Inequality and Political Trust in China: The Social Volcano Thesis Re-examined

Yingnan Joseph Zhou; Shuai Jin

doi:10.1017/S0305741018001297

Inequality and Political Trust in China: The Social Volcano Thesis Re-examined

Published online by Cambridge University Press: 12 November 2018

Yingnan Joseph Zhou and

Shuai Jin

Show author details

Yingnan Joseph Zhou*: Affiliation:
Department of Political Science, University of Texas at El Paso.
Shuai Jin: Affiliation:
Department of Political Science, University of Massachusetts Boston. Email: Shuai.Jin@umb.edu.
*: Email: yzhou5@utep.edu (corresponding author).

Article contents

Abstract
The Social Volcano Thesis and Its Challenges
Challenges Examined
The Social Volcano Thesis Reconsidered
Empirical Analysis
Results
Conclusion
Footnotes
References

Rights & Permissions

Abstract

The social volcano thesis states that the rising inequality in China threatens regime stability. This idea, although widely held in the media and in academia, is backed by little positive evidence but by much negative evidence. Two primary pieces of negative evidence are that the Chinese people trust the central government and that they are highly tolerant of inequality. This paper discusses the shortcomings of the negative evidence and re-examines the thesis in a rigorous and direct way. Our multilevel analysis shows that provincial inequality has negative effects on individuals’ trust in the local government but not in the central government, and this negative effect holds for both the rich and the poor. Because distrust in the local government implies distrust in the central government, we conclude that a social volcano exists.

摘要

“社会火山论” 认为中国严重的收入不平等会威胁政权稳定。此说虽然在媒体和学术届颇具影响，却尚未得到有力的证据支持。相反，反对此说的证据却越来越多。本文梳理了这些反面证据，指出了其中的问题，并对 “社会火山论” 进行了直接并严格的检验。我们的多层分析显示，省级的收入不平等确实降低了民众（包括穷人和富人）对地方政府的信任度，但不平等对民众对中央政府的信任度没有影响。由于民众对地方政府的不信任隐含了他们对中央政府的不信任，所以本文认为 “社会火山” 是存在的。

Keywords

inequality Gini coefficient political trust China multilevel analysis 不平等基尼系数政治信任中国多层模型分析

Information

Type: Research Article
Information: The China Quarterly , Volume 236 , December 2018 , pp. 1033 - 1062

DOI: https://doi.org/10.1017/S0305741018001297 [Opens in a new window]
Copyright: Copyright © SOAS University of London 2018

In a TV interview, Chinese billionaire Wang Jianlin 王健林 expressed his uneasiness about the younger generation’s aspiration to attain a fortune without any planning. “It is best,” advises Wang, “that you begin with a small goal that you can reach, say, to earn 100 million yuan [US$15 million] first.”Footnote ¹

Wang's career advice is perhaps less useful, and certainly less palatable, than that offered by fortune cookies. An ordinary person may hit ten jackpots and still find him or herself a long way off from Wang's “modest” start. However, the advice sends a strong reminder of the growing inequality in China. In 1980, the Gini coefficient of household income was 0.230, one of the world's lowest, but it climbed to 0.412 in 2000 and 0.491 in 2008.Footnote ² In 2008, China was one of the most unequal countries in the world, ranking alongside South Africa, Columbia and India.Footnote ³ Today, it is difficult to call China a socialist country without sounding sarcastic.

This inequality causes concern. Analysts issue warnings about the repercussions of inequality such as high crime rates and high infant mortality rates.Footnote ⁴ But China observers have an additional concern – one that if ignored, could have catastrophic consequences. In a democracy, the resentment over inequality can be periodically vented by elections; in China, without this venting mechanism, the survival of the authoritarian regime is at stake. Scholars have sounded the alarm that China has long crossed the warning line of 0.40 Gini and that the country is at the mercy of a social volcano that threatens to cremate the regime and bury the people.Footnote ⁵

This social volcano thesis has intuitive appeal and theoretical grounds. In China, housing and education are becoming unaffordable, yet both are curiously in high demand. Doctors and civil servants appear to enjoy lifestyles that their official salaries cannot support. The internet is awash with expressed sentiments of relative deprivation stemming from the gap between what people receive and what they think they deserve, sentiments which have been found to fuel rebellions.Footnote ⁶

But the biggest problem with the social volcano thesis is that it contradicts an established finding about China, that is, that the Chinese people have tremendous trust in their authoritarian regime, even in times of increasing inequality.Footnote ⁷ If people trust the regime despite the inequality, where is the social volcano?

The short answer according to scholars such as Xiaogang Wu and Martin Whyte is that there is no volcano.Footnote ⁸ Wu compares people's acceptance of existing inequality in 28 societies and finds that China ranks remarkably high, only behind the Philippines and Hong Kong. Martin Whyte calls the volcano a myth, because although the Chinese people are aware of the huge income gap, they willingly accept it in the belief that income is distributed fairly and based on individual merits.Footnote ⁹ The impression of public anger about inequality, according to Wu and Whyte, is simply unrepresentative.

Wu and Whyte certainly shed new light on understanding the political consequences of inequality in China, yet they fail to provide a sufficient basis for rejecting the social volcano thesis. To begin with, the social volcano thesis proposes a relationship between a dependent variable (regime stability) and an independent variable (inequality), but neither study directly examines this relationship. And instead of directly measuring inequality as a reality independent of people's perceptions, they focus on a different concept: the sense of equity – people's normative judgement on inequality – possibly on the assumption that inequality affects regime stability only when people think it unjust. But, this assumption may be false. First, inequality may cause one to lose trust in the government even though no inequity is detected in the survey. This could happen, for example, when people blame the government not for inequality, but for the side effects of inequality such as high crime rates. In addition, the key findings of both studies are aggregate-level statistics that reveal majority opinions. But the majority's acceptance of inequality is no guarantee of regime stability, as has been long recognized by the social movement literature. Evidence shows that when more than 10,000 people rebel, state repression is often handicapped, and when 5 per cent of the population rebels, few regimes survive.Footnote ¹⁰ Therefore, if people do lose trust in the government because of inequality, the volcano may still exist.

In this paper, we undertake a direct and rigorous test of the social volcano hypothesis. We acknowledge that the Chinese people are more tolerant of inequality than people in other societies, but we take seriously the theoretical instinct of the social volcano thesis. We argue that the Chinese people do lose trust in the government because the populist authoritarian culture so inclines them and because they are aware of the level of inequality in China.Footnote ¹¹ We further argue that the Chinese people lose trust only in local government, not in the central government, because the central government makes popular social policies while the local government is tasked with implementing them. When inequality rises, people hold the local government responsible as it is the implementer. This attribution of responsibility explains the puzzling coexistence of high trust in the central government and high inequality. Our arguments receive substantial support from our multilevel analysis. Because distrust of the local government could eventually erode trust in the central government, we conclude that a social volcano does exist in China.Footnote ¹²

The Social Volcano Thesis and Its Challenges

The social volcano thesis states that economic inequality corrodes regime stability to a breaking point. This view is widely shared in the media and academia. A 2005 report by the Chinese Academy of Social Sciences states that the “rich–poor disparity has led to the intensification of social disputes, mass protests, and criminal cases.”Footnote ¹³ A New York Times article remarks that “Because many people believe that wealth flows from access to power more than it does from talent or risk-taking, the wealth gap has incited outrage and is viewed as at least partly responsible for tens of thousands of mass protests around the country in recent years.”Footnote ¹⁴ Internet users minced no words chastising Wang Jianlin for his outrageous definition of small targets: “Damn! You know where your money is from!” “Pretentious, the day when your original sin is exposed, you'll cry and beg for your life.” In academia, scholars have noted that the Gini coefficient of China has long exceeded 0.40, the internationally recognized warning line.Footnote ¹⁵

However, empirical support for the social volcano thesis is surprisingly scant. Although studies frequently cite mass protests and social disputes, they stop short of showing that inequality is the culprit. To our knowledge, no published research has empirically linked inequality to any indicator of regime stability. The widespread narrative of the 0.40 warning line is dubious. What makes 0.40 the magic number? Does the probability of social unrest drastically increase once the Gini coefficient reaches 0.40? The literature is silent on these questions. A simple Google search of “0.40, Gini, warning line” returned results exclusively about China, casting serious doubt on the warning line narrative.

