The Introduction of Pension, Accident, and Health Insurance in Urban China

Abstract

Since the initiation of reform and opening policies, social protection for urban workers in China has transformed massively. Before the 1980s, state-owned enterprises were responsible for protecting workers from social risks such as old age, accidents, and illness. Today, these three areas are organised as contribution-based social insurance systems with Chinese characteristics. This chapter identifies the causal mechanisms that led to the introduction of insurance schemes in the 1990s and early 2000s. We find three causal mechanisms: (neutral and strategic) policy experimentation, top-leader intervention, and (consensus-based and enforced) elite cooperation. Moreover, we demonstrate that the presence or absence of complementarity between the international environment and the domestic actor constellation had a decisive effect on how those mechanisms played out.

This chapter is a product of the research conducted in the Collaborative Research Centre 1342 “Global Dynamics of Social Policy”, funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

1 Introduction

A previous version of this chapter was published as a working paper by the SOCIUM Research Centre on Inequality and Social Policy and the Collaborative Research Centre 1342 “Global Dynamics of Social Policy”, University of Bremen (ten Brink et al. 2020).

This chapter aims to explain the introduction of three types of contribution-based social insurance in the People’s Republic of China (PRC). We focus on the explanation of three significant policy events, in particular, the introduction of the 1997 Urban Employees’ Basic Pension Insurance, the 1998 Urban Employees’ Basic Medical Insurance, and the 2004 Work-Related Injury Insurance. These three events each represented an important milestone in a longer reform process, resulting in a redesign of the respective insurance models. Furthermore, as Fig. 2.1 illustrates, the three events were all followed by substantial increases in coverage.

Fig. 2.1

Participation in urban employees’ insurance schemes 1989–2008. (Sources: MoLSS 2001, MoHRSS 2018; CHFP 2018; MoH 2005)

To explain the three policy events, we build on recent advances in mechanism-based explanations in political science and sociology, detailed in the Introduction (Chap. 1) to this edited volume (also see Goertz and Mahoney 2012; Hedström and Ylikoski 2010; Kuhlmann et al. 2020). To unpack the causal mechanisms that facilitated the introduction of the three social insurance schemes under scrutiny, we reconstruct the reform dynamics in each national policy subsystem. We start with the initial conditions, that is those structural trends in the economy or the fiscal system, reform pressures arising from demographic change, and other threat perceptions that triggered the reform process in the first place. We then trace relevant actor patterns, denoted in this chapter as activities; key sector-specific domestic events such as compromises between competing actor groups or camps; and events of interdependence, indicating significant international influences on the domestic policy process. Particularly the latter entail the existence (or not) of complementarities between national and international factors, such as established international insurance models serving as domestic frames of reference. The causal mechanisms we analyse mostly operate at the meso-level of organisational actors. We especially focus on ministries, the State Council, as well as the key political leaders within these organisations, and on national-international interaction rather than local issues.¹

In a nutshell, our policy analysis finds three key mechanisms which can plausibly help to explain the three respective policy events and their commonalities and differences. First, a policy experimentationmechanism fostered learning from domestic and international experiences, by testing different variants of insurance schemes at the local level. There are two forms of this mechanism in our case: On the one hand, in a context of political conflict, the policy experimentation mechanism is more strategic, used to either promote different policy options, as in the example of pension insurance, or to delay policy implementation, as in the example of health insurance (see Heilmann 2008). In the absence of conflict, on the other hand, it is a more neutral means of gaining information and learning (work accident insurance).

Second, we find an elite cooperation mechanism. Again, we find two versions: One that originated in a long-term and broad ideational consensus among the governing party-state elite (work accidents) and another that was based on an enforced compromise between competing groups within the party-state (pensions).

The third mechanism we find comes into play only in the health field: The top-leader intervention mechanism. This mechanism involves the attempt by an individual actor within the administration to promote a new policy direction. Moreover, we observe that the presence or absence of complementarity between the international environment and the domestic actor constellation had a decisive effect on how the policy experimentation mechanism and the elite cooperation mechanism in particular played out.

In the next section, we first provide a short historical background and inform about the existing academic literature that our analysis builds on. In the main part, we then reconstruct the key causal mechanisms in each of the cases. In the conclusion, we summarise the findings, present our contribution to the literature, and identify the limitations of the study.

2 Historical Background and Research on China’s Social Security System

The People’s Republic of China (PRC) fundamentally restructured its social security system during the transformation from a command economy into a state-permeated type of capitalism (ten Brink 2019). Before the 1980s, state-owned enterprises (SOEs) were responsible for protecting workers from social risks such as old age, accidents, and illness in urban China (danwei system; see Dillon 2015). But although reform pressure was already high in the 1980s, and many reform proposals were made in the early phases of this transformation, it was only in the 1990s that serious steps were undertaken to adopt a comprehensive social insurance system in urban China.

In the 1990s, in the midst of what China scholars call the second phase of reform (see Naughton 2007), the objective of the party-state leadership was to transform the remnants of the old eroding social security system (especially the danwei system in the cities) into a contribution-based social insurance system partially based on Western and East Asian models. From the 2000s onwards, the aim increasingly became the implementation of universal insurance schemes, and new policies were adopted to cover the rural population and informal workers (see Ngok and Chan 2016; Zhou 2017; Zheng 2008).

