Abstract
How have local governments in China been able to break through central policy restrictions in a unitary and authoritarian political system? Why is China’s official discourse in the reform era often so conservative and unfavorable to reform? The author argues the two issues are components of a signaling game between China’s central government and local officials, in which local officials know that the center may be reformist, but the reformist center imitates the rhetoric of a conservative center to control the pace of local liberalization. The result is a gradualist reform of “signaling left, turning right,” with glaring incongruity of speech and actions in the process.
How do countries in transition undertake socioeconomic liberalization? The two dominant schools of thought, the “J-curve” and the “partial reform equilibrium” theories, focus on postcommunist countries that simultaneously undergo political and economic transitions, and they argue for either insulating the government from the electoral pressure of citizens who bear the cost of initial reforms (Przeworski 1991) or opening the government to voter pressure so as to limit the influence of narrow interest groups that try to block further reform after gaining from early, incomplete liberalization (Hellman 1998). For countries that undertake economic reform while remaining politically authoritarian, a number of scholars have stressed the role of subnational governments in promoting reform and liberalization. Local officials can serve as coalition partners of reformist central leaders (Malesky 2009; Shirk 1993) or initiate policy breakthroughs on their own because of their institutionally guaranteed autonomy (Montinola, Qian, and Weingast 1995), their strong financial position induced by foreign investment (Malesky 2008), or the weakness of the central state (Jones-Luong 2003).
Local governments have also been instrumental in China’s economic and social transformation in the past three decades. Although some reform measures in China were directly championed or encouraged by (some) central leaders, “almost every major step on the path of reform was tried out by a few regions first before being launched nationwide” (Xu 2011, 1107). Many of these local reforms were initiated or protected by subnational governments without central leaders’ involvement or authorization. The central government may endorse such reforms ex post or even popularize them to the entire country, but they often contradicted the central government’s regulations when they were first put forward. Agricultural decollectivization and state enterprise privatization are prominent examples of this recurring phenomenon. But there are important differences between what has enabled local reform in China and the explanations offered in the comparative literature. As will be discussed in more detail later, China’s central state has remained powerful vis-à-vis local governments in the reform era, and local governments do not enjoy institutionally guaranteed freedom from the central state. Still, in many instances local officials unilaterally initiated reform, without being coalition partners of central leaders. How have local governments been able to break through policy restrictions of the central government in a unitary and authoritarian political system? And why does the central government allow local breaches to occur?
Another significant feature of Chinese reform is that while the Chinese central state has accepted many unauthorized local reforms in the past three decades, the official discourse about the country’s political and economic developments, whether in leaders’ speeches or newspaper editorials, is often full of conservative ideological rhetoric that not only is unfavorable to reform but sometimes sharply contradicts reality. Why would the Chinese central state permit (local) reforms on one hand and transmit conservative signals on the other?
I will discuss examples of local reform in the empirical section. Here it is worthwhile to highlight some contradictions between Chinese political rhetoric and the reality of reform. By 2005 the number of state enterprises in China had fallen to about 6 percent of all enterprises, while those explicitly registered as private enterprises had increased to 61 % the state sector’s share in industrial output had fallen to 15.2 percent, and its nonagricultural employment had declined to 13.1 percent (State Council Economic Census Group 2005; National Bureau of Statistics 2006). 1 Politically, the Chinese Communist Party (CCP) has tried to co-opt a broad range of social groups, and many private entrepreneurs are now CCP members (Chen and Dickson 2008). Despite this scale of capitalistic transformation, President Hu Jintao reiterated as usual in a December 2008 speech marking the thirtieth anniversary of the start of the reform that China will “unwaveringly uphold the Four Cardinal Principles” (i.e., the socialist path, people’s democratic dictatorship, leadership of the CCP, and Marxism–Leninism–Mao Zedong thought), and “will never embark on the evil road of changing the flag, but will unswervingly adhere to the path of socialism with Chinese characteristics.” 2 In the entire speech Hu used the term socialism close to 160 times, socialism with Chinese characteristics close to 60 times, and Marx over 30 times. Why does the Chinese government stick to orthodox socialist rhetoric when, after three decades of reform, there is a near social consensus that the reform has no turning back?
This article shows that the two seemingly distinct puzzles are components of a game between China’s central leadership and local officials in the reform era; and I use a simple heuristic model to analyze the central–local interactions. In a nutshell, the argument goes as follows. The existence of a reformist faction in China’s central leadership (or the possibility of the center being overall reformist) has enabled local officials to carry out reforms. The reformist central leadership, however, does not want the reform to go too fast, either to ensure an orderly and gradual social transition or to protect themselves from potential conservative attacks or undesirable outcomes of risky reforms. So it often imitates the rhetoric of a conservative center, both in abstract, general discourses and sometimes in specific policy articulations. The use of similar rhetoric by different types of central leadership makes it difficult for local governments to know exactly if the center is reformist or conservative, constrains the pace and scope of policy liberalizations that local officials can undertake at a time, and often forces them to shield their reforms from publicity or cover them with labels that conform to existing central decrees. Central rhetoric is not always pooled since sometimes the reformist center feels compelled to separate its proclamations from those of the conservative center, but the pooling equilibrium has characterized a considerable part of Chinese reform.
