Do Hawks and Doves Deliver? The Words and Deeds of Foreign Policy in Democracies

Do Hawks and Doves Deliver? The Words and Deeds of Foreign Policy in Democracies Abstract What are the domestic determinants of international conflict? I employ elements of salience theory to build an issue emphasis approach to foreign policy. I argue that parties and candidates in democracies credibly signal their foreign policy position prior to their election and that leaders live up to their foreign policy position. Significant research explains how both the behavior of other states and domestic political institutions may constrain leaders, so there are reasons to doubt leaders may be able to match deeds with words. Some scholars have integrated measurements of partisanship into their theoretical explanations, but extant scholarship has not effectively introduced the foreign policy position of the executive into the equation. Using this approach, we can connect competing foreign policy platforms to conflict behavior in a new way. I estimate initiation of militarized interstate disputes by democracies from 1951–2000 in the empirical test, and the results provide support for the hypothesis. What are the domestic determinants of conflict involvement in democracies? Here, I consider the role of party positions and ideology via issue emphasis in understanding foreign policy outcomes in democracies. Ideology can be thought of in a number of different ways, but Budge and Farlie (1983) argue that we can best understand ideology by measuring how parties emphasize issues. I contribute to the current understanding of ideology and foreign policy by disaggregating ideology and measuring the foreign policy orientation of leaders in democracies. The basic argument is that candidates and parties convey messages about how they will handle foreign policy by selectively emphasizing peace, militarism, or some mixture of these concepts prior to national elections. To date, few scholars have considered the actual foreign policy stance of leaders in the study of conflict. By ignoring the party position, or more specifically, the foreign policy position of leaders, the extant literature provides a compelling yet incomplete set of arguments. Though it may seem self-evident that leaders who send hawkish messages to the public will engage in more conflict, this approach is important given the extant literature on foreign policy constraints in democracies. A number of pieces have shown how either the behavior of other states or domestic political institutions may otherwise keep democratic leaders from living up to their foreign policy position. For example, the strategic conflict avoidance (SCA) literature suggests potential target states may avoid contact with democratic leaders who are likely to lash out (e.g., Smith 1996). In the language of hawks and doves, it may be that hawkish leaders are less likely to engage in conflict because other states deny them that opportunity. At the other end of the spectrum, dovish leaders may be targeted more frequently, meaning pacific campaign promises actually lead to more conflict. Similarly, the formal literature suggests hawkish leaders may actually be better at securing long-term peace (e.g., Cowen and Sutter 1998). Moreover, the literature on domestic institutions (e.g., Auerswald 1999; Howell and Pevehouse 2007) suggests leaders may not be able to use force whenever they please due to domestic institutional constraints. Altogether, there are several reasons clearly identified in the literature that may keep leaders in democracies from behaving as promised. Given these possible constraints, testing whether words and deeds actually match is worthwhile. This research also provides implications for voters in democracies and builds on the previous scholarship. First, a more detailed approach to issue emphasis and foreign policy shows that voters can expect foreign policy to match leaders’ promises. Of course leaders may renege on policy promises in some cases, but there appears to be a relationship between words and deeds regarding foreign policy. Furthermore, this relationship extends beyond left–right politics. Though rightist parties may typically present themselves as hawks, not all hawks present the same policy promises, and some elites may subvert these basic left–right expectations (Whitten and Williams 2011). These findings matter from the academic perspective as well. A more nuanced view of leaders suggests binary measures of partisanship may not tell the whole story when it comes to ideology. Aggregate measures of ideology may be helpful when looking across different policy realms as Fordham (1998) and Arena and Palmer (2009) do, but analysis of foreign policy outcomes can be improved by isolating the foreign policy position of leaders in democracies. To fully contextualize my theory, I consider the extant literature first which allows me to accurately place my approach in the broader constellation of the previous scholarship. Following this, I provide a brief overview of salience theory. From here, a full explanation of the issue emphasis approach to foreign policy is provided. Next, an empirical test is outlined and the substantive results are interpreted. The final section concludes and offers several thoughts on future research. Context The most important set of extant literature that this piece addresses examine whether leaders are actually able to translate policy positions into reality. This research has tended to focus on how and when leaders in democracies are constrained. One strand of this work comes from the research on SCA, and suggests that potential target states will avoid leaders that are domestically vulnerable (Leeds and Davis 1997; Fordham 2005). In other words, states may deny democratic leaders the opportunity to divert from a declining domestic political situation or otherwise use force when they would prefer to. This debate surrounding diversionary theory and SCA is important to the extent that it constitutes one of the key divides within the domestic determinants of foreign policy literature. More specifically, the SCA literature is important to the approach taken here because the general SCA logic suggests that leaders may struggle to live up to their campaign promises if other states selectively avoid hawks and target doves. Another portion of significant scholarship has examined the role of institutional support on foreign policy. Those interested in the US case have turned toward the effects of Congress on the use of force and the likelihood of diverting. Clark (2000) theorizes that as the preferences of the president and Congress converge, conflicts become more likely and tend to last longer. Howell and Pevehouse (2005, 2007) show that interbranch politics are a key component of the equation.1 The frequency with which the president uses force depends on his level of partisan support in Congress. In the same realm of study, Brulé (2006) examines the interactive effect of the economy and Congress, showing that presidents are more likely to divert from a flagging economy when they face a hostile Congress. Carter and Scott (2009) examine a wide variety of foreign policy outcomes and show that individual members of Congress often take a very active role in shaping US foreign policy, often to the displeasure of the executive. Altogether, this literature suggests institutional support is a key component of this research agenda. This work is important to the argument made below because I claim leaders live up to their campaign promises, and this work suggests institutional constraints may keep leaders from doing so. Some new cross-national research highlights the role of government characteristics in accounting for international conflict. Auerswald (1999) argues that because leaders in democracies are vulnerable to elections, votes of no confidence, or both, we can expect institutional context to matter in international crises.2Brulé and Williams (2009) have investigated government characteristics (e.g., minority versus coalition governments, weak versus strong party discipline), economic downturns, and dispute initiation cross-nationally, finding that the level of government accountability influences the likelihood of diversionary conflict. The work of Kaarbo and Beasley (2008) further suggests that coalition governments featuring extremist parties tends to produce extreme (either peaceful or violent) foreign policy behavior. Furthermore, Clare (2010) studies the role of ideology as mediated by political institutions, arguing that when extremist parties become junior members of a coalition they can exercise disproportionate influence on foreign policy, with leftist outliers having a more pacific effect and rightist outliers leading to more aggressive behavior. The perspective provided by the contemporary literature suggests the foreign policy behavior of leaders in democracies is strongly constrained. It may be that domestic institutional support limits foreign policy options, or it may be that other states simply avoid hawkish leaders altogether. Without discounting the possibility of SCA or the importance of domestic political constraints, I argue here that the foreign policy positions of leaders still has a significant effect on outcomes abroad after taking these constraints on behavior into account. I suggest that we can get a better idea of leaders’ foreign policy positions by disaggregating ideology and isolating the foreign policy position of executives in democracies. Some formal work suggests hawkish leaders in democracy are actually more likely to secure long-term peace (Cowen and Sutter 1998; Cukierman and Tommasi 1998). The basic point here is that leaders in democracies may implement policies counter to their ideology in order to help secure reelection. Further formal work by Schultz (2005) delineates the scope and conditions under which we can expect hawks to truly deliver lasting peace. The approach presented here speaks directly to this literature by disaggregating ideology into specific policy components to identify hawks and doves. Moreover, my approach analyzes broad patterns of peace and conflict in democracies in order to see what types of leaders actually deliver lasting peace. Alongside research on institutional constraints and SCA, the literature on how leaders divert from poor domestic conditions and low public approval ratings has developed extensively (Ostrom and Job 1986; James and Oneal 1991).3 A number of skeptics challenge the logic of diversionary theory, arguing that we can explain foreign policy by accounting for international conditions (James and Hristoulas 1994; DeRouen 1995; Meernik and Waterman 1996; Gowa 1998). Following Hibbs’ (1977) logic concerning the economic preferences of Democratic and Republican presidents, Fordham (1998) finds that Democratic presidents are more likely to use force when facing inflation, while Republican presidents have used force more often in the face of high unemployment. Importantly, much of this early work on partisanship and diversionary theory considers only the party of the leader, as opposed to treating partisanship or ideology as continuous concepts. Finally, some authors have considered policy positions and continuous measures of ideology in the aggregate regarding interstate conflict. Palmer, London, and Regan (2004) have considered government position, showing that governments of the left in parliamentary democracies are less likely to engage in international disputes than governments of the right, but more likely to escalate disputes once engaged because they are more sensitive to removal from office in the face of international conflict. Recent cross-national research has also shown that right-leaning governments are willing to prosecute interstate disputes longer on average than left-leaning governments (Koch 2009). Arena and Palmer (2009) find that governments that lean to the right on their one-dimensional scale of ideology are more likely to initiate a militarized interstate dispute (MID), while left-leaning governments are more peaceful. These cross-national studies measure the positions of political elites using aggregate measures of ideology. To this point, there has been no attempt to measure the foreign policy position of the executive in the study of the domestic determinants of conflict. While some pieces treat partisanship dichotomously (e.g., Fordham 1998) and others employ continuous measures of government ideology, one of the key contributions made here is in isolating the foreign policy preferences of leaders by disaggregating ideology. Accounting for Policy Positions In order to better understand why certain leaders in democracies are more likely to initiate conflict than others, we should make some effort to consider their general position on foreign policy issues. That is, we must measure how hawkish or dovish leaders present themselves in order to better predict behavior. Salience theory provides a foundation to help explain how we might measure such a concept prior to an executive taking office and why issue emphasis should matter. Budge and Farlie (1983) introduced the issue emphasis approach, suggesting that parties attempt to emphasize salient issues that favor their strengths relative to their opponents.4 Ideology is itself a basket of issue positions and these issue positions change over time. By allowing ideology and issue emphasis to vary across a number of dimensions, we can see clearer, more detailed images of parties and their candidates. From here, we can isolate the issues of interest and analyze only those policy promises. Because this analysis focuses on foreign policy outcomes, it is important to isolate the foreign policy position of the PM/president in democracies. Recent research by Whitten and Williams (2011) show why this is necessary.5 Klingemann, Hofferbert, and Budge. (1994) build on this approach by introducing the salience theory of party competition. Salience theory as envisioned by these authors is not entirely distinct from Budge and Farlie’s (1983) approach. Their interest lies specifically with political parties and the issues they choose to emphasize. To quote Klingemann, Hofferbert, and Budge., “By stressing certain items and excluding others—without overtly denouncing the latter—parties are, to be sure, implicitly taking pro and anti positions” (25). Through this process, parties put together the package of policies that they intend to offer the voters by emphasizing a set of issues. Extant work has tapped such emphases by analyzing party manifestos (e.g., Budge 1994). While some skeptics might offer that voters are unaware of party manifestos or party stances, Klingemann, Hofferbert, and Budge. (1994) argue that manifestos have an impact. To quote their argument directly, “In most democracies, the manifesto or platform is launched at a press conference, with great publicity, designed to set the major themes for the whole election. It is through the appearance of programmatic themes[…] in the media that the document makes it main impact on electors” (21). Moreover, we can certainly imagine that competing parties loudly publicize aspects of their competitors’ policy platforms that they expect to be unpopular.6,7 As the general issue emphasis approach posited by Budge and Farlie (1983) suggests, politicians highlight their strengths to their opponents’ detriment as best they can. Along somewhat similar lines as the above noted authors, Petrocik (1996) argues that political parties in the United States own certain issues and that the ownership of issues is relatively static over time. McDonald and Budge (2005) agree that parties can be expected to emphasize the same set of issues over time (e.g., we would expect Republicans to consistently emphasize the need for a strong national defense relative to Democrats in the United States). So while parties typically do not leapfrog one another on issues, the extent to which they emphasize different issues varies over time (Budge 1994). This variation over time is critical to my understanding of how leaders and governments make foreign policy. It may be perfectly reasonable to theorize that Democratic presidents in the United States or Socialist presidents in France behave as doves relative to their rightist competition, but it is also vital to account for the variation between different Democratic and Socialist presidents. Theory This piece applies salience theory to foreign policy behavior. I conceive of policy positions and their role in policymaking in a manner consistent with the extant literature. Parties and candidates in all democracies broadly emphasize some set of policies to voters, implicitly taking policy positions. Even in states that feature relatively short campaign periods, parties are expected to provide some sort of policy platform that covers a variety of issues. Once in office, politicians will make some effort to follow up on these policy positions, though countervailing forces may demand compromise. Of course significant events beyond the control of politicians may demand leaders and governments shift policy away from platform promises. However, I assume these significant exogenous events are not the norm. The implication is that we can still expect policy positions to affect policy outcomes on average. As noted above, my particular interest here is the behavior of states abroad. Foreign aid generosity and defense spending can be tied to foreign policy positions, but these correlations have already been well mapped by Budge and Hofferbert (1990) in the American case and by McDonald and Budge (2005) in a cross-national analysis. This piece approaches foreign policy issue emphasis by examining conflict behavior. By studying involvement in international disputes, this piece analyzes a set of outcomes that have not previously been studied in terms of disaggregated policy positions. I analyze conflict from the perspective of leaders, but this approach does not take other elite behavior for granted. A rich literature that shows how junior coalition partners (e.g., Kaarbo 1996; Clare 2010; Oppermann and Brummer 2014) and other sources of institutional support (e.g., Howell and Pevehouse on the US Congress) can influence foreign policy decisions, including interstate disputes. The contention I make here is not that divided or coalition governments have no role to play in the decision-making process. Any empirical test of the domestic determinants of conflict should account for these factors. The theoretical approach here suggests that leaders have an effect on foreign policy outcomes after controlling for sources of institutional support. Ultimately, it is beyond the scope of this paper to test the proposition that the executive dominates foreign policy across different democracies. The argument here is not incompatible with this perspective that elites outside of the executive matter. Indeed, it is the strength of domestic political constraints in democracies that make further investigation of whether hawkish leaders engage in more conflict a worthwhile subject for analysis. If we assume the PM or president has at least some influence over foreign policy, we can see how their foreign policy promises may matter. The translation between foreign policy issue emphasis and the initiation of interstate disputes is clear. The candidate or party who wins control of the executive position in a democracy has clear incentives to follow up on the issues that they have chosen to emphasize. Signaling a hawkish foreign policy orientation represents a type of “tying of the hands” as Fearon (1997) puts it. Leaders who promise to act tough will suffer domestic audience costs if they fail to live up to their promises. These domestic audience costs can be conceptualized as lost votes during a bid for reelection or reputation costs suffered at home. Though Fearon focuses on leaders tying their hands in order to demonstrate resolve to other leaders, this logic should hold true for leaders who outline a pacific foreign policy portfolio during the election campaign. A leader who promises to keep a state out of conflicts and peacefully negotiate in the face of international tension may be punished by voters if he or she engages in conflict unnecessarily. These potential domestic costs as outlined by Fearon should make messages sent during campaigns credible. I also suggest that while parties and leaders emphasize issues that they believe they enjoy a comparative advantage in, they also follow through by shaping policy to match their rhetoric in order to appear responsive to voters. Of course, not all candidates run for reelection, but I suggest that words should match deeds ceteris paribus, because campaign promises reveal in part the preferences of elected leaders on different policy issues. Though parties and elites vary their policy positions to remain competitive, I also suggest here that democratic candidates reveal their preferences for handling foreign policy during campaigns. Messages communicated by candidates are comprised of both strategic vote-chasing elements and personal preferences on individual policy issues according to this line of thinking.8 A concrete example should be helpful in outlining the logic of the theory. Sensing a weakness in incumbent Jimmy Carter’s foreign policy, challenger Ronald Reagan pounced during the 1980 