Abstract
Popular and political concerns about accelerating migration from sub-Saharan Africa (SSA) are often framed through migration transition theory, which posits that as poor regions develop, migration rates first rise and later decline after surpassing a certain development threshold. SSA is widely viewed by scholars, pundits, and policymakers as the region where migration rates have only just begun to climb the upward slope. Adopting a long-run historical perspective, we revise this image in three respects. First, although migration out of SSA has recently increased, Africans overall have not become more migratory since colonial times. The share of Africans living abroad was high in the 1960s, reflecting the legacy of large-scale interterritorial migration during the colonial era. Second, voluntary mass migration within Africa expanded rapidly in an era that is not typically regarded as “developmental,” after the abolition of slavery in the late nineteenth and early twentieth centuries. Large numbers of voluntary migrants, often extremely poor by historical standards, moved toward zones of cash-crop production and mining, and later increasingly to urban destinations. Third, we argue that the recent surge in extracontinental migration reflects a shift in destinations within a much longer migration transition rather than its early stage. This shift was driven by narrowing spatial opportunity gaps within Africa, growing hostility to intra-African migration in the context of nation building, and the simultaneous expansion of capabilities and opportunities for migration beyond the continent.
Introduction
Sub-Saharan African (SSA) migration has become a headline-grabbing and highly politicized issue over the past two decades (Carbone 2017; Asserate 2018; Natale, Migali, and Münz 2018; Smith 2019). Not only have the number of SSA's migrants to the Global North risen (Figure 1), they have also increasingly moved via irregular channels. The challenges, threats, and opportunities arising from such migration are under close scrutiny in destination and origin countries (Castles 2009; Akinola and Bjarnesen 2024). Increasing extracontinental migration rates are often associated with poor governance, conflict, climate change, economic malaise, and demographic growth in the African migrant-sending regions. But paradoxically, the belief that African migration rates have begun to surge is also grounded in a more optimistic narrative: it is African development that spurs emigration.
Sub-Saharan African Migrant Stock per World Region, 1990–2024. Note: We follow the United Nations classification of countries into world regions. Source: United Nations Bilateral Migration Matrix for 2024 (United Nations 2024a).
The latter view, widely accepted by migration scholars, is grounded in “mobility transition” or “migration transition” theory (Zelinsky 1971; Skeldon 1990; 1997; De Haas 2010). The logic of migration transition theory (MTT) is that “development,” broadly captured by improvements in income, health, education, and political freedom, enhances people's aspirations to migrate in general, and their ability to migrate over long distances in particular. Only beyond a certain development-threshold are such migratory aspirations expected to decline. As African income levels (and thus migration capabilities) are still low, such a transition appears in its early stages, and there is ample potential for rates to increase.
What happens when we apply a crude migration transition logic to Africa? Clemens (2014) uses World Bank and United Nations migrant-stock data to show that emigrant stock rates tend to peak at a GDP per capita of circa 10–11,000 US$ in 2017 PPP prices. 1 South Africa had a per capita GDP of 12,790 US$ in 2019, so it has already surpassed its expected peak emigration rate. Yet, virtually all other countries south of the Sahara are far below this inflection point. Assuming an average GDP per capita growth rate of 5% annually (which is about the rate at which Ethiopia and Nigeria have grown over the past three decades), Nigeria's migration rates will peak in 2034, Ethiopia's in 2047 and DR Congo's in 2066. If we assume a peak emigrant stock/population of 15%, in line with Clemens’ global regressions for 2010, and use the UN “medium variant” population projections the peak stock of emigrants will be 42 million for Nigeria, 32 million for Ethiopia, and 44 million for the DR Congo (United Nations 2024b). These numbers are several orders of magnitude above the estimated international migrant stocks in 2024: 2.1 million for Nigeria, 1.2 million for Ethiopia, and 2.1 million for DR Congo. In other words, if African nations were to follow patterns similar to other developing countries like Mexico, Turkey, or the Philippines, we should expect a continued and rising migration within as well as beyond Africa, for decades to come. But do such projections have any meaning? And how well do we understand the nexus of development and migration in SSA?
A migration transition is an inherently contextual process. Analysis of present-day and future trends of African migration should therefore be grounded in an understanding of ongoing migration transitions in the continent itself. While there is a rich theoretical literature that can inspire research on this question, the application of MTT to Africa has so far lacked historical depth. Migration transitions have been documented for parts of Europe and Asia during the “Age of Mass Migration” (1860–1914), and on a global scale since 1960 (Skeldon 1997; Hatton and Williamson 1998; 2005; De Haas 2010; Clemens 2014; Dao et al. 2018; 2020). Historical African migration – pre-1960 or even pre-1990 – has been unduly ignored for reasons of presumed data paucity, or written off as irrelevant due to the context of colonialism which, purportedly, heavily restricted voluntary migration of Africans and pre-empted development. Recent scholarly work has done little to correct this imbalance. Even those works that take history seriously have drawn parallels between Europe's past and Africa's present and future, rather than analyzing Africa's own historical migration patterns (Hatton and Williamson 2002; 2003; Natale, Migali and Münz 2018; Clemens 2020). Several empirical and conceptual misconceptions about historical African migration have therefore remained unchallenged, limiting our understanding of changing SSA migration patterns.
In this study, we re-evaluate long-term migration transition dynamics in SSA and yield three insights. First, Africa's migration transition has much deeper roots than is often presumed. Even when focusing solely on international migration, the share of Africans living abroad was in fact substantially higher in the 1960s than it is today, despite tangible, albeit erratic, long-run improvements in income, health, and education since (Figure 2, Panel A, also see Flahaux and de Haas 2016). 2 As we will show, the large emigrant population in 1960 reflects pervasive interterritorial migration during the preceding colonial era. The subsequent decline of the total migrant population took place despite development: income per head of the population has increased by some 40%, the average years of education have increased from 1.3 to 5.9 years, and life expectancy at birth has increased from 41.4 to 60.8 years. 3 In 1960, 95% of international movers had done so within the continent (De Haas and Frankema 2022b). Only when we look at migration out of the continent do we see a clear increase since independence, and a stronger positive relationship with income growth (Figure 2, Panel B).
