The politics of social protection expenditure and financing in southern Africa

Abstract

Social protection is expanding in southern Africa, but consideration of its fiscal base is usually limited to affordability concerns. Little attention is paid to the different sources of revenue or how the interests of contributors to social protection may affect spending priorities. This article suggests there is a link between revenue source and social protection spending. Aid dependent countries' social protection policy is mostly determined by donors. The governments of countries that rely on natural resources or Southern African Customs Union revenue are relatively free to shape social protection policy. Only in countries that rely on domestic tax-based revenue, where the government must consider the interests of the taxpayer, is there something resembling a social contract for social protection, in which the citizens engage with their government through an exchange-based logic. This article concludes that a broad and diversified tax base is an important mechanism for creating a reciprocal relationship of this kind and thus increasing social spending.

Keywords:

• development aid
• fiscal policies
• resource rents
• social contract
• social protection
• taxation

1. Introduction

Since the mid-1990s many countries in southern Africa have introduced or extended social protection policies, be they safety nets, poverty targeted transfers or categorical provisions such as social pensions. Some of these programmes are supported by donor agencies while others are funded by governments using revenue from natural resources or general taxation of trade and consumption. Only in a few cases does direct taxation of the population constitute a relatively large share of governments' fiscal base for social (and other) spending.

Attention is increasingly being paid to the financing side of social protection policies. However, the issue of financing tends to focus on affordability and the importance of political commitment for social spending (Hickey, 2008; Taylor, 2008; Samson, 2009). Yet when discussing how to create political will and foster ‘social contracts’ for social spending, the literature on social protection has paid little attention to how the choice of social protection policies, and their possible expansion, may be determined both by who benefits from such policies and who pays for them.

This article explores the link between social protection expenditure and financing in southern Africa. Theoretically, it investigates how different sources of financing may affect spending priorities and discusses how the notion of a social contract for social protection expansion can best be understood as an exchange-based logic that includes both beneficiaries of and contributors to social protection policies. Empirically, it examines the link between revenue source and priorities of social protection expenditure in southern Africa. The empirical section also compares four countries in the region (Zambia, Lesotho, Botswana and South Africa), exploring the extent of a social contract for social protection. The empirical analysis suggests that the revenue source tends to determine spending priorities for social protection and that social contracts between states and citizens are fairly rare in southern Africa.

2. Connecting the politics of spending and the politics of financing

The question of social protection expenditure is often considered separately from the challenges of domestic revenue generation. If the issue of financing is taken into account, the concern is usually how best to allocate a given amount of resources. However, if social protection is ‘considered to be more than a residual category that merely compensates for market failures … the financing side has to be treated as an integral part of the problem, and by extension, the solution’ (Hujo & McClanahan, 2009:4). There is a need for more research that links financing, particularly taxation, and expenditure (DiJohn, 2010).

Connecting the spending and financing sides allows us to recognise that the source of revenue is closely connected with the type of expenditure: support for and choice of social protection policies is determined both by who benefits from such policies and who pays. In essence, policy making – whether fiscal or social – is not merely a technocratic exercise but the outcome of a political bargaining process where different social and economic groups will seek to promote their policy preferences.

Many studies have focused on ‘the politics of social spending’; that is, how the interests and mobilisation of different groups and their political negotiations have affected the priority of specific social policies and the extent of social spending (see for instance Esping-Andersen, 1990; Castles, 1998; Huber & Stephens, 2001; Haggard & Kaufman, 2008). It is for instance well established that low income groups tend to support social protection spending whereas the well-to-do are more reluctant to support it – particularly if they do not benefit themselves (Esping-Andersen, 1990). In environments of perceived fiscal austerity, the threat of social spending cuts is always looming as governments have to decide which of many important areas of spending to prioritise. However, retrenchment of social protection policies may be difficult politically (Pierson, 1994), particularly when those policies provide broad coverage of the population. Beneficiaries of social protection policies become a constituency of which politicians need to be conscious – the wider the reach of social policies and the stronger and better organised the beneficiaries, the harder it will be to cut social spending (Skocpol, 1991; Pierson, 1994).