Such a mode of reform involves the concealing of goals and a confusing juxtaposition of rhetoric and actions and can be regarded as a kind of reform by “muddling through” (Lindblom 1959; Huntington 1968). Reforming without an explicit overarching design from the central leadership, but with various local actors pursuing liberalization measures one at a time, often in a guarded and incremental way, is not just necessitated by the searching of optimal policies but is taken when policy makers in China operate in a very politicized and uncertain environment, and the potential impacts of a liberalizing policy can be huge (Chung 2000; Shih 2007a).
I call this approach of reform “signal left, turn right” because of the glaring incongruity of speech and actions during the process. 3 But the metaphor should not be taken to mean that a single driver is performing the two actions. Instead, both the central and local governments are instrumental in China’s reform, with local governments carrying out many actual reforms and the center selecting its signals to manage the pace of reform, as if they are engaged in a tacit collusion in making a gradualist reform. This style of reform may appear bewildering, and the task of this article is to explicate its internal political logic.
The recent and growing literature on authoritarianism has focused on the role of political institutions such as elections, legislatures, parties, and the bureaucracy as tools of buying support, sharing rent and power, and ultimately political control and survival (e.g., Blaydes 2011; Brownlee 2007; Bueno de Mesquita et al. 2003; Geddes 1999; Lust-Okar 2006; Magaloni 2006; Slater 2003). A small formal literature has similarly examined the questions of how authoritarian regimes elicit cooperation among the elites or manage co-optation of the opposition (Dal Bo and Powell 2009; Gandhi 2008; Myerson 2008; Svolik 2009), sometimes developing institutions to do so. The distinguishing feature of this article is twofold: first, it examines a noninstitutional aspect of authoritarian politics (rhetorical activities) and how it relates to changes of policies and institutions (reform); second, the emphasis of the article is not on how a regime maintains its survival but how it engages in economic and social transition while ensuring that the regime remains stable.
In the following section I discuss existing explanations of local reform and political rhetoric in China and why they do not fully account for the above-mentioned puzzles. Then I present the central–local reform game. Afterward I discuss some important examples in Chinese reform as evidence consistent with my theory. The last section concludes with a brief discussion of the nature of China’s gradualist reform.
Existing Literature
This article is related to two distinct streams of literature on Chinese politics. The first deals with how reforms are conducted in China, especially at the local level. An influential argument about China’s reform is the “market preserving federalism” theory (Weingast 1995; Montinola, Qian, and Weingast 1995), according to which local reforms can emerge because China is a de facto, although not de jure, federal state, and so local governments enjoy guaranteed autonomy over their own activities: “Experimentation, learning, and adaptation all follow from the inception of local political freedom over the economy” (Montinola, Qian, and Weingast 1995, 78).
The “market preserving federalism” argument is insightful about local motivations for reform, but it also has some major drawbacks. According to Weingast (1995), a crucial condition of market preserving federalism is that the authorities allocated to subnational governments must be institutionalized or at least durable, and when the central state transgresses against one local constituency, there is a coordination device such as a political constitution with stipulations on the rights of local constituencies that leads them to cooperate and challenge the central state together. There are no such democratic coordination devices in China’s one-party state. In addition, China’s decentralization is neither institutionalized nor firmly durable. China has indeed devolved considerable fiscal and administrative powers to local governments, particularly in the 1980s, but the central government has never released control of the hierarchy of authority or the careers of local officials (Solnick 1996; Xu 2011), and it has always retained the power to change central–provincial relations as it sees fit, including utilizing personnel management power to control excessive local investment and fending off centrifugal challenges rising from trade openness (Huang 1996; Sheng 2007), and reconfiguring the sharing of tax revenues and rebate responsibilities between itself and local governments to its own favor even after the 1994 fiscal reform sought to stabilize central–provincial fiscal relations (Yang 2006). In fact, China’s fiscal decentralization might have worked only because it comes with political centralization (Blanchard and Shleifer 2001).
