The significance of political leaders for social policy expansion in Brazil, China, India and South Africa

ABSTRACT This paper examines the roles that political leaders have played in the formulation and implementation of government initiatives to tackle poverty and inequality in Brazil, China, India and South Africa since about 2002. While research on social policy and welfare expansion in the industrialised world has largely ignored leaders, we stress the importance of politics and political agency, since political leaders often exercised decisive influence even if their decisions to prioritise certain issues and to adopt certain strategies depend on the convergence of other factors. We examine their management of tensions and opposition within their governments and their political systems – as well as what we can discern of their motivations and political calculations. We consider the impact on government initiatives both of their adroit machinations and, at times, of their ineptitude.

The aim of the research project from which this paper is derived was to analyse the political and policy processes that from around 2002 led governments in Brazil, China, India and South Africa to intensify efforts to tackle poverty and inequalityand their outcomes. Politics was central to our investigationsunavoidably, because it had a potent influence on events. Any assessment of politics forces us to devote much of our attention to senior political leaders because they possess significant power, although the limits on their power also need to be carefully delineated. This chapter focuses on themon politicians at the apex of power, and at times on politicians, bureaucrats and others just below, in the second tier.
We are not seeking here to revive the 'great man' (or 'great woman') theory of history. The leaders in these four countries faced multiple constraints, and had personal limitationsthey were not geniuses. They seldom dominated the political and policy processes. But we argue that they were sufficiently powerful and adept to play key roles. Despite the many differences in their countries' political and socio-economic circumstances, in their precise reasons for tackling poverty and inequality, and in the types of policies that they adopted, leaders in all four countries concluded that efforts to address poverty and inequality were necessarybecause they recognised that poverty was pernicious and that economic growth and globalisation had increased inequality. They also reckoned that efforts to tackle these problems were feasible (politically, administratively and financially). They believed that poverty initiatives would ease destitution and discontents, and enhance their popularity and legitimacyso that they would serve their political interests. They therefore took significant risks by investing political and financial capital in innovative policies and programmes. Some policies were more effective than others, and some leaders were more adroit than others. But they all had a substantial impact and succeeded in redistributing wealth, reducing poverty and reaching the poorest of the poor. (Reducing inequality was more difficultsee the paper on 'Outcomes' in this collection.) By examining contrasts and commonalities across the four countries, this comparative analysis seeks, in part, to broaden the perspectives of those who focus on just one of these countries (or on any other single country). It shows that certain important features of a researcher's chosen country, which they take for granted, are in fact distinctive, so that they merit greater attention. It can enhance their understanding of how their political system works, about the balance of forces within it, and about their own habits of mind. Comparisons with other cases reveal policy options and political choices that were rejected or never considered elsewhere. This enables observers of any single country to see how things might have been differentwhich in turn can broaden their view of the range of potential actions and cast what did happen in a fresh light. But (to reiterate), despite the many differences, political leaders in these countries had important things in common. In all four casesin different systems and settings, where different leaders devised different policiesthey evidently thought that poverty and inequality had to be addressed for political if not social reasons, and they demonstrated that it was possible by a diversity of means to make a significant impact. These leaders cannot be ignored.
Although the study of leaders and leadership has declined in political science (but see Ahlquist & Levi, 2011;Jervis, 2013), there is a stronger tradition of examining the role of political leaders in foreign policy and international relations than in social policy. In international relations, work soon evolved from a focus on how leaders' backgrounds, social origins, education and training might shed light on their actions to a focus on their orientations, beliefs and values (see George, 1969;Holsti & Rosenau, 1998) or on their personalities and decision-making styles (e.g. Etheridge, 1978). 1 Some of this work acknowledges, as we argue here, that while it is hard to deny the influence of individual leaders, at the same time, there is extensive evidence of the environmental constraints on their actions.
Leaders have, however, been neglected in research on social policy and welfare state expansion. That research has focused instead on the role of economic development and modernisation (e.g. Wilensky, 1975), political parties, class support and coalitions (e.g. Korpi, 2006), and institutions (see Amenta, Bonastia, & Caren, 2001, and the Introduction to this special issue). As Daniel Béland (2005) has noted in work focusing on the role of ideas, dominant explanations in recent years have been in the historical institutionalist tradition. While our study does not claim to be comprehensive or to systematically evaluate the role of leaders across all social policy areas in our four countries, it highlights the neglected role that leaders have played in the early twenty-first-century expansion of policies to tackle poverty and inequality in these four countries.
The utility of politics, and the issues of path dependency and policy innovation By emphasising politics and political leaders, we differ also from policy analysts with a technocratic turn of mind. They regard politics as an unhelpful force that taints elegant policy blueprints with extraneous concerns, irrationality, untidiness and unpredictability. In their view, the formulation and implementation of policies need to be insulated from politics and politicians, from which little good can come.
They are mistaken on two counts. First, even when technocrats are in command, politicsthat is, the interplay of interests, actors and ideas in the pursuit of poweris always present. Even strategies that try to insulate policy processes are saturated by politics. Second, we have abundant evidencefrom documents, and field research including extensive interviews with practitioners, some of whom were members of our research teamto demonstrate that politicians often make constructive, rational contributions to the design and implementation of policies. They do so mainly because they believe that poverty programmes serve their political interests. 2 They operate in ways that are sensitive and responsive to the leverage, preferences and needs of diverse interests: supporters of their policies, sceptical allies who will nonetheless support them, opponents of their policiesand poor, marginalised groups, even though they may have little political influence. These are not extraneous concerns. The political leaders and processes that we have examined therefore tend strongly to contribute to another, more utilitarian rationalityand to the sustainability of policies and to social justice.