election, outlining a more aggressive foreign policy. As Petrocik (1996) shows, Reagan recognized that the public perceived Republicans as owning national defense, and he sought to take advantage of the issue by offering a militaristic policy alternative to Carter. Interestingly, Reagan actually implemented his foreign policy as advertised. In his first term in office, Reagan initiated twenty-two major international uses of force, contrasted by six uses of major force under the Carter administration.9 At first blush, the wisdom of hawks and doves as Republicans and Democrats holds, but when we note that Dwight Eisenhower opted to use force only twenty-seven times during his entire two terms as president, things become less clear. Reagan and Eisenhower were both Republicans, but Eisenhower and Republicans of his day offered a different and more pacific foreign policy stance than did Reagan and the Republicans of the 1980s. This distinction is an important one and needs to be studied beyond the American example. It is also important to note that my argument assumes foreign policy orientations can be considered distinct from other policy positions (e.g., positions on government spending, social justice, and environmental protection). This line of thinking echoes Wildavsky’s (1966) argument about the dual presidency in the United States, and the argument here suggests all leaders have both domestic and foreign policy orientations that are distinct from one another. I argue hawkish leaders initiate militarized disputes more frequently, while dovish leaders abstain. There are, however, some alternative arguments about how potential target states may shy away from contact with democratic leaders who are likely to engage in conflict. In particular, the SCA perspective suggests states avoid politically vulnerable leaders. In his analysis of diversionary theory, Smith (1996) posits this sort of argument. Smith suggests that diverting is unlikely to occur because potential target states strategically avoid conflict with leaders who may look to divert. There is some mixed evidence for the general SCA argument. For example, Leeds and Davis (1997) show rivals avoid provocative behavior when a rival democracy is experiencing slow growth or approaching an election. Clark (2003) further shows that democracy promotes diversionary incentives while simultaneously making it difficult to act on such impulses. Examining only the United States, Fordham (2005) finds qualified support for this approach, showing that rival states are less likely to publicly challenge the United States when growth is low or inflation is high. Transferring the basic logic of SCA to the study of hawks and doves in democracies, it may be that rival states avoid hawkish leaders who are likely to lash out if targeted. This would imply that dovish leaders are actually engaged in relatively more conflicts. It may even be that dovish leaders are targeted specifically because other leaders view them as weak. The overall effect would be that we can expect foreign policy in democracies to be the opposite of what leaders campaign on. This is a distinct possibility and one that runs counter to my proposition. In other words, the logic of SCA relating to ideology and partisanship provides another good reason to test this argument about the words and deeds of foreign policy in democracies.10 From the perspective presented here, foreign policy tensions may arise for a variety of reasons (e.g., disagreements about fishing rights between Iceland and the United Kingdom in the North Atlantic). The threat or use of military force in these disputes is directly related to the foreign policy position of the leaders in question. Potential target states may want to avoid violence, but hawkish leaders in democracies should be more likely to use force regardless of the target state’s behavior or preferences. Though the foreign policy position approach and the SCA approach both posit plausible explanations about conflict in democracies, their empirical expectations run in opposite directions. From this theoretical argument I derive a testable hypothesis: H: Leaders from parties that emphasize militaristic foreign policy concepts will initiate international militarized conflict more frequently. Research Design I test the issue emphasis approach to foreign policy on a sample of twenty-five democracies from 1951 through 2000. Table 1 describes the cross-national and temporal characteristics of the democracies in the sample. The unit of analysis is the directed dyad quarter. I use quarterly data because many parliamentary systems change governments mid-year. Annual data might otherwise attribute conflicts to the wrong government. The sample of directed dyadic observations includes all politically relevant dyads in all models. I use directed dyad data because it is important to distinguish between MIDs initiated by the democracies in question and those disputes that target the democracies in the sample. In the tests examining MID initiation (Table 2), I consider all those directed dyads where the democracies in the sample may initiate a dispute against another state (e.g., the United States attacking China). In the models examining whether other states target democracies (Table 3), I examine only directed dyadic observations where the democracies in the sample constitute a potential target (e.g., China attacking the United States). Table 1. Sample countries   Sample  Australia  1951–2000  Austria  1955–2000  Belgium  1951–2000  Canada  1953–2000  Denmark  1951–2000  Finland  1951–2000  France  1951–2000  Germany  1971–2000  Greece  1975–2000  Iceland  1951–2000  Ireland  1951–2000  Israel  1951–2000  Italy  1951–2000  Japan  1961–2000  Luxembourg  1951–2000  Netherlands  1951–2000  New Zealand  1951–2000  Norway  1952–2000  Portugal  1976–1978; 1983–2000  Spain  1977–2000  Sweden  1951–2000  Switzerland  1951–2000  Turkey  1951–1971;1973–1979;1983–2000  United Kingdom  1951–2000  United States  1951–2000    Sample  Australia  1951–2000  Austria  1955–2000  Belgium  1951–2000  Canada  1953–2000  Denmark  1951–2000  Finland  1951–2000  France  1951–2000  Germany  1971–2000  Greece  1975–2000  Iceland  1951–2000  Ireland  1951–2000  Israel  1951–2000  Italy  1951–2000  Japan  1961–2000  Luxembourg  1951–2000  Netherlands  1951–2000  New Zealand  1951–2000  Norway  1952–2000  Portugal  1976–1978; 1983–2000  Spain  1977–2000  Sweden  1951–2000  Switzerland  1951–2000  Turkey  1951–1971;1973–1979;1983–2000  United Kingdom  1951–2000  United States  1951–2000  Notes: The start date of the data is constricted by the availability of GDP growth data from the Penn World Table (Germany). Start dates for Greece, Portugal, Spain, Turkey, and Japan are determined by the availability of CMP and government data, as these states experienced democratic transitions during this time period. The end date is determined by the availability of alliance portfolio data. All exclude caretaker governments. View Large Table 2. Foreign policy issue emphasis and conflict initiation, 1951–2000   (1)  (2)  (3)  (4)    All MIDs  All MIDs  Hostile MIDs  Fatal MIDs  Militarism index  0.040***  0.179***  0.043***  0.055***  (0.011)  (0.069)  (0.015)  (0.017)  Relative capability  10.031***  10.096***  8.152***  10.335***  (3.727)  (3.744)  (2.578)  (2.921)  Polity of target  −0.084***  −0.085***  −0.104***  −0.135***  (0.014)  (0.014)  (0.012)  (0.017)  Alliance portfolio  −0.327  −0.319  −0.042  −0.752***  (0.308)  (0.314)  (0.236)  (0.260)  Distance  −0.227***  −0.228***  −0.275***  −0.248***  (0.025)  (0.025)  (0.020)  (0.021)  Other initiated disputes  1.191***  1.205***  0.961***  1.503***  (0.139)  (0.138)  (0.173)  (0.186)  GDP growth  −0.008  −0.004  0.011  0.030  (0.014)  (0.014)  (0.022)  (0.023)  CIEP (% left)  0.238  0.224  0.222  0.653**  (0.181)  (0.180)  (0.273)  (0.302)  Single-party majority  −0.189  −0.232  −0.341  0.213  (0.179)  (0.179)  (0.221)  (0.243)  Coalition  −0.402**  −0.425**  −0.312  0.091  (0.193)  (0.195)  (0.208)  (0.238)  Coalition ideological distance  0.020*  0.022*  0.039***  0.048***  (0.011)  (0.011)  (0.010)  (0.011)  Executive left–right  0.002  −0.002  0.005  0.001  (0.004)  (0.004)  (0.004)  (0.004)  Peace quarters  −0.074***  −0.075***  −0.077***  −0.060***  (0.008)  (0.008)  (0.010)  (0.011)  Peace quarters-squared  0.001***  0.001***  0.001***  0.000**  (0.000)  (0.000)  (0.000)  (0.000)  Peace quarters-cubed  −0.000***  −0.000***  −0.000***  −0.000  (0.000)  (0.000)  (0.000)  (0.000)  Constant  −9.796***  −9.805***  −8.785***  −11.706***  (2.637)  (2.654)  (1.934)  (2.230)  Log likelihood  −1,950.903  −1,951.153  –  –  Observations  112,718  112,718  112,718  112,718  Standardized IV  No  Yes  No  No  Rare events estimator  No  No  Yes  Yes    (1)  (2)  (3)  (4)    All MIDs  All MIDs  Hostile MIDs  Fatal MIDs  Militarism index  0.040***  0.179***  0.043***  0.055***  (0.011)  (0.069)  (0.015)  (0.017)  Relative capability  10.031***  10.096***  8.152***  10.335***  (3.727)  (3.744)  (2.578)  (2.921)  Polity of target  −0.084***  −0.085***  −0.104***  −0.135***  (0.014)  (0.014)  (0.012)  (0.017)  Alliance portfolio  −0.327  −0.319  −0.042  −0.752***  (0.308)  (0.314)  (0.236)  (0.260)  Distance  −0.227***  −0.228***  −0.275***  −0.248***  (0.025)  (0.025)  (0.020)  (0.021)  Other initiated disputes  1.191***  1.205***  0.961***  1.503***  (0.139)  (0.138)  (0.173)  (0.186)  GDP growth  −0.008  −0.004  0.011  0.030  (0.014)  (0.014)  (0.022)  (0.023)  CIEP (% left)  0.238  0.224  0.222  0.653**  (0.181)  (0.180)  (0.273)  (0.302)  Single-party majority  −0.189  −0.232  −0.341  0.213  (0.179)  (0.179)  (0.221)  (0.243)  Coalition  −0.402**  −0.425**  −0.312  0.091  (0.193)  (0.195)  (0.208)  (0.238)  Coalition ideological distance  0.020*  0.022*  0.039***  0.048***  (0.011)  (0.011)  (0.010)  (0.011)  Executive left–right  0.002  −0.002  0.005  0.001  (0.004)  (0.004)  (0.004)  (0.004)  Peace quarters  −0.074***  −0.075***  −0.077***  −0.060***  (0.008)  (0.008)  (0.010)  (0.011)  Peace quarters-squared  0.001***  0.001***  0.001***  0.000**  (0.000)  (0.000)  (0.000)  (0.000)  Peace quarters-cubed  −0.000***  −0.000***  −0.000***  −0.000  (0.000)  (0.000)  (0.000)  (0.000)  Constant  −9.796***  −9.805***  −8.785***  −11.706***  (2.637)  (2.654)  (1.934)  (2.230)  Log likelihood  −1,950.903  −1,951.153  –  –  Observations  112,718  112,718  112,718  112,718  Standardized IV  No  Yes  No  No  Rare events estimator  No  No  Yes  Yes  Robust standard errors in parentheses. *p<0.1, **p<0.05, ***p  <0.01; two-tailed tests. View Large Table 3. Foreign policy issue emphasis and targeted MIDs, 1951–2000   (5)  (6)  (7)  (8)    All MIDs  All MIDs  Hostile MIDs  Fatal MIDs  Militarism index  0.042***  0.072  0.047***  0.056***  (0.014)  (0.062)  (0.017)  (0.020)  Relative capability  8.563***  8.233***  9.835***  11.491***  (2.135)  (2.041)  (1.564)  (2.129)  Polity of initiator  −0.076***  0.017  −0.103***  −0.142***  (0.013)  (0.038)  (0.012)  (0.021)  Alliance portfolio  0.087  −0.317  0.203  −0.092  (0.377)  (0.391)  (0.223)  (0.269)  Distance  −0.186***  −0.181***  −0.216***  −0.233***  (0.028)  (0.030)  (0.019)  (0.023)  Other initiated disputes  2.473***  2.797***  2.146***  2.738***  (0.197)  (0.198)  (0.170)  (0.193)  GDP growth  −0.003  −0.000  0.003  0.000  (0.016)  (0.017)  (0.022)  (0.026)  CIEP (% left)  0.017  0.046  0.026  0.329  (0.171)  (0.169)  (0.231)  (0.278)  Single-party majority  0.017  0.010  0.077  0.084  (0.154)  (0.165)  (0.162)  (0.227)  Coalition  −0.242  0.194  −0.465**  0.010  (0.186)  (0.208)  (0.208)  (0.253)  Coalition ideological distance  0.022*  0.021  0.044***  0.059***  (0.012)  (0.014)  (0.011)  (0.010)  Executive left–right  −0.003  −0.002  −0.002  0.001  (0.004)  (0.004)  (0.004)  (0.004)  Peace quarters  −0.072**  −0.074***  −0.073***  −0.073***  (0.009)  (0.010)  (0.010)  (0.012)  Peace quarters-squared  0.001***  0.001***  0.001***  0.001***  (0.000)  (0.000)  (0.000)  (0.000)  Peace quarters-cubed  −0.000***  −0.000***  −0.000**  −0.000  (0.000)  (0.000)  (0.000)  (0.000)  Constant  −9.278***  −9.037***  −10.380***  −12.600***  (1.590)  (1.484)  (1.184)  (1.605)  Log likelihood  −2,047.825  −2,146.654  –  –  Observations  112,919  112,919  112,919  112,919  Standardized IV  No  Yes  No  No  Rare events estimator  No  No  Yes  Yes    (5)  (6)  (7)  (8)    All MIDs  All MIDs  Hostile MIDs  Fatal MIDs  Militarism index  0.042***  0.072  0.047***  0.056***  (0.014)  (0.062)  (0.017)  (0.020)  Relative capability  8.563***  8.233***  9.835***  11.491***  (2.135)  (2.041)  (1.564)  (2.129)  Polity of initiator  −0.076***  0.017  −0.103***  −0.142***  (0.013)  (0.038)  (0.012)  (0.021)  Alliance portfolio  0.087  −0.317  0.203  −0.092  (0.377)  (0.391)  (0.223)  (0.269)  Distance  −0.186***  −0.181***  −0.216***  −0.233***  (0.028)  (0.030)  (0.019)  (0.023)  Other initiated disputes  2.473***  2.797***  2.146***  2.738***  (0.197)  (0.198)  (0.170)  (0.193)  GDP growth  −0.003  −0.000  0.003  0.000  (0.016)  (0.017)  (0.022)  (0.026)  CIEP (% left)  0.017  0.046  0.026  0.329  (0.171)  (0.169)  (0.231)  (0.278)  Single-party majority  0.017  0.010  0.077  0.084  (0.154)  (0.165)  (0.162)  (0.227)  Coalition  −0.242  0.194  −0.465**  0.010  (0.186)  (0.208)  (0.208)  (0.253)  Coalition ideological distance  0.022*  0.021  0.044***  0.059***  (0.012)  (0.014)  (0.011)  (0.010)  Executive left–right  −0.003  −0.002  −0.002  0.001  (0.004)  (0.004)  (0.004)  (0.004)  Peace quarters  −0.072**  −0.074***  −0.073***  −0.073***  (0.009)  (0.010)  (0.010)  (0.012)  Peace quarters-squared  0.001***  0.001***  0.001***  0.001***  (0.000)  (0.000)  (0.000)  (0.000)  Peace quarters-cubed  −0.000***  −0.000***  −0.000**  −0.000  (0.000)  (0.000)  (0.000)  (0.000)  Constant  −9.278***  −9.037***  −10.380***  −12.600***  (1.590)  (1.484)  (1.184)  (1.605)  Log likelihood  −2,047.825  −2,146.654  –  –  Observations  112,919  112,919  112,919  112,919  Standardized IV  No  Yes  No  No  Rare events estimator  No  No  Yes  Yes  Robust standard errors in parentheses. *p<0.1, **p<0.05, ***p  <0.01; two-tailed tests. View Large I test my hypothesis on three different dependent variables to examine conflict behavior in democracies. The measures for the dependent variables come from the Correlates of War Militarized Interstate Dispute 3.1 data (Ghosn, Palmer, and Bremer 2004).11 I first examine a binary dependent variable that is equal to one in each observation if the democracy in question initiated any MID and zero otherwise. The argument for using this broad indicator is that the foreign policy position of leaders matter for foreign policy in a general way, so an analysis of all MIDs makes theoretical sense. I also examine a narrower set of dependent variables that includes only those MIDs that lead to the use of force (Hostile MIDs), and a final measure of only the initiation of MIDs leading to at least one fatality (Fatal MIDs). These dependent variables are analyzed to make sure the foreign policy position of leaders influences salient events that are likely to capture the public’s attention.12 The empirical approach here not only tests the hypothesis posited above, but explicitly examines whether dovish leaders are more likely to be targeted by other states. This is part of the SCA perspective discussed above. If other states avoid hawkish leaders, we should expect hawkishness to be negatively associated with conflict. Though I offer no explicit hypotheses about whether other states might target dovish leaders more frequently than hawkish leaders, this possibility derived from the SCA perspective is worth investigating given its implications. If dovish leaders are in fact subject to challenges from abroad, such a finding would indicate a universal incentive for democratic leaders to take hawkish positions. I examine the same three dependent variables described above in a set of models that examine whether democratic leaders are targeted by other states on the basis of their foreign policy position (Table 3). These models include all the covariates described below, and, as noted above, these models examine only directed-dyad observations where the democracies from the sample may be targeted. I expect the foreign policy position of the leader to have an impact on foreign policy outcomes, and I measure the foreign policy position of leaders by looking at emphasis of different foreign policy concepts using data from the Comparative Manifesto Project (Volkens et al. 2013). The CMP performs content analysis of party manifestos and codes emphasis of individual issues (e.g., positive statements about free enterprise or positive statements about the welfare state) as a proportion of the entire manifesto document for each party in a given election. I use individual measures of foreign policy emphasis instead of aggregate measures of a government’s right–left position. The theoretical argument focuses on the foreign policy position of the presidential or prime minister’s party and not on the ideological orientation of the government as a whole. I make no particular predictions about the nature of leftist or rightist parties in government. I consider measures of emphasis on three different foreign policy concepts and use these to create an index as the independent variable. I first consider positive militarism statements.13 These include statements that emphasize self-defense, the need to improve the state’s military strength, and the importance of external security among other similar topics. A measure of negative military emphasis is coded the same as the pro-military variable, but counts statements that oppose pro-military topics. A third variable measures emphasis of pacifism and relies on statements that mention peace as a general goal and declarations that peaceful means should be used in solving disputes. I create a basic Militarism Index by subtracting the sum of the pacifism and anti-militarism variables from the value of the positive militarism variable.14 The intuitive meaning of this process is that I capture the proportion of statements referring to the concepts of interest in each manifesto and calculate an index where higher numbers should indicate a more hawkish position. Figure 1 displays the values of the Militarism Index of the major US political parties since World War II. The story this figure tells indicates that the measure has some level of face validity. The parties provide divergent policy choices for the most part, and Republicans appear more aggressive than Democrats in most years. Of specific interest is the 1992 election, in which Bill Clinton and the Democrats actually leapfrogged incumbent George H.W. Bush and the Republicans. This switch may actually be in line with standard thinking about the 1992 election. H.W. Bush is widely perceived as having moderated his tone in the wake of the Persian Gulf War, while Clinton pushed the Democrats further to the right on a number of issues, including foreign policy (Volkens et al. 2013). Figure 1. View largeDownload slide Net emphasis of militarism over time by Major Parties in the United States, 1948–2008 Figure 1. View largeDownload slide Net emphasis of militarism over time by Major Parties in the United States, 1948–2008 More importantly, variation over time between and within parties clearly exists. Some of this variation can be explained via the logic of Nincic (1988) work, who suggests the US public may reward traditionally dovish leaders for acting tough and hawks for moderating their behavior. With regard to the US context, the elections in the 1990s may have represented a strategic move by the Clinton Democrats to gain votes by taking a hawkish position with the expectation that voters would reward such behavior. Variation in policy position can be partially understood as an attempt to gain votes, and the goal here is to see how this strategic maneuvering translates to policy. In general, what we see is that neither party has occupied a static foreign policy position since the 1940s, and the variation displayed above is theoretically important. I include a set of control variables that capture the strategic dyadic relationship in each observation as well some measures of domestic political conditions for each democracy in the sample. Relative Capability, originally developed by Singer (1988), is an indicator of relative power. This measure is coded as the log of the ratio of State A’s CINC score to the sum of State A and State B’s CINC scores, plus one (Singer 1988). I also consider the regime type of potential targets using data from Polity IV (Marshall and Jaggers 2002). In models examining initiation (Table 2), the Polity of Target indicator represents the combined polity2 measure of democracy from Marshall and Jaggers’ data. In models examining whether states target democracies (Table 3), I include the Polity of Initiator variable. A measure of Distance is included, which measures the natural log of distance in miles between state capitals. Finally, the weighted global S score calculated by Signorino and Ritter (1999), which measures alliance portfolio similarity, is included.15 A final variable that helps account for opportunity is the Other Initiated Disputes measure. This is a dichotomous measure that is equal to one if the democracy in question has initiated a MID in another dyad during the same quarter and zero otherwise. The