The Role of Pan-African Ideology in Ethnic Power Sharing

Abstract

What are the conditions under which governments form more ethnically inclusive coalitions? Previous contributions highlight strategic incentives as well as colonial and precolonial legacies as determinants of ethnically inclusive government coalitions but overlook the impact of political mobilization during the decolonization period. We argue that ideological exposure and commitment to the Pan-African anticolonial movement played a vital role in African leaders’ decisions to share power with other ethnic communities. We leverage novel data on African government leaders’ attendance at decolonization-era Pan-African conferences through a unique collection of conference delegate lists. Accounting for rival mechanisms, we find that African political elites who attended Pan-African conferences formed ethnically more inclusive government coalitions when they became government leaders. Our findings imply that the ideological influence and commitment signaled by conference attendance affected political leaders’ approach to form more inclusive governments and that ethnic coalitions have systematically unexplored legacies in the Pan-African decolonization movement.

What are the conditions under which governments form more ethnically inclusive coalitions? This question has gained academic attention because ethnic inclusiveness is robustly linked to peace and economic development.¹ Particularly in Africa, where 38 percent of armed civil conflicts since 1945 have taken place,² many of these conflicts can be attributed to ethnic divisions created or exacerbated by European colonial policies and border drawing. As a result, interest in why some governments are ethnically more inclusive than others has grown. So far, the literature has focused on how strategic elites navigate structural factors such as ethnic demographics to maximize their chances of surviving in office. In these explanations, leaders form ethnic coalitions to minimize threats from within and outside the regime.³

While these explanations play an important role in understanding ethnic power sharing, they impute little theoretical agency to African political elites and their ideological and political preferences. This stands in contrast to scholarship showing that political leaders’ background, formative experiences, and political preferences shape policy decisions.⁴

In this article we identify the decolonization period as a crucial and formative time of political learning and ideological exposure for African elites. During this time, the key ideological divide among African elites was between the Pan-African movement, which demanded rapid and complete liberation, and politicians who were close to (former) colonizers and favored more gradual approaches. Alongside rapid and complete independence, one of the Pan-African movement's core principles was African unity and the overcoming of ethnic divisions.⁵

We argue that African political elites’ exposure and commitment to the Pan-African ideological movement had important consequences for their preferences over postcolonial institutions: on gaining power, those who had been exposed to the Pan-African movement formed ethnically more inclusive government coalitions. To approximate elites’ exposure and commitment to the Pan-African ideological movement, we use attendance at international anticolonial Pan-African conferences. Pan-African conferences can proxy for ideological and political preferences in two ways. First, the conferences influenced their attendees by promoting ethnic unity as a normative ideal as well as an effective political instrument. Second, because of the Pan-African movement's stance on African unity and rejection of ethnic politics, the conferences attracted elites who were committed to ethnic inclusiveness.

To test our argument, we leverage novel archival data on African state leaders’ attendance at international conferences under the Pan-African ideological umbrella. We find that African political leaders who attended more Pan-African conferences built ethnically more inclusive governments.

The case of Kenneth Kaunda, first president of Zambia, illustrates our theoretical mechanism. Kaunda was a central figure in the Pan-African movement and its conferences. On reaching power, he consciously formed governments that represented all politically relevant ethnic groups in the state. Even before leading Zambia to independence in 1964, Kaunda emphasized national unity and an “ideology of togetherness among Africans.”⁶ After independence, when his party, UNIP, struggled to retain power, “UNIP's quest to dominate the political scene was increasingly articulated as a process aimed at national unity.”⁷ This does not mean that ethnicity did not play a role in Zambian politics at the time. Rather, Kaunda explicitly co-opted ethnic interests into UNIP, which included elites from all politically relevant ethnic groups and regionally based ethnic interests.⁸ Other African postcolonial government leaders, such as Niger's first president, Hamani Diori, did not attend Pan-African conferences before independence and were ideologically opposed to the more fundamental decolonization approach favored by Pan-African elites.⁹ Diori's ethnic group had been favored by the French under colonialism. After independence, he was appointed president by the French colonial governor.¹⁰ Diori formed an ethnically highly exclusive minority government, where only his own group was represented.¹¹

This article makes two contributions to current understandings of ethnic power-sharing in Africa. First, it demonstrates that variation in African leaders’ ideological preferences and political understandings is a crucial factor in explaining ethnic power sharing. This stands in contrast to earlier scholarship, which focuses on how office-seeking leaders respond to structural conditions, such as ethnic demographics. Second, it shows that the independence struggle and the Pan-African movement had an important impact on African politics during the postcolonial period by shaping elites’ political preferences. Thus it differs from other work on African politics that does not consider ideological factors as drivers of African political elites’ behavior.

Determinants of Ethnic Inclusion and Exclusion

Ethnic cleavages are relevant to political contestation in electoral, autocratic, and violent contexts. Indeed, the political salience of ethnic identity carries the potential for political violence in the form of coups,¹² armed civil conflict,¹³ and government repression.¹⁴ Postcolonial African states often see particularly politicized ethnic cleavages with a high potential for political violence, which are the consequence of colonial boundary drawing and the construction and exploitation of ethnic identities under colonialism.

Ethnic power sharing is argued to prevent ethnic grievances and armed ethnic conflict in states with politicized ethnic identities,¹⁵ even though ethnically inclusive government coalitions come with an increased risk of coups.¹⁶ Given the conflict-preventing effect of ethnic power sharing, scholars have turned to investigating the causes of ethnically inclusive government. While early scholarship has aimed to explain why specific groups face discrimination and political exclusion,¹⁷ a wave of subsequent scholarship has explained ethnic power sharing as a strategic tool for government leaders to navigate structural conditions such as ethnic demographics,¹⁸ colonial legacies,¹⁹ and precolonial power structures.²⁰ However, differences in individual leaders’ preferences, shaped by their experiences, political understandings, and normative convictions, have not been considered as explanatory factors for ethnic power sharing. In particular, normative and ideological factors can interact with the strategic reasoning of political actors.²¹ We address this limitation in demonstrating theoretically and empirically that ideological movements impact ethnic-inclusion dynamics.

Structural Conditions and Elite-Driven Ethnic Coalition Formation

Previous explanations for ethnic power sharing focus on government elites strategically navigating structural conditions. We identify three structural conditions that are particularly prominent in the literature: ethnic demographics, resource distribution, and colonial and precolonial legacies.

