Jump to ContentJump to Main Navigation
The Politics of Inclusive DevelopmentInterrogating the Evidence$

Sam Hickey, Kunal Sen, and Badru Bukenya

Print publication date: 2014

Print ISBN-13: 9780198722564

Published to Oxford Scholarship Online: January 2015

DOI: 10.1093/acprof:oso/9780198722564.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 14 December 2019

Political Factors in the Growth of Social Assistance

Political Factors in the Growth of Social Assistance

(p.146) 6 Political Factors in the Growth of Social Assistance
The Politics of Inclusive Development

Armando Barrientos

Sony Pellissery

Oxford University Press

Abstract and Keywords

This chapter examines the significance of political factors in the rise of social assistance programmes in developing countries in the last decade. It finds this is a two-way process. Politics is crucial to the adoption, design, and implementation of social assistance programmes. They also have a feedback effect on local and national politics. The chapter develops a framework for distinguishing the different dimensions of influence. It applies this framework to study the development of social assistance in India, Brazil, and South Africa. It employs a comparative perspective to identify key approaches, findings, and knowledge gaps in the politics of social assistance.

Keywords:   social assistance, poverty, politics, India, Brazil, South Africa

A large expansion of social assistance programmes has taken place in developing countries in the first decade of the twenty-first century. Large-scale anti-poverty transfer programmes providing direct transfers to households in poverty have spread to a majority of middle-income countries in the South. In low-income countries, the growth of social assistance has been slower and more speculative, due in part to delivery capacity and financial constraints. Impact evaluation studies indicate that, taken as a whole and in combination with economic growth and basic service infrastructure, well-designed anti-poverty programmes can make an important contribution to the reduction of global poverty and vulnerability (Lopez-Calva and Lustig 2010).1 Social assistance has become an important component of poverty reduction and development strategies in the South. There is a fast-growing literature on anti-poverty transfer programmes in developing countries (Fiszbein and Schady 2009; Grosh et al. 2008; Hanlon et al. 2010). To date, the focus of this literature has been on issues of design and impact. It is widely acknowledged that politics does matter for the adoption, design, and implementation of anti-poverty transfer programmes, but this remains a substantially under-researched topic (Hickey 2008b). The main objective of this chapter is to review and assess the scarce literature on the politics of social assistance, with a view to identifying relevant approaches, knowledge, and knowledge gaps.

(p.147) There is a degree of uncertainty around terminology in the context of international development, thus we start with some definitions.2 Social policy includes the provision of basic services—in the main, education and health care, but also water and sanitation in low-income countries—and social protection. Social protection includes three main components: social insurance, social assistance, and labour market interventions. Social insurance covers contributory programmes covering life-course and work-related contingencies. Social assistance comprises tax-financed programmes managed by public agencies and addressing poverty and deprivation. It has become commonplace to distinguish ‘passive’ from ‘active’ labour market policies, with ‘passive’ interventions aimed at securing basic rights in the workplace and ‘active’ interventions enhancing employability.

Our chapter focuses on social assistance or anti-poverty transfer programmes, that is, tax-financed programmes directed by public agencies with the objective of reducing, preventing, and eventually eradicating poverty (Barrientos 2007). Programmes in low-income countries are sometimes financed from international assistance. They are tax-financed but the taxes are collected in a different jurisdiction.

There is huge diversity in the design of social assistance in developed and developing countries. In high-income countries, social assistance has an income maintenance design, providing income transfers aimed at filling in the poverty gap. In developing countries (Barrientos et al. 2010), social assistance includes a variety of programme designs, including pure income transfers, as in non-contributory pensions or child grants and allowances; income transfers combined with asset accumulation and protection, as in human development conditional transfer programmes or guaranteed employment schemes; and integrated anti-poverty programmes covering a range of poverty dimensions and addressing social exclusion. There is also diversity in scale, scope, and institutionalization in social assistance across countries, and across programmes within countries.

This chapter conceptualizes the politics of social assistance in developing countries as a two-way process. On the one hand, social assistance is shaped by political processes. The extension of anti-poverty transfer in the South reflects growing attention to poverty reduction. In many countries in the South, social policy, but particularly social protection and assistance, has risen in importance in political and policy discourse and debate. Lula’s re-election in 2006 is credited by many to the success of Bolsa Familia (Zucco 2008). In India, the re-election of the United Progressive Alliance (UPA) in 2009 is largely credited to the introduction of the National Rural Employment (p.148) Guarantee Act (NREGA) in their first term (Yadav and Palshinkar 2009). The first important set of issues and questions revolve around the influence of politics on the shape of social assistance in the South. On the other hand, social assistance programmes feed back into politics at the national and local levels. Social assistance programmes have the potential to enhance the political participation of groups in poverty at the local level, align electoral support, and change policy priorities and the effectiveness of service delivery.3 At the national level, social assistance programmes have the potential to lock in left-of-centre or populist coalitions, and perhaps generate wider changes in policy. A discussion of these feedback processes would throw light upon potential issues of dependency and political manipulation by elites. A second important set of questions and issues revolve around the significance and orientation of these feedback processes.

The existing literature on the politics of social assistance is scarce and country or programme specific. A full review of the literature will not be attempted here, but relevant studies will be referred to and integrated throughout the chapter. There is a fast-growing literature focusing on the micro-level interaction of social assistance and politics. This literature is country or programme specific (De La O 2006; Giovagnoli 2005; Pellissery 2008; Zucco 2008). There is a scarce literature examining the macro or structural political factors facilitating or restricting social assistance (Hickey 2008b, 2009). The most significant gap in the literature relates to conceptual frameworks capable of explaining the politics of poverty reduction programmes. The existing literature draws from the conceptual frameworks developed to study the politics of redistribution.4 This framework provides interesting insights, but there are few gains in collapsing processes of poverty reduction into processes of redistribution. For high-income countries, Esping-Andersen has provided the most influential framework for examining the production of welfare in advanced capitalist countries,5 but efforts to adapt the framework to the examination of developing countries have not paid sufficient attention to politics or social assistance. Particularly the challenges to develop an appropriate framework arise from the issues of incorporating different forms of democracy (since democracy has seen uneven development in the global South), identity politics that shapes participation (p.149) in political processes and limited accountability that many of the authoritarian states possess. Our attempt is to address these issues which are neglected in the literature on politics of social protection.

Given the vast ground to be covered and the absence of comprehensive approaches to the issue under investigation, it is important to set out the approach and methodology adopted in this chapter at the outset. It is beyond the scope of the chapter to try to cover all programmes, countries, and regions. It is necessary to be extremely selective on these. Our chapter will focus on three countries: Brazil, India, and South Africa. The justification for selecting these three countries is straightforward. They are three large middle-income countries with a rich experience of social assistance innovations. They are leading countries in their respective regions. Together they provide a range of approaches to the extension of social assistance and also demonstrate the diversity of political institutions. The country selection will inevitably influence the discussion in the chapter.6 It can be argued that the three countries selected have conditions which have led to a rapid growth in social assistance, conditions which are hard to replicate elsewhere. The point of the country selection is to learn about the role of politics in social assistance, not necessarily to ensure representativeness of conditions in the South. The analysis in this chapter will provide insights and perspectives valuable to other developing countries, and the framework presented should enable one to assess the strength and weaknesses of the relevant institutions and to identify feedback effects from, and to, political processes in those contexts.