Some analysts also argue that little new can be expected of politicians and other policy-makers because they tend strongly to be locked into pre-existing patterns of behaviour and unwritten agreements between ruling elites and various social forces. Those analysts stress the iron grip of 'path dependency' and the immense difficulties that stand in the way of policy innovation. The evidence from our study raises serious doubts about such arguments. 3 In three of our four countries -China, India and South Africathe leaders who intensified efforts to address poverty and inequality had to overcome unhelpful political or policy legaciesinheritances from their immediate (and earlier) predecessors. They sometimes had to tackle reluctance among politicians and bureaucrats within their governmentsespecially in South Africa and China where ruling parties did not change, so that fewer new appointees were inserted into senior posts there than after elections in Brazil in 2002 and India in 2004. They also had to challenge, to varying degrees, pre-existing political understandings that favoured prosperous interests who were sometimes in league with complacent bureaucrats. In all four governments, proposed poverty initiatives met with opposition from senior figures, especially in finance ministries, who (like their counterparts across the world) worried about the fiscal implicationsand in some cases, from ministries where initiatives would threaten the prerogatives and interests of bureaucrats. In other words, most of the leaders we assessed had to break free of 'path dependency', and to innovateand they managed to achieve this.
Only in Brazil did presidents Luiz Inacio Lula da Silva, or 'Lula', between 2003 and 2011, and Dilma Rousseff thereafter, inherit a tradition of propoor policies that had been put in place by Lula's predecessor, Fernando Henrique Cardoso (the leader of a progressive rival party), between 1994 and 2002 (Melo, Ng'ethe, & Manor, 2012). But even Lula's government innovated by introducing new initiatives that went beyond Cardoso's. In the other three cases, leaders had to struggle against far less helpful inheritances. In China, Hu Jintao and Wen Jiabao took power in 2002-2003 after more than a decade in which Jiang Zemin and Zhu Rongji's administration had emphasised economic growth with an urban bias. That led to increasing economic inequality, a relative neglect of rural areas and of the plight of poor people (rural and urban), and mounting collective protests over injustices (see Duckett & Wang, 2017). Hu and Wen countered with an array of new initiatives that extended programmes to disadvantaged groups and lifted the onerous tax burden on rural dwellers. 4 These innovations were intended both to spur economic growth and to promote a 'harmonious society'one of the key slogans of their decade in powerby reducing distress, discontent and worrying levels of instability.
In India, the United Progressive Alliance (UPA) led by the Congress Party introduced numerous programmes to benefit poorer groups who had been largely ignored both by its main rival, the Bharatiya Janata Party (BJP), which had led a ruling coalition between 1998 and 2004, and by the Congress itself in earlier periods. The key actors were Sonia Gandhi, who declined to lead the government but headed a National Advisory Council (NAC) whose members (see below) crafted several innovative programmes, and Prime Minister Manmohan Singh who was hesitant about some programmes but whose staff contained progressives who developed some others. These innovations could be generously funded, thanks to a marked surge in government revenues that began in 2003 and continues to this day. The UPA's election victory in 2004 was erroneously attributed in the media to a revolt among the rural poor against the policies of the previous BJP-led government. But the UPA set out to make that myth a reality at the next election. It had some success in achieving that, although that was also over-stated in media assessments in 2009 (Manor, 2011). Its poverty initiatives, only some of which succeeded, added up to a decidedly innovative approach.
In South Africa, the African National Congress (ANC) had introduced certain policies to benefit poor people after achieving dominance in the first fully democratic election in 1994dominance that was not seriously challenged until the municipal elections of 2016. But its initial, strongly pro-poor development strategythe 'Reconstruction and Development Programme' (RDP)was sidelined in 1996 in favour of a more market-and growth-oriented approach, 'Growth, Employment and Redistribution' (GEAR), which prioritised a reduction in the fiscal deficit, tackling inflation, the liberalisation of capital flows and reductions in government expenditure. Resentments against GEAR's de-emphasis on poverty programmes developed within (and beyond) the ANC, under the highly centralised leadership of Thabo Mbeki (President, 1999(President, -2008 (Seekings & Nattrass, 2006). Few major poverty initiatives emerged in that periodalthough one which eventually had a substantial impact (the child support grant, later augmented by grants to other vulnerable groups) came into being thanks to the determination of one entrepreneurial minister (with help from intellectuals and civil society) to go against the grain of the ruling order (Groenmeyer, 2016). When Jacob Zuma succeeded Mbeki, greater efforts to tackle poverty and inequality were madeless as a result of pressure from Zuma than because his relaxation of tight control from the top permitted other senior figures to pursue them. Cumulatively, their efforts amounted to a marked change from the inheritance not just from the apartheid era but also from Mbeki's ANC.
All four of these casesespecially China, India and South Africademonstrate that political agency and policy entrepreneurship, the actions of leading political actors, have the potential to break with 'path dependency' in innovative ways. They achieved this in different ways. Leaders often presided over groups of policy entrepreneurs who drew on ideas and experiences at lower levelswithin sub-national governments that had been controlled by their partiesor within their parties' forums, which were sometimes open to ideas from progressive civil society organisations and scholars. For example, in Brazil, municipal and state governments headed by Lula's Worker's Party had experimented with programmes from which new initiatives were generated. In China, the party-state used experimental methods to test policy innovations at lower levels (Heilmann, 2008). In India, prolonged encounters in Rajasthan state between civil society organisations and state governments (some of which were Congress-led) generated ideas that were used by key members of Sonia Gandhi's NAC to design the crucial Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA).
At other times (especially in India and South Africa where ruling parties had developed fewer programmatic ideas), leaders innovated by commissioning policy entrepreneurswithin and outside officialdomto devise new initiatives. An official in India's Prime Minister's Office (PMO) convened groups of progressives who specialised in the problems of urban slum dwellers and poor forest dwellers to help craft, respectively, the Jawaharlal Nehru National Urban Renewal Mission (JNNURM) and the Forest Rights Act. In South Africa, officials in government ministries accepted a proposal from closely associated civil society activists and scholars for an adaptation of India's MGNREGA.
