Support from Above: International Organizations, Summits, and Leader Survival

Abstract

While existing literature indicates a variety of political consequences caused by IO membership, little attention has been paid to IOs’ impact on leaders. This article argues that IO membership has a positive effect on leader survival through two major mechanisms: IOs carrying information and IOs raising national well-being. The beneficial effect of IOs on leader survival, moreover, is stronger for IOs that regularly or periodically hold leader summits. This is not only because these IOs are more important, but also because IO summits increase leaders’ visibility and serve as an endorsement for leaders’ power-consolidating behavior. Using data on leader failures and IO membership, this article shows that IO membership has a negative effect on leader failures, and this effect is much stronger for IOs that hold regular summits.

Bien que la littérature existante indique la présence de diverses conséquences politiques à l'appartenance aux OI, l'on s'est encore peu intéressé aux répercussions des OI sur les dirigeants. Cet article affirme que l'appartenance aux OI s'accompagne d'un effet positif sur la survie d'un dirigeant par le biais de deux mécanismes importants : la transmission d'informations par les OI et l’élévation du bien-être national grâce aux OI. En outre, l'effet bénéfique des OI sur la survie des dirigeants se renforce pour les OI qui organisent régulièrement ou à intervalles réguliers des sommets de dirigeants. Cette relation s'explique par le fait que ces OI sont plus importantes, mais aussi parce que les sommets des OI accroissent la visibilité des dirigeants et font office de soutiens du comportement de consolidation du pouvoir d'un dirigeant. À l'aide de données sur les échecs de dirigeants et sur l'appartenance aux OI, cet article montre que cette dernière a un effet négatif sur les échecs des dirigeants et que cet effet se trouve fortement renforcé pour les OI aux sommets réguliers.

Si bien la literatura existente indica una variedad de consecuencias políticas causadas por la membresía de las OOII (Organizaciones Internacionales), se ha prestado poca atención al impacto que ejercen las OOII sobre los líderes. Este artículo argumenta que la membresía de las OOII tiene un efecto positivo sobre la supervivencia del líder a través de dos mecanismos principales: El hecho de que las OOII proporcionan información y el hecho de que las OOII aumentan el bienestar nacional. Además, el efecto beneficioso de las OOII con relación a la supervivencia de los líderes resulta más fuerte para aquellas OOII que celebran cumbres de líderes de forma regular o periódica. Esto se debe, no solo, a que estas OOII son más importantes, sino que también se debe a que las cumbres de las OOII aumentan la visibilidad de los líderes y sirven como respaldo para el comportamiento en materia de consolidación del poder por parte de los líderes. Utilizamos datos sobre los fracasos de los líderes y la membresía de las OOII y demostramos que la membresía de las OOII tiene un efecto negativo sobre los fracasos de los líderes, y que este efecto es mucho más fuerte para las OOII que llevan a cabo cumbres de forma regular.

Introduction

Recent decades have witnessed remarkable growth of international organizations (IOs), with purportedly several hundred intergovernmental organizations (IGOs) in existence. Despite a backlash against international integration in recent years, as exemplified by Brexit, countries are generally interested in becoming members of various IOs. What are the consequences of IO membership? While the existing literature shows that membership in IOs generates a variety of favorable outcomes, such as a higher level of democracy and less frequent intrastate conflicts, the impact on the leader—oftentimes the decision-maker who decides whether to join or remain in an IO—is rarely studied.

This article argues that IO membership provides some benefits to leaders and thus has a positive effect on leaders’ political tenure. While IOs vary in their sizes, structures, and functions, I further argue that IOs that hold leader summits regularly have a stronger effect on leader survival. This is because IO summits provide a platform for leaders to increase their international visibility, to engage with other leaders, and to show their determination to consolidate power.

Using data on leaders and IOs in 165 countries from 1975 to 2014, this article finds that IO membership has a negative effect on leader failures, meaning that IO membership helps political leaders survive. Moreover, the helpful effect of IOs on leaders is much stronger when we focus on IOs with leader summits. A two-stage instrumental variable analysis shows that the result is not driven by endogeneity. Also, by disaggregating IOs by their functionality, this article finds that IOs help leaders collectively.