Thus, our three case studies in fact only cover a small, yet important segment of a much larger, and much more complex, process—in which social insurance institutions have been established as parts of the broader structure of a new welfare state. This process of building a welfare state in China is ongoing, albeit at different speeds and with large variation—when it comes to system design, coverage, and generosity, for instance. Indeed, in this chapter, we merely examine social insurance institutions for urban Chinese workers with a formal employment contract. The majority of the population in the 1990s, which either lived in the countryside or as migrant workers in the cities, had only very limited state-run social protection. Social security reform for the rural population and informal workers was initially organised into different systems, and mostly came at a later point. It is important to note that the urban-rural divide so typical for reform China was already established in the 1960s with a strict system of household registration (hukou) that ascribed people either a rural or urban residency status. This restrictive household registration system perpetuated the disparities between urban and rural incomes, standards of living, and social safety nets.

With the establishment of a contribution-based social insurance system with Chinese characteristics, an academic literature emerged which has placed a spotlight on many of its key features and development trajectories. This holds for socioeconomic studies, which stress the effects of market-oriented reforms and related waves of redundancies, and for a social security vacuum which made it necessary to introduce employer-independent contributory pension and health insurance (Liu 2002; West 1999; Whiteford 2003). Additionally, rapid demographic change and a declining significance of the family facilitated the expansion of social policy (Fong 2004; Ngok and Chan 2016). Besides these arguments, political-institutional approaches also focus on the capacity for action among party-state actors, on interest groups, and on other political factors (Duckett 2001; Frazier 2010; Müller 2016). The success or failure of experimentation under hierarchy, that is of local pilot projects, also influenced the reform processes, and previous research has already highlighted the neutral and strategic facets of policy experimentation (Heilmann 2008; Shi 2012). Further studies show a causal relation between social protests and deficits in the legitimacy of the Chinese authorities, and social policy reform (Solinger 1999).

Notably, the existing literature on social insurance reform in China largely refers to national influences as explanatory factors. To capture both national and transnational influences, we tie in with an emerging literature that investigates international influences on the formation of social insurance schemes in China (see Hu 2012; Dillon 2015). For instance, Duckett (2019) reveals the role played by international events and organisations on China’s New Rural Cooperative Medical Schemes, and Leisering et al. (2017) demonstrate how Chinese actors have synthesised disparate ideas from the United States and Western Europe with Chinese traditions to create a distinctive model of social assistance (dibao), thereby complementing the literature on policy diffusion. Moreover, we also build on mechanism-based explanations in the research on Chinese social policy, which has been gaining ground recently (see Müller 2016).

3 Causal Mechanisms in the Introduction of Urban Basic Pension Insurance

In 1997, the Chinese government introduced a new urban employee basic pension insurance framework under State Council Document No. 26, with the goal of adopting a unified basic pension system for all urban employees. The 1997 decision united a system that had previously been highly fragmented. While a basic model of mixing social pooling and individual accounts was already decided by the State Council in 1991, resulting in some coverage increase, it took six years to overcome a heavily localised system with numerous mixes of social pooling and individual accounts. The 1997 unification also triggered a relatively successful rise in coverage.

The causal chain leading to the adoption of a mixed model of social pooling and individual accounts was dominated by two mechanisms. First, since the early 1990s, strategic policy experimentation to promote different model mixes stimulated learning. However, persistent conflict between the two prevailing camps delayed the process of national policy unification. Second, strong complementarities between the domestic pressure to reform SOEs, a centrepiece of economic reform from the mid-1990s onwards, and international influences especially by the then dominant World Bank pension reform model caused a political compromise on basic pensions. An elite cooperation mechanism emerged that fostered policy unification. Moreover, strong national-international complementarities fuelled further reform efforts thereafter (see Fig. 2.2).²

Fig. 2.2

Causal chain pension insurance. (Source: Own presentation)

3.1 Initial Conditions and Threat Perception

Chinese economic reform in the 1980s put the necessity to change pension policy onto the agenda. China had established its urban pension scheme in 1951 as part of its labour insurance system, following the Soviet model with insurance financed solely by enterprises, and therefore largely for SOE employees. Obviously, the old system was no longer functional. In the 1980s, a large proportion of the urban population was exposed to the risks of having no old-age coverage at all, especially in the emerging non-state enterprise sector. “In addition, enterprises supporting a large number of retirees were disadvantaged in market competition, with some finding themselves on the verge of bankruptcy” (Li 2014, 283). The old enterprise-based provision of pensions served as an obstacle to both SOE competitiveness and labour mobility (West 1999).

The field of pensions constituted a problem for social stability from early on in the reform period. Around 1980, there were already two million Chinese workers who had reached pension age but were not allowed to retire by their employers (Chen and Wang 2010). The containment of rising pension costs mostly for SOEs also was high on the agenda at an early stage. Additional reform pressure had arisen from rapid demographic change (West 1999; Whiteford 2003).

In the 1980s, however, different and competing agendas in the responsible government authorities made serious reform difficult. There was an agreement on the underlying rationale to maintain social order and political stability and, as a developing country, to keep the benefit level as low as possible to avoid financial burdens on the party-state. Regarding the design of new schemes, however, there was substantial disagreement. Consequently, policymakers and policy intellectuals learned from competing international models that coexisted at the time, with no one having a clearly dominant position. On the one hand, the Ministry of Labour (MoL) was eager to learn from Western (German) models of social insurance, and more particularly from a pay-as-you-go system, in which the current cohort of contributors pays for pensioners through contributing to a pooled social insurance fund. The MoL was closely connected to the International Labour Organization (ILO) since the 1980s, and an ILO office in Beijing was opened in 1985.

