The Puzzle of Authoritarian Legitimacy
Data from the Asian Barometer Survey show that among fourteen countries surveyed in Waves III and IV (2010–2016), authoritarian regimes enjoy higher Diffuse Regime Support than do democratic regimes. The essay explores the impact on legitimacy of four causal factors: governmental economic and political performance, propaganda, nationalism, and culture. Traditional Social Values (TSV) tend to increase regime legitimacy and Liberal Democratic Values (LDV) to reduce it in both kinds of systems. Persons who adhere to LDV, however, are more likely to be alienated from authoritarian than from democratic regimes—and adherence to these values is increasing across the region as modernization advances.
Studies of Asians' attitudes toward democracy contain a puzzle. Since the Asian Barometer Survey (ABS) began in 2001, it has consistently found that the region's autocracies outscore its democracies in terms of "institutional trust."1 Measuring the public's average expressed level of trust in six governmental institutions across each of the four ABS waves from 2001 to 2016, we see, for instance, that people in China and Vietnam trusted their governments more than people in Japan and Taiwan trusted theirs.2 Many studies focused on China confirm the high level of public support for China's authoritarian regime.3
Why should authoritarian regimes enjoy more trust than democratic regimes, in which governments hold power by consent of the governed? Part of the answer may be that democracies include many "critical citizens"—people who are so strongly attached to democratic principles that they hold democratic institutions to stringent standards.4 The presence of high levels of institutional trust under authoritarian regimes, however, challenges the common assumption that such regimes are sustained only by repression, censorship, corruption, and an ample flow of public goods, with citizens' unforced preference for nondemocratic rule playing no role.
In hopes of making progress toward solving this puzzle, the ABS began measuring regime legitimacy separately from institutional trust with its Wave III survey in 2010. Although institutional trust is often used as a measure of regime legitimacy, the two concepts are not the same. One of the most influential attempts to conceptualize regime legitimacy was that of political scientist David Easton, who renamed the concept "diffuse regime support" and defined it as a "reservoir of favorable attitudes [End Page 158] or good will that helps members to accept or tolerate [regime] outputs to which they are opposed or the effects of which they see as damaging to their wants."5 Easton acknowledged that diffuse regime support would be difficult to measure, partly because the attitude is a complex one with at least two dimensions (which he labeled "trust" and "legitimacy") and partly because its referent, the regime, is difficult for respondents to distinguish from institutions, incumbents, and policies.
Borrowing Easton's concept, the Waves III and IV ABS surveys included a battery of questions designed to gauge the level of Diffuse Regime Support (DRS) expressed by respondents. The DRS questions sought to set aside people's attitudes toward particular governmental institutions, incumbents, and policies in favor of a focus on how respondents feel about their country's type of political system—the regime, in Easton's sense of the term. The battery contains four questions: Does the respondent think that "over the long run, our system of government is capable of solving the problems our country faces"; is the respondent "proud of our system of government"; does the respondent believe that "a system like ours, even if it runs into problems, deserves the people's support"; and would the respondent "rather live under our system of government than any other that I can think of." There are four available responses, ranging from "strongly agree" through "agree" and "disagree" to "strongly disagree." For this essay nonresponses, including "don't know" and "no answer," are coded as neutral.
In the combined Waves III and IV dataset, comprising surveys fielded in different countries at different times from 2010 to 2016, the new DRS measure confirms previous authors' findings that East and Southeast Asia's authoritarian regimes enjoy higher levels of diffuse support than the region's democratic regimes receive.6 The Figure below shows that, in general, the more authoritarian the regime, the stronger is its DRS. Among democratic citizenries, South Koreans appear to be especially dissatisfied with their country's political system, while among people living under authoritarianism, the Vietnamese seem to stand out as especially satisfied. Respondents in the People's Republic of China (leaving aside Hong Kong, which is measured separately) are more satisfied than not, but in comparative perspective are not especially regime-supportive.7
These findings are subject to reasonable doubts on two scores. First, the DRS questions are quite abstract. Can people be expected to offer considered responses to such queries, or will they just respond with "top of the head" answers that would be different were they asked the same questions on another occasion?8 There is no doubt that "top of the head" responses occur, but there are technical reasons, addressed in an online Appendix9 that I have created to supplement this essay, for thinking that the DRS battery is measuring something real. Moreover, as the analysis that follows will show, DRS measurements correlate in interpretable ways with measures of other variables. [End Page 159]
Click for larger view
View full resolution
Second, can people in undemocratic systems be expected to answer such sensitive questions honestly? Are they not likely to dissemble or retreat into the realm of the noncommittal in order to avoid trouble? Scholars who work with surveys taken in China have argued—to my mind persuasively—that political fear does not deter most Chinese respondents from giving honest answers to survey questions.10 Political intimidation in China is real, of course, but its more likely effect is not to cause people to lie about what they think, but rather to shape the thoughts they have.
