The politics of social policy: welfare expansion in Brazil, China, India and South Africa in comparative perspective

ABSTRACT This introductory essay reviews the scholarship on the politics of social policy, and shows the contribution of the special issue to explaining expanded welfare commitments in Brazil, China, India and South Africa in the twenty-first century. Much literature on welfare expansion in lower- and middle-income contexts views it primarily as a policy corrective to the economic dislocations produced by global economic integration. This special issue focuses on the political factors that are critical to understanding the shape social policies have taken and their effectiveness in ameliorating poverty and inequality.

The papers in this special issue compare the political drivers of social policy expansion in four major, emergent economies that are sometimes referred to as 'rising powers': Brazil, China, India and South Africa. In the first decade of the twenty-first century, these four countries developed new programmes to tackle poverty and inequalityat a time when many advanced industrialised economies were questioning whether welfare state commitments were compatible with global competitiveness. The special issue examines why they did so. It also explores cross-national similarities and differences in the politics of social policy expansion so as to better understand the contours of social policy expansion in the global south. Each paper looks at a different dimension of those politics, from the influence of political leaders and the role of ideas, to the effects of institutions such as courts and federal systems on the development of social policy. Collectively, their aim is to highlight the actors, ideas and institutions that have shaped provisions to tackle poverty and inequality in emerging economies, and thus to develop existing comparative theory.
Together, they demonstrate that politics matters in determining how social policy is framed, and for explaining whydespite comparable economic circumstances arising from economic growth and closer global integrationthe shape and outcomes of social policies have differed between countries.
The special issue contributes to a growing literature on the politics of social policy in developing countries that has developed largely on a regional, rather than global, comparative basis. Much existing scholarship focuses on withinregion comparisons, for instance explaining differences between regimes within Latin America (such as Fenwick, 2016;Huber & Stephens, 2012;Niedzwiecki, 2015;Pribble, 2013;Segura-Ubiergo, 2007), East Asia (such as Holliday, 2000;Mok & Lau, 2014) or South Asia (Koehler & Chopra, 2014;Singh, 2016;Tillin, Deshpande, & Kailash, 2015). But studies that attempt cross-regional comparisons are considerably fewer (notable exceptions include Gough & Wood, 2004;Haggard & Kaufman, 2008;Melo, Ng'ethe, & Manor, 2011;Rudra, 2007;Sharkh & Gough, 2010). Other projects that have included multiple country case studies have not moved to the stage of cross-case comparative analysis (for example, Midgley & Piachaud, 2013). As a result, scholars are left with many fine-grained explanations of the political drivers of differences within individual countries or in families of countries with similar political histories or models of economic development, but fewer attempts to explain patterns across countries with starker differences of historical background or indeed regime type. This lack of wider comparative analysis means existing studies may unwittingly overstate the exceptionalism or universalism of their case (or set of cases). They also find it harder to specify which factors have been moreor lessimportant in individual cases. As a result, the political drivers of social policy expansion in lower and middle-income contexts remain under-specified.
In the introduction to this special issue, we survey the existing literature on the politics of social policy within and across regions. Around 70 per cent of the world's poor live in emerging economies that increasingly have the resources to invest in poverty reduction themselves rather than rely on foreign aid (Sumner, 2016). Thus there are compelling reasons to compare the strategies they have taken to reduce poverty and inequality and the political reasons behind such strategies.
The expansion of social provision in the global south Very many countries in the global south expanded social provision from the mid-1990s. The most prominent type of new policy was some form of social assistance, in particular anti-poverty transfer programmes that were generally tax-financed and non-contributory. Opening a recent volume on social assistance in developing countries, Barrientos (2013) underlines the reach of these policies across countries by pointing to the concurrent development of the Dibao, or Minimum Living Standards Guarantee, in urban areas of China in the mid-1990s, the implementation of Bolsa Escola through which cash transfers were linked to school attendance (the forerunner to Bolsa Familia) in some Brazilian municipalities in 1995, and the introduction of the child support grant in South Africa in 1998. 1 Countries as diverse as Kenya, Malawi, Nigeria, Liberia, Paraguay, Honduras, Argentina and Pakistan are currently experimenting with cash transfer programmes (Barrientos, Niño-Zarazúa, & Maitrot, 2010). Many have also introduced, or extended, initiatives that combine income transfers with initiatives intended to build the wider social and political capabilities of the poor to escape poverty over the longer term. These include public works programmes, such as India's Mahatma Gandhi National Rural Employment Guarantee Act. In addition, numerous countries have also introduced new forms of social insurance (such as health insurance or social pensions) that have extended benefits previously reserved for formal sector workers to a wider population.
At an overarching level these new social policies can be seen as part of lower and middle-income country governments' policy responses to the consequences of varying constellations of economic crisis, structural adjustment and globalisation, and recent transitions to democracy. Policies adopted in the 1980s and early 1990s in response to economic crisisincluding public expenditure cuts, privatisation and decentralisationhad often increased poverty and inequality and undermined the quality of service provision. In addition, by the early 2000s, some large emerging economies were experiencing higher levels of economic growth, which increased the resources available for new social policies.
However, the concurrence and basic similarity of experience across such diverse countries may have obscured the need for comparative research into the political origins of social policy change. One of the most common assumptions is that the primary drivers of social policy expansion were economic. Globalisation and rapid economic transformation, including liberalising reforms that increased informalisation, enhanced the vulnerability of the poor in developing countries. As Barrientos and Hulme (2009, p. 443) state: 'Increasing poverty and vulnerability arising from globalization and economic transformation are … key drivers for social protection.' 2 In this context, domestic governments, international agencies including the World Bank, and donor agencies such as the United Kingdom's Department for International Development (DfID) began either to experiment with, or to encourage, new forms of social protection (Barrientos & Hulme, 2009, p. 448). Given the comparable economic transformations across emerging economies and the nascent international discourse of inclusive growth, domestic politics has often taken a back seat in explanations of where social policy initiatives came from, how they were shaped and sustained.