reasoning here is that states have limited military capabilities, and using force against any one target may preclude the use of force in another dyad, even if a leader is relatively hawkish. The model includes two control variables that may constitute incentives for democratic leaders to divert or incentives for potential target states to avoid contact with vulnerable democratic leaders. The first of these is a measure of GDP per capita growth in constant US dollars. This variable is labeled GDP Growth. As Downs and Rocke (1995) argue, leaders with little hope of reelection may gamble for resurrection by extending or escalating conflicts. Because declining economic conditions may represent such an incentive to gamble, the GDP growth indicator represents a necessary inclusion here. It may also be that declining economic conditions have a constraining effect on conflict involvement (Arena and Palmer 2009). This measure comes from the Penn World Tables (Heston, Summers, and Aten 2012). The second measure from the diversionary theory/SCA literature is an indicator of the proportion of the constitutional interelection period remaining for each democracy during each quarter. This is a measure of time remaining until a constitutionally mandated election. This measure, constitutional inter-election period (CIEP), comes from Seki and Williams’ (2014) update of Woldendorp, Keman, and Budge’s dataset on government duration in democracies. As discussed above, institutional factors may empower or constrain a leader. To account for this empirically, I include a set of binary indicators that measure whether a government is a coalition or majority government, with minority governments representing the excluded category (Woldendorp, Keman, and Budge 2000). In the case of presidential (the United States) or semi-presidential (France) systems, governments are coded as minority governments unless they enjoy majority support in all houses of the legislature, in which case they are coded a single-party majority. This coding practice is adopted from Powell’s (2000) approach. These measures are labeled Single-Party Majority and Coalition, respectively. Following Clare (2010), I exclude all caretaker governments in parliamentary democracies. Because the ideological position of vital junior coalition partners matters in parliamentary democracies (Kaarbo 1996), the model includes a measure called Coalition Ideological Distance that captures the distance between the PM in parliamentary democracies and the other governing parties. It may also be that hawkish parties in parliamentary systems form coalitions with other hawkish parties, so including this measure makes it possible to estimate the effect of the leader’s foreign policy orientation while accounting for junior coalition partners. To calculate this, I take the difference between the executive party’s aggregate ideological position and the government as a whole’s ideological position using Comparative Manifesto data (Volkens et al. 2013). Positive values of this variable indicate more conservative coalition partners while negative numbers indicate the presence of junior coalition partners to the left of the PM. This measure is always zero in presidential systems that do not feature coalition governments. Finally, the model includes a measure of the executive party’s left–right position as calculated using the normal CMP equation with all foreign policy variables purged (Volkens et al. 2013). The remaining value proxies the leader’s policy positions on domestic issues (e.g., the welfare state, taxes, expansion of public education, the role of religion in society, etc.). A higher value of this purged measure indicates a more conservative or rightist aggregate position, and this variable is labeled Executive Left–Right. Including this covariate provides evidence that the Militarism Index has an effect when controlling for the executive’s position on other policy issues. I use logistic regression to estimate all models with robust standard errors clustered on the dyad. To handle temporal dependence in the dependent variables, I follow Carter and Signorino’s (2010) method and include a count of peace quarters along with its squared and cubed terms for all dependent variables. To estimate initiation of hostile and fatal MIDs, I employ a corrected rare events logistic estimator because the initiation of these sorts of disputes is relatively rare (King and Zeng 2001). Finally, an alternative measure of the independent variable where the Militarism Index is divided by the standard deviation of each country’s party system Militarism Index during the most recent election is used as a robustness check on the main findings. This alternative provides some indication of how hawkish each leader is relative to their domestic competition. In these models (Models 2 and 6), the standardized Militarism Index is meant to account for country-election specific factors and the general foreign policy environment in each country during each election.16 Analysis The findings support the hypothesis. Political parties and party leaders in democracies signal their foreign policy positions prior to election by selectively emphasizing concepts in their party manifestos. Table 2 presents the results of the empirical test looking at MID initiation across different dependent variables. The results from Models 1–4 provide general support for the proposition about leaders and interstate conflict. In Model 1, we see hawkish leaders are more likely to initiate any sort of dispute. This finding is replicated in Models 3 and 4, which look at hostile and fatal MIDs, respectively. Here, we see again that hawkish leaders are more likely to initiate these sorts of militarized conflicts. The results of the models with the standardized independent variable are displayed in Model 2. The results are consistent with prior findings.17 The coefficient associated with the Militarism Index is still positive and significant. The coefficient is larger in Model 2 than others, but this is only due to the fact that the standardized independent variable scales differently than the regular Militarism Index. The control variables generally behave as expected. The Relative Capability measure is positive and significant across models, suggesting that democracies are more likely to target relatively weaker states. Democracies are less likely to target other democracies as the negative and significant coefficients associated with Polity of Target indicate across Models 1–4. Distance is negative and significant in all models, meaning conflict is less likely between states that are far apart. This effect is consistent across models. The Alliance Portfolio variable is not significant in any model except Model 4, where is it negative as expected. The coefficient for Other Initiated Disputes is surprisingly positive and significant across models, meaning that democracies are actually likely to initiate more disputes once they have initiated one. Turning to the domestic political covariates, the measure of GDP Growth is not significant in any model. CIEP is positive and significant when estimating initiation of disputes that lead to at least one fatality, but not otherwise significant. This indicates that democracies are more likely to initiate salient disputes when elections are relatively far away. This is normatively pleasing as these results provide evidence that leaders do not divert on the basis of the economy, nor do these results suggest leaders in democracies begin disputes near elections to divert the public’s attention. The Coalition variable is negative and significant in Models 1 and 2, implying that coalition governments are less conflict prone than other forms of government. This means coalition governments may be more constrained in initiating disputes. However, this effect does not hold when looking at more salient types of disputes. Interestingly, majority governments appear to show no difference from minority governments. Altogether, the evidence here is mixed regarding institutional support. However, we should exercise care in making inferences here, as Palmer, London, and Regan (2004) suggest government type is a key factor in general conflict involvement for parliamentary democracies, regardless of who initiates or reciprocates a dispute. The measure of Coalition Ideological Distance is significant across models, suggesting more conservative coalition partners increase the odds of a state initiating a dispute that leads to at least one fatality. This finding is in line with the previous quantitative work of Kaarbo and Beasley (2008) and Clare (2010). The purged Executive Left–Right measure is not significant in any model, implying that the position of leaders on issues outside of foreign policy (e.g., taxes, crime, religion, etc.) does not affect interstate conflict.18 Recall the SCA perspective suggests other states may avoid hawkish leaders and target dovish leaders. The results of Models 5 through 8 in Table 3 suggest that if anything, more hawkish foreign policy positions lead states to challenge democracies more frequently. This finding challenges the specific notion that hawkish leaders are able to secure peaceful outcomes without at least threatening or using force. This is true of all disputes (Model 5), disputes leading to the use of force (Model 7), and disputes leading to fatalities (Model 8). However, this is not true in the model where the standardized IV is used (Model 6), meaning this finding is inconsistent when accounting for other domestic party positions. The Militarism Index is positive, but not statistically significant, meaning this effect is not robust across models. The control variables in these models generally behave as expected, with Relative Capability and Distance signed as expected and significant across models. Polity of the Initiator is negative as expected, but only significant in Models 6–8. The results also show that coalition governments are less likely to be targeted than other types of government but only in terms of forceful dispute initiations. The Other Initiated Disputes measure is positive and significant in all models, which is the same finding noted above. Interestingly, more conservative coalition partners also lead governments to be targeted more frequently according to the results regarding Coalition Ideological Distance, though again, this does not hold in Model 6. The measure of GDP Growth is not significant in any model, which suggests low growth in democracies does not deter challengers. The measure of the proportion of the CIEP left is not significant in any model, meaning democracies are not any more or less likely to be targeted at any level of hostility as elections draw nearer. These results challenge the SCA perspective described above that states actively avoid conflict with leaders who are likely to lash out (e.g., Smith 1996). These results also speak to the formal literature that examines whether dovish leaders are capable of creating a lasting peace. Some of this work (e.g., Cowen and Sutter 1998) has suggested that hawks are actually better at securing peace, but the broad empirical patterns outlined here do not support this position. Hawkish leaders in democracies use force more frequently after controlling for a broad array of factors, and other leaders do not appear to systematically avoid conflict with hawkish leaders. If anything, it would seem a hawkish foreign policy position may invite conflict. The conclusions drawn above should not be taken as an outright rejection of the logic behind SCA and diversionary theory. It is certainly possible that these theoretical arguments speak to behavior not perfectly captured by the measures employed here. MIDs are hardly a perfect measure of interstate conflict (Kang and Gibler 2012; Fordham and Sarver 2001). The evidence discussed above simply provides support for the argument that leaders’ foreign policy positions are frequently translated into actual foreign policy after accounting for a broad array of domestic and international constraints. This also does not mean other state leaders ignore the ideology of democratic leaders, as recent work by Clare (2014) shows dovish governments are able to leverage better deals out of rival governments. The other key implication here is that there is not a universal incentive for candidates to take a hawkish foreign policy position in order to deter challengers. Substantive Effects To provide a more intuitive display of the substantive effects of the findings across different simulated scenarios, Table 4 uses results from Model 1 and presents the change in probability and 95 percent confidence intervals associated with a one standard deviation (4.08) shift from the mean (0.58) in the Militarism Index variable in three different scenarios.19 Model 1 is used here as it represents the most general model of conflict and includes all militarized disputes. Specifically, these three simulations represent scenarios in which the risk of conflict is either low, moderate, or high based on different values of the control variables. In order to provide a point of comparison for the impact of the Militarism Index in each scenario, the table also provides the relative effect of a one standard deviation change from the mean of two geostrategic covariates. First the Relative Capability measure is increased in each simulation. Recall this measure is the natural log transformation of the proportion of capabilities that Side A possesses, so it does not vary as much as the other covariates but is still a strong predictor of conflict in Model 1. A decrease from the mean of Polity of Target by one standard deviation is also included. One point worth noting is that the odds of conflict in any dyad-quarter in any scenario are very low, so a small increase in the odds of conflict in all dyads aggregates to a larger overall probability of any conflict for each simulated democracy. Table 4. Change in predicted probability of MID initiation (Model 1)   Change in predicted probability        Low  Moderate  High  Baseline probability  .0001  .0006  .013    (.00005, .00018)  (.0004, .00097)  (.009, .018)  Militarism index  +.00002  +.0001  +.002  (.52→4.6)  (.000006, .00004)  (.00005, .00021)  (.001, .004)  Relative capability  +.00004  + .0003  +.006  (.69→.73)  (.00001, .00009)  (.00008,.00053)  (.0001, .011)  Polity of target  +.0002  +.0005  +.006  (1→ −7)  (.00001, .0003)  (.00035, .00093)  (.003, .009)    Change in predicted probability        Low  Moderate  High  Baseline probability  .0001  .0006  .013    (.00005, .00018)  (.0004, .00097)  (.009, .018)  Militarism index  +.00002  +.0001  +.002  (.52→4.6)  (.000006, .00004)  (.00005, .00021)  (.001, .004)  Relative capability  +.00004  + .0003  +.006  (.69→.73)  (.00001, .00009)  (.00008,.00053)  (.0001, .011)  Polity of target  +.0002  +.0005  +.006  (1→ −7)  (.00001, .0003)  (.00035, .00093)  (.003, .009)    Values of Variables in Scenarios        Low  Moderate  High  Militarism index  −1.3  0  1.8  Relative capability  .69  .69  .70  Polity of target  10  1  −7  Alliance portfolio  .57  .37  .16  Distance (ln of miles)  8.6  8.3  7.2  Other initiated disputes  0  0  0  GDP growth  3.69  2.54  .96  CIEP (% left)  .33  .56  .79  Single-party majority  0  0  1  Coalition  1  1  0  Coalition ideological distance  −1.19  1  0  Executive left–right  −9.1  4.5  18.97  Peace quarters  80  40  8  Peace quarters-squared  6,400  1,600  64  Peace quarters-cubed  512,000  64,000  512    Values of Variables in Scenarios        Low  Moderate  High  Militarism index  −1.3  0  1.8  Relative capability  .69  .69  .70  Polity of target  10  1  −7  Alliance portfolio  .57  .37  .16  Distance (ln of miles)  8.6  8.3  7.2  Other initiated disputes  0  0  0  GDP growth  3.69  2.54  .96  CIEP (% left)  .33  .56  .79  Single-party majority  0  0  1  Coalition  1  1  0  Coalition ideological distance  −1.19  1  0  Executive left–right  −9.1  4.5  18.97  Peace quarters  80  40  8  Peace quarters-squared  6,400  1,600  64  Peace quarters-cubed  512,000  64,000  512  Note: 95 percent confidence intervals in brackets. View Large In the simulation that represents low underlying odds of conflict, the opportunity for conflict is low, and the dyad has been at peace for twenty years. Coalition is held at one here, while Single-Party Majority is equal to zero. All continuous measures are set at either their 25th or 75th percentile values depending on the sign of the coefficient in Model 1 to make the probability of initiation low, while the dichotomous Other Initiated Disputes measure is set to its mode (0). Varying the independent variable from its mean by one standard deviation shows the substantive effect of the foreign policy position of the leader matters even in dyads where the underlying odds of conflict are low. Though a one standard deviation increase in the Militarism Index only increases the probability of conflict involvement by .00002 on average, when one considers that the baseline probability of conflict in any given quarter in this scenario is only .0001, this change represents a 20 percent increase from the baseline probability of initiation. This change is smaller than increasing the relative capability of the initiator by a one standard deviation and the effect of varying the regime of the target. However, it should be noted that these geostrategic variables are increased by large amounts that we would probably not observe within a single dyad from year to year. We would only be likely to observe that sort of variation cross-sectionally. In other words, it would be unlikely to actually observe a target state’s Polity score change from 1 to –7 in one year, but we definitely observe changing values of the Militarism Index comparable to the values used in this simulation within dyads from year to year. Moving to the scenario that represents a moderate risk of conflict involvement, the relative risk increases, and the dyad has been at peace for ten years. Here, all variables are set to their 50th percentile values, except Coalition, which is equal to one, Single-Party Majority, which is equal to zero, and Other Initiated Disputes, which is also equal to zero. Again, the baseline probability of conflict is low (.0006), but a one standard deviation increase from the mean in the independent variable increases the probability of conflict by about 16 percent from the baseline, suggesting that the foreign policy position of the PM or president also matters in this more moderate scenario. Though the baseline probability of conflict is low, a positive shift in the independent variable increases these odds by a meaningful amount. Finally, the high risk of conflict scenario simulates a dyad that has been at peace for two years, while the control variables are adjusted to increase the underlying odds of conflict.20 In this scenario, a one standard deviation increase in the independent variable from the mean increases the odds of conflict involvement by .002, showing a 15 percent increase from the baseline probability of dispute initiation. Another key point here is that the confidence interval for the Militarism Index does not overlap zero in any simulation, meaning we can be relatively confident the leader’s foreign policy position matters across different hypothetical scenarios. Finally, the choice to use a one standard deviation change (4.08 units) in the independent variable is a conservative choice and justified given the real-life choices that voters face and the changes we observe within states. In terms of pivotal elections, the shift from Jimmy Carter’s Democratic party to Ronald Reagan’s Republican party in 1980 is measured as a 17.6 unit shift in the independent variable, while the shift from Yitzhak Rabin’s Alignment alliance to Menachem Begin’s Likud party in Israel in 1977 corresponds with roughly a 20 unit change. The change from James Callaghan’s Labour government to Margaret Thatcher’s Conservative government in Great Britain in 1979 represents a 4.6 unit shift. Pivotal elections in democracies regularly feature policy shifts that are similar to or significantly larger than the one employed in these simulations. In sum, these simulations show that the foreign policy position of the PM/president has a significant impact on the probability of dispute initiation given the underlying odds of conflict. Discussion and Conclusion To reiterate, the results of the empirical tests support the hypothesis. Political parties and leaders that emphasize concepts related to militarism relative to anti-militarism and pacifism are more likely to initiate conflicts. These findings shed light on the process of foreign policy-making in democracies. Leaders and parties in democracies credibly signal their foreign policy preferences by selectively emphasizing some mixture of foreign policy concepts. There is an empirical relationship between emphasis of these concepts in manifestos and actual behavior in the international arena. Even after accounting for a number of domestic and international factors that we expect to constrain leaders, I find evidence that leaders live up to the foreign policy positions they espouse during campaigns. This analysis contributes to the body of research on the domestic determinants of conflict. Previous research has shown that partisanship and ideology matter (Fordham 1998; Palmer, London, and