Does China’s middle class prefer (liberal) democracy?

ABSTRACT Many surveys show that China’s political regime, under the Chinese Communist Party’s authoritarian rules, enjoys a high level of public support. However, it is still uncertain whether China’s emerging middle class will become the “agent of democratization” as suggested by modernization theory. Using the data of Asian Barometer Survey conducted in China in 2011, this article demonstrates that the relationship between class identity and preference for liberal democracy in China may be inverted U-shaped. The Chinese middle class shows a higher preference to features of liberal democratic regimes than its counterparts of the lower- and upper-class. Members of the Chinese middle class also tend to regard democracy as the best form of government. Thus, the middle class has the potential to initiate democratization in China if the Chinese government fails to keep satisfying the middle class’ quest for economic well-being and protection of property rights.


Introduction
One enduring puzzle in the literature of comparative democratization is the coexistence of economic development and resilient authoritarianism in many developing countries, especially the case of China. Modernization theory predicts that economic growth facilitates democratization due to the emergence of the middle class as "the agent of democratization". 1 Yet, the rise of middle class in China casts a serious doubt on the validity of this theory because the Chinese middle class is rather conservative and supportive of the authoritarian regime after three decades of China's double-digit economic growth. 2 Furthermore, the economic development in China results in "capitalism without democracy" 3 where members of the Chinese middle class become the "allies of the state". 4 As a result, the authoritarian rule under the Chinese Communist Party (CCP) seems to become more resilient regardless of its economic achievement and the growth of middle class. 5 Figure 1 illustrates this puzzle. Based on the third-wave Asian Barometer Survey China (ABS3-China) Project conducted in 2011, we calculate the proportion of respondents expressing their support for China's current political regime according to different levels of subject social status. As Figure 1 demonstrates, Chinese middle class' regime support, either measured in an absolute sense (Q82) or in a comparative fashion (Q83), is about the same as it is for other classes. Thus, the Chinese middle class shows no more potential to become "agent of democratization" than its lower and higher counterparts.
Rather than being an agent of democratization, numerous studies have demonstrated that members of the Chinese middle class are major supporters of the authoritarian rule. Due to China's remarkable economic development, the Chinese middle class benefits from the state-led economic development and becomes economically dependent on the state. 6 Consequently, the middle class in China is less likely to support democratization than its counterparts in other late-developing countries, such as Brazil, South Korea, and Taiwan.
Nevertheless, the support for an authoritarian regime does not address whether the Chinese middle class has been instilled with any democratic values after the economic boom in China. If members of the Chinese middle class bear no democratic ideas, their support for the authoritarian rule is consistent with their value system. If they do hold democratic values dearly, their support for the authoritarian rule is in tension with their value system. Failing to respond to the tension will threaten the political stability of authoritarian rule in China. However, existing studies are inconclusive regarding this issue. Some studies find that the Chinese middle class has fewer democratic orientations than other classes, 7 while others offer contradictory evidence. 8 Recent scholarship develops an eclectic approach to reconcile this debate. Members of the Chinese middle class are aware of the values of liberal democracy, such as political rights and freedom, but they prioritize the social and political stability to keep enjoying the economic benefits generated by the authoritarian regime. 9 In this article, we join this eclectic view and argue that the Chinese middle class' high support for the current authoritarian regime does not preclude the prospect of democratization in China. Instead, the emerging Chinese middle class has become more supportive of ideas of liberal democracy than its lower and higher counterparts, because newly enriched citizens prefer more protection of their property rights from state appropriation. According to theories of comparative democratization, ordinary citizens rely on the establishment of liberal democratic institutions, such as representative legislatures or competitive elections, to constrain rulers' temptation to seize their wealth. 10 Since Chinese citizens are more aware of their rights since the 1990s, 11 especially in cases of land expropriation and housing rights, 12 the Chinese middle class would become more supportive of liberal democracy that may better protect property rights.