Estimates of GDP per Capita and Migration From Sub-Saharan Africa, 1960–2020. Note: The migrant stock pertains to all Africans residing outside their country of birth except for refugees. Different definitions of who is a migrant explain the difference between the United Nations (which includes refugees) and World Bank estimates (which excludes them). To improve consistency, we manually subtracted refugees from the United Nations figures (1990–2019) using data from UNHCR (2024). In the absence of comprehensive survey data for many African countries, and given the extensive modelling and projecting that went into the datasets, the estimates should be treated as proximate only. Source: Adapted from De Haas and Frankema (2022a, 382), updated with data from the World Bank Development Indicators (accessed September 2025) and United Nations (2024a).
Second, mass migrations under colonial rule emerged in an era that is not typically regarded as “developmental,” but nonetheless deserves to be analyzed through that lens. Colonial rule, while generating substantial forced mobility and involuntary immobility, did not prevent the emergence of large-scale voluntary migration to both rural and urban destinations. Voluntary mass migration rose in the wake of the abolition of slavery in the late nineteenth and early twentieth century. 4 Large shares of these migrants, including many former enslaved, were extremely poor not only by contemporary, but also by historical standards. Initially, voluntary migrants moved mostly between rural destinations, seeking out places of cash-crop production and mining, the latter mostly controlled by colonial companies, the former in part also developed and controlled by African farmers. From the 1950s onward, migration increasingly centred on urban destinations, while the share of migrants seeking out rural destinations declined, albeit gradually and unevenly (De Haas and Frankema 2025).
Third, by tracing back migration in SSA over the long-run and taking it forward to the present, we argue that the key question for Africa's ongoing migration transition is not why Africans are becoming more mobile overall, because there is no evidence they are. Instead, we argue that the primary explanandum is why SSA migrants have increasingly sought destinations out of the continent after 1960. We contend that the main reason for this shift is the narrowing of the spatial opportunity gaps within Africa that had emerged before and during colonial times. Growing hostility in potential intra-African migration destinations as a result of nascent African nation building plays a role too. In addition, extracontinental migration was facilitated by reduced costs of overseas migration, growing migrant capabilities due to educational expansion and income growth, the growth of migration-facilitating networks at origin, along corridors and at destination, and a fluctuating but persistent demand for migrant labor in extracontinental destination countries.
We proceed in Section 2 by discussing MTT and elaborating on its implications for Africa. In Section 3 we develop the argument why the presumption, implicit in some versions of MTT, that African migration in the (pre)colonial era can be set aside as either insignificant or of a fundamentally different (i.e., traditional, nondevelopment-related) nature is flawed. Section 4 provides three distinct arguments to show that the roots of Africa's migration transition should be traced back to the pre- and early-colonial times, and be studied in the contexts of slavery, colonial rule, uneven rural commercialization and the rise of large mining enclaves. Section 5 revisits the reasons why extracontinental migration has been growing, both in absolute and relative terms, since 1960. Section 6 concludes.
Migration Transition Theory
MTT posits that developing countries experience a migratory life cycle or transition, which takes at least several decades to unfold. The principal driver of such a transition is development in the sending region. Migration transitions have been variously linked to the demographic transition (Zelinsky 1971) or to broader processes of economic and social change (Skeldon 1990; 1997; Hatton and Williamson 1998; 2003; De Haas 2010; 2021). Here, we define development as a combination of rising levels of income, educational expansion, and declining mortality rates followed, with a delay, by declining birth rates. We view political freedom as a fourth, migration-enhancing dimension of development. This pertains to the absence of (legal) restrictions on mobility, as well as gains in political stability and security. With development, MTT predicts that emigration rates first rise and decline after reaching a threshold, creating an inverted U-curve of migration rates. Meanwhile, as development also makes societies more attractive to newcomers, they may ultimately shift from net-emigrant to net-immigrant status. 5 Figure 3 depicts these relationships.
Stylized Depiction of the Migration-Development Nexus. Source: Authors’ own, based on De Haas (2010, 19).
Migration scholars have proposed various explanations for nonlinear, development-driven migration. In his original formulation of the mobility transition, Wilbur Zelinsky (1971, 219–222) argued that mobility was a critical aspect of a broader process of modernization that also set a demographic transition in motion. 6 Without offering a comprehensive explanatory framework, Zelinsky posited that migration moves through five phases, parallel to the five phases of the vital transition: from limited circulation in “premodern traditional societies,” to large domestic rural–rural, rural–urban and international permanent migration and circulation during the early and late transitional phases, to urban–urban and circulatory migration in “advanced societies.” Skeldon (1990, 112) significantly updated Zelinksy's transition model. Skeldon's adaptations acknowledged that mobility was in fact extensive pretransition—a crucial point for our analysis—and further specified the nature of migration in the transitional phases, charting in more detail the shifts from rural, to (small and large) urban, to international destinations, and specifying the sex ratio of these different migrant flows (see also Skeldon 1997; De Haas 2010, 9).
Economists and economic historians have further elaborated the theoretical and empirical underpinnings of MTT by focusing on the role of shifting age structures, educational expansion, and the lessening of financial constraints (Hatton and Williamson 1998; 2002; 2003; Clemens 2014; Dao et al. 2018; 2020). Emigration, then, is expected to rise when an expanding, young, and increasingly educated labor force is confronted with limited domestic opportunities for social mobility. A society 's failure to absorb growing labor supplies in times of rapid population growth incentivizes younger generations to migrate to areas with open land frontiers or tighter labor markets. Educated workers have an additional drive to capitalize on their knowledge and skills elsewhere. Remittances lessen financial constraints and experiences of earlier migrants reduce transaction costs of chain migration. Eventually, as demographic pressures decrease and domestic economies grow, migration rates decline.
For migration sociologists, the development-migration nexus revolves around the concepts of aspiration and capability. Albeit couched in a different academic tradition, these concepts refer to comparable causal mechanisms (Carling 2002; De Haas 2010; 2021; Carling and Schewel 2018). The central idea is that rising welfare levels spur peoples’ aspirations to explore socio-economic opportunities elsewhere that are not (yet) present at home. Meanwhile, the capabilities to act upon such aspirations rise when migrant-origin regions develop. Remittances and the “demonstration effect” generated by returning migrants further increase migration capabilities and aspirations. After reaching a threshold, migration capabilities will continue to rise, but aspirations decline as spatial opportunity gaps narrow. Figure 4 shows these relationships.
Stylized Depiction of Migration Aspirations and Capabilities. Source: Authors’ own, based on De Haas (2010, 17).