Some critics have gone further to argue that 'mainstream' development discourses in the global north have primarily regarded social policy as a corrective to economic dislocations, rather than as a form of collective public action to improve well-being (Adésínà, 2009;Mkandawire, 2004). Such critics suggest that the result has been to marginalise non-Western welfare systems based on cultural traditions of familial or cultural solidarity, and to champion an individualism that is inimical to collectivist traditions (Adésínà, 2009;Midgley, 2004). From this perspective, global efforts to reinvigorate social development in the 1990s, such as the Copenhagen Declaration, crowded out national perspectives. Following such lines of reasoning, the relative paucity of serious attempts to compare the domestic politics driving social policy choices and outcomes in different countries of the global south may in part arise from the assumption that social policy in this era has itself taken on a thinner, 'safety net' quality that is increasingly detached from local traditions and more susceptible to global diffusion.
Thus for many reasons, the structural changes faced by developing countries in an era of globalisationand responses to themhave been viewed in essentially similar terms to those faced by advanced industrialised countries. In the global north, it was argued that globalisation was opening up new social fissures between those with skills and mobility, and those without, meaning that policy-makers at the turn of the millennium had to ensure that 'international economic integration does not contribute to domestic social disintegration' (Rodrik, 2008, p. 2). Such arguments about necessary correctives to the dislocations arising from global integration were also being applied to understand the compulsions driving governments in lower and middle-income countries to expand social provision. For instance, Arvind Subramanian, now chief economic advisor to Indian Prime Minister Narendra Modi since 2014, applied Rodrik's argument to India: Financial integration, whatever its merits, exposes the economy to greater volatility. Cushioning against this requires social insurance mechanismsoften in the form of greater spending, especially during downturns … . In 2008 and 2009, the Indian economy, buffeted by the vagaries of international marketsbecause of India's increasing integration with themhad to deploy countercyclical fiscal policies. One legacy is increased government spending and larger overall deficits. (Subramanian, 2013) Yet an earlier generation of literature had also sought to explain welfare state retrenchment since the 1970s as an outcome of globalisation, especially the increased competition over mobile capital and downwards pressure over labour costs in a competitive international market (for a summary see Swank, 2005). There is thus no consensus on either the causal direction of increased trade openness and capital mobility on social policy, or on the precise connections between these policies. Many studies have highlighted instead the effects of de-industrialisation and the shrinking of the manufacturing sector within countries (Iversen & Cusack, 2000), a shift of the workforce into sectors where productivity improvements are more limited (Pierson, 2001c), the demographic pressures of an ageing population (Hicks & Zorn, 2005) and domestic politics more broadly (see, for example, Huber & Stephens, 2001;Iversen & Cusack, 2000;Swank, 2002Swank, , 2005. Furthermore, social policy changes have taken on clear regional inflections. This suggests that a more localised, rather than straightforwardly globalised, form of policy diffusion may have been at work (Barrientos & Hulme, 2009). For example, cash transfers have been popular in Latin America and parts of Sub-Saharan Africa, but in South Asia there has been greater scepticism about replacing older in-kind subsidies with cash transfers. The existence of regional patterns suggests that domestic political factors and historical trajectories, are important in influencing the shape and consequences of social policy interventions in response to larger economic transformations.
Lastly, social policy initiatives adopted across the global south, including Brazil, China, India and South Africa, have gone beyond simply establishing a 'safety net' or residual function during periods of high economic growth when resources can be redistributed to prevent social dislocation. Instead, as papers in this special issue discuss, many new initiatives have been designed in ways that seek to build social and human capital or the capacities of the poor, as well as alleviating income poverty. In other words, social policies in the global south have both a protective and productive function. 3 These policies are conceived by many policy-makers as long-run contributors to economic growth, and hold a political appeal to leaders of both democratic and authoritarian governments. There are strong indications that policy agendas in lower and middle-income countries increasingly, even if partially, include a concern for elements of what a growing literature calls a 'social investment' regime (see, for example, Deeming & Smyth, 2015;Drèze & Sen, 2013;Midgley, 1999;Midgley & Piachaud, 2013). 4 In some countries, such as India, there has also been a focus on establishing social 'rights' (Harriss, 2013;Jayal, 2013).
Overall, the social policy mix and extent is vigorously debated and contested in most of these countries, with China a partial exception because the debate there is more constrained. This further suggests that domestic actorsincluding political parties, individual leaders, civil society organisations, wider epistemic communities and judges play influential roles. 5 And the outcomes of their debates and actions are necessarily shaped by the institutional contexts within which they operatepolitical regime types, federal arrangements and the constitutional role of courts. All these domestic factors are critical for understanding the emergent landscape of welfare in emerging economies and yet research has hitherto neglected them.
The contributions to this special issue take a comparative approach to analysing the domestic politics that have shaped the social policy of four major polities in the global south since the turn of the twenty-first century. Our analyses start from the assumption that countries were responding to the legacies of liberal economic policies of the 1990s which had reduced social sector investment, and that new social policies were enabled by the revenues generated by faster economic growth. These factors help to explain the timing of new investments in social provision from the late 1990s, intensifying in the 2000s in middle-income countries. But they do not give us a full account of the different approaches and overall policy packages adopted in different countries experiencing similar economic trends. For a more complete picture, we need to turn to the politics.
The remainder of this introduction will provide an overview of the major approaches to understanding the politics of welfare state expansion in both advanced industrialised states and in lower and middle-income countries. We consider how well existing theories capture the departures in social policy witnessed in Brazil, China, India and South Africa and outline the contributions of papers in this collection to understanding the politics of these important transitions in welfare provision. We begin, however, by outlining the rationale for focusing on Brazil, China, India and South Africa and provide details of the collaborative research programme on which the special issue is based.
