How social security policies and economic transformation affect poverty and inequality: Lessons for South Africa

Abstract

This article examines how various characteristics of social and economic policy frameworks affect poverty and inequality levels in developing countries, principally in Botswana and Mauritius. The research findings suggest that poverty and inequality are lower in countries with generous and broad-based – rather than pro-poor – social security policies, and where social policies are complemented by economic policies promoting economic transformation rather than mere economic growth. While South Africa's challenges of combating poverty and inequality are shaped by its own historical context, the lessons from other countries offer the opportunity to reflect on the social consequences of various social and economic policy mixtures. In particular, it may be worth considering how to bridge the divide between the economically productive contributors to social security policies and the economically marginalised beneficiaries of such policies.

Keywords:

• social protection
• economic modernisation
• poverty reduction
• inequality

1. Introduction

It has been observed that the World Bank, the IMF and much of the aid industry, using the UN's Millennium Development Goals approach, have focused only on the poorest of the poor, and erroneously assumed that the problem of poverty can be solved without dealing with related problems of inequity, social exclusion and the state's failure to support developmental projects that will benefit all social groups (Deacon, 2010).

For decades the problem of how best to combat poverty and reduce structural inequality in Africa has been of major concern to international aid agencies, local governments, academia and civil society alike. Remarkably, most of these actors now appear to take the near-hegemonic position that social protection is the best way to address issues of poverty (Adésínà, 2010). The World Bank and IMF acknowledge that markets are not perfect and that social safety nets for the most vulnerable are necessary to avoid human disasters (World Bank, 2001). The UN, along with many other donor agencies, promotes the ‘social protection floor’ advocating that minimum cash benefits to the poor as a useful strategy for poverty reduction (Hickey et al., 2008; ILO & WHO, 2009).

Worthy as these initiatives are, however, there may be a concern that this sole focus on the poor fails to address structural inequalities and instead increases the gap between the poor and the non-poor. Moreover, despite claims to the contrary, the promotion of social protection is still merely secondary to the economic growth agenda. Questions of how social and economic policies could together strengthen developmental efforts towards broad-based economic transformation deserve more scrutiny than they currently receive.

In examining how contrasting mixtures of social and economic policies have affected poverty and inequality levels in developing countries, this paper reveals lessons for South Africa. The initial theoretical discussion considers how policies of social welfare and economic development can ensure that the poor become part of the developmental effort rather than being isolated. The empirical section analyses the effects of social protection policies and economic modernisation (as an alternative to economic growth) on poverty levels in developing countries. The findings of this statistical analysis are complemented by examinations of Botswana and Mauritius, looking at how different combinations of social and economic strategies affect poverty and inequality. Lessons are then drawn for South Africa.

2. Social protection and economic transformation: The theoretical debate

The dominant approach to economic and human development since the mid-1990s has been the pro-poor growth strategy promoted by the World Bank. The main aim of this strategy is to achieve sustained and rapid economic growth benefiting the poor through sound economic policies, political stability and investments in physical and human capital. Social safety nets supplement this strategy by supporting poor and vulnerable groups unable to benefit from the economic growth (Besley & Cord, 2007).

Critics argue that the safety nets component promoted by the World Bank and the IMF is a very narrow definition of social protection. Moreover, as there is a tendency to focus mainly on food-for-work and cash-for-work programmes, many of these safety nets fail to include destitute people who are unable to work (Ellis et al., 2009). Relief work programmes appear to be a favourable strategy, though cash and food transfers for non-working vulnerable groups (the elderly, children and the disabled) are also acknowledged as important elements of poverty relief (Ravallion, 2008). The World Bank is therefore coming closer to the idea of social protection promoted by the UN and others. Social protection in a broader sense may be described as ‘all public and private initiatives that provide income or consumption transfers to all poor, protect the vulnerable against livelihood risks, and enhance the social status and rights of the marginalised’ (Ellis et al., 2009:8). Social protection is therefore a variety of targeted transfers (cash, food, inputs or assets) aimed at empowering the poor, vulnerable, and marginalised and fulfilling their basic needs.

