UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The capitalist peace revisited : a new liberal peace model and the impact of market fluctuations Han, Zhen 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2012_spring_han_zhen.pdf [ 776.97kB ]
JSON: 24-1.0072640.json
JSON-LD: 24-1.0072640-ld.json
RDF/XML (Pretty): 24-1.0072640-rdf.xml
RDF/JSON: 24-1.0072640-rdf.json
Turtle: 24-1.0072640-turtle.txt
N-Triples: 24-1.0072640-rdf-ntriples.txt
Original Record: 24-1.0072640-source.json
Full Text

Full Text

The Capitalist Peace Revisited: A New Liberal Peace Model and the Impact of Market Fluctuations  by Zhen Han B.Sc., University of Electronic Science and Technology, China, 2001 B.A., The University of British Columbia, 2010  A THESIS SUBMITTED IN PARTIAL FUFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in The Faculty of Graduate Studies (Political Science)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  March 2012  © Zhen Han, 2012  Abstract Gartzke (2007) argues that countries which are more open to the global financial market are less likely to have militarized interstate conflicts, and he also argues that democratic peace is spurious once the market openness is taken into consideration. However, this argument has been weakened due to the missing values in his data and flaws in the modeling. Furthermore, the negative impact of financial globalization, such as the lack of surveillance and market fluctuation, need to be considered. This paper proposes a new liberal peace model with careful attention to the problems of missing data and model misspecifications, and the results of the new liberal model shows that while traditional democratic peace and commercial peace remain robust in the new era of global financial integration, the pacifying effect of commercial ties has been weakened by the volatile fluctuations in the international financial market.  i  Table of Contents  Abstract ............................................................................................................................................ i Table of Contents ............................................................................................................................ ii List of Tables ................................................................................................................................. iii List of Figures ................................................................................................................................ iv Acknowledgments............................................................................................................................v 1  Introduction ..........................................................................................................................1  2  Democratic Peace vs. Commercial Peace ............................................................................4 2.1  Democratic Peace.....................................................................................................4  2.2  Commercial Peace .................................................................................................10  3  Methodology ......................................................................................................................17  4  Data and Variables .............................................................................................................20  5  Findings..............................................................................................................................28  6  Discussion and Conclusion ................................................................................................36  Bibliography .................................................................................................................................42 Appendix A: The results of different approaches to control temporal dependence ......................46 Appendix B: The results of comparing different methods to control the missing values in the capital net inflow variable ..............................................................................................................48 Appendix C: The predicted probability of conflicts, under an alternative definition of democracy ........................................................................................................................................................49  ii  List of Tables  Table 1  Descriptive Statistics ..............................................................................................23  Table 2  Missing Values Caused by the Capital Flow Variable ..........................................25  Table 3  Models and Coefficients ........................................................................................29  iii  List of Figures  Figure 1  Predicted Probability of Conflicts in different types of dyads (Model 3.1)...........33  Figure 2  Predicted Probability of Conflicts, on Condition to the Types of Dyads (Model 3.2) .................................................................................................................................33  iv  Acknowledgement I would like to offer my gratitude to the faculty, stuff and my fellow students in the UBC political science department for inspiring my interests in the field of international political economy and supporting me throughout this research project. In particular, I would like to thank my supervisor, Dr. Benjamin Nyblade, as his knowledge and attitude towards scientific research have guided me through my studies. I would like to thank Dr. Gyung-Ho Jeong for reviewing this thesis and encouraging me to pursue this topic further. I would like to thank Dr. Angela O’Mahony , Stewart Prest, and Go Murakami for helping me improving my research by their thoughtful comments and suggestions. Finally, I would like to thank my family for always, financially and morally, supporting my studies in the political science department of UBC.  v  1. Introduction Does global economic integration make the world more peaceful? From Kant to Norman Angel, the school of commercial peace has argued that states are less likely to fight each other if they are integrated by close economic ties1. While this commercial peace argument has received some attentions from scholars and political leaders, the evolving nature of globalization brings some new challenges to this argument. Since the late 1970s, financial liberalization has become an important aspect of the new waves of globalization2. Economic integration is no longer just about commodity trade; therefore, some new questions can be raised for commercial peace theory in the era of financial liberalization. Does financial market integration also produce the same pacifying effect between states, as the traditional commercial peace would? Giving the volatile nature of financial markets, one can reasonably question whether international financial crises can spill over to the political sphere and cause international conflicts. The importance of this question is not simply academic. On September 15th 2011, Jacek Rostowski, the finance minister of Poland, warned the committee of the EU finance ministry that if no prompt actions are taken to fix the 2011 Euro crisis, there will be ―war in 10 years‖ 3. The intensity of the 2011 Euro crisis has reached a high level, and political leaders have openly expressed their concerns of social turmoil or even international conflict in Europe.  1  Angell, 1912. Kant, 1795. For example, Goodman and Pauly argues that since the 1980s, financial liberalization have weakened the power of states and strengthened the power of corporations. Goodman and Pauly, 1993. 3 The report of Rostowski’s speech can be found at http://euobserver.com/18/113625. September 14th, 2011. EUobserver. Retrieved on September 18th, 2011. 2  1  Erik Gartzke’s prize-winning paper ―the Capitalist Peace‖ is a pioneer of testing the relations between market openness in this new era and interstate conflicts by combining commercial peace data and economic data on market openness4. He finds that if two states have a higher level of market openness, they are less likely to have interstate conflicts5. His findings also challenge the democratic peace theory, as he finds that democracy has no significant correlation with peace once control for market openness6. These findings raise questions of the two major components—democratic peace and commercial peace—of the liberal peace model. Some recent studies challenge Gartzke by arguing that 1.) His measurement of democracy is problematic7; 2.) Missing values in his data has created a selection bias in his conclusion; 3.) Temporal dependence and regional dependence are not properly controlled for in his statistical model8. The debate between Gartzke and his critics begs a new liberal peace model—a model in which democracy and market openness are properly measured, and dependence across cases are properly controlled. This thesis tries to build and test such a model with data focusing on later time periods when the data availability is better. In section 2, this paper reviews the current debate on the two pillars of liberal peace—democratic peace theory and commercial peace theory, and suggests some new measurements to improve conventional quantitative studies on liberal peace. Section 3 reviews the debate on methodological issues, such as the proper way to control temporal dependence, and proposes a new statistical model to test the liberal peace theory. Section 4 introduces the datasets used in this study. The missing variable problem, which has poisoned some previous studies on  4  Gartzke, 2007. Ibid. pp.176. 6 Ibid. pp.178. 7 Choi, 2011. 8 Dafoe (2011) suggests the second and the third point here. 5  2  commercial peace, is also discussed in this section. Section 5 reports the findings from the new liberal peace models of this paper, and section 6 provides some discussion for future studies. The results of these new statistical models of liberal peace theorie shows that while traditional democratic peace and commercial peace remain robust in the new era of global financial integration, the pacifying effect of commercial ties is weakened by the volatile fluctuations in the international financial market.  3  2. Democratic Peace Vs. Commercial Peace -  2.1 Democratic Peace  Democratic peace and commercial peace are often considered as the two pillars of the liberal peace models9. The history of democratic peace can be traced back to Kant, as ―the Perpetual Peace‖ has often been cited as the founding piece of the liberal peace idea10. However, ―The Perpetual Peace‖ speaks of the pacifying effects of republican institutions rather than democratic institutions and norms, and this conceptual difference requires some attention in current studies on democratic peace theory11. Democratic peace was ―re-discovered‖ in the 1970s and become an important topic in the studies of International relations12. Political leaders openly claim that democracy should be promoted because the expansion of democracy will make the world more peaceful 13 , but some scholars remain skeptical on two aspects of the democratic peace argument—the causal mechanisms and empirical models. While the empirical evidence supports the basic correlation between democracy and peace, many scholars have argued that the causal mechanisms of democratic peace are still open to debate 14 . Scholars of democratic peace theory emphasize two causal mechanisms—the  9  Scholars have proposed other liberal causal mechanisms of peace. O’Neal et. al. suggest memberships in international governmental organizations lead to peace. Doyle mentions a belief of universal human rights as a pillar of liberal peace. In this paper, I restrain my discussion with the two major components: democratic peace and commercial peace. (Doyle, 2005, O’Neal et. al. 2003). 10 Kant, 1795/96. 11 Kant argues that republicanism is the opposite of despotism, while democracy ―is necessarily a despotism, because it establishes an executive power through which all the citizens may make decisions about (and indeed against) the single individual without his consent‖. Pp 100-101. 