There has been less scholarly focus on the ‘politics of social financing’ (Steinmo, 1993; Kato, 2003; Hujo & McClanahan, 2009), which, interestingly, is quite contradictory to the politics of spending. This contradiction can largely be attributed to inconsistencies in public attitudes towards spending and financing (Steinmo, 1993). Citizens desire public spending, and everyone wants to benefit. Hence, social protection spending may be easy to introduce and will be politically popular among beneficiaries. On the other hand, as mentioned, once introduced such social benefits may be hard to cut. The situation with financing is the reverse. Citizens – though they want to benefit – are loath to pay. Tax cuts are politically popular and relatively easy, whereas it is politically difficult to introduce new taxes – particularly if no rewards are in sight for contributors. Hence, policy making is a political process, where policy choice is largely determined by the preferences and interests of beneficiaries as well as contributors. Crucially, as is elaborated in the next section, continued payment can only be assured in the long term if contributors get something in return.

The above discussion of social policy financing focuses on taxation, but it is likely that other revenue sources equally influence the priorities for social spending. The next section describes some types of revenue source salient in the southern African context and discusses how each is likely to affect social protection spending priorities.

2.1 Revenue sources and priorities of spending

Raising domestic revenue is critical to support sustainable social protection spending. In addition, the actual source of revenue is likely to affect the spending priorities. The first type of revenue source considered here is development aid, which provides substantial financing to the poorer countries in southern Africa. Aid makes important contributions to the budgets of low income countries with weak domestic revenue bases, but it also has drawbacks. For instance, social protection interventions are most effective in the medium to long term, which clashes with the short-term horizon of donors (Barrientos, 2008). Aid is also volatile and unpredictable and disturbs the relationship between the state and its citizens as the government must be accountable to donor agencies rather than the general population (Sindzingre, 2009:123–4).

Development aid may seem altruistic, but it is also motivated by the political and economic self-interests of donor countries (Lancaster, 2007). The revenue source of aid is ultimately taxpayers in donor countries. To ensure continued support, aid agencies must provide evaluation reports that demonstrate developmental impacts, efficient spending and sustainability. Even if they attempt to promote local ownership and sustainable development, donors tend to prioritise short-term projects with visible outputs (De Haan, 2009).

Revenue from natural resources appears to offer a convenient opportunity to minimise donor influence. Governments raising revenue from natural resources benefit from a more stable revenue source and are free to prioritise their own spending. However, countries reliant on natural resources often encounter substantial wastage on power preserving activities by the political elite and general misallocation (McGuirk, 2010). As with donor funds, the government gains revenue independently of the citizens, which renders citizens ‘unable to exert leverage on the government for public service provision and responsible management’ (Moss, 2011:5). There is in fact some indication that high reliance on natural resources causes political leaders to lower the tax burden on citizens, which in turn reduces citizens' demands for democratic accountability (Bornhorst et al., 2009; McGuirk, 2010). Thus, the spending of revenue from natural resources is likely to be determined by the political elites, and citizens may have limited leverage. However, as the political elite aim to ensure continued legitimacy in democratic elections, politicians may choose policies that have wide benefits and immediate and visible impact (Kjær & Therkildsen, 2011).

It has increasingly been noted that, unlike the effects of aid and resource rents, taxation improves democratic representation and state accountability as taxes on the population increase their incentives for public participation and raise their demands for prudent spending (Ross, 2004; Bräutigam et al., 2008; Gupta & Tareq, 2008). Given the recognition that taxation is a stable revenue base and that adequate tax policies can improve public revenue, it is surprising that taxation has received conspicuously little attention in research on social protection (Hujo & McClanahan, 2009; Lesage et al., 2010).