Regan 2004; Arena and Palmer 2009; Koch 2009). However, no studies have systematically examined the foreign policy positions of individual leaders across democracies until now. This research contributes to the study of international conflict in a fine-grained manner. More specifically, this analysis contributes to our understanding of partisanship and foreign policy. What I have argued here is that the foreign policy position of leaders can be seen as independent of their policy positions on other issues. In other words, we must extract a party’s foreign policy preferences from their aggregate ideological position to better predict behavior in office. Second, these results suggest leaders and parties may pursue dovish foreign policy positions without necessarily being concerned that other states will target their state for this reason. This is appealing normatively, as it means leaders and parties can credibly present a wide range of foreign policy options to voters without having a universal incentive to appear hawkish. This finding speaks directly to the SCA literature. Of course there are caveats to this point, as large foreign policy events may incentivize hawkish positions across the party system (e.g., 9/11 in the United States). This finding does not suggest other leaders totally ignore the ideology of leaders. Indeed, forthcoming work by Clare shows that dovish governments have an advantage in bargaining with other states because international adversaries know that future bargaining episodes with stubborn hawkish governments may prove unproductive. Because hawkish governments are more likely to use force according to the findings here, I believe this work is consistent with this new research on ideology and external actors’ expectations. Furthermore, though this work fails to find evidence of SCA by potential target states or diversion by democratic leaders, we can imagine that the foreign policy positions of leaders in democracies might function in the SCA/diversionary theory framework in other ways. For example, though hawkish leaders may not be able to ward off challenges from abroad over salient issues, they may be able to keep potential joiners from bandwagoning against them with threats made credible by their hawkish disposition. In other words, there are a wide variety of hypotheses along these lines that I have not tested, but which future research should. Finally, these findings matter for careful consumers of political information in democracies. Voters can use issue ownership to figure out which parties own certain issues, but they should also be aware of variation over time in issue emphasis. The work of Budge and Farlie (1983) and Carmines and Stimson’s (1989) provides the basis for such an argument. In the United States, the Democrats and Republicans have not occupied static foreign policy positions over time since the end of World War II, and these changes have concrete implications for US foreign policy. The story is much the same in democracies around the world. If we want to know what sorts of leaders are likely to engage in international conflict, we should focus specifically on their emphasis of foreign policy concepts instead of highlighting only binary measures of partisanship. One possible extension of this research might be to examine how foreign policy positions matter in the context of national elections. In some cases, voters seem to reward aggressive foreign policy positions (Reagan in 1980 and 1984), while other elections feature a more dovish winner (Clinton in 1996). Nincic (1988) astutely argues that parties may seek votes by reversing their rhetoric, and this may well play into my general argument about words matching deeds. Further research on this subject is worthwhile, as Nincic and Hinckley (1991) show that voters in the US condition their vote on approval of the executive’s foreign policy. How domestic audiences reward or punish different foreign policy stances is probably conditional on the salience of foreign policy and the international atmosphere. Recent research on this subject has looked both at the United States (Gadarian 2010) and cross-nationally (Williams, Brulé, and Koch 2010; Koch 2011), but more research is needed. In any case the relationship between issue emphasis and foreign policy is worth further investigation. Footnotes 1 In a similar vein, Kesgin and Kaarbo (2010) delineate the scope and conditions under which we can expect parliaments to have an independent effect on foreign policy in parliamentary democracies. 2 See also Williams (2014). 3 More recent research on diversionary uses of force examines what issues or scenarios beyond economic conditions (e.g. rivalry, territorial disputes, or rapidly increasing power in the target) are likely to induce a diversionary attack (Mitchell and Prins 2004; Mitchell and Thyne 2010; Jung 2014). 4 Carmines and Stimson’s (1989) study of racial politics as an issue in the United States provides a compelling argument for us to take a dynamic view of issue competition. Because politicians behave strategically and because exogenous shocks to the political system demand attention, parties are forced to address issues as they evolve over time. The concept of issue evolution itself implies a dynamic restructuring of party politics. I make no claim here that foreign policy issue emphasis has itself reshaped politics across democracies to the degree that racial politics did in the United States during the 20th century. However, the general approach of Carmines and Stimson underlies a significant assumption made here, which is that policy issues are themselves dynamic. More recent research on diversionary uses of force examines what issues or scenarios beyond economic conditions (e.g., rivalry, territorial disputes, or rapidly increasing power in the target) are likely to induce a diversionary attack (Mitchell and Prins 2004; Mitchell and Thyne 2010; Jung 2014). 5 As Whitten and Williams find, the relationship between domestic economic positions and foreign policy orientations may produce unexpected results with regard to defense spending, as dovish parties often compromise by voting for increased military spending as a sort of welfare in disguise. If we only examine aggregate measures of ideology or assume issues remain static over time, we are certain to miss such nuances. 6 I would like to thank an anonymous reviewer for raising this possibility. 7 Several excerpts from the Conservative party manifesto from the 1987 British national election included in the online appendix provide evidence for this point. Moreover, this example provides a prime example of the foreign policy perspective of a leader (Margaret Thatcher) that is typically classified as hawkish. Though the Conservative manifesto discusses nuclear arms reduction, the balance of statements suggests Great Britain should be prepared to take decisive action abroad. 8 For example, the 1987 Conservative Party manifesto is introduced from the perspective of Margaret Thatcher. This implies that the document communicates not only the party’s perspective on individual issues but Thatcher’s as well. 9 These data on use of force come from Fordham, who updates the original Blechman and Kaplan (1978) data for use in his 1998 piece, “Partisanship, Macroeconomic Policy, and U.S. Uses of Force, 1949-1994.” These data include only major uses of force, as is standard practice in the literature. 10 I would like to thank an anonymous reviewer for raising this point. 11 In the online appendix to this piece, I examine whether hawkish leaders are more likely to become involved in international crises using the International Crisis Behavior data. The results suggest more hawkish leaders, as indicated by the Militarism Index, are more likely to become involved in international crises. 12 Though I used the Blechman and Kaplan (1978) data for illustrative purposes above, I use MID data here because the Blechman and Kaplan data only covers the United States, while the MID data provides coverage of all states. 13 The full coding definition of these variables is provided in the appendix. 14 The Pro-Militarism indicator is weakly positively correlated with the Anti-Militarism indicator (r = .10) and moderately correlated with the Pro-Peace indicator (r = .28) implying that leaders who discuss foreign policy tend to mix their language. This can be seen as evidence for using an index in this case, because the goal is to understand where the balance of statements lies. 15 Bennett and Stam’s (2000) EUGene software was used to generate these dyadic control variables. 16 I would like to thank an anonymous reviewer for this suggestion. 17 This standardized measure is also positive and significant when estimating hostile and fatal MIDs as well, and these results are available in the online appendix. 18 I also re-estimated Models 1–4 exchanging the Coalition Ideological Distance and purged Executive Left–Right measures for a measure of Government Ideology. The results of these models are very similar to the ones discussed above and are available in the online appendix. 19 Clarify was used to generate these simulations (Tomz, Wittenberg, and King 2003). 20 Because Coalition is set to zero and Single-Party Majority is set to one, the Coalition Ideological Distance measure is set to zero as well because there are no coalition partners in this hypothetical scenario. References Arena Phillip, Palmer Glenn. 2009. “ Politics or the Economy? Domestic Correlates of Dispute Involvement in Developed Democracies.” International Studies Quarterly  53: 955– 75. Google Scholar CrossRef Search ADS   Auerswald David P. 1999. “ Inward Bound: Domestic Institutions and Military Conflict.” International Organization  53: 469– 504. Google Scholar CrossRef Search ADS   Bennett D. Scott, Stam Allan. 2000. “ EUGene: A Conceptual Manual.” International Interactions  26: 179– 204. Google Scholar CrossRef Search ADS   Blechman Barry, Kaplan Stephen. 1978. Force Without War: U.S. Armed Forces as a Political Instrument . Washington DC: Brookings Institution Press. Brulé David. 2006. “ Congressional Opposition, the Economy, and US Dispute Initiation, 1946-2000.” The Journal of Conflict Resolution  50: 463– 83. Google Scholar CrossRef Search ADS   Brulé David J, Williams Laron K. 2009. “ Democracy, and Diversion: Government Arrangements, the Economy, and Dispute Initiation.” Journal of Peace Research  46: 777– 98. Google Scholar CrossRef Search ADS   Budge Ian, Farlie Dennis J.. 1983. Explaining and Predicting Elections: Issue Effects and Party Strategies in Twenty-Three Democracies . London: George Allen and Unwin. Budge Ian, Hofferbert Richard I.. 1990. “ Mandates and Policy Outputs: U.S. Party Platforms and Federal Expenditures.” The American Political Science Review  84: 111– 31. Google Scholar CrossRef Search ADS   Carmines Edward G, Stimson James A.. 1989. Issue Evolution: Race and the Transformation of American Politics . Princeton, NJ: Princeton University Press. Carter David, Signorino Curtis S.. 2010. “ Back to the Future: Modeling Time Dependence in Binary Data.” Political Analysis  18: 271– 92. Google Scholar CrossRef Search ADS   Carter Ralph D, Scott James M.. 2009. Choosing to Lead: Understanding Congressional Foreign Policy Entrepreneurs . Durham, NC: Duke University Press. Google Scholar CrossRef Search ADS   Clare Joe. 2010. “ Ideological Fractionalization and the International Conflict Behavior of Parliamentary Democracies.” International Studies Quarterly  54: 965– 87. Google Scholar CrossRef Search ADS   Clare Joe. 2014. “ Hawks Doves and International Cooperation”. Journal of Conflict Resolution  58: 1311– 37. Google Scholar CrossRef Search ADS   Clark David H. 2000. “ Agreeing to Disagree: Domestic Institutional Congruence and US Dispute Behavior.” Political Research Quarterly  53: 375– 400. Clark David H. 2003. “ Can Strategic Interaction Divert Diverionsary Behavior? A Model of US Conflict Propensity.” Journal of Politics  65: 1013– 39. Google Scholar CrossRef Search ADS   Cowen Tyler, Sutter Daniel. 1998. “ Why Only Nixon Could Go to China.” Public Choice  97: 605– 15. Google Scholar CrossRef Search ADS   Cukierman Alex, Tommasi Mariano. 1998. “ When Does It Take a Nixon to Go to China?” The American Economic Review  88: 180– 97. DeRouen Karl. 1995. “ The Indirect Link: Politics, the Economy, and the Use of Force.” The Journal of Conflict Resolution  39: 671– 95. Google Scholar CrossRef Search ADS   Downs George W, Rocke David M.. 1995. Optimal Imperfection? Domestic Uncertainty and Institutions in International Relations . Princeton, NJ: Princeton University Press. Fearon James D. 1997. “ Signaling Foreign Policy Interests: Tying Hands versus Sinking Costs.” Journal of Conflict Resolution  41: 68– 90. Google Scholar CrossRef Search ADS   Fordham Benjamin. 1998. “ The Politics of Threat Perception and the Use of Force: A Political Economy Model of US Uses of Force, 1949-1994.” International Studies Quarterly  42: 567– 90. Google Scholar CrossRef Search ADS   Fordham Benjamin. 2005. “ Strategic Conflict Avoidance and the Diversionary Use of Force.” Journal of Politics  67: 132– 53. Google Scholar CrossRef Search ADS   Fordham Benjamin, Sarver Christopher C.. 2001. “ Militarized Interstate Disputes and United States Uses of Force.” International Studies Quarterly  45: 455– 66. Google Scholar CrossRef Search ADS   Gadarian Shana Kushner. 2010. “ Foreign Policy at the Ballot Box: How Citizens Use Foreign Policy to Judge and Choose Candidates.” Journal of Politics  72: 1046– 62. Google Scholar CrossRef Search ADS   Ghosn Faten, Palmer Glenn, Bremer Stuart. 2004. “ The MID3 Data Set 1993-2001: Procedures, Coding Rules and Description.” Conflict Management and Peace Science  21: 133– 54. Google Scholar CrossRef Search ADS   Gowa Joanne. 1998. “ Politics at the Water’s Edge: Parties, Voters, and the Use of Force Abroad.” International Organization  52: 307– 24. Google Scholar CrossRef Search ADS   Heston Alan, Summers Robert, Aten Bettina. 2012. “ Penn World Tables Version 7.1.” Penn World Table Version 7.1, Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania . Hibbs Douglas A. 1977. “ Political Parties and Macroeconomic Policy.” American Political Science Review  71: 1467– 87. Google Scholar CrossRef Search ADS   Howell William G, Pevehouse Jon C.. 2005. “ Presidents, Congress, and the Use of Force.” International Organization  59: 209– 32. Google Scholar CrossRef Search ADS   Howell William G, Pevehouse Jon C.. 2007. While Dangers Gather: Congressional Checks on Presidential War Powers . Princeton, NJ: Princeton University Press. James Patrick, Hristoulas Athanasios. 1994. “ Domestic Politics and Foreign Policy: Evaluating a Model of Crisis Activity for the United States.” Journal of Politics  56: 327– 48. Google Scholar CrossRef Search ADS   James Patrick, Oneal John. 1991. “ The Influence of Domestic and International Politics on the President’s Use of Force.” The Journal of Conflict Resolution  35: 307– 32. Google Scholar CrossRef Search ADS   Jung Sung Chul. 2014. “ Foreign Targets and Diversionary Conflict.” International Studies Quarterly  58: 566– 78. Google Scholar CrossRef Search ADS   Kaarbo Juliet. 1996. “ Power and Influence in Foreign Policy Decision Making: The Role of Junior Coalition Partners in German and Israeli Foreign Policy.” International Studies Quarterly  40: 501– 30. Google Scholar CrossRef Search ADS   Kaarbo Juliet, Beasley Ryan K.. 2008. “ Taking It to the Extreme: The Effect of Coalition Cabinets on Foreign Policy.” Foreign Policy Analysis  4: 67– 81. Google Scholar CrossRef Search ADS   Kang Choong-Nam, Gibler Douglas M.. 2012. “ An Assessment of the Validity of Empirical Measures of State Satisfaction with the Systemic Status Quo.” European Journal of International Relations  Kesgin Baris, Kaarbo Juliet. 2010. “ When and How Parliaments Influence Foreign Policy: The Case of Turkey’s Iraq Decision.” International Studies Perspectives  11: 19– 36. Google Scholar CrossRef Search ADS   King Gary, Zeng Langche. 2001. “ Logistic Regression in Rare Events Data.” Political Analysis  9: 137– 63. Google Scholar CrossRef Search ADS   Klingemann Hans-Dieter, Hofferbert Richard, Budge Ian. 1994. Parties, Policies and Democracy . Boulder: Westview Press. Koch Michael T. 2009. “ Governments, Partisanship, and Foreign Policy: The Case of Dispute Duration.” Journal of Peace Research  46: 1– 19. Google Scholar CrossRef Search ADS   Koch Michael T. 2011. “ Casualties and Incumbents: Do the Casualties from Interstate Conflicts Affect Incumbent Party Vote Share?” British Journal of Political Science  41: 795– 817. Google Scholar CrossRef Search ADS   Leeds Brett Ashley, Davis David R.. 1997. “ Domestic Political Vulnerability and International Disputes.” Journal of Conflict Resolution  41: 814– 34. Google Scholar CrossRef Search ADS   Marshall Monty, Jaggers Keith. 2002. “ Polity IV: Political Regime Characteristics and Transitions, 1800-2002.” Polity IV Manual . McDonald Michael D, Budge Ian. 2005. Elections, Parties, Democracy: Conferring the Median Mandate . New York: Oxford University Press. Google Scholar CrossRef Search ADS   Meernik James, Waterman Peter. 1996. “ The Myth of the Diversionary Use of Force by American Presidents.” Political Research Quarterly  49: 573– 90. Google Scholar CrossRef Search ADS   Mitchell Sara McLaughlin, Prins Brandon C.. 2004. “ Rivalry and Diversionary Uses of Force.” Journal of Conflict Resolution  48: 937– 61. Google Scholar CrossRef Search ADS   Mitchell Sara McLaughlin, Thyne Clayton L.. 2010. “ Contentious Issues as Opportunities for Diversionary Behavior.” Conflict Management and Peace Science  27: 461– 85. Google Scholar CrossRef Search ADS   Mitchell Sara McLaughlin, Moore Will H.. 2002. “ Presidential Uses of Force During the Cold War: Aggregation, Truncation, and Temporal Dynamics.” Journal of Politics  46: 438– 52. Google Scholar CrossRef Search ADS   Nincic Miroslav. 1988. “ The United States, the Soviet Union, and the Politics of Opposites.” World Politics  40: 452– 75. Google Scholar CrossRef Search ADS   Nincic Miroslav, Hinckley Barbara. 1991. “ Foreign Policy and the Evaluation of Presidential Candidates.” The Journal of Conflict Resolution  35: 333– 55. Google Scholar CrossRef Search ADS   Oppermann Kai, Brummer Klaus. 2014. “ Patterns of Junior Partner Influence on the Foreign Policy of Coalition Governments.” British Journal of Politics and International Relations  16: 555– 71. Google Scholar CrossRef Search ADS   Ostrom Charles, Job Brian. 1986. “ The President and the Political Use of Force.” American Political Science Review  80: 541– 66. Google Scholar CrossRef Search ADS   Palmer Glenn, London Tamar, Regan Patrick. 2004. “ What’s Stopping You? The Sources of Political Constraints on International Conflict Behavior in Parliamentary Democracies.” International Interactions  30: 1– 24. Google Scholar CrossRef Search ADS   Petrocik John R. 1996. “ Issue Ownership in Presidential Elections with a 1980 Case Study.” American Journal of Political Science  40: 825– 35. Google Scholar CrossRef Search ADS   Powell G. Bingham. 2000. Elections as IInstrument of Democracy: Majoritarian and Proportional Visions . Yale University Press, New Haven, CT. Schultz Kenneth. 2005. “ Do Hawks or Doves Deliver the Olive Branch?” International Organization  59: 1– 38. Google Scholar CrossRef Search ADS   Seki Katsunori, Williams Laron K.. 2014. “ Updating the Party Government data set.” Electoral Studies  34: 270– 79. Google Scholar CrossRef Search ADS   Signorino Curtis S, Ritter Jeffrey M.. 1999. “ Tau-b or Not Tau-b: Measuring the Similarity of Foreign Policy Positions.” International Studies Quarterly  43: 115– 44. Google Scholar CrossRef Search ADS   Singer J. David. 1988. “ Reconstructing the Correlates of War Dataset on Material Capabilities of States, 1816-1985.” International Interactions  14: 115– 32. Google Scholar CrossRef Search ADS   Smith Alastair. 1996. “ Diversionary Foreign Policy in Democratic Systems.” International Studies Quarterly  40: 133– 53. Google Scholar CrossRef Search ADS   Tomz Michael, Wittenberg Jason, King Gary. 2003. “ CLARIFY: Software for Interpreting and Presenting Statistical Results.” Version 2.1 . Stanford University, University of Wisconsin, and Harvard University. Volkens Andrea, Lehmann Pola, Merz Nicolas, Sven Regel, Werner Annika. 2013. “ The Manifesto Data Collection. Manifesto Project (MRG/CMP/MARPOR).” Whitten Guy, Williams Laron K.. 2011. “ Buttery Guns and Welfare Hawks: The Politics of Defense Spending in Advanced Industrial Democracies.” American Journal of Political Science  55: 117– 35. Google Scholar CrossRef Search ADS   Wildavsky Aaron. 1966. “ The Two Presidencies.” Society  4: 7– 14. Google Scholar CrossRef Search ADS   Williams Laron K., Brule David J., Koch Michael T.. 2010. “ War Voting: Interstate Disputes, the Economy, and Electoral Outcomes.” Conflict Management and Peace Science  27: 442– 60. Google Scholar CrossRef Search ADS   Woldendorp Jaap, Keman Hans, Budge Ian. 2000. Party Government in 48 Democracies (1945-1998) . London: Kluwer Academic Publishers. Google Scholar CrossRef Search ADS   © The Author (2016). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Foreign Policy Analysis Oxford University Press

Do Hawks and Doves Deliver? The Words and Deeds of Foreign Policy in Democracies

Loading next page...
