positions in response to an economic downturn in the middle of their tenures. We briefly describe these examples to explore the economic diversionary process behind the findings of our analysis, outlined in the previous section. President Roh came into office in February 1988 immediately after democratization. In the first top-level meeting with Japanese PM Noboru Takeshita, he publicly announced the priority of establishing a good relationship with Japan (Asahi Shimbun, 26 February 1988). Roh hoped to visit Japan at the beginning of his tenure, but was unable to visit until 1990. During this period, he confronted strong domestic opposition to his visit due to prevailing public criticism of Japan’s colonial occupation of South Korea (Hankyoreh, 22 May 1990). However, his policy stance toward Japan remained moderate, and he made efforts to build trust between the two countries. In response to his first visit, Japanese PM Toshiki Kaifu and Emperor Hirohito formally apologized for Japan’s colonial occupation. Roh requested that Emperor Hirohito visit South Korea and expressed forgiveness for Japan’s colonial occupation of South Korea (Asahi Shimbun, 25 May 1990), which mitigated grievances in South Korea. This demonstrates that a president does not always resort to a hawkish foreign policy due to the hawkish preferences of domestic constituencies. An increasing trade deficit with Japan, however, prompted the diversionary incentives after 1991. The growth of the South Korean economy was driven by export substitution industrialization, in which South Korea relied heavily on the imports of intermediate and producer’s goods from Japan (Watanabe and Kim, 1996, pp. 236–270). Export expansion stimulated more imports from Japan and created a huge trade deficit, which made the South Korean economy more vulnerable to currency crisis (Watanabe and Kim, 1996, pp. 239–240). In a 1991 top-level meeting with Japanese PM Kaifu, Roh attributed this problem to the lack of technology transfer from Japan. Roh requested a settlement of trade deficits and the promotion of technology transfer, but he obtained few results (Asahi 50 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from Shimbun, 11 January 1991). In a 1992 top-level meeting, Roh sought to obtain more concessions from Japan, but Japanese PM Kiichi Miyazawa firmly declined his requests (Asahi Shimbun, 18 January 1992). In the meeting, Roh addressed the comfort women/sex slaves issue to gain more bargaining leverage because five days prior to the meeting, the Japanese newspaper, Asahi Shimbun, reported on Japan’s military involvement in comfort women/sex slaves mobilization during the war (Asahi Shimbun, 11 January 1992). Although Roh did not gain a favorable outcome on the economic issue, Miyazawa apologized for the comfort women/sex slaves issue thirteen times during the two days of meetings (Mainichi, 18 January 1992). Roh achieved political gains by switching the economic issue to the comfort women/sex slaves issue. The political benefits of diversion became clearly shortly after the meeting, as the South Korean public never criticized Roh for failing to obtain improved economic outcomes (Hankyoreh, 18 January 1992). Roh continued to take a hard line toward Japan by repeatedly addressing the historical understanding issues until the end of his tenure (Kyunghyang Shinmun, 19 January 1992). Thus, this example illustrates how economic distress drives a president to divert public attention from an economic issue to a political one. Next, President Kim Young-sam came into office in 1993 and confronted nationalistic grievances against Japan that were provoked by his predecessor Roh. Kim was expected to take a hard line because he was the first president with no military background since the coup détat in 1961 (Mainichi, 19 December 1992). However, his foreign policy stance toward Japan was much more moderate than expected. Kim stated that the South Korean government would not ask for compensation regarding the comfort women/sex slaves issue. In a sign that Kim’s foreign policy stance was more conciliatory in the beginning of his tenure, he reportedly intended to seek a political settlement for both the comfort women/sex slaves and historical understanding