Introduction: The Legacy of Colonialism on Social Protection

The boundaries of empire appear to have had a profound and lasting effect on social protection policy, even into the 2000s. Whether a territory was colonized by the Spanish, French, Germans, Portuguese, Italians or British—or the Russians, Japanese or Americans—seems to correlate with both the onset and the subsequent direction and pace of policy reforms. In 2011, Midgley and Piachaud asserted that “social policy in the developing world cannot be understood without examining the way welfare policies and programmes introduced during the imperial era have continued to influence current policy-making” (2011, 10). Their edited collection included suggestive case studies of colonial influence within the British Empire. Schmitt (2015) demonstrates that the timing of the adoption of different kinds of social security programs differed systematically between French, British and Spanish colonies (and former colonies). French colonies were quick to introduce social insurance in the form of family allowances but rarely (even long after independence) introduced any old-age pensions. Spanish colonies were the first to introduce old-age pensions. British colonies varied, but some of them at least were quick to provide for old age (often through non-contributory social assistance), and almost none introduced family allowances. In her chapter in this volume, Schmitt (2019) shows that the probability of a former French colony having introduced any social assistance program is massively lower and the probability of a former British colony having done so is very much higher than in other countries across the Global South. In previous work, I also found that a history of British colonization was closely correlated with high expenditure on social assistance relative to expenditure on social insurance, that is, a characteristic of the British welfare regime (Seekings 2014). The apparently random geography of imperial conquest—that is, which imperial power colonized any particular territory—appears to have had an enduring effect on public policy.

There are three possible explanations for these enduring differences between social protection in different parts of the world. First, the imperial power might have directly influenced policy during either the colonial or post-colonial periods, and there was some form of path dependence thereafter. Second, the imperial power might have indirectly influenced policy, perhaps through shaping the institutional environment. Third, imperial conquest might not have been entirely random, in that local conditions differed between the imperial empires. There are reasons for taking seriously all three possible explanations.

The easy explanation for inter-imperial variation in policy legacies is that different imperial powers simply imposed their own models on their colonies and these models then persisted over time through some process of path dependency. For example, the French imposed their 1952 Code du Travail across the whole of their empire (Cooper 1996), which certainly helps to explain the enduring prevalence of child allowances in former French colonies (Schmitt 2015). The British, however, did not impose any similarly centralized, monolithic model on their colonies. As Schmitt (2015) emphasizes, Britain’s policy was to decentralize policy-making (and financial responsibility) to colonial governments. Unsurprisingly, Schmitt shows, there was more variation for decades after independence between former British colonies than among former French colonies. Nonetheless, post-imperial aid flows—and the influence that accompanies these—track the former boundaries of empire (see Becker, Chap. 7, this volume). British aid and influence, through its Department for International Development (DFID), is thus concentrated in former British colonies. It would not be surprising if DFID’s policy preferences were most influential in former British colonies.

Imperial legacies might also reflect indirect influences. This appears to have been the case with respect to schooling. Most British colonies in Africa had much higher enrolment rates in the 1950s than either French or Portuguese colonies, and this had both direct and indirect effects long after independence. Close analysis reveals that this inter-imperial variation was due not so much to direct differences in expenditure on education by colonial governments but rather to the indirect effects of colonial policies toward Christian missions. Crucially, it seems, British colonial governments allowed Christian missionaries to compete for converts, which they did in large part through expanding education through mission schools (Frankema 2012). Indirect influences might inform path dependence also. Lange (2009) suggests that variation—perhaps inter-imperial in origin—between direct and indirect rules had enduring consequences on the character of the state, politics and public policy. Schmitt also suggests that colonialism “shaped the institutional arrangements of the state and the power and preferences of actors” (2015, 332).

Differences between former British, Spanish and French colonies might reflect exogenous differences in local conditions. The samples of territories analyzed by Schmitt (2015) comprised Spanish colonies in Latin America or the Caribbean, French colonies that were mostly in Africa and British colonies that were much more widely dispersed across the world. The more pronounced heterogeneity in the date of adoption of welfare programs in British colonies might reflect the greater variation in local conditions as well as the British policy of decentralizing policy-making. The challenges facing post-colonial states in Latin America were very different to those facing their peers in Eastern Europe and East Asia (Haggard and Kaufman 2008). Conditions across most of Africa (and parts of South Asia) were very different to those in these three regions. Africa as a whole differed from large parts of Latin America in terms of the relative sizes of indigenous, slave and settler/immigrant populations. Across most of Africa, but only in some parts of Latin America, for example, indigenous, peasant-based agrarian societies survived colonization. Even within Africa, conditions may have differed between those areas colonized by France and those colonized by Britain. The former had larger Islamic populations and contained a smaller proportion of fertile areas than the latter (which might help explain why there was weaker demand for education in French colonies than in British ones—see Frankema 2012; Cogneau and Moradi 2014; Dupraz 2017). Mkandawire (2015) found that taxation and social expenditure in Africa reflected the economic character of a colony, not the colonial power per se. The crucial difference was between cash crop economies, mostly in West Africa, and the “labor reserve” economies, mostly in East and Southern Africa. The kind of economy was clearly affected by colonial policies of settlement and development but also reflected natural and other differences that were exogenous to colonial policy. In Mkandawire’s analysis, the British cash-cropping colonial economies of West Africa resembled their French colonial neighbors, while Rwanda and Burundi resembled British labor reserve economies. The implication of this is that the evolution of a distinct model or models of social protection in Anglophone East and Southern Africa might reflect similarities in local conditions as much as or more than the fact that these territories were colonized by Britain.

