1. Introduction

Because the legitimacy of the welfare system ultimately depends on citizens’ support, it is vital to understand public welfare attitudes. Such understanding is even more critical for less-developed countries that are undergoing social policy expansion while wrestling with various challenges such as aging populations, economic volatility, and welfare cost escalation. That is the case with China. For such a society, which is experiencing large and rapid societal changes, it is unfortunate that very little research has been conducted on Chinese citizens’ welfare attitudes. This article attempts to address that gap by exploring the popular support for social security in urban China.

Although the structure of the Chinese social security system – composed mainly of contribution-based social insurance programmes – is largely in place, the low level of benefits, high contribution rate, unequal entitlements, and regressive redistribution have continued to be major sources of public discontent (Xu, Guan & Yao et al. Reference Xu, Guan and Yao2011; Liu, Liu & Huang Reference Liu, Liu and Huang2016; Reference TaoTao, 2017). Recent years have witnessed the government's reform efforts, such as lowering contribution rates, integrating the fragmented social insurance system, and reducing the urban-rural disparity in entitlements (Reference Cai and ChengCai & Cheng, 2014; Reference He and WuHe & Wu, 2017). Despite the authoritarian nature of the Chinese political system, public attitudes have become increasingly important considerations in policy-making. It is found that even in political systems without contested elections or an independent civil society, authoritarian governments such as the Chinese authorities are still sensitive to public sentiments in order to avoid social instability and the erosion of political legitimacy (Reference Duckett and WangDuckett & Wang, 2013; Reference Ngok and HuangNgok & Huang, 2014). Therefore, soliciting citizens’ views on the existing system and their expectations for improvement is very crucial to inform China's ongoing social security reforms.

To what extent do the Chinese support the country's social security system? Do their different roles played in the social security system shape different attitudes? What are the factors associated with their attitudes towards social security? This study attempts to answer these research questions with cross-sectional data collected from a survey in a Chinese city. It seeks to make theoretical contribution by extending the theoretical frameworks that have been tested in the Western literature of welfare attitudes to the Chinese context.

2. Literature review

Popular support for welfare has long been studied from the perspectives of self-interest and ideology. The self-interest argument assumes that people's preferences are driven by the pursuit of individual gain (Reference KangasKangas, 1997; Reference Taylor-GoobyTaylor-Gooby, 1999); therefore, those who are socioeconomically vulnerable and thus more likely to become recipients of welfare benefits tend to hold favourable attitudes toward welfare policies (Reference Bean and PapadakisBean & Papadakis, 1998; Reference Hasenfeld and RaffertyHasenfeld & Rafferty, 1989; Reference SvallforsSvallfors, 1997). These interest-based models assume a direct relationship between people's structural position in the society and their welfare attitudes, based on their varying degrees of risk and dependency.

Underpinned by the concept of social location, self-interest explanations tend to use social classes in exploring the differences in people's welfare attitudes. Class is typically measured by income, education, occupational status, and self-identification of class position. Studies in the West have reported strong negative association between income and welfare support. Typically, lower-income earners and unemployed groups tend to demonstrate a stronger preference for state involvement (Reference Baslevent and KirmanogluBaslevent & Kirmanoglu, 2011; Reference EdlundEdlund, 1999; Reference GelissenGelissen, 2000; Reference Hasenfeld and RaffertyHasenfeld & Rafferty, 1989; Reference Yang and BarrettYang & Barrett, 2006). As predicted by self-interest models, people with higher education are less motivated to support welfare because they tend to have higher incomes and a stronger ability to provide for themselves, and hence they would expect to benefit less from welfare than persons with lower education (Reference Baslevent and KirmanogluBaslevent & Kirmanoglu, 2011; Reference GelissenGelissen, 2000; Reference PfeiferPfeifer, 2009; Reference van Oorschotvan Oorschot, 2000; Reference Yang and BarrettYang & Barrett, 2006).