While social services such as promotion of education and health are acceptable parts of the pro-poor agenda, social safety nets are still regarded as secondary to economic goals and it is generally suggested that they be kept to a minimum (Mkandawire, 2004). Proponents of the broader understanding of social protection, on the other hand, often fail to integrate the role of economic policies. They suggest that social protection assists in providing a skilled and healthy work force and is important in creating opportunities for the poor. Nevertheless, they seem to accept the dominant growth strategies (Norton et al., 2002; Adésínà, 2010) and then concentrate on how best to develop, target and implement social protection policies (Ellis et al., 2009). There has thus been a tendency to focus on economic growth and, as an add-on, social protection for the poorest of the poor, while social protection (here termed ‘social security policies’) for the broader population and economic transformation to create meaningful employment and social equity have been neglected.

It seems obvious that social policies aimed at reducing poverty should focus on the poor. However, such strategies not only go against the historical evidence of Western welfare states (Deacon, 2010), they also carry the risk of creating a gap between the poor and the non-poor that may be extremely difficult to close, with detrimental consequences for poverty and inequality in the long term. In the following I elaborate on this possibly contentious argument by considering first the role of social security policies in reducing poverty and inequality and second how social outcomes may be improved when a mix of social and economic policies fosters structural economic transformation.

Various social security policies ensuring social protection, such as social cash transfers, insurance schemes or benefits in kind, may be defined in two ways: how generous they are and which social groups they cover (Bonoli, 1997). We know from research on Western welfare states that countries with a mix of social security policies covering broadly across income groups tend to have more generous social policies and consequently higher levels of social well-being – or, in reverse, that ‘the more we target benefits at the poor only … the less likely we are to reduce poverty and inequality’ (Korpi & Palme, 1998:681).

This may seem counter-intuitive. However, the reason for differences in the character of social security policies across countries lies in the interests and commitments of the non-poor, particularly the middle-class (Esping-Andersen, 1990). This heterogeneous group, lying between the rich and the poor, plays an essential role in promoting economic development and good governance as its members are generally resourceful and largely politically independent of the state – unlike the rich in many developing countries (Birdsall, 2010). They also have interests in adequately-funded social services and social protection (Birdsall, 2010), though they can ensure their own social welfare through both private and public institutions. According to this argument, countries where social security policies cater solely for the poor do not succeed in reducing poverty and inequality as the non-poor do not benefit from such social policies and are therefore less willing to help cover the costs (UNRISD, 2010). Hence, political feedback often causes social security to become more targeted, efficient and parsimonious. Conversely, when the non-poor also benefit, social policies tend to be more generous, causing a more effective redistribution (Gelbach & Pritchett, 2000). Policies that target only the poor risk isolating this group from the mainstream economy. The non-poor may have charitable concerns for the destitute and may offer some minimal social assistance, but little else pushes them to pursue the expansion of social policies from which they are unlikely to benefit.

This paper does not contend that Western countries can provide developing countries with blueprints for social protection. However, it is surprising, and potentially misleading, that international agencies have made suggestions contrary to historical evidence (UNRISD, 2010). Poverty, and more so inequality, has most efficiently been reduced in countries where social welfare has been a national project that cuts across income groups rather than focusing on the poor alone, and where social and economic policies are aimed at transforming the economy and creating employment.

The argument in favour of social protection is that transfers will help the poor to increase their capabilities, i.e. empower them, so they can also benefit from economic opportunities. Research shows that cash transfers improve the livelihoods of the poor, enable them to engage in less hazardous behaviours, and enhance their skills (Samson, 2009). Arguably also, such transfers support, rather than hamper, economic growth (Cichon & Scholz, 2009). Yet important assumptions underlie this strategy. For instance, World Bank researcher Martin Ravallion argues that relief work programmes to which the needy can sign up, especially during periods of crisis, to work for a ‘not too high’ wage rate constitute important safety nets. The advantage is that ‘when the crisis is over, the safety net will no longer be needed for the majority of workers and … they will automatically return to regular work’ (Ravallion 2008:19). However, for that to happen there must of course be jobs available. While relief work programmes set a condition on receiving benefits, cash transfers – even if for non-working groups such as children or pensioners – have the advantage of freeing the direct or indirect beneficiary to seek jobs and improve skills. Once again, this is only useful if there are opportunities to engage meaningfully in the economy.