The Precolonial Origins of Individual Migration in Africa
Misconceptions about African Migration “Before Development”
As illustrated in Figure 3, in the MTT framework both emigration and immigration rates rise from low initial levels. Contemporary migration is thus modeled as a deviation from a supposedly sedentary historical state. Zelinsky (1971, 230) explicitly situated the first stage of the mobility transition in a “premodern traditional society,” with “little genuine residential migration and only such limited circulation as is sanctioned by customary practice in land utilization, social visits, commerce, warfare, or religious observances.” In his iteration of MTT, Skeldon critiqued this assumption of low migration “before development” as the “myth of the immobile peasant,” pointing at the prevalence of short-term circulation (pilgrimage, warfare and navigating complementary ecologies) and permanent movement (slavery and “the opening up of agricultural frontiers”) (Skeldon 1990, 31–32; 112; Skeldon 1997; Lucassen and Lucassen 2009).
However, as pointed out by Bakewell (2008) and Skeldon (2012), scholars and policymakers engaging with the migration-development nexus still work from the assumption that migration rates have only recently begun to rise from a low “pre-development” starting-point. While not necessarily denying that there were earlier forms of migration, such migration is usually written off as categorically different from, and thus irrelevant for, the subsequent transition. A clear example of such presentist bias can even be found in the historical work of Hatton and Williamson who, in their survey of global migration, argue that “many African countries have neighbors who are at similar levels of development. Thus, the incentives for cross-border migration have not been large enough in most of Africa to induce any secular migrant floods” (2005, 252). The idea that African migration before 1960 was limited is further reinforced by the widespread, but—as we will show incorrect, belief that colonial governments were able to effectively control movement within and beyond their empires (Herbst 1990, 186; De Haas, Castles and Miller 2020). What further contributes to this shallow historical depth is that most recent scholarship on global and African migration transitions has focused on international migration, even though it is only one type of migration within a broader migration complex (King and Skeldon 2010). As theorized by Zelinsky (1971), Skeldon (1990; 1997) and De Haas (2010; 2021), international—let alone intercontinental—migration provides only a partial insight into ongoing migration transitions as it tends to gain salience in later phases of the transition (Skeldon 1990).
Although not explicitly using MTT, Marxist scholarship took a somewhat longer view of Africa's migration transition, associating migration with “uprooting” arising from growing global inequalities, engendered by colonialism and capitalism. Samir Amin has provided the most influential version of this argument for Africa. In his view, European colonialism induced a shift from what he referred to as “traditional migration” involving “mass movements of peoples,” to modern migration of laborers taking “their place in an organized and structured host society” (Amin 1974, 66; Amin 1995, 29). For Amin, historical African societies were not immobile altogether, but they engaged in wholly different types of group-based migration, such as nomadic pastoralism and frontier-mobility. Such frontier-group migrations occurred in a non-commercialized, subsistence-oriented economy. In a seminal study, historical anthropologist Igor Kopytof (1987, 7) also rejected the notion that Africans were “mired in timeless immobility” before they were either “uprooted” or “developed” under European rule. At the same time, the core idea that precolonial African migration was “traditional” and involved groups rather than individuals remained largely unchallenged (Kopytof 1987; Ehret 2002; Wang et al. 2020).
More recent migration and development scholarship has only superficially analyzed African migration patterns before and under colonial rule. In online case study material appended to the widely used textbook The Age of Migration: International Population Movements in the Modern World (De Haas, Castles and Miller 2020), it is argued that various forms of African migration in the later decades of the twentieth century were an outcome of the “colonial destruction of African economies and societies,” thus espousing an unqualified underdevelopment explanation for contemporary African migration. Indeed, stylized dichotomies between traditional and modern migration keep resurfacing in recent scholarship on Africa (Baker and Akin Aina 1995; Cohen 1995; Adepoju 1998, 387; De Bruijn and van Dijk 2003, 287; Usman and Falola 2009; Cordell 2013, 180–188; Fernandez 2013, 135; De Haas, Castles and Miller 2020). Schewel and Asmamaw’s (2021) study on Ethiopia's mobility transition is one of very few exceptions which, if not in terms of data, at least in its conceptualization, traces interrelationships between different types of migration (including nomadic pastoralism) going back to the early twentieth century.
Was There a Break between Traditional and Modern Migrations?
We will now argue that precolonial African mobility was not always “traditional” and group-based. Instead, it was often market-driven, responsive to specific historical conditions, and not uncommonly individual and even voluntary. Enslavement, a prime driver of precolonial African mobility, was a dynamic and diverse institution that was widespread across large parts of the continent all the way into the late nineteenth century, and inextricably tied up with migration (as noted by Skeldon 1990). African slavery was both ancient and ever-evolving. Successive surges of the Trans-Saharan, Trans-Atlantic and Indian Ocean trades in the sixteenth to nineteenth centuries greatly intensified slavery in Africa: they raised the number of people and regions involved, spurred investments in military capacity, induced states to specialize in raiding and trading, and enlarged the degree of exploitation to which slaves were subjected. While African slave-holding societies were typically keen to procure female slaves for household duties, agricultural labor and sexual and reproductive roles, male slaves were preferred in the Transatlantic trade, giving rise to a gender division in domestic and export slavery.
To enslave a person implied the deliberate cutting of someone's ties with their relatives and homeland. To become a master's property, a slave was forced to integrate in a new household or community. The market value of captives increased the further they were taken away from their homelands, as this reduced the risk of escape (Lovejoy 2000, 91; Allen 2021). Enslavement, thus, almost by definition, involved migration. Moreover, the prospect of enslavement through warfare and raiding both induced flight and displacement of African communities. Individual, voluntary, and long-distance mobility between communities, however, became hazardous and vulnerable people thus preferred to stay close together in fortified or otherwise protected settlements (Austin 2022). Other forms of voluntary mobility were, however, stimulated by the slave trades. For example, as slaves were transported over long distances, their mobility supported elaborate diasporic trading networks stretching over expansive geographical zones.
Regardless of the brutality, there is little point in writing expanding slave migrations off as traditional forms of migration, or even as premodern, with little relevance for subsequent modern or developmental migration. For sure, slave migration itself was not driven by voluntary explorations of spatial opportunities gaps, neither by personal aspirations or capabilities. At the same time, however, the slave trades were deeply commercialized, capitalist, dynamic and directed toward the integration of subordinated individuals in society. It is telling that some of the routes that slaves travelled for centuries, such as the ones through the Sahara desert, are still in use by migrants today (Scheele 2012; Saleh and Wahby 2022).