On the other hand, the National Economic System Reform Commission (ESRC, a ministry-level authority, and precursor of the National Development and Reform Commission) was attracted to the global epistemic community of (neo)liberal economists and eager to learn from the World Bank. The World Bank was promoting an individual accounts-based pension system as a means to facilitate SOE reform. Intellectual and technical support was provided, for instance the formulation of comprehensive actuarial projections. This promotion of an alternative to a mostly state-run social insurance pension was complemented by the attraction of the Singapore model, in essence compulsory savings accounts for the purpose of a pension. Resembling developments in health (see Sect. 5), the inclusion of the Singapore model on the policy agenda also made it possible to interpret this as a creative mixing of Western and non-Western models and to thereby appease domestic critics of Western models.

From the mid-1980s onwards, the MoL promoted social pooling of pension funds to replace the work unit as the focal point of the social basic pension pillar and to even out dissimilar burdens of enterprises. Experimentation with local social pooling began in a few selected provinces. The ESRC supported experiments as early as July 1989, in Fujian province, which “set up individual accounts on top of social insurance for workers” in non-SOEs (Hu 2012, 620). Interestingly, already in 1991, a basic model of mixing social pooling and individual accounts was promoted by the State Council (Guofa 1991, 33). From then on, the battle was about the “right mix” between social pooling and individual accounts. In the following, we identify two causal mechanisms that have significantly impacted the process.

3.2 Mechanism 1: Strategic Policy Experimentation

Since a number of important players were involved in the competition of ideas and influence in the pension field, and functional authority was overlapping, the debate over suitable reform measures was intense, especially between the MoL and the ESRC. From 1989 onwards, each institution supported numerous local experiments. Policy experimentation here had the following aspects: The local pilots typically were useful for gaining information (“neutral” experimentation) and learning, for example, about different combinations of social pooling and individual accounts. However, most of those experiments also had a strong strategic aspect. They were a means for promoting different policy options in a situation in which both the MoL and the ESRC lacked the political support to produce a collective, binding decision at the national level.

In 1993, the New Socialist Market Economy initiative confirmed the mix of pooling and individual accounts the State Council had already promoted in 1991 (Zhonggong Zhongyang 1993). All local experiments now had to be variants of this model. The MoL’s “pro-pooling” experiments, based on ideas from the ILO, favoured a small individual account (equal to 3–5% of payroll) on top of a larger social pooling component, a model that was later coined the “Guangdong” or “Hainan model”. Conversely, the ESRC’s “pro-individual account” approach led to strategic experimentation with larger individual account components (equal to 16–17% of payroll), later coined the “Shanghai model”. The ESRC branded this approach as a middle way between planning and free market, and also stressed Southeast Asian influences, by emphasising and transferring the Singaporean experience (Gong 2003; Hu 2012). It also had the support of national leader Zhu Rongji, whose promotional efforts however were resisted by the MoL, which was “probably supported by Vice-Premier Zhou Jiahua” (Gong 2003, 247), and thus had a different effect than in the health system where top-leader intervention strongly impacted on the policy process (see Sect. 5).

In 1995, the insights from strategic experimentation were incorporated into State Council Circular No. 6, which required provinces to select one of the two models mentioned above to establish local basic pension schemes. The result was system fragmentation. Both models involved individual accounts and social pooling; however, their proportion was different (larger vs. smaller individual accounts) and system components could be combined and implemented in diverse ways. While in almost all provinces individual accounts now were adopted, it led to the creation of hundreds of schemes all over China, significantly handicapping labour mobility, for example, from SOEs to the non-state sector (Béland and Yu 2004). The Circular also failed to achieve an agreement about the transfer of employer contributions and “delegated local authorities to decide the detailed transfer plan” (Gong 2003, 87). In practice, both benefit and contribution rates varied significantly.

In summary, strategic experimentation led to policy learning and important changes in the urban pension system. However, it also fostered fragmentation due to unresolved conflicts regarding the right pension mix, resulting in coverage stagnation. Another causal mechanism, the elite cooperation mechanism based on a productive complementarity between domestic and international factors, however, fostered another change of direction shortly afterwards.

3.3 Mechanism 2: Elite Cooperation Through Enforced Compromise

If not to the same degree as in health (with implementation delayed by strategic experimentation and substantial implementation gaps at the local level), experimentation with pension schemes still contained enough conflict and opposition between key stakeholders to hinder a reform breakthrough. The second mechanism we identify here however caused a real step forward. Under pressure to reform the SOEs and given a dominant international model for pension reform with enough leeway for compromise upon its transferral to China, elite cooperation and consensus finding developed.

With economic reform accelerating again after 1992, the reform of SOEs soon became a key objective. Increased competition, not only created by the rise of the non-state sector but more and more also by international investors, resulted in a drop in profit margins, particularly for local SOEs, and rising debt levels (Naughton 2007, 304–05). At this point, the government’s economic policy strategists developed a plan to address the situation that went down in the annals of history as “grasping the large, releasing the small” (zhuada fangxiao). From 1995 to 2000, three-quarters of all local SOEs had been subject to restructuring measures ranging from rationalisation to formal privatisation. A dramatic reduction of state-sector employment (as well as in urban collectives) resulted, which led to a serious “conflict over how to distribute the social costs of SOE reforms” (Frazier 2010, 19).