Related to the federalism/decentralization argument is the cross-regional competition thesis. Maskin, Qian, and Xu (2000) and particularly Xu (2011) argue that tournament or yardstick competitions between subnational officials for climbing up the career ladder propel them to pursue growth-enhancing policies. Although the performance-based promotion system indeed provides a powerful incentive for promoting local development, it does not explain why local officials want to embrace risky and career-threatening breaches of central policies. As shown in the case studies below, reforming local leaders are often more concerned about potential punishment than promotion. Yang (1997) and S. Li, Li, and Zhang (2000) hold that when regions compete for economic development or market shares, local governments are forced to liberalize prices to avoid resources flowing to other regions that have liberalized or to privatize state enterprises to boost managerial incentives. While local competition certainly matters for reform, these accounts ignore the center’s political and ideological concerns and simply assume that local reforms will always be accepted by the center. But the center has political and ideological concerns, and whether it will accommodate local initiatives depends on the type of center and the degree of the local breaches. Yang’s (1997) and S. Li et al.’s arguments, therefore, provide a local economic logic but not a political logic for reform.
Some other studies employ the “fragmented authoritarianism” framework of Lieberthal and Oksenberg (1988) that stresses bargaining and persuasion in the Chinese policy-making process in explaining China’s local reforms. Fewsmith (1994), Yang (1996), and Zweig (1997), for example, contain vivid descriptions of interactions between central and local officials over agricultural reforms. Such interactions, however, often occurred after local governments had already initiated the new policies. Similarly, accounts about local lobbying (Zweig 2002) or policy disputes within the central leadership and associated scholars (Baum 1994; Fewsmith 1994) cannot explain how some of the reform measures started without explicit involvement of central officials.
It is widely known that the Chinese government sometimes experiments with new policies in a few localities before promoting them nationally, and China’s multiregional governance form (M-form) may be particularly suitable for small-scale and uncertain experimentations (Qian, Roland, and Xu 2006). Such centrally approved experiments include the policy to give state enterprises enhanced operational autonomy and the establishment of special economic zones in the early years of the reform (Xu 2011). But many reform policies such as agricultural decollectivization and state enterprise privatization emerged without first being approved by the center as experiments. Of course, once reforms by local governments have been instituted and are successful, they may be presented as experiments. But this does not mean that these reforms were introduced by the center as experiments or that they had explicit supporters in the center all along. To treat both purely local initiatives and policy trials sponsored by the center as experiments fails to distinguish their distinct political logics and risks.
Work on power competition within the central leadership is also relevant. Shirk (1993) argues that China’s top leaders have a “reciprocal accountability” relationship with their Central Committee selectorate, many of whom are provincial cadres. Thus, policy makers competing for top central positions must play to the selectorate by promoting reform policies or particularistic contracting with provinces. Shirk’s theory, however, does not deal directly with locally initiated reforms, the main concern here. Cai and Treisman (2006), on the other hand, argue that rival factions in the center sought to win competition by appointing their supporters to certain provinces to initiate and demonstrate the effectiveness of new policies. While explicit backing of central leaders will certainly embolden local governments and is compatible with the separating equilibrium of my model, the theory advanced here does not rely on it. More importantly, in many reforms local officials do not have obvious patrons at the center; they in fact shield their reforms with much secrecy, apparently not trying to impress any central bosses. Furthermore, under China’s “one level down” cadre management system, municipal or other lower-level officials who are often at the reform front are not directly appointed by the central authority (Landry 2008), although they are ultimately accountable to the latter.
The second and smaller literature related to the article is about the persistence and prevalence of conservative and outdated discourse in Chinese politics during the reform era, which has been described as “doing things with words” (Schoenhals 1992) and a “language game” (Link 1992). One obvious potential explanation is the government’s continued need for propaganda (A.-M. Brady 2007; Lynch 1999). But citizens in authoritarian countries often know that their governments are propagandizing and discount the messages (Geddes and Zaller 1989; Mickiewicz 2006). Since the official socialist discourse has been rendered largely obsolete in social life and has been widely received with cynicism and mockery (Link 1992), it is difficult to argue that outdated dogmatic rhetoric is used only or even mainly for ideological indoctrination.
Another argument holds that since the CCP has justified its rule on a communist revolutionary narrative for decades, divesting itself of socialist ideology and rhetoric altogether would amount to rejecting the party’s raison d’être (Creek 2006; Kluver 1996; Shambaugh 2008). Pressures from the more traditional segment of the party and the need to maintain continuity of policies and unity of the leadership in appearance have augmented this consideration (Ding 1994; Misra 1998; Solinger 1993, chap. 2).
This explanation no doubt has important and valid points. But if legitimacy of the CCP rule is the only concern, the party can just pay some lip service to socialist slogans since its current legitimacy is largely dependent on delivering economic growth rather than serving as a revolutionary vanguard, and with the passing of time the historical and ideological burden will fade away. China, however, devotes considerable resources on a daily basis to the production of official discourses and various ideological campaigns (Holbig 2008; Shambaugh 2008), far more proactive than a mere defensive lip service requires. This demands an explanation beyond legitimacy consideration.
