Diversionary war is False!

There is a large literature on diversionary conflict in international relations, but it focuses on material conflicts like militarized interstate disputes rather than foreign policy rhetoric. It is based in social identity theory, which suggests that leaders can increase ingroup affinity by making intergroup distinctions more salient (Coser 1956; Simmel 1955; Tajfel and Turner 1979). A recent review concludes that though the internal logic of diversionary conflict is “compelling and theoretically well supported,” the empirical evidence is “decidedly mixed” (Baum and Philip 2008, 48). Several studies find evidence of diversionary aggression in US foreign policy (Clark 2003; DeRouen 2000; DeRouen and Peake 2002; Fordham 1998a,b; Hess and Orphanides 1995; Howell and Pevehouse 2005; James and Hristoulas 1994; James and Oneal 1991; Levy 1989a, 1989b; Clifton and Bickers 1992; Ostrom and Job 1986) and elsewhere (Bennett 2000; Dassel and Reinhardt 1999; Davies 2002; Enterline and Gleditsch 2000; Gelpi 1997; Heldt 1999; Lebow 1981; Mansfield and Snyder 1995; Oneal and Tir 2006; Russett 1990; Sobek 2007; Tir 2010). Yet skeptics have amassed opposing evidence (Chiozza and Goemans 2003, 2004; Foster and Palmer 2006; Gowa 1998; Johnston 1998; Leeds and Davis 1997; Lian and Oneal 1993; Meernik 2004, 2000; Meernik and Waterman 1996; Moore and Lanoue 2003; Potter 2007). Some cases are hard to reconcile with the theory: in Britain, there were rallies in the Falklands War and the Gulf War but not in other cases in which rallies would be expected, such as the Korean, Suez, and Kosovo wars (Lai and Reiter 2005). Some go so far as to call diversionary aggression a “myth” (Meernik and Waterman 1996).

Others have developed scope conditions for diversionary aggression. It is more likely between states with pre-standing rivalries (Mitchell and Prins 2004), when leaders are accountable (Carter 2018; Kisangani and Pickering 2011), and in mature democracies, consolidating autocracies, and transitional polities (Pickering and Kisangani 2005). It is less likely when states avoid provoking troubled rivals (Clark 2003; Fordham 2005; Leeds and Davis 1997; Miller 1999). Diversion appears more likely to produce a rally when supported by Security Council authorization (Chapman 2011; Chapman and Reiter 2004), when the White House draws attention to a dispute (Baker and Oneal 2001), and in conditions of media attention, popular leadership, divided government, non election years, and first terms (Colaresi 2007). Most recently, scholars have asked whether diversion occurs outside democracies. They find some autocracies, especially single party regimes, divert as well (Carter 2018; Pickering and Kisangani 2010).

This study extends the logic of diversionary conflict to foreign policy rhetoric. There is surprisingly little research on rhetoric in international relations. The international relations literature deems talk “cheap” (Fearon, 1995; Kydd, 2005). The audience cost literature considers rhetoric meaningful, but only if it invokes audience costs through explicit, public threats (Fearon, 1994; Schultz, 2001; Smith, 1998; Tomz, 2007).

However, if foreign policy rhetoric can activate ingroup identity, then it may be appealing for leaders who wish to improve their ratings without incurring the substantial risks of militarized interstate disputes. Domestic political factors like low presidential approval ratings can lead otherwise rational states to choose “suboptimal” foreign policies (Fearon 1998). Diversionary rhetoric represents optimization inside the bounds of domestic political constraints. While it might be “outlandish” for presidents to engage in the impeachable exercise of diversionary war (Meernik and Waterman 1996), hostile foreign policy rhetoric is far less outlandish a risk.

To develop a theory of diversionary cheap talk, this paper draws upon research in political psychology and political communication. These literatures find persuasive evidence that elite statements influence citizen beliefs (Behr and Iyengar, 1985; Bennett, Lawrence and Livingston, 2006; Brody, 1991; Cohen, 1995; Jentleson, 1992; Zaller and Chiu, 2000). I draw on social identity theory to argue that diversionary cheap talk highlights intergroup differences between nations and leads citizens to evaluate their leader favorably. When a leader criticizes foreigners, she cues ingroup identity, which increases citizens’ social attachment to the nation and to herself as its leader. This is a “solidarity mechanism,” through which “[c]ollective group goals and common group identity are highlighted, norms of groupbased altruism are strengthened, punishment and rejection of defectors are increased, and perceptions of the in-group and out-group are manipulated” (Halevy, Bornstein and Sagiv, 2008, 405).