It is also important to note that not all precolonial migration of individuals was linked to slavery. In nineteenth century East Africa, impoverished pastoralists sought refuge among sedentary farmers, working for them and living among them, sometimes attempting to return back to a pastoralist lifestyle (Waller 1985; Hakansson 2022). These were certainly individuals taking “their place in an organized and structured host society,” complicating Amin's boundary between traditional and modern migration. Meanwhile, in the latter half of the nineteenth century, a culture of migratory wage labor emerged among the Nyamwezi caravan porters, who, unlike most of their counterparts in other parts of the continent, were earning wages and transported ivory to the coast and cloth back into the hinterland on a voluntary basis (Rockel 2006). Manchuelle (1997) has shown that among the Soninke of West Africa voluntary cross-community migration of individuals also had deep roots, predating colonial rule.
We have already mentioned trade diasporas, of which there were many across the continent, from the Juula and Hausa in West Africa, to the Swahili and Nyamwezi on the East African coast (Curtin 1984; Rockel 2006; Lydon 2009; Röschenthaler 2023). These diasporas often involved religious and literate individuals. Their numbers might have been comparatively small, but these were clearly not migrants who responded to “traditional” incentives such as soil exhaustion or population pressure. Indeed their relative significance across time and space depended on conflict, revolution and religious renewal, as well as shifts in the demand for commodities such as gold, ivory and slaves. Many of these diasporas have left deep imprints, which still shape migration patterns today, for example across the West African interior and the Sahara, and in the East African corridors that connected the Great Lakes Region with the Indian Ocean coast.
The abolition of slavery during the early decades of colonial conquest and rule (c. 1880-1918) and the commercialization of the African countryside triggered the rise of large-scale voluntary mobility (De Haas and Frankema 2025). Emancipated slaves left their masters en masse, to return to home or seek a new life elsewhere, while others fled to areas where abolition was enforced (Rossi 2014; Rodet 2015; Becker and Liebst 2022). This opened up new migration pathways and connections. The suppression of slave raiding and the “pax colonia” which arose after violent and disruptive European partition and conquest, made it safer for people to be on the move. 7 The abolition of slavery and the expansion of export agriculture created local labor shortages which voluntary migrants could fill (De Haas and Travieso 2022). Hence, adopting the terminology of MTT, the abolition of slavery expanded people's capabilities to respond to growing spatial opportunity gaps, and so it is not surprising to find large parts of the continent in migratory flux by the early twentieth century.
A Migration Transition under Colonial Rule?
Did Colonial Rule Prevent Africans from Moving Voluntarily?
Colonial rule was a major driver of mobility in Africa. Colonial powers were often in the business of forcibly moving people, in order to create nature reserves, to eradicate sleeping sickness, to control nomadic pastoralist mobility, for purposes of white settlement, to reduce population pressure in what were perceived to be overpopulated areas, or to populate areas with high economic potential. In other instances, colonial policies were geared toward restricting mobility, both at the sending and receiving end, for purposes of political control, to maintain a local labor supply or to enforce (urban) racial segregation. A broad range of policy tools were used, including taxation (hut and poll taxes), employment rules (master and servant ordinances), land ownership (native reserves and settler land alienation), mobility passes (“health passes”), (forced) relocation, and (forced) labor recruitment. But can colonial-era migration—as per Amin—be set aside as top-down engineered uprooting?
The rapid development of large, capital-intensive mining areas in South Africa, Northern Rhodesia and the Belgian Congo can shed a first light on this question. The rise of large gold, diamond and copper mines was impossible without the inflow of hundreds of thousands of African migrant workers. In the early stages of their operation, these workers were often forcibly recruited (Juif 2022). European settler farms in South Africa and Rhodesia were also major recipients of labor migrants whose mobility was deliberately engineered by creating crowded “labor reserves” with limited options for income earning beyond a meagre subsistence level, as well as taxation regimes that pushed Africans onto a poorly remunerated labor market (Crush 1995). Cheap labor was one of the reasons for forcibly resettling some 3.5 million black South Africans into rural homelands under apartheid (Jeeves 1995; Abel 2019).
That said, even in such conditions the majority of migrant workers had at least some degree of agency in their mobility decisions. Their migration was often not spontaneous, in the sense that they were regulated through agreements between governments, such as between South Africa and Portuguese Mozambique. Migration of workers from colonial Mozambique to South Africa constituted one of the largest singular bilateral flows in Africa (Juif 2022). They would work in the South African mines for an agreed amount of time, stipulated in a labor contract. They would receive part of their salaries upon return in Mozambique to discourage permanent settlement. Mozambicans were subjected to particularly heavy labour obligations to the Portuguese colonial state (Alexopoulou and Ribeiro da Silva 2022). But migrant workers did enjoy a considerable degree of agency in terms of the decision (where) to move and the terms of their mobility and employment (Dolan 2025; Harries 1994). This was true in other Southern African contexts as well. Even in some of the “whitest” spaces, such as Cape Town, black Africans were able to use mobility as a strategy to improve their lives (Mbem and de Haas 2023). Moreover, as the colonial era progressed, migration tended to become less exploitative. The migrant labor force of the Central African Copperbelt even became some of the best-remunerated African manual workers, with comparatively good access to health and education facilities (Juif and Frankema 2018).
Moreover, as Thaddeus Sunseri (1996; see also Cordell 2013) has argued, Southern African migration regimes do not translate well to other parts of the continent. Colonial states outside of Southern Africa did seek to control labor mobility, especially when mines (such as the tin mines on the Jos Plateau in Nigeria, or the Kilimoto Gold mines in the eastern Congo) or settler farms (such as in the Kenyan highlands or the Kivu region in the eastern Congo) were involved. However, colonial control over migration was often tentative. Colonial officials, imbued with the idea that Africans were sedentary and traditional and would therefore not migrate spontaneously, were oftentimes caught completely off guard when they chose to be mobile. Their ingrained biases prevented them from fully appreciating the migration aspirations that many individuals already harbored (cf. Bakewell 2008). In 1923, in Belgian Ruanda-Urundi, colonial officials pondered about the possibility of mobilizing labor for the Congolese mines, but concluded that the people of what are today Burundi and Rwanda were “too attached to their country to envisage the possibility of taking them out of their environment. 9 However, within several years, some 50,000–100,000 migrants were circulating annually between Ruanda-Urundi and neighboring Uganda, a British Protectorate. They did so voluntarily and without any direction by the colonial state, largely ending up working for African cotton and coffee farmers in the Buganda region of Uganda. Belgian officials observed these moves to their “British neighbors” impotently, and with disapproval. By the end of the colonial period, population census data shows that immigrants from Rwanda and Burundi made up over a fifth of Buganda's population, with a large female share (over a third), suggesting substantial local settlement, mostly in rural areas (De Haas 2019).