The Game of Rhetoric and Reform
In this section I first explain in some detail the setup of the game, which incorporates several stylized features of Chinese politics relevant to the article’s research questions, and then present the results.
Setup
There are two players: the central government (center) and the local government. The two-level government is an abstraction of the multilayer system in reality; no matter which specific level of subnational government initiates a reform in China, ultimately the reform will be subject to the center’s verdict. Furthermore, since China’s central government strongly inhibits lateral communication and coordination between local governments (Pye 1992; Zhu and Huang 2002), the local government can be treated as a singular player rather than being plural in number in this game with the center.
The center can be either reformist or conservative, and its type is private information. A natural interpretation of the center’s type is that it has two factions—the reformist faction and the conservative faction—but the local government does not know with certainty which faction is dominant at a particular time or over a particular policy issue. Depending on which faction is in charge, the center is reformist or conservative in type. Factions in Chinese elite politics may be only loosely structured and fluid in membership over time and across policy issues (e.g., with some leaders being liberal in economics but hard-line in politics), but their existence as well as the continuous conflicts and constant changes of relative power between them have been well documented (e.g., Baum 1994; Fewsmith 1994; Shih 2007b). Even if the center is unified, it can sometimes have reformist tendencies and at other times conservative leanings. The local government does not know perfectly the center’s current attitudes toward certain specific policy issues. Therefore, what this specification captures is the uncertainty that a reforming local government has about whether its intended liberalization of a particular policy at a particular time will be accepted by the center.
There are three levels of reform that can be taken, high (H), medium (M), and low (L). High reform refers to open and uncompromised departures from existing policies. Low reform means making minimal changes or doing nothing. The medium level is something in between, that is, moderate or concealed reform. As explained below, different players have different preference orderings about the levels of reform. One’s most preferred level of reform gives the player a benefit of v, the second preferred level yields a benefit of b, and the least preferred level results in 0, with v > b > 0.
The conservative center would naturally prefer the reform to be as low as possible, either because the conservatives have a more orthodox socialist ideology or because they have significant vested interests in the old system. Hence, this center’s preference ordering is L > M > H. The reformist center wants reform but would like the pace of reform to be controlled, either because the reformers are fearful of attacks from the conservatives, 4 or because they have a concern for maintaining social order and fear of regime instability if the reform, the outcome of which is uncertain, goes too fast. Therefore, this center prefers M the most. Between H and L it prefers H. This is because, as intelligent political players, reformist leaders know that moderate reforms will accumulate into a large transformation, so from their preferring M to L it can be inferred that they also prefer H to L. This is essentially the same assumption as that in the pioneering study of postcommunist democratization and reform by Przeworski (1991, 62-64), where he argues that liberalizers will adopt broadened dictatorship if they prefer democratic transition to status quo dictatorship.
The local government has yet different preferences. While China is a large country and local officials can differ from each other considerably in ideological persuasions and benefits from specific reforms, in this article I model the emergence of local reforms rather than the diversity of local behaviors, and so focus attention on a very reform-minded local government; that is, its preference ordering is H > M > L. Reform- and development-minded officials abound in China given the significant changes in the incentive structure of Chinese local bureaucracy in the reform era, including the sharing of fiscal revenues and the need to self-finance local expenditures (L. Zhang 1999; Jin, Qian, and Weingast 2005), as well as opportunities for family members to participate in business activities resulting from local development (D. Li 1998). Even though they are not elected, local officials can also have community interests in improving the material, cultural, and social life of citizens in their regions. To the extent that they may also have concerns for national stability, each local leader has an incentive to free ride on other regions’ restraints in reform in ensuring that China as a whole undergoes gradual change. Therefore, local officials can often be more liberal minded than central leaders. It is important to note, however, that the theory here does not rely on all local governments being very reformist. As long as some local governments are willing to initiate reforms, and such reforms will then spread to other regions if unpunished, the center needs to strategically decide how to interact with the local governments.
Since the article models local reform, the level of reform is chosen by the local government. The center, on the other hand, can engage in rhetorical proclamations about the country’s general directions and/or specific policy issues. Such rhetoric conveys signals about the center’s type, although the two types may choose the same proclamation, in which case their signals are “pooled” and the local government cannot distinguish one type from the other. Although central rhetoric can come in various shapes, in essence it boils down to two kinds: pro-reform rhetoric and conservative, pro–status quo rhetoric. Touting the urgency of reform, as in Deng Xiaoping’s famous slogan “be a little bolder, go a little faster” during his 1992 tour of south China, belongs to the former, while stressing the supremacy of social stability belongs to the latter. These signals often appear in leaders’ speeches and commentaries in official newspapers, sometimes in an opaque and elliptical manner, and local governments are supposed to read between the lines to “comprehend the spirit of the center” (linghui zhongyang jingshen). For example, with the notion “socialism with Chinese characteristics,” the center makes a conservative proclamation if it emphasizes “socialism” and a pro-reform proclamation if it emphasizes “Chinese characteristics.” For concreteness, I use “socialism” to represent the pro–status quo rhetoric in the model and “Chinese characteristics” the pro-reform rhetoric.