issues (Asahi Shimbun, 14 March 2013), which was not accepted by domestic constituencies. In 1995, Japanese PM Tomiichi Murayama and his cabinet members provoked criticism from South Korea after delivering a speech about the comfort women/sex slaves issue (Kawano, 2001). Later in 1995, Murayama attempted to settle tensions between Japan and South Korea but failed to do so (Kimura, 2013, pp. 20–23). In February 1996, Japan ratified the UN Convention on the Law of the Sea and claimed that its exclusive economic zone included Takeshima/Dokdo and its adjacent South Korean diversionary foreign policy 51 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from waters. South Korea strongly opposed Japan’s territorial claim and escalated disputes by shifting focus from the comfort women/sex slaves issue to the territorial issue (Mainich, 9 February 1996). However, Kim’s nationalistic response was criticized even in South Korea. The head of the largest opposition party, Kim Dae-jung, complained that President Kim Young-sam unnecessarily escalated the dispute (Mainichi, 16 February 1996). After all, Japan and South Korea settled the dispute by laying the territorial issue aside (Mainichi, 24 June 1996). This settlement would have temporarily improved the Japan–South Korea relationship, but Kim continued to take a hard line against Japan. Then, South Korea addressed the comfort women/sex slaves issue again. At the meeting of foreign ministers, South Korea strongly requested that the Japanese government pay official compensation to the former comfort women/sex slaves (Mainichi, 16 January 1997). South Korea also finished building a pier and facilities and enhanced its effective control over Takeshima/Dokdo in November 1997 (Mainichi, 4 November 1997). Kim’s hawkish attitudes can be explained by economic diversionary incentives. First, the South Korean economy started declining after the latter half of 1996 and was hit by financial crisis in 1997. According to our data, GDP per capita growth rate decreased from 1.22 to −1.75 and the misery index increased from 3.31 to 5.84 from the third quarter of 1996 to the fourth quarter of 1997. In 1998, the state of the economy worsened. At the time, Kim was facing economic turmoil and, therefore, had an incentive to divert public attention from financial crisis to the territorial issue. This example also shows that economic distress serves as a driving force behind a president’s diversionary behavior. In short, we summarize two important points in these examples. First, both Presidents Roh Tae-woo and Kim Young-sam sought to divert public attention from their respective economic downturns to Japan–South Korean disputes. Second, Roh succeeded in creating a foreign diversion by obtaining political gains from Japan, whereas Kim gambled and lost his support by unnecessarily worsening the Japan–South Korea relationship. These examples validate our empirical findings as well as our theoretical arguments. 7 Conclusion Since democratization, South Korea has increasingly raised the issues of the Japanese history textbooks, the comfort women/sex slaves in the prewar 52 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from period, and the territorial disputes over Takeshima/Dokdo. To understand the fluctuations in South Korea’s foreign policy, we investigated the political and economic determinants of their foreign policy in the period 1988–2011. Our findings suggest that South Korea’s economic performance is an important factor in shaping its foreign policy behavior toward Japan. The empirical analysis provides striking findings that characterize foreign policy making in a new democracy. First, economic distress drives South Korea’s leaders to divert public attention toward the Japan–South Korea disputes. Second, a drop in public approval ratings does not encourage leaders to resort to a hawkish foreign policy because the public’s loyalty shifts to the potential candidates in the next election due to the five-year presidential term limit. Thus, the president becomes less sensitive to public approval. Third, national elections, whether presidential or legislative, do not stimulate diversionary incentives in candidates to show their competencies in order to further