Scholars of international relations (IR) have different views on IOs. While many believe that IOs mainly reflect stakeholders’ interests and are maneuvered by powerful countries to implement their favored policies (e.g., Stone 2004; Copelovitch 2010; Kilby 2013; Lim and Vreeland 2013), some point out that IOs are characterized by centralization and autonomy (Abbott and Snidal 1998; Barnett and Finnemore 2003), can enhance democracy (Keohane et al. 2009), and can promote peace (Russett et al. 1998; Oneal and Russett 1999). This article contributes to the literature on the political effects of IOs by introducing another domestic consequence of IO membership—longer leader tenure.

In what follows, I first review existing literature on leader survival and the political consequences of IOs. Then I provide a theory on the effect of IO membership on leaders’ survival prospect. The section that follows proposes a research design to test the hypotheses. The empirical results are presented in Section 5. The final section concludes.

Leader Survival and International Institutions

Over the past two decades, scholars of IR and comparative politics have developed an interest in political leaders. By considering political leaders as a unit of analysis, this literature relaxes the conventional IR assumption that the state is a unitary actor. According to the classic selectorate theory developed by Bueno de Mesquita et al. (2005), the tenure of political leaders is affected by two key institutional factors: the size of the selectorate and the size of the winning coalition. The former refers to the pool of people who have the ability to choose the leader, and the latter is a subset of the selectorate whose support would enable the leader to stay in power. In other words, leader survival is determineddomestically.

While leaders’ tenure is essentially shaped by domestic factors, external forces sometimes influence leaders’ survival prospect, either directly or indirectly. Leader changes may be imposed directly by foreign governments or by domestic opposition with foreign support, for example, the 1953 Iranian coup that deposed the then prime minister Mohammad Mosaddegh and the removal of the Iraqi president Saddam Hussein in 2003. Even without direct foreign efforts to overthrow the leader, the prospect of leader survival may be indirectly influenced by other outside factors, such as foreign aid or remittances (Kono and Montinola 2009; Licht 2010; Ahmed 2012), economic sanctions (Marinov 2005; Escriba-Folch and Wright 2010; McLean and Radtke 2018), international conflicts or terrorism (Chiozza and Goemans 2003, 2004; Park and Bali 2017), and diplomatic visits by foreign powers (Malis and Smith 2021).

Despite this growing attention to external determinants of leader survival, little is known about how international institutions affect leaders. Existing literature that looks at the effect of international institutions on leaders mainly focuses on bilateral or multilateral agreements. For example, Hollyer and Rosendorff (2012) find a helpful effect of preferential trade agreements (PTAs) on leader survival; Mazumder (2016) and Arias et al. (2018) indicate a similar effect of bilateral investment treaties, particularly on autocratic leaders. How international institutions in general and IO membership in particular influence leader survival is rarely explored.

This lack of research is particularly curious because the literature shows that membership in IOs has numerous consequences on political regime or state behavior. An important strand of this literature finds that IOs have a democratic effect; i.e., IO membership helps democratic transitions and promotes democratic consolidation (Pevehouse 2002a,b; Keohane et al. 2009; Poast and Urpelainen 2015). Another significant strand of the literature examines the effect of IOs on interstate conflicts, in which the results are inconclusive: While many find that IO membership or IGO networks have a pacifying effect in terms of reducing the incidence or the duration of militarized conflicts (Russett et al. 1998; Oneal and Russett 1999; Dorussen and Ward 2008; Shannon et al. 2010; Lupu and Greenhill 2017), others contend that IOs may instead stimulate conflicts, at least low-severity conflicts in emerging states (Chan 2005; Fausett and Volgy 2010).¹ In addition to promoting democracy and peace, IO membership may produce other favorable consequences, such as human rights practices (Greenhill 2010) and improved risk ratings (Dreher and Voigt 2011). IO membership also leads to positive economic outcomes. Dreher et al. (2015), for example, find that investment-related IOs help a country attract FDI. Baccini and Kim (2012) show that membership in IGOs, whether economic IGOs or non-economic, reduces the level of protectionism. While some studies on IOs implicitly or explicitly suggest that IOs help leaders to survive,² to the best of the author's knowledge, no study has empirically tested the effect of IO membership on leaders’ political survival. This article fills this gap by systematically examining how IO membership affects leader failures.