Studies focusing on ethnic demographics as determinants of ethnic coalition formation extend work on autocracies that highlights leaders’ strategic incentives for elite inclusion and coalition building.²² Leaders need to manage threats from within the regime, such as ethnically motivated coups, and from outside the regime, such as armed mobilization by ethnic groups excluded from government.²³ Scholars have considered how ethnic demographics, in particular ethnic group size, affect how leaders navigate this dilemma. Group sizes impact power balancing within ruling coalitions and the potential threat of groups excluded from coalitions,²⁴ and determine the resources needed to buy the support of group elites.²⁵ Leaders of demographically dominant groups can suppress minority groups motivated by “politics of entitlement” that establish policies of inclusion and exclusion.²⁶ Focusing on ethnic cleavages, Bormann argues that leaders have incentives to include elites representing groups that share cleavage dimensions with their own (the same language or religion, for example) to avoid having them as outside competitors who mobilize the leaders’ own ethnic community.²⁷

The second structural condition pertains to resource distribution within the population and politicians’ access to financing. Financing has been stressed as important in the context of electoral competition, where politicians have strategic incentives to form multiethnic coalitions to be electorally successful.²⁸ Arriola focuses on access to financing and argues that the opposition's chances of forming electorally viable multiethnic coalitions depends on their ability to pay off elites from other ethnic communities before elections.²⁹

The third structural condition determining ethnic coalitions is historical factors, in particular colonial and precolonial legacies. Roessler and Ohls argue that weak institutions stemming from colonial legacies lead to security dilemmas and exclusion due to the risk of coups.³⁰ Hence, colonial legacies determine the strategic environment that determines coalition building. Colonial legacies also shape the ethnic power relationships that impact coalition composition. Here Vogt highlights that ethnic cleavages are shaped by European colonizers, and then fundamentally shape inclusion and exclusion patterns.³¹ In addition to colonial factors, precolonial legacies are increasingly analyzed in the context of ethnic coalitions. Wishman and Butcher highlight the role of precolonial states in shaping current ethnic groups and having long-lasting effects on inclusion and exclusion dynamics.³² Paine focuses on precolonial state groups and argues that they have historically rooted advantages from gaining central power either during or after decolonization.³³

In sum, scholars have generated valuable insights into how structural conditions affect elites’ ability to strategically form ethnic coalitions that can unseat incumbents and ward off external threats, as well as defections from within. Our main argument draws attention to leader-specific variation in preferences for ethnic power sharing. These preferences are shaped by leaders’ experiences, political understandings, and normative convictions, which have been well studied in other contexts.

Leader-Specific Factors, Political ideology, and Elite-Driven Ethnic Coalition Formation

While leader-specific variation is less prominent in explaining ethnic coalition formation, there is ample evidence on how leaders’ preferences and beliefs shape policy decisions. In international relations, studies of leaders’ foreign policy decisions highlight the role of personal beliefs, political orientations, and experiences.³⁴ For example, Colgan and Weeks show that leaders who have led a revolution are more likely to initiate international conflict, especially if they are unconstrained by other elites.³⁵ Similarly, Horowitz and Stam focus on military service as a formative experience and find that autocratic leaders who have served in the military are more conflict prone.³⁶ Leaders’ backgrounds and experiences have also been found to affect other policy outcomes, including economic policies and the level of democracy.³⁷

Another strand of literature has focused on the effect of political ideology on political behavior and policy outcomes. Political ideology is understood as a “set of beliefs about the proper order of society and how it can be achieved.”³⁸ But note that these sets of beliefs are shared collectively. For example, Denzau and North state that “ideologies are the shared framework of mental models that groups of individuals possess that provide both an interpretation and prescription as to how that environment should be structured.”³⁹ The effect of political ideology on elite values and preferences is well established.⁴⁰ Political ideology has been linked to party⁴¹ and to rebel coalitions.⁴² A large body of literature also shows that norms and ideological beliefs can diffuse between countries and elites to influence policies.⁴³ Research on conflict has shown that ideology can motivate the political goals of armed groups,⁴⁴ shape their repertoires of violence,⁴⁵ and structure their institutions.⁴⁶ In violent contexts, Leader Maynard highlights that individuals’ adherence to an ideology can be strategic, genuine, or both.

While scholars have shown that individual leaders and their backgrounds, experiences, influences, and ideologies affect many political outcomes, variation in leaders’ preferences and ideological convictions have not often been considered as explanatory factors for political outcomes that shape African politics, such as ethnic power sharing. For example, Chemouni and Mugiraneza argue that “unlike during the two decades after decolonization, the analysis of political ideologies as a normative engine of political action [in African politics] seems to have receded in favour of a treatment of ideology as the support of actors in their pursuit of material interests.”⁴⁷ Scholarship on African politics usually does not see stark ideological differences between parties and politicians and considers ethnic or regional logics of support more important.⁴⁸ In new democracies, partisanship is argued to have less of an ideological base,⁴⁹ and African parties’ programs do not tend to exhibit strong ideological differences.⁵⁰

We investigate how African political leaders’ exposure and ideological commitment to Pan-Africanism during the anticolonial struggle affects their approach to ethnic power sharing. For many African political elites, the anticolonial struggle was a formative experience that shaped their political understanding. Africans under colonial rule engaged in many different forms of resistance, ranging from mobilization against discriminatory and repressive colonial policies in local institutions and economic, tax-, and labor-related resistance to strikes, mass demonstrations, and armed rebellion.⁵¹ During this time, African elites could align themselves with the Pan-African movement, with its commitment to African independence, African unity, and overcoming ethnic cleavages, or take a more moderate position on decolonization by upholding relations with the former colonizers.

Historical Background

The Pan-African Movement and Conferences

Anticolonial activism played an essential role in the rapid decolonization of Africa. The period following the end of the Second World War saw an upsurge in anticolonial activity and political mobilization, with unprecedented levels of international coordination between African elites.⁵² Pan-Africanism provided the ideological umbrella for demands for rapid decolonization. What made Pan-Africanism a “radical” ideology during this period was that it advocated the total political and economic independence and unity of Africans regardless of cultural, political, or geographic factors.⁵³ During decolonization, the Pan-African movement identified transnational organizing and the formation of national movements crossing ethnic boundaries as key to the liberation of African states.⁵⁴ Colonizers were suspicious of Pan-Africanism and tried to prevent the ideology from spreading in their colonies.⁵⁵

At the time, African political elites could either align themselves with the Pan-African movement and its proximity to socialist states and ideals, or seek the continued support of colonizers by endorsing less rapid independence and subsequent cooperation with former colonizers. This choice is illustrated by two Zambian politicians who were prominent during the anticolonial struggle in Northern Rhodesia, Kenneth Kaunda and Harry Nkumbula. Both were allies and leading figures of the African National Congress (ANC) and the struggle for independence. However, “younger radicals led by Kenneth Kaunda” broke away from the ANC under Nkumbula's leadership because of its moderate, “gradualist” approach to independence and formed the Zambian African National Congress (ZANC).⁵⁶ ZANC was immediately banned by the colonial government (in 1959), but its more radical leaders, under Kaunda, eventually formed the United National Independence Party (UNIP).⁵⁷ UNIP's civil disobedience campaigns against the colonial government and elections isolated them domestically from potential settler support, but led to international backing from the Pan-African movement.⁵⁸ Meanwhile, “Nkumbula's refusal to mobilize [ANC] followers against” a settler-backed constitution “granted their party an incontrovertible badge of moderation,” which, coupled with the ANC's anti-communism and “anti-Pan-Africanism” rhetoric, led to a pact with the settler United Federal Party and expulsion from Pan-African networks.⁵⁹ Harry Nkumbula and the ANC were viewed as “colonial puppets” by most Pan-Africanists.⁶⁰

Pan-African conferences were a key ideological space for the Pan-African movement. These conferences created a forum in which elites of anticolonial movements across Africa could interact, formulate policies, issue demands of colonial administrations, exchange support, and debate postcolonial policies.⁶¹ Prominent elites of African anticolonial movements such as Kwame Nkrumah, Sékou Touré, Frantz Fanon, Kenneth Kaunda, Joshua Nkomo, Félix-Roland Moumié, Tom Mboya, and Patrice Lumumba attended these conferences and often helped organize them.⁶² While earlier Pan-African conferences took place in Europe or North America and predominantly featured intellectuals from the African diaspora,⁶³ the fifth Pan-African Congress, held in Manchester in 1945, marked a distinctive shift in the movement by including more elites from Africa, who formulated a cohesive anticolonial platform with explicit demands for independence.⁶⁴ Subsequent Pan-African gatherings were held in Africa and were all organized and dominated by African politicians and activists, many still in the midst of independence struggles.