Demographic factors are also found to influence people's welfare attitudes. Women are typically found to hold more positive views about welfare policies than men do (Reference EdlundEdlund, 1999; Reference PfeiferPfeifer, 2009; Reference SvallforsSvallfors, 1997; Reference Yang and BarrettYang & Barrett, 2006). This may be explained by cultural differences across both genders in the sense that women tend to adhere more than men do to values of equality and mutual responsibility (Reference GelissenGelissen, 2000). Age is usually found to have a strong positive effect, with older people generally being more supportive of welfare policies than younger people are (Reference Baslevent and KirmanogluBaslevent & Kirmanoglu, 2011; Reference Yang and BarrettYang & Barrett, 2006).

The ideology thesis, an alternative theoretical framework, refines the self-interest models by arguing that there is no direct relationship between individuals’ socioeconomic status and their welfare attitudes – that such effect is mediated by one's ideological orientation (Reference Gevers, Gelissen, Arts and MuffelsGevers, Gelissen, Arts, & Muffels, 2000). As Missinine, Meuleman and Bracke et al. (Reference Missinine, Meuleman and Bracke2013) maintain, the genesis of popular legitimacy is a process that is too complex to be influenced by self-interest alone but is also shaped by normative considerations and value frameworks. If norms and social obligations require it, individuals may act against their own material interests (Muuri Reference Muuri2009). Several empirical studies in both the East and the West have supported this ideology thesis by demonstrating that welfare attitudes are linked to more general ideological dispositions. For instance, endorsement of the concept of social rights and egalitarianism appears to be a key predictor of welfare support (Reference Blekesaune and QuadagnoBlekesaune & Quadagno, 2003; Reference EdlundEdlund, 1999; Reference GelissenGelissen, 2000; Reference HeHe, 2018).

A very important predictor of welfare attitudes is one's perceptions about the causes of poverty. For instance, research conducted in the UK and the US has revealed that people's general belief about the causes of poverty, rather than their own material circumstances, appears to be a primary determinant of their welfare attitudes (Reference FongFong, 2001; Reference LuoLuo, 1998). Wu and Chou (Reference Wu and Chou2017), in their analysis of welfare attitudes in Hong Kong, also elucidated that when poverty is seen as the result of laziness or character failing, people tend not to support welfare programmes. However, a more favourable stance toward welfare may be engendered if the root cause of poverty is understood to be social injustice, rather than personal failings.

Despite the wealth of research reviewed above, three major weaknesses exist. First, most studies were based on the Western context, and there is a dearth of empirical work that investigates welfare support in the East Asian welfare regime – a system that manifests salient characteristics which are distinctive from the Western models (Reference KwonKwon, 1997). Labelled as productive/developmental welfare regimes, China and the Asian Tigers have historically taken a conservative stance in which economic growth takes precedence over social policy (Reference HollidayHolliday, 2000). Therefore, people's welfare attitudes may present rather distinct characteristics.

Second, as Yang and Barrett (Reference Yang and Barrett2006) note, most previous studies tend to lump all welfare programmes into composite measures, at the cost of concealing the intrinsic differences between contributory programmes and means-tested arrangements. Attitudinal patterns with regard to social security programmes may appear to be somewhat different from those concerning other types of social welfare arrangements (Reference Yang and BarrettYang & Barrett, 2006). In China, for example, various social insurance schemes constitute the backbone of the country's social security system and need to be closely examined in terms of people's support of them.

Third, most previous studies tend to reduce welfare support to a single dimension, in particular to the preference for state involvement in welfare, and have ignored the inherent multidimensionality of welfare attitudes (Reference van Oorschot and Meulemanvan Oorschot & Meuleman, 2012). From a theoretical perspective, because individuals perform multiple roles in the social security system – as taxpayers, contributors, service users, and ordinary citizens, one should not expect people's attitudinal patterns necessarily to be identical across different dimensions and in people's different roles. People may have one view as users of social services and another view as financiers of those services (Reference KangasKangas, 1997; Reference MuuriMuuri, 2009). Hence, subjecting these different roles to operationalization and empirical investigation will help elucidate the multidimensionality of welfare attitudes.