IOs, Summits, and Leaders

This article argues that in general IO membership has a positive effect on leader survival, which works through two major mechanisms: IOs as an information conveyor and IOs increasing national well-being. First of all, extensive literature suggests that participation in international institutions conveys information to international actors (such as foreign investors) and/or domestic audience (e.g., Keohane 1984; Stein 1990; Milner 1997; Simmons 2000; Dai 2002; Chapman 2007, 2011). This informational function is helpful to leaders’ survival prospect, because it reveals leaders’ credible commitment to the citizens. Without perfect information, citizens may not be able to discern whether their leaders are engaging in adverse behavior such as rent-seeking or not. As Mansfield et al. (2002, 479) state, “[v]oters…face an informational problem in their attempt to monitor politicians[.]” IOs offer such information to domestic audience by signaling the leader's commitment to following international norms or rules. Citizens who receive this signal are thus more likely to believe that their leaders are doing their job and serving the public interests. This is especially so for countries under democratic transitions, as leaders of these countries need IOs to signal their commitment to democratic reforms (Mansfield and Pevehouse 2006, 2008; Poast and Urpelainen 2015). The signaling effect is also important for authoritarian leaders because they need domestic audience and external actors to believe they are unbiased and reliable (Fang 2008; Fang and Owen 2011).

One may wonder why citizens buy information from IOs. This is because, in general, the public trusts IOs more than their political leaders. Data from the Asian Barometer Survey, for example, show that only 38.6 percent of the respondents said they have “a great deal of trust” or “quite a lot of trust” in their leaders (the President or the Prime Minister). However, when asked about their impression of four IGOs: the European Union (EU), the United Nations (UN), the International Monetary Fund (IMF), and the World Bank, 62.6 percent, 69.4 percent, 64.7 percent, and 68.8 percent of them have a positive view.³ The Eurobarometer conducted in 2021 also shows that 49 percent of the Europeans trust the EU—a historic high since 2008—but only 36 percent trust their national governments.⁴ According to the 2016 Latino Barometer survey, similarly, only 28.2 percent of the respondents have a lot or some trust in their national governments, whereas 73.1 percent, 60 percent, and 59.6 percent of them have trust in the UN, the IMF, and the World Bank, respectively. These public opinion data indicate a nearly global phenomenon that individuals tend to have greater confidence in IOs than their own governments or leaders. We thus can infer that they are more likely to trust information disseminated by IOs.

Indeed, some IOs are heavily influenced by powerful countries. The literature on the IMF, for example, shows that major stakeholders of the IMF, especially the United States, have substantially influenced the IMF lending and conditionality (e.g., Stone 2008; Copelovitch 2010). This informal control of IOs by powerful members may induce negative perceptions towards an IO from people who already have a negative view of the influential member state (Johnson 2011). Despite this bias, existing studies generally show that endorsements by IOs increase public support for foreign policies or other issues such as environmental policies (e.g., Chapman 2009; Grieco et al. 2011; Greenhill 2020). This again indicates that citizens trust IOs more and implies that they are more likely to increase their support for the incumbent government when IOs endorse it.

Second, most of the IOs are established essentially to promote cooperation over economic, security, political, or social issues, which benefits the member countries in a variety of ways. Economic multilateralism can help a member state join the regional or global trade bloc, receive foreign capital and create local employment, enjoy technology transfer from other countries, and ultimately stimulate economic growth. Security multilateralism enables member states to practice collective defense, maintain regional peace, and share the military burden. Multilateral cooperation over social issues such as environmental issues helps member states overcome the collective action problem, improve social functioning, and reduce social problems. All these efforts result in a better society or raise a country's overall well-being, which will be translated into stronger support for the incumbent leader. Therefore, I believe that the leader's survival prospect will be enhanced due to the positive outcomes generated by IO membership. Even when accession to IOs (particularly economic IOs) often creates winners and losers in a country, as long as leaders are able to strategically distribute the benefits resulting from IO membership to their core supporters, their political base can be secured.

One such example is Kazakhstan. Kazakhstan is the largest country in Central Asia, and its economy is heavily reliant on the export of commodities with petroleum accounting for 60 percent of the total export. Openness is therefore beneficial to its economy and to the ruling elites. The former Kazakh President Nursultan Nazarbayev, who had been in power since Kazakhstan's independence in 1991 until his resignation in 2019, actively proposed to create an Eurasian Union, which was partly realized by the foundation of the Eurasian Economic Community in 2000 (later evolving into the Eurasian Economic Union in 2015) (Bohr 2004).⁵ Regional economic integration of Central Asia, particularly the connection with Russia, has helped the Kazakh economy to grow and enabled Nazarbayev to further secure his power.⁶ Nazarbayev's relatives have owned huge business interests in Kazakhstan, including in the oil sector, and despite being blamed for cronyism and corruption, Nazarbayev won over 95 percent of the vote in the 2011 and 2015 elections. Aware of the benefits of accession to economic IOs, Kazakhstan also applied to join the WTO in 1996 and became a full member in November 2015.