The literature on taxation and state-building follows an exchange-based logic, where the state and citizens establish a contract through which representation and accountability are granted in return for taxation (Timmons, 2005; Bräutigam et al., 2008). A few scholars link taxation directly to government spending. Ross (2004:234) points out that ‘citizens ultimately care about the “price” they pay for the government services they receive’. Bates and Lien (1985) emphasise that a revenue-seeking government will find it to their advantage to strike bargains with the citizens they want to tax such that the state defers to citizens' policy preferences in order to induce a greater willingness to pay tax. In fact, it can be assumed that citizens want benefits and services from the state but that ‘they would rather someone else pay’ – the challenge then is to persuade them ‘to see beyond narrow interests and to contribute to the collective welfare through tax’ (Lieberman, 2002:93).

Consequently, states tend to provide more benefits and services to taxpayers who contribute more revenue. In a quantitative study of about 90 countries, Timmons (2005) finds that the state will provide the goods desired by the income group that contributes the bulk of revenue – mainly the protection of property rights in the case of the wealthy and basic public services and improved social welfare in the case of lower income groups. In a similar vein, the more tax-intensive welfare states in the West have a high reliance on taxes from labourers' wages, and in turn provide generous welfare spending in line with the preferences of such lower income groups (Cusack & Beramendi, 2006).

The question is not so much whether the tax system is progressive or regressive as from which social group the revenue is raised. For instance, steeply progressive and narrow taxation may cause government to target spending at the main contributors, i.e. high income groups. Conversely, a tax burden borne by all income groups can generate substantial revenues (potentially more so than from a narrow tax base), which in turn can be spent on generous and broad based social protection policies (Steinmo, 1993; Kato, 2003). Alternatively, as is common in many African countries, a state could have a regressive system in which neither upper nor lower income groups are taxed intensively, with subsequent low levels of public spending (Timmons, 2005).

It is also necessary to pay attention to the financing mix (Barrientos, 2008). If a state is able to diversify its tax base, it widens its engagement with different sectors in society and is less reliant on a narrow interest group (DiJohn, 2010). Furthermore, different types of taxes call for different types of bargaining logic. If a tax is easy to evade, the state will need to give a higher return to ensure compliance (Bates & Lien, 1985). If a tax is invisible (such as consumption tax), citizens are less likely to be aware of their contributions and hence less demanding. If, on the other hand, a tax is visible (such as income, profit and property tax), affected citizens are likely to demand substantial returns. Direct and visible taxation may require greater capacity to enforce, but such taxes also reflect greater levels of state–society relation and cooperation (Lieberman, 2002). In fact, direct taxation is a key mechanism in building a relationship based on reciprocity; it becomes an important element in a social contract between a government and its citizens – more so than other revenue sources, as the next section explains.

2.2 Social contract for social protection expansion

It has been suggested that the extension of social protection ultimately requires the development of a politically sustainable social contract (Barrientos & Hulme, 2008; Hickey, 2008; Devereux & White, 2010). The idea of a contract between the citizens and the state has intuitive appeal. However, the social protection literature remains unclear as to the content of such a contract and who participates in it.

Stemming from the works of Hobbes and Grotius, early ideas of a social contract were built on the principle of reciprocity. Basically a device for mutually gainful cooperation, where both parties gain advantages and fulfil obligations (Haddock, 2008; Sen, 2010), the social contract principle arguably corresponds well with the exchange-based logic discussed above, where a government and its citizens establish a contract through which privileges or benefits are provided in return for tax revenue or other contributions. Following this logic, contributors to and beneficiaries of social protection are equally important to the existence of a social contract. However, given that governments need not rely solely on domestic revenue sources, a social contract between a government and its citizens can in reality become meaningless – or one-sided at best. Take for instance development aid as reflected in Table 1, which is an externally controlled revenue source. Not only do donors have a strong influence on spending priorities, but the extent of a social contract is also limited as recipient governments need to be more accountable to donors than to their own citizens.

