Why are the decisions of international organizations (IOs) sometimes treated as if they are legitimate by observers? Recent work provides a compelling answer: because of pervasive uncertainty surrounding international events, actors look for additional sources of information when making decisions about policy. The domestic public, for instance, may seek information about either foreign policies, or the leaders who propose them, when deciding whether to support their elected leaders.Footnote 1 Audiences in other countries may want reassurance that a state will not act overly aggressively when conducting foreign policyFootnote 2 or may glean information about states' resolve by observing IO activity.Footnote 3 The idea that key audiences may view IO decisions with intense interest points to an important way in which institutions can influence foreign policy, even in the “high politics” realm of security affairs. Namely, institutions might constrain leaders through the indirect channel of public opinion. This challenges traditional realist views, which tend to focus on institutions' lack of direct enforcement mechanisms.Footnote 4 However, these views have generally ignored indirect channels of influence, such as those that might exist if IOs influence public opinion and public opinion, under some conditions, constrains leaders. If IO activity affects public support for foreign policies and elected officials, it follows that leaders will often face a difficult tradeoff between making compromises in order to obtain the benefits of institutional authorization and eschewing multilateral fetters in favor of policy autonomy.Footnote 5
Take, for instance, U.S. President George W. Bush and his administration's efforts to acquire UN Security Council (SC) approval prior to the March 2003 beginning of the Iraq War. The administration invested substantial time and resources in the campaign to acquire SC approval. The internal debate within the White House also reportedly led partially to the resignation of a key foreign policy figure in the administration, Secretary of State Colin Powell. Why did the U.S. government exert so much effort and risk the political costs of failure to secure the approval of the SC if they intended to carry through with the war regardless of authorization?
This behavior is not limited the 2003 Iraq War. President George H. W. Bush and his National Security Advisor General, Brent Scowcroft, expressed hope that UN SC Resolution 678, authorizing the 1991 Persian Gulf War, would help boost support at home as well as give political cover to key Arab allies.Footnote 6 The administration even created a special public relations committee to publicize the authorization and allied support for the war.Footnote 7 President Harry Truman sought SC authorization for the Korean War to build allied support, and later during the Korean War sought a UN General Assembly resolution authorizing the crossing of the 38th parallel.Footnote 8 Even in the hours after the infamous Gulf of Tonkin incident, the administration of President Lyndon Johnson discussed how they could amass evidence to make the case for the war to the SC, although they did not ultimately ask for authorization.Footnote 9 Authorization-seeking behavior is also not limited to U.S. governments. British Prime Minister Tony Blair's insistence on seeking SC authorization for the 2003 Iraq War is well documented,Footnote 10 and some governments make SC authorization a virtual prerequisite for the commitment of military forces. Thompson makes this point strongly, writing, “foreign intervention without some effort to gain [external] approval is now virtually obsolete.”Footnote 11
Constructivist scholarship provides one explanation for this behavior, pointing to the symbolic importance of obtaining authorization or operating through appropriate channels prior to conducting foreign policy.Footnote 12 From this perspective, it is more or less always optimal to seek institutional approval, even if one does not expect it to be forthcoming, because it signals an effort to do what is appropriate for modern, mature states. Yet this explanation does not match the empirical record, as states do not always consult organizations during foreign policy crises. For instance, states brought issues to the SC in only about one quarter of all international crises between 1946–2003.Footnote 13 Moreover, the constructivist view suggests that certain institutions become imbued with legitimacy over time, owing to either their decision-making procedures or to how they are treated by powerful states.Footnote 14 The SC, in particular, has become an important arbiter of policy appropriateness.Footnote 15 This suggests that decisions from such an institution should routinely be taken as important. Yet one can observe variation in reactions to institutional decisions. For instance, European and U.S. audiences viewed the failure to obtain SC authorization prior to the 2003 Iraq War in different ways, and the American public seemed to treat the failure prior to 2003 as less important than the authorization in 1991 for the Persian Gulf War, as indicated by trends in public support for the two wars.Footnote 16
A third approach, however, provides an explanation rooted in the rationalist logic of the information transmission.Footnote 17 This view, labeled the strategic information perspective, suggests that behavior consistent with legitimacy, or the belief that an institution's decisions should be followed, is conditional on perceptions about an institution's motives or the preferences of member states. Of course, from the point of view of the hypothetical uninformed citizen, who cares only about “good” policy, it is optimal to have a source of information that is altruistic, or that also cares only about good policy. If the institution has access to more information, such as vital intelligence and foreign policy experience, but holds preferences that are identical to the uninformed citizen, its decisions should be taken at face value and greatly affect citizens' beliefs about international affairs. But such a situation is unrealistic when one considers contemporary multilateral security institutions. Institutions such as the SC are comprised of states with heterogeneous, and often competing, interests. These bodies are therefore unlikely to be seen as completely neutral or altruistic. Yet this does not necessarily preclude audiences learning from IO decisions. When multilateral security bodies hold aggregate preferences that are perceived to be generally opposed to the foreign policy goals of a proposing state, support or authorization from those bodies can be a powerful signal to an uninformed audience that a proposed policy is worth supporting, precisely because the uninformed audience does not expect authorization. However, while the effect of IO support on public support for foreign policies should increase as an IO is seen as less likely to grant authorization, the effect of IO opposition (or failure to garner support) should decrease with that same perception. In other words, bodies whose interests are seen to be inimical to a proposing state can provide an uninformed audience with a strong signal of legitimacy but may often see their opposition ignored. Bodies perceived as friendly to a state's interests can provide an audience with a strong signal of the dangers of a proposed policy by expressing their (unexpected) disapproval, but their support may have little effect on public support.Footnote 18 The key point is that the effect of institutional authorization and opposition on audiences will vary according to how an institution is perceived.
There is, in fact, some anecdotal evidence that actors react differently to the activity of different institutions (for example, SC versus the Organization of American States (OAS)) and to the same institution at different times (for example, SC pre– and post–Cold War); and different observers react differently to the same institutional decision (for example, American and European reactions to the SC before the 2003 Iraq War). There is no scholarly research on why this should be the case. The informational logic also has counterintuitive implications for foreign policy behavior, in that instead of seeking political “cover” for foreign policy by appealing to organizations that are clearly expected to offer support, governments face some incentive to appeal to institutions that appear likely to oppose them. Winning support from such institutions provides a boon for public support, while opposition from such bodies does not automatically decrease public support.