Through the above two mechanisms, IOs boost leaders’ domestic approval and enhance their survival prospect. Joining IOs signals a government's commitment to domestic reforms or compliance with international norms, and citizens that receive the signal thus trust the government better. A leader, however, will avoid IOs that erode the state's sovereignty or that hurt the leader's core capacity of controlling power. Some IOs are toothless, but some IOs set rules or standards that would constrain their members. Member states that deviate from the rules may be punished or sanctioned, which governments are generally sensitive to. Thus leaders will selectively enter IOs that bring more benefits than harm. As Mansfield and Pevehouse (2008, 270) write, authoritarian states “have reason to enter IOs only if the prospect of punishment is very low or the likely sanctions for violating rules are mild.”

In recent years, due to events such as Brexit and the U.S. withdrawal from the Paris Agreement, scholars have increasingly paid attention to state withdrawals from IOs (e.g., von Borzyskowski and Vabulas 2019; Walter 2021; Choi 2022). While termination of IO membership is not the focus of this article, the above discussions imply that leaders who withdraw from IOs may be punished or experience a shorter term, and real-world cases suggest that this seems true. Gambia's former president, Yahya Jammeh, had pursued anti-West foreign policy since he came into power in 1994. He withdrew Gambia from the Commonwealth in 2013, accusing the institution of being “neo-colonial,”⁷ and announced the plan to leave the International Criminal Court (ICC) in 2016. In the presidential election in December 2016, Yahya Jammeh was surprisingly defeated by the opposition challenger Adama Barrow, who later decided to cancel the withdrawal from the ICC and to rejoin the Commonwealth in 2018. Despite probably not the main reason for losing the 2020 re-election, Trump's decision to pull the United States out of the Paris Agreement was unpopular to American voters.⁸ These cases show that leaving existing IOs may cause a negative impact on citizens’ perceptions toward the leader and thus hurt the leader's survival prospect.

In sum, IO membership signals to the domestic audience that their leaders have the legitimacy to retain office, and this signaling effect will be channeled into a stronger power base for the leader and thus longer leader survival. Favorable consequences such as economic growth due to IO membership will also lift domestic support for the leader. These indicate a helpful effect of IOs on political leaders. Moreover, I expect the effect to be linear; that is, the chance of leader failures decreases as the number of IOs a country participates in increases. This is because a leader can derive some benefits (whether political, economic, or symbolic) from each additional IO, and thus cumulatively, IO memberships help enhance leaders’ survival prospect. The first empirically testable hypothesis therefore is: 

Hypothesis 1 assumes that IOs’ effect on leader survival is homogeneous, but apparently IOs differ from one another—for example in their functions, sizes, and levels of authority. While different types of IOs may affect leader survival to varying degrees, the above discussions suggest that on average IOs help leaders. As the first study that systematically examines the impact of IOs on leader survival, it is not my intention to further explore whether or how certain types of IOs affect political leaders. However, because this article focuses on leaders, I pay attention to a special event that involves government leaders and is held regularly only by a small portion of IOs—leader summits.⁹ I argue that in general IO membership has a beneficial effect on leader survival, and the effect is stronger for IOs that hold regular leader summits for three reasons.

First, IOs have different forms of decision-making and may hold various events with participants from different fields or with different ranks. Many IOs have ministerial meetings, which usually serve as the highest decision-making body (such as the WTO Ministerial Conference). Due to the high opportunity costs of leaders’ time, however, not many IOs regularly hold leader summits. Leaders’ time is valuable, and they have to devote much of their time to handling domestic affairs. Although leaders around the world have traveled more frequently than before, probably thanks to better transportation, and due to the increasing importance of cross-national cooperation, leaders cannot be present at every single IO to which their country is a member. Gathering all leaders from all member states apparently is a difficult task. This suggests that only major and important IOs are more likely to hold regular summits; otherwise, it would not be worth leaders’ time to attend.