12 Schneider and Gleditsch, 2010. 13 For example, President George W. Bush says his administration will promote democracy because ―democracies don’t go to war with each other‖. Mousseau 2009, pp. 54. Cameron, 2005, pp. 190. 14 For examples, Hayes 2010, Mousseau 2009, Gartzke 2007, Rosato 2005, 2003. 4  externalization of peaceful norms and the effects of democratic institutions—to explain why domestic democratic structure can produce a peaceful effect at the interstate level 15 . The externalization argument expects liberal norms, such as trust and mutual respect, will develop at the domestic level of democracies and extend to the interstate level. Rosato challenges this argument by highlighting colonial and imperial wars16. Military interventions in the cold war also suggest that democracies don’t always treat each other with trust and respect17. Gartzke challenges the norm externalization argument with two points: 1.) These pacifying norms of democracy have often become both the phenomena and the causal factor of explanations; 2.) Mutual trust and respect seem more pronounced between developed democracies, while pacifying norms are less salient between poor democracies18. Defenders of the externalization argument argue that the pacifying effect of norms will only work in a reciprocal manner19, thus the externalization effect should be observed more, if not only, between democratic dyads. Doyle argues that democratic norms by themselves are not sufficient to make peace, and they work together with the other two factors— international trade, and respect of basic human rights—to 15  Rosato 2003, pp. 586-7. Ibid. pp. 588. Critics to Rosato argue that Britain and the U.S. in the 19th century are not qualified as modern democracies, and modern democracies had not matured till the end of the WWII. 17 Rosato lists cases such as the military intervention in Chile. A powerful critic from Rosato is that, even one can argue the previous Chile government was not fully democratic, but the government established after U.S. intervention was even less democratic than the former ones. The Cold War anti-communism sentiment is an overwhelming factor in this case, and the argument of shared democratic norms cannot explain why a democracy would intervene to establish an even less democratic regime. Rosato 2003, pp. 588. 18 Gartzke, 2003, pp. 168. In the following paragraphs, Gartzke argues that the pacifying effects are all due to economic factors, and democracy is insignificant in his liberal peace model. 19 Kinsella argues that Rosato ignores the dyadic nature of democratic peace. The peace-norms will only work in a reciprocal manner. Since autocracies do not have the peace-norms, the externalization effects do not work in a democracy-autocracy dyad. But Kinsella’s critic cannot explain the colonial wars and imperial wars mentioned by Rosato, since the more democratized European states were the initiator of conflicts. More discussions on the dyadic nature of democracy are presented in the following sections of this paper. (Kinsella, 2005. Doyle, 2005.) 16  5  create sufficient conditions of interstate peace20. This debate seems to suggest conditional effects of democratic norms. At the monadic level, norm externalization has some difficulties with regards to explaining why democracies are reluctant to extend the norms to their relations with non-democracies. Institutional approaches suggest several mechanisms to explain democratic peace. First, leaders are more accountable to citizens in a democracy. Therefore, political leaders are less likely to wage wars as they need to face the cost of losing a war21. Similar to the accountability argument, the public constraint theory argues that democracy gives the peace-loving public tools to keep the government from aggression 22. Peace is a public good which benefits the whole society, while the gains from war often bring private good to a small winning coalition. Since political leaders in a democracy need to provide public goods to large winning coalitions, they will prefer peace rather than war23. Furthermore, the political opposition in democratic states is more active, and this aspect of democracies leads to two effects: Democracies are less efficient at mobilization, thus slowing the process of crisis escalation, and policy transparency is higher in democracies, thus reducing the chance of ―war by mistakes‖24. Critics reject these institutional arguments. Finel and Lord argue that higher policy transparency does not lead to peaceful interstate relations but can lead to confusion and noise25. Alternatively, other states may behave more aggressively, as they can gain strategic advantage  20  Doyle 2005. Rosato, 2003. 22 Gartzke 2007. Rosato, 2003. 23 Bueneo de Mesquita et. al. 2003 24 Lektzian and Souva, 2009. Gartzke 2007, Rosato 2003.Lekzian and Souva find that information transparency will facilitate peace and prevent the outbreak of war, but will have no pacifying effects in terms of preventing the escalation of the continuation of conflicts. 25 Finel and Lord, 1999. 21  6  by exploiting the policy transparency of the democracy. Rosato raises the point of nationalism and policy circumvention made by U.S. presidents to argue that democracies are not less efficient at mobilizing citizens 26 . He also argues that there is no evidence showing that democratic leaders are more likely to be removed from their office—in other words, be accountable—after losing a war 27 . In conclusion, realists suggest that the democratic peace observed after the WWII is more likely due to other factors, such as economic development28 or a hegemonic peace around the United States29. On the empirical side, while scholars generally agree that the dyadic democratic peace is well supported 30 , the empirical models are still in the process of evolution. Dyadic level of analysis has become the norm in the studies of liberal/democratic peace31, thus most empirical models of democratic peace incorporate two variables—the higher value of democracy and the lower value of democracy of the countries in the dyad32—to operationalize the test of democratic peace33. Challengers of this conventional measurement argue that the democracy-high variable is  26  Rosato, 2003. Ibid. 28 Gartzke finds that the pacifying effect of development is particularly strong for contiguous dyads. Gartzke 2007, pp. 172. 29 Rosato, 2003. 30 Some scholars disagree with this statement and argue the democratic peace is either conditioned on other variables or confounded by other factors, such as economic development. Russett 2010, McDonald 2009, Gartzke 2007. 31 An incomplete list of major dyadic studies of democratic peace includes: McDonald 2009, Gartzke 2007, Souva and Prins 2006, Barbieri 2002, O’Neal and Russett 1999. There are studies focusing on monadic level as well, but the dyadic studies dominate the literatures. 32 The POLITY scores or the Freedom house index has been often used as the quantitative measurement of democracy. And most studies found these different measurements produces identical results. In this paper, the index of democracy is measured by POLITY IV scores. The most updated POLITY IV scores can be found on the website: http://www.systemicpeace.org/polity/polity4.htm 33 Examples of studies using this measurement of dyadic democratic peace includes: Gartzke 2010, 2007, 2003, O’Neal and Russett 1999, Polachek 2005. 27  7  often hard to interpret 34 . The democracy-high variable by itself does not provide a direct measurement of democratic peace theory, because the theory only predicts highly democratized states will not fight each other, but highly democratized states can be more likely to fight with autocracies for their ideological discrepancies35. To make the democracy-high variable easier to interpret, Choi suggested a linear transformation of democracy-high variable, which he calls ―political distance‖ 36 . This variable is measured as the democracy-high value minus the democracy-low values, indicating the ideological distance within the dyad. And the expectation is that bigger democracy distance leads to higher chance of conflicts37. Replacing the democracyhigh variable with democracy distance does not change the models much but makes the models easier to interpret. However, one possible step forward from importing this democracy distance variable leads to the argument that the impact of democracy distance is conditioned by the democratic-low values of the dyad. Suppose that there are two authoritarian states which both score -10 in the POLITY IV index. Democratic peace theory would suggest their chance of fighting is higher than a democratic dyad, in which both states score 10 in their POLITY IV index. In other words, conditioned on the democratic-low value of the dyad, the same democracy distance can produce different consequences. This argument suggests an interaction effect between the democracy 34  Choi 2011. Doyle 2005. In democratization literatures, there are debates on whether democracy can be imposed by military interventions. If democratization can be imposed by military means, one can expect higher chances of conflicts between democracies and autocracies. For the debates on democratization intervention, please see Peceny 1999 and Beetham 2009. 36 Choi 2011. In this paper, I will use the term ―democracy distance‖ to name this variable, as the term ―political distance‖ can catch many aspects of political differences other than democratic ideology and norms. The political structure of countries which score the same Polity index can be quite different—for example, the parliamentary democracy of Britain and the presidential democracy of the U.S., but the political distance of these two states is 0. To maintain the parsimony of the concept, I prefer to use the name ―democracy distance‖. 37 Ibid. 35  8  distance and the democracy-low variable. Since the democracy distance is a linear transformation of the democracy-high variable, one can expect that the conventional models of democratic peace also can be improved by introducing the interaction between democracy-low and democracy-high variable. This leads to Hypothesis 1 and Hypothesis 2 of this paper: H1: If country dyads include at least one authoritarian state, greater democracy distance leads to higher risk of conflict in the dyad. H2: If both states in the dyad are highly democratic, greater democracy distance does not lead to higher risk of conflict. Many scholars have argued that democratic peace essentially works at the dyadic level38. Therefore, researchers need to be cautious about making monadic interpretations through dyadic data and models39. To emphasize this dyadic nature of democratic peace, this paper suggests an indicator to directly measure the nature of the dyad. In other words, dyads are categorized into three categories: democratic dyad, autocratic dyad, and discrepant dyad40. Democracy distance should have conditional impact on the probability of conflicts in different dyads. This leads to Hypothesis 3 and Hypothesis 4 of this paper: H3: Democratic dyads are less likely to fight each other.  38  A number of studies have emphasized this dyadic nature of democracy. An incomplete list of these studies includes: Hayes 2011, Mousseau 2010, Gartzke 2007, Rosato 2005 2003, Kinsella 2003. 39 The question of whether we can extend the dyadic evidence to the monadic level of democratic peace mechanisms is at the center of debate between Rosato and Kinsella. Please see Rosato 2005, 2003 and Kinsella 2003. 40 Discrepant dyad is defined as consisted by one democratic state and one autocratic state. The operationalization of these concepts is explained in section 3 of this paper. 