In order to establish why states appeal to IOs, it is useful to examine the informational perspective's claims regarding public reaction to IOs. In this article, I first discuss trends in ideas about the influence of IOs and highlight the insights offered by the strategic information perspective. Second, I present illustrative polling evidence that conforms to the expectations of a strategic information perspective. This evidence is suggestive, but does not constitute a scientific test of the informational argument. Third, I present a research design to test the alternative hypotheses generated by the symbolic legitimacy and strategic information perspectives. Specifically, I test whether SC support systematically increases the change in presidential approval during military disputes, and whether the size of this effect is conditional on a measure of ideological affinity between the United States and the SC in a given year. The analysis shows that changes in public approval surrounding U.S. military disputes have been systematically larger when accompanied by a SC that is ideologically distant, in terms of foreign policy preferences, from the United States. This result provides support for the strategic information perspective and offers the first quantitative evidence for the initial link in the causal chain that claims that IOs can indirectly influence foreign policy through altering public opinion.
Power, Legitimacy, and Information
The idea that institutions affect foreign policy, even in the power-dominated realm of security affairs, has important implications for broad international relations theory. First, traditional realist views suggest that institutional statements are essentially irrelevant and should have little effect on actors' beliefs and behavior, as multilateral security institutions lack robust enforcement power. Constructivist and legalist work maintains that multilateral authorization carries with it symbolic import or that consulting an international organization prior to using military force is the procedurally appropriate thing to do. A third view suggests that multilateral authorization conveys policy-relevant information to important audiences, thus potentially constraining foreign policy leaders who rely on those audiences for support. Each of these views suggests different empirical patterns, and hence ways to differentiate between them with careful data analysis.
IOs and the Prerogatives of Power
Realist students of international politics have traditionally dismissed the ability of IOs to influence state behavior independent of the power politics.Footnote 19 According to this view, international organizations should be least influential in the area of international security, since the primary goal of states is to preserve their own security. To the extent that IO decisions are enforced, it is because powerful states take it upon themselves to do so, as the United States has done in recent memory, but generally states will ignore IO decisions that are detrimental to their interests, while supporting and complying with those that coincide with their interests.Footnote 20
In point of fact, the rate of compliance with SC decisions is only mixed at best.Footnote 21 According to the realist perspective, since enforcement capacity for most multilateral security IOs is so weak, audiences should disregard institutional decisions as essentially “cheap talk,” communicating little if anything about the costs of policy or likely policy outcomes. This view is certainly compelling, particularly given the history of global security institutions. The first truly large-scale attempt at collective security, The League of Nations, initially championed by President Woodrow Wilson, failed to be ratified by Congress because of domestic isolationist sentiment. Without the presence of the United States, who emerged after World War I as a rising superpower, the League Council was left with four permanent members—the United Kingdom (UK), Italy, Japan, and France. Notably, the League Council operated on a unanimous voting rule, which gave veto power to the four permanent members as well as the nonpermanent members. This had the effect of severely limiting the League's ability to act in cases in which any of these members held conflicts of interest. Although the League settled several minor border disputes, such as those between Albania and Yugoslavia (1921), Greece and Bulgaria (1925), and Iraq and Turkey (1925–26), its legacy is dominated by failures to intervene during the Japanese invasion of Manchuria in 1931, the Italian invasion of Ethiopia in 1935, and during Soviet aggression against Finland in 1938. Germany left the League in 1933 and the League famously failed to intervene during German rearmament or annexation of the Czechoslovakian Sudetenland. The failure of this early attempt at collective security seemed to demonstrate that international organizations could do little to deter powerful states from pursuing their own interests.
Although the UN SC is much longer lived, at first glance its record seems little better than the League. Early optimism about the new postwar organization resulted in some initial activity, such as SC discussions during the 1948 Berlin blockade. The SC authorized military action in the Korean War, although notably in the absence of Soviet representation, as the Soviets boycotted the 1950 session. In the 1960s and the 1970s, the Council played a part in disputes in the Middle East, Africa, and Southeast Asia, but it was generally inactive or passed only weak resolutions where the Soviet Union or United States held stakes in the outcome of conflicts. Indeed, the distribution of SC resolutions over its lifetime is heavily skewed away from the five permanent members' areas of direct interest, and the general perception of the organization during the Cold War was that it was paralyzed by the opposing vetoes of the United States and Soviet Union.Footnote 22 Thus, the Cold War record largely supports the notion that conflicting member-state interests rendered the institution impotent.
The end of the Cold War brought renewed optimism about multilateral cooperation in general and the SC in particular. The Council achieved remarkable consensus in 1990–91, authorizing the expulsion by force of Iraqi forces from Kuwait. The Council also authorized several peacekeeping missions in the 1990s, as well as U.S. intervention in Haiti in 1994. However, mixed success throughout the 1990s weakened the initial post–Cold War enthusiasm. The 1999 U.S. circumvention of the SC in the Kosovo War, and the more unpopular decision in 2003 to conduct the Iraq War without explicit authorization, resulted in renewed charges of institutional irrelevance. These events seemed to echo the claims of Carr and others, who underscored the fundamental primacy of power and interests over institutions and international law.Footnote 23
Yet even post-2003, powerful nations, including the United States, continue to consult the SC over issues such as North Korean and Iranian nuclear programs and the 2006 Israeli-Hezbollah war. States continue to act as if the institution is useful, and majorities of the U.S. public still believe that it is important to gain UN approval when conducting foreign policy.Footnote 24 The effort states invest in securing approval along with a strong public preference for approval is puzzling from a power-centric realist viewpoint, which would generally predict that UN support has little independent effect on foreign policy.
One reason for the inability of the realist tradition to explain this behavior is its overly narrow focus on hard enforcement capacities, at the neglect of considering indirect channels of influence. Yet a large body of work has shown that IOs can influence domestic politics, which in turn affects the decisions of state leaders.Footnote 25 Others have claimed that international agreements are often used for domestic political purposes,Footnote 26 but there has been little investigation as to how this dynamic might work in the context of security affairs. This investigation is important, given the prominence of domestic-constraint explanations in many theories of foreign policymaking.
