Why do Authoritarian Regimes Provide Public Goods? Policy Communities, External Shocks and Ideas in China’s Rural Social Policy Making

Abstract Recent research on authoritarian regimes argues that they provide public goods in order to prevent rebellion. This essay shows that the ‘threat of rebellion’ alone cannot explain Chinese party-state policies to extend public goods to rural residents in the first decade of the twenty-first century. Drawing on theories of policy making, it argues that China’s one-party regime extended public goods to the rural population under the influence of ideas and policy options generated by policy communities of officials, researchers, international organisations and other actors. The party-state centre adopted and implemented these ideas and policy options when they provided solutions to external shocks and supported economic development goals. Explanations of policies and their outcomes in authoritarian political systems need to include not only ‘dictators’ but also other actors, and the ideas they generate.

WHY DO AUTHORITARIAN REGIMES PROVIDE PUBLIC GOODS? MUCH research portrays autocrats as focussed on maximising their own revenues (Niskanen 1997;Acemoglu & Robinson 2000). It expects them to channel private goods to members of their support base in the 'selectorate', but to neglect the 'unenfranchised' majority (Bueno de Mesquita et al. 2003). 1 Other recent studies have found authoritarian regimes sometimes do extend provision beyond their immediate support base-but only to groups who threaten rebellion. Gandhi and Przeworski (2006), for example, argue that well institutionalised one-party regimes make policy concessions so as to co-opt those who might otherwise challenge them. Gallagher and Hanson (2009) similarly suggest authoritarian regimes sometimes try to reduce revolutionary 1 We do not have the space here to discuss the substantial literature categorising authoritarian regimes (Linz 2000), explaining their transitions to democracy (O'Donnell et al. 1986) and examining how they repress their opponents (Lichbach 1995;Wintrobe 1998). threats by offering 'carrots'-and they use as an example the Chinese party-state's provision of social insurance in the late 1990s to urban workers. 2 In the first decade of the twenty-first century, however, China's one-party state adopted a series of policies aimed at providing support to people in ill health and old age, primarily in the countryside. But China's rural residents are not part of the Chinese Communist Party's (CCP's) 'winning coalition' or 'selectorate'. Indeed, they have long been seen as the most politically marginalised and disadvantaged by the economic and social policies of the People's Republic of China (PRC). Despite the CCP's revolutionary rural heritage, after it seized power in 1949 it 'squeezed' agriculture (buying rural goods at low prices to provide cheap food in the cities and fund industrialisation), and it failed to provide rural dwellers with the public housing, assistance with medical expenses and pensions accorded to many urban families. Although land reform reduced inequalities within rural areas, and communes provided some security and public goods, provision was much less generous in the countryside than in the cities. Even when Mao Zedong called for greater focus on rural areas from the late 1950s through the late 1970s (and initiated rural cooperative medical schemes, for example), government spending on officials and urban workers far exceeded that on farmers. Then in the post-Mao period, the party-state taxed farmers highly in comparison with urban wage earners, and new social programmes-minimum livelihood guarantees, re-employment training, unemployment insurance, and then reform of old-age and health insurance-prioritised urban officials and formal sector workers but neglected farmers (Solinger 2001;Duckett & Hussain 2008;Frazier 2010).
What, then, changed into the twenty-first century to lead the Chinese party-state to extend public goods to the countryside? Expert observers of the period did not consider rural dwellers a threat to the Chinese party-state, even if their protests were increasing (O'Brien 2009, pp. 25-9;Saich 2014). We therefore take a bottom-up approach and examine the drivers of China's twenty-first century rural social policies. Policy analysis has shown that in democratic political systems policies are shaped not just by leading politicians or policy makers, but also sometimes by communities or networks of actors (bureaucrats, researchers, activists, journalists and non-governmental organisations) or individual policy entrepreneurs. 3 Political institutions also play a role, for example by giving certain actors access to influence, or by creating veto points where policies can be blocked (Weaver & Rockman 1993;John 2012), while external shocks or 'focussing events' are seen as the main catalysts for policy change (Baumgartner & Jones 1993;Kingdon 2003). Ideas can also play a role, as policy actors interpret shocks, evaluate evidence, invoke values and articulate preferences (Majone 1989;Hall 1993;Béland and Cox 2011).
Drawing on this policy analysis research we adopted an analytical framework to guide our investigation of the factors that have shaped China's twenty-first century rural public goods provision. Thus we tried to identify key actors, political institutions, external shocks and ideas that shaped specific policies. Although most policy analysis research has focussed on democratic political systems in Europe and North America, its findings are relevant to authoritarian systems. Much authoritarian policy making will involve not just 'dictators' but other actors, even if top leaders have ultimate decision-making power. Political institutions may work differently, but they still shape access to policy influence, and not only electoral institutions are important. Finally, policy actors in authoritarian regimes are potentially just as susceptible as their counterparts in democracies to the influence of contingent external shocks and to the complex mix and flow of ideas around them.