Manchuelle (1997), in a monograph aptly titled Willing Migrants, argues that the primacy of migrants’ own aspirations explains the emergence of a widespread and versatile diaspora of the Soninke people of Senegal and Mali. The Soninke diaspora stretched all the way beyond the Senegalese coast and into France and formed the basis of a large Senegalese diaspora there. De Haas and Travieso (2022) give more examples of mass circulation in East and West Africa. They demonstrate that the economic gains of migration (earning higher wages, exploiting seasonal complementarities or engaging in price arbitrage) could be very substantial. Meanwhile, voluntary migration often also contained elements of flight or resistance against colonialism. Upper Volta (today's Burkina Faso) became a major migrant-sending region, first to the Gold Coast (Ghana), and later, after the abolition of forced labor there, to Côte d’Ivoire. Asiwaju (1976) has argued that at least some of this trans-imperial migration should be interpreted as a defiant response to repressive French colonial policies. A similar “voting with their feet” took place from harsh and odious forced labor regimes in Portuguese Africa to neighboring British colonies and South Africa (Alpers 1984).
Table 1 provides an overview of the largest migration flows in colonial SSA. The numbers in the table are mostly based on observations by colonial officials, border statistics, employment data, and census records. The data are scattered, and inevitably come with limitations, biases and a considerable error margin. At the same time, the estimates are an outcome of careful scholarly interpretation and contextualization in each of the individual cases, and thus provide indicative approximation of the actual numbers involved.
The Major Migration Systems of Colonial and Early Postcolonial Sub-Saharan Africa.
Note: For purposes of clarity and brevity, present-day country names are used. The estimation of peak annual flows should be considered indicative only, given that no consistent statistics were collected on annual migration flows in almost all cases and the numbers had to be indirectly inferred from employment censuses, information about contract duration and incidental estimates and observations by contemporaries. Only migration destinations that attracted at least 50,000 migrants annually for a prolonged period of time are considered here. Many more smaller migration flows and systems existed across colonial Africa, such as those pivoting on tea plantations in Nyasaland (Malawi) (Palmer 1986) and sisal plantations in Tanganyika (Tanzania) (Alpers 1984; Sabea 2008), both of which relied on local as well as Mozambican laborers, or the settler coffee plantations in the Kivu area of the Eastern Congo, which relied on labor from across Belgian Africa (Van Melkebeke 2020)—just to mention a few. For a map visualizing the major flows c. 1960, see Hance (1970, 147; redrawn in De Haas and Frankema 2025). The references are a selection. We have prioritized sources that contain useable statistics and targeted information on migration, if these exist (which is not the case for all of the systems listed here)
Of course, people's decisions to move were situated in a setting of colonial intrusion and were partly driven by a lack of opportunity and even regression of livelihoods in the sending regions. Our point here is that this does not render migration under colonial rule fundamentally and categorically different from African migration today, which is equally influenced by poor and often repressive governance and lack of opportunities in migrants' origin regions. Colonial rule hardly curbed migration, nor did it control it to such an extent that we can set this era aside as irrelevant for understanding subsequent migrations. Rather, African migrants already proved highly responsive to uneven spatial opportunity structures in sending and receiving areas.
The research on African historical migration has not evolved sufficiently to precisely estimate migration numbers and rates, but that colonial-era migration rates were high is beyond doubt. Based on earlier population census-based calculations (Zachariah and Condé 1981), De Haas and Frankema (2022b, 7) place a lower-bound estimate at 7.5 million people (17 percent of the region's total population) who permanently resettled from the interior to the coast of West Africa west of Nigeria alone between 1920 and 1970 (see also Mabugonje 1972). For all intents and purposes, a migration transition narrative that starts in 1960 would miss out on momentous migratory dynamics in the preceding half century or more.
Did Low Capabilities Inhibit Mass Mobility?
Central to MTT is the idea that initial gains in income from a low starting point result in growing migration rates, as it increases people's capabilities and aspirations to move. However, a striking aspect of migration in colonial Africa is that voluntary migrants, including those travelling over large distances of over 1,000 km and crossing colonial or even imperial borders, often came from areas where extreme poverty was an almost universal condition. Among these sending regions were Upper Volta (Burkina Faso), Nyasaland (Malawi), the Soudan (Mali), Mozambique, and Ruanda-Urundi (Rwanda and Burundi). Moreover, most of the migrants’ destinations, such as the cash-crop producing zones of coastal West Africa and the Great Lakes region, were also very poor, certainly when viewed from today's standpoint. Demographic trends suggests some improvement of living standards. Although knowledge of mortality and fertility rates before 1950 remains tentative, most scholars point to the second quarter of the twentieth century as the time when populations started to grow, often recovering from a period of stagnation or even decline during the disruptive years of colonial conquest (Manning 2010; Doyle 2013; Frankema and Jerven 2014; Sevdalakis, Bras and Remund 2023).
Was it primarily poverty and adverse livelihood shocks—potentially induced by colonial policy—that drove migration? Or did migrants seek out opportunities despite their limited capabilities? In-depth historical studies of migration systems show a nuanced picture in which both deprivation and aspiration factored into migrants’ decisions. In the case of mass migration from Ruanda-Urundi to Uganda, De Haas (2019) shows that migrant flows involved mostly poor peasants, and were affected by conditions of high population density, commercial isolation and a high degree of socio-economic inequality. The introduction of a colonial tax, the occurrence of major food shortages in the late 1920s and early 1940s, and onerous colonial labor demands instituted during the Great Depression of the 1930s all played a role too. Migrants sometimes walked the entire journey of well over 500 km. Especially during episodes of famine, many migrants arrived at their destination too weak to work for the first days or even weeks. Some perished along the way. At the same time, migrants often returned home after several months or years with substantial amounts of cash and assets, such as textiles, which they could use to establish their own household, buy livestock, and pay bride price, which in turn created a demonstration effect for other prospective migrants. Initially, the Ruanda-Urundi migrants managed to double their wages by migrating to Uganda, but over time the wage premium declined, not in the least due to the effect of mass migration on labor markets at both origin and destination. This was one reason why migrants began to settle at their destination (De Haas 2019).