Symbolic Legitimacy
An alternative view is that certain institutions possess symbolic importance, owing to properties such as fairness, inclusiveness, or procedural appropriateness. Although definitions of legitimacy vary, it usually refers to a normatively held belief that a rule or decision ought to be obeyed because it is the right thing to do, not because obeying that rule will result in beneficial consequences.Footnote 27 This suggests that individuals' beliefs and behavior should be affected by the activity of institutions that are perceived as symbolically legitimate. In turn, IO approval, or the act of consulting an IO, can convince audiences of the appropriateness of proposed policies and increase public support.
Although views differ on whether legitimacy stems from perceptions of procedural fairness,Footnote 28 policy neutrality,Footnote 29 or inclusiveness,Footnote 30 once this relational characteristic is established, this perspective suggests that it should be relatively fixed, or at least slow to change. One should generally observe high rates of compliance with the decisions of institutions that are considered legitimate,Footnote 31 although one may observe differences in rates of compliance between institutions because of the differential legitimacy they are accorded.
Establishing this sort of legitimacy is important because it allows the body to influence actors through means other than material inducements and deterrents.Footnote 32 In other words, it is one possible way in which security IOs can influence states even though they lack robust enforcement power. For instance, IkenberryFootnote 33 argues that in the aftermath of the World War II, the United States sought to create a legitimate international order because relying on coercion to achieve its foreign policy goals would be prohibitively costly. By reassuring European allies and by actively incorporating their input, the United States was able to signal benign intent and imbue the postwar order with a sense of legitimacy, resulting in widespread acceptance and longevity of the order.Footnote 34
The legitimacy perspective is certainly consistent with recent observations about the frequency with which even powerful states appeal to multilateral institutions. It also suggests two possible, and empirically verifiable, patterns. First, if legitimacy derives from following the procedure of consulting an institution, then public support for foreign policies should increase when states consult relevant international bodies, regardless of whether authorization is obtained. Second, if legitimacy derives from receiving authorization from an important institution, the effect of authorization from the same institution should be more or less constant across time, or at least not conditional on audience beliefs about the foreign policy preferences of that institution. I show below that neither of these patterns is consistent with available data.
Information Transmission
A third view suggests that behavior consistent with a legitimacy view, such as the frequent decision by governments to seek multilateral authorization, or a domestic audience reacting to authorization with increased support for their leader, may be due to the information conveyed by institutional activity to key audiences, such as the domestic or foreign public. For instance, VoetenFootnote 35 suggests that the SC possesses few of the characteristics—policy neutrality, inclusiveness, procedural fairness—that tend to lead to perceptions of legitimacy. Instead, Voeten suggests that authorization signals that powerful states will not oppose a foreign policy, which lowers expectations of costs for domestic audiences and signals that policies will not be overly expansive to foreign audiences. ThompsonFootnote 36 offers a related theory, claiming that powerful states seek authorization for coercive purposes because it signals benign intent to foreign audiences, lowering international opposition.
Two formal models, by FangFootnote 37 and Chapman,Footnote 38 make similar arguments. In both models, the domestic public features prominently as an interested, but uninformed, consumer of information about foreign policy. In Fang's model the decision to seek multilateral authorization can signal a leader's competence to domestic audiences. Rational leaders should anticipate whether they will receive IO support or opposition, implying that incompetent leaders will often avoid consulting an IO and revealing their “type.” In turn, the public should be more likely to support leaders that consult an international organization such as the SC, because it signals that they are competent, in the sense that they are likely to implement good policies.
Chapman's model argues that IO authorization can signal whether proposed policies are good, in the sense that they conform closely to the public's preferences, but adds the caveat that the effect of authorization on domestic support for policies should be conditional on an audience's perceptions of the interests of an organization's member states, who necessarily drive organizational decisions. In other words, audiences that are relatively uninformed about international affairs may seek information about foreign policy when forming their opinions, but must “consider the source,” just as they would when listening to politicians or media pundits arguing for domestic policies. A citizen who cares only about good policy may discount arguments about the value of tax cuts from a Republican senator with a record of arguing for “small government,” while the same citizens may take similar arguments from a Democratic senator with a record of supporting public spending as an important signal about the need for tax cuts.Footnote 39 Chapman's analysis suggests an analogous process should operate when citizens observe signals from international organizations. Organizations comprised of member states that are predisposed to veto a given resolution communicate little about policy merit when they oppose policies, as they are expected to do so. However, support from such member states can convince audiences that a policy is worth supporting. The opposite is true for organizations that are comprised of “friends” or traditional allies of a proposing state; opposition is an unexpected and strong signal that a given policy may be costly or overly aggressive, while support conveys little information about policy appropriateness because it is expected.
Of course, in general models of information transmission between a principal, or receiver, and an agent, or sender, “truth-telling” or “information revealing” equilibria occur when the principal and agent have identical preferences. In such a situation what is good for the agent is also good for the principal, and since the agent has no interest in provoking outcomes that are bad, the agent's signals can typically be trusted. This situation may sometimes be true when states receive signals from very close allies. For instance, KyddFootnote 40 points out that Serbian leaders took Russian signals about U.S. resolve very seriously during the 1999 Kosovo War, since Russian and Serb preferences were closely correlated. However, in many cases this is not possible, such as in Kydd's alternative example of the Argentine military junta, which discounted American signals about British resolve during the 1982 Falklands crisis. Since the United States was a close ally to the UK, it could be expected to try to convince the Argentines to back down, regardless of whether the UK actually intended to follow through with a military conflict.
An identical logic can be applied to signals from IOs, whose decisions are driven by member states that often have a stake in the outcomes of the issues under consideration. Opposition is likely to be “noisy,” or reveal little information, as long as it comes from an IO whose members' interests are distant from a state proposing foreign policy action. On the other hand, support from such an institution is likely to be more convincing when relevant members' interests are distant from a proposing state, because the members have no interest in supporting policies that will create bad outcomes from their point of view. From the point of view of ordinary citizens, support from an organization comprised solely of allies has the opposite problem—it is expected to occur under a variety of circumstances, some of which may be prohibitively costly from the point of view of citizens.