Certainly, research on China has already shown that not only top leaders but also ministries and provincial level governments, and institutions such as bureaucratic rank, can shape policy (Lieberthal & Oksenberg 1988). 4 The Chinese political system consists of a sophisticated array of integrated CCP and government organisations that radiate out from the political centre in Beijing to govern one fifth of the world's population. The CCP centre consists of not only the Party General Secretary-its top leader-but also between four and eight others (it has varied) in the Politburo Standing Committee (zhengzhiju changwuhui-PBSC) and about another 18 in the wider Politburo (zhengzhiju-PB). While the PB and PBSC meet frequently, some decisions are also approved by the less frequent gatherings of the CCP Central Committee (Zhonggong zhongyang weiyuanhui-CCPCC) of around 200. One member of the PBSC is Premier and head of the State Council (guowuyuan-SC, effectively the 'cabinet'), made up of (currently) 23 government ministries and commissions that in turn have agencies in every province, city and county. 5 Other PBSC members and Vice-Premiers coordinate broad policy across ministerial spheres of responsibility (Hamrin 1992).
China's top leaders are extremely powerful in this centralised system. The Party General Secretary and members of the PBSC set the overall direction of policy making and determining key priorities as set out in China's overarching five year plans and annual government plans. Under Hu Jintao and Wen Jiabao (2002, for whom 'scientific development' was a key slogan, scholars and other experts were invited to present their work directly at PB 'Study Sessions' . More routinely, however, the central party-state uses sophisticated bureaucratic arrangements that funnel upwards information and intra-ministerial expertise, external advice (from scholars, think tanks, lower levels of the party-state and the media), policy options and feedback from policy experimentation (Heilmann 2008;Mertha 2009).
In this context our essay focuses on the Chinese government's policies to abolish agricultural tax and extend programmes of public provision-rural cooperative medical schemes and new rural pensions-in the first decade of the twenty-first century. These three policies aimed to reduce the tax burden on farmers and deliver to all rural dwellers financial help with medical expenses and basic pensions. Despite their limitations they may have begun to bridge the longstanding divide in the state's treatment of urban and rural dwellers. We chose these policies because they targeted some of the most marginalised in Chinese society-poor, sick and elderly rural residents. They not only benefited a large number of people, they-especially the medical schemes and pensions-constitute 'tough cases' because they provide for those who are the most vulnerable and least able to threaten the party-state.
We used a case study approach, tracing the evolution of the three policies and identifying the factors (focussing on actors, political institutions, external shocks and ideas) that influenced their development and adoption (George & Bennett 2005). We identified and read key policy documents and leaders' speeches, as well as academic papers about each policy, and we also interviewed individuals who had been researching or interacting with government departments during the development of the policy. We established a chronological narrative for each policy case and identified contributing factors. We then developed an explanation for each case, before considering the cases together and whether they influenced each other or were influenced by wider contextual factors. We then fully evaluated competing explanations, especially the 'threat of rebellion' thesis (George & Bennett 2005, p. 91). Finally, we took our descriptive explanations for each case and developed them into an overarching explanation that captures both specific factors and key features of institutional and historical context (George & Bennett 2005, p. 93). It is challenging to identify actors and decisions in China because so many decisions take place in closed forums that do not publish written records, but we have pieced together our evidence with care.

The argument
Our essay argues that the threat of rebellion cannot explain the early twenty-first century expansion of public goods in rural China. Although rural protest was a growing problem in the 1990s and early 2000s, the Chinese party-state focussed on predatory local government activity-ad hoc fee charging-as its most important cause. Public protest was a factor in rural tax reforms but it was not the only one, and it was much less important in the introduction of new rural cooperative medical schemes (xin nongcun hezuo yiliao-NRCMS) and new rural pensions schemes (xin nongcun yanglao baoxian-NRPS). NRCMS and NRPS were adopted nationwide in the twenty-first century not because top leaders thought they would significantly reduce protest, but because a number of different actors-a mix of central and local officials, researchers, international organisations and journalists-put ideas about rural problems and their solutions on the agenda and then sustained and developed policy options through research, debate and experimentation. They did so sometimes under the influence of previous policies, but also persuaded by the ideas they absorbed by considering the experiences and policies of other countries. Top leadership consensus on adopting andmore importantly-driving the implementation of new policies nationwide emerged when these policies were seen to support the wider national economic development paradigm as it evolved through interpretations of external shocks, most importantly: the 1997 Asian financial crisis, the 2003 Severe Acute Respiratory Syndrome (SARS) epidemic and the 2008 global financial crisis. Political institutions had myriad influences but most notably allowed (indeed encouraged) central and local officials to develop and experiment with policies.
This essay next sets out the new social policies adopted and implemented in the first decade of the twenty-first century. It then examines in more detail the three policies we selected for study and explains why each was adopted. It concludes that in theorising about authoritarian regimes, political scientists may need to revise assumptions about the role of 'dictators' and their motivations in policy processes.

Extending public goods to the unenfranchised in the twenty-first century
Having focussed on delivering public goods to urban residents in the 1990s, the Chinese party-state in the first decade of the twenty-first century introduced new social programmes that extended provisions to the poorest, primarily in rural China. In late 2002, it announced policies to reduce the burden of arbitrary fees that local governments charged farmers and to introduce NRCMS (CCPCC & SC 2002a). In 2005 it then abolished agricultural tax and began to expand NRCMS, investing more in subsidies and setting clear targets for local governments to enrol voluntarily the vast majority of villagers (Brown et al. 2009). In 2006, it announced that all children would receive nine years of free education and began promoting this policy first in western and central rural areas and then nationwide. 6 In 2007 the party-state announced that the means-tested 'minimum livelihood guarantee' programme (a household income support initiative that had operated in the cities since the 1990s) would be extended to the countryside (Gao & Zhai 2012). In 2009, it began a national Rural Pension Pilot Programme, and then extended its coverage as it rolled out NRPS nationwide, finding take-up much faster than anticipated.