3. Theoretical framework

This study is built on a bi-dimensional conception of welfare attitudes by distinguishing individuals’ roles as ordinary citizens and as premium contributors, and examines their respective attitudinal patterns and determinants. The first dimension represents people's general support of government's role in social security. Reflecting a citizen's normative expectation towards the appropriate split of responsibility between the state and individuals, this part of welfare attitudes is arguable most fundamental to the legitimacy of a country's social security system (Reference PfeiferPfeifer, 2009). The second dimension – the reasons why people are willing to make regular monetary contribution, are also critically relevant to the legitimacy of social security, given the mandatory nature of the programmes (Reference van Oorschotvan Oorschot, 2000).

In the first dimension, we expect generally high support for governmental responsibility because the Chinese social security system remains at its early stage of development while citizens’ demands for more welfare provision soar. In the meantime, citizens’ preference for governmental responsibility may considerably vary across different welfare domains (Reference NordlundNordlund, 1997). In the second dimension, individuals’ acceptance of monetary payment may not be necessarily determined by one single motive. Therefore, a methodologically sound approach is to distinguish more nuanced motives by letting respondents choose from alternative narratives, in order to gauge their wider motivational disposition as contributors (Reference van Oorschotvan Oorschot, 2000). We expect varying degree of support for these alternative motivations, but the actual variation is subject to empirical test.

The empirical analysis is guided by a theoretical framework depicted in Figure 1. First, we hypothesise that personal characteristics such as gender, age, education, income, and marital status exert major impact on both dimensions of welfare attitudes, as predicted by the self-interest thesis. In particular, people's welfare attitudes may increase or decrease due to their varying degree of dependence on the social security system. For instance, the elderly, retirees, the unemployed, and people in poor health tend to rely more on the system and are thus more likely to support governmental responsibility. Ironically, the retirees may also show higher support for premium contribution than the working population, given their beneficiary status. Second, not only past beneficiary experience may affect people's attitudes toward both dimensions, their access to the social security system may play a greater role, especially in shaping their preference for governmental responsibility. Those excluded from the system may engender negative perceptions against the state and the system itself.

Figure 1. Conceptual framework.

Third, inspired by the ideology argument, we hypothesise that both dimensions of welfare attitudes are shaped by different ideological beliefs held by individuals when they play different roles. Specifically, as citizens, people's preference for state involvement in social security may be more related to their general value frameworks; but as contributors until retirement, people's motivation to contribute may be more related to their recognition of the essential values of social security, such as income redistribution, at large. In the principle of risk-pooling, social security – particularly social insurance programmes – is inherently underpinned by the notions of social solidarity and income redistribution, which may provide a motivational basis for individuals’ contributions.

Fourth, we hypothesise that two salient factors may exert crucial influence on people's motivation for social security contribution, namely, their satisfaction with social services and their trust in government. Specifically, from a self-interest point of view, individuals’ satisfaction with social services determines whether they deem paying premiums to be worthwhile (Reference MuuriMuuri, 2009). Greater satisfaction may further increase contributors’ willingness-to-pay, fostering a healthy cycle, whereas dissatisfaction tends to erode their motivational basis. Moreover, general trust of the state has been recognised to be a crucial source of welfare legitimacy (Reference HeHe, 2018; Reference Rothstein and SteinmoRothstein & Steinmo, 2002). This is particularly so for contributory programmes, given the requirement of long-term regular payment. As Rose (Reference Rose and Shafer1991) illuminates, only if the state is trusted to be caring and effective does it make sense for people to put their welfare in its hands.

These broader factors identified in the theoretical framework will be operationalised into the Chinese context with reference to the international literature. Their explanatory power will be examined against the two conceptual dimensions with quantitative analysis.