One can derive several testable hypotheses from this logic applied to the context of U.S. public reaction to UN SC decisions. Specifically, the effect of SC support on public support for leaders and their policies should increase as the ideological distance between the United States and pivotal members of the SC increases. Second, the effect of appealing to the SC and failing to garner support on public support for foreign policy should decrease as the ideological distance between the United States and pivotal members of the SC increases. The following testable hypotheses make these claims explicit:
H1: All else equal, IO support will increase public support for foreign policy.
H2: The effect of IO support on public support for foreign policy increases as the ideological distance between the United States and the pivotal member of a given IO increases.
H3: The effect of the absence of IO support on public support for foreign policy decreases as the ideological distance between the United States and pivotal member of a given IO increases.
These hypotheses are untested in empirical research on either IOs or public attitudes toward foreign policy. Importantly, they go beyond the well-documented finding that the public generally prefers multilateralism to examine how politics within security organizations affects the public's view of multilateralism. All three run counter to the realist “null hypothesis” of no institutional effects on attitudes (apart from the possible burden-sharing benefits) and H2 and H3, in particular, distinguish the strategic information perspective from the pure legitimacy or procedural legitimacy views, as well as from the realist.Footnote 41
It is worth noting that the logic of the informational perspective echoes general arguments about “cheap talk” in other political contexts, in that the information that is conveyed by a signal depends on a receiver's perception of a sender's incentives. For the present purposes, however, the view that IO signals may inform the public, but that this effect may be conditional on some observable indicators of the IO's general preferences (as represented by those of important member states) suggests a means of empirically differentiating legitimacy- and information-based arguments. The legitimacy view suggests either (1) that public support for foreign policies should be greater when accompanied by support from important multilateral security organization, such as the UN SC, or (2) that public support for foreign policies should always be greater when a government consults such an important multilateral security organization. The strategic information perspective suggests that this effect should be conditional on whether audiences believe the organization in question is ex ante likely to support or oppose the policy. If authorization comes from an unlikely source, or contradicts the perceived biases of member states, it is likely to boost public support for foreign policy. This conditional effect may be particularly important given the voting rules of the SC. In order for a resolution to pass, only the most distant member, in terms of policy preferences, from a proposing state has to support it (this implies that the four other closer states should support), whereas only that same state has to oppose a resolution for it to fail. Thus, vetoes may be “noisy” signals of policy appropriateness, whereas authorization is a particularly useful signal to uninformed audiences.
Political Sophistication and Information
It is important to point out that the informational argument places a burden on the public, in that its response must be conditional on its general perception of the interests of foreign states within an IO. This requirement contradicts some studies of political behavior, which have often concluded that the average citizen lacks a high degree of political knowledge or sophistication.Footnote 42 Yet other studies point out that citizens, in fact, have a vested interest in developing informational shortcuts when forming opinions about policy. These shortcuts allow citizens to make informed decisions about important policy choices without having to know intricate details about that policy. The strategic information argument assumes that a similar logic may be at work when citizens interpret signals from IOs. However, several basic preconditions must be met in order for the informational logic to operate.Footnote 43
First, the public must have some perception of the general foreign policy preferences of an institution, based on its prominent member states. This may come from knowledge of historic events, contemporary media reports, or public statements by member states about their interests and about policies under consideration. Note that the informational argument does not presuppose that this perception is correct. Rather, regardless of the actual interests of member states, the more they are perceived to be inimical to a proposing state, the more likely institutional support will influence public support. The informational argument does suggest that perceptions must be systematic (within some reasonable variation); if perceptions are completely random there should be no evidence of an information transmission effect.
Second, the public must be aware of organizational decisions. This is most likely to occur during SC deliberations of prominent foreign policy issues, such as the 2003 Iraq War. In that case, the media provided extensive coverage of SC deliberations, as well as the threatened vetoes of France, Russia, and China, and the vocal opposition of other traditional European allies. For more obscure events or for organizations with a narrower scope, this condition is less likely to be met. Yet stories denoting lower scale SC activity are often reported in major national newspapers, such as the New York Times, which is a crude, but often employed measure of how aware the public may be of an event.
If these conditions are present, then decisions of multilateral organizations are likely to both influence public opinion and be conditional on public perceptions of organizational preferences. Otherwise the information conveyed by organizational activity will simply go unnoticed or be too “noisy” to systematically alter public support. Ultimately, this is an empirical question that can be uncovered with careful observation and research design.
Anecdotal Evidence of Public Support for Multilateralism
Before examining more systematic evidence, it is illustrative to consider general trends in U.S. public opinion. A preference for working through institutions is borne out in recent research that suggests that international institutions in general and the UN in particular are held in high regard.Footnote 44 Strong majorities in the U.S. support operating through the UN; support for hypothesized uses of force increases when mentioned in conjunction with authorization from the UN.Footnote 45 However, there is interesting variation in public reaction to institutional decisions, both across institutions and across time. For instance, public support is consistently higher for working through the UN as opposed to the North Atlantic Treaty Organization (NATO), which supports the claim that institutional preferences affect how institutions are perceived.Footnote 46
The UN SC is likely considered more conservative with its authorization for U.S. initiatives, given its multiple vetoing members with heterogeneous preferences, although perceptions of NATO's willingness to support the United States may vary across issues. SC authorization appeared to boost public support prior to the first Gulf War.Footnote 47 The Administration also actively publicized the authorization, suggesting recognition of the potential political boost it could provide. For instance, President George H. W. Bush appeared for a press conference in the White House Rose Garden on 17 December 1990, with ambassadors from most of the coalition countries, to note the widespread agreement and to publicize the SC resolution. In his 16 January 1991 television address from the Oval Office, the president also noted the significance of support from the SC and the Arab League. More recently, there is some evidence that SC endorsement contributed to the already high post-9/11 support for invading Afghanistan. Taken together, these facts illustrate that UN decisions generally seem important to the U.S. public and that approval is valued.