With the data of the ABS3-China Project, we demonstrate that China's middle class prefers a political regime with liberal democratic characteristics, such as representation, media freedom, and competitive election. Moreover, we find that the Chinese middle class' preference for a democratic regime is higher than its lower and higher classes. The Chinese middle class also tends to regard democracy as the best form of the government. The findings suggest that the rapid economic growth has sowed the seeds of liberal democratic ideas in the mindset of Chinese middle class. The gap in their support for the authoritarian regime is indeed in conflict with their political values. Thus, the current Chinese political system seems to gain legitimacy and maintain stability by responding to the middle's class quest for economic well-being and protection of property rights, 13 but it may face a risk of democratic transition if it fails to keep satisfying those demands.
Our findings advance the literature on the democratic values of the Chinese middle class. Although some studies have argued that the Chinese middle class has a preference for features of liberal democracy, their empirical evidence is usually limited to the urban area or a few localities in China. With the nationally representative sample collected by the ABS3-China Project, our finding could be generalized to infer democratic values of the Chinese middle class.
In addition, our study clarifies some ambiguities of previous studies on the Chinese middle class' democratic values. As it is straightforward to investigate democratic support by asking respondents how much they support democracy, answers to this type of question could be just a lip service and result in social desirability bias. 14 The multi-faced meaning of democracy may also mislead the inference on respondents' democratic support. 15 Alternatively, we use a set of questions on features of liberal democracy, including representation, competitive election, media freedom, and multi-party competition. Based on respondents' answers, we construct a measure of preference for liberal democracy. We believe that our measure would have a greater validity than other indicators constructed on the basis of a single question.
Our study sheds light on the issue of authoritarian resilience in China. Given their relatively high degree of support for liberal democracy, the Chinese middle class still has the potential to act as the agent of democratizationif democratization in the sense of liberal democracy should ever take place in China. Thus, the next question should be what are the factors that prevent the middle class' preference for liberal democracy from translating into democratization. Scholars should take a careful look at the intervening variables if they wish to explain the absence of democratization in China and the high degree of support for the current regime. For example, the wellbeing of the Chinese middle class may be largely dependent on the state apparatus. Members of the Chinese middle class would suffer from political instability and choose to support the current regime. 16 In addition, the middle class may be deterred by the high cost of coordinating collective actions against the state. 17 Furthermore, the political elites may utilize the state coercive capacityincluding the use of the militaryto deter the middle class from challenging its authoritarian rule. 18 The rest of this paper is organized as follows. In the next section, we review how existing studies investigate the attitude of the Chinese middle class toward democracy. In the empirical section, we present our research design and empirical evidence to demonstrate that the middle class in China is more likely to prefer a democratic regime than the lower or upper class. Finally, we conclude by discussing how our findings shed light on the prospects for democratization in China.

The middle class and democratization in China
Proponents of modernization theory contend that a linear relationship exists between economic development and democratization. Specifically, economic development fosters the middle class, who then demands more political rights to protect property from confiscation by the state and further becomes the agent of democratization. 19 Accordingly, one key implication of modernization theory is that democratization follows the increases of per capita income. 20 Although other studies disagree on this optimistic prediction, 21 recent cross-national studies offer more empirical evidence to demonstrate the positive relationship between income and democracy. 22 The case of China, however, presents a serious challenge to the view that there is a causal relationship between the development of a middle class and democratization. The evidence, either collected from the macro-or micro-level, seems to contradict modernization theory. At the macro-level, three decades of rapid economic growth since 1978 have indeed fostered the rise of Chinese middle class. The size of the middle class, according to different estimates, ranges from 10% to 28% of the total Chinese population in the early 2010s, 23 and it is predicted to increase to more than 40% in 2020. 24 Nevertheless, there is still no sign of democratization in China. Even though China has replaced Japan as the second-largest economy in the world in 2010, it has always been rated as a "not free" country by the Freedom House and has a Polity score of −7 as an autocracy since 1979. Some studies even argue that China could "export" its model of authoritarian rule to other countries and block their quest for democratization when China deems it necessary to sustain its regime survival. 25 The micro-level evidence corroborates this pessimistic view toward the democratic future of China. The Chinese middle class, as illustrated in Figure 1, is actually rather conservative and supportive of China's authoritarian regime. Studies using other sources of data also suggest that the majority of Chinese citizens, especially those belonging to the middle class, are very status quo-oriented, 26 less supportive of democratic values, 27 and detached from democratic reforms. 28 Why does modernization theory fail to explain China? One line of research answers this question by looking for confounding factors in the relationship between the development of a middle class and democratic transition. Those factors include the economic well-being of the middle class, 29 the state-society relationship, 30 the coercive state capacity, 31 and the lack of meaningful political identity among newly enriched Chinese citizens. 32 We will discuss these factors in more details.