4. Evolution of the social security system in urban China

The old social security system in urban China was built on the Communist planned economy, and remained that way until the country embarked on market-oriented reforms in 1980s. As a manifestation of communist superiority, full and life-long employment, job creation, and job assignment – often referred to as the ‘iron rice bowl’ – were guaranteed (Reference GuanGuan, 2000). Economic activities in urban areas predominantly took place around state-owned enterprises (SOEs) that still contributed about 80% of gross industrial output and employed more than 75% of the urban workforce by 1980 (Reference LeungLeung, 2003). Work units (danwei), particularly SOEs, were the cornerstones of both welfare financing and delivery, providing workers and their family members with a comprehensive welfare package (Reference Saunders and ShangSaunders & Shang, 2001). In contrast to the dominance of danwei in this socialist mini-welfare state, the role played by the government was limited to caring for the childless frail elders, the disabled, the homeless, and orphans. Overall, the Chinese society, especially the urban sector, constituted a ‘welfare society in a low-income country,’ under the planned economy (Reference GuanGuan, 2000).

The economic reform exerted a significant impact on social security. First, inefficient and overstaffed SOEs started to suffer from heavy economic losses, so that financing the mini-welfare state became heavy burden. Neither the government nor the SOEs had the financial ability to continue their generous provision of social security; even worse, many SOEs underwent privatization or bankruptcy. Second, blamed as a disincentive for improving productivity, the iron rice bowl was gradually dismantled. A mass layoff of workers in the 1990s dramatically changed labour relations and raised enormous demands for social security. Third, alongside economic liberalisation, private and informal sectors mushroomed, while new professions proliferated, calling for a paradigm shift in social security arrangements. Although the economic transition was gradual and incremental, it swiftly led to the destruction of old social security institutions (Reference Chan, Ngok and PhillipsChan, Ngok, & Phillips, 2008). Only in recent years has China seen a prominent expansion of social policies, in response to rising social instability and the need for the party-state to maintain political legitimacy (Reference Ngok and HuangNgok & Huang, 2014).

Constituting the current skeleton of the urban social security system are five social insurance schemes, covering old age pensions, health care, maternity, work injury, and unemployment, as well as a provident fund for housing finances. Well-known to most Chinese people is the term of ‘wu xian yi jin’ (five insurances and one fund).Footnote ¹Table 1 presents the prevalent contribution rates of all mandatory social insurance schemes, which are predominantly financed by contributions from employers and employees. Also financed by contributions from employers and employees, but without risk-pooling, is the housing provident fund, which is earmarked for enrolees’ housing purposes.

5. Methodology

5.1. Case selection

This study used a survey approach to probe the attitudes of Chinese urban residents toward social security. Given the practical difficulty in conducting a nationwide survey, and due to budgetary constraints, we adopted a single-city cross-sectional design. Zhuhai City of Guangdong Province was chosen for investigation. Located on the western bank of the prosperous Pearl River Delta, Zhuhai is one of the five Special Economic Zones (SEZs) that pioneered China's economic reforms. Largely due to its less aggressive stance toward industrial expansion, Zhuhai is among the few Chinese cities that enjoy a fairly clean environment, and the city's GDP per capita amounted to US$22,100 in 2017. Ranked as one of the most livable cities in China, it has been a hotspot for internal immigration.

The advantages of selecting Zhuhai for investigation are threefold. First, despite its high level of affluence, this city has a population of merely 1.76 million people (including migrant workers) and does not belong among the first-line megacities such as Beijing, Shanghai or Guangzhou, thus making it more representative of medium-sized Chinese cities. Second, the existence of a sizable non-local population in Zhuhai enabled us to contrast the opinions of individuals both with and without local hukou (household registration status). Third, in contrast to its neighbours that have undertaken major social security innovations, such as Shenzhen and Dongguan, the structure of Zhuhai's social security system largely resembles that of the nationwide picture, thus strengthening its representativeness.

Table 2 outlines the general profile of social security in Zhuhai. Key social insurance programmes have achieved considerably high coverage. While medical insurance has almost protected the entire urban population including those without local hukou, other employment-based social insurance programmes have covered the vast majority of the working population employed in the formal sector. The level of benefits offered by Zhuhai's social insurance programmes also stand on the high side. For instance, when compared with those of Zhongshan City, its neighbour that is also an affluent industrial city of Guangdong, most social insurance programmes in Zhuhai offer more generous benefits.