Given these trends, why did opposition from France, Russia, China, and Germany (three vetoing members of the SC and two traditional allies), beginning in the late fall 2002 and continuing after the onset of the war, not discourage initial public support for the war? The results of Gallup poll questions asked as late as March 2003 suggested that support for the war would drop dramatically if the United States did not even consult the institution.Footnote 48 Strong majorities, as high as 65 percent in some polls, indicated that the United States should only go to war if it obtained SC authorization. Yet President George W. Bush received a 13 percent increase in approval ratings following the onset of hostilities in March 2003, even without a second resolution authorizing the war.Footnote 49 Put differently, the public appeared to view SC sanction as critically important but failure to obtain it did not diminish public support.Footnote 50 If large portions of the public viewed acting explicitly through the SC as important, why did majorities support the war without this authorization?
The strategic information perspective offers several answers to this puzzle. First, the informational perspective suggests that failure to obtain authorization from a conservative institution is unlikely to affect public opinion, as it is a “noisy” signal about the merits of proposed policies. In early 2003, the SC was likely seen as such. The most vocal opposition to the U.S. Iraq policy probably came from France, a vetoing member of the SC, including French foreign minister Dominique de Villepin's 20 January declaration that “Nothing! Nothing!” justified war.Footnote 51 Also, media pundits pointed out that both France and Russia may have had economic incentives to oppose the invasion. The public thus likely viewed the failure to obtain an SC endorsement as very noisy.
Second, strategic information models commonly find that the relationship between the principal, in this case the public, and the agent, in this case the president, moderates the effect of third party signals. If the principal believes its agent holds preferences very close to its own, a noisy signal may have little effect on confidence in the agent. President Bush enjoyed moderately high approval ratings in the months prior to the war (between 58 percent and 64 percent in mid-March), suggesting some alignment between the public's preferences and its leadership's.Footnote 52 The perception of similar preferences may have made the noisy signal of SC opposition even less likely to affect the public.
Third, Voeten'sFootnote 53 view suggests that authorization (and, by extension, opposition) is important because it signals something about the likely response of the international community. However, the signal of consequences against a powerful state may often lack credibility, in that many states in the international system derive benefits from cooperating with the United States that limit incentives to impose costs in response to unilateral U.S. activity.Footnote 54 When member states cannot “credibly threaten” to obstruct policy, a signal of opposition is likely to be discounted by the public. In 2003, none of the permanent five (P-5) SC members were likely to implement direct opposition to U.S. action, although their lack of support may have raised the costs of the occupation and reconstruction effort. On the other hand, members still chose to voice opposition and threaten to veto because they faced domestic political opposition to the invasion. This tension helps demonstrates the tradeoff in the ability of security institutions to influence public opinion—those most poised to communicate the merits of policy with their authorization may be those that are least effective at influencing public opinion through their opposition.
Of course, the surge in approval may have been due to simple patriotism following the onset of the war.Footnote 55 Others have found that the presence of war is perhaps the strongest predictor of the “rally 'round-the-flag” phenomena, or the change in approval surrounding the onset of military hostilities.Footnote 56 However, the anecdotal polling surrounding the Iraq War and the general public preference for working through the UN (over other institutions) is consistent with an informational perspective of security institutions. The public should prefer policies that receive explicit authorization from relatively conservative institutions. Institutions with a more homogenous membership are less able to provide legitimacy through their support. Although there has not been systemic evidence gathered on public opinion regarding the membership and decision-making rules of these institutions, at least one 1994 poll found a strong majority of U.S. respondents (73 percent) indicating a desire to keep the veto system in the SC despite its anti-majoritarian nature.Footnote 57 This desire would be consistent with the informational view, as the veto system allows for biased preference outliers to convey policy-relevant information to audiences precisely because the most conservative vetoing member must support a policy for authorization to occur.Footnote 58
Polls prior to the 2003 war also suggest that the public, although desiring a second UN authorization, may have been predisposed to view the Bush administration's action favorably due to prior SC decisions. In a late February 2003 poll, 68 percent of respondents found convincing the argument that Resolution 1441, which promised “severe consequences” if Iraq did not comply with previous resolutions on weapons inspections, provided the necessary approval for war. Thus, the public may have been convinced that institutional support had already been proffered. However, in light of the publicity surrounding the administration's efforts to secure a resolution firmly authorizing an invasion, such as the wall-to-wall media coverage afforded Colin Powell's infamous “smoking gun” speech to the UN, this explanation seems questionable.
Finally, one common realist explanation for a public preference for institutional multilateralism is that the public is sensitive to the costs of foreign policy and has a preference for burden-sharing. In other words, it is not authorization that matters, but the attendant lowering of costs through the acquisition of allies that influences public opinion. Given the limitations of polling data, it is difficult to determine the relative weight of these factors. However, if the public prefers to work through institutions in order to facilitate burden-sharing, researchers should observe no difference in support for nonauthorized actions undertaken with some allies and those that receive the procedural authorization of a security institution. Yet differences are often observed. For instance, prior to the 2003 Iraq War, the percentage of respondents stating that they would support the war with only a few allies was much lower when compared to war with SC authorization. During the Kosovo War, one poll showed that a full 78 percent of the public were “concerned” that the SC had not authorized the campaign, even though it was conducted under NATO auspices. Chapman and ReiterFootnote 59 present more systematic evidence, finding that rallies 'round the flag in the United States from 1946 to 2002 were consistently larger when the SC offered a supportive resolution, while the support of regional organizations or additional allies did not significantly affect the magnitude of rallies. The strategic information perspective suggests a causal argument for why large majorities of the American public prefer the United States to act through international organizations: doing so provides information about likely foreign policy outcomes due to the institution's role as a forum for bargaining amongst member states.