First, scholars argue that China's remarkable economic growth provides regime legitimacy and garners mass political support for the authoritarian regime. 33 As the emerging middle class is benefiting from the economic development, its members are satisfied with the regime and reluctant to see any changes that may result in political and economic instability. 34 Members of the Chinese middle class have become "allies of the state" and support the authoritarian rule of the CCP. 35 In particular, Chen and Lu argue that China, as a late-developing country, has two unique tools to resist the pressure of democratization among the middle class during the pursuit of economic development: the dominance of the CCP and the prerogative state capacity. 36 The majority of the Chinese middle class is involved in the nexus of the party-state and then becomes more dependent on the benefits and privileges offered by the authoritarian regime. As a consequence, the middle class in China are less likely to support any transitions to democracy or regime changes for self-interest.
This argument regarding the Chinese middle class' dependence on the state is further elaborated by Jie Chen. 37 In particular, Chen contends that the Chinese middle class is dependent on the state in two aspects: institutional and ideational. 38 For the institutional aspect, members of the Chinese middle class are less supportive of democratic values and regime changes if they hold the membership of the CCP or are employed by the state-own enterprises. For the ideational aspect, they are more supportive of the current regime than other classes if they share the same political values with the CCP, such as the priority of economic growth and regime stability.
The Chinese middle class' mindset of pursuing economic benefit and political stability is also well-discussed by sociologists. While there are different ways to define the middle class, such as income and occupation, Goodman suggests that the middle class would be best defined by its lifestyle consumption patterns, especially receiving college education and purchasing private houses. 39 Tang also points out that the pursuit of private homeownership makes the Chinese middle class fear the "insecurity caused by the dramatic social changes" and support regime stability. 40 Accordingly, Tang and Unger conclude that "the educated middle class, as a whole, become a bulwark of the current regime" and it "blocks the way" of democratic changes. 41 Therefore, the emphasis on the stabilizing effects of economic benefit enjoyed by the Chinese middle class suggests the importance of protecting property rights, because the quest for economic well-being may not be fulfilled without the protection of property rights.
The coercive capacity of state would also make citizens more cautious and less able to challenge the state. 42 The suppression of the Tiananmen Square protests of 1989 explicitly demonstrates the strong state coercive capacity to resist any challenges from society. 43 Recent studies further suggest that the CCP is well-prepared to censor the Internet in order to prevent collective action against the government. 44 As a result, the middle class would be pragmatically reluctant to challenge a regime with such a high degree of coercive power.
Although previous studies suggest that members of Chinese middle class accept the CCP's authoritarian rules, it does not mean that they have no preference for liberal democracy. Dowd, Carlson, and Shen report that the economic growth has instilled democratic values into the younger generation and more educated groups in China. 45 Meanwhile, Tang, Woods, and Zhao find that the Chinese middle class, compared to its lower counterpart, are more likely to favour democratic ideas, including democratization procedure, democratic affection, and civic value. 46 Since these studies suggest that members of the Chinese middle class have democratic values, it is puzzling why they also have high support for the current authoritarian regime.
Lu and Shi develop an eclectic perspective to explain why the middle class' democratic preference can coexist with the support for authoritarian regime. According to Lu and Shi, 47 the Chinese people see their government as a "guardian democracy" with an emphasis on the meritocracy articulated by the government propaganda. Thus, the Chinese citizens actually equate the support for democracy with the support for the regime, which is "democratic" under the definition of the CCP. However, the argument of guardian democracy does not explain why the Chinese middle class would prefer features of liberal democracy, such as representation and freedom of speech, especially when the empirical evidence suggests that some Chinese people are able to distinguish ideas of liberal democracy from those of guardian democracy. To answer this question, Miao argues that it is because Chinese citizens choose to prioritize social stability over ideas of liberal democracy. 48 Meanwhile, Wang claims that economic development will make China dominated by people with the belief that political rights and freedom of speech are more desirable than material benefits in the future. 49 In other words, economic growth may have fostered the preference to ideas of liberal democracy in China.