Although it focussed on the countryside, the party-state also introduced new social security programmes for the urban 'unenfranchised'. In 2007 it introduced Urban Residents' Health Insurance for the non-working population in the cities who had been excluded from the 1990s urban employee health insurance (Lin et al. 2009). In April 2009, it then announced major reforms of the health system, in which it set the goal of universal access to basic health care by 2020 (CCPCC & SC 2009). Finally, in 2011, the party-state centre announced the pilot Social Pension Insurance for urban residents (Dorfman et al. 2013).
Among this list of policies to extend public goods in China's one-party system, three of the most important were the reform and then abolition of agricultural tax, the introduction of NRCMS, and the creation of NRPS. Each of these policies was both novel and radical: removing farmers' tax obligations and for the first time giving rural dwellers entitlements to state-subsidised provision in ill health and old age. In the sections below we set out the evolution of each and the particular range of key actors, political institutions, external shocks and ideas that put them on the agenda and then resulted in their adoption and implementation.

Reform and abolition of agricultural tax
The Chinese party-state did not tax farmers under the planned economy, but it introduced agricultural taxation from the early 1980s as it decollectivised farming. Taxation became a source of grievance in the Chinese countryside soon after it was introduced (Bernstein & Lü 2003). But more controversial than statutory taxes were the additional fees charged by local government officials, sometimes to fund specific programmes as a result of policy initiatives from central government ministries. The resulting 'farmers' burden' (nongmin fudan) of combined taxes and fees appeared to fuel rising rural protests in the 1990s, particularly after the 1994 tax reforms reduced local government revenues (Bernstein & Lü 2000;He 2000;Li 2006;Kennedy 2007;Zhao 2007). The central party-state was aware of the problem of excessive state extraction from the mid-1980s, and it was reported in the media and scholarly research (Bernstein & Lü 2000). In 1993 the Party Central Committee and State Council (CCP and SC, China's central government cabinet) published a list of 99 fees and fundraising schemes imposed by 24 central ministries and commissions and at the same time criticised local officials for using 'instruments and methods of dictatorship to collect money or goods' from farmers (Bernstein & Lü 2000, p. 746). Reform of central-local tax-sharing in 1994 then exacerbated revenue problems in some areas and may have contributed to further extraction. From at least 1996 the centre also closely monitored the rising 'farmers' burden'-related protests, issuing circulars from 1996 to 2005 that disclosed serious incidents caused by the 'farmers' burden' in various places all over the country (Dong & Wen 2008). 7 Rural tax reform was put firmly on the policy agenda in October 1998, when the State Council (headed since March that year by Premier Zhu Rongji) set up a working group led by the powerful Ministry of Finance. In April 1999 the working group recommended abolishing fees and increasing taxes from 3% to 7%-an attempt to prevent ad hoc fees and arbitrary local government predation (Xu 2004, p. 33). In February 2000 the CCP's very top leaders, the Politburo Standing Committee, approved this 'tax-for-fee' (feigaishui) reform and in March the CCPCC and State Council issued a document that introduced experiments in the province of Anhui. Other localities resisted implementing the experiments, however, because of concerns that local government revenues would suffer. Some parts of the central bureaucracy, such as the Ministry of Education, were also concerned with the potentially detrimental impact on local service provision. In March 2001 the State Council ordered that experiments should not be extended, effectively putting a brake on the reforms (Li 2006).
In March 2002, however, seven months before Hu Jintao became Party General Secretary and a year before Wen Jiabao became Premier, the State Council kick-started the 'tax-for-fee' reforms again and promised fiscal transfers to compensate some localities. By the end of 2002, the centre had extended experiments to 20 provinces, and then in March 2003 the State Council called for them to be adopted nationwide (SC 2003). In 2004 Premier Wen went a significant step further and announced that the agricultural tax rate would be reduced every year and completely abolished in five years. By 2005, 28 provinces had abolished agricultural taxes, and they formed a negligible share of government revenues nationally. The National People's Congress, China's legislature, then ruled that from 1 January 2006 agricultural tax was no longer payable (Pang 2012, pp. 25-7).
What explains China's reform and abolition of agricultural tax? Protests over the farmers' burden certainly played a role, but they were not the only influence (Yep 2004). China has a long history in which excessive burdens on farmers have led to rebellions and the collapse of imperial dynasties. And the CCP leadership surely wanted to prevent protest from escalating. Certainly the central party-state circulars indicate that protests relating to the 'farmers' burden' were of sufficient concern to merit close monitoring, while Zhu Rongji raised them in an internal speech (Zhu 2011). But rural protest was targeted overwhelmingly at local governments rather than the centre. And at the same time, central and local government restrictions on freedom of association, effective measures to prevent people from organising collective action, and arrests of protest organisers, meant that rural protests against local state predation had not seemed an imminent threat to China's authoritarian regime (Bernstein & Lü 2000). Finally, while monitoring began in the mid-1990s, taxes were not reformed until 7 The Party Central Committee Office and State Council Office published reports in 1996, 1997, 1999, 2000, 2001, 2002 and 2005. Each one was called a 'Notice concerning the situation of pernicious cases of farmers' burden in [the previous year]'. The first was published in Hubei zhengbao, No. 7 (1996, p. 9), and the others in Hubei zhengbao, No. 8, (1997, pp. 25-6);Hubei zhengbao, No. 11, (1999, p. 19);Guangxi zhengbao, No. 32, (2000, pp. 18-9); Zhejiang zhengbao, No. 32, (2001, pp. 7-8);Zhejiang zhengbao, No. 30, (2002, pp. 14-8);and Neimenggu zhengbao, No. 6, (2005, pp. 29-30). the early 2000s after local governments blocked experiments, and so clearly protest was not the only factor.