9  H4: In autocracy dyads and discrepant dyads, bigger democracy difference makes states more likely to fight each other. -  2.2 Commercial Peace  The liberal peace argument rests on not only the democratic peace theory but also the commercial peace theory. The economic tradition of liberal peace also can be traced back to Kant, Adam Smith, and Norman Angell41. Unlike the ―re-discovery‖ of democratic peace, the debate on commercial peace, based on international trade, has more constantly been a focus of academic studies. Realists, such as E.H. Carr and Waltz Kenneth, argue that commercial ties can actually increase the chance of conflict42. Recently, quantitative studies have been widely used to test commercial peace theory43. The conceptualization of commercial peace has been re-defined and improved44. Most importantly, the new trend of globalization--financial liberalization since the middle 1980s--has produced profound impact on the studies of commercial peace. Commercial ties are no longer solely defined by commodity trade. Financial markets, and the interdependence created by financial integration, have become a topic of the studies of commercial peace45. However, these new developments in the study of commercial peace require a careful review of the causal mechanisms underlining the commercial peace theory.  41  Smith 1789, Kant 1795/96, Angell 1911. Carr 1939, Waltz 1979. 43 An incomplete list of recent quantitative studies on commercial peace includes: Gartzke and Hewitt 2010, McDonald 2010, 1009, Gartzke 2007, Polachek et. al, 2005. Barbieri 2002, O’Neal and Russett 1999, 1997. 44 For example, Barbieri argues peace is promoted by trade dependence and the salience of trade in the GDP, not simply by the volume of international trade. McDonald argues peace is promoted through free trade, not just trade. Gartzke argues peace is promoted by the integration of financial markets. Mousseau clarifies the term capitalist peace by emphasizing the importance of free markets. Mousseau 2010, Gartzke 2007, McDonald 2004, Barbieri 2002. 45 Dafoe 2011, Gartzke and Hewitt 2010, Gartzke 2007. 42  10  Generally, the causal mechanisms supporting the commercial peace theory can be categorized into three major arguments: the opportunity cost argument, the strengthening domestic ―peace-loving‖ sectors, and the socialization theory. The opportunity cost theory argues that commercial development and integration increase the cost of war and decrease the cost of trade as trade provides easy tools for states to gain what they want; therefore, wars become less preferred as the opportunity cost of conflicts increases46. In ―the Great Illusion‖, Norman Angell argues that: [A]s the only possible policy in our day for a conqueror to pursue is to leave the wealth of a territory in the complete possession of the individuals inhabiting that territory, it is a logical fallacy and an optical illusion in Europe to regard a nation as increasing its wealth when it increases its territory, because when a province or state is annexed, the population, who are the real and only owners of the wealth therein, are also annexed and the conqueror gets nothing.47 Thus the cost of conquering is high, and the gains are small. And if states are rational, conquering shall not be an optimal preference. However, decisions are not always made with rational choices. Treating the population as the only owners of the wealth might also be an oversimplification of the reality. Despite these weaknesses in his argument, ―The Great Illusion‖ occupies an important position in the studies of commercial peace. There is some evidence for the opportunity cost argument, such as Rowe’s finding that increasing international trade in late 19th century England increased the competition in the labor market, and thus made it more difficult for the military forces to recruit personals48.  46  Gartzke 2007, Barbieri 2002, O’Neal and Russett 1999, Angell 1911. Angell, 1911, pp. 31. 48 Rowe 2002. 47  11  The opportunity cost of war can increase as the interdependence between states—facilitated by economic interactions—increases 49 . However, interdependence can also increase the vulnerability of being the target of attack50. In order to secure its resource supply, a powerful and highly dependent state might act aggressively to a weak state which produces the raw material. By emphasizing the opportunity cost argument, Gartzke introduces financial market integration as a causal factor of peace into the commercial peace model 51. He argues that, under financial liberalization, the cost of war has been further increased as the damage to the financial markets easily spread to other nations and cause significant cost to many nations52. On the other hand, global financial markets provide new tools for nations to gain profit, further reducing the cost of trade 53 . An argument associated with this idea is the peace caused by economic development, which can be facilitated by increasing foreign capital investments 54 . However, economists have argued that financial liberalization does not always facilitate development. Broner and Ventura find that liberalization often leads to short-run development in a capital-poor state, but once the state becomes less capital-poor, liberalization can lead to a capital out-flow, as newly-accumulated domestic capital evade risks by escaping to developed countries with better financial institutions, thus hindering further development55. The second causal mechanism of commercial peace theory works at the domestic level. International commercial ties can empower ―peace-loving‖ sectors, and these peace-loving 49  Mousseau, 2010, Gartzke 2007. Polachek et.al. 2005. Barbieri, 2002. 51 Gartzke 2007. 52 Ibid. pp. 173. 53 Ibid. 54 Ibid. pp. 171. Whether financial liberalization leads to development is still under debate. Generally, economists agree that financial liberalization facilitate development under certain conditions. More discussions on this issue is presented in section 6 of this paper. 55 Broner and Ventura, 2010. 50  12  sectors can lobby the state to adopt peaceful policies 56 . Domestic sectors related to tradable goods often benefit from international trade; in order to maintain the benefits of trade, the sectors of tradable goods will lobby the state to maintain peaceful relations with other countries. However, this peace-lobby factor seems to be more effective in democracies than autocracies. In other words, the ―peace-loving sector‖ variable interacts with other domestic institutional variables. On the other hand, since the cost of international trade is often concentrated within some small groups, these protectionist groups are better organized and more efficient at lobbying the state. Therefore, it might be over optimistic to predict that the ―peace-loving‖ group will easily prevail in the policy making. Extending this domestic factor argument to the financial market, one can argue that the nature of financial integration can possibly mitigate this pacifying causal mechanism. The cost caused by commodity trade is often not limited within a small group, but a financial crisis often creates severe income shocks for the whole society. As the history of the great depression shows, economic crises can leads to protectionism and xenophobia 57 . On the other hand, whether financial liberalization can empower a peace-loving sector is also conditioned by domestic institutions. While there is positive evidence showing that increasing foreign direct investment(FDI) reduces the chance of conflicts58, case studies on the impact of the increasing FDI in Rwanda and Sierra Leone show that increasing FDI inflows unevenly benefits local  56  For example, Please see Barbieri 2002. For example, please see Carr E.H. 1939. 58 Polahek et. al. 2005 57  13  warlords more than citizens, enhancing their fighting capacities and increasing the chance of conflicts in these states59. If the peace-loving groups can have significant impact on national policy, one can expect states which are more integrated into the international market to develop similar policy interests, such as more market liberalization. Therefore, this positive feed-back process makes states with similar interests less likely to fight each other60. However, states can adopt the same market liberalization policy for different reasons. Doyle rightly points out that market integration can lead to more conflicts if states wish to strengthen their material strength through trade, and economic integration will produce peace if states just want to exchange for their own well being61. The third causal mechanism is socialization theory. Market integration provides more forums allowing national policy makers to meet, thus policy transparency can be increased, and misinterpretations, which can lead to more conflicts, are reduced62. Similar policy interests can also be developed through the socialization processes 63 . However, Waltz argues that as the number of contracts increases through international market integration, the number of contract default will also increase64; therefore, socialization under globalization can work in a negative way and make conflicts more likely. On the other hand, the socialization effect caused by financial market integration can be limited, as the highly professionalized nature of financial  59  Williams, 2007. He finds that warlords in war-torn states often use their power to manipulate the FDI flows into the state and benefit themselves. Consequently, foreign FDIs empower the war capacities of local warlords. 60 Gartzke 2007, Signorino and Ritter 1999. 61 Doyle, 2005. 62 Gartzke 2007. 63 This is the hypothesis 3 in Gartzke’s Capitalist Peace argument. Gartzke, 2007. 64 Waltz 1979, Barbieri 2002. 14  markets creates interactions only within a small group of experts. Chewieroth suggests that state leaders often have little to say in the norm building of international financial structure, and the self-interested bureaucrats of IMF and other international financial organizations have a significant impact on financial liberalization65. These discussions suggest there are some reasons to argue that the pacifying effects of financial liberalization are not as strong as commodity international trade. The causal mechanisms of conventional commercial peace may not function well with the financial integration. Furthermore, the possible negative impact of liberalization needs to be considered, as liberalization does not always bring stability. Financial market fluctuations, often marked by significant amount of capital inflows and outflows, can destabilize economy and cause further crisis. While the negative impact of large capital outflows, often known as the capital flight, are well recognized, this paper suggests that large capital inflows can be risky too. One can observe a large foreign capital inflow in cases of speculative accumulation, which often leads to financial crises when market confidence starts to collapse. The Asian Crisis in 1997 is an example of this type of crisis66. A large capital inflow also can be observed if the state is consistently borrowing from international financial markets to fix its budget deficits, such as the case in the 2011 Euro crisis. In both cases, large capital net inflows destabilize the economy and causes economic crisis. In the processes discussed above, higher level of financial deregulation provides the tool for states to borrow more from foreign capitals market, and it also encourages foreign capitals to  65  Chewieroth 2008. For example, Cooper argues that the crisis in Thailand was caused by ―excessive foreign currency borrowing by banks, which sought to arbitrage the low interests available on short-term foreign currency loans, using the funds to extend higher interests loans in baht to domestic residences.‖ In this case, Liberalization provides the channel for state to arbitrage the benefit, and the risk of crisis has accumulated through the foreign currency borrowing—a form of capital inflows. Harwood et.al. 1999, pp.3.. Cooper 1999, pp.19. 66  15  take the risk of entering a foreign market, as liberalization guarantee foreign capitals can pull out at any time as they want. For these reasons, it is reasonable to observe increasing capital net inflows67 before a crisis breaks out, and a higher level of liberalization increases the vulnerability of the state. These discussions lead to the Hypothesis 5 and 6 of this thesis. H5: A higher level of financial liberalization leads to a higher chance of having militarized interstate conflicts. H6: A higher level of capital flight leads to a higher chance of having militarized interstate conflicts.  