High-Level Visits as Public Diplomacy

Leaders around the world devote significant amounts of their scarce and valuable time to international travel. Recent U.S. presidents spent up to one third of their time on international trips (Malis and Smith Reference Malis and Smith2019), during which they engaged in extensive public outreach, directly addressing foreign audiences and attending public events in front of TV cameras. For example, when President Trump visited Japan in May 2019, he held a joint press conference with Prime Minister Shinzo Abe, enjoyed a traditional Japanese barbecue dinner, presented an award at a sumo tournament, and met Japan’s new Emperor Naruhito.Footnote ⁴

Although studies of public diplomacy disproportionately focus on the U.S., leaders of other countries also frequently deploy public-diplomacy campaigns. When Indian Prime Minister Modi traveled to France in April 2015, he held a joint press conference with French President Hollande, visited an Airbus facility, attended a reception at the Carrousel du Louvre, and joined Hollande to release a stamp commemorating 50 years of India-French cooperation.Footnote ⁵ During his trip to Uzbekistan in October 2018, Russian President Putin held a joint press conference with Uzbek President Mirziyoyev, laid flowers at the Monument of Independence and Humanism, and joined Mirziyoyev to launch the site-selection project for the country’s first nuclear power plant.Footnote ⁶

Public diplomacy is unlikely to be the primary reason leaders travel internationally, as closed-door negotiations with host leaders affect security (Malis and Smith Reference Malis and Smith2021; McManus Reference McManus2018) and economic (Nitsch Reference Nitsch2007) relationships and the timing and destinations of these trips are often driven by domestic factors (Lebovic and Saunders Reference Lebovic and Saunders2016; Ostrander and Rider Reference Ostrander and Rider2018). Still, given the intense constraints on their time, the fact that visiting leaders often spend significant time on public outreach while abroad suggests they believe this type of public diplomacy brings substantial benefits.

There are multiple possible mechanisms underlying these expected benefits. While the primary goal of this study is not to test any particular mechanism, in the next two subsections, we discuss plausible causal pathways to help illustrate how public diplomacy during a high-level visit reaches targeted audiences, and why individuals in those audiences might change their opinions.

We expect that visiting leaders try to influence foreign public opinion through two processes. First, visiting leaders try to increase awareness of themselves and their country among citizens in the host country. Of the many possible tools of public diplomacy, high-level visits are particularly well suited to gain the audience’s attention due to the large amount of media coverage they tend to draw. Second, through conveying positive messages and images, often focused on the relationship with the host country and its leader, visiting leaders attempt to increase the likelihood that the audience will view them (and, by extension, their country) favorably. High-level visits help leaders achieve this by affording them the opportunity to display their shared goals and values with the foreign audience. For example, leaders often express their shared policy goals with national leaders during joint press conferences, and the cultural activities leaders participate in during their visits are likely meant to signal shared values (Sheafer et al. Reference Sheafer, Bloom, Shenhav and Segev2013).

Host Leader Incentives and Media Environments

Both of the pathways described above—increasing awareness and conveying positive messages—are contingent on (1) how the opportunities for public diplomacy are arranged and (2) how much and in what ways the media cover foreign leaders’ visits. Therefore, to better understand the potentially positive effects of high-level visits on foreign public opinion, we should also consider two additional key actors that shape public diplomacy: host leaders and the domestic media.

Host leaders have the power to shape the effect of international public diplomacy because they play an important role in determining the occurrence, timing, and structure of a visit. Regardless of the relationship between the countries, it is highly unlikely that a visit occurs when either the visiting or host leader is opposed. However, neither the host nor the visitor typically has full control over the public diplomacy activities or messages associated with a high-level visit. Rather, arrangements are often the result of prior negotiation, striking a balance between the messaging goals of each country.Footnote ⁷

In addition to controlling invitations to foreign leaders, hosts can influence public diplomacy efforts through their agenda-setting power over the topics that domestic media focus on. The effect of a high-level visit is essentially the effect that the visit has on the media system and, in turn, on citizens’ consumption of the media coverage. In either market-driven or state-driven media environments, both visiting and host-country leaders try to arrange events and activities to promote specific images that are beneficial for their own country’s objectives.Footnote ⁸ Host leaders usually have their own domestic incentives to produce publicly visible and positive visits, which may shore up their own domestic support (Malis and Smith Reference Malis and Smith2021; McManus Reference McManus2018).Footnote ⁹ As invited guests of the administration in power, visiting leaders typically enjoy some level of endorsement from their host, increasing their legitimacy in the eyes of the foreign public. Therefore, in most cases, the host leader’s ability to direct domestic attention assists the visiting leader in their public diplomacy goals.