In sum, during decolonization, the key ideological divide among African elites was between the Pan-African movement, which demanded the rapid and complete liberation of African states, and more moderate approaches. The Pan-African conferences provided a forum in which elites in the Pan-African movement could interact and strategize toward this goal across the African continent. They created and reinforced connections between anticolonial leaders and demonstrate the extensive coordination that occurred between diverse rival and collaborative movements in the struggle for independence.

Conference Attendance

The elites who organized Pan-African conferences and established permanent institutions intended the movement to be inclusive and representative of anticolonial nationalist movements across Africa. For example, the All-African Peoples Conference Organization (AAPCO), the body responsible for inviting delegates to the All-African Peoples’ Conferences (AAPCs), tried to make them “truly representative” of anticolonial movements across Africa.⁶⁵ As Kwame Nkrumah proclaimed in his opening speech at the first AAPC, “Invitations were sent out to all bona fide political and trade union organizations regardless of their political complexion or the relationships which exist between them in their various countries.”⁶⁶

At the same time, the secretariats of the AAPCO and the Afro-Asian People's Soldiarity Organization (AAPSO), which governed the memberships of the Pan-African and Afro-Asian conferences, respectively, were essentially the gatekeepers of the Pan-African movement. Organizers from the AAPSO and the AAPC screened and monitored elites and could deny or revoke memberships.⁶⁷ Those considered would have to be “active nationalists enjoying the full confidence of the Nationalist Movements in their respective Zones,” and they were “screened by the Screening Committee before a decision is taken.”⁶⁸ These organizations were well informed of ongoing political events and actors across Africa and used their extensive networks to vet and monitor members.⁶⁹ For example, the Bureau of African Affairs (BAA), which was closely linked to the AAPCO, had agents stationed across Africa who forwarded intelligence briefings on political developments.⁷⁰ When an individual or organization asked to join the AAPC, they were screened by the staff of the BAA and/or discussed at high-level government meetings like Kwame Nkrumah's African Affairs Committee.⁷¹ This information was filtered back to the AAPCO and AAPSO organizing bodies to vet the credentials of different individuals and movements and to reasonably deny or revoke the memberships of those they considered colonial puppet governments or against the goals of the Pan-African movement. This internal vetting process means that movements fundamentally opposed to the Pan-African ideal of African unity were less likely to be represented at the conferences.

The case of Harry Nkumbula and the ANC illustrates the backlash one might expect from the Pan-African movement for cooperating with colonial governments. In the months preceding the first AAPC, Harry Nkumbula was increasingly seen as a moderate for his willingness to accept the Northern Rhodesian constitution backed by the colonial governor.⁷² This in part motivated Kenneth Kaunda and other ANC radicals to break away and form ZANC in October 1958. When, in December 1958, both Nkumbula and Kaunda attended the first AAPC in newly independent Ghana, Kaunda became Kwame Nkrumah and Kamuzu Banda's preferred contact person, over Nkumbula.⁷³ Kaunda, who had recently formed the more radical ZANC, enjoyed the support of the Pan-African movement, while Nkumbula became increasingly sidelined by previous allies.⁷⁴ Subsequently, in 1959, Nkumbula and the ANC's requests for financial support were denied by the AAPCO. No ANC delegates attended the 1960 or 1961 All-African Peoples or Afro-Asia Solidarity conferences. As a result of his political isolation from the Pan-African movement, Nkumbula sought support from Moise Tshombe, the discredited leader of the Katanga secessionst state who had central roles in the Congo Crisis and the assassination of Patrice Lumumba.⁷⁵

Pan-African conferences were valued by political elites for three reasons. First, they formed linkages between independence movements across regional, linguistic, and ideological divides. At the conferences, the multitude of veteran politicians with diverse experiences and ideologies provided invaluable knowledge and established networks for inexperienced elites. For example, Kwame Nkrumah's political platform and the sweeping goals of his administration for a united Africa can be traced back to his experiences at the 1945 conference in Manchester.⁷⁶ Then just a young student, Nkrumah interacted with experienced Pan-Africanists like George Padmore and Ras Makonnen, who introduced him to their networks and instructed him on how to operate political organizations.⁷⁷

Second, these conferences created lasting Pan-African institutions that actively maintained networks but also offered concrete material resources and support to anticolonial movements. The most important Pan-African organizations were established by Kwame Nkrumah and his administration during the 1958 AAPC in Ghana, which made Accra a center for anticolonial elites.⁷⁸ For example, at the 1958 AAPC the African Affairs Center and the BAA were established in Accra. The African Affairs Center welcomed “hundreds of African freedom fighters in search of funds, training, and a political platform,” including the exiled UPC party, led by Félix-Roland Moumié, and Amílcar Cabral's African Party for the Independence of Guinea and Cape Verde, which was already in armed conflict with Portuguese colonial administrators.⁷⁹ The BAA provided additional funds, military training, and political support to anticolonial movements.⁸⁰

Third, the conferences were highly publicized platforms from which anticolonial elites could project their aims and legitimacy to both international and domestic audiences. For example, Frantz Fanon and other delegates of the National Liberation Front (FLN) used the AAPCs to garner support from the Pan-African movement and to put international pressure on the French government. FLN members directly participated in the formulation of key resolutions, such as the Resolution on Algeria at the 1960 AAPC, which diplomatically recognized independent Algeria under the FLN and pledged for “all the African Independent Governments the inclusion in their budget of regular financial contribution in favour of struggling Algeria.”⁸¹

Ethnic Inclusiveness in the Pan-African Movement

Ethnic unity and overcoming ethnic divisions were core values of the Pan-African movement. In this section, we outline how Pan-African ideology prioritizes ethnic inclusiveness by overcoming ethnic, religious, and social cleavages that were fortified during colonial rule. The Pan-African movement also identified the formation of multiethnic mass movements as not only an ideological goal but also a critical instrument of liberation.