SOE reforms thus heavily affected, and indeed accelerated, urban pension reform. According to leading policymakers, the unification of pensions across enterprises of various ownership forms and the active participation of the non-state sector with a younger demographic structure could reduce the overall pension costs for SOEs (Gong 2003). Additionally, benefits for pensioners would stimulate domestic consumption more than that of dismissed older workers with lower propensities to consume. Most importantly, unified pension schemes for different enterprise forms, industries, and localities promised higher labour mobility, “particularly from the state sector with privileges to the non-state sector” (Gong 2003, 202).

This strong pressure to move ahead with pension reform was complemented by a change at the international level. Here, the former coexistence of competing pension models was replaced by the advent of one dominant model, propagated inter alia by the World Bank. Its “three-pillar” model now served as the dominant international reference frame. In particular, in addition to the 1994 publication of the book Averting the Old Age Crisis (World Bank 1994), a comprehensive report on China’s Pension System Reform (World Bank 1996a) was published after a mission to China. By criticising the fragmentation of the 1995 system, it became a key reference thereafter and was widely read at the highest levels of the party-state, including by President Jiang Zemin himself (Frazier 2010, 60).

This “event of interdependence” facilitated domestic compromise between the promoters of social pooling and the promoters of individual accounts (see Fig. 2.2).

Upon the World Bank’s criticism on the fragmented pension system, … the State Council required the ESRC and the Labor Ministry to investigate provincial pension systems to find a way to unify them. The two ministries organized a research group … According to the research group, most of the localities agreed to unify the pension system. (Hu 2012, 630)

A new unified model was eventually agreed upon. The State Council Document No. 26 of 1997 on the Establishment of a Unified State Pension Programme for Enterprise Employees required all enterprises to accept the same contribution and entitlement rules. As a compromise, a medium-sized individual account was established.³ Individual accounts, equal to 11% of payroll, would be funded by both enterprises (3%) and employees (8%). Social pooling was funded by enterprises, which could not contribute more than 20% of the total wage as pension funding. It was also promised that the financial burden of pension commitments would be more evenly spread across different enterprise forms (Béland and Yu 2004; Li 2014). While the compromise entailed a strong role for individual accounts, all sides could eventually live with it (on the interpretative leeway in World Bank proposals that left enough room for compromise between the opposing domestic camps, see: World Bank 1996a, 19–31).

Accordingly, the party-state elite could be unified. Rather than the previous inter-ministry conflict, elite cooperation at the national level and between central and local authorities now became effective. Notably, this constellation was prolonged. The unified pension policy paved the way for geographic unification, and until 2000, “state enterprises, collective enterprises, foreign-invested enterprises and private enterprises were all part of the system, which was further fine-tuned later” (Li 2014, 286). In principle, every urban employee with stable, contract-based employment was now covered by some sort of pension scheme (the self-employed were only able to contribute after 2003, on a voluntary basis). And indeed, it became easier for many workers to move between localities, as pension contributions were now managed at the provincial level.

What made for this prolongation and deepening of reform? In a nutshell, the elite cooperation mechanism was reinforced through another event of interdependence, namely the 1997/1998 Asian Financial Crisis (which caused a decrease in demand), and an upsurge in pension protests. The protests were based on the still poor implementation of pension policy leading to delayed or reduced pensions for instance, and one incident was particularly relevant as it shocked the highest echelons of the party-state: On a tour to North East China early in 1998 to visit ongoing SOE reform efforts, a special train carrying then Premier Zhu Rongji and several ministers “was blocked by local pensioners. Some of them lay down on the rail track to plead for a regular payment of pensions. It was a real shock” (Gong 2003, 193).

In conclusion, after the mid-1990s, a “golden opportunity” (Gong 2003, 190) for a basic pension reform compromise emerged. A favourable domestic constellation matched with an international policy environment that provided an elastic reference frame for erstwhile conflicting camps.

4 Causal Mechanisms in the Introduction of Work Accident Insurance

This section investigates the trajectory of work accident insurance in China and the introduction of the Regulations on Work Accident Insurance by the State Council in 2004. While some provisional regulations in this field had been in place since 1996 such as Trial Measures of Work Accident Insurance for Employees, and coverage increased slowly from 1993 onwards, the 2004 Regulation represented a milestone with substantial impact: Through its inclusion of non-SOE employees, and migrant workers (at least theoretically), coverage in this scheme rose from 45 million in 2003 and 85 million in 2005 to 239 million workers in 2018 (MoHRSS 2019).⁴

The causal chain leading to the adoption of the regulations on work accident insurance in 2004 was dominated by two mechanisms: First, beginning in the late 1980s, the neutral policy experimentation mechanism stimulated learning on introducing insurance schemes based on social pooling. Second, the elite cooperation mechanism brought about by a broad consensus among scholars and officials on the design of a work accident insurance scheme facilitated reform since the late 1990s. Strong complementarities between Chinese and international epistemic communities helped to further reinforce elite cooperation and consensus building. A dominant international model of public social insurance as opposed to commercial insurance schemes was especially relevant in this field.

Compared to health and pensions, however, the introduction of work accident insurance came several years later. Why? We argue that reform efforts were halted in the mid-1990s because reform dynamics in this field were not perceived as crucial to social stability. Since the end of the 1990s, finally, the re-employment of the two mentioned mechanisms mobilised enough drive to nationally adopt the insurance scheme—elite cooperation was further encouraged by wider public debates on work injuries, and new experimentation especially fostered learning of technical details, with Germany’s system as a reference frame and the capability to locally adapt these (see Fig. 2.3).