Undoubtedly one of the best ways out of poverty is to get decent employment – to be a productive part of the economy and thereby contribute directly to economic development, and indirectly, for instance through taxation, to social policies from which one may benefit in times of need. There is probably a consensus on this. However, whether the state can – or should – create employment opportunities, and how, is highly controversial. And there are no easy solutions. One approach is to consider whether the poor become integrated parts of the economy rather than remain isolated from it. For instance, to promote relief work or subsistence agriculture does not necessarily make beneficiaries a contributory part of the economy. Indeed, they are likely to remain in a position of economic insecurity, especially since their income, when they have any, is usually minimal. It has been suggested that labour-intensive strategies that tap into the abundant unskilled labour in developing countries is an efficient way to increase growth and reduce inequality (UNRISD, 2010). In addition, while social security policies are mechanisms of redistribution, poverty and inequality can also be addressed through greater equality of market incomes, i.e. earnings before taxes and benefits (Wilkinson & Pickett, 2009:183–4).

To recapitulate: it is laudable to promote cash transfers to the poor, but if such policies are not complemented by strategies of economic transformation and employment creation then it is hard to see how such policies can be sustained and structural inequalities overcome. Cash transfers may improve the well-being of the poor. However, beneficiaries of means-tested policies will still by definition be poor, and distinct from the salaried and better-off population, unless the composition of economic growth is such that it offers decent employment opportunities. This is not to downplay the importance of economic growth – that is a necessary condition – but perhaps (difficult) changes to the structures of a country's economy are just as important? And perhaps social security needs from the beginning to be a national project – ensuring that the majority benefit, but also contribute? The following empirical analyses attempt to tackle these questions.

3. The effects of social policies and economic modernisation on poverty levels

This section offers a statistical analysis which first tests the relation between social policies and poverty and second examines the extent to which economic modernisation, as an alternative to economic growth, affects poverty. The statistical analysis is rudimentary in that it focuses merely on causes of poverty and cannot suggest how social and economic strategies may be linked, or give any indication of the forms that economic transformation may take. The case studies of Botswana and Mauritius provide an opportunity to elaborate on the links between social security policies and economic transformation and the effects on poverty and inequality.

Poverty is generally considered a multi-dimensional concept. That is, being poor is not only about lack of income but also about being short of other basic commodities such as food, water, knowledge, and health facilities (Barrientos, 2010). Of the poverty statistics available, the Human Poverty Index (HPI) best captures this broader understanding of poverty. This index measures the proportion of people below a threshold level in basic dimensions of human development – i.e. health, knowledge, and standard of living. More details about the poverty variable and other variables can be found in Table A1 in Appendix A. The data for the identified variables cover developing countries identified as low- and middle-income countries. The analyses presented in the following provide a set of statistical regressions, conducted in SPSS, which test the robustness of the effects of social security policies and economic factors on poverty. In essence, the test shows whether these variables explain some of the variance in poverty levels across countries, even when other, competing, explanations of poverty are taken into account.

Theoretically we assume that social security policies that are both generous and have broad coverage best reduce poverty. To test this hypothesis we first correlate social policy expenditures (as a percentage of GDP) with HPI. As Table 1 shows, there is a strong correlation between social policy expenditures and poverty such that higher expenditures lead to lower poverty rates. This is consistent with findings by Barrientos (2010), who tests the relations between social policy expenditures and measures for income poverty. However, the adjusted R square in Table 1 is higher than Barrientos's findings, which indicates that social policy expenditures account for greater variation when one uses a broader measure for poverty.

Table 1: Social policies and poverty

Dependent variable	Human Poverty Index (HPI)
Independent variable	Social expenditure	Social coverage	Social security policies
Coefficient (standard error)	–0.865 (0.173)	–0.784 (0.117)	–0.586 (0.083)
Adj. R square	0.300	0.340	0.477
Number of cases	57	87	54

Next we test the relation between social policy coverage (which measures the extent of coverage across income groups) and poverty. Again, this relation is highly significant and in the right direction, which means that the more social policies cover across various income groups the lower the poverty levels. Most importantly, however, when we combine social expenditures and social coverage (into social security policies) the impact is strongest. Thus, for social policies to affect poverty levels positively, it is important not only to increase expenditure levels but also to ensure that social policies have broad coverage.