Leaders of the Pan-African movement called for “self-government throughout Africa and the establishment of democracy under which there would be no discrimination, victimization, or segregation based on color, race, or religion.”⁸² The principle of inclusion was in direct response to the divide-and-rule strategies of colonial administrations, which exploited and created ethnic and cultural cleavages. Pan-Africanists identified ethno-religious divisions as “arbitrary divisions … done to satisfy the greed and avarice of colonial and imperialist powers.”⁸³ This guiding strategy of the Pan-African movement is captured in Abdoulaye Diallo's statement, in his speech at the 1960 AAPC, that “unity within each country, will facilitate unity in Africa as a whole.”⁸⁴

Overcoming the ethnic and cultural divisions created or exacerbated by colonial administrations was a central goal of the Pan-African movement reiterated throughout speeches and resolutions at conferences. For example, one of the main resolutions passed at the first All-African People's Conference, in 1958, was titled “Tribalism and Religious Separatism,” and stated:

We strongly oppose the imperialist tactics of utilizing tribalism and religious separatism to perpetuate the colonial policies in Africa … We are also convinced that tribalism and religious separatism are evil practices which constitute serious obstacles to (i) the realization of the unity of Africa (ii) the political evolution of Africa (iii) the rapid liberation of Africa.

Ethnic unity was not just an abstract ideological principle. Pan-African elites also viewed the formation of mass national movements as a key political goal and a necessary political instrument to overcome colonization. While the understanding that mass action was needed was inspired by socialist liberation politics, Pan-Africanists interpreted this specifically as the need to overcome ethnic divisions sown by colonizers.⁸⁵ The conferences provided a space in which to discuss how nationalist movements could be built. For example, Oscar Kambona's speech at the second AAPC, in 1960, addressed the difficulties of forming a “nationalist movement which cuts across tribal, religious and racial barriers” to unify “tribal organisations scattered all over the country” and thus overcome “a colonial government [that tried] to use [race] to segregate people living in Tanganyika.”⁸⁶

In sum, uniting Africa and overcoming ethnic divisions were core ideological and political goals of the Pan-African movement that were maintained regardless of the later splits within the movement over the use of violence in achieving independence and the economic future of Africa.

Theoretical Mechanism

In this section, we develop our theoretical argument on how exposure and commitment to the Pan-African movement and its values—proxied through attendance at Pan-African conferences—affected political leaders’ preferences for ethnically inclusive governance.

First, we argue that conference attendance serves as an information revelation mechanism: individuals with ideological convictions and political goals in line with Pan-Africanism, including ethnic unity, are more likely to be invited and to self-select into attending. Elites also had many strategic incentives to join Pan-African ideological spaces like the conferences. We argue, however, that to seek strategic support from the Pan-African movement, elites made a conscious choice to commit to one side of a rigid ideological and political divide, which required adherence to specific norms. Thus even strategic attendance signals commitment to the movement's core values.

Second, we argue that the movement and the conferences offered ideological as well as political instruction, affecting attendees’ normative preferences as well as their understandings of effective political strategies. Through norm diffusion and political learning, exposure to the Pan-African movement at the conference shifted political elites’ preferences toward forming ethnically inclusive coalitions, even if their initial motivation for attending Pan-African ideological spaces was predominantly strategic. Figure 1 provides a visualization of our main argument that we further elaborate on in the following subsections.

Figure 1. Theoretical mechanism of ethnic inclusion

Conference Attendance as Revealing Information

The first theoretical mechanism that allows us to identify political elites sharing Pan-African ideological goals and political understandings is selection. We argue that politicians and organizations that were ideologically aligned with Pan-African ideals were much more likely to join the conferences. This implies that politicians, political movements, and parties that were already committed to ethnic inclusiveness and tried to form broad coalitions of domestic support were over-represented among conference attendees. For example, when Felix Moumié, the leader of Cameroon's UPC, joined the 1957 AAPSO conference, his party, the UPC, was already a successful radical nationalist mass movement with close ties to communism—and banned by both the British and French governments.⁸⁷

However, elites also had many strategic incentives to join conferences. Pan-African conferences were attractive for African elites because they could offer support and visibility to individual politicians and anticolonial organizations. For elites striving for liberation, the conferences provided an alternative political space outside the colonizers’ sphere of influence. For members of nascent anticolonial movements seeking recognition and assistance, the conferences were an opportunity to connect with experienced elites and state leaders who could provide practical knowledge, as well as funds.⁸⁸

However, we argue that even those who joined in hopes of strategic benefits were at the minimum sympathetic to the movement and its goals from the outset. This is plausible for several reasons. First, there was extensive gatekeeping and vetting of potential attendees by conference organizers. To gain support from the Pan-African movement and be welcomed at conferences, political elites and their movements had to operate in line with the core values of Pan-Africanism. Pan-African organizers understood individual elites’ and their political organizations’ values, policies, and networks. Strategic elites who wished to gain support from and visibility in the Pan-African movement needed to show commitment to its values and survive vetting processes. Thus their approach to domestic political organizing needed to reflect core values such as ethnic unity.

Political leaders whose aim was to mobilize specific ethnic constituencies exclusively would have found their approach to organizing at odds with prominent Pan-Africanists’ stances against traditional rule and chieftaincy and their condemnation of ethnic distinctions as colonial products. The Pan-African response to the Congo crisis serves as a clear example. At the third AAPC Conference, which took place shortly after Lumumba's murder, political parties viewed as “colonial puppets” or as secessionists, such as the Confederation of Tribal Associations of Katanga under the leadership of Moise Tshombe, were excluded and openly denounced. In a series of resolutions on neocolonialism, conference delegates explicitly condemned “puppet governments represented by stooges … based on some chiefs, reactionary elements, and anti-popular politicians” and warned of “Balkanisation as a deliberate political fragmentation of States,” citing Katanga as a direct example.⁸⁹ Meanwhile, surviving members of Patrice Lumumba's Congolese National Movement were in full attendance at the conference and considered the legitimate national party of the Congo.

Political leaders considered “colonial puppets” were often permanently excluded from Pan-African networks and conferences.⁹⁰ Referring to the multitude of permanent Pan-African institutions established in Ghana after the first AAPC (in 1958) to support liberation movements and exiled anticolonial activists, “Nkrumah reiterated that assistance would be given only to organizations that fully subscribed to the idea of African unity.”⁹¹

Second, there were dynamics that locked in alignment with the Pan-African movement and its values. On the one hand, elites’ visibility in Pan-African spaces revealed information to prospective domestic supporters and voters. As a result, it would be much more difficult for elites who were heavily involved in the Pan-African movement to credibly campaign for the support of one domestic ethnic group in particular after presenting themselves as leaders of nationalist and inclusive movements in international Pan-African spaces.⁹²

On the other hand, placing themselves on the Pan-African side of a polarized ideological divide made it much harder for political elites to garner subsequent support from former colonial powers, in terms of either gaining power or securing post-independence economic support. Once elites were perceived as radical by former colonizers, they needed to maintain support from within the movement to navigate foreign and domestic politics after independence. For example, both Joshua Nkomo and Kanyama Chiume barely escaped arrest by British colonial administrators and were forced to flee because of their “radical” political platforms and unwillingness to negotiate on the timing and form of independence. While they were denied support and legitimacy from their colonial governments, their decision to attend the 1958 AAPC provided them with an alternative network and resources.⁹³ Pan-African elites, on the other hand, may have served as third-party enforcers for ethnic power sharing.⁹⁴ For example, Roessler and Verhoeven show that a Pan-African alliance overthrew Mobutu in the Democratic Republic of the Congo due to his reactionary politics.⁹⁵

In sum, under this mechanism, conference attendance serves as an information revelation mechanism to identify pre-existing ideological commitments to ethnic inclusiveness among political elites, through both self-selection and the conference organizers’ gatekeeping. We argue that this self-selection reflects ideological commitment or, at the minimum, elites’ conscious and long-term choice to align themselves with the Pan-African movement and to adhere to its political goals, which is hard to explain if ideological moderation, cooperation with former colonial powers, and ethnically exclusive organizing were actually preferred.