Table 1 lists 44 IOs that hold regular or periodic leader summits and the year when the first summit started for each. As can be seen, IOs that hold summits are mainly influential and important global or regional IOs, such as the Asia-Pacific Economic Cooperation (APEC), the Association of Southeast Asian Nations (ASEAN), the North Atlantic Treaty Organization (NATO), Mercosur, and the UN. Even without leader summits, membership in these IOs allows member states to enjoy various benefits, such as access to foreign markets, cultural exchange, economic integration, and collective defense. We therefore believe that these IOs have a stronger beneficial effect on the leader through the two mechanisms discussed above. In other words, IOs that are able to hold leader summits are essentially more important than other IOs.

Second, leader summits are normally high-profile events that receive a significant amount of media attention. This high level of media exposure can enhance attending leaders’ international visibility and create positive images for them. In a recent article, Goldsmith et al. (2021) show that high-level visits as a form of public diplomacy increase the public approval of the visiting leader in the host country, and one mechanism they argue is through media coverage. In normal times, citizens may not be aware of the leader's itineraries, but media coverage is usually extensive during important summits, so citizens can easily observe their leader's behavior on the international stage from the media. While the media coverage of the leader may not always be positive, especially in free-market countries, the media may be more self-restrained when the event under the spotlight is a significant international summit where the leader represents the whole nation. Also, even without full control over the media, the government usually has the agenda-setting power that might influence the topics covered by the media. Therefore, unless the leader has surprisingly unsatisfactory performance at the summit, high media exposure often creates increasing feelings of national pride among citizens and thus may raise their support for the leader.¹⁰

Third, attending IO summits means the government leader has to travel abroad (unless the summit takes place in their country). Leaders’ official travel signals a lot of information. It indicates a country's diplomatic preferences, where this country's national interests lie, and which country it is willing to build a relationship with. The diplomatic visits by leaders of powerful countries are especially a key indicator of their foreign policy commitments and evaluations of the country being visited. Kastner and Saunders (2012), for example, use Chinese top leaders’ travel to assess whether China is a revisionist or status quo state. Lebovic and Saunders (2016) show that the travel of the U.S. President and Secretary of State is largely driven by strategic and economic interests. Because powerful countries carefully and strategically select the designation which their leaders travel to, and because of the high opportunity and reputational costs of leaders’ travel, scholars find that their leadership visits often generate economic or political benefits to the host country, such as a higher level of bilateral trade (Nitsch 2007), a reduction in military disputes (McManus 2018), and longer survival of the incumbent leader (Malis and Smith 2021).

The above discussions suggest that leaders’ face-to-face meetings with other national leaders, especially those of foreign powers, help enhance their national interests and their own survival prospect. Although meeting at IO summits is a diplomatic routine and is multilateral in nature, leaders usually take the opportunity to engage in bilateral meetings on the sidelines to discuss issues outside the formal agenda. The data compiled by Lebovic and Saunders (2016) also show that bilateral and multilateral visits by the U.S. President are highly correlated. So it is not unreasonable to expect that the helpful effect of bilateral meetings on leaders can be similarly applied to meetings at multilateral summits. After the trip to an IO summit, the leader is likely to bring home some fruitful results, such as investment or trade deals made during the summit. The citizens are also more likely to give credit to the leader since they observe from the media that the leader participates in bilateral or multilateral talks in person. So attending IO summits helps leaders to have concrete achievements and at the same time may heighten their popularity.

Appearance at multilateral summits, moreover, signifies that a leader is endorsed and supported by other countries, which is particularly important for authoritarian leaders who want to hold on to power firmly. Standing together in front of the camera with other national leaders not only conveys information to the domestic opposition, deterring the anti-government action that might otherwise be carried out, but it also signals to the world that the leader's repressive behavior (if any) is justified. A New York Times analysis, for example, blames leaders of the ASEAN for protecting Cambodia's longest-serving leader Hun Sen, who blocked opposition politicians from entering Cambodia before the 2019 ASEAN Summit.¹¹ This case suggests that IO summits can be a good timing before which leaders engage in power-consolidating behavior, since appearing at the summit later serves to display their legitimacy. Based on the above discussions, I formulate the second hypothesis: 

Research Design

This section proposes a research design to test the hypotheses. I first discuss the data and variables, and then introduce the statistical model.