Naturally, the reformist center would like to talk up the reform while the conservative center prefers to do the opposite. Although rhetoric is often characterized as cheap talk, the production of official discourses and various ideological campaigns in China is not without cost (Holbig 2008; Shambaugh 2008). In addition, saying things that one does not really approve of can incur psychological dissonance costs (Kuran 1995), and Shih (2008) argues that the language Chinese officials employ in public is often “nauseating.” Therefore, I assume that central leaders will pay a relatively small but positive cost c if they make declarations that they dislike, with c < v – b, while the cost of making one’s preferred declarations is normalized to be zero. This means just that, ceteris paribus, central leaders prefer to make announcements that they like rather than those they dislike.
Besides engaging in rhetorical activities, the center can also punish the local government if its action digresses significantly from the former’s preferences. In particular, the local government will receive an expected punishment of p if it chooses the center’s least preferred reform level. Given the size of China and the limited monitoring capacity of the center, it is reasonable to assume that local officials will be punished only if their digression is large and obvious. Consistent with this assumption, some notable local policy deviations in China have occurred in relatively small and remote areas, where reforms are less eye-catching.
I assume p is greater than b and v – b, meaning the punishment has some deterrent effect. On the other hand, p is smaller than v, which reflects the limit to the severity of central punishment over local policy digressions in the reform era (as long as such digressions do not involve power struggles against the center), as the CCP has learned from history to refrain from using cruelty in dealing with intraparty policy conflicts. In 1985, for example, General Secretary Hu Yaobang explicitly stated that although local officials might be fired for committing mistakes, they would “normally not be deprived of their material privileges, nor would their family members and relatives be discriminated against” (quoted in Cai and Treisman 2006, 515). For some officials, being fired from government jobs in the reform era is also not as costly as in the past—they have plenty of options outside the government sector (D. Li 1998; Montinola, Qian, and Weingast 1995).
Besides the likely punishment, the center can also bestow awards to the local government. In particular, the conservative center will give the local government an award for playing L, without which L will be a dominated choice for the reformist local government. This “conservative award,” denoted a, either can be purely honorary—the local government may be praised for “adhering to the socialist road”—or may involve an office promotion. Although the reformist center can also give the locals praises for playing the reformers’ favorite option M, such an honor is less important than the material benefits a local government can reap from local development, and hence will be assumed away for simplicity. I let a < v because if the award is purely honorary, it is simply compensation for playing low reform and so should not mean as much as one’s maximum benefit from reform. If it involves office promotion, there are always more lower-level officials than higher-level positions, and so a lower official choosing the conservative center’s favorite reform cannot be guaranteed a promotion; but the benefit of reform can be enjoyed by every reforming locality. I do not assume a priori the relationship between a and b.
The sequence of the game is as follows (see Figure 1). Nature moves first, choosing a reformist center with probability π and a conservative center with probability 1 – π. After nature’s move, the center chooses to send the conservative message “socialism” or the reformist one “Chinese characteristics.” Finally, the local government selects a level of reform, and payoffs are assigned.
The central–local rhetoric and reform game.
Results and Discussion
This is a signaling game of incomplete information, for which I use the solution concept of perfect Bayesian equilibrium, together with the “intuitive criterion” refinement from Cho and Kreps (1987). Solving the game leads to the following result (for proof, see the online appendix at http://prq.sagepub.com/supplemental/):
The game has a pooling equilibrium and a separating equilibrium. In the pooling equilibrium both types of center proclaim “socialism,” and the local government chooses M on the signal “socialism” and H on “Chinese characteristics.” When a ≤ b, this equilibrium is sustained by beliefs π < (p + b – v)/p and r ≥ (b + p – v)/p; when a > b, the equilibrium is sustained by beliefs (a – b)/(a + p) < π < (b + p – v)/p and r > max {(b + p – v)/p, (a + p – v)/(a + 2p)}, provided that p2 + 2p (2b – v) > a(v – b). In the separating equilibrium, which can occur only when a > b, the reformist center proclaims “Chinese characteristics,” the conservative center proclaims “socialism,” and the local government chooses H on “Chinese characteristics” and L on “socialism.”