Fig. 2.3

Causal chain accident insurance. (Source: Own presentation)

4.1 Initial Conditions and Threat Perception

Since the adoption of the economic reform, labour insurance provided by SOEs and collective enterprises had eroded in China (Hu 2009, 322–77). Protective measures against work-related injuries, accidents, and occupational diseases were fundamentally weakened as well. Although in the 1980s some SOEs still provided some compensation once work accidents had occurred, however, the benefit level for compensation was equivalent to the 1950s’ level and could not cover the real needs of households with family members suffering from accidents. Non-state enterprises regularly offered no protection at all. If a severe work injury occurred, many enterprises were unable to bear the huge costs for compensating injured or dead workers. Some employers even fled with their money and escaped their obligations to compensate or had to declare bankruptcy after the occurrence of deadly work accidents (Hu 2009).

Lack of protection frequently resulted in serious disputes throughout the 1980s: “Disturbance related to work accidents” (nao gongshang) was a euphemism for widows and relatives publicly displaying the corpses of dead workers and blocking companies’ production through verbal quarrels and conflicts with employers, aiming at receiving higher compensation. Enterprises in some high-risk sectors, such as mining, were especially at risk of falling into crisis when deadly accidents occurred (Hu 2009). Taken together, these developments led to a new demand for social protection against work accidents.

4.2 Mechanism 1: Neutral Policy Experimentation

The search for a work accident insurance programme began in the late 1980s, at around the same time as Chinese officials and experts started to explore a health insurance and a pension insurance system. The MoL began to research the matter in 1988, and first experiments were conducted. The process was complemented by regulations on occupational diseases by the Ministry of Health (MoH), as well as disease-specific prevention programmes and sector-specific safety regulations. In 1990, the decision to create an insurance system was integrated into the eighth five-year plan. Moreover, government officials from the MoL searched for a social protection system beyond individual enterprises and preferred a social pooling scheme that was able to collect and pool resources to compensate for injured workers. Notably, from the beginning, transnational interdependence influenced the process. Western social insurance models that pool work-related risks of individual enterprises were especially used as a reference model for establishing a work accident insurance in China (Zheng 2008, 269).

At the end of the 1980s and into the 1990s, various counties and cities in the provinces of Liaoning, Fujian, Hainan, and Guangdong piloted work accident insurance schemes based on social pooling. One of the most successful and thus famous pilot projects at that time was located in Tieling (Liaoning), which already achieved a social pooling scheme at city level in 1991 (Hu 2009, 334). Since the 1990s, similar pilot projects have been further promoted nationwide, mainly designed by the MoL, its local counterparts, and local governments.

In comparison, particularly to piloting in health insurance, experimentation here was mainly to gain information about the operational experiences of a pooling-based accident insurance model. Due to the low conflict intensity between ministerial actors in this policy subsystem, local pilots could be mostly utilised to collect practical implementation experience. Notably, strategic experimentation with substantially different models like in pension or health insurance was the exception to the rule. Only a few actors promoted commercial insurance. When in the early 1990s some officials suggested that the management of work accident insurance could be transferred to commercial insurance companies and some insurance companies provided pilot programmes for employers’ liability insurance, the MoL stepped in and corrected this approach with a firm attitude. Backed by strong support among experts, a consensus emerged that work accident protection should be based on social pooling, not commercial firms (Liu and Leisering 2017, 118). Throughout the 1990s, the debate on a private work accident insurance gradually became weaker in mainstream discourse.

4.3 Mechanism 2: Elite Cooperation Through Consensus

A second aspect that contributed to the policy event of 2004 was elite cooperation, based on a long-term and broad policy consensus. As mentioned above, the initial consensus in the early 1990s set the reform direction towards a pooling-based insurance. Another complementary consensus emerged at the turn of the millennium. A group of key figures from the new Ministry of Labour and Social Security (MoLSS, founded in 1998), the think tanks of the MoLSS, and some elite universities like Renmin University constituted major “pushing hands” of this consensus building. This group co-constituted a “modern” and “rational” work accident insurance system in the early 2000s, which was to reshape the configuration of traditional work accident insurance and move the emphasis of this system from merely monetary compensation to prevention measures and comprehensive recovery programmes after the occurrence of work accidents. Chinese elites began to orient themselves strongly towards the German model of statutory work accident insurance with its emphasis on prevention, compensation, and rehabilitation (Liu and Leisering 2017).

A dominant international model for the feasibility of a public social insurance scheme served as a key reference frame for long-term consensus. The ILO with its partner organisation, the International Social Security Association (ISSA), played a key role here. Both organisations were also strongly influenced by the German model of work accident insurance and both have promoted its worldwide diffusion. Scholars and experts from European social insurance states like Germany, France, and Switzerland constituted the strongest “expert community” in these organisations (Liu 2018). The World Bank did not promote an alternative model in this field.

4.4 Reform Delay: From the Late 1990s to the 2004 Regulations

Although the area of work accident insurance is an arena with low intensity of contestation in comparison to health for instance (see Sect. 5), it was adopted comparatively late. In a nutshell, reform efforts were halted in the mid-1990s because reform dynamics in this field were not perceived as crucial to social stability. In contrast to general social risks like old age and illness, the risks of work accidents are rather contingent and specific, that is, occupation related. Especially workers in high-risk sectors such as mining are exposed to work accidents. Indeed, in the mid-1990s, work accidents were not perceived as an acute threat for society as a whole or as a challenge to social and political stability. For these reasons, it was not put at the top of policymakers’ agenda by the mid-1990s, when health and pension reform absorbed many political capacities. The elite cooperation mechanism thus lost momentum for some time.