This chapter examines these three categories of explanation—and hence when, how and why external actors have been influential—through two case studies of welfare policy-making over time. Both cases were part of the British Empire through the early and mid-twentieth century, and both were in labor reserves rather than cash-cropping territories. Both ended up, by the end of the twentieth century, with variants of the same kind of welfare regime, with a strong focus on means-tested social assistance relative to social insurance (as well as largely tax-financed public health systems alongside large private health sectors). Yet the two cases reveal two very different routes to these outcomes, with external actors playing quite different roles.

First, I consider the case of South Africa which institutionalized a system combining social assistance with “semi-social” insurance (explained below) between the 1920s and 1940s and retains this basic system today, in the 2010s. The South African case is characterized by significant imperial influence in the formative period, followed by strong path dependence, such that the system survived hostility from both the governing National Party under apartheid (from 1948 to 1994) and the governing African National Congress (ANC) following democratic elections in 1994. Second, the chapter examines the case of Botswana, whose welfare state originated in drought relief programs in the 1960s and later evolved into social assistance programs.

The cases of South Africa and Botswana illustrate two very different pathways toward welfare states, with some common characteristics (although the welfare state in Botswana remains more conservative than the South African one, in important respects). While both pathways were shaped by ideas circulating in primarily English-speaking networks, the mechanisms of influence were far more complex than any simple imposition of a British model. Both the initial design of public policies and their subsequent path-dependent expansion reflected the resonance of specific models to local conditions and ideas.

South Africa and Botswana ended up with social protection systems with important similarities—although the South African system is far more generous—and some common features with British models not because they copied British models, but because the basic ideas informing British public policy were shared by political elites in parts of Southern Africa. Crucially, in both South Africa and Botswana social protection policies evolved on the basis of a dominant elite ideology that states should leave activities to the market except if the market fails to provide adequately for deserving groups of poor people. This was “liberal” in the sense that the state’s role in social welfare (defined narrowly as transfers in cash or kind, but excluding public education) was conceived as being residual, provided only for people deemed deserving in that they were unable to provide for themselves (and lacked kin who could provide for them; see Chinyoka and Ulriksen, Chap. 10, this volume). In practice, this meant an emphasis on programmatic support for the specific categories of destitute people considered deserving (especially the elderly and disabled, and less often children, pregnant and breastfeeding mothers and single mothers) as well as (when necessary, especially during episodes of drought) workfare for working-age, able-bodied adults. Local political imperatives drove these initially “residual” programs to expand into universal or quasi-universal ones.

This chapter suggests that all three explanations of the enduring differences in social protection have some relevance in Southern Africa. The case of South Africa suggests that foreign ideas shaped local policy debates, both prior to 1948 (when the most influential ideas came from the UK, Australia and New Zealand) and after 1948 (i.e. under apartheid, when neo-Calvinist theology from the Netherlands exerted significant influence). The design of Botswana’s welfare state was also shaped by external ideas, through the World Food Programme (WFP). Institutional design mattered, indirectly, in that both elements of indirect rule and representative democracy ensured that elites were incentivized to introduce and then conserve pro-poor programs. Both the first and second explanations contribute to understanding path dependency in these Southern African cases. Third, local conditions mattered in both South Africa and Botswana, as they posed specific challenges to the large number of small-scale farmers: climate and ecology (compounded by government policy in the South African case) mattered, framing the expansion of social assistance programs in both cases.

Imperial Influence in the Making of the South African Welfare State

South Africa has long been at the fore of welfare state-building in Africa. As early as 1937, the new Professor of Sociology at the University of the Witwatersrand proclaimed in his inaugural lecture that “Today the provision for [the] European population … is scarcely less complete than that of Great Britain” (Gray 1937, 270). He exaggerated somewhat, even with respect to the “European” minority. The African majority of the population was almost entirely excluded. Nonetheless, the state paid means-tested pensions to elderly white men and women, pensions to white and colored people who were blind or otherwise disabled and cash grants-in-aid to poor people caring for children (as well as to institutions or associations caring for children). During recessions, the state operated workfare programs for otherwise unemployed, mostly white men. Public education and hospital care for white people was financed almost entirely from taxes. In addition, nascent contributory pension schemes were beginning to provide for select groups of workers in formal employment.