Our methodological approach will be twofold. A comparative study of the politics of social assistance in the three countries selected will help identify inductively key issues and questions. This will be preceded by a discussion of how best to model the interaction of politics and social assistance in developing countries. This discussion will help give shape to the main features of an appropriate deductive framework. A process of triangulation will help achieve the main objectives of the chapter: to identify key approaches, findings, and knowledge gaps.

(p.150) The rest of the chapter is divided into three main sections. The second section introduces and discusses a framework for examining the politics of social assistance. The third section presents the main findings from the three country case studies and draws out the main differences and similarities. The final section goes back to the main research question and discusses whether politics matter for the growth and effectiveness of social assistance in developing countries, and draws out the main conclusions.

Approaches to the Politics of Social Assistance

As noted there is scarce literature on the politics of social assistance. By far the most widely used approach assumes a situation in which politicians compete for support among voters, and voters exercise preferences in line with their interests and advantage. Politics in this context has the function of aggregating voter preferences.7 In conditions of an inclusive franchise and a single-issue policy, the preferences of the median voter signal the direction of policy. Applying this perspective to social assistance would suggest that highly unequal societies with high poverty incidence would opt for more generous forms of social assistance (Alesina 1999). In conditions where low-income groups are a minority, coalitions including the middle classes would be required to support greater expenditure on broad-based social assistance (Gelbach and Pritchett 1995). Social assistance would be weakest in conditions of greater equality and low poverty incidence. These predictions are at odds with the facts of the expansion of social assistance in developing countries, and fail to explain both its timing and scale.

An issue is that several features of political systems in developing countries differ from the assumptions of the median-voter model. Political competition is imperfect in situations where the franchise is restricted, or where patrimonial or identity politics filter the aggregation of preferences in favour of elites. In these contexts, politics reflects more directly the concentration of power and influence as opposed to the calculated preferences of the population. The greater the concentration of wealth and power, the weaker will be the policies addressing poverty. In situations where identity politics dominate, the design and implementation of social assistance is likely to be fragmented and unequal (Keefer and Khemani 2003a). Imperfect political competition gives greater autonomy to politicians representing powerful interests and elites.

(p.151) The role of public agencies charged with implementing social programmes becomes more significant after programmes are legislated for and adopted. The degree of autonomy of civil servants, and the nature of their linkages to political elites, can have implications for the reach and effectiveness of social assistance programmes (Alesina and Tabellini 2004; Besley and Ghatak 2007; Mookherjee 2004). In particular, in countries with federal structures, the links between public agencies and local politicians is of some significance for the evolution and implementation of these programmes (Giovagnoli 2005).

Even this brief discussion highlights the complexity of the political influences over social assistance and the variety of potential approaches to studying these influences. Small variations in political structures can have large effects on the direction and relative weight of the predicted effects.

To address this complexity, the chapter develops a phased approach to this issue. Table 6.1 provides a summary. The framework presented in Table 6.1 should be approached as a means of separating out the different dimensions of social assistance to highlight the political influences specifically relevant to a specific dimension. The most productive way of approaching the framework as an analysis tool is to imagine the different columns as sections of a concertina. In practice it is hard to separate out the different political influences on social assistance, and their variation across countries and across time, but by extending the concertina, it is possible to highlight the political issues of greatest relevance in specific dimensions. Once that work is done, it becomes possible to bring all the influences together.

Table 6.1 An incremental framework for understanding the influence of politics on social assistance in developing countries

Adding complexity incrementally

Direct democracy

Representative democracy

Representative democracy with a measure of government autonomy

Representative democracy with a measure of government autonomy and participation

What does the analysis abstract from?

  • Representatives

  • Government

  • Civil society

  • Government

  • Civil society

Civil society

What political influences are included?

Democratic process

  • Democratic process

  • Representatives

  • Democratic process

  • Representatives

  • Government

  • Democratic process

  • Representatives

  • Government

  • Civil society

Focus points/areas

Preference aggregation

  • Social assistance and electoral support

  • Priority of antipoverty policy

  • (De)centralization (responsibility for poverty reduction)

  • Agency competition/cooperation

  • Horizontal and vertical integration

  • Accountability

  • Public support

  • Interest groups

Key social assistance dimension


Design, incidence, and budget


  • Accountability

  • Dynamics/Evolution over time

Practical issues/findings relevant to social assistance

Principles and values (citizenship/contribution/dessert)

  • Design

  • Reach/incidence

  • Budgets

  • Leading agency

  • Beneficiary selection

  • Transfer level and type

  • Coverage

  • Agency coordination,

  • Monitoring and Evaluation

  • Forms of accountability

  • Participatory processes

Feedback effects

Social pacts -social contracts -access order

  • Party dominance

  • Coalition formation/stability

  • Electoral support

  • Pro-poorness

  • Local political support and power

  • Autonomy of local bureaucracies

  • Sustainability of social assistance

  • Priority of poverty reduction and equity policies in public dialogue Influence of civil society & NGOs

Critical issues

  • Social contract -which can be understood as the point at which direct democratic processes operate

  • Social pacts? Coalitions? Influence of donors in LICs

  • Building and maintaining coalitions

  • Interests and values

  • Basis for electoral advantage

  • Credibility and trust

  • Competitive versus identity politics

  • Opportunities for reform/ scale up

  • Federal/Estate/District politics, respective roles, responsibilities, and gains

  • Strengthening technical and political networks coalitions

  • Public perceptions of effectiveness and sustainability of political support

  • Elections and opportunities for change and reform

  • Closed or open source models as far as lower instances

  • Political sustainability

  • Managing political change (changes in government, etc.)

  • Coalition building Feedback into political processes

Source: Authors

The first two rows define which political factors are included and excluded. The first column assumes direct democracy, voters exercise their preferences directly through the political process. Political representatives, government, and civil society are assumed to follow voter preferences. They have no influence over decision-making. The main function of the political process is to aggregate voter preferences. This simple account of the political process is relevant to the adoption of social assistance strategies and programmes. The clearest examples of direct democracy in shaping social assistance arise in contexts where social contracts are being renewed. The new 1988 Constitution in Brazil, for example, was the outcome of a popular decision at the end of twenty years of dictatorship from 1965 to 1985, to reconsider the basis for citizenship in the country. The Constituent Assembly discussions led to the inclusion of a right to social protection based on citizenship, as opposed to the contributory principle which had dominated the development of social insurance.8 The right to social protection based on a (p.152) (p.153) citizenship principle has become embedded in government policies aimed at securing a minimum guaranteed income for all residents in Brazil (Mesquita et al. 2010). This provided the basis for the expansion of social assistance in Brazil. In fact a procedural definition of a social contract can be described in terms of the intervention of direct democratic processes. In practice, single-event social contracts, like the Brazilian experience, are rare. Social pacts and dominant coalitions might lead to the same policy outcomes, but without the presence of direct democracy.9

The important point about direct democracy and the extension of social assistance is this: where political processes revert to the simple function of aggregating preferences, social assistance institutions are likely to emerge. Of course, social assistance can also emerge in conditions where direct democracy conditions are absent. It would be interesting to consider the implications from this point of departure, both in terms of design and scale.10 This will involve collecting and evaluating single events leading to the expansion of social assistance. In the context of the development of social assistance in low-income developing countries, the role of international partners adds a complication to the aggregation of preferences with important implications for social assistance outcomes in these countries.