Expanding, not shrinking, social programmes in Brazil, China, India and South Africa The papers that form this special issue arise from research conducted by four teams of researchers involved in the Economic and Social Research Councilfunded project, 'Expanding, Not Shrinking Social Programmes: The Politics of New Policies to Tackle Poverty and Inequality in Brazil, China, India and South Africa' led by Professor James Manor at the University of London. Each of the four teams studied the political dynamics of initiatives designed to tackle poverty and inequality in Brazil, China, India and South Africa respectively. Fieldwork for the project was conducted between 2012 and 2016, and focused on policies that were initiated between about 2003 and 2013 in each of the countries. These were periods in which governments in all four countries emphasised tackling poverty and inequality after periods when they had concentrated more narrowly on economic growth and macro-economic stabilisation. 6 In most of these countries growth has faltered in the period since 2013, and new governments have faced a combination of fresh fiscal and political pressures that have raised questions about the risks of retrenchment. The special issue focuses primarily on the political dynamics of the expansionary era, but some of the analyses also reflect on the sustainability or otherwise of the gains made in this period.
In Brazil, the advent of a government led by the Worker's Party (Partido dos Trabalhadores or PT) under President Luiz Inácio 'Lula' da Silva (2003-2010) marked a departure in levels of social investment. 7 The PT ruled in a centre-left coalition, maintaining the macro-economic policies of the preceding PSDB (Partido da Social Democracia Brasileiro, or Brazilian Social Democracy Party) government, but overlaying them with a more vigorous focus on social inclusion. The PT's flagship social policy was 'Bolsa Familia', a conditional cash transfer programme that built on earlier municipal experiments and the 'Bolsa Escola' introduced under the Cardoso regime. Under Lula, almost 30 per cent of the population received cash transfers under Bolsa Familia in return for fulfilling conditions around school attendance, childhood immunisations and maternal healthcare. Bolsa Familia is credited with playing an important role in reducing income inequality in Brazil, the only one of the four countries in this special issue to have achieved such a reduction (for a summary of the large literature on this see Pereira, 2015 and Maiorano and Manor, this special issue).
In China, the Hu Jintao-Wen Jiabao administration (late 2002-early 2013) introduced policies to reduce income inequalities and regional disparities, and to improve social fairness and protect the rural disadvantaged. This was a noticeable break with the pursuit of economic reform and growth that had characterised China under previous top leaders Deng Xiaoping and Jiang Zemin (Zhao, 2013). Under Hu and Wen, the Party-state created a number of social programmes focused specifically on rural areas. These included a pension scheme that extended across rural China, and New Rural Cooperative Medical Schemes that reversed the health sector retrenchment of the preceding period (Duckett, 2011). 8 In 2007, the Party-state announced the extension of means-tested poverty assistance, or dibao, to rural areas. It also paid more attention to the social, political and economic inequalities faced by migrant workers in Chinese cities, although progress was slower on this front. As Howell (2014) writes, the persistence and violence of rural protests as well as frequent strikes and demonstrations at coastal factories, may have put pressure on the new leadership to more systematically address poverty and inequality. However, given that rural protests focused on local governments not the centre, and that rural dwellers are generally politically marginalised, the role of policy ideas and external shocks (such as the Asian financial crisis, and the SARS epidemic) may loom larger in explaining the shift in social policy (Duckett & Wang, 2017).
In India, the general elections of 2004 saw the election of the Congress Party-led United Progressive Alliance (UPA, 2004(UPA, -2014. This government interpreted its mandate as reflecting a revolt of the rural poor who had been left out of the benefits of economic liberalisation pursued since the early 1990s. While this was not an entirely accurate account of the social dynamics behind the election outcome, it provided the rationale for a new era of investment in rights-based social programmes (Manor, 2011). The UPA government began with the flagship Right to Information Act 2005 and the National Rural Employment Guarantee Act 2005 (under which rural households could demand up to 100 days' work 9 ), and during its two terms in office also passed the Right to Education Act 2009 and the National Food Security Act 2013 that gave legislative protection to the notion of a 'right to food'. Informed by prominent civil society activists who sat on a new National Advisory Council and helped to draw up legislation, the UPA government sought to put social rights on a legislative footing (Harriss, 2013;Jayal, 2013;Koehler & Chopra, 2014;Ruparelia, 2013).
In South Africa too, a rights-based approach to social policy focused on redistribution and building human capabilities has been a prominent goal of the post-Apartheid African National Congress (ANC) government since 1994 (Patel, 2013). Under Thabo Mbeki's leadership in the mid-1990s, South Africa adopted market-oriented reforms that were widely perceived to have limited the state's role and investment in the social sector (see summary in Friedman & Niekerk, 2016). But even Mbeki raised concerns about, what he called, the 'Two Nations' living within South Africaone largely white, urban and well-connected to the international economy; another largely black, living in peri-urban, informal settlements and rural areas exposed to high levels of crime, unemployment and disease (Andersson & Alexander, 2016). Contrary to the perception that a neoliberal policy shift shrank the room for social policy, the government extended social programmes markedly over the course of the late 1990s and 2000s. It increased social grants, introduced new housing and community works programmes and improved access to anti-retroviral medicines. While the effectiveness of South Africa's social policies in addressing poverty and inequality may be debated (see Friedman & Niekerk, 2016 and Maiorano and Manor, this issue), there is no doubt that they expanded.