To ensure that the relationship between social policies and poverty is robust, a set of control variables are included, and these cover potential institutional and socioeconomic causes of poverty. As indicated in Models 2 and 4 in Table 2, the effects of social policies stay significant and add to the explanatory power of the models.

Table 2: The effect of social policies and control variables on poverty (HPI)

	Model 1	Model 2	Model 3	Model 4
Political institutions
Government effectiveness	3.449 (2.960)	3.382 (3.172)	–3.404* (1.857)	–2.372 (1.956)
Democracy	–0.061 (0.200)	–0.021 (0.212)	0.142 (0.171)	0.119 (0.182)
Socioeconomic factors
HIV prevalence	0.727*** (0.210)	0.529* (0.302)	0.413 (0.255)	0.343 (0.246)
Unemployment	–0.114 (0.209)	0.409 (0.306)	0.215 (0.208)	0.541* (0.274)
Economic conditions
GDP per capita	–0.003*** (0.000)	–0.002*** (0.001)
Economic modernisation			–0.377*** (0.051)	–0.323*** (0.069)
Social policies
Social security policies		–0.203** (0.082)		–0.151** (0.066)
Constant	37.759*** (3.607)	35.490*** (4.341)	39.699*** (3.257)	39.536*** (4.174)
Adj. R square	0.557	0.629	0.698	0.721
F-test	18.119***	12.892***	26.399***	16.491***
Number of cases	69	43	56	37
Notes: Unstandardised coefficient given, numbers in parentheses are standard errors, all tests are two-tailed; ***p < 0.01, **p < 0.05, *p < 0.10. Results are similar when all models have N = 37.

Table 2 furthermore compares the impacts of economic growth and economic transformation on poverty levels. The theoretical argument is that while economic growth is important, it may be equally critical that development is based on some form of structural transformation of the economy. Here we use a simple proxy for structural transformation, namely economic modernisation, which measures the percentage of the population not employed in agriculture. Economic modernisation is argued to be important in relation to poverty reduction, since expanding manufacturing industries and service sectors gives people the opportunity to partake in formal employment at reasonable salary levels, whereas high levels of occupation in farming may imply unsophisticated agricultural production, informal employment and subsistence farming (UNRISD, 2010).

Models 1 and 2 confirm that economic growth (reflected as GDP per capita) is indeed important for poverty reduction. Models 3 and 4 include economic modernisation instead of GDP per capita, and it becomes apparent that economic modernisation also has a significant effect on poverty levels and in fact improves the explanatory power of the models. This by no means disturbs the established consensus that economic growth is a necessary but not sufficient condition for poverty reduction. However, this finding does indicate that we need to better understand how to ensure structural transformation and increase (decent) employment opportunities, rather than concentrate purely on policies of economic growth. Loayza & Raddatz (2010) also suggest that it is not merely the size of growth but more importantly its composition that matters – the expansion of labour-intensive sectors in particular is deemed critical.

The purpose of this analysis is merely to establish the hypothesised effects of social security policies and economic modernisation, rather than to identify all the relevant causes of poverty. Yet it is interesting that the extent of unemployment has little direct effect on poverty. There may be a number of reasons for this. First, being employed does not necessarily guarantee a decent salary. Second, if there are good social security schemes in place, unemployment need not lead to destitution. Third, it is likely that unemployment is associated with economic modernisation in that structural transformation which expands avenues of employment also increases formal unemployment. This suggests that as an economy modernises it becomes critical to identify measures that tackle unemployment and support the unemployed through their (hopefully short) periods of austerity.

4. Experiences from Botswana and Mauritius

So far we have established that broad-based and generous social security policies as well as economic modernisation reduce poverty rates in developing countries. This section concentrates on the cases of Botswana and Mauritius and demonstrates how contradictory experiences of social policy expansion and economic development affect poverty and inequality.