The second column in Table 6.1, entitled representative democracy, envisages preference aggregation through political representatives. This ensures a focus on attracting and sustaining electoral support. The priority given to anti-poverty policy then plays within the political processes selecting representatives and endowing their decisions with legitimacy. This level of abstraction is particularly relevant to issues around the design, reach and incidence, and budget-setting for social assistance programmes. The last two are intimately related. In most developing countries, parliamentarians set the budget for anti-poverty programmes with direct implications for the scale of the programmes.

There are specific features of political processes which have particular relevance at this level of abstraction. The need to build and sustain coalitions often influences the scale and location of anti-poverty transfer programmes, as well as the distribution of programme budgets (de Janvry et al. 2009; Giovagnoli 2005). Historically, rural interests and their representatives show a preference for broad-based, tax-financed programmes, while urban interest and their representatives have a preference for occupational (p.154) employment-related social insurance plans.11 A specific issue in this context is how to explain the fact that selective anti-poverty transfer programmes could secure a broad base of political support in some developing countries. Identity politics can help sustain and enlarge social assistance programmes in some countries like South Africa or India, but limit them in other contexts, as in ethnically divided low-income countries in sub-Saharan Africa. The issue of electoral advantage arising from the perceived effectiveness of social assistance is also an interesting issue at this level of abstraction.

The third column takes account of the degree of autonomy of public agencies.12 This level is particularly relevant to issues of implementation and scaling up of anti-poverty transfer programmes. The degree of autonomy of public agencies is essential to understand the operation of social assistance in the context of decentralized political and programme structures. In many low-income countries, process deficits and fiduciary risks are very significant and strongly influence the scale and scope of anti-poverty transfer programmes. In the context of transfers which are combined with asset accumulation or protection or in the case of anti-poverty programmes, horizontal and vertical coordination is central to the effectiveness of the programmes. Monitoring and evaluation processes provide measures of effectiveness and impact.

The features of the political processes which are important at this level include the linkages between different levels of government and the distribution of political influence and power. In highly centralized countries, the responsibility and political gains for social assistance programmes are located at the federal or central government level; whereas in countries with a decentralized administrative and political structure, the distribution of responsibility and potential electoral gains are outcomes from interlinked processes. These imply the need to pay attention to public perceptions of the effectiveness of the programmes to a greater extent than where programmes are heavily centralized. The relative autonomy of public agencies grants a stronger influence on technocratic networks.

The final column incorporates civil society organizations and their political influence. In developing countries, civil society organizations have not had a strong influence on social assistance growth. Trade unions and employers, for example, have not been directly affected by the extension of the programmes.13 Community and NGOs in some countries are mainly involved in ensuring the rights of potential beneficiaries and managing (p.155) processes of accountability of programme managers. Their engagement has some implications for the reform and extension of programmes.

As noted above, the most productive way to apply this framework as an analytical tool is to imagine the different columns as parts of a concertina. Extending the concertina enables an analysis of the different political influences on social assistance, which need to be brought together at a second stage. Two additional points on the framework are important. Firstly, the framework assumes that national politics are dominant in terms of their influence on anti-poverty programmes. This can be contentious because we find below that local politics sometimes overrides national politics in the context of social assistance. In Brazil, we find local-level dynamics more clearly. In India too, state-level primacy for the emergence of social assistance is clear. In India, separating out design and implementation gives local-level politics an overriding importance at the implementation stage. Local-level politics might need to be examined separately and then integrated within the national level. However, our approach to give primacy for national-level politics is important since we are attempting to find a workable model beyond the case countries presented here. Federal institutions have to be treated as context variables that will bring changes to the model presented. International political influences, for example, through the Millennium Development Goals or through regional commitment and policy diffusion, have not been fully incorporated into the model, except through their influence at the national level.14,15 Secondly, the framework does not specifically address any potential political influences over the dynamics of social assistance, except through issues of scale. It will be important in this context to pay special attention to social assistance as a whole as opposed to particular programmes. The political influences on the institutionalization of social assistance programmes are an important research issue.

The revenue side of the framework is in many respects crucial, both to the shape and scale of social assistance, and to the quality of the political linkages. For example, there are good reasons to hypothesize that those governments able to rely on revenues from natural resources might have a greater degree of autonomy from voters in the design of policy than governments reliant on direct taxation. The extent to which revenues are centralized or dispersed in deferral political systems largely defines the relative influence on the different levels of government and politics. Including the revenue (p.156) side would require a significantly longer chapter, but perhaps not a great deal of change to the framework developed above. Whilst highlighting the importance of the revenue side, this is left out of the discussion here.

Comparative Analysis of the Politics of Social Assistance

This section focuses on case studies constructed to map out and analyse the political dimension of social assistance growth in India, Brazil, and South Africa. These are the main findings.


Introduction of social assistance in India can be traced back to British colonial legacy.16 On the one hand formal social security was introduced for employees in the formal sector after the European model. In effect, this was divisive, as the well-off sections employed in regular jobs were able to gain the welfare benefits, and close to 90 per cent of the labour force (primarily in the agriculture sector) was excluded. On the other hand, in its low-growth period until the early 1990s, central government paid little attention to this issue and largely development, particularly rural development, was the focus. These anti-poverty programmes aimed to provide food and nutrition,17 supply basic services like education, health care, and housing, generate employment through public works programmes,18 and improve natural resources and rural people’s assets through Integrated Rural Development Programmes.19 Anti-poverty programmes, as emerging from the nation-building discourse, dominated the politics of social assistance.20

The Constitution of India had left the issue of social assistance as a ‘desirable activity’ under its directive principles. Therefore, federal states (formed (p.157) according to the regional languages after independence in 1947) undertook initiatives to introduce social security measures. The state of Uttar Pradesh introduced the earliest programme of old-age pension in 1957.21 Different states began to introduce different programmes such as pension for agricultural landless labourers, maternity benefit, disability benefit, relief for educated unemployed persons, and employment guarantee depending on the ‘need’ for the same in respective states. Thus, paternalistic and patrimonial principles dominated the origin of these programmes more so than right to welfare or justice principles (Jayal 2001). Very often, these programmes were also introduced as electoral instruments, with the title of the programme prefixed with a politician’s name, which would signal who should be credited for such a programme. These state-level programmes were aimed at workers in the informal sector, primarily agricultural workers on whom the political class relied for votes. Thus, welfare regimes in India could be classified as ‘clientelist’ or ‘populist’. A couple of exceptional states had the influence of left-leaning politics to demand social assistance (Harriss 2004).

In the last two decades, there was a reversal of the story. Central government enacted a number of social assistance measures by systematically expanding the fundamental rights (such as the right to life, the right to employment) enshrined in the constitution of India. From the social assistance viewpoint, three developments are important. First, in 1995 central government introduced the National Social Assistance Programme (NSAP), under which five different benefits were provided, which complemented what federal states were already providing. These benefits were the Old Age Pension Scheme (reaching 8.3 per cent of elderly households), the Widow Pension Scheme (6.2 per cent of widow households), the Disability Pension Scheme (14.1 per cent of disabled households), the Family Benefit Scheme (one-time relief for the families where the main breadwinner died in an accident), and Annapurna (food for elderly households).22 What triggered this development is very closely tied to the story of liberalization that India followed since 1991. Social sector expenditure for the period 1991 to 1995 showed a distinct decline in state expenditure, primarily since the centre’s aggregate transfer to states got reduced (Guhan 1994, 1995; Prabhu and Chatterjee 1993) in the process of state retrieval. This provided an opportunity for bureaucratic–civil society entrepreneurs to argue for direct transfer through initiating social assistance programmes. Introduction of the programme in proximity to the next general election and prefixing of the (p.158) schemes with the name ‘Indira Gandhi’ are facts worth noting. These programmes were meant for poor households. The identification process for poor households took place every five years, which has been much politically contested both locally (Hirway 2003) and between the federal state and central government in determining the threshold of the number of poor people in a state, since that determines the quota of transfer from the centre to the state. The poor households identified through the survey, classified as ‘Below Poverty Line’ (BPL), are eligible for different NSAP programmes. Often, concentration of poverty in particular social categories helped to create a clientelist politics of its own through this targeted approach.