We follow the eclectic perspective, arguing that members of the Chinese middle class prefer ideas of liberal democracy, a political regime that can better their protect property rights. There are two reasons for the inconsistent findings of previous studies regarding the relationship between the middle class and values of liberal democracy. First, they use different measures to define the "middle class" in China. A common practice is to use income as a proxy of the middle class, but this approach is problematic due to the issue of "hidden income" in the Chinese economy, 50 which usually results in no response to survey questions on respondents' income. For example, the ABS3-China Project has about 20% of missing values on the question of household income, and this proportion of missing values is arguably large enough to threaten the validity of any inference about the political attitudes of the middle class.
Another source of this inconsistency among existing studies is the different meanings of democracy. As discussed before, the Chinese people arguably perceive their government as a guardian democracy with an emphasis on meritocracy. 51 Since democracy became the only legitimate game in town after the third wave of democratization, many authoritarian regimes revised the definition of democracy to reinforce their legitimacy. 52 The Chinese government has framed itself as an output-driven guardian democracy which looks after its people's well-being. 53 Indoctrinated by the idea of guardian democracy, the Chinese middle class may equate the support for democracy to the support for the existing regime. Nevertheless, guardian democracy and liberal democracy are not mutually exclusive, but the former downplays the importance of citizen's input in policymaking, a core feature of liberal democracy that can better protect property rights. Thus, the Chinese middle class may still have a stronger preference to liberal democracy than its lower-or upper-class counterparts.
In this article, we tackle both issues by using subjective measures to define middle class and features of liberal democracy as our analytical framework. While current studies demonstrate that the middle class in China tends to support the current regime as much as other classes do, it does not mean the middle class has a lower level of preference for democracy. Instead, the Chinese middle class's high level of support for the authoritarian regime may be a strategic response partly due to the fear of the state's coercive capacity. In addition, the Chinese government may gain regime support via gradually responding to the middle class' demand of protecting of property rights instead of fully democratizing the regime. Nevertheless, the Chinese middle class would still prefer defining characteristics of liberal democracy than do the lower and upper classes, because a liberal democratic would better protect property rights from state interference.
Based on these discussions, we empirically test the following hypothesis in the rest of this article: The middle class in China has a higher preference for liberal democracy than the lower and upper classes.

Data
To test our hypothesis, we use data collected by the third-wave Asian Barometer Survey (ABS3). The ABS3 includes national representative samples from 13 East Asian countries and territories collected during 2009 and 2011, including Cambodia, China, Hong Kong, Indonesia, Japan, South Korea, Malaysia, Mongolia, the Philippines, Singapore, Taiwan, Thailand, and Vietnam. 54 The ABS3-China Project collected data from 3473 successful face-to-face interviews in 25 provincial-level administrative divisions in China during July and November 2011. One major advantage of ABS3-China Project over other public opinion surveys conducted in the country is its wide coverage of China'S population and territories. The use of such a representative sample of Chinese citizens ensures the external validity of inferences based on this data.

Dependent variable
The dependent variable in our empirical analysis is the preference for liberal democracy. We rely on respondents' answers to questions in the "Regime Preference" section of the ABS3 to measure democratic preferences. The first part of this section consists of seven questions that ask respondents to choose one between two competing statements concerning how their political regime should work. 55 In each of these questions, one statement emphasizes a democratic characteristic and the other an authoritarian alternative. The characteristics of democratic regimes represented in the statements include representation (Q74 and Q75), media freedom (Q76), competitive elections (Q78), and multi-party competition (Q79). Table 1 reports the statements and the frequency with which they were chosen by the respondents. Respondents who chose the first statement over the authoritarian alternative reveal their preference for democracy.
As demonstrated in Table 1, most respondents show a preference for the democratic ideas of representation and competitive elections, but they do not prioritize media freedom (Q76) or multi-party competition (Q79). However, around 63.2% of respondents reveal a preference for competitive electionsa major defining criterion of democracy. 56 Nevertheless, there is a similar percentage of respondents, about 63.6%, who prefer one-party dominance. The contrast between the respondents' preference for competitive elections and their support for one-party dominance suggests that Chinese citizens still hold democratic values, despite living under an authoritarian regime.