Given the mutually agreed nature of the visit and the incentives of both leaders, the media coverage of high-level visits tends to convey positive messages and images about the visiting leader. For example, during Trump’s 2019 visit to Japan, all of the national newspapers gave it front-page coverage. Major news media also frequently reported Trump’s activities in their television and radio broadcasts, social media, websites, and even in video advertisements on trains. These Japanese news outlets tended to emphasize the positive relationship between the two nations in catchy headlines, such as “Japan-U.S. Honeymoon, Built with Golf.”Footnote ¹⁰ Unsurprisingly, some news media were more critical.Footnote ¹¹ But even these critical articles often included photos and videos associated with Trump’s actual activities, which conveyed positive nonverbal messages.Footnote ¹²

In summary, our theory predicts that when public diplomacy visits occur, it is likely that the incentives of the visiting and host leaders are reasonably well aligned and news media are likely to provide some coverage of the visit. As a consequence, a visit raises awareness of the visiting leader among citizens of the host country and conveys positive messages about that leader to them.

However, this does not imply that the visiting leader lacks the ability to sometimes buck the host leader’s preferences and convey public-diplomacy messages based on their own state’s interests and goals. For example, high-level visits can also occur in the context of sharp international rivalry. Nevertheless, the public diplomacy activities may still insert the visitor’s rival messaging goals into the host country’s media (Wiseman Reference Wiseman, Melissen and Wang2019). Recent U.S. Presidents, for example, have engaged in extensive previsit negotiations with China over whether joint news conferences would be held during visits to China and, if so, whether Chinese media would broadcast them live and in full.Footnote ¹³ Chinese leader Hu Jintao carefully crafted his public-diplomacy messages to maximize their influence on local audiences in the U.S. based on regional economic factors (d’Hooghe Reference d’Hooghe2015) and for citizens in different member-states on a tour of the European Union (d’Hooghe Reference d’Hooghe, Jong Lee and Melissen2011). Even among friendly democracies, messaging goals can diverge significantly between the visiting and host leaders. For example, this is seen in Britain’s (van Ham Reference van Ham, Zaharna, Arsenault and Fisher2013) or Canada’s efforts to affect U.S. environmental policies by going “over the head of the US government” to its constituents (Henrikson Reference Henrikson and Melissen2007, 76).

Methodological Challenges

In the scholarly literature, the ability of public diplomacy to increase public support among foreign citizens is assumed, but rarely measured. The paucity of empirical research may be in part due to methodological challenges associated with estimating the effects of public diplomacy—and international visits, in particular—on foreign public opinion. The timing and targets of public diplomacy efforts are strategically driven and therefore nonrandom, which makes it difficult to interpret any estimated effects as causal. We are aware of only one study that attempts to estimate the effects of high-level visits on foreign public opinion, but it does so by controlling for a range of observable variables (Goldsmith and Horiuchi Reference Goldsmith and Horiuchi2009). Because leaders’ strategic considerations are difficult to measure, observational studies that adopt this approach are subject to bias due to unmeasured confounding factors.

An alternative approach uses survey experiments, which offer better insights into the causal link between specific messages from foreign leaders and public opinion outcomes. Existing studies find that the effects of foreign leaders’ messages are contingent on the content of the message, as well as feelings toward the leader, familiarity, and nationality (e.g., Agadjanian and Horiuchi Reference Agadjanian and Horiuchi2020; Balmas Reference Balmas2018; Dragojlovic Reference Dragojlovic2015). However, these findings are predicated on the assumption that leaders’ public-diplomacy efforts reach foreign publics in a form that is similar to what respondents experience on the survey. In real-world settings, leaders compete with many other “noises” in their efforts to improve the images of their country and themselves, including other news events and messages from competing domestic elites. If these noises attenuate the effect of public diplomacy in real-world settings, survey experiments could overestimate the effects of policy messages from foreign leaders.