Table 1: Relation between revenue source, spending and extent of social contract

Revenue source	Priority of spending	Extent of social contract
Externally controlled rents (development aid)	Donor set priorities: short-term spending, visible / developmental outputs e.g. pilots, pro-poor grants	Recipient government oriented towards / accountable to donor
Domestically controlled rents (natural resources / SACU)	Elite prioritise: the elite prefer minimal but broad-based spending to ensure legitimacy	Citizens are beneficiaries, but not contributors: they have limited leverage
Taxation	Citizens (taxpayers) have impact on spending priorities	Interdependent relationship between government and citizens
Note: Rents are understood as non-tax sources of revenue and include revenue from aid, natural resources, and (in southern Africa) the SACU (Southern African Customs Union).

Revenue from natural resources as well as the SACU (Southern African Customs Union)2 may improve the extent of a social contract in that governments need not be oriented towards external actors. Instead, this type of revenue is domestically controlled. Yet the relationship remains lopsided and based on an exchange that requires less commitment. As citizens do not contribute, politicians rely on the citizens merely for electoral support and not as a revenue base. Hence, citizens may receive benefits and strongly support the continued existence of such benefits, but they have limited leverage for determining the priority of spending and ensuring accountability.

Taxation is clearly the instrument where the citizen gains most leverage vis-à-vis the government. However, the character of the social contract will vary with the nature of the taxation system. As explained above, the larger tax contributors have a stronger influence on policy spending – particularly if such taxation is highly visible and hard to enforce. Quite simply, the ‘tax-expenditure nexus signals the fundamental social values of society, the balance of social forces and the kind of “social contract” they have arrived at’ (Mkandawire, 2010:1664).

Southern African countries rely on a number of different revenue sources. The nature of the revenue affects the priority of social protection spending and the extent to which there is a social contract for social protection expansion. This is the focus of the following sections.

3. Social protection programmes in southern Africa: An overview

This section describes social protection programmes and the various sources of financing in southern Africa. Supporting this account, Table 2 offers an overview of general national wealth per capita, social protection expenditure, and the share of income (personal and corporate) taxes, indirect (mostly value added) taxes, and rents of total revenue. Rents are non-tax sources of revenue and include revenue from natural resources, aid, and the SACU. Rents from the SACU are essentially revenue from customs and excise and therefore stem from trade and consumption by people (not only citizens) and companies in the union. However, as it is South Africa that collects the revenue and distributes it to the other member states in accordance with a specific formula, SACU funds to Botswana, Lesotho, Namibia and Swaziland are an indirect revenue source; some may even argue that these countries receive financial support from South Africa through this mechanism.

Table 2: Social protection expenditure and revenue sources in southern Africa

	GNI per capita (PPP 2008 US$)	Social protection expenditure (%)^a	Income taxes, % total revenue	Indirect taxes, % total revenue	Rents,^b % total revenue
Aid dependent
Malawi	911	5	24	25	35
Zambia	1 359	6	32	22	25
SACU beneficiaries
Lesotho	2 021	..	15	11	62
Swaziland	5 132	6	19	10	68
Namibia	6 323	8	23	16	54
Resource dependent
Angola	4 941	..	6	5	80
Botswana	13 204	4	12	12	69
Domestic resource base
South Africa	9 812	15	58	34	..
Mauritius	13 344	20	16	48	1
Sources: IMF country reports and statistical appendixes; UNDP Human Development Report (2010); South African Budget Review (2009); Zambia: Social Protection Expenditure and Performance Review and Social Budget, ILO (2008). Data for most recent year. Comparable data for Mozambique and Zimbabwe could not be obtained.
Notes: GNI = gross national income; PPP = purchasing power parity. ^aExpenditure on social security and welfare as a percentage of total government expenditure. ^bRents include revenue from natural resources, SACU and/or aid.

Table 2 lists the countries according to their main revenue sources. Malawi and Zambia are the poorest and the only ones largely reliant on donor funding. Both have very limited social protection legislation, and only a small fraction of the population is part of any formal social security schemes (Social Security Administration, 2009). In recent years, donors and governments have introduced various types of social transfer programmes aimed at the poor.