67  This paper defines the changes of capital net inflows as the difference between the capital net inflows in the current year and the net inflow of the last year. This definition is consistent with the data used to measure foreign capital net inflow. Detailed definition is discussed in section 4, and can be found in the codebook of World Bank International Development Indicator under the series code ―BN.KLT.PRVT.GD.ZS‖. 16  3. Methodology Binary Time Series Cross Sectional method (BTSCS) has become the norm of quantitative liberal peace studies since Beck, Katz and Tucker (BKT in short) found that temporal dependence needs to be controlled in peace studies68. In simple words, temporal dependence in liberal peace model means two states are more likely to have conflicts if they had a conflict before. Scholars generally agree that temporal dependence needs to be controlled, but debates focus on how to control temporal dependence in liberal peace model. Temporal dependence can be controlled by using a set of dichotomous variables, often called time dummy variables, indicating each time phase which has passed. This approach works well in small T and big N cases, but when number of time phases increases, this approach uses significant amounts of degrees of freedom, and might cause problems of semi-separation69. BKT recommends using cubed splines with knots as the smooth functions of the time phase which has passed70, and this technique has been widely used in most BTSCS studies. However, Carter and Signorino criticize the imprudent using of this approach in many BTSCS social science studies, as many studies using splines without a careful interpretation or using the default knots settings without careful theorization 71 . BKT’s default knots (1,4,7) provide an efficient and accurate prediction in case of small T and big N studies72, but they will be less accurate as the number of  68  Beck, Katz et. al. 1998 Semi-separation happens when the result can be perfectly predicted. For the risk of semiseparation caused by using too many time dummies, please see Carter and Signorino 2010a, 2010b. 70 BKT 1998. 71 Carter and Signorino compare the prediction of using splines with BKT default knots and time dummies, and found they are mostly identical in big-N-small-T studies. In the web appendix, Carter and Signorino provide a list of papers using the splines to control temporal dependence, but few of them provide interpretations to the splines. Carter and Signorino 2010a, 2010b. 72 Carter and Signorino 2010. Pp.11. 69  17  time dummies increase. The splines and knots are hard to interpret, thus Carter and Signorino suggest a simpler approach to control temporal dependence: using time squared and time cubed and assuming there is a non-linear correlation between time and the dependent variable73. While scholars are debating on whether dropping the knots is a right approach 74, this approach provides easier interpretations to link models and theories. The data used by this paper contains 14,792 pairs of dyad, and the time span covers 11 years, thus it fits with the definition of a big-N-small-T study. Details of the data are reported in the next section. According to the discussion above, either time dummies or the BKT approaches with default knots should function well in a dataset like this. With this data, I have tested the three different methods listed above to control temporal dependence; Splines with different combinations of knots are also tested, and the result shows that splines with the default knots provide a reliable and efficient prediction75. Therefore, in this paper I use splines with default BKT knots to address the issue of temporal dependence. Gartzke uses several continental dummy variables to measure the fixed effects of regional (or continental) dependence76. Dafoe criticizes that these variables are under-theorized77, and I agree with Dafoe. For example, while the Asia dummy gives a fixed effect to all Asian Dyads, one can reasonably argue that the Israel-Palestine dyad can behave quite differently than the 73  Ibid. Beck questions the approach of Carter and Signorino, suggesting splines provide a better tool to control the different pattern of temporal dependence. For this debate, please see the interaction between Beck, Carter and Signorino. Carter and Signorino 2010b, 2010c. 75 Details and the results of models using different methods of controlling temporal dependence are reported in Appendix A. Generally, time dummies and default BKT splines produce similar results, and they also have the highest pseudo-R2 for all the models. 76 Gartzke 2007. Not all Gartzke’s model used the continental dummies to control the regional fixed effects. The model 5 of the Capitalist Peace, which is replicated in this paper, does not use these continental dummies. 77 Dafoe 2011. 74  18  South Korea-Japan dyad. The basic model of this paper replicates the Model 5 of Gartzke’s Capitalist Peace article, in which regional dummies are not used78.  78  In the Model 4 of Gartzke’s Capitalist Peace paper, all regional dummies, except the Middle East regional indicator, have no significant impact on conflicts. Therefore, regional dummies were not controlled when Gartzke developed the Model 5. Gartzke 2007, pp.177. 19  4. Data and variables The monadic data of this thesis are collected from published data bases such as the Correlation of War 4.0 (COW), World Bank International Development Indicator (IDI), EUGene project, and the Chinn-Ito index of market openness79. After the monadic data was collected, I merged them into dyadic data. The original dataset and coding do files (STATA format) are available for replication. The dependent variable used in this paper is the militarized interstate disputes (MIDs). MIDs are widely used in peace and conflict studies 80 . Wars are also often used as the dependent variable in liberal peace models81, but wars are relatively rare from 1990 to 2001, which is the period examined in this paper. The estimation of the impact on rare event needs different research designs, and the common BTSCS method can fail82. Intrastate conflicts and non-state wars are not included in the measurement of MIDs. This operationalization can be criticized as too state-centric, but it also provides advantages, such as maintaining the consistency on the level of analysis and producing comparable findings to leading studies to the research design. Independent variables are categorized into several subgroups. The democratic measurements includes the lower value of democracy in the dyad, the higher value of democracy in the dyad, 79  Please see table 1 for the detail of the sources of all variables used in this paper. Table 1 also reports the descriptive statistics of all the variables. 80 Major liberal peace works cited in this paper mostly use MIDs as the dependent variable. For example, please see McDonald 2010, 2009, Gartzke 2007, Polachek et.al. 2005, O’Neal and Russett 1999. The MID data used in this paper are extracted from Correlation of War (COW) 4.0 project. The definition of MID can be found in the COW 4.0 codebook. 81 For example, Gartzke 2007. War data are available in the dataset of this paper, but this paper does not use war as the dependent variable, as there are only two major wars, the Yugoslavia War and the Gulf War, in the 1990s. The extremely low probability of dependent variable can cause the problem of semi-separation, thus making spurious results. 82 As discussed before, I have tested the models using war as the dependent variable, but due to co-linearity, BTSCS logit model failed to converge. 20  the democracy distance measured by the high-democracy value minus the low democracy value, and the indicator of the type of the dyad 83. The scores of democracy are collected from the POLITY IV project84. The economic measurements include: the level of development 85 , market openness measured by the Chinn-Ito index86 , the lower value of trade dependence in the dyad measured by total trade volume over GDP87, the lower value of private capital net inflows, measured by net inflows of private capital through financial markets as the percentage of the GDP88. Hypothesis 6  83  Using the same technique used by Gartzke, I standardize the POLITY IV score into a 21-point scale, in which 0 means the most authoritarian and 10 means the most democratic. The democracy distance variable is also coded in a 21-point scale. If both states score higher than 7 ,or equal 7, on the 21-point standardized scale, this dyad is coded as a democratic dyad (value 1 in the dyad type variable), if both states score lower than 7 on the 21-point scale, this dyad is considered as an autocratic dyad. If one state scores higher than 7 or equals 7, and if the other state scores lower than 7, this dyad is considered as a discrepant dyad. Gartzke uses value 5 as the cutting line to distinguish democracies and autocracies, while scholars (Mousseau 2010) consider this cutting line is relatively low. Some countries which are doubtfully to be called democracies will be counted as democratic, if we use 5 as the cutting point. Examples of these countries include: Russia from 1991 to 1999, Peru under Fujimoto from 1990 to 1995, Albania from 1990 to 1993, Liberia in the 1990s, Iran from 1997 to 2001. The commonly used cutting line is 7. The results reported in Table 3 and Figure 2 of the finding section use 7 as the cutting line, and I also tested the results under the condition that 5 is used as the cutting line of democracy. The results are reported in Appendix C, and they are mostly comparable with the results in Figure 2, except that the confidence intervals overlap more in the appendix C. 84 The codebook, definition of concepts, operationalization and the most updates POLITY IV data can be found on the POLITY IV website: http://www.systemicpeace.org/polity/polity4.htm 85 The level of development is measured by logged GDP per capita, Data are collected from Correlations of War 4.0. Original data can be found on COW project website: http://www.correlatesofwar.org/ 86 The Chinn-Ito index has been widely used as the measurement of financial market openness. Gartzke used Chinn-Ito index in the appendix of the Capitalist Peace paper and found that the result of using Chinn-Ito index is comparable to the IMF measurement of Market openness. The original data of Chinn-Ito index can be found on their website: http://web.pdx.edu/~ito/ChinnIto_website.htm 87 Trade volume and trade dependence are collected from COW 4.0 data. 88 The net inflow of foreign capital is collected from World Bank International Development Indicator. Net inflow of foreign capital includes both FDIs and the volume of portfolio good. For details of definition, please see the codebook of the World Bank IDI. Data and codebook can be found on the World Bank Website: http://data.worldbank.org/indicator 21  of this paper suggests that the bigger changes in the private capital flows leads to more conflicts, but the private capital flows variable from the World Bank IDI only measures the current level of capital flows. Therefore, in order to catch the changes in the level of capital flows, I use the differences between the current capital net inflows and the one-year lagged capital net inflows in the models to capture the impact of changing capital flows. All economic variables are lagged for one year. The control variables include a list of realist factors which are commonly used in statistical analyses of the liberal peace89. They are: a national material power index90, major power status91, military alliance, continuity92, and national policy interest measured by the weighted global Sscore provided by Ritter and Signorino93. Descriptive statistics for all variables are reported in Table 1. The dataset as a whole contains 185,652 observations, although coverage for some variables is incomplete. The data of this paper focus on the years from 1990 to 2001. Focusing on this era gives two advantages to my study of liberal peace model: 1.) It starts from a major event in international relations—the end of the cold war, decreasing the problem of left-censoring94. 