In other cases, politicians became policy entrepreneurs themselvesmost notably when a minister in Mbeki's government in South Africa worked with progressive scholars and civil society organisations to launch the Child Support Act. That entailed cash transfers which, when later extended to other disadvantaged groups, had a major impact.
Leaders in the four countries seldom borrowed ideas from the other three. South Africa was the main exception here. The Child Support Grant was devised with Brazil's cash transfers in mind, and (as noted) they also used India's MGNREGA as a model.

Situating leaders within political systems: coherence and concentrations of power within governments
No government is monolithic. Multiple leaders and forces operate within them, and external groups and power centres sometimes play roles in the formulation and implementation of policies. So, we must consider the coherence of these governments and the array of actors that influenced these processes. (Complexities multiply when we move down to middle and lower levels in these political systems, but since they are assessed in other papers in this collection, we mainly focus here on the upper reaches). In some cases, senior leaders were solidly united behind poverty programmes, but in others, powerful figures were lukewarm or opposedso that contestation and occasional subterfuge was required to launch and sustain an initiative.
Leaders atop these four systems differed from each other in the powers that they and their parties possessedrelative to other leaders and forces. In China, they led a party with no powerful rivals that permeates state structures to the point where scholars commonly refer to a 'party-state'. Top leadersthe Politburo Standing Committee of usually five to nine peopleoperate on the principle of collective leadership and rarely reveal any disunity, but by virtue of their positions as head of the Party and government, respectively, the General Secretary (Hu Jintao) and Premier (Wen Jiabao) were clearly pre-eminent. 5 In South Africa, the ANC competes with other parties, but it was plainly dominant within the political system until mid-2016although as we see, while Thabo Mbeki was president (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008), proponents of poverty initiatives received limited support from him. In India, the Congress Party exercised great power within the ruling coalition that it led between 2004 and 2014, but it was allied to many (mainly regional) parties, some of which had only limited interest in poverty initiatives. As we shall see, Prime Minister Manmohan Singh and some of his key colleagues were sceptical and eventually even hostile to certain poverty programmes that Congress leader Sonia Gandhi supported. In Brazil, the powerful presidency was held by leaders of the Workers' Party (Partido dos Trabalhadores) -Lula between 2003 and 2011, and Dilma Rousseff thereafter. But Brazil's political system divides power between the executive presidency and the legislature or Congress where the Workers' Party has never held anything close to a majority. That system thus lacks the coherence of India's Westminster-style parliamentary system in which the executive and legislature are fusednever mind the South African system in which the dominance of the ANC after 1994 trumped the formal division of powers between executive and legislature, or China's one-party system in which party and state are fused.
However, in Brazil, another informal reality prevented the separation of powers from imposing serious restraints upon the two presidents as they sought to tackle poverty and inequality. This was a broad popular consensus that had existed since the late 1980s in support of government efforts to pursue these goals. So in Brazil, inter-party competition (until 2016) led less to attempts by rival parties to restrain Workers' Party presidents than to experiments by rival parties, in lower level arenas that they controlled, to devise initiatives that would address poverty and inequality. This crucial informalitythe broad consensusleft Lula and Dilma Rousseff less constrained than their Indian counterparts in their creation and implementation of poverty initiatives.
We will shortly examine several key themes: leaders' management of conflicts within their governments; their motivations and calculations; the political gains that they made by tackling poverty and inequality; their approaches to the problems of corruption and clientelism; and their acuity or ineptitude. But to prepare the ground for those discussions, let us first consider how leaders' roles changed as the political and policy processes moved through successive stages.

Leaders' changing roles during three successive phases
It is necessary to analyse three phases in the political/policy process: initiatives' emergence or origins, and then their formulation and their implementation. (Their outcomes are discussed in a separate paper in this collection.) Senior leaders' roles changed as they moved through these three phases. When we consider initiatives' origins and formulation, we find that they had great influence. But implementation mostly occurred at lower levels that were largely out of sight and beyond their reach.
Readers will detect a shortage of comments on China here. That is inevitable since the Chinese political system is much less transparent than the other three. Especially in the Hu Jintao and Wen Jiabao era, the CCP strove to portray its top leadership as acting collectively and hid evidence on the roles played by individuals. Journalists sometimes suggest that their professional experiences, which left them with a patent awareness of rural poverty, helped to persuade Hu and Wen to develop new government initiatives. Certainly, Hu Jintao had been Party Secretary of Guizhou, one of China's poorest, least industrialised interior provinces, while Wen Jiabao had been Vice-premier responsible for rural affairs under the previous leadership, where the farmers' burden (of fees and taxes) had been closely monitored for the protests it catalysed (Duckett & Wang, 2017). But Hu and Wen were careful to portray party-state policy decisions as collectively produced and to limit any suggestion of their personal influence. Moreover, although there was a clear shift in overall policy direction towards providing for rural dwellers under their leadership, there is also evidence that Jiang Zemin had begun that shift from 2001. We are again short of details about their roles during implementation, although it is apparent thatlike their counterparts in Brazil, India and South Africathey had only limited involvement in it.
Government initiatives had complex origins. They emerged in part from pressures applied by organised interests upon governments to address poverty and inequalitysometimes buttressed by public advocates. Governments also faced pressure from within from activists who urged action in order to attract votes (the main emphasis in Brazil and India) or to damp down instability (in South Africa and China). Senior leaders sometimes paid attention to promising experiments with poverty programmes in other countries or, more often, at lower levels within their own systems. Chinese leaders often drew upon the findings from trial runs of policy innovations in certain localities or regions that were tested over extended periodsa practice that took more time than leaders elsewhere had available. 6 But some Chinese policy initiatives that we analysed were not based on extended prior experiments.