Chinese ministries, scholars and the media in the rural policy community also played their part, by getting-and keeping-the idea of the 'farmers' burden' on the agenda. The Ministry of Agriculture's Rural Work Bulletin for example published an article in 1983 on the need to reduce the 'farmers' burden' (Chen 1983, pp. 19-20). Rural policy community members then reported and discussed the farmers' burden issue through the 1980s and 1990s (Bernstein 1999). The phrase 'farmers' burden' was then widely used in party-state documents relating to rural tax reform.
By 2002 and 2003, just as Hu and Wen came to power, the debate over tax reform began to intensify and some scholars argued that agricultural tax revenues were small enough to make even abolition affordable. When others countered that removing the tax obligations of approximately half of the population set a poor precedent and was itself unfair in principle, supporters of abolition contended that farmers-already among the poorest in the countrywere taxed much more heavily than better off urban residents and rural non-agricultural workers (Nongcun shuifei gaige yanjiu ketizu Sun 2003, pp. 16-8;Zhao 2003, pp. 44-6). Finally, some argued that China's 2001 entry into the World Trade Organization would harm agriculture and that it was unfair that while other countries subsidised this sector, the Chinese government still taxed it (Li 2002, pp. 51-5;Duan 2003, pp. 12-5).
External shocks-and their interpretation-also played a part. Influential economic policy actors interpreted the Asian financial crisis of 1997 as demonstrating the dangers for China's economy of relying on export markets. Justin Yifu Lin, an economist and later a deputy director of the World Bank, for example, began to argue that relying too much on overseas markets created risks for the Chinese economy, and that domestic-especially rural-consumption needed to play a greater role in China's growth, in turn requiring higher rural incomes and better social security (Lin 1999(Lin , 2003(Lin , pp. 20-1, 2005Wang 2004;Wen et al. 2013). These ideas were soon found in policy statements. In 1998, for example, the CCPCC, at a meeting that approved rural tax reforms, noted the need to develop domestic demand and promote rural development (CCPCC 1998). In 2002, a CCPCC and State Council document pushing forward rural 'tax-for-fee' reform argued that 'increasing farmers' income is not only concerned with … improving their lives and rural stability, it is also concerned with carrying out the policy of expanding domestic demand and the whole national economy' (CCPCC & SC 2002b, p. 1). In 2006 Minister of Finance Jin Renqing argued that the abolition of rural taxes: … is a force to expand domestic demand [and] maintain the speedy development of the national economy. The countryside is a potentially massive consumer market, [and] the rural population concentrates our country's most numerous and greatest latent consumer group. It is the source of our country's most reliable, enduring driving force for economic growth. (Zhou 2006) Thus for the reform and abolition of rural fees and taxes, the party-state centre appeared to be concerned about protest. But it initiated reform only after the external shock of the 1997 Asian financial crisis apparently led top leaders to accept arguments linking the 'farmers' burden' to lack of consumer power. Agricultural 'tax-for-fee' reform was prioritised when it became associated with the party-state's wider economic development strategy, and it was boosted in 2002 by concerns about the impact of China's WTO entry. Several issues and ideas thus converged in the late 1990s: rising levels of protest, arguments about the need for domestic consumption-led economic growth and concerns about WTO membership, as well as policy-specific proposals from the rural policy community. Top leader Zhu Rongji himself seems to have made a difference, too, acting decisively on fiscal reforms perhaps because the 1997 crisis hit China just as he became Premier. Local leader Hui Liangyu played an important role by sustaining experiments that allowed the policy to develop until China's economy recovered from the crisis and central subsidies were seen as affordable (Pang 2012). The Hu-Wen leadership appears to have been persuaded by policy community arguments that abolishing agricultural tax was affordable as well as contributing to social stability and economic growth. In this it was helped by agricultural tax's decline from 4% of government revenues in 1998 to less than 1% in 2004 (He et al. 2004).

New rural cooperative medical schemes
Unequal urban-rural health insurance provision dated back to the pre-reform Mao era, when state enterprises paid for much of their workers' relatively high quality health care while rural cooperative medical schemes, introduced from the late 1950s, funded villagers' medical treatment much less generously. Then from the early 1980s, even the very basic rural cooperative schemes were allowed to collapse-with only an estimated 5% of villages retaining them by 1984 (Duckett 2011).