 
/lp/ou_press/do-hawks-and-doves-deliver-the-words-and-deeds-of-foreign-policy-in-x9OWxVojL8
Publisher
Wiley-Blackwell
Copyright
© The Author (2016). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1743-8586
eISSN
1743-8594
D.O.I.
10.1093/fpa/orw011
Publisher site
See Article on Publisher Site

Abstract

Abstract What are the domestic determinants of international conflict? I employ elements of salience theory to build an issue emphasis approach to foreign policy. I argue that parties and candidates in democracies credibly signal their foreign policy position prior to their election and that leaders live up to their foreign policy position. Significant research explains how both the behavior of other states and domestic political institutions may constrain leaders, so there are reasons to doubt leaders may be able to match deeds with words. Some scholars have integrated measurements of partisanship into their theoretical explanations, but extant scholarship has not effectively introduced the foreign policy position of the executive into the equation. Using this approach, we can connect competing foreign policy platforms to conflict behavior in a new way. I estimate initiation of militarized interstate disputes by democracies from 1951–2000 in the empirical test, and the results provide support for the hypothesis. What are the domestic determinants of conflict involvement in democracies? Here, I consider the role of party positions and ideology via issue emphasis in understanding foreign policy outcomes in democracies. Ideology can be thought of in a number of different ways, but Budge and Farlie (1983) argue that we can best understand ideology by measuring how parties emphasize issues. I contribute to the current understanding of ideology and foreign policy by disaggregating ideology and measuring the foreign policy orientation of leaders in democracies. The basic argument is that candidates and parties convey messages about how they will handle foreign policy by selectively emphasizing peace, militarism, or some mixture of these concepts prior to national elections. To date, few scholars have considered the actual foreign policy stance of leaders in the study of conflict. By ignoring the party position, or more specifically, the foreign policy position of leaders, the extant literature provides a compelling yet incomplete set of arguments. Though it may seem self-evident that leaders who send hawkish messages to the public will engage in more conflict, this approach is important given the extant literature on foreign policy constraints in democracies. A number of pieces have shown how either the behavior of other states or domestic political institutions may otherwise keep democratic leaders from living up to their foreign policy position. For example, the strategic conflict avoidance (SCA) literature suggests potential target states may avoid contact with democratic leaders who are likely to lash out (e.g., Smith 1996). In the language of hawks and doves, it may be that hawkish leaders are less likely to engage in conflict because other states deny them that opportunity. At the other end of the spectrum, dovish leaders may be targeted more frequently, meaning pacific campaign promises actually lead to more conflict. Similarly, the formal literature suggests hawkish leaders may actually be better at securing long-term peace (e.g., Cowen and Sutter 1998). Moreover, the literature on domestic institutions (e.g., Auerswald 1999; Howell and Pevehouse 2007) suggests leaders may not be able to use force whenever they please due to domestic institutional constraints. Altogether, there are several reasons clearly identified in the literature that may keep leaders in democracies from behaving as promised. Given these possible constraints, testing whether words and deeds actually match is worthwhile. This research also provides implications for voters in democracies and builds on the previous scholarship. First, a more detailed approach to issue emphasis and foreign policy shows that voters can expect foreign policy to match leaders’ promises. Of course leaders may renege on policy promises in some cases, but there appears to be a relationship between words and deeds regarding foreign policy. Furthermore, this relationship extends beyond left–right politics. Though rightist parties may typically present themselves as hawks, not all hawks present the same policy promises, and some elites may subvert these basic left–right expectations (Whitten and Williams 2011). These findings matter from the academic perspective as well. A more nuanced view of leaders suggests binary measures of partisanship may not tell the whole story when it comes to ideology. Aggregate measures of ideology may be helpful when looking across different policy realms as Fordham (1998) and Arena and Palmer (2009) do, but analysis of foreign policy outcomes can be improved by isolating the foreign policy position of leaders in democracies. To fully contextualize my theory, I consider the extant literature first which allows me to accurately place my approach in the broader constellation of the previous scholarship. Following this, I provide a brief overview of salience theory. From here, a full explanation of the issue emphasis approach to foreign policy is provided. Next, an empirical test is outlined and the substantive results are interpreted. The final section concludes and offers several thoughts on future research. Context The most important set of extant literature that this piece addresses examine whether leaders are actually able to translate policy positions into reality. This research has tended to focus on how and when leaders in democracies are constrained. One strand of this work comes from the research on SCA, and suggests that potential target states will avoid leaders that are domestically vulnerable (Leeds and Davis 1997; Fordham 2005). In other words, states may deny democratic leaders the opportunity to divert from a declining domestic political situation or otherwise use force when they would prefer to. This debate surrounding diversionary theory and SCA is important to the extent that it constitutes one of the key divides within the domestic determinants of foreign policy literature. More specifically, the SCA literature is important to the approach taken here because the general SCA logic suggests that leaders may struggle to live up to their campaign promises if other states selectively avoid hawks and target doves. Another portion of significant scholarship has examined the role of institutional support on foreign policy. Those interested in the US case have turned toward the effects of Congress on the use of force and the likelihood of diverting. Clark (2000) theorizes that as the preferences of the president and Congress converge, conflicts become more likely and tend to last longer. Howell and Pevehouse (2005, 2007) show that interbranch politics are a key component of the equation.1 The frequency with which the president uses force depends on his level of partisan support in Congress. In the same realm of study, Brulé (2006) examines the interactive effect of the economy and Congress, showing that presidents are more likely to divert from a flagging economy when they face a hostile Congress. Carter and Scott (2009) examine a wide variety of foreign policy outcomes and show that individual members of Congress often take a very active role in shaping US foreign policy, often to the displeasure of the executive. Altogether, this literature suggests institutional support is a key component of this research agenda. This work is important to the argument made below because I claim leaders live up to their campaign promises, and this work suggests institutional constraints may keep leaders from doing so. Some new cross-national research highlights the role of government characteristics in accounting for international conflict. Auerswald (1999) argues that because leaders in democracies are vulnerable to elections, votes of no confidence, or both, we can expect institutional context to matter in international crises.2Brulé and Williams (2009) have investigated government characteristics (e.g., minority versus coalition governments, weak versus strong party discipline), economic downturns, and dispute initiation cross-nationally, finding that the level of government accountability influences the likelihood of diversionary conflict. The work of Kaarbo and Beasley (2008) further suggests that coalition governments featuring extremist parties tends to produce extreme (either peaceful or violent) foreign policy behavior. Furthermore, Clare (2010) studies the role of ideology as mediated by political institutions, arguing that when extremist parties become junior members of a coalition they can exercise disproportionate influence on foreign policy, with leftist outliers having a more pacific effect and rightist outliers leading to more aggressive behavior. The perspective provided by the contemporary literature suggests the foreign policy behavior of leaders in democracies is strongly constrained. It may be that domestic institutional support limits foreign policy options, or it may be that other states simply avoid hawkish leaders altogether. Without discounting the possibility of SCA or the importance of domestic political constraints, I argue here that the foreign policy positions of leaders still has a significant effect on outcomes abroad after taking these constraints on behavior into account. I suggest that we can get a better idea of leaders’ foreign policy positions by disaggregating ideology and isolating the foreign policy position of executives in democracies. Some formal work suggests hawkish leaders in democracy are actually more likely to secure long-term peace (Cowen and Sutter 1998; Cukierman and Tommasi 1998). The basic point here is that leaders in democracies may implement policies counter to their ideology in order to help secure reelection. Further formal work by Schultz (2005) delineates the scope and conditions under which we can expect hawks to truly deliver lasting peace. The approach presented here speaks directly to this literature by disaggregating ideology into specific policy components to identify hawks and doves. Moreover, my approach analyzes broad patterns of peace and conflict in democracies in order to see what types of leaders actually deliver lasting peace. Alongside research on institutional constraints and SCA, the literature on how leaders divert from poor domestic conditions and low public approval ratings has developed extensively (Ostrom and Job 1986; James and Oneal 1991).3 A number of skeptics challenge the logic of diversionary theory, arguing that we can explain foreign policy by accounting for international conditions (James and Hristoulas 1994; DeRouen 1995; Meernik and Waterman 1996; Gowa 1998). Following Hibbs’ (1977) logic concerning the economic preferences of Democratic and Republican presidents, Fordham (1998) finds that Democratic presidents are more likely to use force when facing inflation, while Republican presidents have used force more often in the face of high unemployment. Importantly, much of this early work on partisanship and diversionary theory considers only the party of the leader, as opposed to treating partisanship or ideology as continuous concepts. Finally, some authors have considered policy positions and continuous measures of ideology in the aggregate regarding interstate conflict. Palmer, London, and Regan (2004) have considered government position, showing that governments of the left in parliamentary democracies are less likely to engage in international disputes than governments of the right, but more likely to escalate disputes once engaged because they are more sensitive to removal from office in the face of international conflict. Recent cross-national research has also shown that right-leaning governments are willing to prosecute interstate disputes longer on average than left-leaning governments (Koch 2009). Arena and Palmer (2009) find that governments that lean to the right on their one-dimensional scale of ideology are more likely to initiate a militarized interstate dispute (MID), while left-leaning governments are more peaceful. These cross-national studies measure the positions of political elites using aggregate measures of ideology. To this point, there has been no attempt to measure the foreign policy position of the executive in the study of the domestic determinants of conflict. While some pieces treat partisanship dichotomously (e.g., Fordham 1998) and others employ continuous measures of government ideology, one of the key contributions made here is in isolating the foreign policy preferences of leaders by disaggregating ideology. Accounting for Policy Positions In order to better understand why certain leaders in democracies are more likely to initiate conflict than others, we should make some effort to consider their general position on foreign policy issues. That is, we must measure how hawkish or dovish leaders present themselves in order to better predict behavior. Salience theory provides a foundation to help explain how we might measure such a concept prior to an executive taking office and why issue emphasis should matter. Budge and Farlie (1983) introduced the issue emphasis approach, suggesting that parties attempt to emphasize salient issues that favor their strengths relative to their opponents.4 Ideology is itself a basket of issue positions and these issue positions change over time. By allowing ideology and issue emphasis to vary across a number of dimensions, we can see clearer, more detailed images of parties and their candidates. From here, we can isolate the issues of interest and analyze only those policy promises. Because this analysis focuses on foreign policy outcomes, it is important to isolate the foreign policy position of the PM/president in democracies. Recent research by Whitten and Williams (2011) show why this is necessary.5 Klingemann, Hofferbert, and Budge. (1994) build on this approach by introducing the salience theory of party competition. Salience theory as envisioned by these authors is not entirely distinct from Budge and Farlie’s (1983) approach. Their interest lies specifically with political parties and the issues they choose to emphasize. To quote Klingemann, Hofferbert, and Budge., “By stressing certain items and excluding others—without overtly denouncing the latter—parties are, to be sure, implicitly taking pro and anti positions” (25). Through this process, parties put together the package of policies that they intend to offer the voters by emphasizing a set of issues. Extant work has tapped such emphases by analyzing party manifestos (e.g., Budge 1994). While some skeptics might offer that voters are unaware of party manifestos or party stances, Klingemann, Hofferbert, and Budge. (1994) argue that manifestos have an impact. To quote their argument directly, “In most democracies, the manifesto or platform is launched at a press conference, with great publicity, designed to set the major themes for the whole election. It is through the appearance of programmatic themes[…] in the media that the document makes it main impact on electors” (21). Moreover, we can certainly imagine that competing parties loudly publicize aspects of their competitors’ policy platforms that they expect to be unpopular.6,7 As the general issue emphasis approach posited by Budge and Farlie (1983) suggests, politicians highlight their strengths to their opponents’ detriment as best they can. Along somewhat similar lines as the above noted authors, Petrocik (1996) argues that political parties in the United States own certain issues and that the ownership of issues is relatively static over time. McDonald and Budge (2005) agree that parties can be expected to emphasize the same set of issues over time (e.g., we would expect Republicans to consistently emphasize the need for a strong national defense relative to Democrats in the United States). So while parties typically do not leapfrog one another on issues, the extent to which they emphasize different issues varies over time (Budge 1994). This variation over time is critical to my understanding of how leaders and governments make foreign policy. It may be perfectly reasonable to theorize that Democratic presidents in the United States or Socialist presidents in France behave as doves relative to their rightist competition, but it is also vital to account for the variation between different Democratic and Socialist presidents. Theory This piece applies salience theory to foreign policy behavior. I conceive of policy positions and their role in policymaking in a manner consistent with the extant literature. Parties and candidates in all democracies broadly emphasize some set of policies to voters, implicitly taking policy positions. Even in states that feature relatively short campaign periods, parties are expected to provide some sort of policy platform that covers a variety of issues. Once in office, politicians will make some effort to follow up on these policy positions, though countervailing forces may demand compromise. Of course significant events beyond the control of politicians may demand leaders and governments shift policy away from platform promises. However, I assume these significant exogenous events are not the norm. The implication is that we can still expect policy positions to affect policy outcomes on average. As noted above, my particular interest here is the behavior of states abroad. Foreign aid generosity and defense spending can be tied to foreign policy positions, but these correlations have already been well mapped by Budge and Hofferbert (1990) in the American case and by McDonald and Budge (2005) in a cross-national analysis. This piece approaches foreign policy issue emphasis by examining conflict behavior. By studying involvement in international disputes, this piece analyzes a set of outcomes that have not previously been studied in terms of disaggregated policy positions. I analyze conflict from the perspective of leaders, but this approach does not take other elite behavior for granted. A rich literature that shows how junior coalition partners (e.g., Kaarbo 1996; Clare 2010; Oppermann and Brummer 2014) and other sources of institutional support (e.g., Howell and Pevehouse on the US Congress) can influence foreign policy decisions, including interstate disputes. The contention I make here is not that divided or coalition governments have no role to play in the decision-making process. Any empirical test of the domestic determinants of conflict should account for these factors. The theoretical approach here suggests that leaders have an effect on foreign policy outcomes after controlling for sources of institutional support. Ultimately, it is beyond the scope of this paper to test the proposition that the executive dominates foreign policy across different democracies. The argument here is not incompatible with this perspective that elites outside of the executive matter. Indeed, it is the strength of domestic political constraints in democracies that make further investigation of whether hawkish leaders engage in more conflict a worthwhile subject for analysis. If we assume the PM or president has at least some influence over foreign policy, we can see how their foreign policy promises may matter. The translation between foreign policy issue emphasis and the initiation of interstate disputes is clear. The candidate or party who wins control of the executive position in a democracy has clear incentives to follow up on the issues that they have chosen to emphasize. Signaling a hawkish foreign policy orientation represents a type of “tying of the hands” as Fearon (1997) puts it. Leaders who promise to act tough will suffer domestic audience costs if they fail to live up to their promises. These domestic audience costs can be conceptualized as lost votes during a bid for reelection or reputation costs suffered at home. Though Fearon focuses on leaders tying their hands in order to demonstrate resolve to other leaders, this logic should hold true for leaders who outline a pacific foreign policy portfolio during the election campaign. A leader who promises to keep a state out of conflicts and peacefully negotiate in the face of international tension may be punished by voters if he or she engages in conflict unnecessarily. These potential domestic costs as outlined by Fearon should make messages sent during campaigns credible. I also suggest that while parties and leaders emphasize issues that they believe they enjoy a comparative advantage in, they also follow through by shaping policy to match their rhetoric in order to appear responsive to voters. Of course, not all candidates run for reelection, but I suggest that words should match deeds ceteris paribus, because campaign promises reveal in part the preferences of elected leaders on different policy issues. Though parties and elites vary their policy positions to remain competitive, I also suggest here that democratic candidates reveal their preferences for handling foreign policy during campaigns. Messages communicated by candidates are comprised of both strategic vote-chasing elements and personal preferences on individual policy issues according to this line of thinking.8 A concrete example should be helpful in outlining the logic of the