For this reason, even when DRS is measured correctly at a given time, it may mean different things in different kinds of political systems. As the "critical citizens" literature implies, low DRS in a democracy may reflect citizens' routine irritation with the political system but not a preference for an alternative type of regime, while high DRS in an authoritarian system may reflect, for some people at least, acceptance of the regime out of the belief that there is no alternative, without a commitment to defend it.11 Even if DRS has different "flavors" (to borrow a term from particle physics) in different political systems, however, it is still a measurable attitude, and the challenge remains of explaining what factors contribute to high DRS in authoritarian systems.
There are four explanations commonly suggested for why some authoritarian regimes enjoy high levels of legitimacy: government economic and political performance, propaganda, nationalism, and [End Page 160] culture.12 The ABS data allow us to test these explanations, although not always as fully as we would like given the limits of space in the survey.
Testing the Explanations
The survey includes four separate measures of government performance. First, it asks respondents to rate the country's current economic condition, to state the overall trend they see over recent years, and to say how they think the national economy will be doing a few years hence. Second, the state of the respondent's "family economy" is probed via the same three questions: How is your family doing today, was it doing better or worse a few years ago, and how do you think it will be doing a few years from now? Next, opinions regarding the government's political performance are collected via questions about both the effectiveness and the fairness of governance. With regard to effectiveness, respondents are asked to rate how good a job the government is doing at fighting corruption, at administering rule of law, at being accountable to the public, at solving what the respondents see as the most important problem facing the country, and so on. Finally, seven other questions probe respondents' views of how fairly the government behaves: For example, does it treat rich and poor alike, safeguard the freedoms of speech and association, and guarantee access to basic necessities such as food, clothing, and shelter?
The impact of propaganda is gauged by two variables. "Media use" gets at how often respondents follow domestic and world news, and use the internet. "Media trust" scores respondents on how much they trust newspapers and television as news sources.
Nationalism is measured by a variable that I label "national pride." This is calculated as the respondent's average score on two questions: "How proud are you to be a citizen of [country]?" and "Given the chance, how willing would you be to go and live in another country?" (The two questions are of course scaled inversely, such that people with high scores on the first and low scores on the second are the most nationalistic.)
Finally, culture is measured with two batteries, one that assesses Traditional Social Values (TSV) and another that assesses Liberal Democratic Values (LDV).13 The nine TSV items were formulated on the basis of a wide range of literature which argues that conflict avoidance, deference to authority, and an inclination to favor the group over the individual are values widely held in traditional societies.14 The questionnaire items refer to general preferences and do not mention politics. The seven LDV items are designed to discern support for core liberal-democratic principles such as the freedoms of speech and association, judicial independence, and the separation of powers. These items present statements [End Page 161] that contravene liberal-democratic values, so that disagreement signals LDV adherence.