Indeed, it is striking that, despite many similarities, the levels of poverty and inequality in these two countries are remarkably different. Botswana and Mauritius have both developed into stable capitalist democracies with remarkable records of economic growth and good governance, despite pessimistic prospects at independence. In fact, they are on similar levels of economic wealth. They score alike, and well, on ratings of economic freedom and indexes of democracy, state capacity and transparency. Yet in Botswana more than 30% are poor according to the HPI, compared to just over 10% in Mauritius; hardly anyone in Mauritius lives on less than US$1 a day, in contrast to the 23% in Botswana; and income inequality in Mauritius stands at a Gini coefficient of about 0.38, whereas the figure is well above 0.5 for Botswana (Ulriksen, 2010). In the paragraphs that follow we consider the mix of economic and social strategies in the two countries and how these may be related to the divergent social outcomes. The focus is purely on social security policies, since the role of social services such education and health is less disputed. It is also beyond the scope of this paper to analyse why the two countries have developed vastly different social and economic policy frameworks.

4.1 Economic growth and residual social policies in Botswana

When Botswana achieved its independence in 1966 it was underdeveloped and entirely dependent on beef exports. Fortunes changed in the 1970s when a lucrative mineral export sector was established. While mineral resources carried high and continuous economic growth and enabled the government to expand infrastructure and social services, social security policies have been a residual part of the developmental strategies – generally reaching the poorest and most vulnerable.

The overall government strategy has been to ensure macro-economic stability and create an environment conducive to the development of the private sector, while leaving it to market forces to develop industry and business for more diversified economic growth (Masire, 2006). In the words of the IMF, ‘Botswana has followed textbook macroeconomic advice on attracting investment inflows’ (IMF, 2007:30). Employers have few contractual obligations towards employees, there are (very low) minimum wages and limited social security requirements, which may increase costs, and labour regulations are fairly restrictive towards the organisation and mobilisation of the salaried class.

Despite impressive growth records, however, economic growth in Botswana is without equity and broad-based transformation (Samatar, 1999; Hillbom, 2008). There have been no structural changes in the patterns of production which would otherwise drive economic transformation. Instead, outside the diamond sector, the economic sectors are characterised by limited technological innovation and low productivity. Economic exports are overwhelmingly dominated by the labour-extensive diamond sector, while industrial exports contribute less than 10% of total exports (Hillbom, 2008).

The agricultural sector most clearly reveals a society with little structural transformation. Agricultural exports account for a mere 2% of total exports; even so, however, roughly 21% of the labour force is still employed in this sector (CSO, 2004:125). The livestock sector dominates agricultural production, but there has been little progress in technological innovation or value-added activities. One significant development in the rural sector has been the increasing exclusion of many rural families from agricultural production. Over the decades, crop production for food has decreased, meaning that many families are no longer self-sufficient. Instead, the rural poor seek to obtain casual labour at stipulated salary levels even lower than the minimum wages for other income groups (Wikan, 2004; RoB, 2008).

Even though formal sector employment has increased over the years, only just above half of all economically active persons are in paid employment, with formal unemployment levels persisting above 20% (CSO, 2004:41). Botswana has therefore been unsuccessful in ensuring economic transformation that offers meaningful and broad-based employment opportunities. In fact, many people in rural and urban areas alike subsist on considerably low salary levels – a monthly full-time minimum wage is about P500–600 (about US$75–90, April 2008), while the food basket in most urban areas is indicated to be well over P2000 (FES, 2008:14; RoB, 2008).

Generally, there is little social security for many Batswana. There is no regulatory framework to ensure compensation at times of unemployment. Sickness and maternity leave and work injury regulations are minimal and likely to depend largely on one's position and place of employment (SSA, 2009:38–9). Needless to say, anyone in informal or temporary employment has few rights when unable to work.

A number of social assistance programmes exist. Some, including public works programmes and the destitute policy, specifically target the needy. Others, such as pension and school feeding schemes, have a broader reach (Seleka et al., 2007). Nonetheless, social transfers are of little relevance to most Batswana. The old age pension is insufficient to cover monthly costs, which means that the elderly must find additional resources elsewhere in order to ensure reasonable income security (Wikan, 2004). The destitute policy only reaches a small segment of the population, with 38 074 registered destitute in 2005 (Ntseane & Solo, 2007:89–91), and no more than about 2% of the population receive destitute allowances. The public works programmes tend to have a broader reach but are temporary and, as with the feeding programmes, mainly ensure that people stay alive.