electoral benefits. Only economic diversionary incentives characterize the fluctuations in South Korean foreign policy. The results also provide implications for managing the Japan–South Korea disputes. First, economic turmoil encourages South Korean leaders to escalate the Japan–South Korea disputes. In the introduction section, we mentioned President Lee Myung-bak’s visit to Takeshima/Dokdo. A failure to renew the temporal bilateral currency swap facilities could worsen the state of the South Korean economy and then subsequently escalate the Japan–South Korea disputes. To mitigate tensions between Japan and South Korea, the Japanese government should take South Korean leaders’ diversionary incentives into consideration. Second, area experts often attribute historical revisionism to hawkish foreign policy actions (Rozman and Lee, 2006), but we confirmed that those policy actions reflect the existence of economic rationality. To avoid provoking grievance on the part of the Japanese public in reaction to the media attention on South Korean nationalist sentiment, our findings are helpful in explaining South Korea’s foreign policy toward Japan since democratization. Third, the degree of US commitment to East Asia is thought to determine the influence of historical antagonism on the fluctuations in the Japan–South Korea relationship because Japan and South Korea are both American allies (Cha, 2000). However, we detected economic diversionary incentives in South Korea that are independent of the US commitment to East Asia. Policy makers need to understand that although a stronger US commitment might worsen the Japan–South Korea relationship by South Korean diversionary foreign policy 53 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from dampening their incentives to cooperate, taming the volatility of the South Korean economy is helpful for managing the Japan–South Korea disputes. References Azar, E.E. (1980) ‘The conflict and peace data bank (COPDAB) project’, Journal of Conflict Resolution, 24(1), 143–152. Brandt, P.T. and Williams, J.T. (2001) ‘A linear Poisson autoregressive model: The Poisson AR(p) model’,Political Analysis, 9(2), 164–184. Brulé, D. (2006) ‘Congressional opposition, the economy, and U.S. dispute initiation, 1946–2000’, Journal of Conflict Resolution, 50(4), 463–483. Brulé, D.J. (2008) ‘Congress, presidential approval, and U.S. dispute initiation’, Foreign Policy Analysis, 4(4), 349–370. Cha, V.D. (2000) ‘Abandonment, entrapment, and neoclassical realism in Asia: The United States, Japan, and Korea’, International Studies Quarterly, 44(2), 261–291. Dalton, R.J. and Tanaka, A. (2007) ‘The patterns of party polarization in East Asia’, Journal of East Asian Studies, 7(2), 203–223. DeRouen, K.R., Jr. (1995) ‘The indirect link: politics, the economy, and the use of force’, Journal of Conflict Resolution, 39(4), 671–695. Downs, G.W. and Rocke, D.M. (1995) Optimal Imperfection? Domestic Uncertainty and Institutions in International Relations. New Jersey: Princeton University Press. Enterline, A.J. (1998a) ‘Regime changes and interstate conflict, 1816–1992’, Political Research Quarterly, 51(2), 385–409. Enterline, A.J. (1998b) ‘Regime changes, neighborhoods, and interstate conflict, 1816–1992’, Journal of Conflict Resolution, 42(6), 804–829. Fordham, B. (1998) ‘The politics of threat perception and the use of force: a political economy model of U.S. uses of force, 1949–1994’, International Studies Quarterly, 42(3), 567–590. Fordham, B.O. (2002) ‘Another look at “parties, voters, and the use of force abroad”’, Journal of Conflict Resolution, 46(4), 572–596. Gelpi, C. (1997) ‘Democratic diversions: governmental structure and the externalization of domestic conflict’, Journal of Conflict Resolution, 41(2), 255–282. Ghosn, F., Palmer, G. and Bremer, S.A. (2004) ‘The MID3 data set, 1993–2001: procedures, coding rules, and description’, Conflict Management and Peace Science, 21(2), 133–154. Goldstein, J.S. (1992) ‘A conflict-cooperation scale for WEIS events data’, Journal of Conflict Resolution, 36(2), 369–385. Haggard, S. (1990) Pathways From the Periphery: The Politics of Growth in the Newly Industrializing Countries. Ithaca, NY: Cornell University Press. 