The social volcano thesis also faces jarring challenges from negative evidence. The thesis flies in the face of a well-known finding about China: the Chinese people have great trust in the central government. In 2008, inequality hit a record high, but China's remained one of the most trusted governments in the world, with 89 per cent of the people saying that they trusted the national government. To compare, the figures were 38 per cent in Germany, 32 per cent in Taiwan and 20 per cent in the United States.Footnote ¹⁶ Although sceptics often challenge the data on China on the basis that people's views are affected by fear, serious investigations suggest otherwise.Footnote ¹⁷ Considering that the citizenry trusts the regime despite severe inequality, the notion of a social volcano seems far-fetched.

Four public opinion studies indeed think so. In his 2009 study, Xiaogang Wu compares people's tolerance of inequality in 28 societies.Footnote ¹⁸ In each of the societies, respondents were asked to estimate the actual incomes of a list of occupations, including both blue-collar and white-collar jobs (do-earn incomes). Respondents were also asked how much they thought each occupation should earn (should-earn incomes). Wu compares the inequality among do-earn incomes, which indicates the inequality in reality, with the inequality among should-earn incomes, which reflects the inequality believed to be appropriate. In all of the 28 societies, people thought that inequality was higher than it should be, but this feeling was less intense in China than in 25 societies, including Russia, France, Japan, Canada and the United States. Only in the Philippines and Hong Kong were people more accepting of the existing inequality. This means that, in perspective, people in China have “very high normative support for income inequality.”Footnote ¹⁹ Thus, it is a misperception to see a social volcano in China but not elsewhere.

In his influential book, Martin Whyte deals a heavy blow to the social volcano thesis. Using data from a 2004 national survey, Whyte makes several observations.Footnote ²⁰ First, there is no massive dissatisfaction with income distribution in China. Most people believe that an individual is poor because of a lack of ability, education and effort, not because of unequal opportunity or discrimination, and they are optimistic about their opportunities to get ahead.Footnote ²¹ Second, echoing Wu's finding, Whyte shows that citizens’ acceptance of inequality in China is greater than that in East European countries and Western democracies.Footnote ²² Third, the radical egalitarianism under Mao has had no lasting imprint on the following generations. Data show no widespread nostalgia for the Maoist era. As a matter of fact, Mao's making-all-poor approach was unpopular, and more important, it failed to achieve its goal, as salaries and benefits in Maoist China were highly unequal across workplaces and occupations.Footnote ²³ It is the individualist elements of Confucianism that have profoundly shaped China's distributive culture.Footnote ²⁴ Whyte updates his findings in two subsequent studies and the conclusion remains: the social volcano is a myth.Footnote ²⁵

Challenges Examined

The current literature makes it tempting to dismiss the social volcano thesis as false; however, this may be too hasty. First, the social volcano thesis has not been tested directly. The thesis proposes a causal relationship between two variables. It states that regime stability (the dependent variable) decreases as a result of rising inequality (the independent variable). To reject the thesis, one must show that there is no relationship or no positive relationship; this has yet to be accomplished. True, the studies show that people in China are more tolerant of inequality than people in other countries.Footnote ²⁶ But this at best shows that there is no imminent volcanic eruption– – not that there is no volcano. We think even this no-eruption interpretation could be over-simplistic. Researchers of social movements have long realized that when it comes to rebellion, what matters most is not the majority but the minority. The majority of the people may be loyal and happy, but a disgruntled minority, if mobilized, can shatter the regime. William Gamson notes that when 10,000 people rebel, state repression is often crippled.Footnote ²⁷ Mark Lichbach finds that when 5 per cent of the population rebels, few regimes can survive.Footnote ²⁸ The 1989 Tiananmen protests presented the biggest challenge to the CCP regime to date, but how many protesters participated? The biggest crowd on the Tiananmen Square was probably less than 0.1 per cent of the population.Footnote ²⁹ So, even if most people in China have positive feelings about the current inequality, there could still be a volcano if inequality negatively affects regime stability.

The existing studies do not directly measure inequality and so do not prove that a social volcano does not exist. Inequality refers to how evenly resources are distributed within a group, and it should be a mind-independent property of a group, not the subjective perception of an individual. Individuals within a group may have completely different perceptions of how things are distributed, while the group-level distribution remains unchanged. That is to say, inequality does not vary from individual to individual, but perception of inequality does. Hence, objective measures of inequality such as the Gini coefficient and the ratio of high incomes to low incomes should be preferable to subjective measures such as how much inequality individuals perceive. Sometimes survey questions are phrased so that they tap into respondents’ normative judgement on income distribution. Take, for example, the question “do you think the gap between the rich and the poor is too large?” In reality, such survey instruments measure not inequality but a sense of inequity, which is a different concept. Because people's normative judgement depends on culture, values and expectations, a sense of inequity is an individual property that is inherently subjective.

Although the term inequality appears in the titles of both Wu's and Whyte's works, their empirical analyses instead focus on the sense of inequity, probably for two reasons.Footnote ³⁰ First, some argue that inequality alone is meaningless. A Gini coefficient of 0.43 tells us nothing, but 80 per cent of the population reporting excessive inequality indicates a regime's frailty. Our response is that this is a problem only when inequality is examined alone. When inequality is examined in relation to regime stability, the problem vanishes. Second, some may argue that inequality affects regime stability only by creating a sense of inequity, so it is unnecessary to measure inequality directly. However, there are many reasons why this is not a universal rule. Inequality may corrupt political systems, erode governance and cultivate a culture of distrust, all of which may trigger distrust in the government even if no overt resentment is expressed. Imagine a poor person living in a segregated city who only associates with similar people. This person may feel little inequity because of his or her homogeneous social contact but may still distrust the government because, in the segregated context, that person has learned since childhood not to trust anyone, including the government. In addition, the idea that a sense of inequity is the only path from inequality to political distrust contradicts what we know about survey methods. There are many ways in which people may lose trust in the government as a result of inequality, even if they express no sense of inequity in the survey. They may believe that the government is responsible for excessive inequality, but what if right before they take the survey, they read an article that convinces them that inequality has been in steady decline? In this case, they may report whatever is top in their mind: inequality is not too high.Footnote ³¹ In short, the existing studies leave open the possibility that a social volcano exists.