In this special issue, authors have stepped beyond their country expertise to engage in cross-country comparisons that allow them to isolate critical themes in understanding the politics that has produced pro-poor policies in some instances and not in others. Brazil, China, India and South Africa are ripe for comparison not only because of the evident similarities in their trajectories of welfare expansion and successful economic growth. They are also all large, complex states that are all now classified as middle-income countries. And they share some political similarities. Legacies of nationalist and freedom struggles have historically loomed large in framing debates about development (in China) and social justice in India and South Africa, while Brazil and South Africa have experienced processes of democratisation within a decade of each other. In Brazil, India and South Africa, socio-economic rights are either included in post-democratisation/independence constitutions, or in India's case, have become to resemble justiciable rights as a result of judicial initiative and legislative action in the more recent period (see Friedman and Maoirano, this issue). Two of the four countries are formally federal (Brazil and India), a third comes close (South Africa) and while China, as an authoritarian state is politically centralised, it is a complex, multi-level polity (sometimes referred to as fiscally federal) with economically powerful local authorities that can shape its approach to policy-making and implementation.
There are, however, challenges in comparing these countries. The biggest comes from the inclusion of China. As an authoritarian regime, the dynamics of policy-making in China look different, with non-state actors, legislatures and courts having much less influence than in the democratic settings of Brazil, India and South Africa. Yet it is important to remember that there are also very substantial differences between democraciesincluding those in our group. Brazil has a presidential system, while India and South Africa have parliamentary democracies. Identity and ethnicity play an important role in political life in India and South Africa, but a lesser role in Brazil. Furthermore, the economies of all four countries look different. The drivers of growth are different: export-led manufacturing in China, service-sector led in India and commodities-led in Brazil. And their income levels also vary substantially: the World Bank classifies Brazil, China and South Africa as upper middleincome countries, but India as lower middle income. Thus we cannot treat any of these comparisons as tightly controlled, but we think there are sufficient commonalities in processes and policy emphases to justify moving from a focus on country cases alone to comparative analysis.

Approaches to understanding the politics of social policy
Having set out an overview of the expansion of social policies in Brazil, China, India and South Africa since the early 2000s, the following section reviews the existing literature on the ways in which politics shapes social policy. We identify three broad established approaches. The first comprises a body of political economy literature focused on 'varieties of welfare capitalism' and suggests that models of capitalist development produce distinct sets of choices about social policy provision. The second approach consists of studiescommonly labelled 'power resources' theoriesthat unpack the ways that the organisation and strength of different interest groups, and particularly trade unions and left-wing parties, influence policy. Finally, the third approach, which followed the institutionalist turn in political science, focuses on political institutions (such as electoral systems and numbers of parties, executivelegislative relations, the role of courts and structures of federalism and/or decentralisation) and policy feedback (the effects of prior policy choices in constraining or shaping the options of later policy-makers). To these three reasonably well-established approaches, we add a fourth that may be critical in developing country contexts but is less well-studied in its own right: the relatively autonomous role of political leaders or 'policy entrepreneurs' in challenging or changing paths of policy development.

Varieties of welfare capitalism
As Pierson (2001a, p. 5) notes, the welfare capitalism literature tends to see the welfare state as a central component of capitalist development rather than simply a Polanyian style 'protective reaction' to the expansion of modern capitalism. Social policies have implications for many of the basic decisions made by firms including the cost of hiring, firing and training workers and the opportunities for collective organisation by workers (Pierson 2001a, p. 5). Much of this literature does not explicitly explore the causal processes that link varieties of capitalism and welfare state development, but rather looks for patterns among similar countries.
In recent years, several studies have built on Esping-Andersen's (1989) work on varieties of welfare capitalism in advanced industrial contexts and looked for welfare regime types among lower and middle-income countries. Within this small literature, a distinction has been drawn between what Rudra (2007) defines as 'productive' welfare states (those that aim to promote the market dependence of citizens through wage labour and investment in basic education and health care) and 'protective' states (those that aim to protect certain individuals from the market through policies that include public employment, social security and pensions, housing subsidies, labour market protections and tertiary education). A third cluster of countries encompasses 'dual welfare states' that combine elements of both.
Some research has explored whether particular welfare policies are linked to the pursuit of an export-led strategy of industrialisation (as in East Asia) as opposed to a more protectionist import substituting industrialisation (ISI) model (as in much of Latin America and India until the 1980s) (Haggard & Kaufman, 2008;Rudra, 2007;Wibbels & Ahlquist, 2011). This work suggests that states that encouraged an export-led strategy encouraged firms to maintain international competitiveness by containing costs and pursued social policies that served the interests of both workers and employers. These states focused primarily on producing productive workers by providing basic health and education and by offering limited benefits such as social insurance at a level that was acceptable to employers competing in the international economy. They often fell back on repressive labour policies that supported the interests of employers over workers. Thus, across much of East Asia (but not China at this time), welfare systems often offered minimal social insurance but placed emphasis on education (Haggard & Kaufman, 2008;Rudra, 2007). 10 By contrast, states that adopted the ISI model in Latin America extended relatively generous protection to urban middle class salaried employees and some blue-collar workers, but largely excluded peasants and informal sector workers (Haggard & Kaufman, 2008). India followed a somewhat similar pattern to Latin America with a set of labour regulations adopted in the 1940s, in a continuation of colonial policies, that protected a small minority of the working class in the formal sector with trade unions largely incorporated by political parties (Agarwala, 2013;Candland, 2007).
Rudra suggests that states that adopted an ISI model that shielded domestic producers from international competition produced welfare models that focused more on protecting workers from the market. The state retained tighter control over the economy and firms did not face the pressures of cost containment arising from international market competition. In such situations, governments could subsidise groups who were empowered by the lack of international market exposurecivil service employees, the military and urban formal sector workers (Rudra, 2007). Recent literature on the expansion of social policy in Latin America concurs that in countries that industrialised this way, an emergent urban working class pressed directly for the expansion of social security or, more often, saw benefits expanded in a process of topdown political co-optation that produced limited, or 'truncated', welfare regimes (Pribble, 2010;Segura-Ubiergo, 2007).