4.2 Purposeful economic transformation and social protection in Mauritius

At independence in 1968 Mauritius was, like Botswana, an extreme example of a mono-crop economy, with high reliance on sugar exports and few other developed economic sectors. Yet the government pursued economic transformation to create employment and gradually expand social security programmes.

From early on, the government followed a two-pronged industrial strategy. On the one hand, the government supported import-substituting industries (ISI) which catered for the small domestic market. On the other, it endorsed export-processing industries for the foreign markets. The export-processing zone (EPZ) promoted low-wage industries with few labour law requirements and provided preferential access to EU and US markets. Government employees, ISI workers, and agricultural labourers (all male dominated) enjoyed higher wages and benefits, whereas the EPZ primarily employed women, whose earnings supplemented family incomes. This strategy promoted employment creation; however, it also reflected gender inequities in the labour market (Wellisz & Saw, 1993:232–4; Bräutigam, 1999:147–52).

Despite economic difficulties in the 1980s and adherence to structural adjustment programmes, the Mauritian Government was successful in following its full employment strategies without cutting social security. Trade liberalisation was gradual, which allowed local industries to adjust, and although wage policies were geared towards enhancing export competitiveness, the lowest wage earners received bigger wage increases than the higher income groups, thereby decreasing wage differences between the two groups. The EPZ industry and the tourism sector grew rapidly, employment increased dramatically and, following the near exhaustion of labour reserves in the early 1990s, wage levels rose (Bowman, 1991:120–1; Wellisz & Saw, 1993:248–50; Anker et al., 2001:19, 29).

A consequence of the economic transformation was that income inequality declined during the 1980s from a Gini coefficient of 0.5 to about 0.37, where it stayed up to the 2000s (Bundoo, 2006). It is also suggestive of the different economic strategies in our two cases that Mauritius has a higher degree of income equity across economic sectors than Botswana – the difference between the highest and lowest income group is twice as large in Botswana as in Mauritius (Table A2 in Appendix A).

During the 2000s Mauritius lost guaranteed prices and market access for sugar and textiles and saw an increase in unemployment; consequently the government has pursued economic diversification towards ICT and offshore business services in order to ensure continued economic growth and employment creation (Bundoo, 2006). The recently introduced Employment Rights Act is suggestive of how the government seeks to balance social and economic strategies. The Act aims to improve worker flexibility by limiting the employees' rights while at the same time assisting in skills retraining and providing unemployment benefits to ensure social security in case of job loss (GoM, 2008).

In fact, social security policies have been an important element of the developmental strategies in Mauritius. Overall, social transfers constitute the second largest source of income, with transfer income amounting to, on average, about 13% of total gross income (CSO, 2007). Unlike Botswana, which spends between 2 and 3% of the government budget on social security, in Mauritius social security policies sit at about 20% of government expenditure (Ulriksen, 2010:175).

As early as the 1970s the Government of Mauritius had established an encompassing and generous pension scheme ensuring a large degree of income security across the board. Whereas Botswana also has a universal, non-contributory pension (in place since 1996), the provision of pension is generally more substantial and far-reaching in Mauritius. Not only is the basic pension in Mauritius larger (about double Botswana's), it is also paid out at age 60, compared to 65 in Botswana. Furthermore, pensions in Mauritius are also available to the disabled, and widows and orphans, but are purely for the elderly in Botswana. Finally, the pension scheme in Mauritius is complemented by a well-developed contributory scheme, which tops up pension for many retired people (Willmore, 2006; Seleka et al., 2007).

During the first decades after independence, sugar tax revenues, in the main, funded the expansion of social transfers and social services. Taxation was generally progressive, though few were exempt from paying tax (Wellisz & Saw, 1993:239; Bräutigam, 2008:154–5). The broad-based contribution to social security policies is quite distinct from the case in Botswana, where mineral incomes have generally funded the limited social safety nets.