54 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from Im, H.B. (2004) ‘Faltering democratic consolidation in South Korea: democracy at the end of the ‘three Kims’ era’, Democratization, 11(5), 179–198. James, P. and Oneal, J.R. (1991)‘The influence of domestic and international politics on the president’s use of force’, Journal of Conflict Resolution, 35(2), 307–332. Kagotani, K. and Iida, T. (2012) Scramble for security: domestic turmoil and tensions in interstate rivalries. mimeo. Trinity College Dublin. Kawano, N. (2001) ‘Kakuryōshitsugen no seiji-gaku [politics of the slips of the tongue of the ministers]’, Kokusai Kyōryoku Kenkyū-shi [Journal of International Development and Cooperation], 7(1), 19–35. Kennedy, P. (2003) A Guide to Econometrics, 5th edn. Cambridge: MIT Press. Kimura, K. (2013) ‘Murayama danwa kara ajiajoseikikin he [from Murayama statement to the Asian Women’s Fund]’, Kiwameru, 30, 20–23. King, G. (1998) Unifying Political Methodology: The Likelihood Theory of Statistical Inference. Ann Arbor: University of Michigan Press. King, G. and Lowe, W. (2003a) 10 million international dyadic events. http://hdl. handle.net/1902.1/FYXLAWZRIA UNF:3:dSE0bsQK2o6xXlxeaDEhcg== IQSS Dataverse Network [Distributor] V3 [Version]. (7 January 2013, date last accessed). King, G. and Lowe, W. (2003b) ‘An automated information extraction tool for international conflict data with performance as good as human coders: a rare events evaluation design’,International Organization, 57(3), 617–642. Kong, T.Y. (2000) The Politics of Economic Reform in South Korea: A Fragile Miracle. London: Routledge. Lai, B. and Reiter, D. (2005) ‘Rally ‘round the union jack? Public opinion and the use of force in the United Kingdom, 1948–2001’, International Studies Quarterly, 49(2), 255–272. Leetaru, K. and Schrodt, P.A. (2013) GDELT: Global data on events, location and tone, 1979–2012. Paper presented at the International Studies Association meetings, San Francisco, April 2013. Levy, J.S. (1989) ‘The diversionary theory of war: a critique’, in M.I. Midlarsky (ed.), Handbook of War Studies, pp. 259–288. Massachusetts: Unwin Hyman. Lian, B. and Oneal, J.R. (1993) ‘Presidents, the use of military force, and public opinion’, Journal of Conflict Resolution, 37(2), 277–300. Lind, J. (2011) ‘Democratization and stability in East Asia’, International Studies Quarterly, 55(2), 409–436. Long, J.S. (2006) Regression Models for Categorical Dependent Variables Using Stata, 2nd edn. Texas: STATA Press. Mansfield, E.D. and Snyder, J. (1995a) ‘Democratization and the danger of war’, International Security, 20(1), 5–38. Mansfield, E.D. and Snyder, J. (1995b) ‘Democratization and war’, Foreign Affairs, 74(3), 79–97. South Korean diversionary foreign policy 55 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from Mansfield, E.D. and Snyder, J. (2002a) ‘Democratic transitions, institutional strength, and war’,International Organization, 56(2), 297–337. Mansfield, E.D. and Snyder, J. (2002b) ‘Incomplete democratization and the outbreak of military disputes’,International Studies Quarterly, 46(4), 529–549. Mansfield, E.D. and Snyder, J. (2005) Electing to Fight: Why Emerging Democracies Go to War. Cambridge, MA: MIT Press. Mansfield, E.D. and Snyder, J. (2009) ‘Pathways to war in democratic transitions’, International Organization, 63(2), 381–390. Maoz, Z. (1998) ‘Realist and cultural critiques of the democratic peace: a theoretical and empirical re-assessment’,International Interactions, 24(1), 3–89. Marra, R.F., Ostrom, C.W. and Simon, D.M. (1990) ‘Foreign policy and presidential popularity: creating windows of opportunity in the perpetual election’, Journal of Conflict Resolution, 34(4), 588–623. McClelland, C.A. (1983) ‘Let the user beware’,International Studies Quarterly, 27 (2), 169–177. Meernik, J. and Waterman, P. (1996) ‘The myth of the diversionary use of force by American presidents’,Political Research Quarterly, 49(3), 573–590. Miller, R.A. (1995) ‘Domestic structures and the diversionary use of force’, American Journal of Political Science, 39(3), 760–785. Moon, W. (2005) ‘Decomposition of regional voting in South Korea: ideological conflicts and regional interests’,Party Politics, 11(5), 579–599. Morgan, T.C. and Bickers, K.N. (1992) ‘Domestic