The Social Volcano Thesis Reconsidered

We argue that a social volcano does exist in China, not in the form of pervasive public indignation but in the form of a negative relationship between inequality and individual trust in the government. We argue that people, regardless of their incomes, distrust the government because they are aware of the high inequality, dislike it, and blame the government for failing to contain it. People dislike inequality through many possible mechanisms: they may dislike it for no articulate reasons; dislike it for its side effects such as high crime rates; or dislike it because it is unjust. We therefore do not specify all those intermediate variables, just as economic voting scholars seldom specify the intermediate variables between a bad national economy and people's dislike.Footnote ³² People blame the government for failing to contain inequality because China's political culture inclines them to do so. Owing to the gap between the promised benefits of the central redistributive policies and poor local implementation, when inequality increases, people only lose trust in the local government. Finally, the rich react to inequality in similar ways to the poor. This is because rich people's political trust is not determined by the relative superiority they may gain from inequality. Certain points in our theory demand further elaboration.

Influence of political culture

Inequality causes the Chinese people to distrust the government because of a political culture that holds the government responsible for excessive inequality. The studies by Wu and Whyte unveil an individualistic facet of China's redistributive culture.Footnote ³³ Nonetheless, this individualism should not lead us to overlook Chinese people's general acceptance of government intervention in the economy. In one of the most recited verses of the Analects, Confucius preaches that political leaders should fear inequality more than they fear poverty.Footnote ³⁴ In a sophisticated treatise on contemporary political culture in China, Wenfang Tang finds that people have high expectations about what the government can and should do to address their needs.Footnote ³⁵ For example, labour disputes are frequently brought before the government for resolution rather than taken to the courts or labour unions. There have even been occasions when Wen Jiabao 温家宝 has supported campaigns on behalf of unpaid migrant workers.Footnote ³⁶ Another glimpse of people's reliance on government intervention may be gained from the online forums of municipal government websites, where internet users frequently demand governmental action on trifles such as a hospital's parking charges and street vendor noise. The Chinese people's need for government intervention is also manifest in their peculiar understanding of democracy. Only a quarter of Chinese people think that elections or freedom of speech are essential to democracy, while 42 per cent think democracy essentially means that there is a narrow income gap or the provision of basic necessities like food, clothes and shelter for everyone.Footnote ³⁷

Citizens are aware of high inequality

Even though the Chinese people have a cultural tendency to hold the government responsible for inequality, they will not do so unless they are aware of high inequality in China. So, are they aware? There are many indications that they are. The domestic media are obsessed with exposing the luxurious lives of the rich, as exemplified by the viral video of Wang Jianlin's small-target remark about wealth. Xi Jinping's 习近平 anticorruption campaign, which mobilized the public to condemn the corrupt lifestyles of toppled leaders, has galvanized this obsession. If these examples are too anecdotal, let us turn to national surveys. The gist of Wu's and Whyte's studies is that the people find China's income distribution acceptable.Footnote ³⁸ But when we zoom in on their findings, we clearly see complaints about inequality, and louder complaints when governmental responsibility is implied. According to Whyte, 72 per cent of people in China think that the national income gap is too wide or somewhat large; 60 per cent agree with the statement that the rich are getting richer and the poor poorer; 51 per cent agree with the statement “The reason why social inequalities persist is because they benefit the rich and the powerful;” and 83 per cent think an unfair economic system exerts some influence, a big influence, or a very big influence on people's poverty.Footnote ³⁹ Wu reports that when asked to what extent six occupations (farmer, peasant worker, urban factory worker, professor, senior government official, and CEO in a large company) in China are overpaid, respondents singled out senior government officials as the most overpaid – and overpaid by 35 per cent.Footnote ⁴⁰

Citizens’ attribution of responsibility

If the social volcano thesis is to be sound, it must be reconciled with the enviable level of trust enjoyed by the Chinese central government. The reconciliation, we argue, lies in the trust gap between the central government and local government. Because of a self-reinforcing cycle of institutional overtasking, credit-taking and blame-shifting, the negative influence of inequality on political trust is mostly absorbed by local government. This trust gap spares the central government and provides it with an invincible facade.

Scholars have long noticed that in China, political trust declines as the government in question moves down the hierarchy. Lianjiang Li compares the percentages of people who report very high or relatively high trust in different levels of Party committees.Footnote ⁴¹ For the Party committees at the central, provincial, county, and township levels, the figures are 91 per cent, 72 per cent, 53 per cent and 38 per cent, respectively. Numerous authors find that the public is mostly dissatisfied with the corruption at the local, rather than the central, level.Footnote ⁴² The root cause of this trust gap is that the central government makes policies that mean well and promise much, or at least appear to; however, the local government, when tasked with implementing these policies with limited resources and in complicated circumstances, is often unable or unwilling to deliver. This is largely how the redistributive policy functions in China. Take, for example, the Minimum Living Security (dibao 低保), the assistance programme for the extremely poor. The State Council decreed the programme in a two-page document.Footnote ⁴³ The document contains little detail. It does state that local government is solely responsible for financing the programme, but it is otherwise so vague that it is virtually useless as a guideline. The quality of local implementation is, unsurprisingly, patchy.Footnote ⁴⁴ In one extreme yet telling case, the local authorities unjustifiably cancelled one poor family's dibao, which led the mother to kill her four children and then herself. The tragedy ended with the father's suicide eight days later.Footnote ⁴⁵ In similar cases, aggrieved people would seek help from the centre that sets the high expectations, and the centre often intervenes on a case-by-case basis.Footnote ⁴⁶ Fuelled by state propaganda, this interaction between the aggrieved and the centre constructs a narrative of rogue local officials defying a benevolent centre.Footnote ⁴⁷ When local implementation is successful, the centre still gets most of the credit. In his case study of the abolishment of school fees in 2004, Xiaobo Lü finds that despite local government's contribution, people credited only the central government with more trust.Footnote ⁴⁸

The conditioning effect of income

Before we hypothesize any relationship between inequality and political trust, we must consider the conditioning effect of income. Do the rich and poor react to inequality in similar or different ways? It may be argued that inequality causes the rich to trust the government more and the poor to trust the government less because higher levels of inequality mean that the rich become better off and the poor worse off, in a relative sense. Economists tell us that people are self-interested and gain more happiness from being richer than their neighbours than from being rich in absolute terms.Footnote ⁴⁹ Also, when being rational, the rich and the poor have different redistributive preferences.Footnote ⁵⁰ Therefore, inequality may increase the rich's political trust because it affords a sense of relative superiority.

However, we believe this view is too simplified. Rational thinking may dominate people's economic decisions, but one's trust in the government is more than an economic decision. We have to consider other determinants of human behaviour such as emotions and values. Research has documented a general human tendency to ensure a minimal living standard for everybody.Footnote ⁵¹ For the rich, this tendency may outweigh their sense of relative superiority supplied by inequality. Even if we concede that rational thinking looms large in the formation of political trust, it is still plausible that the rich distrust the government owing to inequality because their interests are at stake. Inequality increases crime rates, infant mortality rates, general mortality rates, and it may cause civil conflicts.Footnote ⁵² Any of the above could dwarf the rich's sense of relative superiority gained from the existence of inequality. Just imagine a well-to-do family deciding to emigrate. What would they consider when choosing their destination? How likely is it that they would choose a highly unequal country in order to feel richer?