While the Union of South Africa was self-governing (as a British “Dominion”), its political and bureaucratic elites were heavily influenced by the ideas and models concerning welfare policy that circulated through the British Empire during the first half of the twentieth century. Government commissions of inquiry and civil servants collected detailed information on the social insurance and social assistance programs in place across much of the world at the time, but they borrowed most from policies in Britain, as well as in the settler societies of Australia and New Zealand. The tax-financed, means-tested social pensions, introduced for white and colored people in 1928–1929, resembled the pensions introduced in Britain itself in 1908 (see Schmitt, Chap. 6, this volume). Subsequent social assistance programs for the disabled and for single mothers and proposals for social insurance, first for pensions and later for unemployment and health also, drew on imperial precedents. Debate in Britain during the Second World War—including around the Beveridge Report, published in 1942—provided a strong impetus to the expansion of social assistance. In the early 1940s, ideas and models from New Zealand were also highly influential. In 1942, a member of parliament (MP) tabled a motion on social security, asking, “Can our government do what New Zealand has done?”Footnote 1 One of the leading activists in civil society in the early 1940s was from New Zealand (Seekings 2005).

The South African system was idiosyncratic, however, reflecting the adaptation of foreign models to local conditions and values. Policy-makers borrowed but did not copy. First, South Africa never implemented a national system of social insurance (“national insurance”). The British model—providing for insurance against health and unemployment from 1911, and contributory pensions from 1925 (Boyer 2019)—was rejected. The only social insurance was a very modest unemployment insurance program. Instead, successive governments required most white workers to join government-regulated but privately operated, mostly sector-specific pension funds (i.e. “semi-social” insurance). Second, the South African social assistance programs were deeply racialized. African and Indian people were excluded entirely from social pensions until 1944. When pensions were extended to them, they were discriminated against in terms of benefits. Social assistance programs were introduced originally by political parties representing white and “colored” voters with the goal of protecting white and colored South Africans from destitution. For some Members of Parliament, this was motivated explicitly by the perceived need to preserve the racial income hierarchy. As one (white) Member of Parliament put it, “not a single white person should be allowed to go under” because the “small number of whites” had to stand together against the “uncivilised hoards”Footnote 2 (Seekings 2007). Racialized solidarity entailed providing for white men, women and children but excluding or discriminating deeply against the African majority of the population.

Social assistance was preferred to social insurance, in part because the leading (white) politicians, bureaucrats, academics, businessmen, trade unionists and other civil society activists were more firmly rooted in English-language circuits of ideas than in continental European models (although they were certainly aware of the latter, in part through the work of the International Labour Organization [ILO]). More strongly even than in Britain, Australia and New Zealand, the modernist ambitions of socialist and “new liberal” welfare state-builders were reined in by the more classically liberal views of much of the white elite, with their preference for the market and a residual state. Modernist ambitions were also framed and contained by the overt or implicit racism of most white South Africans. “National” (i.e. social) insurance was supported by many technocrats, but the political leadership did not deem it necessary. Privately run pension funds provided sufficiently well for white workers who enjoyed privileged access to better-paid formal employment. As in Australia (Castles 1985), white workers were incorporated politically through the regulation of wages in the private sector as well as through public sector and parastatal employment, and through public services, rather than through corporatist or Bismarckian models (Seekings and Nattrass 2005). This combination of wage regulation, “semi-social” insurance and social assistance made sense given that benefits were largely limited to the white minority.

“Poor whites” were politically important not only because elites sought to bolster a racialized solidarity among Afrikaners or white South Africans more broadly, but also because they had the vote in a highly competitive party system. Few black South Africans had any vote. Nonetheless, local conditions pushed the government to extend social pensions to black men and women in 1944. Just as “poor whites” were often the victims of deagrarianization, so the constraints on peasant agriculture in the 1940s were a major factor pushing local officials and even industrial employers to endorse the expansion of social pensions (Seekings 2005).

Policies were never imposed or simply transferred from Britain to South Africa. Officials in London had no direct influence in South Africa. But (white) reformers in the South African state, political parties and civil society officials were embedded in imperial networks through which both norms and policy models were diffused from both Britain and its other settler Dominions to South Africa. Models were adapted, however, to reflect the norms—including racialized and often racist norms—that were prevalent within the South African elite, as well as the political and economic conditions on the ground.

Path Dependency and the Expansion of South Africa’s Welfare State

As of 2018/19, 90 years after the introduction of old-age pensions, South Africa’s welfare state remains a distinctive outlier in Africa. Monthly grants are paid to or for about 18 million elderly people, disabled people and children, or to one in three South Africans. These are paid for out of general taxation at a cost of about 3.5% of GDP. The state also operates by far the largest school feeding program in Africa, reaching 9 million children, as well as large workfare programs. Public education and health care are free for the poor. Contributory pension and medical aid schemes cover many (but not all) people in formal employment (Seekings and Nattrass 2015). The ILO identifies South Africa as the “front runner” in Africa in terms of the coverage of social protection, with an “effective coverage” of 48% of the population (ILO 2017, 123).