In Zambia donors have supported five pilot schemes of social cash transfers targeted at the neediest. These schemes were implemented by the Ministry of Community Development and Social Services, which is also responsible for other programmes such as the national government-financed Public Welfare Assistance Scheme. There is a stark contrast between these programmes and the pilot schemes, as the latter receive generous donor funding with ample technical support whereas the government-run means-tested social schemes are underfunded. In fact, despite only covering pilot projects the donors' contribution to the social assistance budget is double that of the government (ILO, 2008; Devereux & White, 2010).

Pro-poor food and cash transfer programmes in Malawi have also largely been externally driven and, despite their perceived significant impact, have remained outside the country's mainstream social protection discourse and without firm government partnership. The government itself has, despite donor antipathy, implemented a Farm Input Subsidy Programme to improve food security for the poor and vulnerable; the programme has been popular among the rural poor (Devereux & White, 2010; Chirwa & Dorward, 2010).

In Lesotho, Swaziland and Namibia, domestic taxes (income and indirect) account for a relatively small share of total revenues, even proportionally less than in Zambia and Malawi. Unlike their poorer neighbours, these countries receive little aid; instead their governments' revenue comes mostly from the SACU. In Swaziland and Lesotho, incomes from the SACU account for more than 60% of total government revenue. These two southern African kingdoms have limited social protection legislation, though both introduced a non-contributory, universal and government-funded old age pension in 2005. There is little to indicate that the introduction of the pensions was due to any form of popular pressure. Nevertheless, once introduced the small rights-based pensions became popular amongst the electorate (Pelham, 2007; Social Security Administration, 2009; Devereux & White, 2010).

At independence in 1990 Namibia already had some social protection in place. This has subsequently been expanded: the old age and disability pensions have been extended, and the child maintenance grant has increased 10-fold since 2003. The pensions are not means-tested, and though the child grant is, the conditions do not appear to be strongly enforced (Levine, 2010). From 2007 until 2009 a coalition of civil society organisations funded and implemented a pilot project that provided a BIG (Basic Income Grant) to all residents below the age of 60 in Otjivero-Omitara, a typical Namibian town. Although it has been lauded as a great success in fighting poverty and fostering social development, the Namibian government has not been willing to adopt the programme and introduce BIG nationwide (Haarman et al., 2009; Van den Bosch, 2011). The Namibian government receives 37% of its revenue from SACU and 17% from natural resources. Revenue from income taxes is somewhat larger than in Swaziland and Lesotho, though the actual tax base is narrow – conservatively calculated, less than a quarter of the working population pays taxes (NPC, 2006; Weidlich, 2007).

Direct and indirect taxes also play a limited role in the resource dependent countries of Angola and Botswana. In Angola about 80% of government revenue comes from oil exploitation, whereas mineral wealth in Botswana accounts for about half of government revenue (with an additional 18% from the SACU). While limited institutional capacity may reflect inability to enforce taxation in Angola, this is hardly the case in well-administrated Botswana. Instead, about two thirds of the Batswana households have such low incomes that they are exempt from paying income taxes (Ulriksen, 2010).

The extent of social protection in Angola is unclear, but it is assumed that the country has limited social programmes and focuses on building health and education services instead. Given Botswana's national wealth, stability, democracy and good governance, it is something of a surprise to find that of all southern African countries, this country spends the smallest percentage of its government revenue on social protection. Of course, a larger overall budget provides more actual money to spend. Even so, most of Botswana's social cash transfers are means-tested, and the universal non-contributory pension offers payments much smaller than those offered in Namibia, South Africa and Mauritius.3 Botswana has prioritised spending on social services but has focused little on ensuring social protection. Only the well-paid and the formally employed receive social security, and the government offers minimal relief for the poor and vulnerable through various programmes (Nthomang, 2007; Ulriksen, 2010). This is in stark contrast to the only other two countries in the region – South Africa and Mauritius – that are economically and developmentally on a similar level to Botswana.