In other words, in the pooling equilibrium, both types of central government engage in conservative rhetoric, and the local government chooses intermediate reform. This equilibrium occurs when the local government believes that the center can be reformist but the probability has an upper bound; a deviating center that makes the pro-reform proclamation, however, is believed highly likely to be a reformist center. This equilibrium can hold regardless of the value of the conservative award, although when the value is high, both the prior and off-equilibrium posterior beliefs about the probability of the center being reformist are more stringent. The separating equilibrium can occur only when the social environment is such that the conservative award is well valued (a > b). In this equilibrium the reformist center distinguishes itself from the conservative and delivers pro-reform signals, leading the local to play high reform, while the conservative center sticks to pro–status quo rhetoric, leading the local to play low reform.
The intuition behind the result is the following. The conservative center will always want to declare “socialism.” When the ideological award for playing conservative is not well valued, the local government will choose high reform if it knows the center is reformist and medium reform if it knows the center is conservative; low reform is a dominated choice. Because the reformist center prefers medium reform the most, it will pretend to be the conservative center and announce “socialism” too. Given the pooling of signals, local governments cannot tell precisely if the center is reformist or conservative. When the prior probability that the center is reformist is below an upper bound, the local government will choose medium reform to be safe. If the prior probability is high, however, this pooling of signals cannot be an equilibrium because the local government will turn to high reform and then the reformist center would rather announce “Chinese characteristics” to save the signaling cost.
When the ideological award is well valued, two things may happen. Either the reformist center distinguishes itself from the conservative center and declares “Chinese characteristics” to induce the local government to play high reform (the local government will play low reform if the rhetoric is “socialism”), or, when the local government’s prior belief about the center being reformist is within an intermediate range, the reformist center knows that the local government will play medium reform if its type is not revealed, and so decides to again mimic the conservative center’s rhetoric.
The current model assumes that the local government will be punished when it chooses the center’s least preferred reform level, regardless of the center’s message. Sometimes it may be the case that the local government will be punished only if it has chosen the center’s least favored reform and the center had already signaled it not to do so. With this modified assumption the same equilibria still obtain, and the beliefs supporting the pooling equilibrium are in fact more relaxed: when a ≤ b, the pooling equilibrium in which both types of center signal “socialism” is sustained by the belief π < (p + b – v)/p; when a > b, the equilibrium is sustained by the belief (a – b)/a < π < (b + p – v)/p.
The above equilibria occur when the local government’s preference ordering is H > M > L. Generally speaking, when the local government is less reform minded, there is no pooling equilibrium. The only equilibrium is separating in which the reformist center proclaims “Chinese characteristics,” the conservative center proclaims “socialism,” and the local government chooses M, unless its top preference is L and the center has proclaimed “socialism.” 5 Although this result is not part of the model here, it suggests that the rhetoric-reform framework can also help explain the diverse behaviors of local governments in China; in particular, for less liberal-minded local governments to adopt reform, the higher authorities must send strong and explicit instructions urging reform.
Empirical Illustrations
The most conspicuous feature of the rhetoric-reform game is that in the pooling equilibrium the reformist center imitates the conservative center and proclaims “socialism.” This accords well with the prevalence of conservative rhetoric in the reform era, recurring ideological campaigns, and the existence of official news programs with their notoriously obsolete and dogmatic reporting style. This pooling of rhetoric is reflected in both general discourses and discussions about specific policy issues, as shown in the examples below. Given the pooling of the signals and the uncertainty about the center, a reformist local government will still carry out reform but has to limit or guard it in some way, which contributes to a gradualist style of reform in China.
In this section I discuss several prominent examples of local reform without central authorization to illustrate the logic of the pooling equilibrium in the model, state enterprise privatization, agricultural decollectivization, and township-level elections, each of which is a milestone in China’s reform. By describing the contexts and tracing the sequences of the local reforms, such case studies can often clarify the causal mechanism and adjudicate among alternative theoretical explanations better than quantitative data (H. Brady and Collier 2010); a comprehensive data set that could be used for a quantitative test of the article’s theory is also not available at this time. In addition, I provide an example of the separating equilibrium, which does not speak directly to my initial research questions and is an unexpected outcome of the model. It offers, however, a valuable instrument for model testing: one’s confidence in the theoretical model will increase if there is empirical evidence for its unintended implications (King, Keohane, and Verba 1994). These cases have been well documented in the literature, and this article provides a coherent analytical explanation and a reinterpretation.
Local reforms without central authorization are ubiquitous in China. In at least one instance the State Council, China’s cabinet, even explicitly encouraged local officials to adopt a “doing without asking” tactic: “While implementing the coastal economic development strategy, many policy issues will be involved. [Local governments] should do more and say less, act first and speak later, or not talk about certain things at all.” 6 This shows that the logic of the theory here has broad applicability, not just in a few prominent examples.