2.) The data  89  Examples of these studies are listed in footnotes no. 79. The index of national material power is collected from COW 4.0. This index mostly measures national power by economic indicators such as the amount of iron and energy production. Military power in not directly included in the index. For detail, please see the COW 4.0 dataset. 91 Collected from COW 4.0. Major powers in this era are: The U.S., Russia, Britain, France, Germany, Japan, China. 92 Collected from EUGene project. The detail of EUGene project can be found at http://www.eugenesoftware.org/ 93 The S-score is collected from EUGene project. For the original paper, please see Signorino and Ritter 1999. 94 The truncated data problem needs to be considered in time-series studies. Starting from a major event can help to partially address the left-censoring problem. Box-Steffensmeier and Jones 2004. 90  22  Table 1: Descriptive Statistics  Variable Name  Range  Mean  Standard Deviation  # of Observatio n  Source  Militarized Interstate Disputes  Dummy  0.0036  0.059  185,632  COW 4.0  Democracy_low  1-10  2.78  3.28  185,646  POLITY IV  Democracy_high  1-10  7.42  2.99  177,273  ..  Democracy distance  1-10  4.57  3.32  177,273  ..  Trade dependence low  -9.87 to 16.13  0.26  0.53  185,652  COW 4.0  Ito-Chinn Index of market openness  -1.84 to 2.48  -0.58  1.22  181,330  Ito-Chinn Index  The log of GDPPC, lagged  4.16 to 11.46  6.75  1.27  184,838  COW 4.0  measured in the log of current dollar  Private Capital net inflow in % of GDP  -54.36 to 43.59  -0.0996  3.2  114,632  World Bank IDI  Private Capital Net inflow missing  Dummy  0.38  0.49  185,562  for definition, please see the notes. Higher value means more net inflow The indicator of missing values in the private capital flow variable  National Material Power index  0.000001 to 0.14  0.001  0.0027  185,562  COW 4.0  National policy interests  -0.34 to 1  0.71  0.27  185,562  EUGene, S-score  Alliance  Dummy  0.05  0.07  185,562  COW 4.0  Major Power status  Dummy  0.07  0.26  185,562  COW 4.0  For definition of major powers, please see the notes  Contiguity  Dummy  0.02  0.13  185,562  EUGene  1 means two states are contiguous to each other  Notes Dummy, 1 means an MID happened 1 means the most authoritarian, 10 means the most democratic .. Democracy_high minus Democracy_low higher value means being more dependent on trade Higher value means being more open to foreign capital  National Material Power as the percentage of the whole world Global Weighted S-Score, higher value means similar policy interests 1 means in a formal alliance  availability, particularly the availability of economic data, in this period is much better than in the earlier era. A disadvantage of focusing on this period is that the findings from this paper might be specific to this period, and readers are reminded that the findings from this paper shall 23  not be generalized to other periods. The data ends in year 2001 as the most updated version of the dependent variable—militarized interstate disputes—ends in the year 200195. The missing value problem needs serious attention for the students who study liberal peace models. Dafoe finds that missing values in Gartzke’s models are systematically associated with its major explanatory variable—market openness 96 , thus leads to a biased conclusion. For example, China, the U.S.S.R, and North Korea were involved in several militarized interstate conflicts, but a significant part of the market openness is missing for these countries 97 , and excluding these cases from the model leads to a bias. While Dafoe assigns value 1 (least open to financial market) to all the missing values of China, U.S.S.R and North Korea, he finds that market openness lost its significance and democracy become significant again 98. But Dafoe’s approach can be problematic as well, because these nations may be open to each other while staying closed to the west or the global financial markets. In case of North Korea, foreign capital from the U .S.S.R and China are pivotal to the survival of the regime. This discussion suggests care in checking whether missing values in the independent variables are systematically associated with the dependent variable is necessary, and care in handling missing values, as there is no perfect solution to the missing value problem. Even with 95  The most updated COW dataset contains MIDs from 1815 to 2001. There are other data projects contain more updated conflicts information, such as the UCDP/PRIO project of the Uppsala University (updated to 2010). However, the conceptualization of conflicts in the UCDP/PRIO is different from the COW project. Intrastate conflicts and non-state are included in the UCDP/PRIO. While the different conceptualization brings interesting research prospects, using it in this study might also lead to results that are not comparable with the leading studies. Therefore, this paper stays with the traditional measurement of MIDs, even the data availability of COW 4.0 is limited and no data are available after 2001. 96 Dafoe, 2011. 97 Dafoe 2011, pp 252. The list of these missing non-democratic nations includes: China, Vietnam, North Korea, Angola and Soviet Union. For these nations, all data are missing before 1965. The data for China are missing till 1980. 98 Dafoe 2011, pp. 253. 24  greater data availability in recent years, missing economic data cam still damage a research project as they are systematically related to conflicts, the dependent variable. It is likely that there is no efficient governance in a war-torn country, thus no functioning public or private agencies exist in these countries to collect economic data. Therefore, it is likely to observe that missing values of economic data is systematically associated with conflicts—the dependent variable. H5 and H6 of this paper suggest that economic variables such as the foreign capital flows can influence the probability of conflicts, so careful examination of whether there is a systematic missing problem in the foreign capital flow variable is necessary. In total, the private capital flow variable contains 114,632 observations, consisting of 61.7% percent of the full data. If the nearly 40% missing values of the data are random, the results are simply less efficient but still reliable; however, the results can lose the reliability if these missing values are systematically associated with MIDs. Table 2 shows that in total, there are 661 cases of MID-years in this period. 622 cases are used in the basic model of this paper, and 39 cases of MID-years are excluded due to the missing values in the independent variables. However, when the foreign capital flow variable is added to the basic model, 421 further cases of MID-years are Table 2. Missing values caused by the capital flow variable MIDs in the model  MIDs excluded due to missing IVs  Full Model  661  .  Basic Model  622  39  Basic Model + Capital Flow  240  421  25  dropped from the model, and only 240 cases are left. This significant drop of the dependent variable can clearly damage the validity of the results. To acquire some comprehensive view of these MIDs dropped due to missing values in the foreign capital flow variable, I listed out these 421 cases which are excluded, and find out they include 148 MID-years involving Yugoslavia, 84 MID-years involving Iraq, and 27 MIDs-years involving Pakistan. Yugoslavia and Iraq were two major hotspots of conflicts during the 1990s, and excluding these two nations from the model is likely to invalidate any findings from the models. Therefore, the missing value problem of the capital flow variable needs to be addressed before the Hypotheses 5 and 6 are tested. The simplest way to handle this missing value problem is to assign some values to the missing cases, but the values assigned need to be carefully theorized. One approach is to assign value 0 to the missing cases. Zero in this case indicates that the capital inflows and outflows are equal. However, one can guess that sensitive private capitals are more likely to leave the state if the capital owners feel a military conflict is about to happen, therefore, assigning value 0 might not be a proper approach. Another option is to assign the expected value –the mean value of the foreign capital flow variable—to the missing cases. The average value of the foreign capital flow variable is -0.09, and the negative sign indicates a net capital outflow. Therefore, it fits with the expectation that a capital outflow can precede the outbreak of a military conflict. One can make another assumption that large capital net inflows can destabilize the economy and cause economic and political crisis, therefore, positive capital net inflows can precede conflicts. In this sense, assigning the negative mean value to missing cases is questionable, and a positive value,  26  such as the 75 percentile of the capital flow variable (1.465), is a proper guess for the missing cases. Instead of assigning one value to all missing cases, these approaches can be further improved by assigning a set of normally distributed values to the missing cases. I try to assign a normal distribution centered at the mean or the 75 percentile of the non-missing values, and I assume the standard deviation of the assigned values equals the standard deviation of the nonmissing cases. In the findings section, this paper reports the results based on the assumption that the missing cases in the capital flow variable have the same mean and standard deviation as the non-missing cases. Other approaches listed above are tested, and most of the results are comparable to the results presented in Table 3 in the next section99.  99  The results of these tests are reported in the Appendix B. Six different approaches are tested. They are: 1,) assigning value 0 to missing cases, 2,) assigning the mean to missing cases, 3,) assigning 25 percentile to missing cases, 4,) assigning 75 percentile to missing cases, 5,) assigning a normal distribution (the mean and the standard deviation equals the mean and the standard deviation of non-missing cases), 6,) assigning a normal distribution ( the means equals the 75 percentile of the non-missing cases, and the standard deviation equals the standard deviation of the non-missing cases). 27  5. Findings The coefficients of all models are reported in Table 3. The coefficients of time splines are not reported in this section. Please refer to the appendix A for the coefficients of time variables. Model 1 replicates Model 5 of Gartzke’s capitalist peace paper100. A major difference between the findings of Model 1 and Gartzke’s capitalist peace Model 5 is that, Model 1 of this paper shows that higher level of financial market openness is positively associated with more conflict, while Gartzke finds his market openness index is negatively associated with more conflict101. As Dafoe points out, Gartzke’s finding can be damaged by the missing values in his market openness variable, and the temporal dependence and cross-sectional dependence are not properly controlled 102. Model 1 pays close attention to these problems, and finds that, at least in this period, market openness is positively associated with more conflicts. As the data of this paper focuses on a different time period, this result does not suggest Gartzke is wrong, but further explanation of why market openness is positively associated with more conflict is necessary. The low value of democracy is negatively associated with conflicts, and this finding is consistent with the argument of democratic peace theory. The positive impact of the high value of democracy possibly shows that a discrepant dyad—when the democracy low value is controlled—is more likely to fight each other. As Choi points out, the interpretation of the  100  Gartzke 2007. A small difference between Model 1 and Gartzke Model 5 is that, Gartzke coded the openness variable by himself and didn’t use the Chinn-Ito Index. In the appendix of the Capitalist Peace article, Gartzke tried to test the same models with Chinn-Ito index and found that the results are identical with his own openness variable. Moreover, both Gartzke’s measurement of openness and the Chinn-Ito index are based on the IMF Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). So in this paper, I consider Chinn-Ito index as a proper substitute of Gartzke’s openness measurement. 