5. Comparing Botswana and Mauritius, and lessons for South Africa

A comparison of Botswana and Mauritius supports and further substantiates the findings of the statistical analysis. The divergent social outcomes in the two countries can largely be explained by the extent of social security policies and how well the countries' growth ventures have been complemented by economic transformation. In Mauritius, development strategies have not been about the poor, but rather about creating economic transformation, employment and social security for the whole population. In Botswana, economic growth has been driving developmental strategies; poor people struggle to become part of a productive economy and the most vulnerable have come to rely on meagre social safety nets.

While this paper suggests that broad-based social policies and economic transformation decrease poverty and inequality; it is still a puzzle how to advance economic transformation, and thus increase employment opportunities, without hampering economic growth. An equally critical question is how to ensure public commitment to social policy expansion and financing. Theoretically, this paper has suggested that political feedbacks frame policy development. In Mauritius social and economic policies have been forged by compromises of high-, middle- and low-income groups, such that everyone benefits but also contributes. In fact, the alliance between low-income earners and the middle class was essential in building a coalition for broad-based social protection (Ulriksen, 2012). In addition, while the generous social policies in Mauritius improve well-being, creating meaningful employment and keeping wage differences fairly small has also helped reduce poverty and inequality. Conversely, in Botswana, lack of economic transformation has caused a shortage of decent employment. There is arguably a split in Botswana between those who benefit from formal employment and have private social insurance schemes and the rest who struggle in low-paid jobs where the most vulnerable receive minimal social protection. The better-off and politically influential groups benefit little from social welfare, hence they may be unwilling to bear the costs of potential social policy expansion (Ulriksen, 2010).

From an economic strategy perspective, South Africa appears more like Botswana than Mauritius, as its political objectives have generally been to achieve economic growth, and as there is a split between the modern economy and the largely rural areas where economically meaningful livelihoods are few. From a social policy perspective, South Africa may be closer to Mauritius, as these two countries have the most expansive social security systems in the region (Triegaardt & Kaseke, 2010). South Africa has quite extensive social protection policies, with close to 14.5 million grant beneficiaries in 2010 out of a population of about 49 million (SASSA, 2010). However, unlike the case of Mauritius, social security policies are generally targeted and therefore exclude not only the middle class but surely also many with lower incomes. Moreover, with little more than five million taxpayers (National Treasury, 2008), South Africa's resource base for financing social policies is fairly small. This suggests there is another split – between the beneficiaries of and the contributors to social policies. It is equally noticeable that while poverty has decreased in South Africa since the end of apartheid, inequality has not been reduced. Hence, it appears that the well-to-do have benefited from economic growth strategies while the poorest have benefited from social grants. In between is a large population who have not benefited much from either (Bhorat & Westhuizen, 2010).

Keeping in mind that structural inequality built up over three centuries is not easily overcome, South Africa's challenges of poverty and inequality may be confronted by asking how to bridge these social and economic divides, and asking not just how the poor can benefit, but rather how all income groups that suffer some, or occasional, income insecurity can benefit as well as contribute. If policy-making is regarded as an outcome of political bargaining, it is worth considering how social and economic policies can become part of negotiated strategies for sustainable transformation. For instance, a broadening of the tax base to include middle and lower income groups currently exempt from income taxation could offer an opportunity to introduce social security policies benefiting these same groups. Equally, if labour flexibility is necessary for businesses to improve market adaptability and employment creation, businesses may offer better social security and/or decent salary levels in order to get labour unions to accept increased flexibility in employment terms. South Africa must find its own solutions, but experiences from other developing countries certainly offer the opportunity to reflect on the social consequences of different social and economic policy frameworks.

6. Conclusion

In its attempts to reduce poverty and inequality South Africa has tended to focus on economic growth and, as a secondary strategy, social security policies targeted at the poor and vulnerable who do not benefit from such growth. Through a brief theoretical discussion, a cross-case statistical analysis, and a comparative case study of Botswana and Mauritius, this article contends that social security policies need to play a larger role, and one that is complementary to economic policies. To effectively reduce poverty and inequality, economic strategies need to support economic transformation and decent employment creation and be supplemented by generous and broad-based social security policies.