Based on the foregoing theoretical arguments, we propose our hypothesis:

Inequality makes an individual lose trust in the local government but not in the central government, and this effect holds for both the rich and the poor.

Empirical Analysis

Data and measures

To test our hypothesis, we need data for both inequality and individuals. The individual-level data come from the Chinese General Social Survey (CGSS) 2010. The CGSS uses multistage stratified random sampling to draw a representative sample of the Chinese adult population. The survey includes a total of 11,783 completed interviews, with a response rate of 73.15 per cent. The data on inequality are based on the provincial Gini coefficient of household income calculated by Weimin Tian.Footnote ⁵³ We also use other provincial economic statistics. Gross provincial product per capita comes from the 2009 Statistical Yearbook. Data on the degree of marketization in 2009 are compiled by Fan Gang, Wang Xiaolu and and Zhu Hengpeng.Footnote ⁵⁴

Dependent variable: trust in the central and local governments

The CGSS 2010 asks respondents to rate their trust in the central and local governments on a 5-point scale (1 “no trust at all”; 2 “little trust”; 3 “neither trust nor distrust”; 4 “some trust”; and 5 “total trust”).Footnote ⁵⁵ Nationwide, 52 per cent of respondents trust the central government completely, while only 24 per cent trust the local government completely. At the other end of the scale, 3 per cent have little or no trust in the central government, while 16 per cent report similar distrust in the local government.

As Figures 1 and 2 show, levels of political trust vary across provinces. The two figures show the percentages of respondents who have total trust – darker colours mean higher percentages. Two patterns emerge. First, provinces with a high degree of trust in the central government tend to have high local trust. Second, in each province, trust in the centre is significantly higher than trust in the local government. For instance, people in both Gansu province and Zhejiang province trust the central government more than they do the local governments, but both central and local trust are considerably higher in Gansu (88 per cent and 35 per cent) than in Zhejiang (33 per cent and 18 per cent). The variance across the provinces calls for a provincial-level explanation.

Figure 1:

Complete Trust in the Central Government across Provinces

Figure 2:

Complete Trust in Local Government across Provinces

Independent variables: economic inequality and socio-economic status

Economic inequality is measured by the provincial Gini coefficient. Some may question whether provincial inequality has real impact on political trust, because the inequality people experience daily is likely to concern smaller communities like counties or neighbourhoods. Ideally, we could examine and compare provincial inequality and community-level inequality, yet such national data do not exist. Nonetheless, since cross-national studies frequently use national inequality to explain political behaviour, we are confident that provincial-level inequality has political significance.Footnote ⁵⁶

The provincial Gini coefficients are calculated by Tian based on the annual disposable income of urban households and net income of rural households, both gathered from the yearbooks of each province.Footnote ⁵⁷ The 2009 provincial Gini coefficient ranges from 0.28 in the city of Beijing to 0.48 in Guizhou (see Figure 3, Gini index multiplied by 10 for easier interpretation). We centre the Gini coefficient by subtracting the mean so that zero means the average inequality.

Figure 3:

Inequality across Provinces

Source: Tian 2012. Notes: Gini index in 10-point scale. Provinces not covered: Tianjin, Jilin, Shandong, Hunan, Hainan, Yunnan and Tibet.

To test whether the rich and the poor both lose trust owing to inequality, we need a variable on individual income. We constructed a composite index of socio-economic status (SES). We chose SES instead of an income variable because the latter does not fully capture people's real incomes. For many people working in the public sector, a significant portion of their real incomes has ambiguous legality and is unlikely to be reported on opinion surveys. But the SES index, which we constructed based on income, educational attainment and self-assessed social status, can reflect people's financial statuses more effectively without the need to ask sensitive questions. The income variables are personal and household annual income in 2009. Self-assessed social strata include both current status and status from ten years ago. Educational attainment is an ordinal variable corresponding to the educational system in China. We combine these variables into a standardized SES index with high values indicating higher statuses. The index has a reliability scale of 0.753, assuring high reliability.

Control variables

We control for some individual-level covariates that may influence political trust. The first group of control variables measures the sense of inequity. We include these controls to demonstrate that inequality may influence political trust independently from inequity, a point we made earlier. These control variables are: how fair is society (1 – not at all; 5 – totally fair); how fair is your own income (1 – not at all; 5 – totally fair); how severe is the conflict between the rich and the poor (1 – not at all; 5 – very severe); inequality is caused by the manipulations of the people with power (1 – strongly disagree; 5 – strongly agree). We expect the first two variables to have a positive influence on political trust and the last two variables to have a negative influence. The correlations between Gini coefficients and these four measures of inequity are low (see Table A2 in the Appendix), reinforcing the case that inequity cannot substitute for inequality in the evaluation of the social volcano thesis.

We also control for media use or primary sources of political information. Newspapers, magazines, radio and TV are classified as traditional media, and the internet and mobile phones as new media. The government has more control over the traditional media than it does over the new media, so we expect the use of new media to decrease political trust.Footnote ⁵⁸ We also control for whether respondents have experienced protests. Because exposure to protests gives people first-hand experiences of the real conflicts and injustice in society, we expect this variable to have a negative influence on political trust. Other individual-level control variables are employment status, age, gender, marital status CCP membership, and household registration status (hukou 户口).

Provincial-level control variables include gross provincial products per capita in 2009, the level of marketization in each province in 2009, and provincial spending on culture and media, public security, education, healthcare and environment protection in 2009 (percentage of total provincial expenditure). The marketization index captures the maturity of a province's private market since 1978.Footnote ⁵⁹ The marketization index in 2009 ranges from 0.38 for Tibet to 11.8 for Zhejiang, the latter province being more developed. We expect the first two provincial control variables to have a positive influence on political trust. The provincial spending variables can indicate the provision of social services, but they can also indicate a deficiency and a need for more spending. Therefore, we do not expect a specific direction for the effect of provincial expenditures on political trust.

The multilevel models are specified as follows:Footnote ⁶⁰

$$\eqalign{& Trust_{ij} = \gamma _0 + \gamma _1Socioeconomi\; status_{ij} + \gamma _kInd.\; Control_{ijk} + \varepsilon _{ij} \cr & \gamma _0 = \gamma _{00} + \gamma _{01}Gini_j + \gamma _{0k}Pro.Control_{\,jk} + \mu _j \cr & \gamma _1 = \gamma _{10} + \gamma _{11}Gini_j} $$

The dependent variables are trust in the central government and trust in local government. We employ multilevel analysis to control for provincial-level factors not already mentioned, such as population, geography, religion, ethnicity. In the equations, these factors are captured by μ _j.Footnote ⁶¹ We chose the multilevel linear model for its simplicity and ease of interpretation.

Results

Table 1 presents the results of four hierarchical linear models. The dependent variable of models 1 and 2 is trust in the central government, and the dependent variable of models 3 and 4 is trust in the local government. Models 1 and 3 are the baseline models, and models 2 and 4 have cross-level interactions between SES and provincial inequality. In model 1, neither SES nor the Gini coefficient is statistically significant, which suggests that inequality does not affect the general population's trust in the central government. Because our hypothesis has predictions for both the rich and poor, we still need to examine the interaction between SES and the Gini coefficient.