However, a threat perception was back on the agenda in 2002/2003, this time with enough pressure for the elite cooperation mechanism to play a role. An increase in work accidents, especially in mining, attracted a great deal of attention—with increased coverage by both national and international media on the suffering of miners. A “naming and shaming” effect, with increased communication among Chinese officials and scholars on these issues, finally accelerated the adoption of the regulations on work accidents insurance in 2004.

5 Causal Mechanisms in the Introduction of Urban Basic Health Insurance

In 1998, the State Council issued a decision on the Urban Employees’ Basic Medical Insurance. It resembled the decision on pension insurance in 1997. Most notably, both decisions made a mixed model of social pooling and individual savings accounts the national norm. But overall, the coverage of health insurance remained low far into the 2000s for two reasons: First, health insurance appeared on the agenda later than pension insurance and rose in priority only in the context of crisis episodes. Second, due to bureaucratic conflict, the expansion proceeded overall more slowly in the 1990s.

The causal chain leading to the adoption of a mixed model of social pooling and Singaporean-style Medical Savings Accounts (MSAs) was dominated by two mechanisms. First, the introduction of MSAs came onto the policy agenda through a mechanism of top-leader intervention. Second, various ministries⁵ attempted to get it off the agenda again by delaying the process through strategic experimentation with alternative models. Moreover, there was an absence of complementarity with the international environment in healthcare due to the lack of a neat, simple, and practical model induced by the complexity of the policy subsystem (see also: Weyland 2007) and the minimalist universalism the World Bank promoted at the time, which contributed to the overall lower coverage of health insurance.

5.1 Initial Conditions and Threat Perception

Health reforms for cost control appeared on the policy agenda very early in the reform process. The introduction of user fees actually dates back to the Cultural Revolution period, when the central government ordered several rounds of price decreases in healthcare to make health services more affordable. During this period, hospitals were allowed to run their own pharmacies and charge user fees for the drugs sold, part of which they could keep as a profit to fund their operations. Starting in the late 1970s, the Ministry of Health moved towards allowing individual practitioners in basic medical services (Dong et al. 1999; Li 2008).

In risk protection, the Ministry of Health (MoH) oversaw a gradual and stepwise reform approach in the two main protection systems for urban residents: Labour Insurance for workers (under the jurisdiction of the MoL), and Government Employee Insurance for civil servants and employees in government and administrative organs (under its own jurisdiction). The latter provided very generous benefits funded from government budgets, and the former provided less generous benefits for workers in SOEs and some collective enterprises. For the most part, these initiatives were dominated by cost control measures and the introduction of co-payments. In 1988, the State Council asked a group of eight ministries led by the MoH to experiment with reforms of the health insurance systems. Two lines developed under this initiative: First, special health insurance systems for pensioners; and second, pooled insurance for serious illnesses of employees, which was tested in four pilot localities in 1988.⁶ The effectiveness of the latter line of reform on cost containment was disputed (Duckett 2001, 2011; Zheng 2008).

The Tian’anmen protest and its suppression in 1989 became a game-changer for health reforms. Large protests in urban China had substantial national and international repercussions. Within the PRC, the leadership of the Communist Party was suddenly sensitised for the destabilising potential of urban unrest and became more careful with economic and labour market reforms. Research on workers’ opinions in the aftermath of 1989 revealed that healthcare was a major concern for them. As Duckett (2011, 137–38) reports, this substantially raised central government’s threat perception with regard to the field of healthcare. Under these circumstances, the previous line of health reforms under the MoH became less acceptable. In particular, the MoH’s reforms were seen as ineffective in containing escalating health costs and too protective of hospitals and the medical profession in profiting from rising costs, at the expense of workers’ social security. Furthermore, it was clear that if SOE reforms were to be broadly implemented, the protection of workers’ health risks would need to be taken care of.

The year 1989 also saw a significant event of interdependence between national and international factors, which changed the course of development. The international repercussions of the Tian’anmen protest were far-reaching, including a widespread diplomatic isolation and a temporary stop of ongoing development projects. Beijing began to intensify its relations with developing countries as a response, and it began emphasising non-Western models of development. Singapore was one of the clearly identifiable success cases of development in Asia that had preserved an authoritarian political system. Its welfare practices were dominated by the provident fund model of mandatory individual savings accounts. The case of healthcare here is particularly insightful, because Singapore’s MSAs were a singularity, whereas capital-based schemes were more common in a global context.⁷ So after 1989, Party leaders’ threat perception in healthcare rose, while at the same time the Singapore model became a more attractive and politically acceptable policy option (Hu 2012).

5.2 Mechanism 1: Top-Leader Intervention

Fundamental policy change was soon underway. In 1991, liberal politician and previous mayor and party-secretary of Shanghai Zhu Rongji was appointed Vice Premier of the State Council. The year before, he had gone to Singapore to investigate the provident fund⁸ and became its leading advocate thereafter. The State Council decided that pension insurance would henceforth combine a social pooling component with individual savings accounts. In 1992, Deng Xiaoping gave a strong signal for returning to the course of economic reforms after the short interregnum that followed 1989, which meant that SOE reforms were back on the agenda. In the same year, the authority over health reforms at the State Council was reshuffled: The Ministry of Health lost its leadership role and was replaced by the ESRC for a brief period before the MoL took the main initiative (Zheng 2002, 135).