South Africa and Mauritius are exceptional in southern Africa as both countries have more diversified income bases and rely heavily on domestic taxes (incomes from individuals and companies as well as indirect taxes on goods and services). Though Mauritius traditionally relied heavily on progressive income taxes, the country has increasingly, and in line with international trends, shifted towards indirect taxes (Bräutigam, 2008; Ulriksen, 2010). In South Africa, on the other hand, income tax from individuals and companies still constitutes the largest share. Revenues from mining are included in corporate tax, of which minerals account for about 13%. Personal income tax amounts to 29% of total revenue. However, like Namibia, South Africa's tax base is narrow, with about five-and-a-half million taxpayers out of a population of 49 million (SARS, 2009).

Mauritius and South Africa are also remarkable for prioritising social protection. Mauritius has a long tradition of social welfare that combines social security schemes with social cash transfers across income groups. For instance, all citizens of 60 years or older receive a generous pension that can be complemented by a contributory pension. In addition, Mauritius is the only country that caters for the working age with a government-funded and strictly targeted unemployment benefit4 (Bunwaree, 2007; Ulriksen, 2011). South Africa also has a long tradition of social protection – although under apartheid the level of benefits depended on race. Since the mid-1990s the South African government has dramatically extended social protection schemes. There are various types of grant, with the Child Support Grant, Old Age Pension and Disability Grant being most widely used. Though grants are means-tested, the income threshold is fairly high so that many South Africans qualify for a grant. The number of grants beneficiaries is close to 14.5 million (SASSA, 2010).

The southern African examples reflect a pro-poor focus on social protection and a trend of sparse domestic revenue generation, particularly from direct taxation. Many different factors may influence the priority of social protection spending, but it is evident from the southern African countries that differences in social protection spending are also related to the source of revenue. Hence, the aid dependent countries experience a split between donor preferences and national programmes, whereas countries reliant on domestically controlled ‘rents’ have more freedom to prioritise social protection spending. South Africa and Mauritius (and Namibia to some extent) are interesting in that both taxation and social protection spending are relatively comprehensive for the region.

4. The revenue–expenditure nexus and social contracts: Four case studies

Despite differences across the southern African countries, it is possible to discern some trends regarding the link between revenue source and priority of spending. In order to explore further how this revenue–expenditure nexus affects the extent of a social contract for social protection, I compare two low income countries with different revenue sources (Zambia and Lesotho) and two middle income countries with different revenue sources (Botswana and South Africa). These comparisons are based on the analytical framework shown in Table 1, and the findings suggest that social contracts vary depending on the dominant revenue source and related spending priorities.

4.1 Aid revenue compared with SACU-based revenue in low income countries

It has been argued that in aid dependent countries such as Zambia, the politics of social protection spending are donor driven, largely because most social programmes are financed through aid (Taylor, 2008; Niño-Zarazúa et al., 2010). As discussed earlier, donors in Zambia have promoted social protection pilot schemes targeted at vulnerable groups. Such schemes can be considered well-funded ‘islands of excellence’ where outputs are visible and the effectiveness of the schemes can be evaluated (by comparison with districts that are not part of a similar scheme).

Donor funded programmes in Zambia are unconnected to government programmes. However, given their resource dominance donors have substantial influence on social protection priorities and the actual implementation of programmes. Donors tend to favour new initiatives rather than build on existing policies (Niño-Zarazúa et al., 2010); government-run social schemes in Zambia are underfunded and neglected. At the same time, donor-funded pilots tend not to take into account local politics and in fact hamper the potential for building a social contract between the government and its citizens. The Zambian government has been reluctant to scale up the projects to the national level (Devereux & White, 2010), for instance, and citizen influence on social protection priorities seems negligible. It is likely that beneficiaries of the pilot schemes would advocate for project continuity, but for the Zambian government it is easy to pass the responsibility to donors. The government itself does not appear to see the political benefit of extending donor projects and including them in government policy, thereby assuming the funding responsibility. As it stands, the future of social protection transfers in Zambia remains unclear (Devereux, 2010).