The challenges of combating poverty and inequality in South Africa are daunting, but there are lessons to be learned from elsewhere. Specifically, it may be worthwhile considering how to bridge the divides between those who benefit from social security policies, those who contribute to social spending through direct taxation and the group in the middle who, it seems, are neither beneficiaries of nor contributors to social security policies, and may also be outside the formal labour market. If such bridges could be built, social coverage as well as its financing might be extended to cover the population more broadly, just as economic strategies might be made to focus more on how to create decent employment in a business-friendly environment.

Acknowledgements

The author would like to thank Isabela Mares for making available her database on social protection coverage.

Appendix

Table A1: Explanation of variables for statistical analysis

Variable	Explanation	Year	Source
Countries	All developing countries (middle and low income) with HPI value (except Palestine)	2007	World Bank (2007)
Government effectiveness	Measures quality of public services, quality of the civil service and degree of its independence from political pressure, quality of policy formulation and implementation, credibility of government's commitment to such policies. Scores from –2.5 to 2.5, where higher scores = better government effectiveness	2004	Kaufman et al. (2007:3, Table C3)
Democracy	Polity IV scheme consisting of six component measures recording key qualities of executive recruitment, constraints on executive authority, political competition. Score from –10 to 10	Most recent	Polity IV (2009)
HIV prevalence	HIV prevalence as % of aged 15–49	2006	UNDP (2007: Table 9)
Unemployment	Unemployment rate as % of total labour force	1996–2005	UNDP (2007: Table 21), supplemented by data from World Bank (2008)
GDP per capita	GDP per capita PPP US$	2005	UNDP (2007:Table 1)
Economic modernisation	% of employed work force not working in the agricultural sector, i.e. value turned: 100 minus % employed in agriculture	1996–2005	UNDP (2007: Table 21)
Social expenditures	Social security expenditures as % of GDP; covers expenditures on pensions, health care, employment injury, sickness, family housing and social assistance benefits in cash and in kind, incl. also administrative expenses	1996 (1990)	ILO (2000a:320–1, Table 14), supplemented by data from ILO (2000b)
Social coverage	‘Social policy protection index’ – covers the scope of social insurance coverage (on old-age, sickness, disability and unemployment) and redistribution that these particular social policies undertake across occupations. Higher score = more policies cover across groups	2005	Personal communication, Isabela Mares, Associate Professor, Department of Political Science, Columbia University, New York, 22 May 2008; Mares (2005)
Social security policies	Social expenditures + social coverage, calculated so that values go from 0 to 100
Human Poverty Index (HPI)	Measures proportion of people below a threshold level in basic dimensions of human development – living a long and healthy life (probability at birth of not surviving to age 40), having access to education (% of adults who are illiterate), decent standard of living (% not having sustainable access to water and % of children below age 5 who are underweight). Values from 0 to 100, higher scores = greater incidence of poverty	1995–2005	UNDP (2007:355–7, Table 3)

Table A2: Monthly wages per sector in Botswana and Mauritius, 2005

	Botswana		Mauritius
Economic activity	Pula	US$*	Rupees	US$**
Agriculture, hunting, forestry	697	137	9 824	332
Mining, quarrying	4 518	886	5 588	189
Manufacturing	1 219	239	7 798	263
Electricity, gas, water supply	6 124	1 201	19 457	657
Construction	1 138	223	12 042	407
Wholesale and retail trade	1 725	338	12 772	431
Hotel & restaurants	1 015	199	9 881	334
Transport, storage, communication	3 585	703	15 982	540
Financial intermediation	6 461	1 267	21 478	726
Real estate, renting & business activities	2 294	450	12 822	433
Public administration, defence, compulsory social security***	3 489	684	15 066	509
Education	5 010	982	15 008	507
Health & social work	2 328	456	16 628	562
Other community, social & personal activities	2 629	515	11 427	386
Average	2 141	420	12 067	408
Difference between highest and lowest income groups	5 764	1 130	15 890	537
Source: ILO (2010).
*Average exchange rate for 2005: 5.1 BWP to 1 US$.
**Average exchange rate for 2005: 29.6 MUR to 1 US$.
***For Botswana: Average monthly cash earnings for central government (Bank of Botswana, 2007).