Zhu Rongji directly intervened in the health reforms, pushing the Singapore model. The ESRC was a forceful institution in favour of the Singapore model in pension reforms, but in healthcare, Aitchison (1997) reports that all ministries in the leading small group on health reforms disliked Zhu’s reform plan and subsequently negotiated him down to prolonged policy experimentation. At the same time, as we will see below, the inter-ministerial competition between the MoH and the MoL arguably weakened their bargaining position. Pilot projects joining MSAs and social pooling subsequently developed under the jurisdiction of both ministries, while at the same time, the implementation of MSAs in other insurance programmes and localities was often neglected.⁹

5.3 Mechanism 2: Strategic Experimentation to Delay

There was a significant difference between the pension and health policy fields: In pensions, there were already a substantial number of countries applying capital-based models of old-age protection, and the World Bank also began to promote a mix of methods in the 1990s. In healthcare, however, capital-based social protection was an exception, and the international policy subsystem was chaotic in the sense that there was no consensus about an optimal approach to protection (Howlett and Ramesh 1998). While there was a general normative consensus on universal access and universal coverage, there are multiple ways to realise this with fundamentally different distributive consequences. Furthermore, the systems of risk protection and service provision in healthcare are closely interdependent, which makes for a much larger number of possible models than in pensions, and there is no universally agreed-upon system of classification. Some scholars even generally doubt the utility of “models” in healthcare due to the high complexity of health systems (Bali and Ramesh 2017; see also: Weyland 2007; for an overview of existing classifications and their fit for the Global South, see: de Carvalho et al. 2021). The World Bank promoted a universalist, minimalist approach of “essential health services” funded through taxation and focusing on benefits for the poor (Kaasch 2015). But this option was fundamentally incompatible with the political aim of pacifying the privileged SOE workers precisely because of its redistributive, minimalist universalism. SOE workers would have perceived it as social decline to lose their privileged status in social protection after decades of preferential treatment. Furthermore, tax funding was a problematic option at the time of deepening fiscal crisis. The absence of a dominant international model in healthcare that could be reconciled with the divide-and-rule approach of the central government prohibited a productive complementarity between domestic bureaucratic conflict and international ideas, as occurred in pension insurance.

As noted above, at ministerial level, the MoH and the MoL both disliked Zhu Rongji’s proposal of MSAs. However, both were also competing with one another for jurisdiction over health insurance. The MoH was in charge of health protection for civil servants as well as health insurance in rural areas, whereas the MoL was in charge of labour health insurance. Shenzhen was the first city to test an insurance scheme joining social pooling and MSAs run by the local Bureau of Health in 1992. The city had received the mandate to reform social insurance from the State Council in 1989. Preparations had already begun in 1991, and the administrative organ for health insurance had already been set up at the Bureau of Health. The insurance combined social pooling and MSAs, as in the MoL’s proposal, but the local management was by the health administration (World Bank 1996b, 60–61; Zhou 2017, 137–38). Also in 1992, the MoL issued two reform proposals, one about health insurance for urban employees (zhigong yiliao baoxian) including MSAs, and the other under the name of the former MoH initiative: Catastrophic disease pooling (dabing tongchou) (See also: Zheng 2002).¹⁰ A number of pilot cities were subsequently chosen for further experiments: The MoL selected Pingdingshan in Henan Province and Chongqing, the MoH selected the city of Jiujiang in Jiangxi,¹¹ and the city of Zhenjiang in Jiangsu Province volunteered as a pilot (Aitchison 1997, 89–91). The experiments in Jiujiang and Zhenjiang gained a high profile, and the bureaucratic actors at national level reached a compromise on the mixed model of pooling and MSAs in the wake of their evaluation in 1996, ignoring explicit criticism by the World Bank. This led to the extension of pooling to 57 more cities, and the collective decision followed in 1998.

The process of experimentation delayed a broader implementation of MSAs. Throughout the 1990s, various Chinese cities operated social pooling funds, largely without the implementation of MSAs. As Aitchison notes about the catastrophic health pooling scheme in Xicheng District in Beijing around 1995:

The central government stipulated that any health care reform plan should involve individual accounts (…) Neither Xicheng district Bureau of Labor, nor Beijing city Bureau of Labor, seemed enthusiastic in implementing the individual account. In Xicheng district, the Bureau of Labor recommended that individual accounts should be set up by each enterprise separately, but did little to ensure that this was done. At the city level, the city Bureau of Labor planned to implement the individual account “in about two years” [from 1994]. (Aitchison 1997, 148–49)

Reports on similar schemes in other cities often make no mention of MSAs,¹² and the classic Labour Insurance schemes continued operating without MSAs as well. Catastrophic health pooling remained listed as a separate insurance programme in official statistics until 2000 (MoLSS 2001), indicating that the implementation of a unified model proceeded slowly even after the 1998 decision. More generally speaking, the implementation was difficult “because social insurance departments are unable to overcome vested interests and ensure the participation of enterprises and officials, audit enterprises, regulate health service providers and prevent fraud” (Duckett 2001, 302). As Fig. 2.1 illustrates, coverage picked up somewhat in 2000, and only increased more substantially since 2001—when the priority of the health system rose in general, and preparations for extending health insurance to rural China were initiated (Müller 2016) (Fig. 2.4).