Given the significant regional disparity and the fragmentation of China's social security system, we must acknowledge our modest ambition of generalising the findings of this study. The relatively high socioeconomic development of Zhuhai and its status of SEZ give this city some distinctive characteristics that are not present in most other Chinese cities, and may alter people's welfare attitudes to certain extent. Its geographic proximity to capitalist Hong Kong and Macau has long embedded Zhuhai into the global economy that may also exert influence on people's perceptions. Given these contextual characteristics, we position the contribution of this study as a pilot to investigate welfare attitudes in wealthy coastal parts of China.

5.2. Sampling

Data were collected from April to May 2016. Ethical approval was obtained from the corresponding author's university. Ten student assistants with prior survey experience were recruited as enumerators from the first author's university and were further trained. A three-stage sampling strategy was adopted. In the first stage, the three administrative districts of Zhuhai City – namely, Xiangzhou, Jinwan and Doumen – were used as the proportionate sampling frame, in reference to their respective populations. Three to five neighbourhoods (jiedao) were randomly selected from each district in the second stage. As the urban-rural divide in hukou status may exert crucial impact on people's attitudes, we purposefully included two neighborhoods in each district that had larger population of migrant workers, in order to collect a substantive sample of residents without local status. The third sampling stage was a systematic selection of households (including university hostels) with a random starting point and equal probability method. The survey was conducted during weekends when the chance of exclusion was low. One adult per household who was willing to be interviewed was then invited to answer the questions.

A total of 1471 individuals were approached and 1078 accepted our invitation. Prior verbal consent was sought from all respondents. The final sample contained 965 respondents, after invalid questionnaires (mainly incomplete ones and those with significant logical errors) were excluded. The profile of respondents is presented in Table 3. Because only those who were formally employed by the time of the survey were required to indicate their monthly salaries while the unemployed and retirees left that answer blank, data about income were missing for a considerable number of respondents. That restriction reduced the sample to approximately 600, when monthly income had to be included in the multivariate analysis.

6. Results

6.1. Preference for governmental responsibility

Descriptive results suggested that respondents’ expectation for governmental responsibility in various areas of social security was moderately high (mean = 11.75, standard deviation = 2.61). This support was highest for health care (mean = 1.65, standard deviation = 0.60) and lowest for job guarantee (mean = 2.13, standard deviation = 0.76). The dependent variable was then subjected to multivariate analysis by using both an ordinary least squares (OLS) model and an ordinal regression model. Significant Chi-squared statistics suggest that the set of factors in each model reliably predicted the dependent variable while the percentage of variance explained was moderate. Because both estimations produced very similar statistical patterns, only the OLS results are presented (Table 4).

The dependent variable was regressed in a phased fashion in order to test different theoretical frameworks individually and in combination. Model 1 included personal characteristics and variables related to one's access to the social security system as well as actual receipt of benefits, in reflection of the self-interest thesis. Personal characteristics included into regression models were gender, education, age group, employment status, marital status, monthly wage, and subjective health status. Pertinent to social security in the Chinese context, possession of a Zhuhai social security card and status as a recipient of benefits were also included into this model, since the actual beneficiaries were commonly understood to be in greater favour of social security.

Poor health, being age (27-40), possession of a local social security card, and a history of receiving benefits turned out to be associated significantly with a preference for governmental responsibility in social security. Those in poor health are at greater physical and economic risk and hence tend to favour welfare arrangements due to their dependency. It is not difficult to understand that individuals who are entitled to social security benefits – particularly people who have received benefits – were more supportive of governmental responsibility, based on self-interest considerations.