In other cases of colonial-era mass migration, such as from Upper Volta (Burkina Faso) to the Gold Coast (Ghana) or from the Soudan (Mali) to Senegal, aspiration played a more important role than deprivation. Here, the wage premium was much higher than in the case discussed above, with migrants able to triple or sometimes even quadruple their earnings (De Haas and Travieso 2022, 238). As noted earlier, the ways in which migrants, particularly in West Africa, exploited seasonal complementarities by migrating in the agricultural slack season also testifies to the opportunity-seeking nature of their mobility. In all of these cases, people migrated to obtain and bring back specific items, such as foreign currency (shillings) and textiles. Migration also interfaced in important ways with expanding modern education and Christianity—carried to a large extent by European missionaries as well as a fast-growing number of African converts (Meier zu Selhausen 2019)—as it had done with Islam in the nineteenth century. Christianity and modern education expanded people's world view beyond the local thus fostering curiosity and openness toward distant places. They also created new bonds of morality and understanding between previously distant peoples, contributing to the spread of new ideas, sometimes ending up subverting the colonial order (Peterson 2012).
Recent studies on the long-run impact of voluntary migration in colonial Africa have shown various positive effects on development in origin regions related to remittances. For circular migration from Malawi to the South African mines, Dinkelman and Mariotti (2016) have observed a positive effect on education, and Dinkelman, Kumchulesi and Mariotti (2024) on nonfarm employment. Denton-Schneider (2024) finds lower HIV rates in the regions of Mozambique that formerly were recruiting grounds for the South African mines, which he links to migration providing the capital necessary to pay the bride price and marry young. Dupas et al. (2023) find that regions of Burkina Faso exposed to forced labor migration to Côte d’Ivoire under French rule have lower fertility rates today. They argue that it is not the forced labor itself, but subsequent voluntary migration from the same regions that drove the fertility decline, as it reduced the reliance on subsistence farming and child labor, which in turn lowered fertility preferences.
The conclusion that Africans responded to spatial opportunity gaps also lines up well with GDP per capita estimates in 1950, as shown in Table 2. Incomes in the sending regions were close to a bare-bones survival level at the time. Angus Maddison, who first compiled this dataset considered $400 (at 1990 prices) to be the subsistence minimum. Others have set the absolute minimum at $300 (see Milanovic, Lindert and Williamson 2011, 262, fn 17). The only migrant-sending country considered here that appears substantially above this level is Mozambique. This is partly driven by the fact that Mozambique had a substantial European settler community, and that its southern regions benefitted from close integration with the South African mining economy, not in the least through large-scale labor migration and remitted wages. The average level is misleading, however. Unskilled wages and rural incomes in central and northern Mozambique were much lower, and labor regimes there were characterized by a very high degree of coercion (Alpers 1984; Ishemo 1995).
GDP per Capita in Country Pairs with Major Migration Flows in 1950.
Note: GDP per capita is expressed in 1990 (year) international Geary-Khamis dollars. Source: Maddison Project Database v. 2013 (Bolt and van Zanden 2014).
On the destination end, Table 2 shows a variety of average income levels. For the cases of Ghana, Nigeria and Uganda, Broadberry and Gardner (2022) have extended the constant-price GDP series back in time into at least the early twentieth century. Their data suggests substantial income gains in all three cases: 111% for Ghana between 1885 and 1960, 61% for Nigeria between 1885 and 1960, and 45% for Uganda between 1904 and 1960. For French West Africa, a federation of colonial territories including Côte d’Ivoire, Mali and Senegal, Cogneau, Dupraz and Mesplé-Somps (2021) estimate GDP growth of 94% between 1905 and 1960. The annual per capita growth rates that underpin these gains are not spectacular—1.21% for French West Africa, 1.01% for Ghana, and lower rates for Nigeria and Uganda—and were heavily based on agricultural commodity export expansion. They do suggest, however, that expanding opportunities in some colonies induced intra=regional migration flows.
The Migration-Development Nexus before 1960
If we want to understand ongoing migration transitions in Africa we must engage more critically with the fact that African individuals chose to migrate in large numbers before 1960 and, in light of that revision, also reconsider what drives migration shifts post-1960. Even in the nineteenth century, before colonial rule, large spatial opportunity gaps existed across the African continent. In that context, migration was often involuntary, although, as noted, the slave trades also induced sizeable voluntary mobility. In the twentieth century, spatial opportunity gaps widened, as slavery was abolished and some regions became major exporters of agricultural and mineral commodities, attracting large flows of voluntary migrants. These migrant flows were initially centred on agricultural and mining areas. From 1950 onward, migrant destinations became increasingly urban, as cities grew fast. Urbanization rates rose from 11% in 1950 to 23% in 1980 and over 40% in 2020 (United Nations 2018). Most of this rapid urbanization was driven by rural–urban migration, especially in its early stages (Meier zu Selhausen 2022).
These overlapping developments largely predated—but cannot be separated from—the sustained increase of extracontinental migration since 1960. Thus, we argued that migration transitions in Africa have much deeper roots than is commonly considered. Even though extreme poverty was near universal in migrant-sending regions during the colonial period, intraregional spatial opportunity gaps were growing, the abolition of slavery and the “pax colonia” increased peoples’ capabilities to move, and the demonstration effects generated by early generations of migrants as well as new religious ideas and expanding modern education fostered people's aspirations to migrate.
Why Is Migration out of the Continent on the Rise? A Re-Evaluation
Thus far we have argued that Africans were already engaging in voluntary migration on a large scale in the half century before independence. International migration rates postindependence have not increased, despite tangible, albeit erratic, improvements in income, health and education over the long-run (cf. Flahaux and de Haas 2016). Instead of seeing the rise of migration out of SSA as evidence of growing migration rates due to development, this rise primarily constitutes a shift in destinations from within to outside the continent. This changes the explanandum: why has African migration become increasingly centred on extracontinental destinations since 1960? Our aim in this section is not to survey the extensive literature that exists on this more recent period, but to review six main reasons why migration destinations—rather than migration rates per se—have changed.