Conference Attendance as Changing Elites’ Preferences

The second theoretical mechanism focuses on the influence of Pan-African conferences on attending elites, deepening commitment among those who were already committed and influencing those who attended for strategic reasons. We argue that Pan-African conferences promoted ethnic unity both as an ideological, normative principle and as an effective political instrument, even to those who attended strategically to gain support. We discuss these two types of influence in turn.

Elite Socialization and the Diffusion of Political Ideology

We argue that Pan-African conferences act as ideological spaces to diffuse norms of ethnic unity. Exposure to Pan-African values on ethnic unity at Pan-African conferences impacts the socialization of political elites in line with theories of elite socialization through elite political culture⁹⁶ and exposure to political ideology in their formative years.⁹⁷ As a result, politicians who attended conferences developed normative preferences for ethnically inclusive politics. Pan-African conferences enable repeated interactions with actors and institutions that promote Pan-African ideology and thus opportunities to socialize political elites. Elite socialization takes place through political elites adhering to the expected norms of the community to facilitate social interactions with other group members.⁹⁸ Over time, however, adhering to these norms can become a habit that affects behavior⁹⁹ and is internalized by individuals through greater group identification.¹⁰⁰ Thus, exposure to environments promoting specific norms can lead to normative changes and the internalization of “new understandings of appropriateness” through a process of normative persuasion¹⁰¹ or identification.¹⁰²

Research on norm diffusion outlines similar mechanisms of normative adoption. Elkins and Simmons conceptualize norm diffusion as processes of adaptation and learning among international actors that can lead to clustered policy reforms, as state actors update their preferences based on the behaviors of their international peers.¹⁰³ Through different forms of learning, say Simmons, Dobbin, and Garrett, “policy innovation spreads in the wake of the diffusion of a shared fund of … knowledge among elites.”¹⁰⁴ Some authors have applied this logic specifically to the impact of intergovernmental organizations,¹⁰⁵ which provide venues for the exchange of ideas between representatives of different states, which fosters discussion and debate. These exchanges can lead to socialization and norm diffusion between states.¹⁰⁶

Pan-African conferences were able to diffuse political ideology as African political elites met repeatedly in both formal and informal contexts and exchanged ideas and discussed policy issues. Both AAPCO and AAPSO held multiple conferences and established permanent institutions like steering committees which provided continuity between conferences and additional spaces for elites to interact. Beyond the conference halls, delegates also interacted through off-the-record meetings and social events that provided more opportunities to exchange ideas.¹⁰⁷ Interactions at conferences created new relationships that led to further ideological exchanges. For example, at the 1958 AAPC, Patrice Lumumba, Frantz Fanon, and “other important political leaders established contacts with Nkrumah's government” and began “to collaborate with the ‘Pan-African’ institutions of Ghana.”¹⁰⁸ Pan-African institutions like the African Affairs Center sustained contact between delegates and further disseminated Pan-African ideas, in part functioning as a “ideological training centre” where anticolonial elites solidified their “[Pan-]African ideology.”¹⁰⁹

Political learning

Pan-African conferences not only served as spaces for normative instruction but also provided practical instruction on how to build successful political movements. Theories of policy diffusion suggest that diffusion is most likely if previous adopters are similar in ideology.¹¹⁰ The shared ideological goal of rapid and radical decolonization may have made attendees receptive to the promoted strategies for achieving this goal.

Leading Pan-Africanists promoted bridging ethnic divides and forming nationalist mass movements as a key political strategy to gain independence and ward off neocolonial incursions.¹¹¹ For example, “the goal of Nkrumah's Pan-African policy was to realize the unity of the continent by creating a net of parties which could embrace Pan-Africanism and which could follow the example set by CPP and its successful independence struggle. In particular, the new nations had to be freed by a mass nationalist movement.”¹¹²

African elites who attended Pan-African spaces and conferences in their formative years came to understand nationalist politics and the formation of mass movements as a key political instrument. Here, they developed their understanding of how to mobilize and sustain political support and how to use revolutionary politics and nationalist mass appeals to build a following. Nkrumah himself was instructed by experienced Pan-Africanists like George Padmore and Ras Makonnen, whom he interacted with at the 1945 conference in Manchester and who introduced him to their networks as well as instructing him on how to operate political organizations.¹¹³ As a result of this direct exposure to Pan-African ideology abroad, Nkrumah “came to see the divisions in [Ghana's] nationalist movement as a drag on progress” and that only a “mass movement and the creation of institutions responsive to the needs of the people” could overcome colonialism and ethnic tensions.¹¹⁴ Once in power, Nkrumah aimed to overcome tribalism¹¹⁵ and opposed regional parties.¹¹⁶

Similarly, Joshua Nkomo credited the formation and mobilization of the Zimbabwe African People's Union in 1961 to the training he received in Accra and Cairo with the support of the AAPCO and the AAPSO. In Ghana, Nkomo reflected that “the most important thing … was my new friendship with a young man from Uganda, John Kale, a brilliant organiser who helped me a great deal” and whom he met at the 1958 AAPC.¹¹⁷ The following year, Nkomo opened an office in Cairo, with funding from the AAPSO, where he reconnected with John Kale and also met Felix Moumié, who “showed [him] all the techniques needed for running a political office.”¹¹⁸

Once in power, elites from within Pan-African circles extensively used mass appeals to ethnic unity as a tool for political mobilization and to ward off political competition. In Zambia, Kaunda carefully orchestrated UNIP's internal politics to ensure representation and balancing of Zambia's different regional factions within the party.¹¹⁹ He tried to brand political opponents, including his most threatening political rival, Simon Kapwepwe, as “tribalists”¹²⁰ and “sought to use the allegation of ‘tribalism’ to de-legitimize all criticism of his government and policies.”¹²¹

Political learning likely affected the institutionalization of these mass movements as well, encouraging organizational structures that created self-enforcing ethnic power sharing.¹²² For example, leaders who attended at least one Pan-African conference are more likely to rule over party-based autocratic regimes than leaders who did not attend any conference (61 percent versus 21 percent of country-years).¹²³

Empirical Implication

On the basis of the theoretical discussion, we expect that previous attendance at Pan-African conferences is positively associated with African state leaders’ forming more inclusive governments. This association hinges on our argument that Pan-African conference attendances can serve as a proxy for Pan-African ideological commitment (1) prior to attendance or (2) because of the diffusion of political ideology and political understandings at the conferences themselves. We expect that political elites who have adopted Pan-African ideological and political understandings will act according to those values after becoming government leaders. Because Pan-African ideology promotes ethnic inclusion, we predict that once in government office, such leaders will enable more inclusive ethnic coalitions.

Hypothesis State leaders who attended more Pan-African conferences lead ethnically more inclusive governments.