While racial discrimination in eligibility and benefits has been abolished, the basic design of the welfare state in the 2010s remains much the same as it did in the 1930s: the state supported the same categories of deserving poor through tax-financed, means-tested grants and able-bodied, working-age adults through workfare programs. Moreover, no social insurance had yet been introduced for either retirement or health care, and the unemployment insurance system covered very few of the unemployed. On the face of it, the post-apartheid welfare state in South Africa retained its original colonial-era design, with essentially parametric reforms that have allowed for the expansion of coverage of existing programs but not the introduction of new programs.

This apparent path dependence is especially striking given that the governing parties both under and after apartheid were ambivalent if not hostile toward the welfare state. In 1948, the more extreme wing of the National Party was elected into office. Not only was the National Party committed to the deepening of racial discrimination and segregation (i.e. apartheid), but its views on the welfare state were also deeply influenced by reformed Protestantism articulated by Dutch neo-Calvinist theologian and politician Abraham Kuyper. This foreign ideology was adapted to the local context in that the National Party rejected entirely the idea of a welfare state for the African population. On the basis of a variety of arguments—including that old-age pensions would encourage “the native to degenerate”,Footnote 3 that they were culturally and socially inappropriate and that they were paid for by white taxpayers—provision for African people was severely restricted. Even with respect to its white citizens, however, the National Party government was ambivalent about its welfare programs. Shaped by Kuyper’s ideas, the party’s leaders argued that families should take responsibility for providing for family members: working people should save for their old age, adult children should take responsibility for their aged parents and adult husbands and fathers should take responsibility for their wives and children. In this deeply conservative view, excessive public provision—that is, the “welfare state”—was “socialistic”, undermined independence and responsibility and fostered instead dependency and delinquency. “Our State is a social welfare State as opposed to a socialist State”, one MP explained; “We must take care of the paupers, the indigents who need assistance. But do not let us give everyone the right to be taken care of, because we deprive our people of their sense of responsibility.”Footnote 4 The National Party’s values were rooted in a neo-Calvinist conception of individual and familial responsibility, combined with an emphasis on a racialized volk. This was much more conservative than the kind of concern with social harmony that characterized social Catholicism and Christian Democracy at the same time in post-war Europe. Even the (white) opposition United Party retreated into a more conservative stance, advocating a residual welfare regime along American lines (Seekings 2020).

Despite their repeated declarations (influenced by religious ideas from Europe) that they were opposed to a welfare state, National Party governments for the most part expanded both social assistance and the semi-social insurance system, especially as they began to retreat from their grand apartheid project in the 1980s. While they sought to strengthen white families, they were unable to ensure that white families accommodated and supported elderly parents. Expenditure grew rapidly on residential institutions for the elderly. Similarly, despite their enthusiasm to abolish social pensions for elderly African men and women, the importance of pensions to most African people created a very large vested interest, while the apartheid-style devolution of responsibility to compliant African leaders gave an effective veto power to players whose modest legitimacy would vanish if they were complicit in the abolition of the pensions. As international and domestic pressure intensified on the apartheid state, persistent racial discrimination in benefits became less and less defensible. In the 1980s the National Party presided over major increases in the real value of the social pension paid to African people as it moved toward parity in benefits. Racial discrimination in benefits was finally ended in 1993, on the eve of the country’s first democratic elections (although racial discrimination remained with respect to public expenditure on education and health care). The semi-social insurance system expanded to accommodate most skilled and semi-skilled African workers in provident funds managed by newly legal trade unions. The existing programs thus not only created directly vested interests but also provided both symbolic and material incentives to groups of African people to demand their inclusion in the welfare state, creating even stronger vested interests.

When an ANC government was elected in South Africa’s first democratic elections, in 1994, it therefore inherited a welfare state designed more than 50 years earlier. While committed to the reduction of poverty among its mostly African support base, the ANC was ideologically ill-disposed to the idea of a welfare state. Without unmanageable debts and independent of foreign aid, the new government was not beholden to any international organizations or aid donors. It was, however, influenced by the developmentalist ideology that had been hegemonic across much of Africa since the 1950s. In his inaugural speech as newly elected President in May 1994, Nelson Mandela himself made it clear that the ANC would implement a developmental approach rather than distribute “handouts”. Repeatedly over the following 20-plus years, ANC leaders reiterated that South Africa needed a “developmental state” rather than a “welfare state”, valued the family over excessive individualism and worried about “dependency”. ANC governments did rein in expenditure on residential institutions for the elderly, considered abolishing grants for poor, single mothers and rejected calls for a basic income grant. They also proposed shifting the emphasis of social protection from social assistance and provident funds to social insurance, through national pension and health insurance programs. Civil society activists pushed for programmatic expansion. Faced with electoral competition, the ANC did expand massively grants for poor mothers (and other caregivers), but they rejected a series of other proposed reforms and (as of 2018) failed to implement their promised national insurance systems (Lund 2008; Proudlock 2011; Patel 2015; Seekings and Nattrass 2015; Button et al. 2018).