Fig. 2.4

Causal chain health insurance. (Source: Own presentation)

6 Conclusion

This chapter has identified the following causal mechanisms to explain the introduction of three contribution-based social insurances in China: The policy experimentation mechanism, the top-leader intervention mechanism, and the elite cooperation mechanism. Moreover, our findings demonstrate that the presence or absence of complementarity between the international environment and the domestic actor constellation had a decisive effect on how the policy experimentation mechanism and the elite cooperation mechanism, in particular, played out.

In the urban pension insurance scheme, we observe a mechanism of strategic policy experimentation, resulting from experimental competition between “pro-pooling” and “pro-individual account” groups, which fostered learning in the 1990s. Furthermore, an elite cooperation mechanism, triggered by a forced compromise between the two competing groups who anticipated future conflicts in this field, facilitated the functional unification of a heavily fragmented system as well as coverage increase. A strong complementarity between the domestic pressure to reform SOEs and international influences, especially from the then dominant World Bank pension model which had enough leeway to facilitate compromise on its adaptive transferral to China, helped to enforce cooperation.

In the work accident insurance scheme, policy experimentation was neutral, and not strategic. Rather than experimenting with competing models or contrasting configurations of distinct models, pilot projects fostered the learning of technical details and the capability to adapt these to local circumstances. Additionally, elite cooperation through a broad consensus resulted in the comprehensive adoption of a functional insurance model and in coverage increase. A dominant international model on the feasibility of public social insurance served as a key frame of reference for long-term consensus in this policy subsystem.

In the urban health insurance scheme, top-leader intervention by Vice Premier Zhu Rongji aimed to achieve policy reform by using a hierarchical approach. This intervention, however, was opposed by the ministries involved in health reform. Implementation was delayed by a mechanism of strategic policy experimentation, and the ultimate compromise of mixing social pooling and MSAs may have succeeded only because Zhu exploited the rivalry of the MoL and the MoH regarding jurisdiction over health insurance. More research is needed to illuminate the details of this process. Throughout the 1990s, the classic labour insurance and various other urban health insurance funds continued to operate without MSAs, and coverage of the Urban Employees’ Basic Medical Insurance only grew slowly even after the collective decision of 1998. All this was facilitated by a lack of national-international complementarity due to the absence of an international consensus on specific health protection models.

Regarding the role of complementarity between the international environment and the domestic actor constellation, we find that the presence of a dominant international model facilitated domestic elite cooperation in pensions, whereas the absence of such a dominant model resulted in prolonged conflict in healthcare. Similarly, due to the existence of an established international model promoted by the ILO, policy experimentation was more neutral in accident insurance and geared towards the generation of knowledge. Political and academic elites reached a consensus on a public social insurance scheme, which facilitated cooperation. In pension insurance, domestic promoters of social pooling and advocates for individual accounts both found points of reference in the international environment. In health insurance, the situation was similar. However, pensions and healthcare differed with regard to international complementarities: In the former, a dominant international model created productive complementarity which mediated bureaucratic conflict. In this context, with strategic policy experimentation, it was possible to establish a dominant model which combined elements of the competing camps, leading to enforced cooperation between elites. In health insurance, conversely, top-leader intervention to establish a similar mixed model failed to produce enforced cooperation due to lack of complementarity with the international environment, where there was no dominant model that matched local demand. Implementation was delayed by a prolonged phase of experimentation, and coverage only grew slowly even after the collective decision.

Our chapter thereby contributes to the literature on China’s social security reform, and to an emerging literature on mechanism-based explanations of transnational social policy dynamics (Leisering 2019; Obinger et al. 2018; Orenstein 2008; but also see Benz and Dose 2010, who distinguish between different governance mechanisms that resemble those described in this chapter). Notably, our findings complement insights into mechanisms which explain the expansion of social protection programmes, such as the crisis management by going further mechanism (Chap. 3). In fact, the transition from the company protection model of Maoism to the social insurance system of the twenty-first century can be interpreted along these lines. The pressure for expansion here was strongest in pensions, as a growing number of citizens reached pension age and needed social support. As the average age of workers in SOEs differed depending on when they were founded, the introduction of pooling across public and, later, private enterprises helped consolidate social protection in old age. A similar argument can be applied to accident insurance, with the difference that the distribution of risk across companies depended more on the type of occupation (e.g. high-risk occupations such as mining). In healthcare, in contrast, risk was distributed in a comparatively random way, and only a small share of any company’s workers were affected by serious illnesses each year. In pension and accident insurance, comparatively homogeneous groups such as elderly employees or miners had some potential for collective action. In healthcare, on the other hand, the individuals affected tended to be more isolated and less concentrated in social groups. This situation reduced the potential for immediate collective action.¹³

Regarding the limitations of this study, for reasons of accessibility and space, we were unable to reconstruct the causal chains in the three case studies with all their case-specific aspects and idiosyncrasies. More research is needed, for example, on the role of the insurance industry. In future research, we would also be keen to include unemployment and maternity insurance, which constitute the two other areas of insurance that have influenced social policy arrangements in China. Moreover, the different social nature of the three forms of risk we analyse, and its implications, should be more thoroughly addressed. Last but not least, it should be investigated whether the mechanisms identified here can also be identified in other countries, especially in emerging economies with non-liberal political systems or in defective democracies. The same applies to in-depth comparative research to explore variation in the way national-international complementarities impact on social policy dynamics.