Both the second and third important developments took place in 2004 when the Congress Party government had been voted to power in a coalition with left-wing parties. A wider civil society movement that pressed for food security got a significant policy voice through civil society actors who were appointed as members of the National Advisory Council. This civil society activism was successful in getting the bill enacted by parliament in 2005 for the Employment Guarantee Act that ensures 100 days of employment for every rural household.23 The programme reaches approximately 33 per cent of rural households. Unlike the NSAP of 1995, this was legislated as a right and parliamentary scrutiny was very high. Left parties, being in coalition, also had pressed for social security programmes for the vast majority of unorganized sector workers. Eventually, in 2009 a Social Security Board was legislated. The health insurance programme (Rashtriya Swasthya Bhima Yojana), designed particularly for the workforce in the unorganized sector, has already provided insurance against hospitalization for forty million households.

A key challenge faced when all social assistance programmes were introduced has been the assertion from right-wing advocates that social assistance expenditure is both ineffective and wasteful. The inequality discourse has been effective in countering such positions. The growth story of India has widened inequality rather than bridging the gap. Therefore, the introduction of social assistance would be helpful as an inclusive instrument for the poorer sections of the population. Socio-economic divisions within the country demanded programmes that could create support from across different sections. For instance, the NREGA programme has been devised to provide labour for wage-seeking households during the lean agricultural season, without harming the landlord–labourer relationship. However, when agricultural wages rose, landlords showed resistance to NREGA itself. At the (p.159) same time since the NREGA programme improves community infrastructure such as roads or irrigation facilities this acted as an incentive for landlords. Taxpayers in urban areas were allured by the promise that when rural households were provided with employment in the villages, they would refrain from migration, bringing ease to the already hugely pressured urban infrastructure (Ambasta et al. 2008; MacAuslan 2008).

Lower levels of administration such as district or village government, although elected bodies, enter into the politics of social protection when the programmes are being implemented. In the interpretation of eligibility criteria or selection of beneficiaries, local politics play a critical role (Raabe et al. 2010; Shankar et al. 2011). However, in designing the programme and finance, it is the politics at federal level and central government level which matters. Thus, implementation deficits and corruption in the programme indicate an absence of government autonomy from social forces. Introduction of the social assistance programmes hugely shapes local politics since local elites act as brokers to facilitate access for the target population.24 This ability to satisfy local elites by the national- or state-level elites through the introduction of policies acts as a feedback mechanism.

Acting as a countervailing force to such a nexus between local elites and national elites is the role of NGOs and other civil society institutions. The politics of civil society organization is primarily about how these organizations are able to score a goal against local politicians and bureaucrats. Therefore, being able to provide access to a social assistance programme, which was denied by a politician or a bureaucrat, increases the credibility of an NGO. In the case of NREGA this has come to the fore when social audits conducted by NGOs reached contradictory conclusions to government agencies. In other words, accountability achieved is through the participation of citizens as mediated through civil society organizations.

Accountability at the central government level is diffused due to an absence of single-ministry or autonomous agency in India. There are over 300 different types of anti-poverty schemes spread over thirteen different ministries. There is hardly any coherence amongst these programmes. No attempt is made at integration at national level. Often, NGOs have been successful in achieving integration at local level, since they prioritize the needs of the local population and claim for complementary anti-poverty programmes for one particular locality. This has helped NGOs to remain a sustaining force. On many (p.160) occasions, the government has used the help of NGOs to carry out this role in a concerted manner throughout the state by creating NGO consortiums.25

To sum up, the expansion of social assistance in India through central government initiatives has to be seen as a process of change management due to economic liberalization and pressures from left parties for inclusive growth. A wider social contract is still elusive due to identity politics, and linguistic identities are further reinforced by the federal structure. When the same elected political party is in power at both central and state level, there is synergy for social assistance delivery for brief periods.


Brazil provides one of the most important examples of effective delivery of social assistance in the developing world (Barrientos 2011). Not only has Brazil introduced important innovations in social assistance (Bolsa Escola is the precursor of human development transfer programmes in Latin America) and important technological innovations in programme implementation, like the Single Registry, it has also managed to make a large reduction in poverty over time and also to reduce inequality. More importantly, it has managed this in the context of, until recently, low growth performance. The policy process in Brazil has been intensely political, predominantly in a positive sense. The feedback effects from social assistance are also important, especially in the electoral success of Lula in 2006.

The starting point in tracing the rise of social assistance in Brazil is the 1988 Constitution, which followed a long period of right-wing dictatorship from 1965 to 1985. The Constitution was intended as a new social contract extending citizenship to all. The Constitution enshrined a right to social protection, and led to a rethink of the role and scope of social security and the role of government to provide it. Prior to 1988, the role of government was to support private organizations and NGOs in the provision of social assistance (Jaccoud et al. 2009, 2010). The Constitution recognized social assistance for the first time as an area of government responsibility. It also introduced the citizenship principle behind social assistance, access was a right for all Brazilians; in contrast to the dominant contributory principle behind the development of social insurance from the 1920s.26

(p.161) The Constitution emphasized assistance to vulnerable groups, in particular older people and people with disabilities living in households in poverty. The Constitutional right led to the reform and expansion of two non-contributory pension schemes, the Previdencia Social Rural and the Beneficio de Prestacao Continuada. They had been introduced in a different guise in the 1970s, but with very limited reach and effectiveness. These were re-shaped, expanded, and extended. A separate initiative developed from municipal activism, it was grounded on minimum-income guarantee proposals combined with education. In 1995 municipalities began to introduce Bolsa Escola, transfers with schooling conditions. Their origins are in proposals for a guaranteed minimum income, again following the lead established in the Constitution, addressing poverty. The Workers Party, and Senator Suplicy, the first elected political representative of that Party, campaigned for this policy after the fall of the dictatorship. The proposal was taken up in several municipalities run by Workers Party politicians, but with an important change in connecting direct income transfers with conditions relating to schooling. The view was that without the links to basic services, the transfers would have very little effect in the medium and longer run, and much less policy and political traction. Bolsa Escola is a hybrid minimum-income guarantee and human development instrument, which reflected thinking on the left and centre that income transfers were not sufficient in the context of persistent intergenerational poverty. Bolsa Escola spread to other municipalities and in 1997 the government provided counterpart funding. In 2001 Bolsa Escola became a federal programme. Scaling up during Cardoso’s presidency was probably because its Plan Real to address hyperinflation had great short-term adverse effects, therefore it needed to be balanced by poverty activism in the short run. This led to scaling up Bolsa Escola and eventually making it into a federal programme.

During the Cardoso administrations in the late 1990s and early 2000s, addressing poverty through direct transfers became the new orthodoxy, in part through political competition within the government (e.g. Jose Serra creates the Auxilio Gas to compete with Bolsa Escola), but also competition with the Workers Party. This led to a proliferation of transfer programmes, with overlapping target populations. Together with other federal transfer programmes, it became Bolsa Familia in 2003.