The ANC after 1994 found it as difficult as the National Party before 1994 to resist the expansion of the welfare state. The dominant norms within the ruling party might be ambivalent or hostile to welfare statism, but institutions, interests and popular ideas all contributed to path dependency. Institutionally, the consolidation of social assistance programs within the Department of Welfare (later renamed Social Development) created an institutional vested interest. ANC party structures (and MPs), fearing a popular backlash against programmatic retrenchment, wielded a potential veto. Faced with persistent poverty, the ANC expanded programs to maintain its electoral support. The existing programs created a massive vested interest in their continuation—except for residential institutions for elderly white people who comprised a politically weak constituency. Every suggestion that the trade union-run provident funds would be incorporated into a national pension fund system was met with blanket opposition from the politically powerful trade unions. The promised national health insurance system was welcomed by the public sector unions but was resisted by the black as well as white middle classes, who had migrated to private health care after 1994, and prompted ambivalence even among sections of the working class who were covered by sector-specific or other medical aid schemes. Perhaps most importantly of all, the priority attached to deracialization generally meant that programs that had benefitted white South Africans primarily should be extended to all South Africans. The very idea of social assistance had become so taken for granted that grants could be extended to African single mothers and even a basic income grant could be put on the agenda, despite widespread concern over the payment of grants to undeserving individuals.

At no point between the 1920s and 2010s did reforms to South Africa’s welfare state entail simple policy transfer from elsewhere. South African policy-makers were consistently determined to adapt foreign ideas and models to suit local circumstances. While foreign actors never had the power to impose their preferred policies, South African policy-makers were deeply influenced by foreign ideas. Between the 1920s and 1940s, policy-makers drew on some of the ideas and models in Britain (and other British Dominions) in introducing programs to provide primarily for South Africans of European descent. In the South African context, however, these programs served, inter alia, the racialized objective of securing white supremacy. In the early 1940s, debates in Britain and elsewhere influenced some policy-makers to extend some programs to African people, albeit on a discriminatory basis. From 1948, the National Party tried to reverse these reforms, in part because of the influence of a new set of foreign ideas, that is, deeply conservative neo-Calvinist ideas from the Netherlands. In the 1980s and 1990s global opprobrium over racial discrimination contributed to the National Party’s slow deracialization of welfare programs. After 1994, ANC leaders drew on global developmental ideology (as well as their own conservative views about family) to resist the expansion of so-called handouts. Despite the ebbs and flows of ideas about welfare, the welfare programs introduced in the second quarter of the twentieth century were not only never retrenched to any significant extent but tended to expand through parametric reforms that extended the reach of existing programs. South Africa’s welfare state was reformed down a path that depended on its origins.

External Actors, Drought Relief and State-Building in Botswana

The British colony (or, more precisely, Protectorate) of Bechuanaland (renamed Botswana at independence in 1966) was a very different context to its neighbor South Africa. The territory was extremely poor and heavily dependent on remittances sent by migrant mineworkers in South Africa. There were very few European settlers or immigrants. Moreover, British colonial officials enjoyed considerable power until independence, although they chose to devolve considerable authority to Tswana chiefs under indirect rule. At the same time as South Africa was introducing and expanding welfare programs, the British colonial government in Bechuanaland made almost no provision for their subjects.

The absence of welfare programs in Bechuanaland was typical of British colonies in Africa. Britain might have devolved responsibility for policy-making to colonial officials on the ground, but it, nonetheless, generally provided those officials with clear guidance as to what policies they should make. When, during the Second World War, the Beveridge Report had prompted global interest in welfare state-building, Britain’s Colonial Office insisted that Beveridge’s proposals were inappropriate in societies in “early stages of development” where the poor were provided for by kin or community and the priority was to increase agricultural output so as to raise the general standard of living. Even in some industrializing societies—such as the Copperbelt in Northern Rhodesia (later Zambia)—the fact that most industrial workers retired to rural villages obviated the need for “more sophisticated” social welfare programs. Only in societies where there was little or no peasant agriculture, as on islands such as Mauritius, might old-age pensions or other such programs be considered appropriate. In societies without large white settler or immigrant populations, the Colonial Office actively discouraged the kinds of reform already introduced in South Africa (and in some Caribbean colonies). The challenge of poverty among black people would be addressed primarily through agricultural and other economic development, which would generate the resources needed to expand the “social services”, that is, especially education and health care (Seekings 2013).