When other variables are controlled in a regression equation, the correlation between government performance and DRS is strong. Among the four performance variables, however, some aspects outweigh others in contributing to DRS.15 Contrary to conventional wisdom, respondents give no statistically measurable weight to the impact of government policies on their families' economic welfare, apparently crediting or blaming themselves rather than the government for however well or poorly their families are doing. What respondents do take into account on the economic front is the health of the economy as a whole.16 But in nine of the fourteen countries, including both democracies and autocracies,17 citizens give even more weight to government fairness than they do to economic performance: Even in authoritarian regimes, people care more about whether the government treats them evenhandedly than they care about economic affairs. And the variable of government effectiveness is yet more important, having the strongest effect on DRS among the four performance variables across ten countries.18
The authoritarian regimes of East and Southeast Asia tend to score higher on the two governance variables (effectiveness and fairness) than do their democratic neighbors, and this contributes to the robust DRS found in the authoritarian polities. To the extent, however, that authoritarian regimes are more susceptible than democracies to corruption and abuses of power, this finding points to a long-term threat to authoritarian legitimacy. Authoritarian regimes, like democracies, are being watched and evaluated by their citizens, who still expect—even if they cannot act on the expectation by freely voting—that the governments in their countries will act effectively and fairly.
As for propaganda, regression analysis reveals that the intensity of citizens' media use has almost no influence on DRS (there is a statistically significant relationship only in Cambodia and China, and it is small). The more influential variable is media trust, which exerts a statistically significant positive (but again small) influence on DRS in all but three of the countries. This probably reflects different dynamics in democracies and authoritarian systems. In the democracies, where media are diverse and often carry criticisms of the government, trust in media may help citizens to feel that they understand better why the government does what it does. In authoritarian systems, where the media are government-run or at least heavily government-influenced, the influence of media trust signifies that citizens who believe government propaganda are more likely to offer diffuse support to the regime.
Nationalism, or national pride, has a substantial positive influence on DRS in all countries except for Cambodia and the Philippines. In both democratic and authoritarian systems, citizens who express pride [End Page 162] in their country and say that they are unwilling to move elsewhere are more likely than others to express support for the regime.
The final possible explanation for higher DRS under authoritarianism is culture. Both TSV and LDV have strong impacts on DRS in most countries, but their effects are opposed. In all countries except Burma and Taiwan, regardless of regime type, persons who affirm traditional values tend to accord greater legitimacy to the regime under which they live. The attitudes that are expressed through the TSV question battery, such as acceptance of hierarchy and deference to authority, incline citizens who hold these attitudes to support the existing political system regardless of what type it is, how it performs, how effective its propaganda is, or how proud these citizens feel of the nation. The LDV impact is the opposite. In all countries except Thailand, there is a statistically significant relationship between affirming liberal-democratic values and being more critical of one's government than fellow citizens who do not score as highly on LDV measures. This is true whether the regime is democratic or authoritarian, and holding all other variables constant.
But the impact of LDV is not uniform across regime types. On average, a given quantum of LDV among the populace has a stronger downward pull on DRS in authoritarian regimes than in democracies. To be sure, the LDV attitude cluster is more widely distributed in democratic as opposed to authoritarian countries. For example, the average per-item agreement with the seven LDV items is 63.4 percent of all respondents in Japan compared to 24.2 percent in Burma and 34.4 percent in China. But because persons who adhere to LDV are more likely to be alienated from the government if they live under an authoritarian regime than if they live in a democratic system, the spread of liberal-democratic values presents more of a threat to authoritarian regimes than it does to democracies.
Who Supports the Regime?
So what predicts adherence to liberal-democratic values? In all countries except the Philippines and Vietnam there is a significant positive correlation between LDV and level of education. Urban residence correlates positively with LDV in eight of the twelve countries where the urban variable was measured (the ABS considers all Hong Kong and Singapore residents as urban, so there is no urban variable in our data from those societies). In eight of the societies, male gender is associated with LDV adherence. In nine of the societies, youth is so associated. And everywhere except South Korea and Vietnam, the more integrated a person is into social networks and organizations, the more likely he or she is to be an LDV adherent. (This latter variable, labeled "social capital," is a scale that includes how many people the subject has contact with during an average week, how many people the subject thinks [End Page 163] would help the subject in case of need, and how many people the subject would help.)