Research Design
This section presents a quantitative test of the above hypotheses. Ideally, one would be able to compare support for foreign policy proposals across events. Unfortunately, polling questions with comparable wording across foreign policy scenarios is not available, precluding an aggregate comparison. However, scholars have often dealt with this data limitation by examining the change in presidential approval levels during times of crisis, which has been measured frequently since the 1930s. This change is also known as the “rally 'round-the-flag” effect, a phenomena initially noted by Mueller.Footnote 60 The study of rallies, typically measured as the change in presidential approval surrounding the onset of a foreign policy event, is useful precisely because rallies are comparable across events.Footnote 61 Presidential approval is a reasonable proxy for public support for policies, particular when measured at the time of important foreign policy events, and ever since Mueller's seminal study of the rally effect, scholars have used approval ratings to study public reaction to foreign policy events.Footnote 62
Although scholars have studied rallies in a number of ways, there is relatively little knowledge about how IO activity affects the rally effect. Rallies are usually small, but tend to be larger for wars, when the opposition supports the president, when elites make public statements in support of an administration, and when the SC offers a favorable resolution.Footnote 63H2 and H3 provide additional testable predictions about the conditional nature of SC activity.
Data and Measurement
The research design extends that of Baker and OnealFootnote 64 and Chapman and Reiter.Footnote 65 The dependent variable is the short-term change in public approval of the president from immediately prior to a Militarized Interstate Dispute (MID) to immediately after the onset. The unit of analysis is therefore a MID. A MID is “a set of interactions between or among states involving threats to use military force, displays of military force or actual uses of military force.”Footnote 66 The analysis covers the years 1946 to 2001, with data on public approval of the president taken from Gallup polls. Rallies have been positive and negative, large and small. On average, rallies are barely noticeable (the mean in this sample is .419 percent). However, rallies have occasionally been very large, such as those following the September 11, 2001, terrorist attacks and the onset of the Afghanistan War (33 percent), and the onset of the 1991 Persian Gulf War (18 percent). Figure 1 displays the distribution of rallies during MIDs from 1946 to 2001.
FIGURE 1. Distribution of rallies
Data on IO activity surrounding disputes is drawn from newspaper sources and the crisis summary descriptions provided by the International Crisis Behavior Project.Footnote 67 Additional data on SC decisions is from the UN SC documentation page and data on NATO and OAS activity is available from their respective websites.Footnote 68sc authorization is coded as a dichotomous variable based on explicit resolutions that authorize the use of force. A list of this activity is included in the Appendix, but such authorization is rare, occurring only six times in this sample. The clearest cases are the 1991 Gulf War, the 1950 Korean War resolution, the 1994 Haiti resolution, and the 2001 resolution authorizing an American response to the September 11 attacks and condemning the harboring of the Taliban. The other cases are a resolution in 1992 to authorize the policing of no-fly zones in Iraq and a 1992 resolution authorizing peacekeeping involving U.S. troops in the Yugoslav Civil War.Footnote 69 However, the United States formally submitted a resolution, asked the Council to discuss an issue, or convened the Council to discuss an international dispute fourteen times during the period under study. These cases did not always produce an SC resolution. In some instances, such as during the Cuban Missile Crisis, the United States convened the SC to discuss the situation, but a formal resolution was prevented by the opposing positions of the Soviet Union and the United States.
The record for regional organization activity is similar; a regional organization participated in or authorized the use of force in only six instances. The United States has rarely appealed to these organizations for support. As examples, the United States garnered regional organization support prior to the 1999 Kosovo intervention and the 1965 intervention in the Dominican Republic, although the Dominican Republic intervention is not included in the MID dispute dataset. In the other six cases, the OAS, NATO, or SEATO backed the United States through a resolution or public statement, but there is not clear enough evidence to code these instances as formal appeals in which the United States publicly asked for organization support or authorization for action. In several cases, members of a regional organization did contribute troops under regional organization auspices to U.S. efforts. In the 1962 Pathet Lao episode, for instance, SEATO troops were moved to the border of Thailand at Thailand's request. During the 1961 Berlin Wall crisis, NATO troops were stationed in West Berlin and NATO publicly backed the U.S. position. In the 1955 Costa Rica and Nicaragua dispute, the OAS convened at Costa Rica's request and an OAS fact-finding mission found Nicaragua at fault in the dispute, prompting the United States to send military aid. Prior to the 1983 U.S. invasion of Grenada, CARICOM authorized sanctions against Grenada and asked for U.S. intervention. The other two disputes coded as having regional organization involvement are NATO operations in Kosovo and NATO participation in the 2001 Afghanistan War. In the following analysis, I include a measure of whether members of a regional organization participated with the United States under organizational auspices, although I focus primarily on SC appeals and decisions, as these have occurred with more frequency and exhibit more variance in their effects on rallies.
To examine the conditional hypotheses H2 and H3, I include a measure of the similarity of preferences between the United States and the SC in a given year. The best existing measure of foreign policy similarity is Signornino and Ritter'sFootnote 70“S” score, which measures the absolute distance between two country's aggregate voting records in a multidimensional issue space.Footnote 71 This measure ranges from −1 (most distant) to 1 (perfect agreement).Footnote 72 The data are from Gartzke and JoFootnote 73 and updated with data from Voeten.Footnote 74 Given that the pivotal member of the Council, from the point of view of the United States, is that member least likely to support U.S. foreign policy proposals, I take the minimum of this score in a given year as an indication of the distance of the Council from the United States. With China and Russia holding vetoes in the Council, one might think that this measure varies little throughout the sample, particularly during the Cold War. In practice, however, this measure varies from −1 in 1951 and 2003, to .009 in 1972. The empirical prediction for these years would be that the United States was most likely to try to acquire SC support in 1951 and 2003 and was least likely to attempt to acquire SC support during the 1970s, although the threat of a Soviet veto during this time may have made the potential informational benefit of SC support an irrelevant factor. In other words, even though authorization was quite rare during the Cold War, this period provides important information for the study precisely because H3 predicts that the failure to obtain authorization when it is expected to be difficult will not diminish rallies. Since the Soviet veto posed a large obstacle to the passage of U.S. sponsored resolutions during the Cold War, the theoretical expectation would be that rallies during this period in which the United States consulted the SC will be no different than those in which the United States did not approach the Council. Note that this is a different expectation than the procedural legitimacy view, which would predict larger rallies in instances in which the United States consulted the Council—the “procedurally appropriate” course of action.Footnote 75
I test H2 by interacting S with sc authorization. I test H3 by including the original S. The joint interpretation of the interaction and original terms allows a comparison of cases of support and no support conditional on similarity to the pivotal member of the Council.