In some cases, socialist or progressive ideals long espoused by ruling parties played a partalthough pragmatic calculations about the political utility of poverty programmes predominated outside Brazil. In South Africa and India, ruling parties' time-honoured rhetorical commitments to socialist or social democratic ideals had not previously been translated into much robust action. And the same could be said of China and the ruling party's communist ideals. But that rhetoric, in South Africa and Brazil at least, inspired popular expectations of government actionwithin and beyond ruling parties. Only in Brazil did a broad consensus exist in favour of efforts to tackle inequality, and with it of course, poverty.
In all four countries, decisions to respond to these varied 'origins' were mainly made by senior leaders. Even in China, where politics at the top are opaque, the shift to tackling rural poverty and inequality has been attributed to Hu and Wen grappling with their socialist inheritances (Holbig, 2009). In Brazil, Lula's government after 2002 set about demonstrating that it could deliver on their (and his) long-standing promises to the poor by providing even more than the progressive social democrats had during the preceding eight years. In South Africa, there were clear limits on the fulfilment of the ANC's long-standing progressive rhetoric while Mbeki was presidentalthough some headway was made when one minister devised a cash transfer scheme that acquired enough momentum to persuade the lukewarm president to acquiesce (Groenmeyer, 2016). When Zuma succeeded Mbeki in 2008, his less control-oriented approach enabled progressives among ministers, senior bureaucrats and civil society leaders to go further. In India after 2004, the traditional social democratic posturing of the Congress Party (including Indira Gandhi's promise to 'abolish poverty'), which had seldom been matched by substantive action, was again treated as a point of origin and was given substance for the first time.
The formulation of poverty initiatives in these countries was also strongly influenced by senior leaders, who in nearly all casesespecially outside South Africahelped to drive these processes forward, and had to approve them before they came to fruition. But diverse actors below the very apex of power both within and outside government also made important contributions. The identities of these key actors varied somewhat between countries. In China, for example, lower level influence was limited, and formulation was left largely to ministries (even if they regularly draw on their research offices and other think tanks as well as on some individual researchers). There is, however, evidence that in the case of NRCMS, vice-premiers pushed the policy forward while ensuring that the State Council Office coordinated the complex inter-ministerial process of negotiating and drafting the policy documents (see Duckett, 2017).
In Brazil, Lula and his colleagues drew on intellectuals within and close to the party and in progressive think tanksand to a degree on civil society organisations. Senior leaders in Brazil had, since 1984, been more amenable to civil society than their counterparts in the other three countries. But many of the programmes and processes that they introduced were modelled closely on experiments that the Workers' Party had undertaken when it held power at lower levels of governmentin states and municipalities. Senior bureaucrats (who, when governments change, are fresh appointees to a greater degree than in the other three countries) responded to what they viewed as a new dawn in an entrepreneurial spiritand so did Lula's new ministers (Melo, Pereira, & Pereira, 2016a, 2016b. In India, the central government gave progressive intellectuals and civil society leaders potent roles in the policy-making process for the first timemost notably in the NAC that Sonia Gandhi chaired. A detailed plan for that government's flagship programme, the MGNREGA, emerged swiftly from the NAC, thanks to detailed preparations made in advance by two of its members -Aruna Roy, an eminent civil society leader, and Jean Drèze, a scholarly advocate of poverty reduction who had worked closely with Amartya Sen. Other such advocates from civil society organisations were appointed to the Planning Commission. But the formulation of some new initiatives was driven by progressive bureaucrats, notably one in the PMO -R. Gopalakrishnanwho had co-authored a successful pro-poor education guarantee scheme in a state that the Congress Party had governed. His co-author of that scheme, Amita Sharma , was put in charge of the day-to-day running of the MGNREGA. He facilitated the formulation of the Forest Rights Act and the JNNURM by convening groups of progressive policy advocates, public interest lawyers and civil society leaders to assistand then by lobbying within the PMO for these measures. 7 In South Africa, the process was at times less open to actors from those three categories. But progressive social scientists, backed by pressure from civil society and elements of the ANC itself, played a decisive role in the formulation and enactment of provisions for the Child Support Grant (Friedman 2016a). And a civil society leader who worked closely with key government ministries was crucial to the creation of rural employment programme, loosely modelled on India's MGNREGAagain with strong backing from progressives within the ANC and from academics (Andersson & Alexander, 2016).
The discussion above of the origins and formulation of poverty initiatives focused mainly on the upper reaches of the political and administrative systems. That was appropriate because, while the origins included pressures from below, nearly all the key decisions to proceed were made at or near the apex of power. The same can be said of the formulation phase. But things change when we consider implementation. It mainly occurs at intermediate and lower levelsarenas which leaders at the top found difficult to monitor and to influence.
One major problem for senior leaders (less so in China) was their inability to obtain adequate, reliable information on the implementation of programmes. It came from a diversity of sources, official and non-official. Formal monitoring committees were sometimes constituted, but their findings seldom resulted in action by governments to change things. One particularly forlorn example, which is not untypical, was the panel appointed to oversee India's JNNURM. It found executive agencies so unresponsive that it sought its own dissolution. 8 Government ministries also passed information up to senior leaders, but it was often tailored to show that implementation is proceeding satisfactorily, so that those leaders were left flying half-blind.
Opposition parties offered some critiques of the implementation of government initiatives, but their impact was quite limited. There are no meaningful opposition parties in China, and the South African opposition was usually imperiously ignored. In India, it was (surprisingly) rather unusual for opposition parties to voice sharp criticisms of implementationespecially of the crucial MGNREGAbecause state governments that they controlled in that federal system were themselves engaged in implementing poverty programmes, often enthusiastically. 9 In Brazil, opposition parties were loath to complain too bitterly about poverty initiatives because doing so risked unpopularity by challenging the popular consensus in favour of such programmes.