102 Dafoe 2011, Choi 2011. 101  28  Table 3. Models and Coefficients Militarized Interstate Conflicts  Model1  Model2  Model 3.1  Model 3.2  Model 4  Independent Variables  Gartzke's Model 5  Choi's Democracy Model  Democracy low interacts with Democracy distance  Dyad type interacts with Democracy Distance  The new capitalist peace model  -0.036 (0.029)  -0.144*** (0.041)  0.058** (0.0274)  -0.002 (0.029) 0.061*** (0.012)  Democracy Low Democracy High  -0.094*** (0.022) 0.058** (0.027)  Democracy Distance Demolow * Democracy Distance  -0.088** (0.042)  0.063 0.115  0.770** (0.329) 1.183** (0.390)  Dyad type 2 (Autocracy) dyad type 3 (Discrepant dyad) The level of net capital inflow Δ of Capital net inflow The missing of capital net inflow Trade dependence Financial Market openness GDP PC low Contiguity GDPPC * Contiguity Major Power status Formal alliance National material power Policy interest (S-Score) Constant Numbers of observation R2  0.019 (0.032) 0.050*** (0.012)  -0.094*** (0.026) 0.112** (0.048) 0.531*** (0.071) 9.731*** (0.824) -0.779*** (0.111) 0.797** (0.258) 0.065 (0.400) 30.130** (9.557) -0.774** (0.260) -8.515*** (0.569) 173,211 0.3453  -0.094*** (0.026) 0.112** (0.048) 0.531*** (0.071) 9.731*** (0.824) -0.779*** (0.111) 0.797** (0.258) 0.065 (0.399) 30.130** (9.557) -0.774** (0.260) -8.515*** (0.569) 173,211 0.3453  -0.129*** (0.028) 0.108** (0.047) 0.580*** (0.071) 9.610*** (0.829) -0.763*** (0.113) 0.770** (0.260) 0.162 (0.430) 27.762** (10.632) -0.648** (0.269) -8.805*** (0.567) 173,211 0.3545  -0.103*** (0.027) 0.115** (0.048) 0.568*** (0.075) 9.575*** (0.836) -0.761*** (0.113) 0.760** (0.262) 0.079 (0.433) 29.544** (10.101) -0.828** (0.264) -9.391*** (0.666) 173,211 0.3490  0.113** (0.036) 0.075** (0.028) 0.700*** (0.145) -0.268** (0.103) 0.076 (0.051) 0.483*** (0.072) 9.402*** (0.852) -0.721*** (0.117) 0.821** (0.267) 0.170 (0.461) 25.535** (12.378) -0.891** (0.276) -8.636*** (0.581) 155,872 0.3621  *** means significant at 99% level, ** means significant at 95% level  29  democracy high variable is often difficult, but it seems the democratic peace theory is well supported by this data. The traditional commercial peace theory, which focuses on the trade dependency created by international commodity trade, is also supported by this model. Development makes noncontiguous states more likely to fight each other, as the development facilitated the capacity of states to project power to a longer distance, but development also makes contiguous states less likely to fight each other103. This finding supports that the interaction effect between contiguity and development is also robust in this period. Being a major power makes the state more likely to be involved in conflicts. Similar to this finding, a state is more likely to be involved in MIDs if its national power index is higher. However, formal alliances have no significant impact on the probability of MIDs. Model 2 replaces the high value of democracy with the democracy distance variable 104 . Since the democracy distance variable is a linear transformation of the high value of democracy105, this replacement produces identical results to Model 1, but the interpretation of democratic peace in this model is much easier. The positive and significant impact of the democracy distance variable supports the expectation from Choi: politically different countries— the authoritarian states and the democratic states—are more likely to fight each other 106 . Different political ideology can be the underlining reason for tension.  103  Gartzke 2007, pp. 172. Choi, 2011. 105 This variable is calculated as the higher value of democracy minus the lower value of democracy in the dyad. 106 Choi 2011, pp. 763-4. 104  30  As this paper suggests before, since many pacifying mechanisms available for democracies do not exist in autocratic and discrepant dyads, the same democracy distance should have different impact in different types of dyad. Model 3.1 applies this proposal and makes the lower value of democracy interact with the democracy distance variable. The findings are impressive: The negative coefficient of the lower value of democracy becomes significant again; the coefficient of democracy distance loses its significance, but the interaction effects between these two variables are positively significant. This finding supports the democratic peace argument: countries are less likely to fight if they both are highly democratic, but this pacifying effect has been mitigated if the democracy distance is getting bigger. Figure 1 presents a prediction of the probability of conflict based on Model 3.1. It shows that the probability of conflict is almost the same for autocratic and discrepant dyads, and both of them are much higher than the probability for democratic dyads. Model 3.2 replaces the low value of democracy with a three-category indicator of dyad type107 and makes the dyad type indicator interacting with the democracy distance variable. The result shows that, compared with the base category (democratic dyad), the risk of fighting is higher in the other two types of dyads. In the base category, democratic distance does not have significant impact on their chance of fighting. Figure 2 shows how the predicted probability of conflict, based on Model 3.2, changes across different dyad types. The predicted probability shows that one can confidently claim that democratic dyads are more peaceful than other types of dyad, but the upward trend, which is similar to the trend showing in the predicted chance of fighting for autocracies, shows that bigger democracy distance leads to more conflicts in these  107  There are three types of dyad: democracy, autocracy, and discrepancy. For the definition of these three types, please see footnotes no. 83. The results of the test under the alternative definition of democracy (using POLITY score 5 as the cutting line) are reported in Appendix C. 31  two types of dyads. The discrepant dyad group generally behaves similarly to the autocracy group, except that the downward trend of the curve, showing that instead of fighting for different democratic ideology, shows discrepant dyads often fight for other reasons. However, the confidence interval of the discrepant dyad group largely overlaps with the confidence interval of the autocracy group, so more data are needed to distinguish whether discrepant dyads behave differently from autocracy dyads. In conclusion, this paper argues that the democratic peace model can be improved by interacting the democracy distance variable with the other democracy measurement of the dyad. Findings from these interaction models support the dyadic claim that ―democratic countries are unlikely to fight each other‖, but they also suggest one cannot extend this claim to the monadic level. Democratic countries are not more peaceful, as the chance of conflicts is high in a discrepant dyad. Increasing ideological differences, as measured by the democracy distance variable in these models, can increase the chances of conflicts. On the commercial peace aspect, Model 1 of this paper suggests that higher market openness can lead to more conflicts. This positive correlation might be explained by the spillover effect of market fluctuation. In order to capture the impact of market fluctuation, Model 4 added a set of variables related to the measurement of foreign capital net inflows to the model. The results show that, once the capital flow factors are considered in the model, the market openness variable loses its significance, and a higher level of capital net inflow is positively associated with more interstate conflicts.  32  Figure 1: Predicted Probability of Conflicts in different types of dyads (Model 3.1)  0.005 0.0045 0.004 0.0035 0.003 0.0025 0.002 0.0015 0.001 0.0005 0  Figure 1. Predicted chance of MIDs in different types of dyads.  Aotocratic dyad (Demolow=0, Demodist=0)  Discrepant dyad (Demolow=0, Demodist=10)  Democratic dyad (Demolow=10, Demodist=0)  Predicted Probability Confidence Interval  Figure 2: Predicted Probability of Conflicts, on Condition to the Types of Dyads (Model 3.2)  0.008  Figure 2: Predicted Chance of MIDs, on condition to types of dyads  0.007 0.006 0.005 0.004 0.003 0.002 0.001 0 0  1  2  3  4  5  6  Democracy Distance  7  8  9  10  Democracies Predicted Probability Democracies Confidence Interval Autocracies Predicted Probability Autocracies Confidence Interval Discrepant dyad Predicted Probability Discrepant dyad Confidence Interval  The missing value indicator of capital net inflows is included in the model to control the damage caused by missing data in the capital net inflow variable. This missing indicator is positive and significant, suggesting that missing economic data are systematically associated  33  with militarized conflicts. The lagged capital net inflows variable, measured as the percentage of GDP, is included in the model, and higher level of capital net inflows is associated with a higher risk of conflicts. The change of capital net inflow variable, which is measured by the level of current capital net inflows minus the level of the one-year lagged capital net inflows, is also positively associated with more conflicts, meaning the risk of conflict is higher if there are more foreign capitals pouring into the country. These findings support the theory of this paper that large capital inflows can destabilize the domestic economy and cause crises, but they are also contrary to the conventional understanding that foreign capital will leave the conflicting region. However, it can be explained by the following reasons. First, capital flight is often the consequence, not the causal factor, of the collapse of market confidence. Thus, significant amount of capital leaves the state after a crisis breaks out, and in many cases, it does not precede the conflict108. Secondly, MIDs often do not last for a year, and capital net inflows can increase again after the MID is over, and this increasing after the MIDs has been captured by the annual inflow data used in this paper. Related to this point is that the minimal unit of time in this data is one year, but capital flight can take place in a few weeks, or even a few days. Thus annual data are not perfect to capture the impact of swift capital flight, and some weekly based or monthly based data are needed to test this possibility. On the other hand, capital net inflows caused by speculative accumulation and long-standing budget  108  Therefore, the one-year lagged economic variable may not capture the impact of capital flights which often happen swiftly after some political crises. Instead of using one-year lagged net inflow(t-1) as the independent variable, I try to replace it with the net inflows in the next year(t+1) to see how capital inflows response to the onset of an MID. The findings show no significant correlation. This finding does not directly support the expectation, but this may also due to the fact that the annual net inflow data which is used in this paper may not be capable to capture the impact of capital flight, which often happens in days or weeks. 34  deficits—two major factors which destabilize domestic economy—are more long-run phenomena which are better captured by the annual data used in this paper. In conclusion, model 4 calls to revisit the financial capitalist peace model. The financial market openness variable loses its significance after the capital net inflow variables are controlled, suggesting that: 1.) The pacifying effect of financial market openness is spurious; 2.) more likely, market fluctuation caused by financial crisis may lead to more conflicts. The missing value problem is addressed by assigning the expected value of capital flows to the missing cases, but the robustness of this model can be significantly increased if economic data of war-torn states are available.  