Table 1:

Determinants of Trust in the Central and Local Governments

Notes:

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1. Observations per group: minimum: 84, maximum: 590, average: 356.8.

In model 2, both SES and the interaction term are significant. Because the Gini coefficient is rescaled so that zero represents the average level of inequality, the negative effect of SES means that in a province with an average level of inequality, people with a high SES are less trusting of the central government than people with a low SES. Because SES also has a mean of zero, the insignificant coefficient of the Gini coefficient means that for a person with an average SES, living in a more unequal province does not affect trust in the central government. However, we cannot directly interpret the effect of inequality when SES takes on other values, so it is wrong to interpret the significant interaction as “rich people trust the central government more if the objective inequality is severe.”Footnote ⁶² Correct interpretation of interaction terms requires calculating the marginal effects of one variable conditional on each value of the other variable, as well as the confidence intervals of these marginal effects.Footnote ⁶³ We employ simulations to calculate the marginal effects of inequality on trust in the central government across socio-economic status. The results are presented in Figure 4. Because the 95 per cent confidence intervals all cross zero, the figure shows that economic inequality has no influence on trust in the central government for either the rich or the poor, supporting our hypothesis. To determine whether inequality decreases trust in the local governments, as the hypothesis predicts, we turn to models 3 and 4.

Figure 4:

Effects of Increasing Gini Coefficient by 0.1 on the Level of Trust in the Central Government across Socio-economic Status

Note: Whiskers are 95% confidence intervals.

In model 3, the Gini coefficient is negative and statistically significant. This means that for the general population, living in an unequal province decreases trust in the local government. But does it hold across all SES groups? The interaction term in model 4 indicates that it does. Figure 5 presents the marginal effects of inequality at each level of SES. It shows that the negative effect of inequality does not change with different levels of SES. In other words, inequality decreases the trust in local governments of both the rich and the poor. At this point, our hypothesis receives full support.

Figure 5:

Effects of Increasing Gini Coefficient by 0.1 on the Level of Trust in Local Government across Socio-economic Status

Note: Whiskers are 95% confidence intervals.

Yet we still want to probe how much of an influence inequality exerts. Because the mean of SES is zero, the -0.6 coefficient of Gini means that for a person with an average SES, a 0.1 increase of Gini coefficient decreases local trust by 0.6 on a 5-point scale. As mentioned, in 2009, Guizhou was the most unequal (0.48 Gini) and Beijing was the least unequal (0.28 Gini). There is a 0.2 difference in their Gini coefficients, which according to our simulations, may result in a 1.2 decrease of local trust on a 5-point scale. That is to say, from a “total trust” to “some trust,” from “some trust” to “neither trust nor distrust,” and so forth.

Some other results merit comments. Socio-economic status has no influence on central trust (model 1, significant at a 90 per cent level) but it has a negative influence on local government (model 3). This is contrary to the economic voting spirit that sees people reward the government for good personal circumstances, and it casts doubts on the effectiveness of the CCP enhancing regime legitimacy through economic development.Footnote ⁶⁴ But why do people with a higher SES have less trust in local government? Maybe people with a higher SES tend to have higher expectations of government performance. Or, maybe these people are more informed of government performance. Variables of inequity behave as expected across the models: the sense of inequity decreases both central trust and local trust. The fact that inequality exerts influence independent of inequity highlights the risk of confounding the two concepts in inequality research. Exposure to protests has the expected negative effects on both central and local trust. Use of new media reduces both central and local trust. As for the provincial-level variables, gross provincial product per capita has a negative effect on both local trust and central trust, again raising doubts about the effectiveness of economic development for strengthening regime legitimacy. Provincial marketization has no effect on central trust but a negative effect on local trust – another instance of the higher the level of development, the lower the trust in local government. The positive effect of cultural and media spending on both central and local trust demonstrates that propaganda works, which is consistent with the result of the media type variable. That public security (gonggong anquan 公共安全 / weiwen 维稳) spending has a negative impact on central trust but not on local trust is an intriguing result that demands further research. The negative influence of healthcare and environmental spending on local trust looks odd at first glance; however, correlations between provincial Gini and provincial spending provide a clue to this result. Table A7 in the Appendix shows that expenditures on education, healthcare and the environment are all positively correlated with Gini coefficient – the more unequal a province, the more it spends on education, healthcare and the environment. Since it is highly improbable that social spending makes a province more unequal, it must be that more unequal provinces feel the pressure to increase spending to remedy past inequalities. This interpretation is consistent with our theory that people hold local government responsible for inequality.

Conclusion

Our theoretical arguments and empirical results have made a positive case for the social volcano thesis, which has been all but buried by previous findings that the Chinese are good at enduring inequality. We show that inequality reduces people's trust in the local government but not in the central government. Our finding fits the big picture of Chinese politics painted by decades of studies: the central government enjoys unswerving trust while distrust is deflected to the local government. But, if inequality only affects the trust in local government, how dangerous is it? Is it worthy of the name social volcano?

We believe it is. First, inequality may indirectly threaten the regime by spurring local unrest. It has been repeatedly shown that local dissatisfaction is responsible for the vast majority of protests.Footnote ⁶⁵ Although protests seldom challenge the central government, they often morph and spread. They are especially dangerous when the centre is expected to provide the final solution but fails to do so. The central government's hyper responsiveness to citizens’ demands attests to the volatile nature of protests in China.Footnote ⁶⁶

Second, distrust in the local government reflects latent distrust in the central government. Lianjiang Li compares the attitudes of those who have high levels of trust in both the central and local governments with those who have more trust in the central government than in local government (Li calls this latter group “holders of hierarchical trust”).Footnote ⁶⁷ The comparison shows that although both groups of people report high levels of trust in the central government, holders of hierarchical trust have more negative assessments of the central government in other issues, such as China's democratic performance. Li further suggests that underneath the hierarchical trust is a belief that the central government is well meaning but totally incompetent, and this belief can be a source of instability.Footnote ⁶⁸ In short, inequality may be continually eroding people's faith in the regime even though the central trust remains strong on the survey.

Inequality is politically dangerous also because it causes not only the poor but also the rich to lose trust in the government. Poor people's loss of trust is not compensated for with an increase in trust held by rich people. The negative effect is uniform across SES groups. Deng Xiaoping 邓小平 famously justified rising inequality by declaring that “it is appropriate to let some people get rich first” (rang yibufen ren xian fu qilai 让一部分人先富起来), but today the mantra seems to have lost its appeal to even the supposed beneficiaries of inequality.

Our study has taken a small step towards a fuller understanding of the political consequences of inequality in China. Owing to data constraints, we have only considered the provincial Gini coefficient. Future studies may examine the Gini coefficients below the provincial level and also look at indicators of regime stability other than political trust. But, now we conclude our study by affirming that a social volcano does indeed exist.

Acknowledgements

The authors thank Wenfang Tang, John Kennedy, Lianjiang Li, Frederick Solt, Rene Rocha, Julianna Pacheco and the two anonymous reviewers for their comments, suggestions and encouragement.

Biographical notes

Yingnan Joseph Zhou is an assistant professor of political science at the University of Texas at El Paso. His research interests include governance, Chinese politics, public opinion and behaviour, and authoritarian politics.

Shuai Jin is an assistant professor of political science at the University of Massachusetts, Boston. She studies economic inequality and redistribution, quantitative methods, Chinese politics and public opinion.