In 2010, the two main non-contributory pension programmes reached about ten million households with a budget of around 1.5 per cent of GDP, while Bolsa Familia reaches over twelve million households with a budget of 0.4 per cent of GDP.

Parliamentary oversight and policy formulation has been a significant feature in Brazil (Britto and Soares 2010). Parliament’s active role in defining policy initiatives in poverty reduction goes back to the 1991 (p.162) proposal for a minimum guaranteed income by Senator Suplicy from the Workers Party; since that date several proposed bills and amendments have been presented and discussed in Congress every year. The proposals and amendments cover the spectrum, from right-of-centre politicians attempting to reduce the scope and reach of Bolsa Familia to left-of-centre parliamentarians aiming to expand the programme. Few bills successfully become law. Parliamentary activism reflects strong public opinion and interest in social assistance. There is much less parliamentary attention on the non-contributory pension programmes, in large part because their constitutional recognition implies that discretion over their implementation is very limited. Effectively the government is required to provide entitlements to all Brazilians who qualify for the benefits. Budgets simply reflect these entitlements. In the recent past parliamentary attention has made several attempts to change the target population, for example by redefining the scope of households for the purposes of defining entitlements, or restricting entitlements to the Previdencia Social Rural to residents in rural areas.

There are three levels of government in Brazil: federal, estate, and municipal. In Brazil municipalities are federal agencies with the same standing as the federal institutions. The estate level has not been active in social assistance. The federal government has influence over policy formulation, budgets, and implementation. Relationship between federal agencies and municipalities works through agreements and joint financing. An important federal tool to stimulate quality and performance is a Decentralization Index, which ranks municipalities according to their effectiveness and performance with implications for the federal financing streaming down. The Index is both a carrot and a stick (Lindert et al. 2007). It supports municipalities with deficient capacity and penalizes underperforming municipalities. The Index is a technocratic response to principal–agent issues, but increasingly modulates the partnership between federal and municipal levels. The Single Registry collects information on all households applying to any social assistance programme; the database enables the selection of beneficiaries and provides information on their progress through time. It also enables a stronger coordination among programme agencies, as it can be accessed by all agencies involved in the programmes.

The federal government allocates a fixed number of places for Bolsa Familia to municipalities. Local politicians and officials are responsible for registering potential beneficiaries. The information is assessed by the federal government and a score for each household determines eligibility. Local politicians and officials have some influence over the implementation of Bolsa Familia, through adding further interventions to Bolsa Familia or raising benefit levels, and also through ensuring the programme is implemented (p.163) effectively (e.g. whether they have filled the federal government allocations). The implication is that feedback effects are significant at the local level too. Politicians who can demonstrate effectiveness in implementing Bolsa Familia receive electoral support and recognition. Civil society and NGOs have a limited role in ensuring accountability of the programme at the local level, but the direct political accountability is more significant.

It could be argued that the expansion of social assistance in Brazil, and particularly Bolsa Familia, has extended the life of centre-left (Cardoso) and left (Lula) ruling coalitions. They are a dominant political force. Importantly, the expansion of social assistance developments has been shown to be consistent with fiscal responsibility, and retains a large measure of political support. Perhaps the most significant feedback effect from social assistance to politics is the rise of social policy and social assistance to the top of the political agenda. Poverty reduction has a high profile, and delivers electoral support for pro-poor politicians.

South Africa

In South Africa, social assistance can be traced back to the 1920s, with the introduction of the non-contributory pension for poor whites. Social assistance followed the European model of developing income transfers for groups of deserving poor facing acute vulnerability, but with the filter of racial politics. Social pensions were restricted to whites initially, but later incorporated Indians and coloureds and then blacks. The conditions of entitlement and benefit levels were differentiated along racial lines, until the mid-1990s when discrimination was abolished (Barrientos 2008; Lund 2008; Seekings 2008; van der Berg 1997). Over time, the range of direct transfer programmes expanded to include disability and family grants. By the time the first ANC government came to power in 1994, social assistance was fragmented due to the homelands policy of apartheid, and acutely under-resourced (Lund 2008).

The fall of apartheid led to a new Constitution in 1996 which reaffirmed a commitment to social assistance. Section 27 states that ‘everyone has the right to access . . . (i) social security, including, if they are unable to support themselves and their dependents appropriate social assistance’ (Seekings 2008; Woolard and Leibbrandt 2010). The ANC government took steps to review and strengthen social assistance provision. It established the Lund Committee which led to proposals for a Child Support Grant (CSG) to replace the Family Maintenance Grant. In 1997 the government also published a White Paper on Social Welfare which stated the objective of replacing poverty relief with a developmental approach to welfare. The CSG was initially designed to address child malnutrition and was focused on children (p.164) up to six years of age. Over time, it was extended to include children up to seventeen years of age. These measures led to a significant expansion of the reach of social assistance grants (Woolard and Leibbrandt 2010). By 2010, one in every two households had a social assistance beneficiary, and the budget has doubled since 1994 to over 3.5 per cent of GDP. Social assistance is the main policy instrument addressing poverty, vulnerability, and exclusion in South Africa. The grants are widely perceived to be effective in reducing poverty and vulnerability, to promote social inclusion and equity, and to have facilitated a difficult transition from apartheid rule.

The politics of the ANC have dominated the expansion of social assistance since 1994 (Nattrass and Seekings 2001). Initially, the challenge for the ANC government was to manage the transition from apartheid, while maintaining credible economic policies and fiscal responsibility. The government of national unity which directly followed the fall of apartheid in 1994 constituted more of a compromise than a new social contract.27 However, several initiatives which strengthened social assistance like CSG, the constitutional recognition of the right to social assistance, and the White Paper on Welfare Policy developed within a context of wide-ranging support for transformation within the parameters of fiscal responsibility.

The next significant political debate around social assistance came with the discussions surrounding the Taylor Committee of Inquiry into a Comprehensive System of Social Security for South Africa in early 2000s.28 A proposal for a basic income was receiving a substantial amount of attention from researchers and the trade unions. Intriguingly, the proposal for a basic income was supported by the National Party and the Communist Party (Matisonn and Seekings 2003). The arguments for a basic income in South Africa emphasized its advantages as a citizenship instrument, important in the context of racially segregated South Africa; as well as the more operational advantages of not requiring targeting and complex administrative implementation. The Taylor Committee supported the basic income proposal. The ANC rejected the basic income on three main grounds. Firstly, the White Paper on Social Welfare had argued for a change in the orientation of social assistance in South Africa, from poverty relief to a more developmental function. The basic income proposal was a step back from this objective. Secondly, there was no support within the ANC and outside for extending grants to the white population, especially given the large income differentials between them and the black population. Thirdly, maintaining fiscal (p.165) responsibility would have meant reducing the scope and generosity of social assistance in order to finance even a low level of the basic income.

Natrass and Seekings argue that while the ANC came to power committed to redistribution, and was/is expected to bring to effect significant redistribution, its capacity for redistribution has been limited by ‘policies that keep the economy growing along an inegalitarian path, with a large section of the poor being shut out of income generating activities’ (Nattrass and Seekings 2001, p. 495). There is debate among researchers on poverty trends in South Africa, but the general view is that poverty rates have remained broadly stagnant since 1994, while demographics has ensured that the numbers in poverty have risen (Leibbrandt et al. 2006; Leibbrandt et al. 2010). Social assistance has been extremely significant in preventing poverty from increasing. Overall, poverty and inequality outcomes remain problematic in South Africa, especially taking account of the fact that expenditure on social assistance has doubled as a proportion of GDP since 1994.