State Enterprise Privatization
China’s privatization of state-owned enterprises (SOEs) began as early as 1992–93 in pioneering localities such as Yibin of Sichuan Province, Zhucheng of Shandong Province, and Shunde of Guangdong Province (Cao, Qian, and Weingast 1999). Locally managed SOEs had suffered significant financial losses at the time, and so privatization could relieve local governments of a heavy financial burden. But although individuals had been allowed to set up small private businesses in the 1980s, privatizing SOEs was very much a taboo in the early 1990s. Even Deng Xiaoping’s southern tour in 1992 did not touch on privatization. The first sign of relaxation in ideological constraints came only at the end of 1993, after the reform had already started in the aforementioned localities, when the third plenary session of the Fourteenth Party Congress allowed some small SOEs to be contracted out, leased, or sold.
The aforementioned local governments nevertheless pushed the reform, privatizing most of their SOEs. But, consistent with the model’s prediction, the local governments also tried to make the reform (appear) moderate. Officials in Shunde, for example, adopted a “four-no principle”: no report, no promotion, no review, and no publicity (Cao, Qian, and Weingast 1999). This strategy had occurred earlier. In the 1980s, when private businesses were discriminated against if not illegal, local governments often gave them a “red hat” by registering them as “collective enterprises” or enterprises “affiliated” with governmental entities, hence privately encouraging the emergence of private businesses in China before SOE privatization.
To control the pace of reform, however, the central government did not endorse privatization even after the practice had become more widespread in 1994. Limited and vague endorsement from the center came in 1995, when General Secretary Jiang Zemin stated in a party congress session that “we should concentrate our energy in helping the large state enterprises do well, and the sector of small and ordinary state enterprises can be further opened,” hence the slogan “grasping the large and releasing the small.” But what “releasing the small” really meant was not clearly specified by the center. Even in 1996 a prominent central leader declared that “China will never go in for privatization, though it supports the development of a private economic sector” (Beijing Review 1996, quoted in Sun and Pannell 1999, 279-280). Only in 1997, during the Fifteenth Party Congress, did the central authority make it a national policy that the state should divest itself of the majority of small- and medium-sized state enterprises by sales, mergers, or bankruptcy (Cao, Qian, and Weingast 1999; Morris, Hassard, and Sheehan 2002).
But even after 1997, the central authority did not officially acknowledge the country’s relatively gradual but massive privatization; instead, it has insisted that “keeping public ownership as the mainstay” is still a basic principle of China’s economic reform. In official discourses and media articles, privatization is usually not called privatization, but rather zhuan zhi (ownership change) or, in the 1990s, “corporatization,” the conversion of a state enterprise into a shareholding company, which was usually followed by securitization and subsequent sale of shares on the stock market. 7 Private enterprises, at the same time, were routinely called people-run enterprises (minqi) 8 rather than “private enterprises.” And the debate on public versus private ownership is couched in terms of the survival of Chinese national industry amid competition with foreign firms (Gallagher 2002). Only in 2004 was the protection of private property formally written into China’s constitution. All these puzzling language games and the sequence of events during the SOE privatization are consistent with the prediction of the model, namely, local governments conduct cautious and guarded privatization in the unfavorable ideological environment, while the center maintains discouraging rhetoric to restrict the pace and spread of the reform.
Agricultural Decollectivization
Although there existed earlier instances of decollectivization, China’s agricultural reform started in 1978 in earnest, when desperate farmers in the Xiaogang Village of Anhui Province divided their team land into separate plots to be farmed by individual households. This household responsibility system, called baochan daohu or baogan daohu, was almost heretical at the time, and the farmers promised in a written agreement to take care of the production team leader’s children until age eighteen if the leader was to be jailed for what they were doing. The central leadership immediately and explicitly prohibited the new practice through policy circulations such as the Regulations on the Work in Rural People’s Communes (Draft for Trial Use). Even more moderate forms of reform such as “contracting everything to the group” were warned against as contravening central regulations by the People’s Daily in the form of an editor-endorsed “letter from readers.” Local officials could have bowed to such pressures and stopped the farmers’ initiative, but they instead lent their backing. Provincial Party Secretary Wan Li, for example, told his subordinates that “I myself support baochan daohu and think it can be fully experimented with. Some are worried that this might violate the center’s decision, but in fact it is in line with the center’s ‘spirit’ that stresses practice as the only criterion of truth” (Chung 2000, 97). Thus, with local protection, household farming diffused rapidly in Anhui and over time received central approval and spread to the whole country (for detailed accounts of the agricultural reform, see Chung 2000; Fewsmith 1994; Yang 1996; Zweig 1997).