Outcome Variable

To test whether IO membership affects leaders’ political survival, the outcome variable is whether there was a leader failure in a country in a given year. The unit of analysis is country-year.¹² I consider a leader change as a leader failure when it meets one of the four criteria: (1) the leader change is irregular (i.e., the leader is ousted by the military/rebels), (2) the incumbent runs in an election and loses, (3) the leader retires or resigns under severe political pressure, and (4) the successor of the leader is a challenger. The data are from the Archigos data on political leaders (Goemans et al. 2009) and the Regular Turnover Details dataset recently compiled by Licht (2022). The former provides information on all leaders and whether each leader change is regular or irregular, and I rely on the latter to code whether each regular leader change is a leader failure or not.¹³ The time period under investigation is from 1975 to 2014, and the sample includes 165 countries. A list of countries that are included in the analysis can be seen in the Online Appendix.

Explanatory Variables

The key explanatory variable is the number of IOs to which a country is a member in a given year. The data are taken from the International Governmental Organization (IGO) Data Version 3 (Wallace and Singer 1970; Pevehouse et al. 2020), which provides information on IGO membership from 1815 to 2014. An IO is defined as an IGO when it has at least three member states and has an indication of institutionalization, such as headquarters or permanent staff (Pevehouse et al. 2004). I focus on existing IO membership rather than new IO membership because the helpful effect of IOs should be persistent over time. During the time period under investigation, France is the most active IGO participant, being a member of 125 IGOs in 1999 and of 123 IGOs in 2000 and 2002, followed by the Netherlands in 1993 and Sweden in 1999, both having 108 IGO memberships.

One may argue that counting the number of IO memberships is problematic because it assigns equal weight to each IGO. As argued above, the goal is to examine on average how IOs influence leader survival, so it is not this article's intention to differentiate between various types of IOs. However, to test Hypothesis 2, I focus on IOs that hold regular summits, and contend that they are not only of more importance than other IOs, but also have a stronger helpful effect on leader survival. To code whether an IO has regular summits or not and, if so, when the first summit started, I checked each IO's official website, media news, and other reliable open sources. The coding was done by two research assistants separately and cross-checked by the author. We choose to be conservative and assume that an IO has never held summits when no information was found. This also fits the theory because one of the reasons I argue why IO summits help leaders is through extensive media coverage. If no proper information can be found from the media or any other open sources, the influence of an IO summit (if held) is probably very limited. When we could not find information on the first year when the summit started, we simply used the founding year as the starting year. IOs that we determine have regular summits, and the starting years are listed in table 1.¹⁴ The second explanatory variable is the number of IOs that hold regular summits to which a country is a member in a given year.

Control Variables

I include a battery of control variables that may affect leader failures. The logged value of GDP per capita (in constant 2010 US dollars) is used to test whether economic development helps prolong a leader's survival. Economic growth is the growth rate of annual GDP, which measures the short term economic performance. Foreign aid may serve as external support for the incumbent (Kono and Montinola 2009; Licht 2010), so I control for foreign aid, which is the logged value of the total net official development assistance and official aid (in constant 2018 US dollars) received by a country in a given year. The data for the above variables are from the World Bank's World Development Indicator database.

The standard Polity index is included to control for regime type, as authoritarian leaders tend to survive longer than democratic leaders. In democracies, leader changes usually occur after elections, so I include a variable to indicate whether there was an executive or legislative election held in a country in a given year. The data on both types of elections are from the Database of Political Institutions (Beck et al. 2001; Keefer 2010). Internal threat is a weighted sum of eight forms of domestic conflicts: assassinations, strikes, guerrilla warfare, government crises, purges, riots, revolutions, and anti-government demonstrations. This variable is very important since a leader's survival can be in serious danger when such anti-government activities prevail. The data are from the Cross-National Time-Series Data Archive (Banks and Wilson 2014). I also control for oil production (in thousand barrels, logged) in a country-year, as both oil revenues and foreign aid represent unearned income to the leader that helps maintain regime stability (Morrison 2009). The data on oil production are from the BP Statistical Review of World Energy.¹⁵

A leader's age is also controlled for to test whether seniority affects leader change. Lastly, a time period indicator for the post-Cold War period is included because during the Cold War, superpowers may support some country leaders due to geopolitical concerns.¹⁶ All the explanatory and control variables, except for election, leader's age, and post-Cold War, are lagged one year behind the outcome variable to avoid the simultaneous effect or reverse causality. The summary statistics of all the variables are presented in the Online Appendix.

Statistical Model

The outcome variable is a dichotomous indicator of leader failures, and I utilize a logit model with country-fixed effects.¹⁷ The inclusion of country fixed effects makes the model a within estimator, which enables us to test the hypothesis that the more IOs a country joins, the less likely that its leader will fail.¹⁸ To model temporal dependence, I include the cubic polynomials for the number of leaders’ previous years in office, which makes this model a grouped survival model (Beck et al. 1998; Carter and Signorino 2010). A negative coefficient means that a leader failure is less likely to occur, which also means that the leader is more likely to survive that year.