Once established, as new institutionalist theories would suggest, these early regimes developed staying power and distributional coalitions with a vested interest in maintaining existing benefits and protections (Rudra, 2007, p. 392). Haggard and Kaufman (2008, p. 2) suggest that the 'welfare legacies' generated by these early choices influenced the politics and economics of social policy as countries in Latin America, East Asia and Eastern Europe democratised. Prior policies had created both constituencies of support for existing provisions, as well as having fiscal implications. There were strong pressures for retrenchment in Latin America in the 1990s where existing policies for formal sector employees placed a heavy burden on the exchequer, compared to East Asia where new democratic governments had relatively few welfare commitments and thus greater room to expand (Haggard & Kaufman, 2008).

Left-wing political organisation, the threat of revolt and social policy
Many accounts of the politics of social policy, including the literature on Latin America discussed above, explicitly or implicitly combine the varieties of welfare capitalism approach (recognising the existence of broad clusters of social provision) with a 'power resources' approach. They do so by suggesting that patterns of political organisationor approaches to the political incorporation of urban working classes and other marginalised groupsproduce different social policy outcomes within similar economic contexts.
The 'power resources' approach has dominated explanations of welfare state policies in advanced industrial economies. This approach sees the size and structure of the welfare state as shaped by the historical strength of the political left, along with its alliances with middle classes (Esping-Andersen, 1989;Huber & Stephens, 2001;Iversen & Stephens, 2008;Korpi, 1983). There is strong evidence in advanced industrialised countries that greater left-wing control of government produces more redistribution, and that the presence of unions and coordinated wage bargaining reduces wage inequalities or increases wage compression (see summary in Iversen & Stephens, 2008). Furthermore, once created, welfare states create new interest groups of 'welfare state clients' who can organise against retrenchment of their benefits (Korpi & Palme, 2003).
In addition to partisanship and the role of left-wing governments, regime type has been another important variable in considering how demands on the state are organised and filtered. The wider literature suggests that democracies are likely to invest more in social provision because politicians use social policies to attract votes (Lake & Baum, 2001;Mares & Carnes, 2009, p. 96). Previous studies have shown strong associations between democratisation and social spending in Latin America, particularly spending on health and education which reach larger parts of the population than social security for formal sector workers (Kaufman & Segura-Ubiergo, 2001).
But within developing country democracies, the importance of left-wing parties and organised labour on changes to social welfare are debated. The presence of large, weakly organised informal sectors and weaker social democratic tendencies among political parties present a different context to that of welfare state expansion in Western Europe (Avelino, Brown, & Hunter, 2005). This may be why research across Latin American countries between 1970 and 2000 found partisanship to have little effect on variation in overall social welfare expenditure (Huber, Mustillo, & Stephens, 2008). Some studies report, however, that countries with a strong record of democracy, and stronger left influence in the legislature have a better record in reducing poverty (Pribble, Huber, & Stephens, 2009), perhaps because while they do not influence overall levels of expenditure, left-wing governments do influence the overall content of expenditure, for instance favouring more progressive measures such as non-contributory, conditional transfers, school feeding programmes and preventive health care (Huber et al., 2008).
In the democratic settings of Brazil, India and South Africa, the extent to which new social programmes were driven by electoral concernsor had electoral feedback effects for the incumbent governments who introduced or administered themare subject to vigorous debate. 11 In Brazil, there was a good degree of continuity and policy consensus around the combination of macro-economic stability and social inclusion between the PSDB and PT administrations led by President Lula. The existence of a cross-partisan consensus or 'dominant belief' in social inclusion with fiscal responsibility is explored more closely by Carlos Pereira and Frederico Bertholini in this special issue. In India, by contrast, while the Congress-led UPA government was led by politicians with a more social democratic outlook, it is an amorphous party that lacks organisational or ideological coherence. This was clear, for instance, in the prolonged debates over the National Food Security Act which failed to find strong or vocal support within the Congress Party (Tillin, unpublished manuscript).
Drawing too sharp a binary distinction between democracies and nondemocracies may not be productive. Many of the welfare gains within democratising states build on social policies initiated by earlier non-democratic regimes (going back to Bismarck's social insurance). As Mares and Carnes show, most old age, disability and sickness insurance policies adopted in developing countries were first adopted under authoritarian regimes (Mares & Carnes, 2009, p. 97). Furthermore, authoritarian regimes have adopted policies that are very limited in reach, as well as near universal, reflecting the variety of political strategies that authoritarian regimes deploy to retain power (Mares & Carnes, 2009). In authoritarian China, leaders did not have electoral concerns, but they did nonetheless monitor the many localised urban and rural protests, and were concerned about the potential for 'political instability' if poverty and inequality were not tackled (Duckett & Wang, 2015). Despite this, the 'threat of rebellion' thesis, often held to explain why authoritarian regimes extend public goods provision, is not sufficient to explain why the Chinese party-state expanded social provision in rural areas in the early twenty-first century, not least because rural protest had been a growing problem for two decades and most outside observers had not seen it as threatening to the regime (Duckett & Wang, 2017).
In any case, across much of the global south, the pervasive influence of clientelism has reinforced a view that electoral dynamics do not necessarily drive improved social provision. Because of weaknesses in implementation or state capacity, politicians' promises to deliver new social programmes or public goods may lack credibility, and they may therefore have to rely on direct clientelist ties, rather than policy pledges, to win elections, (Keefer & Vlaicu, 2008). There may also be high barriers to political organisationand thus sustained political pressureamong poor, predominantly informal sector workers. Meanwhile trade unions representing formal sector workers, have frequently not been champions for wider pro-poor or pro-worker policies (Agarwala, 2013; see also Friedman & Groenmeyer, 2016). These features of the organisational landscape of electoral politics mean that individual leaders, wider policy communities and the role of other actors such as courts and non-governmental organisations, all play an important role in putting new policy ideas on the agenda.