Being unmarried was associated with a negative expectation for extensive governmental responsibility. This reinforces previous findings in the West that attribute this group's negative expectation to their lower vulnerability (Reference Yang and BarrettYang & Barrett, 2006). As Busemeyer, Goerres, and Weschle (Reference Busemeyer, Goerres and Weschle2009) illuminate, one's position in the life cycle shapes his/her preference for social security. Being single implies not only lower health and economic risks, but also fewer family responsibilities to spouses or children – responsibilities that tend to engender a greater expectation for state involvement. This life cycle explanation also sheds light on the effect of being in the age group 27–40. Urban Chinese people in this bracket are typically at the stage of family formation, childrearing and career development, and thus are in increased need of social protection. In addition, a crucial factor peculiar to this group is the skyrocketing price of housing in China and the stronger need for real property among this group of adults. A further bivariate scrutiny suggested that this group indeed showed a significantly higher expectation for governmental responsibility in providing everyone with affordable housing, vis-à-vis other age groups.

Contrary to what one might expect, neither old age nor being a retiree explained the variance in the dependent variable in a statistically significant manner. That may be because senior citizens in Zhuhai are more content with the coverage and entitlement of social security and therefore do not hold significantly higher expectations.Footnote ² Moreover, urban residents are exempted from a premium contribution upon retirement, so the elderly in essence become net beneficiaries of the system. Bivariate statistics reveal that the elderly showed the highest satisfaction with social services, while their younger peers (age group 27–40) were least satisfied, so the latter may understandably expect stronger governmental intervention. Gender, education, and income yielded no statistically significant association with the first dependent variable. Overall, Model 1 explains approximately 8% of the variance in the dependent variable (R ² = 0.0793).

Two ideology-related variables were entered into Model 2 to test the ideology framework. While the self-interest variables that were significant in Model 1 remained robust, respondents’ endorsement of the statements concerning income gap and social injustice was significantly associated with support for governmental responsibility, thereby reinforcing the ideology argument. Model 2 improved the model fit by 35% and explained approximately 11% (R ² = 0.1070) of the variance in the dependent variable.

6.2. Motivations for monetary contribution

Table 5 exhibits the distribution of respondents’ levels of endorsement of different motivations for contributing to social security. Clearly, most respondents indicated agreement with all four statements. Although the distribution varies across different items, the general level of agreement remains quite high. Personal benefit received the highest endorsement among all. A remarkable finding is that a much larger percentage of respondents (30.3%) did not agree with the fourth statement, which said that they contribute to social security simply because of statutory requirement. People's motivation appeared to be predominantly underpinned by a mixture of self-interest concerns and altruism.

Ordered probit models were employed in multivariate analysis to examine factors associated with each motivation, based on our theoretical framework presented above. Significant Chi-squared statistics once again suggest that the variables in each model reliably predicted the dependent variable while the percentage of variance explained was moderate. Each dependent variable was first regressed on variables related to self-interest factors (including personal characteristics and beneficiary status) in Models a, including gender, education, age group, employment status, health status, and prior receipt of benefits. As noted above, people's level of satisfaction with social services may either strengthen or undermine their support for social security. This was also considered, along with other self-interest variables, in Models a. In Models b, ideology- and trust-related variables were added to test the ideology argument. First, one's recognition of the importance of income redistribution was included to capture his/her ideological standing towards social security. The second variable considered the degree of trust that individuals put on government (Table 6).

Similar to the results reported in Table 4, gender and education, again, showed no statistically significant relationship with any of the dependent variables. The age group 41–60 appeared to be a very powerful predictor of ‘social solidarity’ and ‘mandatory requirement,’ which are two competing narratives. As Models 3a, 3b, 4a, 4b suggest, this group of respondents tended to hold a fairly cynical stance toward social security contribution. People in this age bracket typically started working before China's embarkation on social insurance programmes in 1990s, so there mindset towards contributory programmes may not have been fully adapted. Similar cynicism was also observed in the lowest income bracket, ¥<3000/month (Models 2a, 2b, 3a, 3b). That group's lower labour market status and higher economic vulnerability tend to make them net recipients of social protection; this status may, in turn, reinforce their dependence on the welfare system.