The possibility of re-establishing cooperative medical schemes, or some form of rural health insurance, was explored again from the late 1980s through Ministry of Health projects with international organisations. From 1985 to 1993 the Ministry of Health, in collaboration with the Rand Corporation, researched rural health insurance in two counties in Sichuan. In 1987 the Ministry of Health and Anhui Medical University jointly researched rural cooperative medical and health systems. From 1988 the Ministry of Health's Policy and Management Research Experts Committee carried out research on rural health that established the basis for cooperative medical schemes reform (Cai 2009). In 1993 the Ministry of Health, with the backing of UNICEF, began a '7-year applied policy study on provision and financing of health care for the rural poor' (Liu et al. 2002, p. 4). Through the 1990s the Ministry of Health was also involved in collaborative projects with Chinese and foreign researchers, international organisations such as the World Health Organization, the World Bank, the United Kingdom government's Department for International Development, The Rockefeller Foundation and the Asian Development Bank. These projects experimented with different policy options and demonstrated that large numbers of rural dwellers in China were becoming impoverished by the high costs of their health care because they had no health insurance (Liu & Rao 2006).
Amid all this research activity, voluntary cooperative medical schemes (involving villager contributions but no state funding) rose up the central party-state's policy agenda. In 1991, Premier Li Peng backed cooperative medical schemes, and then in 1993, a State Council Research Office study concluded that the government should support the 'revitalisation' of cooperative medical schemes. In 1996, Jiang Zemin backed it, and in 1997 a CCPCC and State Council Decision on cooperative medical schemes set a target of establishing voluntary cooperative schemes across rural areas by 2000 and announced local experiments (CCPCC & SC 1997). Despite this central party-state decision, however, efforts to extend risk protection failed in the 1990s because voluntary contributions conflicted with central partystate directives aimed at stopping local governments charging ad hoc fees to rural residents (Zhang et al. 2014). Although high on the agenda, it was not high enough for the party-state centre to put weight behind financing or driving implementation nationwide.
Again in 2002, however, Jiang Zemin lent support to cooperative medical schemes, and again a Party Central Committee and State Council Decision announced experiments (Liu & Rao 2006). This 'new' rural cooperative medical system involved county governmentorganised schemes to fund people's health care costs from contributions by households and government. It differed from its predecessors in part because it was backed with central government subsidies to poorer localities, and crucially, because the central party-state created a system of coordinating committees to implement the schemes (Manuel 2016). Experimentation (2003Experimentation ( -2005 focussed on refining the policy and overcame concerns about local embezzlement of centrally transferred funds (Zhang et al. 2014). The programme was rolled out nationally in 2006.
What then explains the eventual adoption of NRCMS? It was not due to protest-in fact cooperative medical schemes were hindered in the 1990s because contributions were associated with 'predatory' local government fee charging. (For this reason, the party-state centre was unwilling to make cooperative medical schemes mandatory as health specialists advised.) Instead, cooperative medical schemes and rural health insurance (both were explored in the 1980s and 1990s) were put on the agenda in part by a health policy community comprising the Ministry of Health, international organisations, and domestic and foreign researchers. With Mao-era cooperative medical schemes a precedent, they could argue that some form of rural health scheme was both feasible and necessary. This was sufficient to get top leader Jiang Zemin's backing and a party-centre decision on cooperative medical schemes in 1997, but not enough to secure implementation or funding-at least until 2002.
The party-state issued a second Decision in 2002-again with the backing of Jiang Zemin-in part because external shocks pushed rural health up the agenda. Top leaders had been embarrassed in 2000 by China's low ranking (188 out of 191 nations) in a World Health Organization international league table that compared 'the fairness in financial contribution' of health systems across the globe (WHO 2000;Wang 2008, pp. 128, 130). The low ranking was in large part because most rural dwellers paid 'out of pocket' for their health care. It was a shock for top leaders because in the past China had received much praise for its health system, including the 1970s rural cooperative schemes (Zhang et al. 2014).
At the same time, officials and researchers helped to keep cooperative medical schemes on the agenda by arguing that inadequate health protection was contributing to rural impoverishment (Liu & Rao 2006;Wang 2008, pp. 128, 130). China's leaders were keen to maintain their international reputation for tackling poverty, and were concerned about falling rural incomes after 1998. Also as we have seen, members of the economic policy community had begun to argue that rural domestic demand-limited by low incomes-was important for balanced growth (CCPCC 1998). Health researchers' evidence on the health system causes of rural deprivation therefore helped portray cooperative medical schemes as solving not just a health problem but also wider poverty and economic problems.
Central party-state consensus on the importance of cooperative medical schemes might nevertheless have been difficult to sustain had it not been for the SARS epidemic that spread across China and then beyond its borders from late 2002 to mid-2003. The SARS crisis thrust China's problem-stricken health system into the national and international spotlight, and revealed the dangers of poor health risk protection: rural dwellers afraid to seek costly diagnosis and care, and an epidemic out of control. SARS struck just as Hu and Wen took over as top leaders from Jiang and Zhu. Had they been unable to contain the disease, their leadership-and perhaps CCP rule itself-might have been in question. Thus, although the epidemic was, by summer 2003, brought under control, it meant the gap between rural and urban health provision remained high on the Hu-Wen policy agenda (Duckett 2011). Whether Hu and Wen would otherwise have pushed through NRCMS implementation we do not know. It was before this (in January 2003) that the party centre set up coordinating committees to drive implementation, but large-scale experimentation and nationwide adoption came later.