Colonial states were under pressure to protect colonial subjects against one specific risk: famine, which in Africa usually arose from drought and, more occasionally and locally, from pests (such as locusts) or flooding. The British Empire had a long experience with drought and famine in India. Official Famine Codes set out how colonial officials should respond to famine. Informed by nineteenth-century liberal thought in Britain, the Codes stipulated that free food should be distributed only to the truly “destitute” (including the elderly), while the able-bodied poor should earn minimal rations of food through food-for-work programs. The Famine Codes were sometimes catastrophically inadequate (as evident in the Bengal Famine of 1942) but were generally regarded as useful enough that they (and subsequent operational manuals) were embraced by Indian governments after independence. Colonial officials in Africa faced regular famines and sometimes responded with limited relief programs, although only in the Sudan did officials draft a Famine Code (De Waal 1989; Iliffe 1990). British colonial officials replicated the kind of approach developed in India: wherever possible, drought should be addressed through the market, in that members of poor rural families could work for wages as migrant laborers, remit their earnings and their rural families could then buy food that had been supplied to rural areas by merchants; if this was insufficient, local authorities (typically chiefs) should intervene; colonial governments themselves should intervene only in dire emergency. In the mid-twentieth century this approach proved broadly effective in some territories (including Southern Rhodesia: see Iliffe 1990) but not in others (most notably Nyasaland/Malawi in 1949: see Vaughan 1987).

Bechuanaland, although especially vulnerable to drought, experienced no major droughts in the 1940s or 1950s. In the early 1960s, however, several years of drought threatened mass famine. The colonial government—in the process of transferring power to local leaders—reacted very slowly. In 1965, the Bechuanaland (later Botswana) Democratic Party (BDP) headed by Seretse Khama and Quett Masire was elected as the territory’s new government, and the following year Botswana achieved independence. The BDP seized control of drought relief, securing massive food supplies from the WFP, a then newly established United Nations agency. Food aid was distributed to between one-third and one-half of the population through feeding schemes. The scale of these emergency programs was unprecedented in the territory.

It quickly became apparent that continued food aid from the WFP required a shift in approach. While the WFP could provide emergency assistance for only a limited period, it could support longer-lasting “developmental” programs that helped to increase production as well as keep people alive. The WFP was happy to support school and other feeding schemes for “vulnerable” groups (school and pre-school children, pregnant and breastfeeding mothers) and destitute rations for the elderly and disabled. It was unwilling to provide food to able-bodied adults. The new government of Botswana therefore introduced workfare programs, providing food for work to men and women, with the intention of supporting whole families. By the end of the 1960s, the Botswana government had put in place the key features of the Indian Famine Codes, although there is no evidence that it was even aware of them. Insofar as the design of the programs was influenced from elsewhere, it appears to have been the WFP rather than the Colonial Office in London (Seekings 2016a). While the WFP remains one of the least researched international organizations (but see Shaw 2001), it appears that its approach was framed by the developmentalist ideology that was hegemonic at the time: food might be handed out in short-term emergencies, but the medium- and long-term priority was “development” to improve long-term food security.

Seretse Khama, the first president of Botswana, was steeped in broadly liberal British values. He had been educated at a mission school, then at South African and British universities, had been living (in exile) in Britain for several years, married a British woman and employed British advisors and speech writers (Tlou et al. 1995). His ideology was, however, rooted in Tswana norms. In the late 1960s and 1970s (before his death in 1980), he articulated a benign conservative ideology of responsibility for the poor. He and his successor, Quett Masire, were respectful of the private sector and wary of enlarging the state. But they saw the new state as assuming the responsibilities for the deserving poor that chiefs, communities and kin had shouldered hitherto. In this ideology, the deserving poor were those poor people who were unable to support themselves: the destitute elderly and infirm, children and mothers. Expectations of reciprocity underlay public responsibility. Anyone who could—that is, able-bodied adults—had a responsibility to work, to support their dependents and to contribute to the common good, hence the emphasis on workfare. This ideology of welfare was rooted in agrarian conditions. Historically, the availability of land meant that people’s welfare depended primarily on work, such that relief was required only when people could not work (and could not be supported by working kin) or when natural disasters such as drought meant that, in effect, there was no work. While the ideology has some similarities to more liberal versions of conservatism in Britain (specifically, “one nation” conservatism), its roots lay in a society very different to British society (Seekings 2016b).

While most of the 1970s were years of good rains, Botswana persisted with the programs introduced in the previous decade: school and other feeding schemes (supported by the WFP), destitute rations and (when necessary) workfare. These welfare programs were integral not only to resilient support for the BDP among voters but also to the construction of a modern state. When drought recurred in 1978–1979, the government and WFP together provided drought relief for about 80% of the population. The government formalized its public works program (as the Labour-Intensive Public Works Programme) and introduced a new National Policy on Destitute Persons.