discontent and the external use of force’, Journal of Conflict Resolution, 36(1), 25–52. Narang, V. and Nelson, R.M. (2009) ‘Who are these belligerent democratizers? Reassessing the impact of democratization on war’,International Organization, 63(2), 357–379. Ostrom, C.W. and Job, B.L. (1986) ‘The president and the political use of force’, American Political Science Review, 80(2), 541–566. Richards, D., Morgan, T.C., Wilson, R.K., Schwebach, V.L. and Young, G.D. (1993) ‘Good times, bad times, and the diversionary use of force: a tale of some not-so-free agents’, Journal of Conflict Resolution, 37(3), 504–535. Rozman, G. and Lee, S.-w. (2006) ‘Unraveling the Japan-South Korea “virtual alliance”: populism and historical revisionism in the face of conflicting regional strategies’, Asian Survey, 46(5), 761–784. Shirai, K. (2009) ‘Kankoku kokkai no tainichidōkō[trends in South Korean legislative behavior towards Japan]’, Gaikoku no Rippō[Lawmaking in Foreign Countries], 240, 6–7. Shugart, M.S. and Carey, J.M. (1992) Presidents and Assemblies: Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University Press. Smith, A. (1996) ‘Diversionary foreign policy in democratic systems’, International Studies Quarterly, 40(1), 133–153. 56 Koji Kagotani, Kan Kimura, and Jeffrey R. Weber at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from Tarar, A. (2006) ‘Diversionary incentives and the bargaining approach to war’, International Studies Quarterly, 50(1), 169–188. Thompson, W.R. and Tucker, R. (1997) ‘A tale of two democratic peace critiques’, Journal of Conflict Resolution, 41(3), 428–454. Ward, M.D. and Gleditsch, K.S. (1998) ‘Democratizing for peace’, American Political Science Review, 92(1), 51–61. Watanabe, T. and Kim, C.-N. (1996) Kankoku Keizai Hatten-ron [Theory of South Korean Economic Development]. Tokyo: Keiso Shobo. Appendix: PAR(p) model Brandt and Williams (2001) derived a pth-order linear autoregressive process and extended it to a pth-order Poisson autoregressive (PAR(p)) model. When Yt represents all the prior information about the series of South Korea’s hostile action at time t, assume that yt is realized from a time-homogeneous Markov process with the conditional transition probability Pr(yt |Yt–1 ) and that E[Y0 ] =μ<∞. The model is described as a stationary linear AR(p) process: E½yt jYt1 ¼Xp i¼1 ri Yt1 þl; ð1Þ where ρi and λ are any real numbers. When we can find iterated expectations for the AR(p) process in the equation (1), the generalized result is as follows: E½Yt  ¼ E½E½yt jYt1 ¼ E Xp i¼1 rYti þl " # ¼ Xp i¼1 E½rYti  þl: ð2Þ Since the equation above is a geometric series for ρi , we can derive the following equation: limt!1 E½Yt  ¼ l 1 Pp i¼1 ri ; m: ð3Þ We transform equation (1) into a stationary linear AR(p) with the admissible set of autoregressive coefficients ρi : E½yt jYt1 ¼Xp i¼1 ri Yti þ 1 Xp i¼1 ri !m: ð4Þ South Korean diversionary foreign policy 57 at Columbia University Health Sciences Library on January 30, 2014 http://irap.oxfordjournals.org/ Downloaded from This linear AR(p) model is combined with the assumption that the event counts follow a Poisson distribution in order to define a new transition equation for a state-space model with an AR(p) process. This Poisson autoregressive or PAR(p) model is defined as follows. First, the observed event counts, yt for t= 1, 2, …, T, are drawn from a Poisson distribution conditional onmt : Prðyt jmt Þ ¼ m yt t e mt yt ! : ð5Þ The state variable is defined by the assumption that mt is the conditional mean of the linear AR process of E[yt |Yt–1 ]. Given the Poisson distribution, the state density is in the exponential family and is characterized by its mean mt =E[yt |Yt–1] and σt =Var[yt |Yt–1]. Finally, we assume that the density of the state variable has a gamma distributed conjugate prior: Prðmt jYt1Þ ¼ Gðst1 mt1; st1Þ; for mt1 . 0; and st1 . 0: ð6Þ The prior distribution is a gamma with mean mt–1 and variance mt1=st1: We introduce the covariates by assuming μ= exp(xt β). The short-term effect of a one-unit change in the regressor xt is as follows: @mt @xt ¼ 1 Xp i¼1 ri !expðxt bÞ b: ð7Þ The long-term effect of a one-unit change in the regressor xt is as follows: @mt =@xt 1 Pp i¼1 ri   ¼ expðxt bÞ b: ð8Þ