Despite being almost as poor as Zambia, the government of Lesotho has introduced its own non-contributory old-age pension scheme. The pensions are funded not by donors – who have remained sceptical about the introduction of this policy – but rather by domestically controlled resources (Devereux & White, 2010), which for the most part comprise revenue from SACU. In contrast to the cash transfer pilot projects in Zambia that are limited in scale, time-bound and discretionary, social pensions in Lesotho are rights-based, legally enforceable and permanent (Devereux & White, 2010).

From a social contract perspective, the citizens of Lesotho are not direct contributors to social protection, and it cannot be argued that the social pension was introduced as part of a contract where citizens receive a benefit in return for tax contributions. Instead, as in Swaziland, the decision to commence the social pension came from the political leadership (Devereux & White, 2010). Interestingly, however, and in line with my theoretical discussion, once introduced the pension scheme became a vehicle for social mobilisation. The pension was an important discussion point during the 2007 elections, and many voters revealed, in a post-election survey, that their electoral choice was based on parties' commitment to the pension (Devereux & White, 2010; Niño-Zarazúa et al., 2010). Compared to aid dependent countries, governments in countries reliant on SACU revenue, such as Lesotho, are more able to prioritise social spending. Pressure to expand social protection does not necessarily come from citizens, whose extent of influence and mobilisation relates to the continued existence of benefits.

4.2 Natural resources compared with domestic revenue in middle income countries

In many ways, the politics of social protection in Botswana are similar to those in Lesotho and not, as might be expected, to those in countries that are on a similar developmental and economic level, such as South Africa. Despite the extent of national wealth in Botswana, the overall spending on social protection is small. Policy making in Botswana has been dominated by the top political leadership. Over time, the government has introduced various means-tested social protection programmes targeted only at the very poor and vulnerable. In 1996, the government introduced a universal non-contributory old-age pension. As was the case in Lesotho, this pension was a government initiative introduced without much prior debate at a time when opposition parties were gaining ground in the rural areas that used to be the stronghold of the ruling party (Ulriksen, 2010).

As is the case in Lesotho, only a small proportion of the population actually pay any direct income tax in Botswana, so the social contract is limited. The divided and weak opposition in Botswana has since the 1980s advocated for the introduction of a comprehensive social protection scheme, but this has not been transformed into any substantial citizen demand. The extensive revenue from natural resources has given the Batswana government substantial freedom to prioritise spending, and the government has ensured continued legitimacy through social service spending. It is likely that the ruling party regained some popularity in the mid-1990s by introducing the pension scheme, though a clearer victory in the following elections can also be attributed to a split in the opposition party alliance (Ulriksen, 2010). Although civic mobilisation in Botswana is limited, a political attempt to reduce or means-test the pension could cause dissatisfaction and even protest.

Compared to Botswana, and indeed the southern African region as a whole, South Africa has – like Mauritius – quite an extensive social protection system with a long history. For instance, the old age pension was introduced in 1928 and was amended in 1944 to provide pensions for all South Africans of age 65 or older (subject to means-testing). However, rates differed according to race as it was argued that ‘natives should receive lower benefits because they paid lower taxes and had a lower standard of living’ (Devereux, 2007:542–3). In fact, there appear to have been strong links between revenue source and priorities of spending during the apartheid era. Modelled on European state welfare policies (Patel, 2005), South African social protection was originally introduced to prevent whites from falling into abject poverty. From a financing perspective, the strategy was to ensure that tax collected from whites also benefited this group. In the same way, spending in the so-called homelands relied on taxes collected by local governments in these areas – more than 90% of local government revenue in the homelands came from their own sources (Mkandawire, 2010).