Based on the respondents' answers to these five questions, we construct a new variable that measures respondents' preference for a democratic regime. Since the choices are binary, not continuous, we use principal component analysis based on the polychoric correlation of these questions. 57 The estimation results derived from the polychoric correlation matrix suggest that there is only one factor with eigenvalues larger than 1, a common criterion for retaining factors. We call this factor Preference for Democracy and normalize it from 0 to 100, with 0 and 100 indicative of the lowest and highest preference for democracy, respectively.

Key explanatory variable: subjective social status
We operationalize class using the measure of subjective social status reported by respondents in the ABS3-China Project, in which Question SE12 asks the following: People sometimes think of the social status of their families in terms of being high or low. Imagine a ladder with 10 steps. Those with the lowest status stand on step 1 and those with the highest status stand on step 10. Where would you place your family on that scale? While we assume that people who place themselves in the middle range of this scale are middle class, we do not group any values on this scale into the category of middle class. If middle-class respondents have a greater preference for democracy than either lower-or upper-class ones, we would expect to be able to depict the relationship between subjective social status and preference for democracy as inverted Ushaped. In our empirical estimation, we create a squared term of subjective social status to capture the nonlinear relationship between the two variables. On the basis of our hypothesis, we expect Subjective Social Status to be positive and its squared term to be negative.
One may argue that Subjective Social Status is an invalid indicator for measuring an individual's class status because class is usually defined by objective criteria, such as income or occupation. However, previous studies criticize the use of income to identify the Chinese middle class, because the amount of "hidden income" would not be reported by respondents. 58 There are also too many missing values in ABS3-China Project for us to be able to use the reported income to define someone as the middle class. Among the 3473 successful interviews, 1261 respondents (36.3%) declined to report their household income. Since there are no theoretical reasons or empirical evidence leading us to assume that these observations are missing at random, we are concerned that the missing data issue would bias our empirical analysis if we used this measure to define respondents' class. Nevertheless, it is reasonable to assume that people with higher income would be more likely to give themselves a higher social status, 59 so self-identification may be a proxy of income.
We are fully aware of other ways to circumvent the limitation of using income to define what the middle class is. For instance, Chen and other studies use occupation when defining the middle class in China. 60 However, the released data of the ABS3-China Project do not include the information regarding respondents' partisanship or employed sector, so we are unable to use the same strategy.
We would like to note that this article is not the first study to use subjective evaluation of social class in investigating Chinese citizens' democratic attitudes. Tang, Woods, and Zhao also use respondents' own perceptions of their social class to measure the size of the middle class in China, 61 and about two-thirds of their respondents identify themselves as middle class. What differentiates our measure from that of Tang, Woods, and Zhao is that our survey question about subjective social status does not use the term "middle class", 62 so we may to a large extent avoid the problem that membership of the middle class tends to be socially desirable, especially for those ruling or upper-middle classes with affluent lifestyles in China. 63 Figure 2 is a histogram of Subjective Social Status for the respondents. About 58% of respondents place themselves somewhere between 4 and 6 on a scale of 1 to 10, with only a few of them thinking that they belong to an extremely low or high class. The distribution of Subjective Social Class is in line with the conventional wisdom that the middle class in China is growing.
With the operationalization of preference for democracy and social class, Figure 3 illustrates the relationship between Subjective Social Status and Preference for Democracy. As shown in Figure 3, people who perceive themselves as the lower class have a lower preference for democracy. Respondents with a higher perceived social status are more likely to prefer democracy. Additionally, Figure 3 suggests that the relationship between social class and preference for democracy may be nonlinear, since those who perceive their class status to be in the middle range tend to have a higher preference for democracy than those with either a lower or a higher perceived status. In the following pages, we employ OLS regression models to further investigate this relationship.