Appendix

Table A1:

Summary Statistics of All Variables

Table A2:

Correlations between Variables

Table A3:

Correlations between Local Trust, Central Trust and Trust in the Judicial System

Table A4:

Cross-tabulation between Trust in the Judicial System and Trust in the Central Government

Table A5:

Cross-tabulation between Trust in the Central Government and Trust in the Local Government

Table A6:

Determinants of Trust in the Central Government

Table A7:

Summary Statistics of Province Fiscal Expenditure Variables

Footnotes

1 Wang Reference Wang2016.

2 State Council Information Office 2013.

3 Solt Reference Solt2016.

4 Merton Reference Merton1938; Shaw and McKay Reference Shaw and McKay1942; Becker Reference Becker1968; Kelly Reference Kelly2000; Pamuk Reference Pamuk1988; Wennemo Reference Wennemo1993; Mayer and Sarin Reference Mayer and Sarin2005; Hosseinpoor et al. Reference Hosseinpoor, Van Doorslaer, Speybroeck, Naghavi, Mohammad, Majdzadeh, Delavar, Jamshidi and Vega2006.

5 Ma Reference Ma2005; Kahn Reference Kahn2006; Hu and Hu Reference Hu and Hu2007; Shirk Reference Shirk2008, 30–31.

6 Reference GurrGurr 1970.

7 Shi Reference Shi2000; Reference Shi2001; Chen, Jie Reference Chen2004; Kennedy Reference Kennedy2009; Tang Reference Tang2016; Zhou and Ou-Yang Reference Zhou and Ou-Yang2017.

8 Wu Reference Wu2009; Whyte Reference Whyte2010; Whyte Reference Whyte2016; Whyte and Im Reference Whyte and Im2014.

9 Whyte's (Reference Whyte and Im2014; Reference Whyte2016) findings about income inequality are consistent with his findings in Whyte Reference Whyte2010. Yet, in the 2014 and 2016 studies, he changed the phrase no volcano to dormant volcano.

10 Gamson Reference Gamson1975; Lichbach Reference Lichbach1998.

11 Tang Reference Tang2016.

12 Li Reference Li2016.

13 Ma Reference Ma2005.

14 Kahn Reference Kahn2006.

15 Ma Reference Ma2005; Kahn Reference Kahn2006; Hu and Hu Reference Hu and Hu2007; Shirk Reference Shirk2008, 30–31.

16 Figures for people who fully or somewhat trust the national government are: 89% in China (2008 China Survey); 38% in Germany (Eurobarometer 2010); 32% in Taiwan (ABSIII); 24% in Britain (Eurobarometer 2010); 20% in the United States (EGSS 2010); 13% in France (Eurobarometer 2010); and 12% in Japan (ABSIII).

17 Chen, Xueyi, and Shi Reference Chen and Shi2001; Shi Reference Shi2001; Chen, Jie Reference Chen2004; Tsai Reference Tsai2007; Tang Reference Tang2016.

18 Wu Reference Wu2009.

19 Footnote Ibid., 1047.

20 Whyte Reference Whyte2010.

21 Footnote Ibid., 48.

22 Wu Reference Wu2009.

23 Whyte Reference Whyte2010.

24 Footnote Ibid.

25 Whyte and Im Reference Whyte and Im2014; Whyte Reference Whyte2016.

26 Wu Reference Wu2009; Whyte Reference Whyte2010.

27 Gamson Reference Gamson1975.

28 Lichbach Reference Lichbach1998.

29 Zhang Reference Zhang, Nathan and Link2002.

30 Wu Reference Wu2009; Whyte Reference Whyte2010; Whyte and Im Reference Whyte and Im2014.

31 Converse Reference Converse and Apter1964; Zaller Reference Zaller1992.

32 Lewis-Beck and Stegmaier Reference Lewis-Beck, Stegmaier, Dalton and Klingemann2007.

33 Wu Reference Wu2009; Whyte Reference Whyte2010.

34 See Confucian Analects, Book XVI: Ke She, Ch. 1

35 Tang Reference Tang2016.

36 Liu Reference Liu2011.

37 Zhou Reference Zhou2016.

38 Wu Reference Wu2009; Whyte Reference Whyte2010.

39 Whyte Reference Whyte2010, 44, 45, 48.

40 Wu Reference Wu2009, 1043.

41 Li Reference Li2004.

42 Pei Reference Pei2002; Manion Reference Manion2004; O'Brien and Li Reference O'Brien and Li2006; Tang Reference Tang2016. The “centre” (zhongyang) is how the Chinese refer to the central government. Here, we use the centre and the central government interchangeably.

43 “Guowuyuan guanyu zai quanguo jianli chengshi jumin zuidi shenghuobaozhang zhidu de tongzhi” (State Council's notice about establishing the institution of urban minimum living standards), 2016, http://www.gov.cn/zhengce/content/2016-10/19/content_5121479.htm. Accessed 7 September 2018.

44 Whyte Reference Whyte2016, 35.

45 Hernandez Reference Hernandez2016.

46 O'Brien and Li Reference O'Brien and Li2006; Tang Reference Tang2016.

47 Li Reference Li2004; Kennedy Reference Kennedy2009; Lü Reference Lü2014; O'Brien and Li Reference O'Brien and Li2006; Tang Reference Tang2016; Cui et al. Reference Cui, Tao, Warner and Yang2015.

48 Lü Reference Lü2014.

49 Easterlin Reference Easterlin1974.

50 Meltzer and Richard Reference Meltzer and Richard1981.

51 Bowles and Gintis Reference Bowles and Gintis2000; Dion and Birchfield Reference Dion and Birchfield2010.

52 Merton Reference Merton1938; Shaw and McKay Reference Shaw and McKay1942; Becker Reference Becker1968; Kelly Reference Kelly2000; Pamuk Reference Pamuk1988; Wennemo Reference Wennemo1993; Mayer and Sarin Reference Mayer and Sarin2005; Hosseinpoor et al. Reference Hosseinpoor, Van Doorslaer, Speybroeck, Naghavi, Mohammad, Majdzadeh, Delavar, Jamshidi and Vega2006; Wilkinson Reference Wilkinson and Wilkinson1986; Reference Wilkinson1990; Kawachi et al. Reference Kawachi, Kennedy, Lochner and Prothrow-Stith1997; Russett Reference Russett1964; Gurr 1970.

53 Tian Reference Tian2012.

54 Fan, Wang and Zhu Reference Fan, Wang and Zhu2011.

55 The survey question is vague about which level of local government is being asked about in urban areas. For the urban respondents, it only asks about the “local government,” but for the rural respondents, it asks about the township government. Without knowing whether and how the interviewers specified the level of urban government, we can only interpret the urban data as trust in local government in general.

56 Solt Reference Solt2008; Reference Solt2010; Rosset, Giger and Bernauer Reference Rosset, Giger and Bernauer2013; Dion and Birchfield Reference Dion and Birchfield2010; Andersen and Fetner Reference Andersen and Fetner2008.

57 Tian Reference Tian2012. Hunan, Hainan, Jilin, Shandong, Tianjin, Yunnan and Tibet are excluded from the analysis because Gini coefficients for these provinces are unavailable.