The post-apartheid government has endeavoured to equalise rates of social protection so not to discriminate on the basis of race and yet still to redistribute across income groups. As illustrated by the old age pension, social protection is tax-funded and reaches almost exclusively the poor black majority (Devereux, 2007; Niño-Zarazúa et al., 2010). Like Namibia, South Africa saw a link between revenue source and spending priorities in the past, yet today there is a separation between those contributing to social protection and those benefiting from it. There may still be a social contract in South Africa in the sense that contributors accept generous social protection spending as a means to rectify past inequities (and this being the case, taxpayers may still feel that they get something in return for their contributions). However, in the long run, taxpayers may become loath to continued contributing if they are uncertain about the returns. In a recent report, the National Planning Commission supports this argument by arguing:

South Africa has had an implicit compact in which the wealthy pay taxes and the government uses these taxes to deliver services and effect redistribution. This compact will be at risk if people believe that the tax revenues that they contribute are being spent inefficiently. (NPC, 2011:15)

In summary, it is evident from this brief comparison of four southern African countries that the revenue-expenditure nexus affects the extent of a social contract. In Lesotho and Botswana the governments are relatively free to introduce social protection policies as they deem fit, though once in place policies are more difficult to change. In Zambia, the priorities of the donor community prevail. South Africa comes the closest to having a social contract.

5. Conclusion

While concerns have been raised about the fiscal sustainability of the social pension, its future as a social welfare programme in southern Africa and elsewhere depends mainly on political commitment from governments and taxpayers. (Devereux, 2007:539)

It is often acknowledged that political commitment is critical to ensure sustainable expansion of social protection policies. The role of those providing the fiscal base for spending – taxpayers for instance – is not acknowledged as often. In fact, the important links between the politics of spending and the politics of financing have received insufficient attention.

This is one of the first articles to examine the issues of revenue and expenditure for social protection in an integrated fashion. The above examination of southern African countries offers evidence that revenue source affects the priorities for social protection spending. Furthermore, it is clear that, when the concept of a social contract is based on an exchange-based logic, only a few countries in the region have something resembling a social contract for social protection. The main distinction should be drawn between tax-based revenue and non-tax-based revenue (also called rents). One can only talk about a reciprocal social contract in countries that mainly rely on domestic taxation; in these cases citizens receive benefits from government in return for contributions. In countries reliant on rents – whether aid, natural resources or SACU revenue – the citizens have much less leverage vis-à-vis the government.

The different types of rents affect the priority of spending in distinct ways. In the aid dependent countries of Malawi and Zambia where large parts of the revenue are externally controlled, social protection is divided between underfunded national programmes and donor funded pilot projects. As a recipient government is mostly oriented towards donor requirements, the social contract between the government and its citizens is limited. In Lesotho, Swaziland, Botswana and Angola, social protection spending is largely dependent on elite priorities. SACU revenue is a peculiarity of the region, but in many ways it has characteristics similar to natural resource revenue. These types of domestically controlled rents are not a direct contribution from citizens or donors. In this sense, the government is not bound to commitments made in an exchange agreement (where certain spending is promised in return for contributions), and it has more freedom to prioritise its spending. If there is a social contract in these countries, it is one-sided – citizens do not engage with the state in a relationship based on reciprocity but mostly as beneficiaries of social protection.

Compared to the countries discussed above, most of which rely on different types of rents, South Africa and Mauritius, where the domestic revenue source is also most significant, have more comprehensive social protection spending. Namibia is something of a mixed case as its revenue base has less dominant features (both rents and taxation are prominent).

In conclusion, a broad and diversified tax base is an important mechanism for creating a reciprocal relationship between citizens and the state that may build social contracts for increased social protection spending.

Notes

²SACU is a revenue source specific to the southern African region, as is discussed further in Section 3.

³Even allowing for some uncertainty in calculations due to exchange rates and annual changes to pension rates, Botswana's pension rate is closer to those of Lesotho and Swaziland. Monthly pensions in US$ are as follows: Botswana: 25, Lesotho: 28, Namibia: 42, Mauritius: 90, South Africa: 141 (maximum), and Swaziland: 14 (author's calculations).

⁴South Africa has an unemployment insurance fund and Mauritius has recently introduced a workfare programme providing social security and training for the unemployed, but these two programmes are social insurance schemes and not government-funded social protection programmes, which are the focus here.