Control Variables
As noted above, an alternative realist argument is that organization activity might strengthen rallies because of burden-sharing. The support of an international organization signals to the public that the United States will likely receive allied support, making a rally more likely. Although the informational account and an alternative burden-sharing account may be difficult to separate empirically, I attempt to do so by controlling for the number of allies on the side of the United States in a dispute. As a robustness check, I substituted measures for the number of major power allies and measures of the aggregate and average national capability scores of allies. These measures did not change the substantive results. I therefore report the models with the simpler count of allies.
The other control variables follow Chapman and ReiterFootnote 76 and Baker and Oneal.Footnote 77 First, I control for characteristics of the dispute that might influence public opinion. These characteristics include the hostility level of the MID (ranging from 1 to 5), and the severity of the conflict, which is an index composed of the number of actors in a dispute, the extent of great power involvement, the salience of the issues at stake, the level of violence during the crisis, and the heterogeneity of the actors in terms of military, economic, and cultural differences.Footnote 78 I also control for whether the United States was an originator of the dispute and if the aim of the use of force was to alter the international environment to further vital national interests (revisionist), which are jointly thought to be an indicator that the public will be more likely to support foreign policy actions because it perceives the United States to be protecting vital interests. Such instances also involve the administration “going public,” an important component of the “opinion leadership” thesis.Footnote 79 Finally, I include a dummy measure of whether the dispute was a war, as others have found that wars are the single largest predictor of rally size. To capture the possibility of public “war fatigue” I include a dummy measure of whether the dispute took place in the midst of an ongoing war.
Second, I control for domestic factors that might influence public opinion trends. Following previous literature, I include a measure of whether the New York Times (nyt) ran stories about the dispute on the front page (2), not on the front page (1), or whether no such story appeared in the Times (0), as one precondition for a public reaction to foreign events is that they are aware of it.Footnote 80 I also control for whether the opposition party (bipartisan support) supported the administration (1), was neutral (0), or publicly opposed administration proposals (−1). The opposition may play an important role in opinion leadership,Footnote 81 and moreover may play a similar informational role as international organizations. In fact, the public may take opposition statements of support as particularly informative because the opposition faces incentives to point out the flaws of administration policies.Footnote 82 Related to the opinion leadership hypothesis, official statements (administration statement) by the president or high-ranking members of the administration may influence public support. I therefore include a dichotomous variable indicating whether the president or administration officials made official public statements regarding the dispute.
Finally, to control for the economic and electoral climates at the time of a dispute, I include the public's business confidence during the next twelve monthsFootnote 83 and a measure of the number months until the next election. I also include the prior level of public approval (prior popularity) of the president because rallies will likely be smaller when approval is already high. In other words, the inclusion of this variable corrects for bias that might be introduced because of the fact that unpopular presidents have room for major improvement.
Analysis
The analysis is performed with a truncated regression estimator. This procedure generates results similar to ordinary least squares (OLS) but is a more appropriate estimation technique given the censored range of the dependent variable.Footnote 84 The standard errors are estimated as robust standard errors clustered by presidential administration to control for unobserved heterogeneity in approval trends across presidents.
First consider Table 1. The data support H1 and H2, in that IO support increases rallies and that this effect decreases dramatically as the ideological similarity between the IO and the United States decreases. The coefficients on the interaction term and its components cannot be interpreted independently.Footnote 85 The coefficient for UN authorization should be interpreted as the effect of UN authorization when the S score is equal to 0. In other words, the predicted rally when the S score is equal to 0 is much smaller than the mean. This supports H2, as the logic of strategic information transmission suggests that authorization is uninformative as the S score reaches 0, close to the maximum value in this sample.
TABLE 1. Determinants of rallies, 1946–2001
Notes: NYT = New York Times. Significance levels: ‡ 10%; * 5%; **1%.
The findings also support H3—that the failure to obtain multilateral approval has a much smaller effect on rally size as the ideological similarity between the IO and the proposing state decreases. The coefficient for the S variable should be interpreted as the effect of the S score when the United States does not obtain UN authorization (either because the UN was not approached or the UN was approached and the United States failed to obtain authorization). When the United States does not obtain UN authorization, an increase from −1 to 0 in the similarity score (roughly from the minimum to the maximum in this sample) decreases rallies by about 2 points, but the effect is statistically insignificant. This lack of significance is predicted; the strategic information perspective suggests that failure to obtain authorization should only affect rallies if the similarity score is positive. When the S score is less than 0, the informational perspective predicts no effect of a failure to obtain authorization, and thus a coefficient that is statistically indistinguishable from 0.
What about the joint affect of authorization and similarity? The interaction term shows a strong negative and statically significant coefficient; suggesting that when UN authorization occurs and the interaction term is “switched on,” positive movement in the similarity score (towards more similar) reduces rallies. Rallies with UN authorization are only larger than average when the pivotal member is ideologically distant from the United States. This provides strong support for the informational rationale for IO legitimacy. Figure 2 displays the marginal effect of UN SC authorization at various levels of affinity (S score) with the SC.Footnote 86 Clearly, the effect of authorization on rallies decreases as similarity increases: foreign policy actions that receive authorization from a less conservative institution receive similar rallies to those that do not receive authorization from an IO.Footnote 87
FIGURE 2. Marginal effect of UN authorization by affinity with the Security Council
Note: Dashed lines give 95 percent confidence interval.
Notably, simply appealing to or consulting the SC has no discernable effect on rally size independent of receiving support. This is consistent with the informational perspective, as appeals without receiving support may be discounted. However, this finding is potentially at odds with the claim that public support increases when leaders simply follow procedural appropriateness. Rather, the signal and the ideological preference of the forum appear to each be important determinants of how the public will respond. The public also does not seem to punish the president for not seeking multilateral approval, as there is no discernable difference in rallies in which the SC was consulted and those in which it was not. Finally, these findings seem to be at odds with claims that neutrality is always a desirable property for IOs. If public support of the president during military disputes is a valid indicator of whether the public views a particular policy as worth supporting, then these findings suggest that public support is actually enhanced by approval from a more biased SC.