Information also emerged at times from within ruling parties, mainly in South Africa and India. Critiques sometimes came from progressives. In South Africa, the Congress of South African Trade Unions (which is part of the ANC) voiced concerns, but their complaints had less to do with implementation than with the failure to formulate more poverty programmes. 10 In India, progressive Members of Parliament from the ruling Congress Party helped to ensure that an attempt by executive agencies (which Congress dominated) to dilute the provisions of the MGNREG Act at the formulation stage was thwarted before it became law. Thereafter, they sometimes applied pressure to ensure proper implementation of that and other programmes. Progressive ministerslike some of their counterparts in Brazil, South Africa and perhaps China (where evidence is harder to come by)sought to make the poverty initiatives that they administered work well (Groenmeyer, 2016;Pereira, 2016).
But once implementation began, most critiques from within India's (internally heterogeneous) Congress Party came from conservativesfrom people representing prosperous industrialists or landowners. They voiced doubts about substantial spending on poverty initiatives in general and, more specifically, about the MGNREGA, which allegedly drove up the cost of agricultural labour. That programme was so popular that landowners' critiques were dismissed. But their anxieties about heavy expenditures were shared by Prime Minister Manmohan Singh and officials in the finance ministry, and helped to trigger cuts in the outlay for several poverty programmes, including the MGNREGA, in 2013-2014. 11 (This was a very rare example in these four countries of a reduction in expenditure on a poverty initiative.) Industrialists' lobbying also persuaded Singh to take radical action to undermine the implementation of the Forest Rights Act in that same periodto grant industrialists access to sites within forested areas. In doing so, he overrode both the law and two progressive cabinet ministers. 12 The media also provided informationand at times, misinformation. Those which specialised in financial mattersespecially in South Africa and Indiavoiced scepticism about the cost and the allegedly faulty implementation of programmes. Some of these interventions were wildly erroneous. 13 But the print media in Brazil, India and South Africa also gave space to progressive commentators who supported poverty initiatives, and whose critical comments on implementation were intended to enhance their impact. Even in China, where the media are more constrained, journalists also acted as advocates for the rural poor (Bernstein & Lü, 2000).
Civil society organisations also provided reports based on their research and activities in the fieldin Brazil, South Africa and especially India. (There was precious little of this in China where such organisations are tightly constrained.) These were for the most part supportiveand their criticisms were, again, aimed at improving implementation. Very occasionally, civic organisations on the left published damning analyses of poverty initiatives because they did not meet their unrealistic expectations of perfection, 14 but these counted for little.
The judiciary occasionally intervened to tackle implementation problems. But courts' importance was limited in China where the party-state has vastly more power, and in Brazil where the Workers' Party organisation counted for more. Judges were more active in South Africa, but they stopped short of exerting a major influence on social policy (Friedman, 2016b). Only in India did an activist and unusually intrusive Supreme Court appoint its own commissioners to monitor implementation. 15 When information reached those at the apex of power indicating the need for action to overcome problems, long delays sometimes occurred because complex consultations had first to take place. For example, when civil servants managing the implementation of India's MGNREGA informed the PMO of a serious design fault that had opened the way for corruption near the local level, it took fully 11 months to get cabinet approval for corrective actionand even then, it was watered down by sceptical senior ministers. Bureaucrats made several other quite crucial changes in the implementing ministry, but that occurred beyond the purview of leaders atop the system. 16 Action on information ensued more swiftly in the other three countries, but this remained something of a problem.
One last source both of information on implementation, and of assistance in acting to improve it, needs to be noted, partly because it existed in strength only in Brazil. When elected councils at lower levelsat local and (more importantly in this connection) at intermediate levelswere given significant powers and resources, they could feed information to leaders at higher levels and then act to address implementation problems (Manor, 2013). Brazil's councils, which had been strong since the 1990s, were able to serve as the eyes and ears of politicians and civil servants closer to the apex of power and to tackle problems once those leaders urged action . In a small number of Indian regions, they also had the capacity to do this. But the empowerment of elected councils attracted resistance from legislators and politicians so that in a telling instance, the senior politician who undertook most promising experiment with such generous devolution was eventually forced to disempower elected councils. This deprived him of reliable information at the intermediate level and of the help of elected councils in tackling problems, a change which contributed to his defeat at the next election (Manor, 2004). In China and South Africa, elected councils or other office holders could make no such contribution. The Chinese party-state experimented briefly with elections in townshipsthe lowest administrative levelbut then abandoned them. 17 In South Africa, local councils created in 2000 were provided with such inadequate powers and resources, and were subjected to such onerous technocratic tasks (as a result of elite distrust of the rustics who would be elected as councillors), that they were crippled at birth (Siddle & Koelble, 2012).

Divisions and tensions within governments
Tensions and even outright contestation within ruling parties over poverty programmes were most evident in India and South Africa. In India's parliamentary system, Sonia Gandhi declined ministerial posts, but served as president of the Congress Party and convenor of the multi-party ruling coalition that it ledthe UPA. As we have noted, she inserted Manmohan Singh as Prime Minister (Ruparelia, 2015). By drawing on civil society leaders and public interest lawyers in the development of several poverty programmes, India's UPA provided greater space for multiple actors than did its counterparts in Brazil, China and South Africa.