NRCMS was developed through the activities and ideas generated by the health policy community. And as with agricultural tax reform, it was when NRCMS was interpreted as helping solve problems created by external shocks and contributing to wider economic development strategy that the party-state centre prioritised it for national implementation. The role of leaders in this policy was not straightforward: although top leader Li Peng backed cooperative medical schemes in 1991, and then Jiang Zemin backed them in 1996 and again in 2002, Premier Zhu Rongji's concern with economic reform and fiscal policy apparently won out in the 1990s. Hu and Wen clearly backed NRCMS after SARS, but may have acted more decisively because the epidemic had put health policy centre stage for the first time in the post-Mao period.

New rural pension schemes
Unlike the many urban residents who had enjoyed employer pensions introduced from the 1950s, China's rural dwellers had never had access to pension schemes. Instead, they had relied in old age on their families and, failing that, basic assistance from rural collectives. Schemes began to appear in the mid-1980s, however, when according to Tony Saich in the absence of central government direction, 'rural localities began to experiment with new forms of old-age provision and these were pulled together by the Ministry of Civil Affairs' (Saich 2008, p. 154). This Ministry then 'won approval' to experiment with rural pensions and proposed testing six models across five provinces. In 1991, it set up a Temporary Office for Rural Social Insurance Pensions and extended experimentation to over 1,000 counties by late 1992 (Saich 2008).
Experimentation with rural pensions continued through the 1990s and covered 82 million farmers nationwide by 1998 (Saich 2008). But as with the rural cooperative medical schemes at the time, most of these 'old rural pensions schemes' were voluntary and received no government funding. As a result, they tended to attract only the well-off rural dwellers and failed to enrol a significant proportion of the rural population. Because they involved local officials collecting contributions from rural dwellers they-like cooperative medical schemes-conflicted with the Ministry of Agriculture's efforts to reduce local government fees and the 'farmers' burden' (Wang 2008, pp. 128, 130). Indeed, Zhu Rongji, after the Asian financial crisis, blocked pension reform as he launched a 'rectification' of the insurance industry, apparently concerned about corruption, financial mismanagement and the fiscal risks of having to bail out failed programmes (Saich 2008, p. 155).
In autumn 2002, however, with local 'tax-for-fee' reform now almost completed and fees abolished, the CCP's 16th Party Congress proposed 'exploring and creating' a rural old-age care insurance system. The Rural Social Security Division of Ministry of Labour and Social Security (laodong he shehui baozhang bu-MoLSS)-since 1998 charged with rural pensions-pushed forward experiments. As with NRCMS, although the government approved experimentation, it was not yet fully committed to the initiative. Instead, it allowed exploration, discussion and testing of different possible models to assess their viability and cost (MoLSS Research Group 2007, p. 1).
Members of the rural policy community-government officials as well as researchersmeanwhile debated whether rural pensions should be set up nationwide. Some made the longstanding argument that rural dwellers had land and could rely on their children for old-age care, so that the government has no need or responsibility to create a pension scheme for them. Others contended that the government had not even been able to provide good urban pensions, let alone create schemes for rural dwellers. Even those who supported the introduction of national rural pension schemes in principle questioned whether the timing was right when China still had so many farmers. A 2002 ADB report, however, supported rural pensions, arguing that some European countries had introduced rural old-age insurance at similar or even earlier stages of development (Saich 2008, p. 157). Others pointed out that agriculture was only 14% of GDP and China's GDP per capita was now high enough to make pensions viable (Chen 2004, pp. 20-2).
In November 2003 the policy received a boost after a Central United Front Department delegation reported to (then) Vice-Premier Hui Liangyu that new rural pensions were 'feasible' and might serve to 'ease potential social conflicts and promote rural social stability' (MoLSS 2003). Premier Wen Jiabao then in turn expressed cautious support for further 'exploring' the initiative locally (Zou 2008, pp. 92-5). The Rural Social Security Department of MoLSS seized the opportunity created by Premier Wen's support and pushed for localities to set up the schemes. But despite Wen's backing for it, the United Front Department delegation linking it to the issue of social unrest and MoLSS encouraging it, the scheme was not yet given backing for nationwide implementation.
At this stage, a research project funded by the Japan International Cooperation Agency (JICA) played an important role. Partnered with MoLSS (which, in 2008, became the Ministry of Human Resources and Social Security (renli ziyuan he shehui baozhang bu-MoHRSS)) and running from 2006 to 2008, the project examined schemes in eight counties across seven provinces. 8 It concluded with a workshop attended by officials from not only the MoHRSS itself and the HRSS departments of 20 provinces, but also the National Development and Reform Commission, the Ministry of Finance and China's central bank. This workshop was followed by other meetings, and led to further local experimentation in 464 counties (Tong 2009, p. 1).
The central party-state then moved quickly, in June 2009 approving the MoHRSS's rural pension programme and announcing pilot implementation in 10% of counties across the country. The State Council now set a target of including all eligible rural residents by 2020 (SC 2009). Take-up and implementation, however, dramatically exceeded expectations and the programme had been adopted nationwide by the end of 2012 (Wen 2012a, p. 3). Once central leaders were behind the scheme and approved government funding for it, 9 local governments 8 'Zhong Ri hezuo nongcun shehui yanglao boaxian xiangmu zai Jing qidong', Xinhuanet, 15 February 2006, available at: http://news.xinhuanet.com/society/2006-02/15/content_4184945.htm, accessed 12 June 2014. 9 The central government fully subsidised basic pensions in middle and western provinces and funded 50% for eastern provinces. Local governments also subsidised farmers' personal contributions.
were more willing to implement it-helped by the fact that implementing rural pension reform experiments was made important in the performance evaluations of local leaders. The threshold payments for joining the scheme were modest-at 100 yuan a year for the lowest level-so that even rural residents in poor areas could afford to join Lu 2010). And because local governments matched personal contributions, the scheme was not seen as a form of state predation.