Results

Table 2 presents the results. In Model 1, I only include election, leader's age, post-Cold War, and the cubic polynomials for leaders’ time in office as controls to keep the model parsimonious and to obtain a larger sample size. As can be seen, the coefficient for IO membership is negative and statistically significant at the 99 percent level. This suggests that the more IOs a country participates in, the less likely that the leader will be replaced, or the more likely that the leader will survive a year. Other things being equal, an additional IO membership makes the leader 4.3 percent more likely to stay in power.

One may argue that less stable and poorer countries tend to have frequent leader changes and, at the same time, are less likely to participate in IOs. Without these important confounders properly controlled for in the model, the finding of Model 1 could be spurious. In Model 2, therefore, I include other control variables, although doing so causes the number of countries to drop to 153 due to missing values for some variables. As Model 2 shows, the coefficient for IO membership remains negative and statistically significant at the 95 percent level. Other things being equal, an additional IO membership makes the leader 3.9 percent less likely to be replaced. This finding lends support to the first hypothesis.

In Model 3, I include the number of IOs that hold regular summits to which a country is a member as the explanatory variable. As can be seen, the coefficient is negative and statistically significant at the 99 percent level, suggesting that IOs with leader summits also have a helpful effect on the leader. Compared to the effect of IO membership in Models 1 and 2, the effect of IOs with leader summits is much stronger. Other things being equal, being a member of an additional IO that holds regular summits makes the country leader 24.3 percent less likely to fail. Model 4 is the full model, including all control variables, and the results remain similar. These findings lend support to Hypothesis 2 that IOs that regularly hold summits benefit leaders more than other IOs.

In Model 5, I disaggregate a country's IO memberships into those in IOs with summits and those in IOs without leader summits, and include two variables in the same model. The results show that the effect of IOs with regular summits outweighs that of IOs without leader summits. Both coefficients are negative, but only the former achieves statistical significance. It is likely, however, that countries that possess more IO memberships also have a higher probability of joining important IOs that hold leader summits, which suggests that these two variables are highly correlated.¹⁹ In Model 6, therefore, I only include IOs without leader summits to explore its sole effect. As it shows, the coefficient is negative and statistically significant at the 90 percent level, suggesting that IOs without leader summits still have a beneficial effect on leader survival. This effect, however, is not as strong as that of the aggregate IOs shown in Model 2. While in general IOs help leaders, those without leader summits are less helpful because they do not attract as much spotlight as IOs that hold regular summits do.

In addition to the impact of IO membership, table 2 indicates some factors that are important determinants of leader failures. The level of internal threat is positively related to leader failures, which makes a lot of sense since leaders are much more likely to be deposed when domestic political discontent is strong. The leader's age has a positive effect in Models 1 and 3, meaning that a leader's chance to fail increases as they get older. Lastly, leader failures are less frequent in the post-Cold War period, although this finding only appears in Model 3.

Additional Analysis

I also perform three additional analyses to test the robustness of the results. First, the hypothesis that the more IOs a country participates in, the longer its leader will survive leads us to the use of the count of IO memberships. While this measure also appears in other studies such as Sandholtz and Gray (2003), one may question the assumption that one additional IO is equivalently helpful across the range of all possible numbers of IO memberships. I thus use two other operationalizations as robustness checks. The first is a binary variable equal to 1 when the number of IOs to which a country is a member is above the mean value of IO memberships across all country-years and 0 otherwise. The other is a log-transformation of IO memberships, and I use log(IO memberships + 1) because the data include zeros.

Second, IOs vary substantially in their functions. Mansfield and Pevehouse (2008) distinguish between standards-based, economic, and political IOs, and argue that democratizing countries tend to join the first two. Ingram et al. (2005) classify IOs into general, political, economic, and social ones. Poast and Urpelainen (2018) provide a detailed classification of IO functionality, which contains four broad categories: political, economic, technical, and others. To see whether the beneficial effect of IOs on leaders is driven by a particular type of IOs, I use the data from Poast and Urpelainen (2018) and classify IOs into political ones, economic ones, and others (including technical and other IOs).²⁰