The theory generates observable implications about when leaders use diversionary cheap talk and who they target. I follow the consensus in the diversionary conflict literature in focusing on poor economic conditions as the most important source of public disapproval for leaders. Low approval ratings limit leaders’ ability to advance their domestic agenda. Therefore, when the economy deteriorates, leaders will criticize foreign nations to improve their approval ratings and restore the political capital necessary for them to govern.

Second, a key observation from social identity theory is that the depth of intergroup differences is important for group attachment. Therefore, consonant with recent empirical findings in the diversionary conflict literature (Mitchell and Prins 2004), I expect diversionary rhetoric to be most effective when it targets threatening outgroups. In the context of foreign policy, these are best represented by historical rivals. Finally, I expect diversionary cheap talk to have differential partisan effects. The solidarity mechanism suggests that because national identity cues widen the tent of the political ingroup, diversionary cheap talk should be most effective at boosting support among the leader’s nonpartisans: conservative voters for Democratic presidents, and liberal voters for Republican presidents. However, this solidarity mechanism is mediated by citizens’ foreign policy preferences. Liberals prefer more dovish foreign policy and hence might interpret diversionary rhetoric from Republican presidents as too extreme. In contrast, Republicans prefer more hawkish foreign policy and therefore are likely to reward Democratic presidents for hostile rhetoric. Therefore, diversionary rhetoric should be most powerful for Democratic presidents, whose non-core voters are most likely to reward such rhetoric.

I test these hypotheses with the American Diplomatic Dataset, an original record of over 50,000 US diplomatic events between 1945 and 2010 drawn from New York Times articles on foreign affairs. I used tools from computational social science to classify bilateral interstate interactions into hundreds of specific types and four aggregate categories: rhetorical cooperation, rhetorical conflict, material cooperation, and material conflict. This is by far the most historically extensive event dataset. As such, it allows an exploration of US foreign policy behavior across a variety of administrations and economic crises. Crucially, the granularity of the dataset allows me to restrict attention to events which the sitting US president initiated and directed at foreign state actors.

I find robust evidence of diversionary cheap talk in US foreign policy. First, I establish that US presidents face incentives to divert verbally rather than materially: while militarized interstate dispute initiation does not affect presidential approval ratings, hostile foreign policy rhetoric is associated with increased ratings, especially for Democratic presidents among Independent and Republican citizens. Responding to this incentive, Democratic presidents between 1945 and 2010 typically diverted in the form of words, not deeds. Simulations indicate that as the misery index of unemployment plus inflation varies from its observed minimum to its observed maximum, predicted hostile foreign policy rhetoric doubles. Throughout this study, estimates are conservative: I operationalize conflict as events the United States initiated, although findings are robust to a redefinition of conflict as events the United States participated in. The rhetorical statements in the dataset are high profile and likely to be noticed by the American public: all appeared in the headlines of the New York Times, the American newspaper of record in the postWorld War II era.

This study contributes to existing scholarship in several ways. First, it demonstrates that US foreign policy rhetoric may be shaped by domestic economic conditions. International relations scholars should therefore continue to focus more seriously on the communicative aspects of foreign policy, and in particular its relationship to domestic politics (Johnston 2001, 2008; Kurizaki 2007; Ramsay 2011; Sartori 2002, 2005; Trager 2010, 2011, 2016). The American Diplomacy Dataset will enable researchers to further explore the communicative aspects of foreign policy, and their relationships to material and economic factors, in more detail than existing datasets permit.

Second, this study contributes to the diversionary conflict literature by showing that in many cases where diversionary theory predicts conflict initiation, leaders may instead choose rhetorical hostility. In this sense, leaders may have their cake and eat it too: They benefit from an ingroup rally without inviting an international crisis. The mixed empirical findings in the diversionary conflict literature may be partly due to the fact that existing scholarship considers only the most serious forms of diversion like militarized interstate disputes. It is possible that a wide range of diversionary behavior takes place at less extreme levels, such as the rhetorical hostility documented in this paper.1