35  6. Discussion and Conclusion In conclusion, this paper finds some evidence to support the argument that while both the conventional democratic peace and commercial peace theory continue to facilitate peace in the era of financial liberalization, financial market fluctuations, caused by increasing net capital inflows, have been positively associated with more conflicts. On the democratic peace side, H1 to H4 are supported by findings, which suggests that democratic peace remains robust in this period. However, the new democratic peace models suggested by this paper call more attention to the consistency in terms of the conceptualization and the unit of analysis in the studies of democratic peace. The conditioned effects from Model 3.1 and Model 3.2 suggest that democratic peace is better conceptualized as the statement that ―democratic countries don’t fight with each other‖, since this statement is parsimonious in terms of being consistent with the same level of analysis. The democratic peace theory shall not be generalized as ―democratic countries are more peaceful‖, or at least, this statement shall not be tested with the dyadic liberal peace data. Many scholars have noticed the gap between the statistical results and causal theories of democratic peace 109 , this paper suggests that the inconsistency between dyadic data and monadic causal mechanisms might be a cause of this confusion. One solution suggested by this paper is to clarify the conceptualization of democratic peace and make it consistent with the dyadic empirical studies. Another solution can be changing the level of analysis in the empirical studies and focus more on the monadic level data. Some  109  Rosato 2003, 2005. Also see Hayes, 2011. 36  monadic models of democratic peace have been proposed, but they often suggest contradictory findings110. On the commercial peace side, this paper finds some evidence to support H5 and H6. Financial liberalization is positively associated with conflicts, and Model 4 suggests that this destabilizing effect can be explained by market fluctuations caused by large capital inflows. As discussed in the finding section, these findings shall not be viewed as directly against Gartzke’s Capitalist Peace arguments, but they raise interesting questions for future research. Why financial liberalization was negatively associated with conflicts from 1950 to 1992 (the period studied by Gartzke), and why liberalization becomes positively associated with conflicts from 1990 to 2001? One can go further to challenge Gartzke by asking whether it is proper to test the correlation between liberalization and conflicts under the Bretton Woods system, when liberalization has not yet been widely accepted till later in the 1980s. Combining the findings from this paper and the critics to Gartzke from Dafoe and Choi111, I am more inclined to the argument that the risk of having economic and political crises has increased in the era of liberalization. And then, more questions can be asked: What are the institutions which constrain the negative impact of liberalization and facilitate positive impact? What are the causal links connecting economic crises with political crises? What are the conditions to control the damage of economic crisis and prevent social and political unrest? The issue of global governance, particularly governance and cooperation over financial markets, has attracted significant attentions in the past years, but no solid conclusions have been made on this issue. Scholars are still debating on what is good governance of financial markets, 110  For example, Barbieri finds some evidence contradict to the argument that democracies are more peaceful. Barbieri 2002. 111 Dafoe 2011, Choi 2011. 37  and some contribution from peace and conflicts studies are necessary, as this paper suggests that financial liberalization is not simply a economic matter—it will have some impact on issues related to peace, conflict, and security. In this paper, the links between economic crisis and political crisis are not explored in detail, and some future studies are needed to find out these links in the new era of financial liberalization. One possible link might be the uneven development caused by liberalization, and the study of Broner and Ventura is an adventure in this direction112. Another possible cause of interstate conflicts can be the uneven distribution of the benefits from financial liberalization, and this uneven distribution of cost and benefit is not simply an economic phenomena--It often interacts with other variables, suggesting a range of research perspectives for the future. Carr, in his classical studies of the Great Depression, argues that liberalization leads to an economic interest harmonization, which often leads to the suffering of the disadvantaged groups 113. Interest harmonization is associated with the elimination of the culture and life style of the disadvantaged groups, further complicating the issues related to the negative impact of market liberalization. The uneven distribution also can interact with ethnicity114, as certain ethnic groups benefit more in the liberalization process, thus the probability of ethnic conflicts can increase as well. To protect themselves from the uneven distribution of liberalization, developing states often adopt certain levels of protectionism115, which is often attacked by the promoters of liberalization.  112  For example, please see Broner and Ventura 2010. Carr 1939, pp. 49. 114 For the discussion on the interaction effects between the uneven distribution and ethnicity, please see Chua 2003. 115 Scholars have argued that most developed states have adopted protectionism at the early stage of their development. Therefore, in Chang’s term, imposing liberalization to all developing states is like ―kicking off the adder‖ of development. Gerschenkron provides a historical review of the protectionism in the early stage of developed states. Chang 2003, Gerschenkron 1962. 113  38  These different policy interests can be a source of tension. The tension around the exchange rate between Chinese Yuan and U.S. Dollar is a salient empirical case showing this tension between the developed and developing countries. In terms of modeling and methodology, the study of this paper can be improved by more nuanced discussion on issues related to the dependent variable of this paper. As mentioned in the data and variable section, the MID measurement of COW data can be criticized as too statecentric. In fact, comparative cases of how democratic states and autocratic states deal with intrastate conflicts are likely to provide some strong evidence for the causal mechanisms of democratic peace at the monadic level, giving the democratic norms and institutions, which are only available in democracies, are powerful tools to constrain the violent behavior of both the government and citizens. On the other hand, one major difficulty to peace and conflict studies is the question of ―who initiates‖—being involved in an MID does not means the democratic state aggressively initiates the conflicts116. If a democratic state is involved in an MID to defend itself, this self-defense shall not be considered as a challenge to the causal theories of democratic peace; however, to distinguish the aggressor and the defender requires some improvement in current peace and conflict data. Another issue on the dependent variable is that the onset of conflict shall not be treated as the same with the continuation/escalation of conflicts. Lektzian and Souva make an adventure to test how the causal mechanism of democratic peace—democracy norms, policy preference, and  116  Hayes, 2011. Mousseau 2010, 39  information transparency—function differently in the onset and the escalation of conflicts 117 . Then, a similar question can be raised for the commercial peace theory: Would the causal mechanisms of commercial peace, particularly the impact of Liberalization, work differently at the onset and the escalation of conflicts? Due to the small number of continuing MIDs in the dataset118, this paper does not provide a test of how democracy variables behave differently in continuing conflicts. Some more data is necessary for this test. In terms of empirical cases and policy implications, this paper suggests political leaders to be more cautious on promoting financial liberalization. Leaders need to be more aware of the possible negative impact of liberalization, while at the same time, since this paper finds that the pacifying effects of conventional commercial peace are robust in the era of financial liberalization, nations shall not abolish their promotion to international trade. Weede argues that give that the democratic peace is not working between the U.S. and China, commercial ties between these two states are particularly important in terms of avoiding hegemony wars119. The trade disputes between China and the U.S. provide interesting empirical cases for the questions raised by this paper. As mentioned previously, liberal peace based on commercial ties is particularly important for the U.S. and China. On the other hand, the U.S. dollar reserve held by China is an interesting empirical case which might challenge the liberal commercial peace theory. Commentators have noticed that the huge U.S. dollar reserve have become a double-edge sword preventing both governments from adopting aggressive policies. The causal mechanism of  117  Lektzian and Souva, 2009. To test the different impact of the onset and the continuation of an event, Markov Transition method is needed. This data provide little cases of continuing MIDs, and it does not define whether the continuing MIDs are escalating. Future research project equipped with better data may provide some answers to these questions. 119 Weede 2010. 118  40  peace in this case is more like a neo-realists peace, and the underlining reasoning of this neorealists peace is similar to the reasoning behind the Mutually Assured Destruction from the cold war era. This case might be an interesting challenge to the liberal approach--financial integration creates peace as it gives states tools to destroy each other’s financial market. Last but not the least, even though this paper finds some negative impact of liberalization, it does not argue protectionism is the solution. The question is how to maintain the gains from global liberalization while controlling its damage at the same time. Possible solutions to this question can include better financial surveillance, global governance and cooperation, and liberal institutional reform of the global financial structure. A lesson from the fail of the first wave of globalization (late 19th century to early 20th century) is that protectionism means trying to repair the damage of economic crisis at the cost of other nations, and that will only lead to more distrust and conflict in the future.  41  Bibliography Angell, N. (2010). The Great Illusion. New York, Cosimo Inc. Barbieri, K. (2002). The Liberal Illusion : Does Trade Promote Peace? Ann Arbor, University of Michigan Press. Barbieri, Katherine, Omar Keshk, and Brian Pollins. (2008). Correlates of War Project Trade Data Set Codebook, Version 2.0. October 28th, 2010, CQ Press. Retrieved on November 20th, 2011. <http://correlatesofwar.org> Beck, N., J. N. Katz, et al. (1998). "Taking Time Seriously: Time-Series-Cross-Section Analysis with a Binary Dependent Variable." American Journal of Political Science 42(4): 1260-1288. Beetham, D. (2009). "The Contradictions of Democratization by Force: The case of Iraq." Democratization 16(3): 433-454. Bennett, D. Scott, and Allan Stam. 2000. ―EUGene: A Conceptual Manual.‖ International Interactions 26:179-204. Bennett, D. Scott, and Allan Stam (2008). EUGene Version 3.204 [Computer Software]. July 17th, 2008, EUGene. Retrieved on November 30th, 2011. <http://www.eugenesoftware.org/> Box-Steffensmeier & Jones. 2004. Event History Modeling: A Guide for Social Scientists. Cambridge, UP. Broner, F. A. and J. Ventura (2010). "Rethinking the Effects of Financial Liberalization." National Bureau of Economic Research Working paper No. 16640. Bueno de Mesquita, B., A. Smith, et al. (2003). The Logic of Political Survival. Cambridge, MIT Press. Cameron, F. (2005). U.S. Foreign Policy after the Cold War: Global Hegemon or Reluctant Sheriff. London, Routledge. Carr, E. H. (1939). The Twenty Year's Crisis, 1919-1939. New York, Harper and Row. Carter, D. B. and C. Signorino (2010). ―Back to the Future: Modeling Time Dependence in Binary Data‖. Political Analysis Advance Access. Chang, H.