First and foremost, since independence, and especially as the majority of African economies came crashing down during the 1980s (see Figure 2), the spatial opportunity gaps that had opened up within the region during colonial times, shrunk during the last quarter of the twentieth century. Declining exports of cash-crops and minerals in the wake of collapsing world market prices reduced the demand for immigrant labor in former migrant-receiving regions. Decades of growing (circular) migration now turned into an oversupply, putting downward pressure on migrant wages (De Haas and Travieso 2022). Rapid urbanization and population growth put urban public goods provision under pressure, and led to rising urban un(der)employment and expansion of slums, thus also reducing opportunities for lucrative rural–urban migration. Accelerated expansion of schooling in a context of limited job opportunities for skilled labor, also put pressure on skill premiums, reducing the attractiveness of migration to cities for skilled workers (Frankema and van Waijenburg 2023). Indeed, the role of migration in urban growth has declined in the third quarter of the twentieth century, sometimes even resulting in counterurbanization (Beauchemin and Bocquier 2004; Potts 2009; Menashe-Oren and Bocquier 2021; Meier zu Selhausen 2022). Meanwhile, the economic decline of formerly growing economies like Ghana (in the 1970s) and Nigeria (in the 1980s) reduced opportunities for intraregional international migration as well. There were some exceptions to this general pattern, most importantly Côte d’Ivoire (until the 1990s), Gabon, and South Africa (Flahaux and de Haas 2016). But all taken together, the prolonged and widespread economic depression incentivized migrants to look for places elsewhere—further afield, and often outside the continent.
Second, even if African migrants wanted to move to neighboring countries, their status there often remained precarious. As argued above, people had moved across cultural, ethnic and territorial borders with relative ease during the colonial period. Labor shortages in the most commercially developed regions were often acute and there were no citizenship rights and few public resources to call upon. Postcolonial states, seeking to strengthen the social contract through the provision of jobs and public services became more reluctant to tolerate “strangers,” especially in periods of economic downturn or civil conflict (Cohen 2019; Frankema 2022). While intra-African trade is widely viewed as an opportunity, support for mobility remains much more contentious and fickle (Fernandez 2013; Moyo, Laine and Nshimbi 2021). Given that people had often moved without official recognition, they could be made illegal (Adepoju 1984; Bredeloup 1995). For example, while ECOWAS members support free circulation in principle, this did not stop Nigeria from expelling millions of migrants in 1983 and 1985, amid falling oil prices and economic decline. The specter of precarity, harassment and even mass expulsion, added a further incentive for migrants to look for places elsewhere, although many continued to move intraregionally (Flahaux and de Haas 2016). South Africa is an important case, which continued to attract large numbers of migrants despite rising xenophobia, marginalization, and violence (Kok et al. 2006).
Third, technological change induced long-distance migration by increasing both migrants’ aspirations and capabilities. Largely, this is a story of reduced economic and noneconomic costs of migration. Most obviously, the proliferation of modern transportation such as airplanes, has made long-distance journeys cheaper. Perhaps even more importantly, modern information technologies—social media, mobile phones, mobile money, etc.—have expanded the possibility for long-distance communication. This has not only increased knowledge about distant places as potential destinations (and thus migrants’ aspirations to move there), but also gave people better opportunities to maintain multi-sited lives, for example through phone and voice calls and sending remittances electronically. Thus, technology did not so much increase aspirations and capabilities to migrate per se, but it did substantially increase aspirations and capabilities to migrate intercontinentally.
Fourth, Africans’ growing educational attainment has increased both aspirations and capabilities for long-distance migration. Although attainment rates and educational quality lag behind other world regions, schooling for boys and girls has expanded massively since the colonial era (Baten et al. 2021). One reason for educated Africans to migrate out of the continent is to seek further education and pursue studies in North America and Europe (including the former Soviet Union), but also in various Asian and other countries. As job prospects and the general socio-political situation worsened in many African countries during the last quarter of the twentieth century, many African students decided to stay abroad, while others were keen to join the diaspora. Improved educational attainment of course also increases opportunities to find stable and well-paying jobs in places like France, the United Kingdom, Canada or Qatar. Many African doctors and nurses have left Ghana, Nigeria, Kenya, and Uganda. Even irregular migrants from Africa, who rely on smuggling networks and dangerous migration routes, tend to be comparatively well-educated compared to their nonmigrant peers, probably because their education enhanced their capabilities and aspirations to attempt one or multiple crossings. Tellingly, 85% of all irregular migrants in Europe interviewed for UNDP's Scaling Fences project came from towns and cities (UNDP 2019, 29), and migration origins were strongly concentrated in the most developed and urbanized parts of the countries of origin.
Fifth, network effects operated in receiving and sending countries. Initially, the vast majority of Africans migrating out of the continent did so to the former colonial metropole, responding to specific labor market opportunities that existed there, or to follow education in the language of instruction at colonial and missionary schools (Lucas 2015). Today, destinations have diversified and links with former colonizers have become less pronounced, although they are still visible in the bilateral migration statistics (Zeleza 2008; Kosack and Sargent 2025). Senegalese migrants, for example, initially moved almost exclusively to France, but the prime destination of new migrants shifted to Italy in the 1980s and Spain in the 1990s (Beauchemin et al. 2018). In 2019, only 50% of African extracontinental migrants from former French African colonies lived in France; for migrants from former Portuguese colonies living in Portugal the share was 59% (cf. Lucas 2015, 1452; United Nations 2019). Refugees, whose presence outside of the continent became more marked from the 1980s onward, followed a different path. Large Ethiopian, Eritrean and Somali diasporas fanned out across the global North. In each of these cases, network effects were crucial, with initial bridgehead communities stimulating the arrival of further migrants from the same national, ethnic, or local background. These network effects were also at play at the place of origin, driven by return migrants, smuggling networks, and other facilitators of migration. Benin City in Edo State Nigeria is the key example of a migrant-origin network effect that results in large-scale intercontinental African migration. Even though Benin City is a relatively small city of 1.5 million inhabitants and Edo State hosts less than 3% of Nigeria's total population, the region perhaps accounts for close to half of all irregular Nigerian migrants in Europe—a clear example of migration as regional economic specialization (Beber and Scacco 2022).