Empirics

In this study, we investigate the effect of government leaders’ previous Pan-African conference attendance on ethnic inclusion during their tenure. In line with related work,¹²⁴ we operationalize ethnic inclusion as the degree to which more ethnic groups and their respective populations are represented in government. We focus on independent African states between 1946 (or independence) and 2010, and the unit of analysis throughout is the country-year. Summary statistics for all variables can be found in Table A1 in the online supplement.

Outcome Variables: Inclusion of Ethnic groups and Population

In the African context, political cleavages often form around ethnic identities, which at times have colonial legacies that the Pan-African movement wanted to overcome. Hence, we expect that countries whose leaders were more influenced by the Pan-African movement should display, proportionally, more included ethnic groups. Ensuring that the inclusion of multiple groups also translates into larger shares of the population being included, we use the proportion of included ethnic population as an alternative outcome.

Thus, we provide two measures of ethnic inclusion. First, we use the proportion of the politically relevant ethnic groups included in government. Second, we use the proportion of the ethnic population that is included in government. Our primary data source is the Ethnic Power Relations (EPR) data set,¹²⁵ which operationalizes political inclusion of groups as meaningful access to the state's executive, which includes “control of the presidency, the cabinet, and senior posts in the administration, including the army.”¹²⁶

Explanatory Variable: Government Leaders’ Past Pan-African Conference Attendance

The coding of our main explanatory variable relies on the Pan-African Conferences Dataset, an original data collection of the delegate lists of twelve Pan-African conferences, which we compiled through archival work in Ghana (Public Records and Archives Administration Department, Accra; George Padmore Research Library on African Affairs, Accra) and the United Kingdom (British Library, London; SOAS Special Collection, London; University of Manchester Special Collections, Manchester). This unique data set allows us to test the hypothesis that government leaders who were previously involved in Pan-African conferences are more likely to rule over ethnically inclusive polities. Up to 1965, we identify state leaders using Archigos,¹²⁷ while from 1966 onward we rely on WhoGov.¹²⁸ Leveraging these data, we code how many relevant Pan-African conferences the current state leader has previously attended. In addition, we code a binary version of this variable, which says only whether a state leader attended at least one of these conferences.

More specifically, to measure exposure and commitment to Pan-African ideology, we consider attendance at eight of the twelve Pan-African conferences in the Pan-African Conferences Dataset between 1945 and 1965. These eight conferences are (1) conferences under the Pan-African umbrella that centered on African politicians and leaders, and (2) conferences that focused on decolonization and not on post-independence international cooperation between African states.¹²⁹

The first criterion excludes earlier conferences organized by Du Bois, which were dominated by elites from the African diaspora.¹³⁰ Accordingly, we start with the fifth Pan-African Congress, held in Manchester in 1945, which marked a distinctive shift in the movement by including more elites from Africa, who formulated a cohesive anticolonial platform with explicit demands for independence.¹³¹ Subsequent Pan-African gatherings, such as the All-African People's Conferences (AAPCs) of 1958, 1960, and 1961 and the Afro-Asian People's Solidarity Organisation (AAPSO) Conferences of 1957, 1960, 1963, and 1965, were held in Africa and were all organized and dominated by African elites.

The second criterion excludes conferences such as the Organization of African Unity Conference Summit (May 1963) and the first and second Conferences of Independent African States (1958 and 1960), which focused more on the interactions between independent African states, as they were widely attended by African state leaders and thus are not a strong signal of ideological commitment. Smaller, regional, and economic conferences are also excluded because of their limited connection to the wider Pan-African movement. Table 1 provides an overview of the conferences used to operationalize our variable of interest, alongside conferences in the data set that did not fit our criteria for inclusion.

Table 1. Conferences in the Pan-African Conferences Dataset

Note: Some conferences are excluded from the main analysis because they are diaspora-focused, post-independence international conferences or are regional and have limited connection to the wider Pan-African movement.

Our digitized collection of Pan-African delegate lists allows us to match government leaders from the Archigos and WhoGov databases to Pan-African conference attendees. The merging of information on delegates and government leaders was supported by automatized string matching and verified by human coding. Having matched Pan-African delegates with government leaders, we calculate each leader's Pan-African conference attendance. Figure 2 plots our main explanatory variable (range, 0 to 3): African government leaders’ attendance at the eight selected Pan-African conferences. There is variation within and between countries in how many conferences the current state leader has previously attended. In our empirical analysis, we particularly exploit within-country variation, while also considering between-country variation.

Figure 2. Pan-African state leaders in countries over time

Control Variables

Our main empirical analysis includes country and year fixed effects, which account for time-invariant country-level confounding variables (e.g., previous colonizers and precolonial dynamics) that could correlate with government leaders’ conference attendance and ethnically inclusive regimes.

One important time-variant confounder to consider is whether the effect of conference attendance is merely driven by the first postcolonial generation of leaders, who had to be more inclusive to stabilize their young nations. We control for whether the first leader is currently in power by including a dummy variable in all our models and coding the historical count of leaders (first, second, third, and so on), as well as the leaders’ current time in office. We control for ongoing civil conflict in a given year using the UCDP Armed Conflict Dataset¹³² and the time since the last coup,¹³³ as conflict might decrease the inclusive effect of Pan-African conference attendance, while high coup risks might also incentivize more exclusive polities over time. We also account for more structural factors such as GDP per capita,¹³⁴ Polity score,¹³⁵ and whether the state has a personalist, military, or party-based autocratic regime.¹³⁶ Controlling for regime type is particularly important because leaders who participated in Pan-African conferences could be more likely to implement particular regime types that are driving the inclusiveness of the political system. Hence, we are controlling for an important mediating variable, leading us to potentially underestimate the total effect of conference attendance.

Our measure of inclusiveness could be driven by the leader's ethnic group's size, as high measures of inclusiveness could indicate large ethnic groups holding exclusive power. We therefore include the size of ethnic groups that are senior partners, dominant, or have a monopoly on power according to the EPR data set and manually code the ethnic group of the executive leader in the few cases where two senior partners are coded in the EPR data.

Estimation Approach

Our estimation approach is motivated by the concern that unobserved country and temporal effects could correlate with government leaders’ past Pan-African conference attendance and ethnically inclusive governments. Thus, we estimate linear regression models with country and year fixed effects. The fixed effects also imply that the estimated effects of government leaders with Pan-African conference experience stem from these leaders’ either entering or leaving office and being replaced by leaders with less or no Pan-African conference attendance.¹³⁷

We present our main results in two steps. We first present a set of models that focus on the relationship between inclusive governments and past Pan-African conference attendance by the current government leader while controlling for structural country-level variables and leader characteristics. We then add information on the time since the last coup.

Results

The estimates of the main model results are provided in Table 2. In models 1 to 4 the main independent variable is a count of conference attendance, while in models 5 to 8 it is a binary indicator (yes or no) of past conference attendance by the current government leader. The first two models estimate the effect of past Pan-African conference attendance by the current government leader on the proportion of ethnic groups (model 1) and their respective population included in government (model 2). These models also include structural country-level control variables and leader characteristics. We find that as state leaders’ Pan-African conference attendances increases, the proportion of included groups (model 1) and the proportion of ethnically included population (model 2) both increase. On average, attending one additional Pan-African conference increases the proportion of included ethnic groups and politically relevant ethnic population by about 10 percent (model 1) and 11 percent (model 2), respectively. Predicted values corresponding to models 1 and 2 are visualized in Figure 3. Across all models, the estimated effect of the government leaders’ past conference attendance is statistically different from 0 at standard levels of significance (p < 0.01). These results are in line with our theoretical expectation that leaders with more Pan-African conference exposure will form more inclusive regimes. Including the time since the last coup slightly increases the estimated effect.