There have been other important refinements to 'power resources' theories. For instance, Rothstein, Samanni, and Teorell (2012) argue that citizens' views of the quality of government (as trustworthy, impartial and non-corrupt) is an important influence on whether those citizens' support the expansion of social insurance by the state. Examining OECD countries between 1984-2000, 12 Rothstein et al. (2012) find that where the quality of government is higher, the size and generosity of the welfare state is likely to be largerand working-class mobilisation is likely to be more effective in demanding more generous benefits. They extend Skocpol's insights about the development of social policy in the United States where the corruption and clientelism that characterised the administration of pensions for veterans of the Civil War in the late nineteenth century delegitimised future expansion of the welfare state, helping to explain why the United States developed comparatively limited, targeted and not very redistributive welfare provision (Rothstein et al., 2012;Skocpol, 1992). By contrast, those European countries that developed fully-fledged welfare states in the twentieth century had improved their quality of governance during the preceding century (Rothstein et al., 2012).
The impact of the quality of government on attitudes towards social policy expansion is likely to be a significant factor for understanding the dynamics at play in middle-income countries. The quality of government might have a significant bearing on the extent to which demands for the expansion of anti-poverty programmes are successful. In India, much of the opposition to expanding social protection among vocal middle and upper classes represented in the media (a constituency that has influence on both the Congress Party, and the other main national party, the Bharatiya Janata Party (BJP)) has focused on inefficiencies in public spending and corruption in service delivery (Tillin, unpublished manuscript). A recent survey by Samuels and Zucco (Odilla, 2016) in Brazil identifies around 11 per cent of the electorate as 'anti-PTistas' who oppose the PT but do not identify with any other party and are disillusioned with democracy more generally. These wealthier voters oppose the Bolsa Familia and quotas for black people in public universities introduced by the PT, although they favour reducing social inequality, and would tolerate paying higher taxes to fund health and education. This may reflect a backlash against a programme perceived as the instrument through which a corrupt government bought the support of the poor. It may also reflect what Rothstein et al. (2012, p. 10) describe as a concern for procedural justice: a citizen may support the policy goal of universal health care, but still oppose it politically because of a belief that the government is incapable of implementing it in a way that will be procedurally just. 13 Across Brazil, China, India and South Africa in the 2000s, therefore, it is debatable as to whether organised societal pressures from below, and specifically via left-wing political organisations, served as the critical factor in propelling governments to pay greater attention to social policy. But left-oriented governments and individual political leadersand the wider policy communities that influence themhave had an important impact on the design of policy, as various contributions to this special issue explore (Friedman and Maoirano, Manor and Duckett).

Institutions, policy legacies and layering
Both 'varieties of welfare capitalism' and 'power resources' approaches suggest that initial policy choices create longer term path dependencies that shape later social policy decisions. There is also a rich body of historical institutionalist literature focused on welfare state development overall, and on the trajectories of specific areas of social policy (see especially Pierson, 1996Pierson, , 2001bPierson, , 2004Steinmo & Thelen, 1992). Historical institutionalists look both to the formal 'rules of the game' (including electoral systems, party system structures, political institutions, inter/intra-governmental relations, formal structures of representation for trade unions) as well as more informal institutions and conventions about policy-making. They are particularly interested in the way that institutions structure political battles and, as a result, may shape their outcomes (Steinmo & Thelen, 1992, p. 3). Most historical institutionalists are interested in policy change, although they have been criticised for failing to produce good accounts of when and why change from established paths take place. Historical institutionalist accounts often fall back on 'critical junctures', or contingent, exogenous shocks to a political system, to account for major change (Capoccia & Kelemen, 2007;Collier & Collier, 2002).
In terms of formal institutions, several accounts have looked at the effects of the dispersal or fragmentation of power implied by different institutional structures for welfare states. These studies have mostly focused on advanced industrial countries 14 and suggest that welfare state expansion is likely to be more challenging in fragmented institutional settings such as federal systems which introduce more veto players and fragment political interests (Bonoli, 2001;Pierson, 1995;Swank, 2001). And yet, once established, for similar reasons, federal systems may make welfare state retrenchment more difficult (Obinger, Leibfried, & Castles, 2005;Pierson, 2001b). The impact of federalism on social policy formulation and implementation is examined in the comparative study of Brazil and India in this special issue.
Much of the institutionalist literature developed at a time in the 1990s and 2000s when welfare state retrenchment in OECD countries was a major source of debate. The consensus in much of this writing was that policy retrenchment, in practice, was less visible than the rhetoric of retrenchment suggested because of the political stickiness of earlier policy choices that generated social and political constituencies in favour of policy maintenance. While cuts may occur in times of fiscal constraint, the underlying social policy frameworks were deemed more secure (a good summary is given in Hacker, 2004). Yet as Hacker (2004) and others argued, by focusing on the absence of changes to formal policy rules, institutionalist analyses of welfare state retrenchment often missed the subtler point that changes within the bounds of existing policies had made welfare policies less well-equipped to protect citizens from the wider range of social risks they faced or that changes within existing policy frameworks altered them fundamentally from within. Hacker (2004, p. 244) notes that: 'actors who wish to change popular and embedded institutions in political environments that militate against authoritative reform may find it prudent not to attack such institutions directly'. There is a parallel with what Jenkins (1999), discussing the politics of economic reforms in India, described as reform by 'stealth'.
Another means by which policy change can be achieved while avoiding a direct challenge to the status quo is through policy 'layering' whereby new policies are introduced without dismantling earlier entitlements. Hacker (2004) uses this frame to understand Republican efforts in the United States from the 1980s to the 2000s to introduce tax breaks for private pensions that competed with public programmes. In the cases of Brazil, India and South Africa, policy layering was central to the phase of social policy expansion that took place from the late 1990s. None of these countries (unlike others in their regions) had seen a roll-back or significant reform to the provisions already entrenched for formal sector, salaried workers. 15 Rather new provisions for those working in the informal sector or previously disenfranchised, were layered on top of existing provisions. Barrientos suggests that the combination of labour market informalisation and the introduction of new forms of social assistance based on socio-economic position not occupational status, have given a new character to Latin American welfare regimes combining 'a down-grading of protection for workers in formal employment and the emergence of weakly institutionalised social assistance programmes' (Barrientos 2009, p. 103). Even in China, which has seen a significant reform of Mao-era social provisions, there were continuities in urban formal sector worker provisions on top of which programmes for informal and rural workers were layered. Thus a combination of policy layering, and wider economic change, have produced changes to the substance of social policy provision without always implying a frontal assault on existing policy frameworks.