The other control variables mostly perform as expected. Prior popularity reduces the size of rallies. Bipartisan support slightly increases rallies, as does an official Administration statement and greater consumer business confidence. Wars and disputes in which the United States was the originator and had revisionist aims received larger rallies. Major power opposition tended to reduce rallies, consistent with a view of the public as preferring low cost disputes. Interestingly, however, a greater number of allies tended to slightly reduce rally sizes, which contradicts the burden-sharing hypothesis. This effect is not strong but is statistically significant.
Additional Tests
Although the data support H1, H2, and H3, there are several limitations to the analysis. First, it is possible that the role of the SC as well as public perceptions of it have changed since the end of the Cold War. Several scholars have suggested that the organization gained legitimacy after 1990, and particularly after the successful authorization of the 1991 Persian Gulf War.Footnote 88 The basic argument is that the SC became more active in the post–Cold War period because the superpower-induced deadlock that characterized earlier periods was alleviated, and citizens afforded it more legitimacy in the aftermath of this stalemate. To address the concern that the small number of post–Cold War authorizations drives the results, Model 2 in Table 2 displays estimates restricted to the Cold War sample only. The variable sc authorization and its interaction with the S score is dropped from the model when estimated on only the Cold War cases because there is only one instance of a clear authorization—the Korean War. However, the S score has a highly significant and negative coefficient in this sample (−4.431, p < .05), indicating that as the S score became more positive—indicating greater preference similarity between the United States and SC—the absence of authorization tended to decrease rallies during this period. This is exactly what the informational argument predicts: when a government fails to obtain authorization from an organization that is closer in its preferences, it raises questions about the how appropriate a foreign policy action is. As S increases, the failure to get external approval from the SC should raise a flag of caution with the public, while failure to get approval should have less effect as S becomes more negative.
TABLE 2. Robustness tests
Notes: NYT = New York Times. Significance levels: ‡ 10%; * 5%; **1%.
A second concern is that given the small number of authorizations in this data, the findings may be driven by several outliers. Figure 1 displayed the distribution of rallies, with the post-September 11/Afghanistan War rally clearly an extreme outlier. Model 3 in Table 2 displays the estimation when this outlier—the post-September 11 rally—is excluded from the analysis. The results are robust to excluding this case.Footnote 89
Notably, the two largest rallies—the 33-point 9/11 rally and the 18-point Persian Gulf War rally—occurred when the Council was relatively distant ideologically from the United States, with S scores of −.52 in 1991 and −.51 in 2001, which supports the informational hypothesis. The theoretical appropriateness of excluding the outliers may therefore be questionable. This case likely contains important information for testing the central argument—that rallies are largest when authorization occurs, only if the SC is viewed as relatively distant.
Finally, one might wonder if pooling all MIDS is appropriate, given that the public reaction to very minor disputes is likely to be muted. I therefore reran the estimation excluding MIDS with the lowest two hostility levels, which involve only threats to use force and minor displays of force. Model 4 in Table 2 displays the results of this analysis. The coefficients for UN authorization and the interaction term are similar in size and significance to the coefficients estimated for the full sample. The elimination of minor disputes does not dampen the results.
Conclusion
Recent events have refocused attention on the question of the sources and implications of international legitimacy. The UN SC has been at the center of these debates. Despite the traditional view of multilateral security institutions as “toothless,” recent work suggests that SC decisions matter because they confer legitimacy or signal information about policies to key audiences. However, the symbolic legitimacy and strategic information approaches suggest different empirical patterns. The symbolic legitimacy account suggests that legitimacy should be relatively fixed or slow-changing, whereas the strategic information account suggests that behavior consistent with legitimacy—such as behaving as if an institution's decisions should be trusted and followed—is likely to be conditional on the observer's beliefs about the incentives of actors driving those decisions. In the context of the SC, those actors are key veto-wielding states. Thus, the reaction to SC decisions should be conditional on audiences' beliefs about the interests of those states.
This article presented the first test of two hypotheses, based on the logic of strategic information transmission. The hypotheses predicted that the effect of SC decisions on U.S. public opinion is conditional on public perceptions of the Council. The primary test of these hypotheses yielded strong support even after controlling for a range of commonly included determinants of presidential approval. This evidence largely remains even after excluding several outlying cases, focusing on only disputes of higher hostility level, and limiting analysis to the Cold War subsample. These findings lend considerable support to the informational theory. The potential benefit of appealing to an unbiased institution is clear from the above evidence. On one hand, the added boost in public support dramatically increases as an institution becomes more biased against authorization. However, a public backlash is less likely to occur due to IO opposition as an institution becomes more biased. The risk-return tradeoff is straightforward: it is less risky and more of a return to appeal to multilateral security IOs when they are biased against authorization.
This evidence has implications for larger theoretical debates about the influence of institutions on state behavior. The realist tradition suggests that IOs matter least in security affairs, where power politics is thought to dominate. These results suggest that security IOs can matter in an unexpected way (from the realist point of view)—by influencing public opinion, which can in turn constrain or influence leaders. This suggests that more attention should be focused on how the presence and behavior of IOs can affect statecraft through indirect channels, such as the anticipation of public reaction to IO decisions. The constructivist tradition has offered a different interpretation of the effects of IOs and legitimacy in world affairs. According to constructivists, IOs can enhance legitimacy because it is procedurally appropriate to consult IOs prior to conducting foreign policy. The analyses above, however, show that the effect of IO authorization is, in fact, conditional on public perceptions. The legitimacy effect, in other words, may vary according to the prevailing view of an IO's preferences. This is a worthwhile addition to both constructivist and rationalist ideas about institutional influence.
Finally, the evidence provides support for the strategic information approach, despite the heavy burden this places on the attention and sophistication of domestic audiences with regards to international events. In general, there is little micro-level evidence of how audiences might respond to different types of international activity, although behavioralists have traditionally found that citizens are ill-informed about the specifics of international politics. Recent experimental work goes far in addressing this lacunae, in that experimental approaches may control stimuli in order to determine whether the average person reacts in ways predicted by many theories that involve domestic constraints.Footnote 90 Future work should continue to fill this gap in order to illuminate how IOs might indirectly influence leaders through the channel of public opinion.
Appendix
Appendix List of Consultations in Sample
Note: *Not included in analysis due to missing polling data.