In authoritarian and democratic systems alike, the social constituency that provides the most DRS to the regime is the opposite of the LDV constituency. Wherever the correlations are statistically significant, regime supporters are less educated, less urban, and older. In most countries, they also possess less social capital. (The relationship of gender to DRS is statistically significant in only five societies, and its effects are mixed.) The persistence of TSV is a key reason that authoritarian regimes in Burma, Cambodia, China, and Vietnam still enjoy wide popular support even amid social change so rapid and disorienting that it might be expected to breed instability. An average of 64.1 percent of Chinese respondents, for example, expressed agreement with the nine items that make up the TSV scale, compared with 46.8 percent in Japan.
The incidence of TSV tends to decline, however, under the influence of the same social forces that drive rising LDV. Already, LDV shows surprising strength in some of East and Southeast Asia's authoritarian systems. An average of 34.4 percent of Chinese respondents, for instance, took the liberal position on the seven items in the LDV scale. This includes 46.9 percent who disagreed with the proposition, "If we have political leaders who are morally upright, we can let them decide everything." In Burma, that figure was 24.2 percent, and in Vietnam it was 20.7 percent. Although these numbers are much lower than the numbers for Japan, South Korea, and Taiwan (63.4 percent, 57.7 percent, and 57.3 percent, respectively), they suggest that even in the region's most authoritarian countries, LDV adherence is already substantial.
The big-picture implications are consistent. Education and urbanization are proceeding—albeit at varying rates—across all East and Southeast Asian countries. As modernization and generational succession proceed, LDV adherence should spread.19 And that is bad news for authoritarian regimes. Of course, authoritarian regimes are not without resources to retain citizen support even as values evolve. As we have seen, strong policy performance can do a lot to generate DRS. Moreover, even a marked decline in DRS among the rising middle class in authoritarian systems may not lead to near-term defections from the regime. China's new middle class, for example, is too tenuous, insecure, and dependent on the regime to withdraw support just yet.20
The Authoritarian Dilemma
Political developments in East and Southeast Asia's authoritarian regimes will be the product of many factors: How well will the elites manage their factional squabbles? How will ethnic tensions be handled? [End Page 164] Will international conflicts, natural disasters, or health crises crucially shape events? But regime legitimacy—diffuse regime support—is likely to play a role in the futures of both democratic and authoritarian regimes. It will do so directly, by influencing civil order, and indirectly, through its effect on elite coherence and the loyalty of security agencies and militaries.
In this long-term perspective, the ABS data contain more troubling signs for authoritarian than for democratic regimes. As noted, the importance of government fairness and effectiveness in generating regime support points to a way in which authoritarian regimes are more vulnerable than democratic regimes. This is not because the performance of authoritarian regimes on these measures is necessarily inferior to that of democratic systems, but because the performance flaws of democratic systems are baked into their DRS scores by virtue of the open character of their politics: Dissatisfaction is readily generated and continues as a chronic condition, but it reaches crisis levels relatively seldom. By contrast, performance flaws in authoritarian systems are hidden until an economic slowdown, an environmental or public-health emergency, an international humiliation, or a corruption scandal comes along and makes them glaringly apparent.21 This can cause authoritarian legitimacy to decline suddenly and drastically.
Without the positive effects of regime performance buttressed by propaganda, the regime must fall back on its citizens' values to generate legitimacy. But authoritarian regimes' development policies are causing values to change. In order to benefit from the legitimacy-enhancing effects of regime performance, as well as for purposes of national security, authoritarian regimes pursue economic growth and social modernization. The more they succeed in their modernizing mission, the more they undermine the prevalence of TSV among their populations.22 The more their populations become urban and educated, the more they believe in freedom of speech and limits on government authority.
People who adhere to LDV tend to express less regime support in both democratic and authoritarian systems. But the implications of this finding are different for the two types of regime. Only in authoritarian systems, not in democracies, is high LDV adherence associated with a desire for an alternative kind of regime. This is shown by two measures. The ABS assesses Detachment from Authoritarianism (DA) by scoring whether respondents disapprove of three forms of authoritarian rule: "We should get rid of parliament and elections and have a strong leader [End Page 165] decide things," "Only one political party should be allowed to stand for election and hold office," and "The army should come in to govern the country."23 Under all types of regimes, LDV is positively correlated with DA: That is, liberal citizens in both authoritarian and democratic regimes are unattracted to authoritarian alternatives. In effect, neither group sees an authoritarian alternative to what they already have.