Lastly, one may argue that the findings in table 2 suffer from potential endogeneity. This is definitely a challenge, and I identify two possible sources of endogeneity in this study. First, some types of countries may be more likely to have long-tenured leaders and at the same time more likely to join IOs, which is an omitted variable bias. Second, long-ruling leaders may be more likely to engage in international cooperation, which is a reverse causality problem. I deal with the first issue by including as many important control variables as possible, such as those measuring political regime and economic conditions. To deal with the second issue, I use a two-stage instrumental variable approach. While a valid instrument for IO membership is hard to find, I use the average number of IOs to which other countries in the same region are members. Indeed, this instrument may not be perfect because arguably regional stability could also influence the density of IOs in the region as well as domestic stability and leader survival. It is, however, similarly used by other scholars (e.g., Büthe and Milner 2008) and is the best instrument I can find given the difficulty of finding a truly exogenous instrument in social science studies. So while I acknowledge the limitation of using this instrument, the instrumental variable analysis is still useful to address the endogeneity issue.

Due to space constraints, I present the results of the additional analyses in the Online Appendix, but briefly discuss the findings here. The first analysis produces similar results, although the variable on IO membership fails to reach statistical significance when it is log-transformed. This probably suggests that a linear specification (and the underlying assumption that one additional IO has an equally beneficial effect) is more appropriate. I think this fits the theoretical effect proposed in this article. The number of IOs to which countries are members varies widely across countries, as some countries are active participants and others are not. When the data are log-transformed, we assume that an increase in larger numbers (say from 110 to 120) has a much smaller effect than an equal increase in smaller numbers (say from 40 to 50), which is not an appropriate assumption for countries that already participate in many IOs. Logging (and scaling down) the data may underestimate the effect of an additional IO for these countries. A linear specification thus aligns better with the theoretical expectation that each additional IO membership matters regardless of how many IOs a country already joins. When a binary indicator for IO membership is used, the coefficient remains negative and statistically significant at the 90 percent level. The variable on IOs with summits is also negative and statistically significant at the 95 percent level, whether in a logged form or a binary form. So while the results are not entirely robust to different coding schemes for IO membership, overall they support the argument that IOs (especially those with summits) help leaders. The second analysis shows that all three groups of IOs have a negative and statistically significant effect on leader failures when they enter the model individually, meaning that all IOs help leaders regardless of their functions. When all three types of IOs are included in one model, however, none of them has a statistically significant effect. This is probably because they are highly collinear with one another, but it also suggests that IOs help leaders collectively. The third analysis shows that instrumented IO membership has a negative and statistically significant effect on leader failures. This confirms that IO membership benefits leaders and suggests that endogeneity is less of a concern in the main analysis.

Conclusion

Existing literature indicates a variety of political consequences caused by IO membership, but little research has been devoted to studying the leaders. In this article, I argue that IO membership has a positive effect on leader survival because IOs can provide information to citizens and can generate economic or social benefits. The beneficial effect, moreover, is stronger for IOs that hold leader summits regularly, as IO summits provide a platform for leaders to gain visibility and engage in negotiations with other influential leaders.

To test the hypotheses, I draw upon data on leader failures, IO membership, and IO summits. Using a grouped survival model, I find that the more IOs a country participates in, the longer the leader will stay in power. After IOs are partitioned into those with regular summits and those without, I also find that the former has a much stronger effect on leader survival. The findings are robust to a two-stage instrumental variable analysis and an alternative measure for IO memberships. I also disaggregate IOs by their functions, and find that the helpful effect is not solely from a specific type of IOs, which suggests that IOs benefit leaders collectively.

The findings of this article contribute to the literature on IOs and provide important implications. While research on IOs is extensive, the effect of IOs on political leaders is understudied. This article fills this gap by examining how collectively IO memberships help leader survival. The results provide an explanation of why countries are keen to join IOs—because IO membership has a helpful effect on leaders. The good news for international society is that the potential beneficial effect of IOs on leaders may incentivize government leaders to participate in IOs, thus promoting international cooperation. Joining or remaining in IOs, however, may also empower authoritarian leaders and thus hinder democratic development. This suggests that reform may be necessary for some IOs, especially those with leader summits, as it is important to prevent summits from simply being a big show for leaders without achieving any substance for citizens.

Author Biography

Chia-yi Lee is Professor at the Department of Diplomacy, National Chengchi University, Taipei, Taiwan. She received her PhD in political science from Washington University in St. Louis. Her research focuses on international political economy and non-traditional security issues such as terrorism and energy.

Footnotes

References