Political leaders and policy entrepreneurs
The observation that substantial change can be achieved within existing institutions and policy frameworks underlines the importance of political leaders and policy entrepreneurs, although these figures are too rarely the focus of analyses of the politics of social policy. As Grindle (2004) notes, writing about the odds stacked against reforms to education policy in Latin America in the 1990s, formal political economy is often interested in the way that winners and losers mobilise around proposed policy changes. In this model, losers are clearly aware of their potential losses and quick to oppose change, while winners are much less likely to benefit in the short term or be aware of long-term gains. Losers have incentives to protect the status quo; winners lack clear incentives to organise for change and therefore face difficult problems of collective action. Further, reform is politically difficult because electorally sensitive politicians have incentives to postpone it, given imbalances between the power of winners and losers … The model provides a powerful and intuitive explanation for the existence of opposition to changes that might be socially beneficial. But reform outcomes are not a simple matter of weighing the interests that support and oppose change. (Grindle, 2004, pp. 11, 12, emphasis added) The role of leaders and policy entrepreneurs can be critical in understanding how the interests of different groups can sometimes be negotiated, or sidestepped, in order to achieve policy change. Research on other policy arenasnotably foreign policyhas long included analysis of leaders and how their beliefs, perceptions, motives, decision style and interpersonal style affect foreign policy decisions (George, 1969;Hermann, 1980). And studies of political survival have shown leaders to use public policy and especially public expenditures to prolong their time in power (Ames, 1987;Mesquita, Morrow, Siverson, & Smith, 2002). But research on social policy has tended to ignore actors, especially individual leaders, when looking at policy change. Some work on political parties does discuss particular leaders, and some studies do mention leaders as influential key actors in their 'thick' descriptions, but they do not theorise about leaders' roles.
As Merilee Grindle (2004, pp. 12, 14) argues, we need to understand the behaviour of reformerspolicy entrepreneurswho 'commit themselves to change despite the odds', especially in developing countries where institutions are 'less strong and durable' and there may be more opportunities for manoeuvring around their constraints. Focusing on the efforts of President Cardoso in Brazil, Chief Minister of the central Indian state of Madhya Pradesh, Digvijay Singh and President of Uganda, Yoweri Museveni, Melo et al. (2012) argue that all three are examples of politicians who devised shrewd political strategies to manoeuvre within existing institutions and political constraints to enact poverty reduction policies. They did so by building trust in their governments and by presenting anti-poverty policies as something other than zero-sum games. They often drew poor people into the political and policy process so that pressure from below reinforced their efforts from above. The role of particular individuals and political leaders in the process of negotiating new anti-poverty policies stands out in many of our analyses for this project in all four countries. The article by Manor and Duckett in this special issue analyses their role directly.

Papers in the special issue
The first paper of the special issue by Diego Maiorano and James Manor sets the achievements of Brazil, China, India and South Africa in context by considering the burden of poverty and inequality faced by these countries, their economic resources, and the headway made by governments towards the reduction of poverty and inequality in this era of social policy expansion. The authors find that all four governments made progress in reducing income poverty, but performed differently on other facets of human development. All countries faced challenges in ensuring that government initiatives reached the poorest of the poor. While some of the poorest were reached in China, a larger proportion were reached in South Africa, and yet larger proportions in Brazil and India. The most difficult challenge faced by all four countries was that of reducing inequality. Here, only Brazil made significant headway. Lastly, the paper considers the extent to which government initiatives have made an impact on poverty more broadly defined, and specifically on interventions that have invested in the political capacities of the poor. Over the long run developing poor people's political capacities may improve democratic engagement but also enable material gains.
In the second paper, James Manor and Jane Duckett argue that top leaders in Brazil, China, India and South Africa have played significant roles in supporting and shaping social policies to tackle poverty and inequality. For all of them, whether democratic or authoritarian, such policies became politically expedient in the twenty-first century not only because they sought to reduce protest or opposition and generate support but also because leaders themselves seemed to accept arguments about the economic development benefits of more even social policy provision. Whatever their motivations, leaders were often instrumental in pushing forward policies and overcoming path dependencies. Manor and Duckett thus seek to bring back into social policy analysis these important but overlooked actors and to encourage others to incorporate leaders into theories of welfare state development.
Louise Tillin and Anthony Pereira's paper contributes to institutionalist research with a study of the effects of federalism and multi-level elections for social policy in Brazil and India. While work on advanced industrialised countries has shown that federalism can create multiple veto points that make welfare state expansion difficult, Tillin and Pereira argue that federalism must be analysed in conjunction with party systems. In Brazil, social policy expansion occurred in an era of fiscal recentralisation and greater party system nationalisation. This helped to facilitate wide territorial coverage of new programmes. By contrast, within India's more decentralised federation, while the central government also introduced major new social policies, state governments were able to take the credit for them. State governments remained powerful players in implementing new national social policies meaning thatunlike Brazilthe outcomes of social assistance programmes have varied substantially across regions.
The paper by Steven Friedman and Diego Maiorano focuses on the courtsinstitutions often neglected in work on social policy and welfare state expansion. They argue that courts in India and South Africa were able to successfully influence social policies not by challenging government but by supporting collective action by the poor and their civil society allies. The courts acted differently in India and South Africa with the former prescribing policy outcomes and the latter preferring to introduce procedures. The authors find that the latter approach tended to produce more sustainable policies and strengthened democratic decision-making. Despite these differences, however, and their caution that the courts are among many actors in the social policy arena, Friedman and Maiorano demonstrate their hitherto neglected role, and one that deserves greater research attention.