theory. Sensing a weakness in incumbent Jimmy Carter’s foreign policy, challenger Ronald Reagan pounced during the 1980 election, outlining a more aggressive foreign policy. As Petrocik (1996) shows, Reagan recognized that the public perceived Republicans as owning national defense, and he sought to take advantage of the issue by offering a militaristic policy alternative to Carter. Interestingly, Reagan actually implemented his foreign policy as advertised. In his first term in office, Reagan initiated twenty-two major international uses of force, contrasted by six uses of major force under the Carter administration.9 At first blush, the wisdom of hawks and doves as Republicans and Democrats holds, but when we note that Dwight Eisenhower opted to use force only twenty-seven times during his entire two terms as president, things become less clear. Reagan and Eisenhower were both Republicans, but Eisenhower and Republicans of his day offered a different and more pacific foreign policy stance than did Reagan and the Republicans of the 1980s. This distinction is an important one and needs to be studied beyond the American example. It is also important to note that my argument assumes foreign policy orientations can be considered distinct from other policy positions (e.g., positions on government spending, social justice, and environmental protection). This line of thinking echoes Wildavsky’s (1966) argument about the dual presidency in the United States, and the argument here suggests all leaders have both domestic and foreign policy orientations that are distinct from one another. I argue hawkish leaders initiate militarized disputes more frequently, while dovish leaders abstain. There are, however, some alternative arguments about how potential target states may shy away from contact with democratic leaders who are likely to engage in conflict. In particular, the SCA perspective suggests states avoid politically vulnerable leaders. In his analysis of diversionary theory, Smith (1996) posits this sort of argument. Smith suggests that diverting is unlikely to occur because potential target states strategically avoid conflict with leaders who may look to divert. There is some mixed evidence for the general SCA argument. For example, Leeds and Davis (1997) show rivals avoid provocative behavior when a rival democracy is experiencing slow growth or approaching an election. Clark (2003) further shows that democracy promotes diversionary incentives while simultaneously making it difficult to act on such impulses. Examining only the United States, Fordham (2005) finds qualified support for this approach, showing that rival states are less likely to publicly challenge the United States when growth is low or inflation is high. Transferring the basic logic of SCA to the study of hawks and doves in democracies, it may be that rival states avoid hawkish leaders who are likely to lash out if targeted. This would imply that dovish leaders are actually engaged in relatively more conflicts. It may even be that dovish leaders are targeted specifically because other leaders view them as weak. The overall effect would be that we can expect foreign policy in democracies to be the opposite of what leaders campaign on. This is a distinct possibility and one that runs counter to my proposition. In other words, the logic of SCA relating to ideology and partisanship provides another good reason to test this argument about the words and deeds of foreign policy in democracies.10 From the perspective presented here, foreign policy tensions may arise for a variety of reasons (e.g., disagreements about fishing rights between Iceland and the United Kingdom in the North Atlantic). The threat or use of military force in these disputes is directly related to the foreign policy position of the leaders in question. Potential target states may want to avoid violence, but hawkish leaders in democracies should be more likely to use force regardless of the target state’s behavior or preferences. Though the foreign policy position approach and the SCA approach both posit plausible explanations about conflict in democracies, their empirical expectations run in opposite directions. From this theoretical argument I derive a testable hypothesis: H: Leaders from parties that emphasize militaristic foreign policy concepts will initiate international militarized conflict more frequently. Research Design I test the issue emphasis approach to foreign policy on a sample of twenty-five democracies from 1951 through 2000. Table 1 describes the cross-national and temporal characteristics of the democracies in the sample. The unit of analysis is the directed dyad quarter. I use quarterly data because many parliamentary systems change governments mid-year. Annual data might otherwise attribute conflicts to the wrong government. The sample of directed dyadic observations includes all politically relevant dyads in all models. I use directed dyad data because it is important to distinguish between MIDs initiated by the democracies in question and those disputes that target the democracies in the sample. In the tests examining MID initiation (Table 2), I consider all those directed dyads where the democracies in the sample may initiate a dispute against another state (e.g., the United States attacking China). In the models examining whether other states target democracies (Table 3), I examine only directed dyadic observations where the democracies in the sample constitute a potential target (e.g., China attacking the United States). Table 1. Sample countries   Sample  Australia  1951–2000  Austria  1955–2000  Belgium  1951–2000  Canada  1953–2000  Denmark  1951–2000  Finland  1951–2000  France  1951–2000  Germany  1971–2000  Greece  1975–2000  Iceland  1951–2000  Ireland  1951–2000  Israel  1951–2000  Italy  1951–2000  Japan  1961–2000  Luxembourg  1951–2000  Netherlands  1951–2000  New Zealand  1951–2000  Norway  1952–2000  Portugal  1976–1978; 1983–2000  Spain  1977–2000  Sweden  1951–2000  Switzerland  1951–2000  Turkey  1951–1971;1973–1979;1983–2000  United Kingdom  1951–2000  United States  1951–2000    Sample  Australia  1951–2000  Austria  1955–2000  Belgium  1951–2000  Canada  1953–2000  Denmark  1951–2000  Finland  1951–2000  France  1951–2000  Germany  1971–2000  Greece  1975–2000  Iceland  1951–2000  Ireland  1951–2000  Israel  1951–2000  Italy  1951–2000  Japan  1961–2000  Luxembourg  1951–2000  Netherlands  1951–2000  New Zealand  1951–2000  Norway  1952–2000  Portugal  1976–1978; 1983–2000  Spain  1977–2000  Sweden  1951–2000  Switzerland  1951–2000  Turkey  1951–1971;1973–1979;1983–2000  United Kingdom  1951–2000  United States  1951–2000  Notes: The start date of the data is constricted by the availability of GDP growth data from the Penn World Table (Germany). Start dates for Greece, Portugal, Spain, Turkey, and Japan are determined by the availability of CMP and government data, as these states experienced democratic transitions during this time period. The end date is determined by the availability of alliance portfolio data. All exclude caretaker governments. View Large Table 2. Foreign policy issue emphasis and conflict initiation, 1951–2000   (1)  (2)  (3)  (4)    All MIDs  All MIDs  Hostile MIDs  Fatal MIDs  Militarism index  0.040***  0.179***  0.043***  0.055***  (0.011)  (0.069)  (0.015)  (0.017)  Relative capability  10.031***  10.096***  8.152***  10.335***  (3.727)  (3.744)  (2.578)  (2.921)  Polity of target  −0.084***  −0.085***  −0.104***  −0.135***  (0.014)  (0.014)  (0.012)  (0.017)  Alliance portfolio  −0.327  −0.319  −0.042  −0.752***  (0.308)  (0.314)  (0.236)  (0.260)  Distance  −0.227***  −0.228***  −0.275***  −0.248***  (0.025)  (0.025)  (0.020)  (0.021)  Other initiated disputes  1.191***  1.205***  0.961***  1.503***  (0.139)  (0.138)  (0.173)  (0.186)  GDP growth  −0.008  −0.004  0.011  0.030  (0.014)  (0.014)  (0.022)  (0.023)  CIEP (% left)  0.238  0.224  0.222  0.653**  (0.181)  (0.180)  (0.273)  (0.302)  Single-party majority  −0.189  −0.232  −0.341  0.213  (0.179)  (0.179)  (0.221)  (0.243)  Coalition  −0.402**  −0.425**  −0.312  0.091  (0.193)  (0.195)  (0.208)  (0.238)  Coalition ideological distance  0.020*  0.022*  0.039***  0.048***  (0.011)  (0.011)  (0.010)  (0.011)  Executive left–right  0.002  −0.002  0.005  0.001  (0.004)  (0.004)  (0.004)  (0.004)  Peace quarters  −0.074***  −0.075***  −0.077***  −0.060***  (0.008)  (0.008)  (0.010)  (0.011)  Peace quarters-squared  0.001***  0.001***  0.001***  0.000**  (0.000)  (0.000)  (0.000)  (0.000)  Peace quarters-cubed  −0.000***  −0.000***  −0.000***  −0.000  (0.000)  (0.000)  (0.000)  (0.000)  Constant  −9.796***  −9.805***  −8.785***  −11.706***  (2.637)  (2.654)  (1.934)  (2.230)  Log likelihood  −1,950.903  −1,951.153  –  –  Observations  112,718  112,718  112,718  112,718  Standardized IV  No  Yes  No  No  Rare events estimator  No  No  Yes  Yes    (1)  (2)  (3)  (4)    All MIDs  All MIDs  Hostile MIDs  Fatal MIDs  Militarism index  0.040***  0.179***  0.043***  0.055***  (0.011)  (0.069)  (0.015)  (0.017)  Relative capability  10.031***  10.096***  8.152***  10.335***  (3.727)  (3.744)  (2.578)  (2.921)  Polity of target  −0.084***  −0.085***  −0.104***  −0.135***  (0.014)  (0.014)  (0.012)  (0.017)  Alliance portfolio  −0.327  −0.319  −0.042  −0.752***  (0.308)  (0.314)  (0.236)  (0.260)  Distance  −0.227***  −0.228***  −0.275***  −0.248***  (0.025)  (0.025)  (0.020)  (0.021)  Other initiated disputes  1.191***  1.205***  0.961***  1.503***  (0.139)  (0.138)  (0.173)  (0.186)  GDP growth  −0.008  −0.004  0.011  0.030  (0.014)  (0.014)  (0.022)  (0.023)  CIEP (% left)  0.238  0.224  0.222  0.653**  (0.181)  (0.180)  (0.273)  (0.302)  Single-party majority  −0.189  −0.232  −0.341  0.213  (0.179)  (0.179)  (0.221)  (0.243)  Coalition  −0.402**  −0.425**  −0.312  0.091  (0.193)  (0.195)  (0.208)  (0.238)  Coalition ideological distance  0.020*  0.022*  0.039***  0.048***  (0.011)  (0.011)  (0.010)  (0.011)  Executive left–right  0.002  −0.002  0.005  0.001  (0.004)  (0.004)  (0.004)  (0.004)  Peace quarters  −0.074***  −0.075***  −0.077***  −0.060***  (0.008)  (0.008)  (0.010)  (0.011)  Peace quarters-squared  0.001***  0.001***  0.001***  0.000**  (0.000)  (0.000)  (0.000)  (0.000)  Peace quarters-cubed  −0.000***  −0.000***  −0.000***  −0.000  (0.000)  (0.000)  (0.000)  (0.000)  Constant  −9.796***  −9.805***  −8.785***  −11.706***  (2.637)  (2.654)  (1.934)  (2.230)  Log likelihood  −1,950.903  −1,951.153  –  –  Observations  112,718  112,718  112,718  112,718  Standardized IV  No  Yes  No  No  Rare events estimator  No  No  Yes  Yes  Robust standard errors in parentheses. *p<0.1, **p<0.05, ***p  <0.01; two-tailed tests. View Large Table 3. Foreign policy issue emphasis and targeted MIDs, 1951–2000   (5)  (6)  (7)  (8)    All MIDs  All MIDs  Hostile MIDs  Fatal MIDs  Militarism index  0.042***  0.072  0.047***  0.056***  (0.014)  (0.062)  (0.017)  (0.020)  Relative capability  8.563***  8.233***  9.835***  11.491***  (2.135)  (2.041)  (1.564)  (2.129)  Polity of initiator  −0.076***  0.017  −0.103***  −0.142***  (0.013)  (0.038)  (0.012)  (0.021)  Alliance portfolio  0.087  −0.317  0.203  −0.092  (0.377)  (0.391)  (0.223)  (0.269)  Distance  −0.186***  −0.181***  −0.216***  −0.233***  (0.028)  (0.030)  (0.019)  (0.023)  Other initiated disputes  2.473***  2.797***  2.146***  2.738***  (0.197)  (0.198)  (0.170)  (0.193)  GDP growth  −0.003  −0.000  0.003  0.000  (0.016)  (0.017)  (0.022)  (0.026)  CIEP (% left)  0.017  0.046  0.026  0.329  (0.171)  (0.169)  (0.231)  (0.278)  Single-party majority  0.017  0.010  0.077  0.084  (0.154)  (0.165)  (0.162)  (0.227)  Coalition  −0.242  0.194  −0.465**  0.010  (0.186)  (0.208)  (0.208)  (0.253)  Coalition ideological distance  0.022*  0.021  0.044***  0.059***  (0.012)  (0.014)  (0.011)  (0.010)  Executive left–right  −0.003  −0.002  −0.002  0.001  (0.004)  (0.004)  (0.004)  (0.004)  Peace quarters  −0.072**  −0.074***  −0.073***  −0.073***  (0.009)  (0.010)  (0.010)  (0.012)  Peace quarters-squared  0.001***  0.001***  0.001***  0.001***  (0.000)  (0.000)  (0.000)  (0.000)  Peace quarters-cubed  −0.000***  −0.000***  −0.000**  −0.000  (0.000)  (0.000)  (0.000)  (0.000)  Constant  −9.278***  −9.037***  −10.380***  −12.600***  (1.590)  (1.484)  (1.184)  (1.605)  Log likelihood  −2,047.825  −2,146.654  –  –  Observations  112,919  112,919  112,919  112,919  Standardized IV  No  Yes  No  No  Rare events estimator  No  No  Yes  Yes    (5)  (6)  (7)  (8)    All MIDs  All MIDs  Hostile MIDs  Fatal MIDs  Militarism index  0.042***  0.072  0.047***  0.056***  (0.014)  (0.062)  (0.017)  (0.020)  Relative capability  8.563***  8.233***  9.835***  11.491***  (2.135)  (2.041)  (1.564)  (2.129)  Polity of initiator  −0.076***  0.017  −0.103***  −0.142***  (0.013)  (0.038)  (0.012)  (0.021)  Alliance portfolio  0.087  −0.317  0.203  −0.092  (0.377)  (0.391)  (0.223)  (0.269)  Distance  −0.186***  −0.181***  −0.216***  −0.233***  (0.028)  (0.030)  (0.019)  (0.023)  Other initiated disputes  2.473***  2.797***  2.146***  2.738***  (0.197)  (0.198)  (0.170)  (0.193)  GDP growth  −0.003  −0.000  0.003  0.000  (0.016)  (0.017)  (0.022)  (0.026)  CIEP (% left)  0.017  0.046  0.026  0.329  (0.171)  (0.169)  (0.231)  (0.278)  Single-party majority  0.017  0.010  0.077  0.084  (0.154)  (0.165)  (0.162)  (0.227)  Coalition  −0.242  0.194  −0.465**  0.010  (0.186)  (0.208)  (0.208)  (0.253)  Coalition ideological distance  0.022*  0.021  0.044***  0.059***  (0.012)  (0.014)  (0.011)  (0.010)  Executive left–right  −0.003  −0.002  −0.002  0.001  (0.004)  (0.004)  (0.004)  (0.004)  Peace quarters  −0.072**  −0.074***  −0.073***  −0.073***  (0.009)  (0.010)  (0.010)  (0.012)  Peace quarters-squared  0.001***  0.001***  0.001***  0.001***  (0.000)  (0.000)  (0.000)  (0.000)  Peace quarters-cubed  −0.000***  −0.000***  −0.000**  −0.000  (0.000)  (0.000)  (0.000)  (0.000)  Constant  −9.278***  −9.037***  −10.380***  −12.600***  (1.590)  (1.484)  (1.184)  (1.605)  Log likelihood  −2,047.825  −2,146.654  –  –  Observations  112,919  112,919  112,919  112,919  Standardized IV  No  Yes  No  No  Rare events estimator  No  No  Yes  Yes  Robust standard errors in parentheses. *p<0.1, **p<0.05, ***p  <0.01; two-tailed tests. View Large I test my hypothesis on three different dependent variables to examine conflict behavior in democracies. The measures for the dependent variables come from the Correlates of War Militarized Interstate Dispute 3.1 data (Ghosn, Palmer, and Bremer 2004).11 I first examine a binary dependent variable that is equal to one in each observation if the democracy in question initiated any MID and zero otherwise. The argument for using this broad indicator is that the foreign policy position of leaders matter for foreign policy in a general way, so an analysis of all MIDs makes theoretical sense. I also examine a narrower set of dependent variables that includes only those MIDs that lead to the use of force (Hostile MIDs), and a final measure of only the initiation of MIDs leading to at least one fatality (Fatal MIDs). These dependent variables are analyzed to make sure the foreign policy position of leaders influences salient events that are likely to capture the public’s attention.12 The empirical approach here not only tests the hypothesis posited above, but explicitly examines whether dovish leaders are more likely to be targeted by other states. This is part of the SCA perspective discussed above. If other states avoid hawkish leaders, we should expect hawkishness to be negatively associated with conflict. Though I offer no explicit hypotheses about whether other states might target dovish leaders more frequently than hawkish leaders, this possibility derived from the SCA perspective is worth investigating given its implications. If dovish leaders are in fact subject to challenges from abroad, such a finding would indicate a universal incentive for democratic leaders to take hawkish positions. I examine the same three dependent variables described above in a set of models that examine whether democratic leaders are targeted by other states on the basis of their foreign policy position (Table 3). These models include all the covariates described below, and, as noted above, these models examine only directed-dyad observations where the democracies from the sample may be targeted. I expect the foreign policy position of the leader to have an impact on foreign policy outcomes, and I measure the foreign policy position of leaders by looking at emphasis of different foreign policy concepts using data from the Comparative Manifesto Project (Volkens et al. 2013). The CMP performs content analysis of party manifestos and codes emphasis of individual issues (e.g., positive statements about free enterprise or positive statements about the welfare state) as a proportion of the entire manifesto document for each party in a given election. I use individual measures of foreign policy emphasis instead of aggregate measures of a government’s right–left position. The theoretical argument focuses on the foreign policy position of the presidential or prime minister’s party and not on the ideological orientation of the government as a whole. I make no particular predictions about the nature of leftist or rightist parties in government. I consider measures of emphasis on three different foreign policy concepts and use these to create an index as the independent variable. I first consider positive militarism statements.13 These include statements that emphasize self-defense, the need to improve the state’s military strength, and the importance of external security among other similar topics. A measure of negative military emphasis is coded the same as the pro-military variable, but counts statements that oppose pro-military topics. A third variable measures emphasis of pacifism and relies on statements that mention peace as a general goal and declarations that peaceful means should be used in solving disputes. I create a basic Militarism Index by subtracting the sum of the pacifism and anti-militarism variables from the value of the positive militarism variable.14 The intuitive meaning of this process is that I capture the proportion of statements referring to the concepts of interest in each manifesto and calculate an index where higher numbers should indicate a more hawkish position. Figure 1 displays the values of the Militarism Index of the major US political parties since World War II. The story this figure tells indicates that the measure has some level of face validity. The parties provide divergent policy choices for the most part, and Republicans appear more aggressive than Democrats in most years. Of specific interest is the 1992 election, in which Bill Clinton and the Democrats actually leapfrogged incumbent George H.W. Bush and the Republicans. This switch may actually be in line with standard thinking about the 1992 election. H.W. Bush is widely perceived as having moderated his tone in the wake of the Persian Gulf War, while Clinton pushed the Democrats further to the right on a number of issues, including foreign policy (Volkens et al. 2013). Figure 1. View largeDownload slide Net emphasis of militarism over time by Major Parties in the United States, 1948–2008 Figure 1. View largeDownload slide Net emphasis of militarism over time by Major Parties in the United States, 1948–2008 More importantly, variation over time between and within parties clearly exists. Some of this variation can be explained via the logic of Nincic (1988) work, who suggests the US public may reward traditionally dovish leaders for acting tough and hawks for moderating their behavior. With regard to the US context, the elections in the 1990s may have represented a strategic move by the Clinton Democrats to gain votes by taking a hawkish position with the expectation that voters would reward such behavior. Variation in policy position can be partially understood as an attempt to gain votes, and the goal here is to see how this strategic maneuvering translates to policy. In general, what we see is that neither party has occupied a static foreign policy position since the 1940s, and the variation displayed above is theoretically important. I include a set of control variables that capture the strategic dyadic relationship in each observation as well some measures of domestic political conditions for each democracy in the sample. Relative Capability, originally developed by Singer (1988), is an indicator of relative power. This measure is coded as the log of the ratio of State A’s CINC score to the sum of State A and State B’s CINC scores, plus one (Singer 1988). I also consider the regime type of potential targets using data from Polity IV (Marshall and Jaggers 2002). In models examining initiation (Table 2), the Polity of Target indicator represents the combined polity2 measure of democracy from Marshall and Jaggers’ data. In models examining whether states target democracies (Table 3), I include the Polity of Initiator variable. A measure of Distance is included, which measures the natural log of distance in miles between state capitals. Finally, the weighted global S score calculated by Signorino and Ritter (1999), which measures alliance portfolio similarity, is included.15 A final variable that helps account for opportunity is the Other Initiated Disputes measure. This is a dichotomous measure that is equal to one if the democracy in