Sonia Gandhi's partnership with Prime Minister Singh was often complementary. But polite contestation also occurred because he (and the executive agencies that he oversaw) tended to be sceptical about the radical character and financial cost of some proposed initiatives. At times, that scepticism inspired actions which impeded the workings of some poverty initiativesespecially when Singh undermined one of them, the Forest Rights Act (noted above). He also allowed his finance minister to curtail funding for the crucial MGNREGA as the 2014 election approachedto reduce government debt. This bordered on political insanity since many poorer voters were affected. 18 Political ineptitude sometimes matters in our analyses (see below), and the UPA suffered a significant defeat at that election. 19 South Africa offers a more vivid example of a leader at the very apex of power who was hesitant or even hostile to anti-poverty initiatives: Thabo Mbeki (1998Mbeki ( -2008 pursued market-oriented policies (Aron, 2011;Gevisser, 2009, chapter 30). Mbeki exercised what appeared to be airtight, top-down control. But even during his presidency, some efforts were made to address poverty, in, for example, the provision of water and housing to impoverished townships and squatter colonies, both of which Mbeki supported. And one major initiative, an important case study in our analysesthe Child Support Grantemerged. Mbeki and his colleagues atop the system were lukewarm towards this policy, but one minister was determined to bring it into being, in part to burnish his personal legacy. He received crucial assistance from civil society organisations and an official panel headed by a progressive social scientist, Francine Lund, which provided evidence of the promise of such a grant. That initiative was then supplemented by grants to other disadvantaged groups, and the cumulative impact on income poverty has been significant (Friedman, 2016a;Groenmeyer, 2016;Seekings & Nattrass, 2006). (See the study in this collection of 'Outcomes'.) After Mbeki was ousted by opponents within the ANC in 2008, Jacob Zuma succeeded him as president. He was less hesitant about poverty initiatives, but the main change that attended this transition arose less from his efforts to promote such programmes than from the space that Zuma provided to other leaders to undertake them. The ANC remained dominant within the political system until municipal elections in mid-2016, but it and the government that it led became less monolithic and more amenable to poverty initiatives under Zuma.
In China, between 2003 and 2013, President (and, more importantly, party General Secretary) Hu Jintao and Premier Wen Jiabao worked in tandem, with neither man playing a dominant roleas current President and General Secretary Xi Jinping has done. As with their predecessors, President and General Secretary Jiang Zemin and Premier Zhu Rongji, they differed in their tasks (with the Premier heading the government and leading on economic and social policy) and their personal styles, and they were complementary figures. Under Jiang and Zhu, however, social policy focused on the cities and especially targeted urban state sector workers, though towards the end of their administration they introduced the Western Development Project to benefit underdeveloped inland regions, announced 'new' rural cooperative medical schemes, and began rural tax reforms (Birney, 2014;Duckett, 2017;Wang, 2017).

Motivations: leaders' altruism and calculations
Any discussion of the motivations of senior politicians is fraught with difficulties. We cannot see into their minds orin most casesaccurately gauge the strength of their commitment to various ideals and principles. We can consider their public statements, but words do not always reflect actual beliefs. It is safer to judge them by their actions. But even if we focus there, it is hard to see whether certain decisions are mainly driven by long-standing progressive idealism (which motivated only some of these leaders) or short-to medium-term political calculations (which usually predominated).
Let us take a few tentative steps into this minefield. The political leaders who oversaw the intensified efforts to tackle poverty and inequality in these four countries were neither strangers to altruism nor saints. Most of them took some satisfaction from creating initiatives that would help the poor, but they also made hard-headed reckonings about the political gains which would ensue for them as individuals and for their parties and governments. Considerations of self-interest and partisan advantage usually outweighed socialist or humanitarian idealism. (See the next section on 'Good Politics'.) But as cases from all four countries demonstrate, enlightened approaches to poverty and inequality emerged.
Senior politicians were also motivated by a belief that poverty initiatives were feasible, especially financially. They rightly concluded that government revenues were sufficient to the task. Revenues were surging in India (as they continue to do), and abundant in China. They were less plentiful in Brazil and South Africa, but leaders regarded them as adequate. (Recent economic downturns have created strains in both countriesexcruciating strains in Brazilbut poverty initiatives have not suffered cuts as a result, although those who ousted Dilma Rousseff in 2016 have begun this). So, politicians introduced such initiatives in part becausefinanciallythey could.
Administrative feasibility was sometimes less certain, especially in South Africa and India, but senior politicians there believed that instruments existed or could be created that would enable implementation. This proved to be an especially thorny problem in India. Its flagship programme, the MGNREGA, was shrewdly designed (and then amended once implementation began) to ensure administrative feasibility, although it was attended by some disappointments (Jenkins & Manor, 2017). But the availability of ample revenues led leaders there to pour money into some other poverty initiatives without adequate attention to programme design or administrative structures. 20 That undermined their effectiveness. (See the paper in this collection on 'Outcomes'.) Sonia Gandhi was not intimately familiar with administrative mattersand that contributed to this problem. But her motivations are worth considering. They emerged in part from the years in which she had shared a house with her mother-in-law, Indira Gandhi, who made (mostly empty) promises to tackle poverty. That and her years as the wife of Rajiv Gandhian inept, confused Prime Ministerhad acquainted her with the progressive rhetoric of the Congress Party. The policies that emerged from the NAC that she chaired, and in some cases from government ministries, went much further than previous Congress governments in addressing poverty, and she regarded them as efforts to lend substance to the dynasty's claims to be progressive.
In South Africa, the ANC government's adoption of the market-oriented GEAR development strategy demonstrated Mbeki's reluctance to prioritise social programmes. Despite that, actions made by some ministries to tackle poverty and inequalitylike the creation of the Child Support Grantwent against the grain of that government. Only when Zuma replaced Mbeki in 2009 did senior leaders find space to pursue a wider array of poverty initiatives unencumbered from atop the system. Socialist idealismnot least among powerful interests within the ANCplayed some part in this, as did the strong progressive commitments of civil society leaders and pro-poor campaigners. And as ANC anxieties slowly grew about threats from rival parties, poverty initiatives were increasingly perceived as politically advantageous.