Public Opinion about the Visiting Country

We measure our outcome variable using GWP (version November 23, 2018). The raw data include annual surveys administered in more than 160 countries from 2005 to 2018.Footnote ¹⁶ An important advantage of using this database is that the date of interview is recorded for each respondent. Therefore, with a set of reasonable assumptions, we can estimate the causal effect of a time-specific salient event on public attitudes by comparing respondents interviewed just before the event with those interviewed just after.

To estimate the influence of a high-level visit on public opinion, we use the following GWP question: “Do you approve or disapprove of the job performance of the leadership of (the name of a country)?” While respondents likely do not have much (or any) information about foreign leaders’ actual job performance, we assume that they use their overall favorability toward (or “image” of) the leader to answer this question. In the Supplementary Materials we provide further evidence of the close association of respondents’ assessments of a leader’s job performance and their general attitudes toward foreign leaders and the countries they represent.Footnote ¹⁷

From 2008 to 2018, Gallup asked this question—with the same format—about the leadership of 32 countries. Questions about the leadership of some major countries, such as the United States, Russia, and China, were included in all of the GWP annual studies. Other countries, such as Japan, the United Kingdom, and India, were included in most years. Several countries (e.g., Brazil, Turkey, and Iran) were included in the survey only in some years and/or in specific regions. In this study, we use questions about the job performance of the leadership of nine major countries—Brazil, Canada, China, Germany, India, Japan, Russia, the U.K., and the U.S. We selected these nine countries primarily based on their standing as a diverse set of major countries in international affairs, but also on the availability of high-level visit data and practical constraints (e.g., language ability) on gathering detailed information about these visits.

High-Level Visit Data

For these nine countries, we collected, cleaned, and standardized numerous quantitative and qualitative data concerning high-level visits by political leaders.Footnote ¹⁸ We focused on the most important and visible leader (or leaders) for each country. In most countries, this is the head of government. For China, since power is centered in the head of the Communist Party (i.e., the General Secretary, who usually also holds the ceremonial post of President), we collected visits by Xi Jinping, and his predecessor Hu Jintao. For Russia, we collected visits by both presidents (head of state) and prime ministers (head of government) because Vladimir Putin remained the de facto political leader when he stepped down from the presidency to become Prime Minister (from May 7, 2008 to May 7, 2012).

We identify the cases in which the leader of Country $ a=\left\{1,2,\dots, 9\right\} $ visited Country $ b $ ( $ \ne a $ ) during the sampling period of a particular year’s GWP (in Country $ b $ ) that included a question about the job performance of the leadership of Country $ a $ . Therefore, for each of the nine countries, we focused our data collection efforts on the years in which Gallup included the question about the leadership of the corresponding country. For example, since the question about the job performance of the leadership of Brazil was included only in the 2008–2010 GWP studies, we did not collect information about the Brazilian presidents’ travels after 2010.Footnote ¹⁹

We selected all GWP respondents in each host country interviewed within a window of 30 days before or after the first day of a high-level visit to that country.Footnote ²⁰ We removed cases in which the total number of respondents interviewed before or after the visit is less than 10. We then carefully checked all the cases using the original sources (Section B of the Supplementary Materials) and other news articles and excluded the visits for which there was no opportunity for public diplomacy activities, such as stopping over for refueling with no high-level meeting, as well as those that involved only a multilateral meeting (e.g., the G20) with no bilateral meeting with the host leader.

There are 86 visits by 15 political leaders from nine countries to 38 foreign countries that met these initial screening criteria. Table 1 shows the number of valid visits for each leader in our sample. Section C in the Supplementary Materials includes tables listing all visits, their dates, and destinations.

Table 1. High-Level Visits: The Number of Valid Cases by Visitors

Note: See Section C of the Supplementary Materails for details.

It is important to acknowledge that our sample of high-level visits is not a representative set of such visits by all leaders around the world during the period of investigation. However, it is highly unlikely that our selection of valid cases is systematically correlated with strategic considerations underlying the high-level visits. More specifically, it is implausible that the location and timing of each visit is systematically associated with the GWP sampling period. For example, the U.S. President’s travel is predicted by domestic politics as well as the trade and security relationship with destination countries (Lebovic and Saunders Reference Lebovic and Saunders2016; Ostrander and Rider Reference Ostrander and Rider2018). The timing of the next GWP survey in the host country should not be part of planning for high-level visits.