Table 2. Main models with different coding of conference attendance

Notes: Linear fixed effects models include the count or binary indicator of conference attendance. Outcome variables are proportion of included groups or population. Unit of analysis is the country-year. All models include year and country fixed effects. *p < .10; **p < .05; ***p < .01.

Figure 3. Effect of Pan-African conference attendance on ethnic inclusion

In a second set of models (models 5–8), the main independent variable is a binary indicator of past conference attendance. Given the few instances of high conference attendance (see Figure 2), there could be a concern that our results are driven by outliers with high conference attendance (see also our outlier analysis, later on). To check, we reduce the count to a binary indicator. The model specification matches the main analysis (models 1–4). These results (models 5–8) are in line with our empirical expectation that government leaders who attended Pan-African conferences in the past are associated with ethnically more inclusive ruling coalitions.

Overall, these findings suggest that the ideological legacy of Pan-African organizing during the decolonial struggle affects political elites’ approach to ethnic power-sharing and inclusive politics. These findings hold when accounting for regime types and leader characteristics, as well as time-invariant country-specific factors (such as former colonizer), which we account for in our fixed-effects specification. Thus, we are able to demonstrate support for the theoretically implied relationship between Pan-African conference attendance and more inclusive governments in postcolonial African states.

Separating Strategic Incentives from the Effect of Ideology: Sources of Support

In this section, we address alternative mechanisms that could link leaders’ Pan-African conference attendance—or absence—to their approach to ethnic inclusion when in power. These alternative mechanisms focus on political support for politicians competing for power in newly independent African states. Politicians with support from sources outside the Pan-African network (such as previous colonizers) had fewer incentives to join conferences and gain Pan-African network support. At the same time, these alternative sources of support could have helped them form and maintain more exclusionary governments. We consider support from (a) ethnic communities and (b) European colonizers.

Ethnic elites who had gained power under colonialism had less incentive to attend Pan-African conferences, as Pan-African leaders opposed ethnic cleavages created by former rulers. A resolution of the 1958 AAPC illustrates this, stating that institutions such as the “chieftaincy do not conform to the demands of democracy” and that it “actually supports colonialism.”¹³⁸ The antagonism between ethnic cleavages and Pan-African ideology is illustrated by Kwame Nkrumah's stripping chiefs of their power or co-opting them when gaining state power.¹³⁹ Similarly, in the former Belgian Congo, ethnic elites feared Patrice Lumumba's nationally oriented Pan-African movement as competition for support in their ethnic constituencies.¹⁴⁰ British indirect rule, in particular, allowed rural ethnic elites to consolidate power and to mobilize electoral support from their ethnic communities.¹⁴¹ As a result, they had little need to solicit support from Pan-African networks and little incentive to attend conferences that were ideologically opposed to their continuing influence. At the same time, leaders who could draw on ethnic support had lower incentives to form national coalitions and thus more inclusive governments.

During decolonization, colonizers repressed movements they considered too “radical” and helped politicians they approved of take positions of power.¹⁴² Thus African politicians had incentives to take a moderate stance to avoid repression and gain support from the colonial administration.¹⁴³ Often, colonizers supported ethnic elites to counter Pan-African movements. For example, the French colonial administration repressed and subsequently banned the Pan-African UPC in Cameroon and helped Ahmadou Ahidjo take power.¹⁴⁴ In the early 1950s, the UPC was the only party with nationwide support, while all other parties had ethnic or personalist support.¹⁴⁵ Thus, conference attendance could just be a proxy for the degree to which African politicians during the decolonization period were favored by and could expect support from the colonial administration. At the same time, state leaders who had the backing of the former colonial administration had better chances of staying in office even without broad ethnic coalitions.

To rule out our findings being driven by such dynamics, we follow Wucherpfennig, Hunziker, and Cederman's approach and consider whether a state leader's ethnic group's position was strengthened by the British approach to indirect rule. In British colonies, “autonomous ethnic leaders had consolidated power thanks to British indirect rule and its focus on customary institutions,” while “under French indirect rule, local conditions and institutions were often deliberately ignored.”¹⁴⁶ Thus, elites from rural groups in British colonies were more likely on average to have ethnic support and had fewer incentives to take more radical anticolonial or Pan-African stances, which in turn increased their chances at support from colonial administrations.

To account for leaders from groups that consolidated power under British indirect rule, we include two variables from Wucherpfennig, Hunziker, and Cederman in our models.¹⁴⁷ First, we include a dummy variable for whether the country gained independence from Britain. Second, we include the natural logarithm of the leader's ethnic group's distance from the coast. Operationalizing this distance for ethnic groups, Wucherpfennig, Hunziker, and Cederman calculate “the distance between the group's settlement area (centroid) and the colonial center (i.e., the coast). For landlocked countries without direct access to the sea, [they] subtract the minimum (country) distance to the coast in order to arrive at a standardized measure.” Importantly, and still following Wucherpfennig, Hunziker, and Cederman, we also include the interaction between colonizer and the leader's group's distance to the coast. As we expect that rural groups in former British colonies have advantages in their access to power due to ethnic and colonizer support, we would also expect leaders who can claim these identities to have less incentive to attend Pan-African conferences during decolonization and thus to form less inclusive ethnic coalitions.

Table 3 shows results from adding the variables for ruralness and British colony, as well as their interaction, to our original specification. As expected, the interaction term is negative and statistically significant, indicating that the effect of coastal distance on inclusiveness is significantly decreased in British colonies (models 9 and 10). Importantly, however, the coefficient of the variable for conference counts remains positive and highly significant. This suggests that leaders’ Pan-African conference attendance does not purely approximate a selection effect by which leaders without ethnic or colonizer support had incentives to find alternative sources of support in the Pan-African network and form a broader ethnic coalition; rather, ideological factors play a role in explaining ethnically inclusive government.

Table 3. Models addressing strategic incentives to attend Pan-African conferences stemming from lack of ethnic and/or colonizer support

Notes: Sample includes only countries with French or British colonial legacy. Linear fixed effects models. Outcome variable is proportion of included groups or population. Unit of analysis is the country-year. All models include year and country fixed effects. *p < .10; **p < .05; ***p < .01.

Separating Strategic Incentives from the Effect of Ideology: External Threats

External threats can incentivize elites to foster greater unity internally.¹⁴⁸ During the African independence struggle, European settlers posed a considerable threat to decolonization.¹⁴⁹ To rule out our results being driven by leaders facing greater settler populations, who then require support from the Pan-African movement and form larger coalitions, we draw on additional cross-sectional data from Paine to account for countries with high European settler population,¹⁵⁰ whether decolonization was highly violent,¹⁵¹ and whether the country was a British or French colony. Models 11 and 12 (in Table 4) provide estimates for our main models, but only including year fixed effects, while models 13 and 14 show all years of the first leader's tenure. All models include only year fixed effects to enable estimation of time-invariant variables.