What then explains the adoption of NRPS? Although an official report had linked it to 'social stability'-in other words, preventing social unrest-there is scant evidence that the absence of rural pensions was contributing to rural protest. Perhaps the United Front Department delegation thought that introducing pensions would contribute to easing more general rural discontent. Or perhaps it used the threat of instability to secure support for a policy it backed for other reasons. Certainly, if the threat of protest did play a role in introducing NRPS, it was not a central one, though in the middle of the first decade of the twenty-first century MoLSS (2006) did encourage local governments to enrol rural dwellers who had lost their land in existing pension schemes (urban or rural) so as to reduce dissatisfaction and potential instability.
Rural pension policy was put on the agenda by local governments, soon with the support of the MoCA, and then nurtured through experimentation and research that sometimes involved international organisations. Perhaps the debate over rural pensions in the policy community meant that the need for pensions was less clearly articulated in the early 2000s. As with NRCMS, it was only when it was interpreted as helping solve problems (revealed by external shocks) and contributing to the wider national development strategy that the partystate centre prioritised pensions policy for national implementation. The need to stimulate domestic demand, an idea that had begun to gain credence from the late 1990s, became even more prominent after the global financial crisis of 2008 once again hit China's exports. Top leaders soon began to link pensions to increasing rural consumption. As Premier Wen Jiabao noted at the launch meeting of the new rural pension pilot scheme in 2009: against the background of the current financial crisis, a signal is sent to rural dwellers that the government will create a full-payment minimum basic pension for millions in the countryside, who will enjoy old-age care. Without fear of trouble in their life, they will dare to consume and spend more, which will produce a strong and durable drive for promoting rural markets and enlarging domestic consumption demand. (Wen 2009, p. 2) NRPS was also increasingly seen as helping deal with another national economic development problem: China's ageing population. The party-state set up a Commission on Ageing as early as 1999, and through the 2000s the sense of crisis grew over care for the elderly due to China's one-child policy and the outward migration of young people from the countryside (Zeng 2001, pp. 3-9;Zhu 2012, pp. 42-3). Arguments about the need to tackle this crisis were sustained not only by parts of the bureaucracy (the Rural Social Security Department of the MoLSS/ MoHRSS) with an interest in expanding provision, but also by researchers and foreign aid agencies (such as the JICA) who together kept it on the policy agenda with joint projects.
Overall, then, local governments and the rural policy community generated ideas and then experimented with policy options that were gradually seen as solutions to not only rural but also national economic and social problems as members of China's economic policy community interpreted external shocks like the 2008 global financial crisis. China's top leaders played a significant role-both negatively and positively-in NRPS. In the late 1990s, Premier Zhu Rongji blocked and 'rectified' rural pensions because they conflicted with rural tax, finance and insurance policies that were pushed up the agenda when the Asian financial crisis focussed attention on the fiscal as well as political risks of rural local government financial mismanagement and predation. He was apparently uninterested in rural social security, stating at the end of his term as Premier in 2003 that social security reform had been basically achieved (Zhu 2002). Hu and Wen, however, went on to develop rural social security as their 'new socialist countryside' agenda developed from NRCMS through to pensions.

Explaining authoritarian public goods provision: beyond the 'threat of rebellion'
We do not claim that the expansion of public goods under Hu and Wen removed inequalities and transformed Chinese society. Indeed, there were many limitations to the policies adopted, many problems implementing them, and their outcomes have been mixed. The abolition of rural taxes, for example, has not been fully compensated for by fiscal transfers. It has therefore reduced local government funding and in some localities has hit spending on education, health and social services ). NRCMS, while it has successfully enrolled the vast majority of China's rural population, is under-funded and unable to pay for costly care. Rural pensions are meagre in comparison with those of many urban residents. Nevertheless, the expansion of public goods to rural areas has aided many families and has established the principle of universal entitlement to basic provision for the first time since 1949-even if 'basic' has not been defined. Provisions and government spending on these programmes has grown since they were extended nationwide.