Employment status has left its mark on some dependent variables. Retirees were more likely to endorse the narrative of moral obligation, whereas those currently unemployed tended to support the solidarity principle. This finding is in line with the self-interest arguments as a result of these groups’ stronger needs for protection. Yet, it must be noted that their answers to these questions are of a counterfactual nature because neither group is required to contribute under the current policy framework in China. ‘Satisfaction’ showed a statistically significant relationship with ‘personal benefit’. Clearly, a higher satisfaction with the actual social services received enhances people's motivation to pay for social security.

The endorsement of income redistribution stood out as a very powerful predictor of all of the dependent variables, suggesting that the central principle underlying the social security institution has been well supported by people in the sample, regardless of their dominant motivation for supporting the system. A higher degree of trust in the government led to a stronger motivation from ‘personal benefit,’ ‘moral obligation,’ and ‘social solidarity’ but not from mandatory requirement.’ Individuals who saw the government as trustworthy were more willing to put their money into the social security funds by accepting a regular payment requirement, because concerns over misuse or insolvency tended to be low.

7. Discussion

This study has sought to operationalise a bi-dimensional conception of popular support for social security in China, by analysing primary data collected in Zhuhai City. As ordinary citizens, respondents’ expectations for governmental responsibility in social security appear to be high. As contributors to the system, their willingness to pay premiums is also on the high side. Self-interest and ideology models were both tested in the Chinese context. A few theoretical insights with reference to the international literature can be drawn.

In the first dimension of welfare attitudes – preference for governmental responsibility, the conventional self-interest arguments from the West merely received partial endorsement in our study. Many important factors that were found of high explanatory power in the West, such as gender, employment status, income, and education (Reference EdlundEdlund, 1999; Reference GelissenGelissen, 2000; Reference SvallforsSvallfors, 1997; Reference Yang and BarrettYang & Barrett, 2006) yielded no significant explanatory power in our study. Instead of those social-location-related factors, the life-cycle perspective provided more plausible contextual explanations for the significant factors that emerged from empirical results, such as marital status, age group, and health status. These explanations still broadly echo the self-interest thesis that individuals’ support of welfare arrangements is explained by one's vulnerability and potential dependency (Reference Hasenfeld and RaffertyHasenfeld & Rafferty, 1989; Reference Taylor-GoobyTaylor-Gooby, 1999). This study has suggested that Chinese people's attitudinal cleavage toward government's role in social security is not created by social location as often found in the West, but instead is predominantly an individual-level phenomenon, at least with the Zhuhai sample.

Despite the rigid social stratification and income polarization in China, those societal changes do not seem to systematically alter the expectation for social security among different classes, as can be seen in many Western societies (Reference PfeiferPfeifer, 2009; Reference Baslevent and KirmanogluBaslevent & Kirmanoglu, 2011). This could be explained by the low baseline for social security provision since China's market transition, making most people tend to expect more. Another possibility is related to China's ideological legacies from Communism; studies have found that individuals in former socialist countries tend to have stronger preferences for governmental involvement than do people in Western nations (Reference Corneo and GruneCorneo & Grune, 2002).

The ideology argument received strong endorsement in Zhuhai. As can be seen in the West that ideological dispositions powerfully shape one's welfare attitudes (Reference EdlundEdlund, 1999; Reference FongFong, 2001; Reference GelissenGelissen, 2000), pro-redistribution values are strongly associated with Zhuhai people's support for state responsibility, providing a solid attitudinal foundation for the city's social security system.

The second half of our analysis attempted to explain the motivations behind people's willingness to pay for social security. The analysis offered a battery of more nuanced findings, by examining people's endorsement of four alternative motivations. The findings revealed a fairly high solidarity spirit, as well as strong evidence of self-interest motives. Differing from the patterns shown in the first dimension, here social location played a salient role in shaping people's willingness to contribute. Whereas the first dimension measured people's general preferences as citizens, the second dimension hinged explicitly on their role as payers; that tangible financial relevance may have prompted respondents to voice a position that better reflected their material position and interest. Overall, social location seems to better explain the second dimension of popular support for social security (Reference Yang and BarrettYang & Barrett, 2006), in contrast to the first dimension, which is found to be mainly a function of individual interest.