Table 4. Linear-time fixed-effects models focusing on between-country variation

Notes: Outcome variables are proportion of included groups or population. Unit of analysis is the country-year. All models include year fixed effects. Models 11 and 12 include all leaders, while models 13 and 14 include only first leaders of a country. *p < .10; **p < .05; ***p < .01.

In models 11 to 14, the effect of Pan-African conference attendance remains positive and significant. We find evidence that decolonization wars are associated with more inclusive ethnic ruling coalitions and that high settler levels are associated with more exclusive ruling coalitions.¹⁵² These models also demonstrate that our results hold up when analyzing cross-country variation and when considering just first leaders; thus the findings in models 11 and 12 do not just reflect a generational shift between early leaders, who had a higher likelihood of attending conferences than later leaders.

Further investigating first leaders, we also estimate separate models for each year they were in office. Thus we run a model on the first leaders’ first year in office and every following year (until the tenth year). Figure 4 presents the estimates for Pan-African conference attendance for both our dependent variables. The number of observations in each model is just over fifty country-years and declines over time, as not all first leaders remain in power for ten years. The estimates indicate that after seven years in office the positive effect of conference attendance count becomes more uncertain. Overall, Pan-African conference attendees did not immediately turn away from inclusive coalitions after winning power, supporting the argument that coalitions were not just formed to overcome the immediate internal and external threats to independence and new governments.

Figure 4. Effect of Pan-African conference attendance on ethnic inclusion

Alternative Ideologies: Majority Domination

An alternative ideological explanation for ethnic power sharing, or its lack, is majority “political domination.”¹⁵³ It implies that elites of majority groups have few incentives to share power with minorities. This is stronger in states with a “ranked” ethnic hierarchy, where groups that represent demographic majorities are also potentially perceived as dominant over other minority groups and “relations between ethnic superiors and subordinates” become embedded in the political structure of the state.¹⁵⁴ Within this framework, policies of exclusion are determined by the majority group through a “politics of entitlement” that reaffirms their dominant position.¹⁵⁵ In states with a dominant majority group, it is possible that the group size of a leader's ethnic group is driving variation in ethnic inclusion. This could be a threat to our inference, if leaders representing nonmajority groups turn to Pan-Africanism and inclusive politics as a source of legitimacy. Therefore, we test for whether a state leader's ethnic group is the largest in the population (Table A2 in the online supplement). The effect of conference attendance remains almost unchanged, but we find that if government leaders are representing the largest ethnic group, they include a greater share of ethnic groups in a country, but seem to join with smaller ethnic groups as the overall proportion of included population declines.

Outlier Analysis

Our empirical analysis leverages information from African countries, and our results rely on a subset of countries whose government leaders have attended Pan-African conferences. Hence, there is a risk that our results are driven by a few cases. We address this concern in three ways. First, we plot observed versus predicted observations, initially showing little visual evidence for extreme outliers (Figure 5). Second, we implement a jackknife approach, where we drop one country at a time and re-estimate the main models with the conference attendance count (Table 2, models 3 and 4) and attendance indicator (Table 2, models 7 and 8). Results are shown in Table A2 in the online supplement. Estimates are very stable across iterations, except for three cases. Dropping Egypt or Zambia increases the main effect, which in the Zambian case can be explained by the persistence of inclusive ruling coalitions even after Kaunda left office. The most obvious estimate change happens when dropping Guinea, particularly in the attendance-count models. This is explained by the stark change in ethnic inclusiveness when Ahmed Sékou Touré left office. We believe that the Guinean case is a meaningful observation and should not be disregarded, but even when that case is dropped the size of the effect is meaningful and significant for both the binary indicator (conference attendance leading to 10 percent more included groups and about 15 percent more included population) and attendance count (6 percent more included groups and 7 percent more included population per conference attended).

Figure 5. Predicted versus observed values of the dependent variable

Conclusion

Our contribution highlights the importance of Pan-African organizing in diffusing political ideas and norms that directly impacted the politics of postcolonial states. We find that African state leaders who attended Pan-African conferences built more ethnically inclusive governments. This suggests that structural conditions do not fully explain state leaders’ coalition choices. Rather, ideological exposure and commitments during the anticolonial struggle affected state leaders’ approach to governance after independence.

Our findings have implications for studies of ethnic power sharing and of policies and governance in postcolonial African states more broadly. They suggest that the norms underpinning the ideological framing of state leaders are important in shaping policies of ethnic inclusion and exclusion. We identify Pan-Africanism as a key ideological and political influence on the subsequent political understandings of African elites. Future research could investigate whether African political elites’ Pan-African ideological influence and commitment also affected political decisions and approaches to government in other areas. For example, it could investigate whether Pan-African ideals of ethnic unity also extended into more equal public goods provision and campaign strategies focusing on positional appeals, as opposed to the mobilization of societal cleavages. By including commitment to the Pan-African movement as an important ideological fault line, such research would complement previous approaches to African politics that focus on rational elites navigating structural conditions and discount elites’ ideological leanings.

By taking seriously the actions and norms of anticolonial African elites in developing postcolonial governments, our contribution moves beyond the prevailing approach of analyzing postcolonial states through the structures and policies of colonial administrators. While the legacy of colonial institutions is important, they are not the only determinant of postcolonial outcomes. We highlight that anticolonial African leaders were not just isolated actors responding to colonial institutions within their own states, but often interacted with elites from other states to discuss and formulate postcolonial policies. Future research could investigate policy and norm diffusion between African leaders who established linkages when attending and interacting at Pan-African conferences.

While we have shown that involvement in an ideological movement influences leaders’ subsequent approach to governance, we are not able to distinguish between self-selection, political learning, and the diffusion of normative convictions. If leaders’ ideological exposure affects their policy choices, understanding more about how these convictions formed and how malleable they were in response to different ideological environments and influences, especially at different times in their political career, is important. Struggles for independence were a formative time for many African political leaders and continued to shape African states. Elites in the period of decolonization had to navigate different strategic incentives and ideologically opposed camps to lead their country to independence. Better understanding how African leaders formed their ideological convictions and approaches to political competition and control during this formative time will provide new insights into how they approached power and decided on policy after independence.

Our findings show that exposure to ideas, ideologies, and training at political fora during their formative years play an important role in elites’ decision making. They also show that relevant ideological divides—and by extension political fora—do not always translate easily onto classical left and right scales, but rather depend on the historical and regional context of elites’ formative years. Identifying the most relevant ideological divides in specific regions at specific times can reveal ideological and political learning and influence that would otherwise be missed. For example, a similar shared ideological framework impacting the development of postcolonial states existed among decolonization movements in Asia.¹⁵⁶ Repeated interactions between activists at international fora strengthened their ideological commitment to Asian solidarity and enabled attendees to exchange strategies of resistance to (neo)colonialism.¹⁵⁷ Future research could explore how political and normative instruction at these fora shaped policy and structures in postcolonial states in Asia and how attending elites influenced each other.