Identification Strategy

To estimate the effects, we focus on respondents who were interviewed just before or just after the first day of each visit. The unit of observation is each respondent $ i $ for each visit $ j $ . Our treatment variable is whether respondent $ i $ was interviewed after the first day of a high-level visit $ j $ ( $ {X}_{ij}=1 $ ) or before it ( $ {X}_{ij}=0 $ ). We exclude respondents who were interviewed on the first day of a visit because we cannot determine their treatment status.

The variable of interest in the GWP survey is categorical (ordinal)—whether a respondent $ i $ approves of the job performance of the leadership of a visiting country around the time of a specific high-level visit $ j $ , disapproves of it, or chooses neither.Footnote ²¹ The third category includes “Don’t know” and “Refused.” Both of these responses indicate unwillingness to choose a position either due to lack of knowledge or due to mixed and contradictory views.Footnote ²² In this case, raising awareness, adding positive information, or both could lead to a change from this third category to another category (specifically, Approve).

Although this variable is ordinal, we do not run ordered probit (or logit) regression for our analysis because, as we discuss below, it is essential to include visit-specific fixed effects. We would face the well-known incidental parameter problem (Neyman and Scott Reference Neyman and Scott1948) if we added fixed effects to a nonlinear model, including an ordered probit model. We therefore transform this variable into three dichotomous outcome variables: (1) whether a respondent approves of the job performance of a visiting leader, (2) whether a respondent disapproves of it, and (3) whether a respondent neither approves nor disapproves.Footnote ²³

There is a deterministic relationship between the treatment variable and the number of days since the first day of a visit $ j $ for a respondent $ i $ (i.e., a running variable, $ {Z}_{ij} $ ): if $ {Z}_{ij}>0 $ , then $ {X}_{ij}=1 $ . If $ {Z}_{ij}<0 $ , then $ {X}_{ij}=0 $ . However, we do not apply a standard regression discontinuity (RD) design because the running variable is categorical.Footnote ²⁴ As we narrow the “bandwidth” to enhance the validity of our causal inference, specifically for a shorter period of $ k $ days before and after each visit, the running variable takes only k × 2 distinct values. Without using a continuous variable, we are not able to satisfy the crucial assumption of continuity for RD estimation (de la Cuesta and Imai Reference de la Cuesta and Imai2016)

Instead, for each of the outcome variables, we estimate the difference of means between the treatment group ( $ {X}_{ij}=1 $ ) and the control group ( $ {X}_{ij}=0 $ ) given a small value of $ k $ . Specifically, we estimate the average treatment effect by running an ordinary least squares (OLS) regression with the treatment variable ( $ {X}_{ij} $ ) and visit-specific fixed effects. It is critical to include the fixed effects in our model because we need to compare foreign public opinion before and after the first day of a specific high-level visit. For example, to estimate the effect of Barack Obama’s visit to Denmark in December 2009, we should take the difference in the outcome variable between Danish respondents interviewed just before and just after the first day of this visit. An advantage of our data and approach is the ability to pool these visit-specific effects across many visits and estimate the overall treatment effect.

The length of the bandwidth for the analysis poses a trade-off between causal inference and statistical power. As we decrease the bandwidth, respondents in the control and treatment groups become more similar in terms of the context in which they respond to the questionnaire, although we inevitably reduce the number of observations. The longer the bandwidth, the larger the number of observations but the more likely that information environments could be different for respondents in the control and treatment groups for reasons not associated with the visit. Considering these costs and benefits, we decided to use k = 5 days. This relatively narrow bandwidth errs on the side of ensuring the validity of our causal inference.Footnote ²⁵ The number of observations within this bandwidth, 32,456, is still sufficiently large for statistical analysis.

The key identification assumption is the lack of any systematic difference between respondents in the control group and those in the treatment group. This assumption holds as long as respondents interviewed before or after the first day of a visit can be considered as-if randomly assigned within each visit. There may be some (unexpected) systematic differences between the two groups for any given case. However, when we pool all 86 cases for analysis, it becomes decreasingly plausible that confounding variables systematically affect the differences in the outcome variable between the control and treatment groups.