The construction of a welfare state in Botswana was driven forward by further drought in the 1980s. Although cattle died on a massive scale, not a single person died, as huge volumes of WFP food aid were brought into and distributed within Botswana. In 1985, in the middle of the drought, a study of drought relief programs concluded that:

Drought relief is coming to assume a role in Botswana politics comparable to education and welfare in the industrialised countries. Indeed, it is already so popular that the leaders of the BDP have resisted pressures for cuts from bureaucrats. It will be difficult for the Ministry of Finance and Development Planning, which quite naturally concerns itself with balancing the budget, to find a politically acceptable way of reducing the various relief programmes, once the drought is over. (Holm and Morgan 1985, 476)

When the drought finally broke, in the late 1980s, the WFP resumed discussions with the government of Botswana about withdrawing from the country. By then, as Holm and Morgan had already recognized, it had become difficult for the government of Botswana to reverse or to step away from the path it had followed hitherto.

External resources and ideas were thus central to the origins of the welfare state in Botswana. WFP food aid and accompanying ideas shaped what the new state provided, for whom and with what conditions. This was a very different set of origins to those in neighboring South Africa. Programs in Botswana were not aimed at European immigrants or settlers. They did not draw on legislation from Britain (or other settler societies). They were not linked to any racialized political project, but rather to indigenous African norms and values. Their relevance was rooted in local conditions: specifically, the risks of drought in an agrarian society comprising mostly small farmers. As in South Africa, however, they were associated with a modernist project of state-building, they were residual in that poverty reduction should be achieved primarily through economic growth (or “development”) and they were focused on deserving categories of people who could not support themselves (especially the elderly, disabled and children) and only on working-age adults through workfare.

Path Dependency and the Institutionalization of Social Assistance in Botswana

Since the mid-2000s, foreign actors have played a key role in putting ideas about social assistance onto the agenda in many African countries. Scholars and activists have argued for “just giving money to the poor”, on the grounds that it reduces poverty, is “developmental” (or at least redistributes resources in a sustainable way) and is politically “transformative” (Devereux and Sabates-Wheeler 2004; Hanlon et al. 2010; Ferguson 2015). International organizations and aid donors have funded consultants to assist with writing policy documents, spent large sums on study tours, seminars and other events to build coalitions of reform-friendly politicians and officials and embedded advisors in government departments (see Devereux and Kapingindza, Chap. 11, this volume). They have funded programs and sometimes established parallel, quasi-state bureaucracies. Reform has often appeared to be driven by donors (Devereux 2010). While international organizations and aid donors have shared a common enthusiasm for social assistance, their priorities have varied. The World Bank has favored targeted (i.e. means-tested) and conditional cash transfers. The ILO calls for “social protection floors”. UNICEF promotes support for children, while HelpAge International promotes pensions for the elderly (von Gliszczynski and Leisering 2016). In many former British colonies, DFID became a powerful player in promoting the social protection agenda, providing advice, financial aid and technical support. Its preferred model was the unconditional cash transfer with broad reach, not unlike the social pensions and other social assistance introduced in Britain itself in the first half of the twentieth century (Hickey and Seekings 2019).

The institutionalization of social assistance in Botswana did not, however, reflect the influence of DFID or any other international organization or aid donor. It resulted from domestic factors, although it was precipitated in part by the decision of the WFP to withdraw from Botswana (see Chinyoka and Ulriksen, Chap. 10, this volume). The WFP’s withdrawal was a reaction to the mining-fuelled rapid economic growth in Botswana that lifted the country out of the ranks of “low-income” countries. The withdrawal compelled the government of Botswana to choose between assuming full responsibility for welfare programs and retrenching them. Not only did the government take over the programs, but it expanded and institutionalized them further over the following decade.

The first clear statement of the government’s position was in its 7th National Development Plan of 1991, which recognized that Botswana suffered from “a structural poverty problem” in that poverty persisted even in years of good rain (Botswana 1991, 17). Social justice and the decline of extended family support required government action: “Government food aid, its drought relief and recovery programmes, and other aid measures targeted for the destitute are intended to supplement the incomes of the very poor in order to ensure that their disposable incomes, both cash and in kind, provide them with a minimum standard of living” (ibid., 33). The Plan institutionalized funding for drought relief and welfare.

The government assumed full responsibility for massive school and other feeding programs, introduced universal old-age pensions (in 1996) and then provision for orphans and reformed its policy on (other) “destitutes”. These reforms preceded the embrace of social protection among international organizations and aid donors. In the 2000s, international organizations did lobby the government of Botswana to expand its cash transfer programs. The government resisted, although it did expand its workfare programs.

By the 2000s, Botswana was providing modest support, in cash or in kind, to half or more of the country’s population. The country’s elite worries endlessly about “dependency”, but the persistence of poverty—despite economic growth—raises the political costs of retrenching programs. Social change resulted in pressure to expand provision. The introduction of old-age pensions in 1996 was in part a technical response to the challenge of poverty among the elderly, modernizing the existing and clearly insufficient provision of destitute relief in the face of diminishing familial support. It was also a political response to the BDP’s electoral vulnerability. Having won elections comfortably hitherto, the BDP was shocked in 1994 when it won barely 53% of the vote. The expansion of workfare in the late 2000s also reflected electoral anxiety (Seekings 2019; Hamer 2016).