question has initiated a MID in another dyad during the same quarter and zero otherwise. The reasoning here is that states have limited military capabilities, and using force against any one target may preclude the use of force in another dyad, even if a leader is relatively hawkish. The model includes two control variables that may constitute incentives for democratic leaders to divert or incentives for potential target states to avoid contact with vulnerable democratic leaders. The first of these is a measure of GDP per capita growth in constant US dollars. This variable is labeled GDP Growth. As Downs and Rocke (1995) argue, leaders with little hope of reelection may gamble for resurrection by extending or escalating conflicts. Because declining economic conditions may represent such an incentive to gamble, the GDP growth indicator represents a necessary inclusion here. It may also be that declining economic conditions have a constraining effect on conflict involvement (Arena and Palmer 2009). This measure comes from the Penn World Tables (Heston, Summers, and Aten 2012). The second measure from the diversionary theory/SCA literature is an indicator of the proportion of the constitutional interelection period remaining for each democracy during each quarter. This is a measure of time remaining until a constitutionally mandated election. This measure, constitutional inter-election period (CIEP), comes from Seki and Williams’ (2014) update of Woldendorp, Keman, and Budge’s dataset on government duration in democracies. As discussed above, institutional factors may empower or constrain a leader. To account for this empirically, I include a set of binary indicators that measure whether a government is a coalition or majority government, with minority governments representing the excluded category (Woldendorp, Keman, and Budge 2000). In the case of presidential (the United States) or semi-presidential (France) systems, governments are coded as minority governments unless they enjoy majority support in all houses of the legislature, in which case they are coded a single-party majority. This coding practice is adopted from Powell’s (2000) approach. These measures are labeled Single-Party Majority and Coalition, respectively. Following Clare (2010), I exclude all caretaker governments in parliamentary democracies. Because the ideological position of vital junior coalition partners matters in parliamentary democracies (Kaarbo 1996), the model includes a measure called Coalition Ideological Distance that captures the distance between the PM in parliamentary democracies and the other governing parties. It may also be that hawkish parties in parliamentary systems form coalitions with other hawkish parties, so including this measure makes it possible to estimate the effect of the leader’s foreign policy orientation while accounting for junior coalition partners. To calculate this, I take the difference between the executive party’s aggregate ideological position and the government as a whole’s ideological position using Comparative Manifesto data (Volkens et al. 2013). Positive values of this variable indicate more conservative coalition partners while negative numbers indicate the presence of junior coalition partners to the left of the PM. This measure is always zero in presidential systems that do not feature coalition governments. Finally, the model includes a measure of the executive party’s left–right position as calculated using the normal CMP equation with all foreign policy variables purged (Volkens et al. 2013). The remaining value proxies the leader’s policy positions on domestic issues (e.g., the welfare state, taxes, expansion of public education, the role of religion in society, etc.). A higher value of this purged measure indicates a more conservative or rightist aggregate position, and this variable is labeled Executive Left–Right. Including this covariate provides evidence that the Militarism Index has an effect when controlling for the executive’s position on other policy issues. I use logistic regression to estimate all models with robust standard errors clustered on the dyad. To handle temporal dependence in the dependent variables, I follow Carter and Signorino’s (2010) method and include a count of peace quarters along with its squared and cubed terms for all dependent variables. To estimate initiation of hostile and fatal MIDs, I employ a corrected rare events logistic estimator because the initiation of these sorts of disputes is relatively rare (King and Zeng 2001). Finally, an alternative measure of the independent variable where the Militarism Index is divided by the standard deviation of each country’s party system Militarism Index during the most recent election is used as a robustness check on the main findings. This alternative provides some indication of how hawkish each leader is relative to their domestic competition. In these models (Models 2 and 6), the standardized Militarism Index is meant to account for country-election specific factors and the general foreign policy environment in each country during each election.16 Analysis The findings support the hypothesis. Political parties and party leaders in democracies signal their foreign policy positions prior to election by selectively emphasizing concepts in their party manifestos. Table 2 presents the results of the empirical test looking at MID initiation across different dependent variables. The results from Models 1–4 provide general support for the proposition about leaders and interstate conflict. In Model 1, we see hawkish leaders are more likely to initiate any sort of dispute. This finding is replicated in Models 3 and 4, which look at hostile and fatal MIDs, respectively. Here, we see again that hawkish leaders are more likely to initiate these sorts of militarized conflicts. The results of the models with the standardized independent variable are displayed in Model 2. The results are consistent with prior findings.17 The coefficient associated with the Militarism Index is still positive and significant. The coefficient is larger in Model 2 than others, but this is only due to the fact that the standardized independent variable scales differently than the regular Militarism Index. The control variables generally behave as expected. The Relative Capability measure is positive and significant across models, suggesting that democracies are more likely to target relatively weaker states. Democracies are less likely to target other democracies as the negative and significant coefficients associated with Polity of Target indicate across Models 1–4. Distance is negative and significant in all models, meaning conflict is less likely between states that are far apart. This effect is consistent across models. The Alliance Portfolio variable is not significant in any model except Model 4, where is it negative as expected. The coefficient for Other Initiated Disputes is surprisingly positive and significant across models, meaning that democracies are actually likely to initiate more disputes once they have initiated one. Turning to the domestic political covariates, the measure of GDP Growth is not significant in any model. CIEP is positive and significant when estimating initiation of disputes that lead to at least one fatality, but not otherwise significant. This indicates that democracies are more likely to initiate salient disputes when elections are relatively far away. This is normatively pleasing as these results provide evidence that leaders do not divert on the basis of the economy, nor do these results suggest leaders in democracies begin disputes near elections to divert the public’s attention. The Coalition variable is negative and significant in Models 1 and 2, implying that coalition governments are less conflict prone than other forms of government. This means coalition governments may be more constrained in initiating disputes. However, this effect does not hold when looking at more salient types of disputes. Interestingly, majority governments appear to show no difference from minority governments. Altogether, the evidence here is mixed regarding institutional support. However, we should exercise care in making inferences here, as Palmer, London, and Regan (2004) suggest government type is a key factor in general conflict involvement for parliamentary democracies, regardless of who initiates or reciprocates a dispute. The measure of Coalition Ideological Distance is significant across models, suggesting more conservative coalition partners increase the odds of a state initiating a dispute that leads to at least one fatality. This finding is in line with the previous quantitative work of Kaarbo and Beasley (2008) and Clare (2010). The purged Executive Left–Right measure is not significant in any model, implying that the position of leaders on issues outside of foreign policy (e.g., taxes, crime, religion, etc.) does not affect interstate conflict.18 Recall the SCA perspective suggests other states may avoid hawkish leaders and target dovish leaders. The results of Models 5 through 8 in Table 3 suggest that if anything, more hawkish foreign policy positions lead states to challenge democracies more frequently. This finding challenges the specific notion that hawkish leaders are able to secure peaceful outcomes without at least threatening or using force. This is true of all disputes (Model 5), disputes leading to the use of force (Model 7), and disputes leading to fatalities (Model 8). However, this is not true in the model where the standardized IV is used (Model 6), meaning this finding is inconsistent when accounting for other domestic party positions. The Militarism Index is positive, but not statistically significant, meaning this effect is not robust across models. The control variables in these models generally behave as expected, with Relative Capability and Distance signed as expected and significant across models. Polity of the Initiator is negative as expected, but only significant in Models 6–8. The results also show that coalition governments are less likely to be targeted than other types of government but only in terms of forceful dispute initiations. The Other Initiated Disputes measure is positive and significant in all models, which is the same finding noted above. Interestingly, more conservative coalition partners also lead governments to be targeted more frequently according to the results regarding Coalition Ideological Distance, though again, this does not hold in Model 6. The measure of GDP Growth is not significant in any model, which suggests low growth in democracies does not deter challengers. The measure of the proportion of the CIEP left is not significant in any model, meaning democracies are not any more or less likely to be targeted at any level of hostility as elections draw nearer. These results challenge the SCA perspective described above that states actively avoid conflict with leaders who are likely to lash out (e.g., Smith 1996). These results also speak to the formal literature that examines whether dovish leaders are capable of creating a lasting peace. Some of this work (e.g., Cowen and Sutter 1998) has suggested that hawks are actually better at securing peace, but the broad empirical patterns outlined here do not support this position. Hawkish leaders in democracies use force more frequently after controlling for a broad array of factors, and other leaders do not appear to systematically avoid conflict with hawkish leaders. If anything, it would seem a hawkish foreign policy position may invite conflict. The conclusions drawn above should not be taken as an outright rejection of the logic behind SCA and diversionary theory. It is certainly possible that these theoretical arguments speak to behavior not perfectly captured by the measures employed here. MIDs are hardly a perfect measure of interstate conflict (Kang and Gibler 2012; Fordham and Sarver 2001). The evidence discussed above simply provides support for the argument that leaders’ foreign policy positions are frequently translated into actual foreign policy after accounting for a broad array of domestic and international constraints. This also does not mean other state leaders ignore the ideology of democratic leaders, as recent work by Clare (2014) shows dovish governments are able to leverage better deals out of rival governments. The other key implication here is that there is not a universal incentive for candidates to take a hawkish foreign policy position in order to deter challengers. Substantive Effects To provide a more intuitive display of the substantive effects of the findings across different simulated scenarios, Table 4 uses results from Model 1 and presents the change in probability and 95 percent confidence intervals associated with a one standard deviation (4.08) shift from the mean (0.58) in the Militarism Index variable in three different scenarios.19 Model 1 is used here as it represents the most general model of conflict and includes all militarized disputes. Specifically, these three simulations represent scenarios in which the risk of conflict is either low, moderate, or high based on different values of the control variables. In order to provide a point of comparison for the impact of the Militarism Index in each scenario, the table also provides the relative effect of a one standard deviation change from the mean of two geostrategic covariates. First the Relative Capability measure is increased in each simulation. Recall this measure is the natural log transformation of the proportion of capabilities that Side A possesses, so it does not vary as much as the other covariates but is still a strong predictor of conflict in Model 1. A decrease from the mean of Polity of Target by one standard deviation is also included. One point worth noting is that the odds of conflict in any dyad-quarter in any scenario are very low, so a small increase in the odds of conflict in all dyads aggregates to a larger overall probability of any conflict for each simulated democracy. Table 4. Change in predicted probability of MID initiation (Model 1)   Change in predicted probability        Low  Moderate  High  Baseline probability  .0001  .0006  .013    (.00005, .00018)  (.0004, .00097)  (.009, .018)  Militarism index  +.00002  +.0001  +.002  (.52→4.6)  (.000006, .00004)  (.00005, .00021)  (.001, .004)  Relative capability  +.00004  + .0003  +.006  (.69→.73)  (.00001, .00009)  (.00008,.00053)  (.0001, .011)  Polity of target  +.0002  +.0005  +.006  (1→ −7)  (.00001, .0003)  (.00035, .00093)  (.003, .009)    Change in predicted probability        Low  Moderate  High  Baseline probability  .0001  .0006  .013    (.00005, .00018)  (.0004, .00097)  (.009, .018)  Militarism index  +.00002  +.0001  +.002  (.52→4.6)  (.000006, .00004)  (.00005, .00021)  (.001, .004)  Relative capability  +.00004  + .0003  +.006  (.69→.73)  (.00001, .00009)  (.00008,.00053)  (.0001, .011)  Polity of target  +.0002  +.0005  +.006  (1→ −7)  (.00001, .0003)  (.00035, .00093)  (.003, .009)    Values of Variables in Scenarios        Low  Moderate  High  Militarism index  −1.3  0  1.8  Relative capability  .69  .69  .70  Polity of target  10  1  −7  Alliance portfolio  .57  .37  .16  Distance (ln of miles)  8.6  8.3  7.2  Other initiated disputes  0  0  0  GDP growth  3.69  2.54  .96  CIEP (% left)  .33  .56  .79  Single-party majority  0  0  1  Coalition  1  1  0  Coalition ideological distance  −1.19  1  0  Executive left–right  −9.1  4.5  18.97  Peace quarters  80  40  8  Peace quarters-squared  6,400  1,600  64  Peace quarters-cubed  512,000  64,000  512    Values of Variables in Scenarios        Low  Moderate  High  Militarism index  −1.3  0  1.8  Relative capability  .69  .69  .70  Polity of target  10  1  −7  Alliance portfolio  .57  .37  .16  Distance (ln of miles)  8.6  8.3  7.2  Other initiated disputes  0  0  0  GDP growth  3.69  2.54  .96  CIEP (% left)  .33  .56  .79  Single-party majority  0  0  1  Coalition  1  1  0  Coalition ideological distance  −1.19  1  0  Executive left–right  −9.1  4.5  18.97  Peace quarters  80  40  8  Peace quarters-squared  6,400  1,600  64  Peace quarters-cubed  512,000  64,000  512  Note: 95 percent confidence intervals in brackets. View Large In the simulation that represents low underlying odds of conflict, the opportunity for conflict is low, and the dyad has been at peace for twenty years. Coalition is held at one here, while Single-Party Majority is equal to zero. All continuous measures are set at either their 25th or 75th percentile values depending on the sign of the coefficient in Model 1 to make the probability of initiation low, while the dichotomous Other Initiated Disputes measure is set to its mode (0). Varying the independent variable from its mean by one standard deviation shows the substantive effect of the foreign policy position of the leader matters even in dyads where the underlying odds of conflict are low. Though a one standard deviation increase in the Militarism Index only increases the probability of conflict involvement by .00002 on average, when one considers that the baseline probability of conflict in any given quarter in this scenario is only .0001, this change represents a 20 percent increase from the baseline probability of initiation. This change is smaller than increasing the relative capability of the initiator by a one standard deviation and the effect of varying the regime of the target. However, it should be noted that these geostrategic variables are increased by large amounts that we would probably not observe within a single dyad from year to year. We would only be likely to observe that sort of variation cross-sectionally. In other words, it would be unlikely to actually observe a target state’s Polity score change from 1 to –7 in one year, but we definitely observe changing values of the Militarism Index comparable to the values used in this simulation within dyads from year to year. Moving to the scenario that represents a moderate risk of conflict involvement, the relative risk increases, and the dyad has been at peace for ten years. Here, all variables are set to their 50th percentile values, except Coalition, which is equal to one, Single-Party Majority, which is equal to zero, and Other Initiated Disputes, which is also equal to zero. Again, the baseline probability of conflict is low (.0006), but a one standard deviation increase from the mean in the independent variable increases the probability of conflict by about 16 percent from the baseline, suggesting that the foreign policy position of the PM or president also matters in this more moderate scenario. Though the baseline probability of conflict is low, a positive shift in the independent variable increases these odds by a meaningful amount. Finally, the high risk of conflict scenario simulates a dyad that has been at peace for two years, while the control variables are adjusted to increase the underlying odds of conflict.20 In this scenario, a one standard deviation increase in the independent variable from the mean increases the odds of conflict involvement by .002, showing a 15 percent increase from the baseline probability of dispute initiation. Another key point here is that the confidence interval for the Militarism Index does not overlap zero in any simulation, meaning we can be relatively confident the leader’s foreign policy position matters across different hypothetical scenarios. Finally, the choice to use a one standard deviation change (4.08 units) in the independent variable is a conservative choice and justified given the real-life choices that voters face and the changes we observe within states. In terms of pivotal elections, the shift from Jimmy Carter’s Democratic party to Ronald Reagan’s Republican party in 1980 is measured as a 17.6 unit shift in the independent variable, while the shift from Yitzhak Rabin’s Alignment alliance to Menachem Begin’s Likud party in Israel in 1977 corresponds with roughly a 20 unit change. The change from James Callaghan’s Labour government to Margaret Thatcher’s Conservative government in Great Britain in 1979 represents a 4.6 unit shift. Pivotal elections in democracies regularly feature policy shifts that are similar to or significantly larger than the one employed in these simulations. In sum, these simulations show that the foreign policy position of the PM/president has a significant impact on the probability of dispute initiation given the underlying odds of conflict. Discussion and Conclusion To reiterate, the results of the empirical tests support the hypothesis. Political parties and leaders that emphasize concepts related to militarism relative to anti-militarism and pacifism are more likely to initiate conflicts. These findings shed light on the process of foreign policy-making in democracies. Leaders and parties in democracies credibly signal their foreign policy preferences by selectively emphasizing some mixture of foreign policy concepts. There is an empirical relationship between emphasis of these concepts in manifestos and actual behavior in the international arena. Even after accounting for a number of domestic and international factors that we expect to constrain leaders, I find evidence that leaders live up to the foreign policy positions they espouse during campaigns. This analysis contributes to the body of research on the domestic determinants of conflict. Previous research has shown that