58 Kennedy Reference Kennedy2009; Tang Reference Tang2016.

59 See Fan, Wang and Zhu Reference Fan, Wang and Zhu2011 for how the index is calculated.

60 The models include individual-level sampling weights. They are also adjusted for the complex survey design. The results are robust to these adjustments.

61 Our results are robust to the coding of the dependent variables (binary, ordinal or linear) and the corresponding models (multilevel logistic model, multilevel ordered logistic model or multilevel linear model). We did not choose the logistic models because sampling weights cannot be applied and interpretation based on predicated probabilities is needlessly convoluted.

62 Model 2 is:

$$\eqalign{Trust{\rm \;} in{\rm \;} Central{\rm \;} Gov. =\;& {\hat{\beta}} _0 + {\hat{\beta}} _1SESindex + {\hat{\beta}} _2Gini + {\hat{\beta}} _3SES{\rm \;} X{\rm \;} Gini \cr & + \sum {\hat{\beta}} _iControl_i = {\hat{\beta}} _0 + {\hat{\beta}} _1SESindex \cr & + \lpar {{\hat{\beta}}_2 + {\hat{\beta}}_3SES} \rpar Gini + \sum{\hat{\beta}} _iControl_i} $$

We interpret the effect of change in Gini on the level of trust: $\lpar {{\hat{\beta}}_2 + {\hat{\beta}}_3SES} \rpar $, and this effect varies across SES levels, as does the statistical significance of this effect. The standard error of $\lpar {{\hat{\beta}}_2 + {\hat{\beta}}_3SES} \rpar \; $ changes with the value of SES:

$$\eqalign{se\lpar {\widehat{{\beta_2}} + \widehat{{\beta_3}}SES} \rpar & = \sqrt {var\lpar {\widehat{{\beta_2}} + \widehat{{\beta_3}}SES} \rpar } \cr & = \sqrt {var\lpar {\widehat{{\beta_2}}} \rpar + SES^2var\lpar {\widehat{{\beta_3}}} \rpar + 2SEScov\lpar {\widehat{{\beta_2}},\; \widehat{{\beta_3}}} \rpar }} $$

We visualize the size of this effect with its corresponding 95% confidence intervals at 7 points of SES in Figure 4.

Brambor, Clark and Golder (Reference Brambor, Clark and Golder2006) articulate that focusing on the statistical significance of the interaction term could only be misleading; its significance does not necessarily imply any effect of the combined term.

63 As Brambor, Clark and Golder Reference Brambor, Clark and Golder2006 shows, even if none of the base variables and the interaction term are statistically significant, the marginal effects of one variable may still be statistically significant conditional on certain values of the other variable.

64 Downs Reference Downs1957; Lewis-Beck Reference Lewis-Beck1985; Sigelman, Sigelman and Bullock Reference Sigelman, Sigelman and Bullock1991.

65 Pei Reference Pei2002; Manion Reference Manion2004; O'Brien and Li Reference O'Brien and Li2006; Tang Reference Tang2016.

66 Tang Reference Tang2016.

67 Li Reference Li2016.

68 Footnote Ibid., 116.

Notes:

Pearson chi2(16) = 9.6e + 03 Pr = 0.000.

Notes:

Pearson chi2(16) = 8.0e + 03 Pr = 0.000.

Notes:

Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

References

Andersen, Robert, and Fetner, Tina. 2008. “Economic inequality and intolerance: attitudes toward homosexuality in 35 democracies.” American Journal of Political Science 52(4), 942–958.Google Scholar

Becker, Gary S. 1968. “Crime and punishment: an economic approach.” Journal of Political Economy 76, 169–217.Google Scholar

Bowles, Samuel, and Gintis, Herbert. 2000. “Reciprocity, self-interest, and the welfare state.” Nordic Journal of Political Economy 26(1), 33–53.Google Scholar

Brambor, Thomas, Clark, William Roberts and Golder, Matt. 2006. “Understanding interaction models: improving empirical analyses.” Political Analysis 14(1), 63–82.Google Scholar

Chen, Jie. 2004. Popular Political Support in Urban China. Washington, DC: Woodrow Wilson Center Press.Google Scholar

Chen, Xueyi, and Shi, Tianjian. 2001. “Media effects on political confidence and trust in the People's Republic of China in the post-Tiananmen period.” East Asia 19(3), 84–118.Google Scholar

Converse, Philip E. 1964. “The nature of belief systems in mass publics.” In Apter, David E. (ed.), Ideology and Discontent. New York: The Free Press of Glencoe, 206–258.Google Scholar

Cui, Ernan, Tao, Ran, Warner, Travis J. and Yang, Dali L.. 2015. “How do land takings affect political trust in rural China?” Political Studies 63(S1), 91–109.Google Scholar

Dion, Michelle L., and Birchfield, Vicki. 2010. “Economic development, income inequality, and preferences for redistribution.” International Studies Quarterly 54(2), 315–334.Google Scholar

Downs, Anthony. 1957. “An economic theory of political action in a democracy.” The Journal of Political Economy 65(2), 135–150.Google Scholar

Easterlin, Richard A. 1974. “Does economic growth improve the human lot? Some empirical evidence.” Nations and Households in Economic Growth 89, 89–125.Google Scholar

Fan, Gang, Wang, Xiaolu and Zhu, Hengpeng. 2011. Zhongguo shichanghua zhishu: gediqu shichanghua xiangdui jincheng 2011 nian baogao (NERI Index of Marketization of China's Provinces 2011 Report). Beijing: Jingji kexue chubanshe.Google Scholar

Gamson, William A. 1975. The Strategy of Social Protest. Homewood, IL: Dorsey Press.Google Scholar

Gurr, Ted Robert. 1970. Why Men Rebel. Princeton, NJ: Princeton University Press.Google Scholar

Hernandez, Javier C. 2016. “Mother's killing of children in rural China spurs debate about inequality,” New York Times, 12 September, https://www.nytimes.com/2016/09/13/world/asia/mothers-killing-of-children-in-rural-china-spurs-debate-about-inequality.html.Google Scholar

Hosseinpoor, Ahmad Reza, Van Doorslaer, Eddy, Speybroeck, Niko, Naghavi, Mohsen, Mohammad, Kazem, Majdzadeh, Reza, Delavar, Bahram, Jamshidi, Hamidreza and Vega, Jeanette. 2006. “Decomposing socioeconomic inequality in infant mortality in Iran.” International Journal of Epidemiology 35(5), 1211–19.Google Scholar

Hu, Lianhe, and Hu, Angang. 2007. “Pinfu chaju shi ruhe yingxiang shehui wending de?” (How does the gap between rich and poor influence social stability?). Jiangxi shehui kexue 9, 142–151.Google Scholar

Kahn, Joseph. 2006. “China makes commitment to social harmony,” New York Times, 12 October.Google Scholar

Kawachi, Ichiro, Kennedy, Bruce P., Lochner, Kimberly and Prothrow-Stith, Deborah. 1997. “Social capital, income inequality, and mortality.” American Journal of Public Health 87(9), 1491–98.Google Scholar

Kelly, Morgan. 2000. “Inequality and crime.” Review of Economics and Statistics 82(4), 530–39.Google Scholar

Kennedy, John James. 2009. “Maintaining popular support for the Chinese Communist Party: the influence of education and the state-controlled media.” Political Studies 57(3), 517–536.Google Scholar