It is also important to understand that in expanding provision in the early twentyfirst century, China's central party-state benefitted from an enabling economic and fiscal environment. As Peter John (2012) has noted in other policy contexts, budgetary concerns often constrain the space for policy options. But equally, budgetary confidence may inspire generosity. So when the Chinese economy had recovered from the effects of the late 1990s Asian financial crisis, and revenues grew, the party-state found itself in a better position to subsidise programmes, especially in poorer parts of the country (Brys et al. 2013). Government revenue in 2001 surged by 22%, the highest rate since 1993 (Zhu, 2002), 10 and in 2002 the central party-state announced funding compensation for 'tax-for-fee' reform as well as government spending of NRCMS. 11 Although we challenge 'threat of rebellion' explanations of China's twenty-first century public goods provision, we do not argue that Chinese leaders are unconcerned by challengers. The central party-state certainly was (and remains) very concerned about threats to its rule, and as many others have shown, uses a range of increasingly sophisticated methods to identify dissatisfaction, pre-empt and suppress protest in both urban and rural areas, emasculate any opposition and shut down criticism. But not all China's policies are aimed directly at maintaining its one-party rule. of government areas, from foreign affairs, the economy, banking, trade, energy, science and technology, through to cultural affairs, education, population, infrastructure investment, forestry management, environment, irrigation, social security and health. While some of these policies (and other decisions) help sustain CCP rule by preferentially benefitting its core elite, many others do not. In our specific case studies, we show that the party-state monitored rising protest in the post-Mao period, and in the 1990s and early 2000s it interpreted rural protest as catalysed by local government predation, especially fee extraction. The 'threat of rebellion' therefore was a factor in the calculations of top leaders when they reformed the rural tax system, though it was only one among several. It was an even less important factor in relation to rural health and pensions reforms. Official documents sometimes noted that these policies contributed to maintaining 'social stability', but there is no evidence that leaders perceived protest to be fuelled by demands for health schemes or pensions (O'Brien 2009, pp. 25-9;Saich 2014). When in the 2000s local government land grabs started to generate protest, the State Council did, however, encourage local governments to ensure that rural dwellers who had lost their land had access to existing social security provisions (MoLSS 2006).
As well as top leaders, we show the influence of actors usually excluded from authoritarian regime theories that focus on individual dictators. We reveal how-even if they are bureaucratically dominated-policy communities can be influential in autocracies. China's ministries, local governments, journalists, academic researchers, and the domestic and international organisations they were working with put rural social policies on the agenda, and developed policy ideas and options through debates, research and experimentation.
In discussing the influence of these actors, we emphasise the importance of policy ideas in public goods provision. Ideas are now well established as policy influences in democracies, but remain neglected in authoritarian theory, even though other comparative work has acknowledged their influence-at least in communist regimes. Stephan Haggard and Robert Kaufman for example have noted that 'Communist regimes radically redistributed resources not because of their responsiveness to underlying social pressures, but because of political and ideological commitments to a socialist transformation' (Haggard & Kaufman 2008, p. 357). Similarly, Michael Miller has acknowledged that 'Communist regimes have highly distinct policy platforms' (M. K. Miller , p. 1543. But instead of incorporating the influence of ideas into authoritarian public goods provision, some have dismissed 'communist regimes' as exceptions or even excluded them from its models (Acemoglu & Robinson 2001, p. 939).
We are able to show that specific policy ideas are connected to the CCP's wider claims to rule and that neither should be ignored. The ideas and options developed by specific policy communities-around rural or health policy for example-were much more likely to rise up the policy agenda when they offered solutions to economic problems. This is because there has been an elite consensus through the post-Mao period that promoting economic development is high priority (Naughton 2015, p. 165). Top leaders and the wider party-state centre (the Party Central Committee, State Council and ministries) have informally staked the CCP's claims to rule-its legitimacy-on its ability to deliver national economic growth. Hu and Wen extended those claims to rule by incorporating social development into the national agenda (Holbig 2009). China's top leaders apparently fear not just threats by distinct groups of challengers, but a more diffuse undermining of their legitimacy and this has encouraged them to support policy innovation in tandem with continually evolving their claims to rule.
External shocks have long been seen as important in democratic policy making but also remain excluded from analyses of authoritarian public goods provision. We show that ideas about the Asian financial and global crises and solutions to them were important in developing not only economic but also social policy. Another external shock, the 2003 SARS crisis, played a role, particularly for NRCMS but also for other social policies, by showing the dangers of their neglect. It appears to have transformed the Hu and Wen leadership's views on national development-encouraging them to see not only economic but also social development as important. Talking about the influence of the SARS crisis on the government, Wen has said, for example, that '[d]uring the fight against SARS in 2003 we received much enlightened guidance, among which the most important was that it is necessary to plan economic and social development as a whole' to enable balance between the two (Wen 2010).
Finally, political institutions also played a role in the expansion of public goods, though not in the way that theories of authoritarianism anticipate. Authoritarian regime theories are institutional in the sense that they see autocratic institutions as having distinctive effects on politics-including public policies and resultant public goods provision. With their focus on dictators, however, they tend to assume that institutions play a certain role-for example by making those dictators sensitive to the demands of their narrow 'selectorate', and insulating them from the demands of wider society. While political institutions in China did indeed limit the influence of rural dwellers and other civil society actors, they also had other effects, notably shaping the access of bureaucratically dominated policy communities and incentivising them to generate new policy ideas and options.
Our study focuses on only one authoritarian regime: one with a communist party in power and an extensive bureaucracy. This might put it in a sub-group of communist or other authoritarian states. And China may be unusual in its longevity, in its people's strong sense of nation, and in having been mostly spared from the ravages of colonialism. These factors may have contributed uniquely to shaping its institutions and giving bureaucratically dominated policy communities and their policy ideas influence in public goods provision. As a result, our arguments may be limited to China itself. We think this is unlikely, but we are more interested in explaining political and policy outcomes than in categorising the Chinese political system. Having demonstrated the importance of policy communities, external shocks and ideas in such outcomes in one authoritarian regime, we encourage others to explore their influence in others.