partisanship and ideology matter (Fordham 1998; Palmer, London, and Regan 2004; Arena and Palmer 2009; Koch 2009). However, no studies have systematically examined the foreign policy positions of individual leaders across democracies until now. This research contributes to the study of international conflict in a fine-grained manner. More specifically, this analysis contributes to our understanding of partisanship and foreign policy. What I have argued here is that the foreign policy position of leaders can be seen as independent of their policy positions on other issues. In other words, we must extract a party’s foreign policy preferences from their aggregate ideological position to better predict behavior in office. Second, these results suggest leaders and parties may pursue dovish foreign policy positions without necessarily being concerned that other states will target their state for this reason. This is appealing normatively, as it means leaders and parties can credibly present a wide range of foreign policy options to voters without having a universal incentive to appear hawkish. This finding speaks directly to the SCA literature. Of course there are caveats to this point, as large foreign policy events may incentivize hawkish positions across the party system (e.g., 9/11 in the United States). This finding does not suggest other leaders totally ignore the ideology of leaders. Indeed, forthcoming work by Clare shows that dovish governments have an advantage in bargaining with other states because international adversaries know that future bargaining episodes with stubborn hawkish governments may prove unproductive. Because hawkish governments are more likely to use force according to the findings here, I believe this work is consistent with this new research on ideology and external actors’ expectations. Furthermore, though this work fails to find evidence of SCA by potential target states or diversion by democratic leaders, we can imagine that the foreign policy positions of leaders in democracies might function in the SCA/diversionary theory framework in other ways. For example, though hawkish leaders may not be able to ward off challenges from abroad over salient issues, they may be able to keep potential joiners from bandwagoning against them with threats made credible by their hawkish disposition. In other words, there are a wide variety of hypotheses along these lines that I have not tested, but which future research should. Finally, these findings matter for careful consumers of political information in democracies. Voters can use issue ownership to figure out which parties own certain issues, but they should also be aware of variation over time in issue emphasis. The work of Budge and Farlie (1983) and Carmines and Stimson’s (1989) provides the basis for such an argument. In the United States, the Democrats and Republicans have not occupied static foreign policy positions over time since the end of World War II, and these changes have concrete implications for US foreign policy. The story is much the same in democracies around the world. If we want to know what sorts of leaders are likely to engage in international conflict, we should focus specifically on their emphasis of foreign policy concepts instead of highlighting only binary measures of partisanship. One possible extension of this research might be to examine how foreign policy positions matter in the context of national elections. In some cases, voters seem to reward aggressive foreign policy positions (Reagan in 1980 and 1984), while other elections feature a more dovish winner (Clinton in 1996). Nincic (1988) astutely argues that parties may seek votes by reversing their rhetoric, and this may well play into my general argument about words matching deeds. Further research on this subject is worthwhile, as Nincic and Hinckley (1991) show that voters in the US condition their vote on approval of the executive’s foreign policy. How domestic audiences reward or punish different foreign policy stances is probably conditional on the salience of foreign policy and the international atmosphere. Recent research on this subject has looked both at the United States (Gadarian 2010) and cross-nationally (Williams, Brulé, and Koch 2010; Koch 2011), but more research is needed. In any case the relationship between issue emphasis and foreign policy is worth further investigation. Footnotes 1 In a similar vein, Kesgin and Kaarbo (2010) delineate the scope and conditions under which we can expect parliaments to have an independent effect on foreign policy in parliamentary democracies. 2 See also Williams (2014). 3 More recent research on diversionary uses of force examines what issues or scenarios beyond economic conditions (e.g. rivalry, territorial disputes, or rapidly increasing power in the target) are likely to induce a diversionary attack (Mitchell and Prins 2004; Mitchell and Thyne 2010; Jung 2014). 4 Carmines and Stimson’s (1989) study of racial politics as an issue in the United States provides a compelling argument for us to take a dynamic view of issue competition. Because politicians behave strategically and because exogenous shocks to the political system demand attention, parties are forced to address issues as they evolve over time. The concept of issue evolution itself implies a dynamic restructuring of party politics. I make no claim here that foreign policy issue emphasis has itself reshaped politics across democracies to the degree that racial politics did in the United States during the 20th century. However, the general approach of Carmines and Stimson underlies a significant assumption made here, which is that policy issues are themselves dynamic. More recent research on diversionary uses of force examines what issues or scenarios beyond economic conditions (e.g., rivalry, territorial disputes, or rapidly increasing power in the target) are likely to induce a diversionary attack (Mitchell and Prins 2004; Mitchell and Thyne 2010; Jung 2014). 5 As Whitten and Williams find, the relationship between domestic economic positions and foreign policy orientations may produce unexpected results with regard to defense spending, as dovish parties often compromise by voting for increased military spending as a sort of welfare in disguise. If we only examine aggregate measures of ideology or assume issues remain static over time, we are certain to miss such nuances. 6 I would like to thank an anonymous reviewer for raising this possibility. 7 Several excerpts from the Conservative party manifesto from the 1987 British national election included in the online appendix provide evidence for this point. Moreover, this example provides a prime example of the foreign policy perspective of a leader (Margaret Thatcher) that is typically classified as hawkish. Though the Conservative manifesto discusses nuclear arms reduction, the balance of statements suggests Great Britain should be prepared to take decisive action abroad. 8 For example, the 1987 Conservative Party manifesto is introduced from the perspective of Margaret Thatcher. This implies that the document communicates not only the party’s perspective on individual issues but Thatcher’s as well. 9 These data on use of force come from Fordham, who updates the original Blechman and Kaplan (1978) data for use in his 1998 piece, “Partisanship, Macroeconomic Policy, and U.S. Uses of Force, 1949-1994.” These data include only major uses of force, as is standard practice in the literature. 10 I would like to thank an anonymous reviewer for raising this point. 11 In the online appendix to this piece, I examine whether hawkish leaders are more likely to become involved in international crises using the International Crisis Behavior data. The results suggest more hawkish leaders, as indicated by the Militarism Index, are more likely to become involved in international crises. 12 Though I used the Blechman and Kaplan (1978) data for illustrative purposes above, I use MID data here because the Blechman and Kaplan data only covers the United States, while the MID data provides coverage of all states. 13 The full coding definition of these variables is provided in the appendix. 14 The Pro-Militarism indicator is weakly positively correlated with the Anti-Militarism indicator (r = .10) and moderately correlated with the Pro-Peace indicator (r = .28) implying that leaders who discuss foreign policy tend to mix their language. This can be seen as evidence for using an index in this case, because the goal is to understand where the balance of statements lies. 15 Bennett and Stam’s (2000) EUGene software was used to generate these dyadic control variables. 16 I would like to thank an anonymous reviewer for this suggestion. 17 This standardized measure is also positive and significant when estimating hostile and fatal MIDs as well, and these results are available in the online appendix. 18 I also re-estimated Models 1–4 exchanging the Coalition Ideological Distance and purged Executive Left–Right measures for a measure of Government Ideology. The results of these models are very similar to the ones discussed above and are available in the online appendix. 19 Clarify was used to generate these simulations (Tomz, Wittenberg, and King 2003). 20 Because Coalition is set to zero and Single-Party Majority is set to one, the Coalition Ideological Distance measure is set to zero as well because there are no coalition partners in this hypothetical scenario. References Arena Phillip, Palmer Glenn. 2009. “ Politics or the Economy? Domestic Correlates of Dispute Involvement in Developed Democracies.” International Studies Quarterly  53: 955– 75. Google Scholar CrossRef Search ADS   Auerswald David P. 1999. “ Inward Bound: Domestic Institutions and Military Conflict.” International Organization  53: 469– 504. Google Scholar CrossRef Search ADS   Bennett D. Scott, Stam Allan. 2000. “ EUGene: A Conceptual Manual.” International Interactions  26: 179– 204. Google Scholar CrossRef Search ADS   Blechman Barry, Kaplan Stephen. 1978. Force Without War: U.S. Armed Forces as a Political Instrument . Washington DC: Brookings Institution Press. Brulé David. 2006. “ Congressional Opposition, the Economy, and US Dispute Initiation, 1946-2000.” The Journal of Conflict Resolution  50: 463– 83. Google Scholar CrossRef Search ADS   Brulé David J, Williams Laron K. 2009. “ Democracy, and Diversion: Government Arrangements, the Economy, and Dispute Initiation.” Journal of Peace Research  46: 777– 98. Google Scholar CrossRef Search ADS   Budge Ian, Farlie Dennis J.. 1983. Explaining and Predicting Elections: Issue Effects and Party Strategies in Twenty-Three Democracies . London: George Allen and Unwin. Budge Ian, Hofferbert Richard I.. 1990. “ Mandates and Policy Outputs: U.S. Party Platforms and Federal Expenditures.” The American Political Science Review  84: 111– 31. Google Scholar CrossRef Search ADS   Carmines Edward G, Stimson James A.. 1989. Issue Evolution: Race and the Transformation of American Politics . Princeton, NJ: Princeton University Press. Carter David, Signorino Curtis S.. 2010. “ Back to the Future: Modeling Time Dependence in Binary Data.” Political Analysis  18: 271– 92. Google Scholar CrossRef Search ADS   Carter Ralph D, Scott James M.. 2009. Choosing to Lead: Understanding Congressional Foreign Policy Entrepreneurs . Durham, NC: Duke University Press. Google Scholar CrossRef Search ADS   Clare Joe. 2010. “ Ideological Fractionalization and the International Conflict Behavior of Parliamentary Democracies.” International Studies Quarterly  54: 965– 87. Google Scholar CrossRef Search ADS   Clare Joe. 2014. “ Hawks Doves and International Cooperation”. Journal of Conflict Resolution  58: 1311– 37. Google Scholar CrossRef Search ADS   Clark David H. 2000. “ Agreeing to Disagree: Domestic Institutional Congruence and US Dispute Behavior.” Political Research Quarterly  53: 375– 400. Clark David H. 2003. “ Can Strategic Interaction Divert Diverionsary Behavior? A Model of US Conflict Propensity.” Journal of Politics  65: 1013– 39. Google Scholar CrossRef Search ADS   Cowen Tyler, Sutter Daniel. 1998. “ Why Only Nixon Could Go to China.” Public Choice  97: 605– 15. Google Scholar CrossRef Search ADS   Cukierman Alex, Tommasi Mariano. 1998. “ When Does It Take a Nixon to Go to China?” The American Economic Review  88: 180– 97. DeRouen Karl. 1995. “ The Indirect Link: Politics, the Economy, and the Use of Force.” The Journal of Conflict Resolution  39: 671– 95. Google Scholar CrossRef Search ADS   Downs George W, Rocke David M.. 1995. Optimal Imperfection? Domestic Uncertainty and Institutions in International Relations . Princeton, NJ: Princeton University Press. Fearon James D. 1997. “ Signaling Foreign Policy Interests: Tying Hands versus Sinking Costs.” Journal of Conflict Resolution  41: 68– 90. Google Scholar CrossRef Search ADS   Fordham Benjamin. 1998. “ The Politics of Threat Perception and the Use of Force: A Political Economy Model of US Uses of Force, 1949-1994.” International Studies Quarterly  42: 567– 90. Google Scholar CrossRef Search ADS   Fordham Benjamin. 2005. “ Strategic Conflict Avoidance and the Diversionary Use of Force.” Journal of Politics  67: 132– 53. Google Scholar CrossRef Search ADS   Fordham Benjamin, Sarver Christopher C.. 2001. “ Militarized Interstate Disputes and United States Uses of Force.” International Studies Quarterly  45: 455– 66. Google Scholar CrossRef Search ADS   Gadarian Shana Kushner. 2010. “ Foreign Policy at the Ballot Box: How Citizens Use Foreign Policy to Judge and Choose Candidates.” Journal of Politics  72: 1046– 62. Google Scholar CrossRef Search ADS   Ghosn Faten, Palmer Glenn, Bremer Stuart. 2004. “ The MID3 Data Set 1993-2001: Procedures, Coding Rules and Description.” Conflict Management and Peace Science  21: 133– 54. Google Scholar CrossRef Search ADS   Gowa Joanne. 1998. “ Politics at the Water’s Edge: Parties, Voters, and the Use of Force Abroad.” International Organization  52: 307– 24. Google Scholar CrossRef Search ADS   Heston Alan, Summers Robert, Aten Bettina. 2012. “ Penn World Tables Version 7.1.” Penn World Table Version 7.1, Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania . Hibbs Douglas A. 1977. “ Political Parties and Macroeconomic Policy.” American Political Science Review  71: 1467– 87. Google Scholar CrossRef Search ADS   Howell William G, Pevehouse Jon C.. 2005. “ Presidents, Congress, and the Use of Force.” International Organization  59: 209– 32. Google Scholar CrossRef Search ADS   Howell William G, Pevehouse Jon C.. 2007. While Dangers Gather: Congressional Checks on Presidential War Powers . Princeton, NJ: Princeton University Press. James Patrick, Hristoulas Athanasios. 1994. “ Domestic Politics and Foreign Policy: Evaluating a Model of Crisis Activity for the United States.” Journal of Politics  56: 327– 48. Google Scholar CrossRef Search ADS   James Patrick, Oneal John. 1991. “ The Influence of Domestic and International Politics on the President’s Use of Force.” The Journal of Conflict Resolution  35: 307– 32. Google Scholar CrossRef Search ADS   Jung Sung Chul. 2014. “ Foreign Targets and Diversionary Conflict.” International Studies Quarterly  58: 566– 78. Google Scholar CrossRef Search ADS   Kaarbo Juliet. 1996. “ Power and Influence in Foreign Policy Decision Making: The Role of Junior Coalition Partners in German and Israeli Foreign Policy.” International Studies Quarterly  40: 501– 30. Google Scholar CrossRef Search ADS   Kaarbo Juliet, Beasley Ryan K.. 2008. “ Taking It to the Extreme: The Effect of Coalition Cabinets on Foreign Policy.” Foreign Policy Analysis  4: 67– 81. Google Scholar CrossRef Search ADS   Kang Choong-Nam, Gibler Douglas M.. 2012. “ An Assessment of the Validity of Empirical Measures of State Satisfaction with the Systemic Status Quo.” European Journal of International Relations  Kesgin Baris, Kaarbo Juliet. 2010. “ When and How Parliaments Influence Foreign Policy: The Case of Turkey’s Iraq Decision.” International Studies Perspectives  11: 19– 36. Google Scholar CrossRef Search ADS   King Gary, Zeng Langche. 2001. “ Logistic Regression in Rare Events Data.” Political Analysis  9: 137– 63. Google Scholar CrossRef Search ADS   Klingemann Hans-Dieter, Hofferbert Richard, Budge Ian. 1994. Parties, Policies and Democracy . Boulder: Westview Press. Koch Michael T. 2009. “ Governments, Partisanship, and Foreign Policy: The Case of Dispute Duration.” Journal of Peace Research  46: 1– 19. Google Scholar CrossRef Search ADS   Koch Michael T. 2011. “ Casualties and Incumbents: Do the Casualties from Interstate Conflicts Affect Incumbent Party Vote Share?” British Journal of Political Science  41: 795– 817. Google Scholar CrossRef Search ADS   Leeds Brett Ashley, Davis David R.. 1997. “ Domestic Political Vulnerability and International Disputes.” Journal of Conflict Resolution  41: 814– 34. Google Scholar CrossRef Search ADS   Marshall Monty, Jaggers Keith. 2002. “ Polity IV: Political Regime Characteristics and Transitions, 1800-2002.” Polity IV Manual . McDonald Michael D, Budge Ian. 2005. Elections, Parties, Democracy: Conferring the Median Mandate . New York: Oxford University Press. Google Scholar CrossRef Search ADS   Meernik James, Waterman Peter. 1996. “ The Myth of the Diversionary Use of Force by American Presidents.” Political Research Quarterly  49: 573– 90. Google Scholar CrossRef Search ADS   Mitchell Sara McLaughlin, Prins Brandon C.. 2004. “ Rivalry and Diversionary Uses of Force.” Journal of Conflict Resolution  48: 937– 61. Google Scholar CrossRef Search ADS   Mitchell Sara McLaughlin, Thyne Clayton L.. 2010. “ Contentious Issues as Opportunities for Diversionary Behavior.” Conflict Management and Peace Science  27: 461– 85. Google Scholar CrossRef Search ADS   Mitchell Sara McLaughlin, Moore Will H.. 2002. “ Presidential Uses of Force During the Cold War: Aggregation, Truncation, and Temporal Dynamics.” Journal of Politics  46: 438– 52. Google Scholar CrossRef Search ADS   Nincic Miroslav. 1988. “ The United States, the Soviet Union, and the Politics of Opposites.” World Politics  40: 452– 75. Google Scholar CrossRef Search ADS   Nincic Miroslav, Hinckley Barbara. 1991. “ Foreign Policy and the Evaluation of Presidential Candidates.” The Journal of Conflict Resolution  35: 333– 55. Google Scholar CrossRef Search ADS   Oppermann Kai, Brummer Klaus. 2014. “ Patterns of Junior Partner Influence on the Foreign Policy of Coalition Governments.” British Journal of Politics and International Relations  16: 555– 71. Google Scholar CrossRef Search ADS   Ostrom Charles, Job Brian. 1986. “ The President and the Political Use of Force.” American Political Science Review  80: 541– 66. Google Scholar CrossRef Search ADS   Palmer Glenn, London Tamar, Regan Patrick. 2004. “ What’s Stopping You? The Sources of Political Constraints on International Conflict Behavior in Parliamentary Democracies.” International Interactions  30: 1– 24. Google Scholar CrossRef Search ADS   Petrocik John R. 1996. “ Issue Ownership in Presidential Elections with a 1980 Case Study.” American Journal of Political Science  40: 825– 35. Google Scholar CrossRef Search ADS   Powell G. Bingham. 2000. Elections as IInstrument of Democracy: Majoritarian and Proportional Visions . Yale University Press, New Haven, CT. Schultz Kenneth. 2005. “ Do Hawks or Doves Deliver the Olive Branch?” International Organization  59: 1– 38. Google Scholar CrossRef Search ADS   Seki Katsunori, Williams Laron K.. 2014. “ Updating the Party Government data set.” Electoral Studies  34: 270– 79. Google Scholar CrossRef Search ADS   Signorino Curtis S, Ritter Jeffrey M.. 1999. “ Tau-b or Not Tau-b: Measuring the Similarity of Foreign Policy Positions.” International Studies Quarterly  43: 115– 44. Google Scholar CrossRef Search ADS   Singer J. David. 1988. “ Reconstructing the Correlates of War Dataset on Material Capabilities of States, 1816-1985.” International Interactions  14: 115– 32. Google Scholar CrossRef Search ADS   Smith Alastair. 1996. “ Diversionary Foreign Policy in Democratic Systems.” International Studies Quarterly  40: 133– 53. Google Scholar CrossRef Search ADS   Tomz Michael, Wittenberg Jason, King Gary. 2003. “ CLARIFY: Software for Interpreting and Presenting Statistical Results.” Version 2.1 . Stanford University, University of Wisconsin, and Harvard University. Volkens Andrea, Lehmann Pola, Merz Nicolas, Sven Regel, Werner Annika. 2013. “ The Manifesto Data Collection. Manifesto Project (MRG/CMP/MARPOR).” Whitten Guy, Williams Laron K.. 2011. “ Buttery Guns and Welfare Hawks: The Politics of Defense Spending in Advanced Industrial Democracies.” American Journal of Political Science  55: 117– 35. Google Scholar CrossRef Search ADS   Wildavsky Aaron. 1966. “ The Two Presidencies.” Society  4: 7– 14. Google Scholar CrossRef Search ADS   Williams Laron K., Brule David J., Koch Michael T.. 2010. “ War Voting: Interstate Disputes, the Economy, and Electoral Outcomes.” Conflict Management and Peace Science  27: 442– 60. Google Scholar CrossRef Search ADS   Woldendorp Jaap, Keman Hans, Budge Ian. 2000. Party Government in 48 Democracies (1945-1998) . London: Kluwer Academic Publishers. Google Scholar CrossRef Search ADS   © The Author (2016). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Foreign Policy AnalysisOxford University Press

Published: Jan 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off