Right-of-centre politicians and business leaders are increasingly questioning the effectiveness of this large component of public expenditure and commonly voice concerns about potential dependency effects from the grants. Trade unions have pointed out the fact that unemployed groups constitute a gap in the social assistance grants system. On the left, some NGOs have explored the potential for judicial routes to expanding the grants (Seekings 2008). This has tested the position of the government on welfare policy. The equalization of the age of entitlement to the Old Age Grant to sixty for both men and women (it was previously sixty-five for men), for example, followed a challenge in the courts over possible gender discrimination associated with a differential age of entitlement. NGOs like the Black Sash play an important role in monitoring and facilitating the implementation of social assistance on the ground. They also perform a key role in ensuring welfare rights are fully exercised.

The potential for change and reform of social assistance is limited by the type of politics which sustain the ANC, and the ANC’s support for the existing social assistance architecture. Nattrass and Seekings argue persuasively that voter loyalty to the ANC depends on partisan identification with its role in bringing about political change in South Africa. To date, opposition parties are treated with considerable suspicion because of their association with apartheid; and few breakaway ANC groups have prospered. As Natrass and Seekings (2001, p. 488) put it: ‘Popular discontent over unemployment and job creation was partly offset by relatively more positive assessment with respect to other issues, including fiscal and social policy, even though the electorate regarded unemployment as the most important’. Political conditions therefore preclude any large-scale reform to social assistance, due to the strength of partisan support for the ANC. At the same time, social (p.166) assistance is important in maintaining and strengthening this partisan support, particularly in rural areas of the country. The feedback effect of social assistance on politics is therefore significant.

It is important to round up this assessment with a brief discussion of fiscal space. The government of South Africa has benefited from a significant improvement in fiscal revenues over time, from a high base. This has enabled a large expansion of the grants without affecting other areas of government expenditure. There are concerns that further expansion of grants expenditure could place pressure on service provision, especially in the context of the impact of the financial crisis (Van Der Berg and Siebrits 2010). Social assistance appears to have hit a ceiling in this respect. On the other hand, the government has committed itself to the introduction of a Comprehensive Social Security System in South Africa, replacing the patchwork of occupational pension plans with a government-supported, national, social insurance scheme. This large-scale social investment will add to fiscal pressures and will have implications for social assistance. It is not surprising that support for a national social insurance scheme comes strongly from trade unions and urban groups.

Triangulating the Case Studies

Tables 6.2 and 6.3 provide respectively a summary of key differences and similarities across the case studies. The cells on top of the diagonal summarize similarities/differences in policymaking and feedback effects. The cells below the diagonal summarize similarities/differences in implementation and programme dynamics.

Table 6.2. Similarities in political apparatus of social assistance delivery in case countries



South Africa

Policymaking and feedback effects


Policy implementation and programme dynamics


  • Multiple political parties and fragmented opposition.

  • Local/sub-national governments initially provide socia national governments extend them.

  • Social assistance investment opposed since it is not a tool for economic growth.

  • Direct evidence on political parties re-elected to power due to introduction of social assistance policies.

  • Debates around new constitution generated new set of common values and social contract.

  • Coordination for different social assistance programmes achieved in progressive manner in both countries.


  • Constitutional recognition brings higher parliamentary scrutiny for Bolsa Familia and NREGA in Brazil and India respectively and non-contributory social assistance programmes do not receive such scrutiny.

  • High level of autonomy for lower levels of administration.


  • Segmented approach to social assistance (in SA for whites and in India for employees of organized sector) exists early on. Later poverty and inclusive growth become key arguments for introduction of social assistance.

  • Concentration of poverty in social groups brings political mileage for policy intervention (identity politics).

  • Social assistance instrumental for managing change: in SA transition from apartheid rule; in India raising inequality due to liberalization.

  • Cooptation of academics in the process of knowledge creation on poverty-linked policies.

South Africa

Basic income debates form the basis against which social assistance programmes are weighed.

  • Cross support from different sections of socially divided voting population generates facilitative politics for social assistance.

  • NGOs play significant role in assisting/acting as countervailing force to government programmes.


Source: Authors.

Table 6.3. Differences in political apparatus of social assistance delivery in case countries



South Africa

Policymaking and feedback effects


Policy implementation and programme dynamics


  • Political party campaigning in Brazil, compared to India where civil society campaigns, which forces introduction of social assistance.

  • Decentralization index brings credibility to politics as a feedback mechanism in Brazil compared to demand generation through identity politics in India.

Municipal initiatives translated as federal level policy in a consolidation process in Brazil. In South Africa, social assistance is introduced at the time of new constitution as a strategy to support the black population.


Autonomous disbursement agency in Brazil compared with India’s decentralized governance system (which intensifies clientelist politics).


High-level coordination under single ministry in South Africa compared to large number of anti-poverty programmes spread over numerous ministries in India.

South Africa

Joint financing by local government and central government is present in Brazil. In South Africa, social assistance is centrally funded and delivered by provincial governments.

Limited or no autonomy for local-level government in South Africa compared to high level of discretion for lower levels of administration and politicians in India.


Source: Authors.

The expansion of social assistance from small privileged groups (whites in South Africa, organized sector employees in India, those able to afford contributory insurance in Brazil) to a broader target population, in line with universal principles, involved intense political process in all case countries. In India an active civil society movement through the ‘right to food’ campaign formed the medium through which this was articulated. The civil society representatives acting as advisors to a coalition government persuaded the government from within. In Brazil the fragmented opposition parties, such as the Workers Party, came together to campaign for the expansion of municipal initiatives to federal level, resulting in constitutional recognition of social rights for all Brazilians. In South Africa, the ANC, the party that spearheaded the transition from apartheid rule, immediately facilitated the right to social security to feature in the new constitution. In all these democracies, the parliament plays a critical role in steering the given fiscal capacity for expanding the social rights to all citizens. Since the right to social assistance is constitutionally guaranteed, the judiciary has become a (p.167) (p.168) new stakeholder in all the case countries, and from time to time the government is challenged to stand up to the promises.29

At the level of programme implementation there are important lessons for each other. Unique features of ranking different municipalities for reaching the target population (holding the politicians accountable) and the establishment of an autonomous disbursement agency are powerful tools (p.169) to create credibility politics in Brazil. In India, decentralized elected bodies are responsible for the delivery of social assistance, while financing is from central government. This has left too much room for manoeuvring policy at the implementation level. Thus, evidence on effectiveness is conflicting. Social audits conducted by NGOs and government agencies have reported conflicting outcomes. The South African system of implementation has a high level of coordination at central ministry level, leaving little discretion and autonomy to local governments. The NGO role in implementation is crucial in India and South Africa, compared to Brazil. In all three countries, the role of multilateral agencies is found to be extremely important, especially in providing technical support for implementation.