-j. (2003). Kicking Away the Ladder: Decelopment Strategy in Historical Perspective. London, Anthem Press. Chewieroth, J. M. (2008). "Normative Change from Within: The International Monetary Found's Approach to Capital Account Liberalization." International Studies Quarterly 52: 129-158. Chinn, Menzie D. and Hiro Ito (2008). "A New Measure of Financial Openness". Journal of Comparative Policy Analysis, 10:3, pp. 309 – 322.  42  Choi, S.-W. (2011). "Re-Evaluating Capitalist and Democratic Peace Models." International Studies Quarterly 55(3): 759. Chua, A. (2003). World on Fire: How Exporting Free Market Democracy Breeds Ethnic Hatred and Global Instability. New York, Anchor Books. Cooper, R. N. (1999). The Asian Crisis: Causes and Consequences. Financial Markets and Development: The Crisis in Emerging Markets. A. Harwood, R. E. Litan and M. Pomerleano. Washington D.C., Bookings Institution Press: 17-29. Finel, B. I. and K. M. Lord (1999). "The Surprising Logic of Transparency." International Studies Quarterly 43(3): 315-339. Gartzke, E. (2007). "The Capitalist Peace." American Journal of Political Science 51(1): 166191. Gartzke, E. and J. J. Hewitt (2010). "International Crisis and the Capitalist Peace." International Interactions 36(2): 115-145. Gerschenkron, A. (1962). Econoimc Backwardness in Historical Perspective. Cambridge MA, Belknap Press. Ghosn, Faten, Glenn Palmer, and Stuart Bremer. (2004). "The MID3 Data Set, 1993–2001: Procedures, Coding Rules, and Description." Conflict Management and Peace Science 21:133-154. Harwood, A., R. E. Litan, et al. (1999). Introduction. Financial Market and Development: The Crisis in Emerging Markets. A. Harwood, R. E. Litan and M. Pomerleano. Washington D.C., Brookings Institution Press: 1-17. Hayes, J. (2011). "The Democratic Peace and the New Evolution of an Old Idea." European Journal of International Relations 17(2): 1-25. Hegre, H., J. R. Oneal, et al. (2010). "Trade Does Promote Peace: New Simultaneous Estimates of the Reciprocal Effects of Trade and Conflict." Journal of Peace Research 47(6): 11. International Monetary Fund (2010, various years). Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER), Washington, D.C.: International Monetary Fund. Kant, I. (1795/96). Toward Perpetual Peace: A Philosophical Sketch. Kinsella, D. (2005). "No Rest for the Democratic Peace." American Political Science Review 99(3): 453-457. Lektzian, D. and M. Souva (2009). "A Comparative Theory Test of Democratic Peace Arguments, 1946-2000." Journal of Peace Research 46(1): 17-37. Marshall, Monty G. and Keith Jaggers. (2002). Polity IV Project: Political Regime Characteristics and Transitions, 1800-2002. Version p4v2010. November 12, 2010. POLITY IV Project, Center of Systemic Peace. Retrieved on November 30th, 2011. < http://www.systemicpeace.org/inscr/inscr.htm> 43  McDonald, P. J. (2009). "Capitalism, Commitment, and Peace." International Interactions 36: 146-168. Mousseau, M. (2009). "The Social Market Roots of Democratic Peace." International Security 33(4): 52-86. Mousseau, M. (2010). "Coming to Terms with the Capitalist Peace." International Interactions 36(2): 185-213. Oneal, J. R. and B. Russett (1999). "The Kantian Peace: The Pacific Benefits of Democracy, Interdependence, and International Organizations, 1885-1992." World Politics 52(1): 1-37. Peceny, M. (1999). "Forcing Them to Be Free." Political Research Quarterly 52(3): 549-582. Polachek, S., Carlos, et al. (2005). Globalization and International Conflict: Can FDI Increase Peace? 10th Annual International Conference on Economics and Security. Philadelphia. Rosato, S. (2003). "The Flawed Logic of Democratic Peace Theory." American Political Science Review 97(4): 585-602. Rosato, S. (2005). "Explaining the Democratic Peace." American Political Science Review 99(3): 467-472. Rowe, D. M., D. H. Bearce, et al. (2002). "Binding Prometheus:How the 19th Century Expansion of Trade Impeded Britain’s Ability to Raise an Army." International Studies Quarterly 46(4): 551-578. Russett, B. (2009). "Democracy, War, and Expansion through Historical Lenses. ." European Journal of International Relations 15(1): 9-36. Russett, B. (2010). "Capitalism or Democracy? Not So Fast." International Interactions 36(2): 198-205. Schneider, G. and N. P. Gleditsch (2010). "The Capitalist Peace: The Origins and Prospects of a Liberal Idea." International Interactions 36(2): 107-114. Signorino, C. and J. M. Ritter (1999). "Tau-b or Not Tau-b: Measuring the Similarity of Foreign Policy Positions." International Studies Quarterly 43: 115-144. Singer, J. David, Stuart Bremer, and John Stuckey. (1972). "Capability Distribution, Uncertainty, and Major Power War, 1820-1965." in Bruce Russett (ed) Peace, War, and Numbers, Beverly Hills: Sage, 19-48. Waltz, K. N. (1979). Theory of International Politics. Reading, MA, Addison-Wesley. Weede, E. (2010). "The Capitalist Peace and the Rise of China: Establishing Global Harmony by Economic Interdependence." International Interactions 36(2): 206-213. Williams, P. (2007). "Peace operations and the international financial institutions: insights from Rwanda and Sierra Leone." International Peacekeeping 11(1): 20.  44  The World Bank (2011). World Development Indicators, December 15th, 2011. World Bank Group. Retrieved on December 30th, 2011.<http://data.worldbank.org/data-catalog/worlddevelopment-indicators>  45  Appendix Appendix A: The results of different approaches to control temporal dependence Militarized Interstate Conflicts  Treatment 1  Treatment 2  Independent Variables  BKT Default Splines  Democracy Low Democracy Distance Demolow * Democracy Distance The level of net capital inflow Changes of Capital net inflow Missings of capital net inflow Trade dependence Financial Market openness GDP PC low Contiguity GDPPC * Contiguity Major Power status Formal alliance National material power Policy interest (S-Score) Peace Year Spline1 Spline2 Spline3 Spline4  Treatment 3  Treatment 4  T and T  Time Dummies  Splines with (1, 2, 5 ) knots  -0.090** (0.042) 0.018 (0.032)  -0.128** (0.047) -0.038 (0.034)  -0.090** (0.042) 0.018 (0.032)  -0.087** (0.042) 0.020 (0.032)  Splines with (1, 3, 6, 9 ) knots -0.088*** (0.042) 0.019 (0.032)  0.050***  0.063***  0.050***  0.049***  0.050***  0.050***  (0.012) 0.112** (0.036) 0.075** (0.029) 0.696*** (0.145) -0.265** (0.103) 0.075 (0.051) 0.480*** (0.072) 9.376*** (0.851) -0.717*** (0.117) 0.828** (0.267) 0.207 (0.458) 25.446** (12.422) -0.895** (0.275) -1.614*** (0.170) -0.330*** (0.058) 0.275*** (0.069) -0.302** (0.110) 0.200 (0.104)  (0.016) 0.078** (0.028) 0.059** (0.022) 0.905*** (0.164) -0.567*** (0.113) 0.021 (0.057) 0.592*** (0.078) 9.722*** (0.945) -0.740*** (0.131) 0.615** (0.272) 0.437 (0.520) 26.965* (14.077) -1.496*** (0.326)  (0.012) 0.112** (0.036) 0.075** (0.029) 0.694*** (0.145) -0.266** (0.102) 0.075 (0.051) 0.480*** (0.072) 9.377*** (0.851) -0.717*** (0.117) 0.829** (0.267) 0.207 (0.460) 25.544** (12.436) -0.897** (0.275)  (0.012) 0.111** (0.036) 0.077** (0.028) 0.697*** (0.145) -0.271** (0.103) 0.080 (0.050) 0.485*** (0.072) 9.439*** (0.856) -0.727*** (0.118) 0.808** (0.266) 0.147 (0.463) 25.522** (12.337) -0.891*** (0.074) -0.855*** (0.074) 1.188*** (0.144)  (0.012) 0.113** (0.036) 0.078** (0.028) 0.696*** (0.145) -0.276** (0.104) 0.081 (0.051) 0.488*** (0.072) 9.460*** (0.859) -0.729*** (0.118) 0.796** (0.266) 0.156 (0.467) 25.354** (12.330) -0.906** (0.276) -0.741*** (0.075) 1.791*** (0.416) -2.998** (0.950)  (0.012) 0.117** (0.037) 0.079** (0.028) 0.695*** (0.145) 0.284** (0.104) 0.081 (0.050) 0.493*** (0.072) 9.485*** (0.864) -0.732*** (0.119) 0.777** (0.264) 0.186 (0.472) 25.041** (12.367) -0.938** (0.277) -0.613*** (0.049) 0.642*** (0.078)  2  3  Treatment 5  Treatment 6 Splines with (1, 5, 9 ) knots -0.089** (0.042) 0.017 (0.031)  46  Militarized Interstate Conflicts  Treatment 1  Treatment 2  Independent Variables  BKT Default Splines  T and T  2  3  Time squared Time Cubbed Temporal Dummy 1 Temporal Dummy 2 Temporal Dummy 3 Temporal Dummy 4 Temporal Dummy 5 Temporal Dummy 6 Temporal Dummy 7 Temporal Dummy 8 Temporal Dummy 9  Numbers of observation R2  Treatment 4  Treatment 5  Treatment 6  Time Dummies  Splines with (1, 2, 5 ) knots  Splines with (1, 3, 6, 9 ) knots  Splines with (1, 5, 9 ) knots  2.090*** (0.281) 0.631** (0.311) 0.215 (0.318) 0.516* (0.303) 0.101 (0.324) -0.218 (0.364) -0.363 (0.355) -1.055** (0.438) 0.249 (0.324) omitted Omitted -10.640*** (0.657) 155,872 0.3660  -8.696*** (0.582) 155,872 0.3598  -8.738*** (0.581) 155,872 0.3587  -8.797*** (0.582) 155,872 0.3576  0.045 (0.184) 0.015 (0.041) 0.0009 (0.003)  Time passed  Temporal Dummy 10 Temporal Dummy 11 Constant  Treatment 3  -8.568*** (0.582) 155,872 0.3647  -10.656*** (0.708) 155,872 0.3138  *** means significant at 99% level, ** means significant at 95% level  47  Appendix B: The results of comparing different methods to control the missing values in the capital net inflow variable Miliarized Interstate Conflicts Independent Variables Democracy Low Democracy Distance Demolow * Democracy Distance The level of net capital inflow Changes in Capital net inflow Missings of capital net inflow Trade dependence Financial Market openness GDP PC low Contiguity GDPPC * Contiguity Major Power status Formal alliance National material power Policy interest (S-Score) Constant Numbers of observation R2  Treatment 1  Treatment 2  Treatment 3  Treatment 4  Treatment 5  Treatment 6 Normal Distribution (75 percentile and SD)  Assign Value 0  Assign Mean Value  Assign 75 percentile  Assign 25 Percentile  Normal Distribution( m ean and SD of non-missings)  -0.088** (0.042) 0.019 (0.032)  -0.088** (0.042) 0.019 (0.032)  -0.087** (0.042) 0.019 (0.032)  -0.089** (0.042) 0.019 (0.031)  -0.089** (0.042) 0.019 (0.032)  -0.090** (0.042) 0.017 (0.032)  0.050***  0.050***  0.049***  0.050***  0.050***  0.050***  (0.012)  (0.012)  (0.012)  (0.012)  (0.012)  (0.012)  0.114***  0.113**  0.121**  0.110**  0.035*  0.051**  (0.036)  (0.036)  (0.036)  (0.037)  (0.020)  (0.020)  0.076**  0.075**  0.092**  0.069**  0.028**  0.041**  (0.028)  (0.028)  (0.027)  (0.028)  (0.013)  (0.014)  0.690***  0.700***  0.535***  0.740***  0.703***  0.596***  (0.145) -0.267** (0.103) 0.076 (0.051) 0.483*** (0.072) 9.401*** (0.851) -0.721*** (0.117) 0.821** (0.267) 0.169 (0.461) 25.537** (12.375) -0.891** (0.276) -8.637 (0.581) 155,872 0.3621  (0.145) -0.268** (0.103) 0.076 (0.051) 0.483*** (0.072) 9.402*** (0.852) -0.721*** (0.117) 0.821** (0.267) 0.170 (0.461) 25.535** (12.378) -0.891** 0.276 -8.568*** (0.582) 155,872 0.3647  (0.152) -0.259** (0.104) 0.076 (0.051) 0.482*** (0.072) 9.386*** (0.850) -0.719*** (0.117) 0.824** (0.267) 0.153 (0.464) 25.573** (12.327) -0.889** (0.276) -8.645*** (0.582) 155,872 0.3625  (0.146) -0.269** (0.103) 0.075 (0.051) 0.483*** (0.072) 9.405*** (0.851) -0.721*** (0.117) 0.820** (0.267) 0.175 (0.461) 25.532** (12.391) -0.891** (0.276) -8.633*** (0.581) 155,872 0.3620  (0.144) -0.268** (0.104) 0.075 (0.051) 0.481*** (0.071) 9.435*** (0.846) -0.724 (0.116) 0.808** (0.266) 0.248 (0.459) 26.750** (12.560) -0.882** (0.275) -8.621*** (0.577) 155,872 0.3608  (0.150) -0.264** (0.104) 0.072 (0.051) 0.482*** (0.072) 9.408*** (0.850) 0.721*** (0.117) 0.816** (0.265) 0.215 (0.461) 25.690** (12.581) -0.879** (0.275) -8.621*** (0.580) 155,872 0.3617  *** means significant at 99% level, ** means significant at 95% level  48  Appendix C: The predicted probability of conflicts, under an alternative definition of democracy  0.009  Appendix C. The predicted probability of conflicts (Countries score higher than 5 of standardized POLITY score are counted as democracies)  0.008 0.007 0.006 0.005  Democracies Predicted Probability  0.004  Democracies Confidence Interval  0.003  Autocracies Predicted Probability  0.002  Autocracies Confidence Interval  0.001  Discrepant dyad Predicted Probability  0 0  1  2  3  4  5  6  Democracy Distance  7  8  9  10  Discrepant dyad Confidence Interval  49  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items