Lastly, external demand for African migrant labor has been a key driver too. Sub-Saharan Africans have been employed in Europe since the early twentieth century, albeit in very small numbers. 8 In the postwar decades of the 1950s and 1960s, attempts to moderate wages in rapidly growing Western European economies, drew large numbers of actively recruited “guest workers” from Europe's internal and external peripheries, particularly Algeria, Turkey and Morocco. At the time, there was free movement between (former) colonies and the metropoles, a right that was gradually abolished from the 1960s to the mid-1980s, when the implementation of free circulation within Europe induced greater barriers at Europe's external borders (Vickstrom 2014). By this time, sizeable African bridgehead communities had already formed across Europe. While migrants have always found their way into “dirty, dangerous, and difficult” jobs, Europe's labor markets tightened again from the 1990s onward. This intensified demand for migrant labor in agriculture, care work and the informal economy. Job market opportunities in the United States and Canada also expanded and the number of African migrants in North America has increased substantially over the past decades (for the Ghanaian case, see Arthur 2017). The Gulf Region has also become a major migration destination due to its oil wealth and concomitant ability to attractively remunerate labor, albeit mostly for temporary work. Ironically, the most prominent narrative about migration to all these three regions is one of restriction and rejection. Yet, as De Haas (2008) has argued, despite restrictive migration policies and narratives, African migration to Europe is catalyzed by the demand for precarious, cheap and flexible migrant labor, and there is no genuine intent on the part of either sending or receiving countries to stop migration altogether. Overall, policy restrictions in Europe have tended to increase African irregular migration and make the migrants’ moves more permanent, rather than reducing migration altogether (Beauchemin, Flahaux and Schoumaker 2020). On their side, African governments who benefit from incoming remittances, have been reluctant to curb labor migration to Europe, or even actively promote and support it, for example through bilaterial labor recruitment arrangements and diaspora engagement policies (Turner and Kleist 2013).
Conclusion
The widespread belief that African migration rates have only recently started to increase and should be expected to continue climbing is hampered by a knowledge gap about the region's long-term migration trajectories. Adopting a historical perspective, and by zooming in on migration dynamics within Africa, we have aimed to correct the presentist misconception that African countries, due to their history of poverty and colonialism, have only recently (after 1960 or even 1990) begun to experience voluntary and development-driven migration on a large scale. Although systematic quantification of migration prior to 1960 remains wanting, we have invoked a sizeable literature demonstrating that large spatial opportunity gaps existed prior to, and further widened during, the colonial period, and that migrants had the capabilities and aspirations to respond to such gaps.
The fact that demographic growth, educational expansion and economic growth have roots going back to the early colonial era, and that these processes were already intertwined with large scale rural–rural and later rural–urban migration, deserves acknowledgement and closer research. In this dynamic, the abolition of slavery also plays a crucial rule. If there ever was a fast expansion of voluntary international migration rates in a context of (uneven) economic growth, it was in the century before 1960, not thereafter. It is helpful to summarize our argument visually. In Figure 5, we plot migration to rural destinations, urban destinations and extracontinental destinations. Albeit tentative and essentially qualitative, these trends represent our reading of the extensive historical literature on intra-African migration before 1960, and the available migration statistics since 1960. More research, well beyond the scope of this paper, can further enhance our understanding of rates and trends. The future projections are, of course, even more tentative.
Stylized Depiction of Voluntary African Migration to Various Destinations, 1850–2050. Note: Drawn up by the authors. Large but temporary extracontinental flows of African migrants during the two world wars (Killingray 2022) have not been factored in.
A long-term secular analysis of African migration has repercussions for how we view Africa's ongoing migration transition. Africans’ growing presence outside their continent is not a sign of growing migration rates, but rather of shifting destinations. This shift is hardly the first one in African history. With the abolition of the trans-Atlantic slave trades, African migration shifted decidedly inwards, inaugurating the “age of intra-African migration” (De Haas and Frankema 2022b; De Haas and Frankema 2025). Subsequently, and aided by the abolition of slavery within Africa, African migrants converged upon commercially vibrant rural areas where cash crops were grown and minerals extracted for export. Toward the middle of the twentieth century, urban destinations (sometimes emerging out of mining sites) became dominant. The shift toward extracontinental destinations can thus be viewed as another shift in this long-term story of proactive opportunity seeking through migration.
What is also not new is the ability of African migrants to tolerate precarity and hardship, en route and at the destination. Indeed, historically, most migrants originated from some of Africa's poorest countries and regions. They used simple means to migrate—often their feet—and poverty did not mean that they lacked the necessary aspirations and capabilities to undertake such moves. Further research will be required to develop a more precise picture of trends across time, and a greater understanding of regional differences within the continent.
Does a fuller consideration of the past affect the way we view the future? We wouldn’t go as far as to argue that future economic growth—which certainly remains to be desired in an African context to alleviate poverty—will not expand overall migration rates, and especially migration out of the continent. However, given the deep roots of opportunity-seeking migration in Africa, and given the pragmatism with which Africans have long sought out opportunities through migration, it is helpful to envision other scenarios as well. For example, intra-African migration might regain its momentum, if some countries and regions pick up the pace of economic development much faster than others. It is also possible that overall migration rates will not change at all, as they hardly have over the past six decades—or even the past century.
The extent to which the emergence of large-scale voluntary migration in the early twentieth century despite low levels of development is peculiar to African history deserves more (comparative) research. One possible conjecture, albeit speculative, is that compared to other world regions, Africa's migration transition took off at a much lower level of development, will peak at lower average levels of income than has been the case elsewhere, and may therefore already be much further advanced than is often presumed. Much remains to be studied and learned about the long-run patterns. A closer mutual engagement between historians of Africa and migration scholars today can help to address this knowledge gap.
Footnotes
Acknowledgements
We thank Hein de Haas, Kerilyn Schewel, editor Holly Reed, three anonymous reviewers, and participants of the “West African Migration and Diversity Seminar” at the WZB Berlin Social Science Center for their valuable input.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We gratefully acknowledge financial support from the Netherlands Organization for Scientific Research (NWO) for the projects “South–South Divergence: Comparative Histories of Regional Integration in Southeast Asia and Sub-Saharan Africa since 1850” (Vici Grant VI.C.201.062, Frankema) and “A Good Crisis Gone to Waste? How the 1930s Great Depression Deepened Africa’s Primary Commodity Dependence” (Veni Grant VI.Veni.211F.047, De Haas).
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.