At the level of programme dynamics the unique features of case countries were more prominent than similarities across case countries. This is to be expected since the interaction of programmes with a unique socio-cultural context may produce incomparable dynamics. For instance, corruption that pervades other segments of life and business in India is also remarkably present in all its social assistance programmes. The corruption is politically steered and bureaucratically carried out. Corruption structures access to the programme. On the other hand, in South Africa, the history of apartheid rule and the emergence of ANC leaves limited space for political competition. The economic inequality prevailing between the white and black population has restricted support for basic-income grant proposals within the ANC ranks. In Brazil the political competition between the Workers Party and Cardoso’s poverty activism led to the mushrooming of various schemes which were later consolidated into Bolsa Familia replacing erstwhile Bolsa Escola. Despite these differences, there is commonality on debates around the stabilization of social assistance programmes. In all three countries general revenue finances the programme. Thus, debate on providing a stable finance to social assistance is active.30

In all three cases, feedback politics is significantly high. Although there had been initial resistance to the grants (compared to development funds) in both South Africa and India, once the programme is introduced, they gained huge popular support. In Brazil and India the introduction of social assistance has served to overcome anti-incumbency trends in the 2006 and 2009 general elections. In South Africa, the partisan support for the ANC has been strengthened because of social assistance. However, depending on (p.170) the architecture of governance, the feedback effect is different on local-level politics. Absence of coordination at central level for different social assistance programmes (and anti-poverty programmes) along with local autonomy has resulted in huge expenditures without convergence attempts in India. This has allowed local elites to turn the resultant divergence into political mileage. At the other extreme, there is a high level of coordination in South Africa by bringing all social assistance programmes under one single ministry. This gives no room for any feedback effect on local politics. The Brazilian system allows feedback on local politics as long as politicians can earn credits for reaching the targeted groups.

Discussion: Does Politics Matter?

In this section we return to the main questions of the chapter, and draw out the main conclusions.

Does Politics Matter for the Delivery of Effective Social Assistance?

The conclusions from the analysis in the chapter are that politics has played a central role in the expansion of social assistance in the three countries selected for detailed examination. This is a two-way process. On the one hand, political factors are at the core of the adoption, design, and implementation of social assistance. On the other hand, social assistance feeds back into political processes, helping reshape them.

What is the influence of politics in shaping social assistance in the South? Our case studies suggest that the influence of politics is strong. At one level, this is an obvious finding. At its core, social assistance is a manifestation of solidarity values in particular countries and communities, values which are themselves political.31 Social assistance represents institutions established with the objective of addressing poverty and vulnerability. Their scale and scope reflects shared understandings and the social priority attached to poverty. The main reason why this finding has purchase on current development discourses arises from the techno-managerial approach often employed by international organizations (Devarajan and Widlund 2007). This approach often ignores political influences on social protection institutions. The relevance of the politics for social assistance delivery is relatively under-researched. Explaining the rapid emergence of social assistance in developing countries has to address this knowledge gap.

(p.171) To achieve this, a stronger conceptual framework is needed. This framework will need to separate out, as we do in this chapter, the different dimensions in which politics matter, and the different dimensions of the influence of politics on social assistance delivery. The examination of three case countries in this study reveals that traditional approaches to this question focused mainly on investigating incentive structures as a motivation for elected politicians and non-elected bureaucrats (Alesina and Tabellini 2004). This is too limited as a tool for capturing the political dynamics around social assistance in Southern democracies. Social contracts and pacts, key events, ideology, and knowledge are important too. Unlike conventional wisdom, our case studies suggest several dimensions in which political influences need to be studied. A sole focus on elites or on full political competition are also limited as approaches to social assistance (Krishna 2006). These different dimensions need to be studied separately and then integrated to form a more comprehensive picture.

Social assistance institutions also help shape political processes. As noted in this chapter, these institutions feed back into political processes at different levels and in different ways. Many studies have discussed, and measured, the role of social assistance in aligning electoral support for incumbents. This dimension, if overemphasized, can reduce social assistance to a purely instrumental function (Hall 2008). Social assistance should also be considered as a legitimate form of aligning party coalitions and political support around pro-poor policies (Stokes 2004).

The growing institutionalization of social assistance as rights through intense political struggle is the story in all three case countries. This redefines groups in poverty as citizens (social citizens). A deepening of democracy follows. In India, for example, informal labourers had little to give and take from the state. Thus, the state’s legitimacy itself was limited. Social assistance challenges this relationship and re-establishes legitimacy.

The case studies reveal a complex interaction between local and national politicians and the electorate, as in India and Brazil, but also point to the role of identity politics and filters, as in South Africa. At a higher level of abstraction, social assistance can embody, and therefore strengthen and develop, shared values and preferences. These institutions can firm up and develop social contracts and shared notions of social justice. In many countries, but especially in India, Brazil, and South Africa, social assistance is widely perceived as an instrument of inclusion. Just as in low-income countries with poorly developed social protection institutions, social assistance can be perceived as an instrument of exclusion. Feedback processes from social assistance and politics have not been studied sufficiently in the literature.

The delivery of social assistance cannot be de-linked from local dynamics and politics. Increasingly, political credibility requires procedural fairness. (p.172) At the implementation stage, efficiency gets redefined as egalitarianism once politics takes centre stage as a means to explain the delivery of social assistance. New centre–local relations are being created (e.g. joint financing, joint monitoring, joint issue-making, joint definition of targets) through the politics of social assistance. Leakages due to poor implementation are part and parcel of political resistance to redistribution by local elites.

In India there is a stark contrast between the politics of development (roads, drinking water, sanitation) and the politics of social assistance (pension, employment), indicating that politics of social assistance is better conceived as a process than as an arena.

On each component of this two-way process, further research is needed to provide deeper insights. Contrasting the autonomy exercised by Brazilian institutions for delivering social assistance with that of India and South Africa it is important to focus on how effective service delivery strengthens political credibility. It is also important to examine in more detail the influence of identity politics, as observed in South Africa and India, in reducing the role of service delivery in affecting credibility-based politics (Keefer and Khemani 2003b). Identity politics structures social assistance through stratified mechanisms. Identity politics could also prevent the expansion of social assistance. This is an important area for further research, the linkages existing between (in)effective delivery of social assistance and political credibility.

Decentralization of power to lower levels of administration could intensify politics around the delivery of social assistance. In India the highest impact of feedback processes is through local politics. In Brazil, though decentralization is significant, the influence of local politics is reduced through strong autonomous institutions at the national level that regulate the delivery of social assistance. In South Africa, high level of coordination of social assistance programmes under a single ministry creates an environment through which local politics gets distanced. In other words, decentralization has to work with other countervailing forces. If not, as in India, the main outcome is a situation in which authority diffusion is rampant allowing the capture of social assistance programmes by local elites.

Knowledge Gaps

We conclude this chapter by noting some glaring knowledge gaps. Conceptual frameworks developed to study redistribution in welfare states can provide important insights into the expansion of social assistance in developing countries; but they need to be adapted in at least two important respects: i) redistributive models, especially those relying on welfarist assumptions, are ill suited to the study of non-welfarist poverty reduction; and ii) neither ‘ideal’ conditions of political competition nor ‘elite capture’ (p.173) models are appropriate to the study of the politics of the expansion of social assistance in developing countries.

  • • There is a literature providing country-specific or programme-specific information and discussion of social assistance. The main knowledge gap, and a key priority, is the need to develop comparative approaches and models.

  • • The feedback from social assistance to political processes has been studied to some extent in middle-income countries for specific countries/programmes. The knowledge gap is at the cross-country comparative level, and at the cross-state comparative level. Cross-country comparative studies can generate knowledge on the role of political institutions, democratization, left–right coalitions, and donors, inter alia, in the specific processes and outcomes as regards social assistance delivery. Cross-country studies, institutional/qualitative/quantitative, can support useful generalizations on the role of political institutions and processes on the expansion of social assistance. Infra-national comparisons can generate information on the relative influence of national–local politics and bureaucracies.