It is difficult to credit Chinese leaders with socialist idealism when their policies after 1979 turned one of the most egalitarian countries on earth into one where inequality had increased markedly. The concentrations of wealth in private handsoften in the hands of members of the ruling party eliteraise questions about the use of the word 'Communist' to describe that party. When Hu Jintao and Wen Jiabao introduced changes that benefited poor people, they were almost certainly aware that rising inequalities sat uneasily with the CCP's formal ideology. But they also sought to cultivate personal legacies as leaders who had mitigated problems created during the preceding Jiang Zemin administration. Under Jiang, taxes and informal fees had increased the burdens on rural dwellers (both poor and prosperous) as local governments became more predatory. That created dangerous tensions between cadres of the party-state and ordinary folk, and eroded the legitimacy of the regime (Duckett & Wang, 2017). Jiang's government had also paid too little attention to the distress of tens of millions of rural-urban migrants who posed a potential threat to public order (Howell, 2016). When these problems began to be seen also as a drag on economic growth, Hu and Wen responded with an array of ameliorative initiatives, with considerable confidence that the party-state possessed the administrative capability to make them work tolerably well (Duckett & Wang, 2017).
In Brazil, President Lula took power in 2002 after a long political career grounded in promises to meet the needs of the poor. His commitment to that ideal is doubted by few observers, and in any case, his emphasis on this theme had created widespread expectations of actions to pursue itso he also faced popular pressure. He succeeded a social democratic predecessor who had introduced significant pro-poor initiatives before him, so he also felt compelled to go further in order to justify his claim to be the principal champion of the poor (Melo et al., 2016a). Thus, for Lula and his chosen successor, Dilma Rousseff, pro-poor commitment and political compulsions reinforced one another. Both leaders assumed, largely accurately, that institutions that had implemented poverty programmes since the mid-1990s were capable of delivering adequately.

'Good politics'
The leaders who undertook or reinforced initiatives to tackle poverty and inequality thought that this amounted to 'good politics'that they, their governments and their parties would benefit politically as a result. They anticipated that their popularity and legitimacy would be enhanced, that votes (if they needed them) would be won, and that popular discontents among the poor (a cause of instability) would be eased. Our evidence indicates that for the most part, at least some of the gains that they anticipated duly followed, although several initiatives proved to be disappointmentsnot least in India.
The leaders' expectations presupposed the financial and administrative feasibility of such initiatives. We have found that those presuppositions were, with certain exceptions, well founded. As noted above, all four governments had sufficient revenues to undertake these initiativesespecially in China and India, but also in Brazil and South Africa. Budgets have become tighter still in those two countries, amid recent economic downturns, but resources were adequate through the period up to 2015 that we have studied. On the administrative front, incapacities created problems in the implementation of certain programmesespecially in South Africa and India.
Incapacities also generated anxieties in those two countries. Two types of worries arose in South Africa. There was scepticism about the capacity of the less-than-agile bureaucracy to manage certain programmes, but these were mitigated somewhat by the relatively straightforward design of the key social grants initiatives (cash transfers, favoured there and in Brazil, and used in China but not in India) (Friedman, 2016a;Groenmeyer, 2016;Seekings, 2011). There were also concerns about the implementation of the Community Work Programme. It was loosely modelled on India's MGNREGA, but South Africa's elected local councils had been starved of resources and autonomy, so they were too weak to play the key roles that their stronger Indian counterparts performed. Instead, civil society organisations operating at the grassroots were drawn in, and the system worked tolerably well.
In India, the MGNREGA functioned reasonably well where state governments in that federal system made efforts to implement it effectivelyas many, though not all, did. This owed much to the programme's strong transparency mechanisms (see details in the following section), which impeded malfeasance among government employees and elected local councillors at lower levels. But in a small number of states, severe problems in the bureaucracy led to dismal implementation. Most notable was Bihar, where political leaders pressed for effective implementation but found that they could spend little more than half of the funds provided from New Delhi because nearly half of the key administrative posts at sub-district level stood vacant. 21 Administrative problems proved quite damaging in three other Indian programmes. Poor drafting of the Forest Rights Act created opportunities for the hostile Forest Service to thwart the implementation of nearly all of its provisions (and the Prime Minister himself flagrantly violated the Act after 2011) (Manor, 2016). The anti-poverty provisions of the JNNURM suffered in its early years because a senior bureaucrat mainly implemented one of them at the expense of the others (Chiriyankandath, 2016). 22 Government support for the Integrated Child Development Services consisted mainly of pumping greatly increased funds into a set of institutions which had long malfunctioned and in which proposed administrative reforms had only a limited impact, with disappointing results (Maiorano, 2016).
These problems in South Africa and India did not prevent the leaders and governments from reaping some political benefits from their efforts to tackle poverty and inequality. Very large numbers of poor South Africans benefited from social grants, and non-trivial numbers gained from other initiatives. The ANC was not (until mid-2016) as acutely concerned with attracting votes as were the ruling parties in the more competitive systems in Brazil and India, but it took satisfaction from the small increase in its already substantial vote share at the 2014 parliamentary election. In India, vast numbers of poor villagers earned decent wages under the MGNREGAtens of billions of person-days were worked after 2006. The media erroneously claimed that that programme won re-election for the ruling alliance in 2009, but it helped to ensure a majority that would survive until 2014 when scandals and political blunders sent it down to a severe election defeat (Manor, 2011).
In Brazil, with its broad popular consensus in favour of efforts to address poverty and inequality, the Workers' Party initiatives were extremely important at two presidential elections. They facilitated Lula's substantial re-election triumph in 2010 and Dilma Rousseff's narrower victory in 2014. (Those initiatives played little part in the attacks on those two leaders seen in 2016, and the consensus will restrain any successor who contemplates slashing those programmes.) China's various initiativesnot least the lifting of the onerous tax burden on rural dwellerslent substance to the pursuit of a 'harmonious society', stressed by Hu and Wen. They did not reverse the increase in collective protestsover 100,000 in 2012, according to official statementsthat had persuaded leaders to spend more on coercive agencies to deal with 'instability' than on national defence. But they may have eased some of the discontents that generated such ructions.