Lewis-Beck, Michael S. 1985. “Pocketbook voting in US national election studies: fact or artifact?” American Journal of Political Science 29(2), 348–356.Google Scholar

Lewis-Beck, Michael S., and Stegmaier, Mary. 2007. “Economic models of voting.” In Dalton, Russell and Klingemann, Hans-Dieter (eds.), The Oxford Handbook of Political Behavior. Oxford: Oxford University Press, 1–38.Google Scholar

Li, Lianjiang. 2004. “Political trust in rural China.” Modern China 30(2), 228–258.Google Scholar

Li, Lianjiang. 2016. “Reassessing trust in the central government: evidence from five national surveys.” The China Quarterly 225, 100–121.Google Scholar

Lichbach, Mark Irving. 1998. The Rebel's Dilemma. Ann Arbor, MI: The University of Michigan Press.Google Scholar

Liu, Weibing. 2011. “Qinli zhe huiyi Wen Jiabao zongli wei nongmingong tao gongzi quan guocheng” (Witness recollects Premier Wen Jiabao's efforts to get peasant workers paid), 6 July, http://news.ifeng.com/history/zhongguoxiandaishi/detail_2011_07/06/7463853_0.shtml. Accessed 10 March 2017.Google Scholar

Lü, Xiaobo. 2014. “Social policy and regime legitimacy: the effects of education reform in China.” American Political Science Review 108(2), 423–437.Google Scholar

Ma, Josephine. 2005. “Wealth gap fueling instability, studies warn,” South China Morning Post, 22 December.Google Scholar

Manion, Melanie. 2004. Corruption by Design: Building Clean Government in Mainland China and Hong Kong. Cambridge, MA: Harvard University Press.Google Scholar

Mayer, Susan E., and Sarin, Ankur. 2005. “Some mechanisms linking economic inequality and infant mortality.” Social Science & Medicine 60(3), 439–455.Google Scholar

Meltzer, Allan H., and Richard, Scott F.. 1981. “A rational theory of the size of government.” The Journal of Political Economy 89(5), 914–927.Google Scholar

Merton, Robert. 1938. “Social structure and anomie.” American Sociological Review 3, 672–682.Google Scholar

O'Brien, Kevin, and Li, Lianjiang. 2006. Rightful Resistance in Rural China. New York: Cambridge University Press.Google Scholar

Pamuk, Elsie R. 1988. “Social-class inequality in infant mortality in England and Wales from 1921 to 1980.” European Journal of Population/Revue Européenne de Démographie 4(1), 1–21.Google Scholar

Pei, Minxin. 2002. “China's governance crisis.” Foreign Affairs 81(5), 96–119.Google Scholar

Rosset, Jan, Giger, Nathalie and Bernauer, Julian. 2013. “More money, fewer problems? Cross-level effects of economic deprivation on political representation.” West European Politics 36(4), 817–835.Google Scholar

Russett, Bruce M. 1964. “Inequality and insurgency: the relation of land tenure to politics.” World Politics 16, 442–454.Google Scholar

Shaw, Clifford, and McKay, Henry. 1942. Juvenile Delinquency and Urban Areas. Chicago: University of Chicago Press.Google Scholar

Shi, Tianjian. 2000. “Cultural values and democracy in the People's Republic of China.” The China Quarterly 162, 540–559.Google Scholar

Shi, Tianjian. 2001. “Cultural values and political trust: a comparison of the People's Republic of China and Taiwan.” Comparative Politics 33(4), 401–419.Google Scholar

Shirk, Susan L. 2008. China: Fragile Superpower. New York: Oxford University Press.Google Scholar

Sigelman, Lee, Sigelman, Carol K. and Bullock, David. 1991. “Reconsidering pocketbook voting: an experimental approach.” Political Behavior 13(2), 129–149.Google Scholar

Solt, Frederick. 2008. “Economic inequality and democratic political engagement.” American Journal of Political Science 52(1), 48–60.Google Scholar

Solt, Frederick. 2010. “Does economic inequality depress electoral participation? Testing the Schattschneider hypothesis.” Political Behavior 32(2), 285–301.Google Scholar

Solt, Frederick. 2016. “The Standardized World Income Inequality Database.” Social Science Quarterly 97(5), 1267–81Google Scholar

State Council Information Office. 2013. “0.47–0.49 guojia tongji ju shouci fabu shi nian jini xishu” (National Bureau of Statistics publishes the Gini coefficients for the first time in a decade), http://www.scio.gov.cn/zhzc/2/32764/Document/1421797/1421797.htm. Accessed 20 March 2017.Google Scholar

Tang, Wenfang. 2016. Populist Authoritarianism: Chinese Political Culture and Regime Sustainability. Oxford: Oxford University Press.Google Scholar

Tian, Weimin. 2012. “Shengyu jumin shouru jini xishu cesuan jiqi biandong qushi fenxi” (Provincial income Gini coefficient: levels and changes). Jingji kexue 2, 48–59.Google Scholar

Tsai, Lily L. 2007. “Solidary groups, informal accountability, and local public goods provision in rural China.” American Political Science Review 101(02), 355–372.Google Scholar

Wang, Jianlin. 2016. “Behind Wang Jianlin's a hundred million small target: eight thousand million will also do,” Nengliang Media, 30 August, https://www.youtube.com/watch?v=M_V3l3znyAk&t=12s.Google Scholar

Wennemo, Irene. 1993. “Infant mortality, public policy and inequality – a comparison of 18 industrialised countries 1950–85.” Sociology of Health & Illness 15(4), 429–446.Google Scholar

Whyte, Martin. 2010. Myth of the Social Volcano: Perceptions of Inequality and Distributive Injustice in Contemporary China. Stanford, CA: Stanford University Press.Google Scholar

Whyte, Martin King. 2016. “China's dormant and active social volcanoes.” The China Journal 75(1), 9–37.Google Scholar

Whyte, Martin King, and Im, Dong-Kyun. 2014. “Is the social volcano still dormant? Trends in Chinese attitudes toward inequality.” Social Science Research 48, 62–76.Google Scholar

Wilkinson, Richard G. 1986. “Income and mortality.” In Wilkinson, Richard G. (ed.), Class and Health: Research and Longitudinal Data. London: Tavistock, 88–114.Google Scholar

Wilkinson, Richard G. 1990. “Income distribution and mortality: a ‘natural’ experiment.” Sociology of Health & Illness 12(4), 391–412.Google Scholar

Wu, Xiaogang. 2009. “Income inequality and distributive justice: a comparative analysis of mainland China and Hong Kong.” The China Quarterly 200, 1033–52.Google Scholar

Zaller, John. 1992. The Nature and Origins of Mass Opinion. New York: Cambridge University Press.Google Scholar

Zhang, Liang, edited by Nathan, Andrew J. and Link, Perry. 2002. The Tiananmen Papers. New York: PublicAffairs.Google Scholar

Zhou, Yingnan Joseph. 2016. “How does voting in grassroots elections influence assessment of democracy in China?” Journal of Chinese Political Science 23(2), 1–25.Google Scholar

Zhou, Yingnan Joseph, and Ou-Yang, Ray. 2017. “Explaining high external efficacy in authoritarian countries: a comparison of China and Taiwan.” Democratization 24(2), 283–304.Google Scholar