In other parts of Africa, reforms of social protection have generally been negotiated between external organizations and local political actors. External organizations rarely get far, unless their ideas appeal to—or fit with the existing ideas of—at least some important local actors (Hickey et al. 2019). In Botswana, the WFP’s withdrawal was an important factor pushing the government to institutionalize drought relief programs into a welfare state, but there is no evidence that the WFP or any other international organization shaped significantly this process of policy reform. The process in Botswana following the WFP’s withdrawal was an entirely domestic one. Indeed, when the World Bank and UNICEF did (later) propose the introduction in Botswana of a general child grant, the government of Botswana declined to do so (Chinyoka 2019).

Both the expansion of the welfare state in the 1990s and early 2000s and the subsequent rejection of a proposed child grant reflected domestic factors: the BDP government had expanded the welfare state in line with its conservative ideology of responsibility and in response of its perceived political interests, but its norms of responsibility did not entail usurping the responsibility of the family to provide for non-orphaned children nor did it perceive its political interests lying in such a reform (perhaps because it assessed that many voters shared these conservative views). The domestic politics of policy reform was, however, path dependent. The introduction of old-age pensions, most notably, was imaginable only because the idea of poor relief was deeply entrenched through the provision for destitutes and other, originally drought-driven programs.

Conclusion

The cases of South Africa and Botswana entail two different but parallel paths to welfare states built primarily around social assistance programs. While the welfare state in Botswana is more conservative than the South African one in some respects—including the generosity of benefits, the role accorded to family and the preference for paying benefits in kind rather than cash—the basic design is common to both. Social assistance programs provide for the elderly, disabled and some children, with workfare for able-bodied adults. Social insurance remains limited. It is tempting to assume that these neighboring countries have similar systems because of a shared history, including common external influences.

In neither South Africa nor Botswana, however, were foreign models imposed. Insofar as there was policy transfer, it was with considerable adaptation to local norms and conditions. External influences entailed primarily the diffusion of ideas which were combined with local ideas to shape policy outcomes. At different times in both cases, foreign ideas meshed with local factors to drive reform. In the South African case, at other times powerful external ideas served to constrain welfare state-building. In the case of Botswana, external actors in the 2000s proposed expansionary reforms that were resisted by the government.

Diffusion is less surprising in the South African case, where the welfare state originated in initiatives to provide for South Africans of European origin in ways comparable to provision in Europe itself. Even in this case, however, foreign models and ideas were adapted. Adaptation was required because the kinds of risks that were most pressing in the Southern African context were not the same risks that faced workers as a result of industrialization and urbanization in the more industrialized countries of Europe or Latin America. Welfare states in East and Southern Africa, including South Africa, were framed by changes in the countryside. This was most obvious in Botswana, where the welfare state originated in programs of drought relief, to tide rural households through periods in which they could not feed themselves. In South Africa, also, many of the “poor whites” of the 1920s and 1930s—and poor African people in the 1940s—were in or from rural areas, and their poverty stemmed more from changes in rural areas (deagrarianization) than from industrialization and urbanization (see Künzler, Chap. 4, this volume). In both countries, local conditions encouraged policy-makers to prioritize economic growth or “development” and to see welfare programs as providing a residual safety net. The similarities between public policies in South Africa and Botswana thus reflect, in part, similar local conditions.

Policies in South Africa and Botswana developed along paths that led to a common focus on social assistance not primarily because they were exposed to the kinds of ideas associated with British public policy (and the similar developmentalism underpinning the WFP) but because those ideas could easily be adapted to fit local conditions and norms. In both cases, targeted social assistance allowed not only for the provision of support to selected deserving categories but also for the exclusion of other categories of people deemed not to be deserving on the basis of need, custom or capacity. Insofar as we can identify a common model of social protection in Southern Africa, this resulted not from the imposition of any specific foreign model but rather from a process of contestation over ideas and models and the adaptation of these to local conditions.

All three possible explanations of the policy outcomes in these two countries have some relevance. Foreign ideas about the state, welfare and “development”, rooted in or linked to Britain or its empire, were influential. The experiences of welfare reform (and wage regulation) in Australia and New Zealand and of drought relief in India and elsewhere, as well as British models, shaped reforms in Southern Africa. Institutional design was also consequential. The legacy of colonial institutions of indirect rule in Botswana meant that the post-independence state sought to legitimate itself through institutionalizing drought relief. Apartheid-style indirect rule in South Africa meant that the state was unable in the 1950s and 1960s to retrench its social pension program. Finally, local conditions were of crucial importance. Botswana remained an agrarian society of small farmers for most of the twentieth century. Deagrarianization transformed South Africa much earlier, but racialized policies and incomplete industrialization under apartheid meant that the market failed to replace the former agrarian safety net. In both Botswana and South Africa in the late twentieth century, high dependency rates resulted in strong pressures for the continued expansion of social assistance. British imperial influences on policy itself and on political institutions therefore combined with local conditions to steer apparently dissimilar territories down quite different paths to similar outcomes.