Dealing with corruption and clientelism
Most senior leaders who undertook poverty initiatives, or in some cases, people who worked closely with them, tended strongly to recognise the dangers posed by two things: corruption and clientelism. If subordinates fleece programmes for private profit, they reduce the resources that can make an impact on poverty and inequality. The same happens if resources are diverted by subordinates to distribute among their clients. That may strengthen the bonds between a ruling party and those clients (who often have networks of sub-clients). But poverty initiatives seek to benefitand to cultivate popular political support frommuch larger groups of recipients than can be reached through even the most elaborate clientelist networks.
For these reasons, senior politicians often seek to protect poverty programmes from corruption and clientelism. In doing so, they turn those programmes into 'post-clientelist initiatives' (Manor, 2013b). Leaders in three of our four countries used different devices to achieve this. In South Africa's key social grants initiative, and in Brazil's crucial Bolsa Familia programme, they used advances in information technology to channel cash transfers to recipientswith the aim of cutting out middlemen who might skim off funds. 23 In India's most important and successful initiative, the MGNREGA, an elaborate IT system was used to disburse payments to workers' bank accounts for labour performed on public works projects. But that programmewhich sought to give any rural household the right to demand 100 days of work per year as a hedge against destitutionwent further, using IT to enhance transparency as a counterweight to corrupt acts at and just above the local level. Every worker had a paper 'job card' showing the dates, hours worked and wages earned on the programme, and a similar electronic record for each worker enabled checks for disparities. Other transparency mechanisms were also usedmost notably 'social audits', local-level mass meetings to evaluate (inter alia) conformity with regulations. These transparency mechanismswhich were progressively tightened over timehad considerable effect, so that it became harder to steal from the MGNREGA than any other Indian programme (Jenkins & Manor, 2017). But such mechanisms were scarcely used in the other three countries, or in other Indian initiatives.
In China, the initiatives of Hu and Wen were less concerned with clientelism, but where possible, used new technologies to reduce the chances of local officials siphoning off funds. The reform and then the abolition of rural taxation was inspired in part by concerns about predation by local officials, and it was in part such concerns that had led Zhu Rongji in the 1990s to dismiss proposals for rural pensions and rural cooperative medical schemes (Duckett & Wang, 2017). Under Hu and Wen, it was precisely the use of experimentation when developing or gradually rolling out new initiatives that enabled the leadership to evaluate possible problems due to local government incapacity or corruption.

Leaders: adroit or inept?
When we discuss the impact of senior leaders' political dexterity or missteps on social programmes, we need to make a key distinction noted above. Leaders loom large during the formulation phase, but their influence diminishes once implementation begins. When initiatives are being formulated, senior political leaders often seek out constructive proposals and then strengthen them and make them more feasibleadministratively, politically and financially. Or, less often, they may seek to dilute them. They often exert pressure to bring programmes into being by mustering support from allies and discouraging opponents from undermining them. If they are adroit, they can have a decidedly positive impact. President Lula shrewdly steered his poverty initiatives through Congress, Brazil's national legislature, where his party never enjoyed a majority. This was a complex business that required great subtlety in negotiations, horse-trading, and a degree of intimidation. 24 But sometimes, leaders' inattention to detail or blunders caused damage during the formulation phasealthough occasionally and ironically, these things proved constructive. In India, Sonia Gandhi unwisely decided that the Forest Rights Bill should be rushed into law, to deny opponents time to muster resistance. But things proceeded with such haste that the resulting Act was poorly drafted. It thus included one clause that largely negated its constructive impact (Manor, 2016). In South Africa, by contrast, Jacob Zuma's accession to the presidency in 2009 brought in a leader who was much less attentive to the details of subordinates' actions than his predecessor, Mbeki, and even though he was not a strong advocate of poverty reduction, that opened up space and opportunities for them to devise and introduce poverty programmes.
Once the implementation of initiatives begins, however, politicians operating from atop large, complex systems seldom have much influence because that process is largely beyond their reach. They may adjust such programmes to make them more effective, and they may take action to undermine opponents of those initiatives. But day-to-day developments mostly occur in lower-level arenas where subordinate politicians, bureaucrats and other actors shape events beyond the purview of leaders at the apex of power.
There are a few exceptions to this generalisation, but only a few. Consider first a positive and then two negative examples from India. Prime Minister Manmohan Singh persuaded cabinet colleagues in 2010 to make a change in the rules governing India's MGNREGA that undermined efforts by actors at lower levels to siphon off funds from the programme. But then, by contrast, he permitted cuts in funds for that same programme after 2011 (which, as noted above, contributed to his party's defeat at the 2014 election), and he responded to pressure from industrialists by undertaking gross violations of the Forest Rights Act. 25 Such interventions during the implementation phase are, however, quite rare. Usually, the most that senior politicians can do is to encourage subordinates to pursue programmes effectively, apply pressure for compliance, monitor events as best they can, and hope for the best.
The actions of political leaders had a major impact on intensified efforts to tackle poverty and inequality in these countries. 'Political agency' mattered. Innovations in three of the four countries enabled leaders and governments to break free of 'path dependency'. As a result, 'politics'which technocrats bemoanwas turned to constructive purposes in the struggle to promote social justice.
Leaders in these four countries differed in their backgrounds, inheritances and outlooks. They engaged differently with different political systems and socio-economic circumstances. The capabilities of their party organisations and governments varied. There were further variations in the powers that they wielded, and in their management of tensions within their governments, and within their political systems more generally. They tackled poverty and inequality in somewhat different ways.
But they also had things in common. They all felt driven to develop poverty initiatives by rising inequality. They all believed that this was feasible, and that it would prove politically advantageousas indeed it did. Hard-headed political calculations almost always predominated over altruism in their minds, but the two things did not entirely contradict one another. Some were more adroit than others, but all were sufficiently skilled and shrewd to make significant headway in redistributing wealth, reducing poverty, and reaching the poorest of the poor.