Empirical estimation
To fully investigate the nonlinear relationship between social class and preference for democracy, we estimate a series of OLS regression models. 64 We use robust standard errors to address the issue of heteroskedasticity. Our explanatory variables include demographic factors, such as age, gender, type of residence, and education. According to previous studies, younger people are more likely to show a preference for democracy. 65 Respondents are more likely to be pro-democracy if they are male, live in urban area, or have been educated to a higher level than their fellows. 66 We also include respondents' evaluation of their family's economic situation in our empirical models to control for (dis)satisfaction with the current regime.
In addition, we consider other political variables that may influence respondents' political attitudes, including the perception of corruption and trust in local and national governments. Specifically, corruption is so pervasive in local government in China that respondents may prefer a political regime that can address this issue via democratic practices, such as media freedom or political competition. Similarly, a low level of trust in national or local governments may foster the preference for democratic alternatives. 67 We report the estimation results in Table 2. The results suggest that the Chinese middle class is more likely to express a preference for a democratic regime than other classes. In Model 1, we estimate a baseline model that excludes the squared term of Subjective Social Status. The positive and statistically significant coefficient of Subjective Social Status at the p < .01 level in Model 1 indicates that respondents' preference for democracy increases with their subjective social status. However, Model 2 includes a squared term Subjective Social Status and suggests that its coefficient is negative and statistically significant at the p < .05 level. The statistically significant signs of the coefficients of Subjective Social Status and its squared term indicate that the effect of perceived social status on preference for democracy is not linear but inverted U-shaped. In other words, the middle class in China tends to prefer democracy more than the lower or upper classes.
Our finding in Model 2 remains unchanged if we use a different operationalization of middle class. We divide the social class into three groups on the basis of respondents' subjective social status: low class (1-3), middle class (4-7), and upper class (8-10). We include a dummy variable for Middle Class in Model 3. Again, the positive and statistically significant coefficient of Middle Class at the p < .05 level implies that members of the middle class in China are more likely to express a preference for democracy than members of the other classes combined as the baseline category. This finding is consistent with the result of Tang, Woods, and Zhao. 68 In Model 4, we include an additional control variable: Extent of Democracy. This variable is constructed on the basis of respondents' answers to the following question: "In your opinion, how democratic is China? Is it (1) a full democracy, (2) a democracy with minor problems, (3) a democracy with major problems, or (4) not a democracy?" We use an ordinal variable to measure respondents' perceptions of the extent of democracy in China, with 1 indicative of "not a democracy" and 4 "full democracy". The inverted U-shaped relationship between preference for democracy and subjective social status remains the same in Model 4. Furthermore, respondents who believe that China already has a high degree of democracy tend to have a lower preference for the characteristics of liberal democracy. This result echoes the finding of Lu and Shi 69 regarding the different conceptions of democracy among Chinese citizens.
To illustrate our finding, we use the Stata package Clarify to calculate predicted values of preference for democracy across different levels of social status on the basis of Model 4. 70 The predicted values with their 90% confidence intervals are shown in Figure 4, where a non-linear relationship exists between preference for democracy and social status. In particular, preference for democracy increases as perceived social status increases from 1 to 7. However, if the respondent's social status is higher than 7, then preference for democracy is reduced. This result confirmed our hypothesis that the middle class in China is more supportive of democracy than the lower and upper classes.

Robustness checks
We conduct additional models as robustness checks and report the results in Table 3. First, we use reported household income level as the proxy of class level. The results of Model 5 suggest that preference for democracy increases with Income. However, we do not find a non-linear relationship between the two variables in Model 6 when we include a squared term of Income. This result may come from the missing values of Income. About 36% of the respondents refused to report their household income and, among those, 25% placed their social status at 5 and 41% placed it at values higher than 5. Thus, we suspect that respondents with high incomes are responsible for the missing values, and this type of respondent would have a lower preference for democracy. To test this conjecture, we use a dummy variable to indicate respondents who declined to report their household income. In Model 7, we find that this group of respondents favours democracy less than those who did report their household income. This finding indirectly supports our conjecture. 71 We also compare the explanatory power of Subjective Social Status and Income. In Model 8, we include Subjective Social Status, its squared term, and Income among the right-hand-side variables. Subjective Social Status, as indicated in Model 8, has a better explanatory power than Income, since the former and its squared term remain statistically significant but the latter does not. Therefore, subjective social status is a better indicator of class status than income in our sample.