But LDV adherents living in authoritarian systems do see an alternative. The ABS measures Democratic Regime Preference by asking respondents four questions about the attributes they prefer in government: for example, does the respondent believe that "Government is our employee, the people should tell government what needs to be done," or does she believe that "The government is like a parent, it should decide what is good for us"? In all types of political systems, adherents to LDV display a preference for regime characteristics associated with liberal democracy. This is not surprising in itself, as the LDV and regime characteristics batteries are quite similar. But the implications are encouraging for the stability of democratic systems and discouraging for that of authoritarian systems. In democracies, citizens may be dissatisfied with what they have but do not prefer an alternative. In authoritarian regimes, as LDV spread, so does a preference for democratic regime characteristics.
Authoritarian regimes may try to slow the erosion of deferential values, as China has done with its campaigns to revive Confucianism and promote nationalism. These efforts promote a kind of double-mindedness, with younger and more educated citizens feeling proud of their country's traditions and accomplishments and keen to contribute to their country's success, yet determined to assert their individuality in their private lives, to protect their personal and property rights, and to learn about the outside world. Meanwhile, the shift away from TSV and toward LDV continues—and the better the regime performs, the more rapidly the shift occurs. The high trust and DRS scores enjoyed by authoritarian regimes in East and Southeast Asia will not protect them indefinitely from challenges to their survival. [End Page 166]
Andrew J. Nathan is Class of 1919 Professor of Political Science at Columbia University and a member of the steering committee of the Asian Barometer Survey. His books include China's Search for Security (with Andrew Scobell, 2012).
1. The ABS conducts surveys in East and Southeast Asian political systems (all of which this essay calls "countries" for convenience, although Hong Kong is part of China and the status of Taiwan is contested). The ABS's Wave I (2001–2003) surveyed eight countries. Waves II and III (2005–2008 and 2010–12) covered thirteen, and Wave IV (2014–16) covered fourteen. Wave V is now in progress. The ABS is headquartered in Taipei, at the Hu Fu Center for East Asia Democratic Studies of National Taiwan University. (The ABS should not be confused with the AsiaBarometer, which operates from Japan and covers Central and South Asia as well as East and Southeast Asia.) In each country where the ABS is conducted, a local team carries out the survey, in most cases with local funding, using a common core questionnaire. The survey focuses on attitudes toward democracy. In order to allow the tracking of changes over time, the questionnaire remains the same from wave to wave, aside from minor adjustments and added questions. For more, see www.asianbarometer.org.
2. For details of this finding and of other findings reported in the article, as well as technical information on statistical procedures, please see the online Appendix at www.journalofdemocracy.org/articles/supplements. Full details on how I coded the variables used in this article (using Stata software) are available upon request to email@example.com.
3. See, for example, Tianjian Shi, The Cultural Logic of Politics in Mainland China and Taiwan (New York: Cambridge University Press, 2015), ch. 5; Bruce J. Dickson, The Dictator's Dilemma: The Chinese Communist Party's Strategy for Survival (New York: Oxford University Press, 2016), ch. 5; Wenfang Tang, Populist Authoritarianism: Chinese Political Culture and Regime Sustainability (New York: Oxford University Press, 2016); and Jie Chen, Popular Political Support in Urban China (Washington, D.C.: Woodrow Wilson Center Press and Stanford: Stanford University Press, 2004).
4. Pippa Norris, ed., Critical Citizens: Global Support for Democratic Government (Oxford: Oxford University Press, 1999).
5. Quoted in David Easton, "A Re-Assessment of the Concept of Political Support," British Journal of Political Science 5 (October 1975): 444.
6. I combined the data from the ABS Wave III and Wave IV to create larger country samples capable of dampening the random measurement errors always found in survey research. Combining the two waves is acceptable for the present investigation because, even if DRS levels shift, their determinants are unlikely to have changed much in the handful of years between these two waves. The combined dataset has 40,103 subjects.