Reinforcing the findings of many previous studies based on other societies (Reference HeHe, 2018; Reference Rothstein and SteinmoRothstein & Steinmo, 2002; Reference Yang and BarrettYang & Barrett, 2006), ideological recognition of the redistribution principle and general trust of the state were found to be very powerful predictors of most motivational items. These ‘soft’ elements of popular support seem to have transcended national boundaries and have become universal sources for the legitimacy of modern social security institutions.

Social security, by its very nature, is a statutory institution; contributing to it is a legal obligation for everyone with an income. Yet, this study in Zhuhai produced little evidence that enforcement is the only source motivating the Chinese to pay for social security. Contributing to social security seems to be accepted on the grounds of self-interest, solidarity, and legal obligation. Remarkably, close to 80% of the respondents cited that they were motivated to contribute on moral grounds. Only a very small percentage explicitly disagreed with this. Remarkably, the figures are comparable with those reported in an empirical study in the Netherlands, a far more mature Western welfare state (Reference van Oorschotvan Oorschot, 2000). Clearly, the social security system of Zhuhai has a strong base of legitimacy.

8. Concluding remarks

Welfare attitudes are intrinsically context-dependent. There is, unfortunately, a paucity of knowledge with regard to welfare attitudes in the East Asian context. Conducted in Zhuhai City, this study focuses on Chinese people's attitudes toward contributory social security programmes. Aside from the empirical findings discussed above, two policy implications are in order.

First, this study has found fairly strong popular support for social security in Zhuhai. Although self-interest motives and social equity values coexist, the latter's impact seems to be more consistent and robust. The wide acceptance of pro-redistribution ideologies provides a solid motivational foundation for sustaining the social security system, but it also raises people's expectations for the provision of more and better services. Health care and public housing, in particular, top the list of public concerns on which greater government efforts are necessary. In addition, this study also found that as a whole, various levels of government and civil servants receive a fairly high level of trust from citizens. Trust, in turn, strengthens people's willingness to pay for social security. In order to reinforce this healthy cycle to serve social security reforms in the future, the key principles of good governance, such as transparency, accountability, and participation, must be honoured. The success of social security reforms ultimately hinges on astute political management of public opinions.

Second, dividing the Chinese populace into agricultural and nonagricultural categories, the hukou system has long been known for creating a rigid urban-rural dichotomy in the Chinese society. Associated with this divide is a vast disparity in social, occupational, and economic entitlements between urban and rural residents. Migrant workers, in particular, live in a rather underprivileged position in cities (Xu et al., Reference Xu, Guan and Yao2011). Although we initially hypothesised that hukou status might affect individuals’ attitudes toward social security, the results Footnote ³actually showed that this variable was statistically insignificant on all dependent variables. Rather, having a local social security card mattered a great deal, as revealed earlier. In response to calls for more inclusive social protection, the Zhuhai Municipal Government, like local governments of some other developed localities, has allowed residents without a local hukou to join their respective social security schemes, provided that the applicant meets certain requirements.Footnote ⁴ Cross tabulation suggested that as many as 44.1% of our respondents who did not have a local hukou actually held a Zhuhai social security card. The average level of satisfaction with social services was also comparable for both groups of respondents. The recent policy developments towards a more inclusive social security system may have contributed to the strong popular support observed in Zhuhai.

This study is certainly not without limitations. The sheer size of China defeats the generalisibility of most city-level case studies. It is not the intent of this study to explain the diverse situation in China. As one of the few empirical studies in the English-language literature that investigate welfare attitudes in East Asia, this research represents a meaningful attempt to test the major theoretical frameworks in the Chinese context and explain the context-dependent observations. Aside from limited generalisibility, there are certainly limitations in the methodological rigour of this study. For instance, wage income may no longer be an accurate proxy of actual income in developed regions of China. The measurement of government trust may also be compromised due to China's authoritarian social atmosphere. These weaknesses will be addressed in the authors’ ongoing investigation on public attitudes toward social security in China.