One may criticize that our dependent variable is unreliable, because it is constructed on the basis of five questions that may not perfectly capture the characteristics of democratic regimes. Therefore, we estimate two additional models that use respondents' answers to the question concerning their support for democratic regimes. In particular, Q128 of ABS3-China asks respondents whether they agree or disagree with the following statement: "Democracy may have its problems, but it is still the best form of government". Obviously, respondents who agree with this statement have a stronger preference for democracy than those who do not. We code those agreeing with this statement as 1 and those who do not as 0. With this binary measure of preference for democracy, we run a probit model with the same independent variables used in Model 2. The signs of the coefficients for Subjective Social Status and its squared term in Model 7 are the same as those in the previous models. They are also statistically significant. Meanwhile, Income still has no statistically significant relationship with Preference for Democracy in Model 10. Therefore, preference for democracy is highest among those who place themselves in the middle range of the 1-10 scale of the subjective social status. The economic development has sowed the seeds of democratic ideas in the mindset of Chinese middle class.
Before concluding this section, we would like to note that the Chinese middle class' higher preference for ideas of liberal democracy does not imply more support of changing the current political regime immediately. As demonstrated by other studies, 72 the Chinese middle class is relatively conservative and acceptant to authoritarianism due to their dependence on the state and the threats of state coercive capacity. Accordingly, they are more reluctant to change the status quo than other classes. In Model 12, we empirically test this argument on the conservativeness of the Chinese middle class, with the support for regime change as the dependent variable. 73 The results suggest that the Chinese middle class is more reluctant to change the current political system than its lower and other counterparts, because the coefficients of Subjective Social Status and its squared term are negative and positive, respectively.
To further illustrate the contrast between the Chinese middle class' democratic values and support for the the current authoritarian regime, we plot the predicted probabilities for both political attitudes on the basis of Model 11 and Model 13. Respondents who self-identified with middle social status, as Figure 5 illustrates, have the highest support for democracy. Meanwhile, their support for regime change is the lowest. This gap between democratic values and regime support of the Chinese middle class confirms the eclectic perspective we have discussed in the previous section. The current Chinese political system may respond to the middle class' quest for economic well-being and protection of property rights in satisfactory ways, but it may face threats of democratic transitions if it fails to keep doing the good work.

Conclusion
After three decades of rapid economic development, the rise of a middle class in China has aroused expectations that this class will act as an agent of democratization. However, the middle class in China is conservative and supportive of the authoritarian regime. As a contribution to the debate over whether the Chinese middle class is likely to play that role, we provide micro-level evidence from the China section of the thirdwave Asian Barometer Survey to investigate the Chinese middle class's preference for liberal democracy. We find that the relationship between class identification and preference for features of liberal democracy among Chinese respondents is inverted Ushaped. Chinese citizens who identify themselves as the middle class express a stronger preference for liberal democracy than those in either a higher or lower class, and they also tend to regard democracy as the best form of government. Our findings shed light on the issue of authoritarian resilience in China. If middleclass citizens have a stronger preference for liberal democracy than citizens from other classes, they have a greater potential to become agents of democratization in the future as the Chinese government fails to address the discrepancies in their political values beyond the reach of the existing authoritarian regime. The Chinese middle class' preference for liberal democracy might be mitigated by the current government's efforts of creating economic growth and protecting property rights. However, the Chinese middle class may become an agent of democratization if the Chinese government fails to keep doing the good job.
Based on our findings, future studies may investigate other factors that mitigate the middle class' preference for democracy and authoritarian resilience in China. For instance, it may be the high cost of collective action or instability that prevents the Chinese middle class from demanding further democratization, 74 or it may be the case that the political elites rely on coercive capacity, including the military, to deter the middle class from challenging authoritarian rule. 75 Lastly, this study investigates the Chinese middle class' preference for the ideas of liberal democracy, but it is also important to explore how the political preferences transform into political behaviour. 76 Specifically, is the middle class with ideas of democratic values more likely to take any actions to initiate transitions to democracy? Or does its support for the current regime prevent it from challenging the state? As we argue that the newly enriched citizens would prefer liberal democracy to secure their property rights in China, future studies could further investigate how the CCP responds to this quest without democratizing the regime.