We test this assumption by examining the balance of basic demographic questions included in the GWP studies.Footnote ²⁶ Figure D.1 in the Supplementary Materials shows that for most variables, we find no statistically significant difference between the treatment and control groups. We later present the results of a robustness test, reestimating the effects with the imbalanced variables as controls. We also conduct a range of additional tests to further scrutinize the causal effects of high-level public diplomacy on mass opinion.

Hard Edge to Soft Power?

We first examine the relationship between hard-power resources and the effect of a high-level visit. Evidence that soft-power resources can be built independent of hard-power capabilities supports the contention that power in international relations can be “multifaceted” and “exerted more subtly and gradually” (Baldwin Reference Baldwin2016; Kelley and Simmons Reference Kelley and Simmons2019, 504). However, a high-level visit’s ability to increase soft-power resources might be conditioned by a visiting state’s hard power. This would be the case if citizens of a host country pay more attention to visiting leaders from powerful states—that is, if they view more powerful visitors as better able to bestow benefits, or cause hardship, through their policies. Similarly, host-country citizens may be more receptive to the powerful visitors’ images and messages due to a process of motivated reasoning (Little Reference Little2019). If they understand the potential for substantial inducements or threats from the visiting leader, they may see it as in their interest to take on a shared perspective with the visiting country. This greater attentiveness and openness may be enhanced by greater media attention to more powerful countries (Chang Reference Chang1998).

In short, there may be a hard edge to soft power: even without mentioning explicit threats or inducements, powerful countries may hold advantages in gaining attention and changing public opinion through their public diplomacy campaigns. This possibility poses a fundamental challenge to the soft-power framework. If hard-power capabilities are crucial to building soft-power resources, this supports criticisms that soft power is essentially subordinate to military and economic capabilities (Cohen Reference Cohen2017). Empirically, if this contention is valid, then the effect of a visit would depend on how much more powerful the visiting leader’s country is relative to the host country.

Our analysis shows limited evidence for this. We use the Composite Index of National Capabilities (CINC, Version 5.0) from the Correlates of War project to assess the power ratio between the visiting and host countries (Singer Reference Singer1988). We first split the sample into relatively equal groups, exploiting a natural break in our data: there are no cases with CINC ratios between 6 and 9. For the 45 visits with relatively low ratios (less than 6), the average is 2.9, indicating that the visiting country is on average about three times as powerful as the host. The most imbalanced dyad in this set is Germany visiting Belgium (5.97). For the 40 visits with relatively high ratios (greater than 9), the average is 47.6. The least imbalanced dyad in this group is the U.S. visiting the U.K. (9.08). Figure 5 shows no difference between the treatment effects for these two groups with low and high power ratios.

Figure 5. OLS Regression Results by the Power Ratio of the Visiting Country Relative to the Host Country

Note: The figure shows treatment effects using the respondents interviewed within five days before/after the first day of a high-level visit. Vertical bars are 95% confidence intervals. Estimates statistically significant at the 0.05 level are highlighted in black. To divide the cases into two groups, we use the Composite Index of National Capabilities (CINC) from the Correlates of War project. A ratio greater than 6 is defined as “high.”

However, when we compare cases of extreme imbalance, we do find some supporting evidence for treatment-effect heterogeneity. Specifically, focusing on the highest third of imbalanced cases, with an average ratio of 65.4, we find a substantially larger treatment effect compared with the lower two thirds (see Figure L.2 in Section L of the Supplementary Materials for full results). This includes 27 visits ranging from the U.S. visiting Egypt (14.29 ratio) to China visiting Rwanda (317.56 ratio). With a very high degree of imbalance, there does seem to be a hard edge to soft power, although we emphasize that this assessment of heterogeneous effects cannot firmly establish causation. Still, even visits in the lowest third of power imbalance result in significantly increased approval for visiting leaders. Our best theoretical interpretation from this evidence is that any advantage in public diplomacy due to hard-power capabilities would be limited to vastly more powerful visitors, with capabilities at least a dozen times greater than those of the host.