  • • There is scarce literature on the extent to which social assistance generates synergies and constraints in the delivery of other services and public goods in developing countries. In the context of human development conditional transfer programmes and employment guarantees, there is a prima facie case for examining these interactions. They have also been observed in pure transfer programmes, like South Africa’s and Brazil’s social pensions in the context of the bundling of services and transfers.

  • • There is a need to distinguish, analytically, social contracts and pacts, from scaling up and reform and evolution of social assistance, as was done in the framework above. Different forms of politics apply to these.

  • • Despite the usual ‘development hubris’, effective delivery of social assistance and the politics associated can be productively studied in high-income countries. Pace the significance of registration (Szreter 2007); the role of left parties in social assistance (Esping-Andersen 1990); the shift to ‘active’ labour market policies following persistent high unemployment in the 1980s (OECD 2003). It is much less productive to focus on welfare states, which at any rate are a post-Second World War phenomenon predicated on full employment and ‘standard’ family structures. It would be useful to take a hard-headed view (no nostalgia) on actual social assistance in high-income countries and see what can be learned from this by developing countries. (p.174)


(1) They can also make a marginal contribution to the reduction of inequality. A recent cross-country study on declining inequality trends in Latin America identifies two main explanations: (i) a fall in the premium to skilled labour; (ii) the impact of higher and more progressive government transfers. See (López-Calva and Lustig 2010).

(2) Uncertainty over terminology and scope is greatest in international policy debates.

(3) A good case in point is NREGA introduction in India. With the introduction of the programme, poor people got the opportunity to interact at the sites of work location, and to organize themselves as ‘NREGA workers’. Some state governments facilitated this potential opportunity to bring accountability in to the work systems and to improve governance by using the feedback from the NREGA workers (Pellissery and Jalan 2011).

(4) See, for a recent survey, Robinson (2010). The classic source is Meltzer and Scott (1981).

(5) Politics and social assistance played a larger part in his earlier work (see Esping-Andersen 1990). Attempts to adapt the framework to developing countries have focused on institutions, and ignored politics (see Gough and Wood 2004).

(6) The selection of cases can also get complicated when we note that different programmes within a single country can show different levels of political support. Indonesia has introduced large-scale social protection programmes since the financial crisis in 1998 to help families cope with the effects from financial shocks. They include food subsidies (Raskin), Unconditional Cash Transfer (Bantuan Langsung Tunai), Conditional Cash Transfer (Program Keluarga Harapan), Cash for Work and School Assistance. Among these, unconditional cash transfer programmes have been the least politically acceptable both at local level and national level, primarily due to targeting errors (see Bambang Widianto ‘The Political Economy of Social Protection in Indonesia’, a paper presented during the international conference on ‘Reforming Social Protection Systems in Developing Countries’ held 20–1 October 2011 at the Institute of Development Research and Development Policy, Rurh Universitat Bochum, Germany).

(7) ‘What institutions or policies a political system generates depends on the distribution of power in society and how political institutions and mobilized interests aggregate preferences’ (Robinson 2010, p. 39).

(8) For a discussion of the citizenship and contributory principles in the context of social protection and the social security reforms in the 1980s in the UK, see Plant (2003).

(9) The ‘War on Poverty’ in the USA in the 1970s or the building of welfare states in European countries in the post-Second World War period provide examples of an expansion of social assistance based on the emergence of a broad consensus across governing coalitions but without originating in a single event. Rawls’s ‘veil of ignorance’ is an abstract construct which can be interpreted as a counterpart to direct democracy (see Rawls 1971).

(10) This applies to the distinction between Bismarckian and Beveridgean social security, for example.

(11) See, for example, the origins of the Swedish pension system.

(12) By autonomy we understand the extent to which public agencies can influence the shape of social assistance in line with their own particular preferences or interests.

(13) Private sector employers of agricultural labour in India are probably an exception. The National Rural Employment Guarantee in India has helped raise market wages in this sector.

(14) Global social policy approaches give international organizations a dominant role (see Deacon 1997).

(15) Regional influences might be significant in certain contexts, especially where clusters of specific types of programmes can be observed at the sub-regional level. In Southern Africa, pure transfers to groups considered to be acutely vulnerable are dominant, especially non-contributory pensions and child grants (see Devereux 2007). Human development transfer programmes are dominant in South America (see Cecchini and Martínez 2011).

(16) For a long view, see Osmani (1991).

(17) The largest of the programmes, the Public Distribution System, which distributes essential food items and non-food items (e.g. kerosene) through a network of fair-price shops, incurred expenses close to 1% of GDP primarily as subsidy. This was largely untargeted until 1990s. Apart from this, specific nutritional programmes for children and pregnant women were in place under an umbrella programme of the Integrated Child Development Scheme (ICDS).

(18) Since the 1960s a large number of public works programmes which provided cash or kind in exchange for labour was a predominant mode of relief especially in times of drought in rural areas.

(19) These sets of programme were initiated in 1952 and the community was consulted while designing what kind of programme could bring developmental changes. Thus, intervention varied from providing assets such as milk animals, to improving dry land, to providing market linkage or finance for small business.

(20) There is considerable scholarship studying the advantages accruing to the political class from the introduction of such anti-poverty programmes. See for instance, among others, Kohli (1989); Harriss (2004).

(21) See Dev (1998) for a list of programmes showing the year they were introduced and the states in which they were introduced.

(22) The figures for reach are based on the World Bank (2011). General evaluation of these programmes concludes that though the benefits are small, it makes a big difference to the households (Dutta et al. 2010).

(23) It is important to note that the prefix of ‘Mahatma Gandhi’ was introduced to the programme after the government was voted to power without coalition forces, particularly left parties in 2009.

(24) Since the elected body of panchayat (village-level government) has control over huge amounts of money allocated for NREGA works, panchayat elections in many northern states have become more contested (with more money spent by candidates during elections) since the introduction of NREGA. A study has shown that participants in the NREGA programme are more likely to be selected along caste lines when a member of their caste is elected as head of a panchayat. For the same effects on social pensions see Pellissery (2008, 2005).

(25) The experience of Andhra Pradesh, which leads in the implementation of NREGA, has a state-wide NGO consortium where regular monthly meetings are held between NGOs and government agencies at state level, district level, and sub-district level. This has been instrumental in coordinating natural resource management and NREGA in the state. On how this helped to improve food security in the state see Pellissery and Sanju (2011).

(26) ‘What institutions or policies a political system generates depends on the distribution of power in society and how political institutions and mobilized interests aggregate preferences’ (Robinson 2010, p. 39).

(27) The minister responsible for social assistance in the national unity government was in fact from the National Party (Lund 2008). The National Party and Inkatha left the Government of National Unity in 1996.

(29) In South Africa the judiciary pressed the government to grant equal entitlement ages for the old-age grant to be from age sixty for men (from age sixty-five). The government of India has been challenged in the court to increase the wages of employment guarantee to take account of inflation. In Brazil, the government has been taken to court to make the threshold of social pensions one quarter of the minimum wage.

(30) Comparable is the case of Thailand where significant attempts to introduce social protection programmes have taken place in recent times. Many have argued that populist policies are a structural element of inequality (for instance, see Jitsuchon ‘How can Thailand Escape a Vicious Cycle of Populist Policies?’, a paper presented during the World Economic Forum on East Asia 2012 during 30 May–1 June in Bangkok). Inequality forces political parties to focus on populist policies at the time of elections. In Thailand such loss of macroeconomic prudence has left the finances so weak that a comprehensive social welfare system could not be built.

(31) Rawls’s (1971) ‘political conception of justice’ is a comprehensive statement on this point.