Lastly, Carlos Pereira and Frederico Bertholini analyse data from a survey of Brazilian legislators' preferences over budgetary spending on social policies. They find that legislators consistently supported increased spending on social policy, regardless of their ideological position on the rightleft political spectrum. On this basis, they argue that there is a widespread and dominant belief in the need for greater social inclusion in Brazil, though this is moderated by perceived limits to raising taxation. This article reveals the ways in which ideas, formed by past political experience, and distinct from ideological differences, can play a role in extending and sustaining welfare states in the global south.
The four countries focused on in this issue are important in their regions and significant beyond themoften referred to over recent decades as among the world's 'rising powers'. Each has very different political cultures, systems and issues, as well as very different patterns of inequality and poverty. Understanding the politics behind the expansion of their social policies not only contributes to the frameworks used to analyse social policy expansion but may also influence social policy elsewhere in the global south. In particular, the papers draw attention to the importance of actors and institutions previously neglected in studies of social policy in lower and middle-income countries. In terms of actors, we move beyond left-wing parties and organised labour movements to include the role of leaders and their political calculations. We also consider the role of ideasor 'dominant beliefs'that sometimes span party lines. In terms of institutions, we look beyond electoral dynamics to focus on federalism and multi-level political dynamics in shaping social policy, as well as the role of the courts in supporting and sometimes enabling pro-poor policy. In addition, in this essay, we have highlighted other factors that have influenced the overall development of social policy in these countries. Longer term policy legacies have been important in shaping what is conceived as politically possible. Furthermore, citizens' perceptions of the quality of governance more generally is very likely to play a role in mediating the likelihood that advocates of more propoor investment succeed in achieving their demands. Most of all, we have shown that while closer global economic integration provides the context in which all of these countries expanded their social provision, it is not possible to explain the shape, reach and effectiveness of social policy without focusing on political dynamics. In finding similarities as well as differences across these major emerging economies, we hope to encourage comparative research to go beyond exploring regional clusters of economically similar countries to develop more ambitious analytical frameworks suited to a wider range of countries. Notes 1. Barrientos (2013, p. 13) defines 'social assistance' policies as direct transfer programmes that target poverty and vulnerability, whereas 'social protection' would typically be understood to include social insurance, social assistance and labour market policies. Under the umbrella of social assistance fall both direct income transfer programmes, as well as what Barrientos calls transfers combined with asset accumulation, designed to support human, financial and physical asset accumulation. 2. This view lies within a longer intellectual tradition of 'embedded liberalism' going back to Karl Polanyi, and with seminal texts by Cameron, Stephens, Katzenstein and Ruggie arguing that the expansion of the post-Second World War welfare state enabled governments to reduce the risks associated with greater international openness. See Swank 2005 for a succinct summary. Studies among non-OECD countries have found contradictory results about the impact of greater trade openness on social expenditure (Avelino et al., 2005;Kaufman & Segura-Ubiergo, 2001). 3. This is true both of conditional cash transfers which are designed to boost school and clinic attendance, as well as policies such as the Mahatma Gandhi National Rural Employment Guarantee Scheme in India. 4. In China, policies designed to reduce rural poverty were linked the desire to boost domestic consumption and reduce reliance on export-led growth (rather than to increase social investment) and they were also inspired by concerns that rising inequality might endanger 'political stability'which makes the Chinese case somewhat different from the other three under examination, as we shall discuss below. 5. The role of the private sector within middle-income countries is also very important, although this is not a principal focus of the papers in this special issue. 6. Papers by the South Africa team have been published in a special issue of Transformation: Critical Perspectives on Southern Africa, Vol. 91 (2016), and are forthcoming by members of the other country teams. 7. Note however that it built on the macro-economic stabilisation achieved under President Cardoso's preceding regime, and earlier experiments with conditional cash transfers (Melo et al., 2012). 8. These reforms built on experimentation and signals of a change in policy direction that preceded the Hu-Wen administration thus we should be careful to posit too sharp a discontinuity between particular leaders (see Duckett & Wang, 2017). 9. The programme was initially targeted toward the poorest districts, but later expanded to cover all rural households. 10. The literature on East Asian welfare states has usually excluded China, which provided social protectionpensions, medical coveras well as housing and full employment to urban dwellers before the market reform period. 11. The debate is especially vigorous with regards to Bolsa Familia in Brazil. Scholars such as Hunter and Power (2007) and Zucco (2013) have argued that incumbents have seen electoral benefits from Bolsa Familia, but this has also been disputed (Bohn, 2011;Corrêa, 2015;Corrêa & Cheibub, 2016). 12. Although they state that they expect to see an even stronger relationship between quality of government and welfare state expansion in non-OECD countries where clientelism and corruption are greater problems. 13. This builds on earlier normative work by Rothstein (1998) who argues that support for the welfare state in Northern Europe follows a moral logic dependent on citizens' appraisal of the substantive justice (who are the deserving poor), procedural justice (can government deliver) and fairness of social policy. He argues that there is likely to be more disagreement about these issues in 'liberal' welfare states, where benefits are residual and targeted, than in universal welfare states. 14. Though there are exceptions such as Duckett (2011). 15. In Latin America, structural adjustment in the 1980s led to the weakening of employment protection and social insurance models with a shift towards individual savings instruments offered by private providers (see, for instance, Madrid, 2003). But countries varied in the extent of reforms from the radical neoliberal reforms of the Pinochet dictatorship in Chile to the more moderate case of Brazil where greater opposition from Congress (especially from the PT) and unions helped to block reforms (Huber & Stephens, 2012, p. 157).

Disclosure statement
No potential conflict of interest was reported by the authors.