Soft Power and Foreign Coattails?

A second category of heterogeneous effects we explore relates to the role of host leaders. Specifically, we might be concerned that high-level visits are primarily directed at benefiting the host leader. To explore this, we examine the subset of cases where we think it is most likely that host leaders’ incentives are driving the effects of the visits. In particular, we consider situations in which a domestically unpopular host-country leader might seek to leverage the high popularity of the visiting leader to advance their own domestic interests, such as bolstering their electoral support or promoting their policy agenda. For example, Barack Obama argued forcefully for the U.K. to remain in the European Union during his 2016 visit prior to the U.K.’s “Brexit” referendum. His host, Prime Minister David Cameron, opposed Brexit but had an approval rating below 40%, while Obama was very popular in the U.K.Footnote ³⁴ If unpopular host-country leaders have these motivations, they could ride the coattails of the popular visitor by using the foreign leader’s visit to gain media exposure, produce more favorable images and messages, and ultimately improve their own popularity.

We assess this possibility by identifying all visits in our dataset for which the previsit job-performance approval of the host is low while the previsit approval of the visiting leader is high.Footnote ³⁵ Specifically, we separate all visits for which the previsit net approval for the host leader is negative (disapproval greater than approval) and the net approval for the visiting leader is positive (approval greater than disapproval). If less-popular hosts are seeking to capitalize on more-popular foreign leaders’ visits, we expect the approval of the host leaders in these cases to increase in response to the visits. Parallel to our main analysis, we use respondents’ answers to the question “Do you approve or disapprove of the job performance of the leadership of (the name of a country)?”

Figure 6 shows the results of this analysis. There is no evidence suggesting that high-level foreign visits translate into a boost in approval for the host leader (circle dots).Footnote ³⁶ At most, there is an increase in indifferent responses about the host leader. We note that host leaders may still believe that they or their policies would see increased popularity by grabbing onto the coattails of a popular foreign visitor—but we find no evidence to support such a belief.

Figure 6. OLS Regression Results by the Job-Performance Approval of the Host-Country and Visiting-Country Leaders

Note: Treatment effects using respondents interviewed within five days before/after the first day of a high-level visit. Vertical bars are 95% confidence intervals. Estimates statistically significant at the 0.05 level are highlighted in black. To divide cases in two groups, we use the difference between the percentage of approval and the percentage of disapproval in the control group. We define a host-country leader or a visiting-country leader as popular if the difference in approval and disapproval is positive.

Dividing our sample in this way also allows us to explore whether the positive effect of public diplomacy for visiting leaders is conditional on the ex ante popularity of the host leader. Yet we find that our main effects on visiting leaders’ approval do not appear to be conditional on the host leader’s approval: the visiting leader sees a significant increase in approval within both subsets (triangle dots). Perhaps unsurprisingly, the effect of a high-level visit by a popular foreign leader on their own approval is large. While the heterogeneous effects shown in Figure 6 should not be interpreted as causal, taken together these patterns suggest that the primary effect of public diplomacy we identified is not simply driven by the incentives of the host leader.

Soft-Power Honeymoon?

Finally, we consider whether visiting leaders’ time in office conditions the effect of their visit. On one hand, leaders with who have established track records and relatively high familiarity among global audiences may be more effective public diplomats. This would indicate that leaders’ reputation and credibility are important in generating soft-power resources. On the other hand, if people start less informed and develop their opinion based simply on the number of messages they have received on a topic (Zaller Reference Zaller1992), their less-entrenched opinions may be more malleable in response to visits by relatively new foreign leaders.

Our results indicate that new leaders have a public-diplomacy advantage. Within the first year of their tenure, leaders appear to have roughly double the effect on host-country approval levels that they have after five or more years in office (Figure 7). This “honeymoon” effect may suggest that public diplomacy’s influence relies to a considerable extent on low levels of information held by foreign citizens rather than mainly being the product of hard-earned reputations for credibility developed by experienced leaders.

Figure 7. OLS Regression Results by the Number of Years since a Visiting Leader Assumed Office

Note: Treatment effects using the respondents interviewed within five days before/after the first day of a high-level visit. Vertical bars are 95% confidence intervals. Estimates statistically significant at the 0.05 level are highlighted in black.