7. Countries are arranged from left to right according to their 2013 Freedom House scores. DRS is a factor score ranging from -2.89 to +1.92. Factor scoring is a way to combine several items into a single variable in a way that gives more weight to their shared property than to other properties that may affect each component item. Factor scores are harder to interpret than mean scores. A score below zero in this Figure does not necessarily indicate a legitimacy crisis. It is best to interpret the scores as simply comparative; a higher score indicates stronger average DRS than a lower score. For more details, see the Appendix.
8. John R. Zaller, The Nature and Origins of Mass Opinion (New York: Cambridge University Press, 1992), 36.
9. See the Appendix as listed in note 2 above.
10. See Shi, Cultural Logic, ch. 3 and Appendix B; and Xuchuan Lei and Jie Lu, "Revisiting Political Wariness in China's Public Opinion Surveys: Experimental Evidence on Responses to Politically Sensitive Questions," Journal of Contemporary China 26, no. 104 (2017): 213–32.
11. This is what Jon Elster refers to as "adaptive preference formation" in Sour Grapes: Studies in the Subversion of Rationality (Cambridge: Cambridge University Press, 1983), ch. 3.
12. See Bruce Gilley, The Right to Rule: How States Win and Lose Legitimacy (New York: Columbia University Press, 2009), ch. 2; Yang Zhong and Yongguo Chen, "Regime Support in Urban China," Asian Survey 53 (March–April 2013): 369–92; Timothy Kaying Wong, Po-san Wan, and Hsin-Huang Michael Hsiao, "The Bases of Political Trust in Six Asian Societies: Institutional and Cultural Explanations Compared," International Political Science Review 32 (June 2011): 263–81.
13. For a discussion of the persistence over time and distribution across social groups of these two sets of cultural attitudes in China, see Andrew J. Nathan, "From the Classics to Today: How Much Have Values Changed?" Journal of East-West Thought 8 (December 2018): 17–32.
14. For an example of this literature, see Alex Inkeles and David H. Smith, Becoming Modern: Individual Change in Six Developing Countries (Cambridge: Harvard University Press, 1974).
15. The impacts of different variables are compared based on standardized regression coefficients; see the Appendix.
16. For a similar finding about opinion in the United States, see Donald R. Kinder and D. Roderick Kiewiet, "Sociotropic Politics: The American Case," British Journal of Political Science 11 (April 1981): 129–61.
17. The countries where fairness in governance affects DRS more strongly than the country's perceived economic performance are Burma, Cambodia, China, Hong Kong, Japan, Malaysia, South Korea, Taiwan, and Thailand.
18. The ten are Burma, Cambodia, China, Indonesia, Japan, the Philippines, Singapore, South Korea, Taiwan, and Vietnam.
19. For recent restatements of this classic claim of modernization theory, see Ronald Inglehart and Christian Welzel, "How Development Leads to Democracy: What We Know About Modernization," Foreign Affairs 88 (March–April 2009): 33–48 and Christian Welzel, Freedom Rising: Human Empowerment and the Quest for Emancipation (New York: Cambridge University Press, 2013). There is of course a chance that people could shift away from LDV and toward TSV as they age, but given the deep social changes now underway throughout the region, it seems a fair guess that such shifts will not happen with enough force and frequency to be decisive at the societal level. We cannot test this intuition, however, without either panel data or a series of surveys stretching over a longer period.
20. Andrew J. Nathan, "The Puzzle of the Chinese Middle Class," Journal of Democracy 27 (April 2016): 5–19.
21. Benedict J. Tria Kerkvliet suggests that the Vietnamese communist regime enjoys "considerable legitimacy" in part because Vietnamese authorities allow citizens openly to voice complaints and criticisms to officials. See Kerkvliet, Speaking Out in Vietnam: Public Political Criticism in a Communist Party–Ruled Nation (Ithaca: Cornell University Press, 2019), 147.
22. This is consistent with the analysis of Zhengxu Wang, "Before the Emergence of Critical Citizens: Economic Development and Political Trust in China," International Review of Sociology 15 (March 2005): 155–71.
23. Only two of the four questions were asked in China in Wave III; Burma was not surveyed in Wave III; the battery was not used in Vietnam in Wave IV.