Many scholars explained the reform by noting that Wan Li was an ally of Deng Xiaoping, who was then emerging as China’s paramount leader. But there was no evidence that Deng played a role in setting agricultural policy at the time or even favored household farming initially. Different studies have put the time Wan first talked to Deng about Anhui’s reform as either 1979 or 1980 (Chung 2000; Yang 1996), after it had already started. Besides, agricultural decollectivization without central authorization was not an idiosyncratic event unique to Anhui; other provinces such as Guizhou and Gansu had also consistently gone ahead of official regulations in implementing the household responsibility system.
What emboldened local officials, in light of the present model, was that the Cultural Revolution was over at the time and the central leadership was split into two factions, the conservative “whatever” faction associated with Hua Guofeng and the more reformist group associated with Deng Xiaoping and Chen Yun, with a large number of pragmatic cadres stigmatized or persecuted in earlier political campaigns being rehabilitated. In addition, the attack on “ultra-leftism” in official news media and the “truth criterion” debate after the fall of the “Gang of Four” had somewhat opened the social and political environment (Chung 2000). Therefore, local officials began to perceive that the likely punishment for policy innovation would be less harsh than before. Furthermore, provincial leaders such as Wan were careful not to challenge the center too aggressively; he sometimes toned down the reform and instead of promoting household farming, he merely encouraged loosening control of rural production (Yang 1996). All this evidence is consistent with the present model, in that (local protection of) reforms occurred when the center was no longer monolithically conservative and the reformers were ascending; furthermore, local leaders were cautious in conducting reform while the center continued signaling conservatively.
Township-Level Elections
The logic of the theoretical model can be applied not only to China’s economic reform but also to the political sphere. The township-level direct election is a case in point. Because of the nature of political reforms, however, central conservative opposition will be stronger and local governments’ benefit will be less tangible, and so one should expect such reforms to be more difficult and/or infrequent.
At present China’s grassroots elections are limited at the village level. Township heads are selected by the county party committee, although it goes through the formality of an election in the people’s congress of the township. In December 1998, however, officials of the Shizhong District of Suining City, Sichuan Province, organized China’s first widely known direct election of a township head in its Buyun Township (for detailed accounts of the Buyun election, see He and Lang 2001; L. Li 2002; F. Li 2003). Besides the obvious unlawfulness of such an election, local officials also knew that the Political Bureau Standing Committee of the party had discussed the issue not long before and had decided against it. The National People’s Congress had also rejected an appeal from Shenzhen to hold direct township elections.
What drove local officials to undertake the reform was its potential benefit—besides trying to alleviate a local governance crisis resulting from serious corruption among appointed officials, throughout the process they were conscientiously aiming at having China’s first government executive election (village committees are formally considered self-governing organizations in China), just like Anhui’s Xiaogang Village had been immortalized as the pioneer of economic reform. They also knew the likely punishment from the center would not be too severe—they were prepared to enter the private sector in case they would be labeled as “counterrevolutionary elements” and lose government jobs (F. Li 2003). At the same time, local officials were careful not to publicize the election and defy the central authority openly. They chose Buyun for the election because it was small and remote from urban areas, and so would attract less attention. They packaged the election in such rhetoric that it followed the “spirit” of the party leadership and the constitution such as “people are the masters of the country” (J. Zhang 2002; J. Zhang and Ma 2004). And they banned outside journalists and “anyone from Beijing” to be present at the election (F. Li 2003). After the election, however, the news still broke out because of local media competition. A commentary in Legal Daily, the newspaper of the Ministry of Justice, immediately criticized Buyun’s election as violating the constitution, but, importantly, the central government did not nullify the election result.
Such a reform was not unique to Buyun. In fact, it turned out that technically speaking the township’s 1998 election was not even the first in China—a few weeks earlier Sichuan provincial government had organized a similar election in a township named Nancheng. That election was held with even more secrecy so that it was not known to the outside world until 2001. Since the practice started in Sichuan, a number of other local governments have followed suit and let the masses participate in the election of township-level heads, although they modified the election method in various ways to nominally satisfy the central regulation so that there was a final formal voting in the local people’s congress (L. Li 2002; Dong 2006). All these situations occurred when the central government had not permitted direct township-level elections.
With regard to central rhetoric, although there have been advocates of direct township elections among intellectuals, the center’s lid on township reform had been tight. The most relevant central policy articulation prior to the Buyun election was Jiang Zemin’s report to the Fifteenth Party Congress in September 1997, in which he vaguely stated that China should “further extend the scope of socialist democracy and improve the socialist legal system” (Cheng 2001; L. Li 2002). Not only did this call lack concrete terms and was really cliché, it also came with “the precondition of adhering to the Four Cardinal Principles.” In the more concrete sentence including “grassroots organs of power and self-governing mass organizations in both urban and rural areas should establish a sound system of democratic elections,” the emphasis was on promoting administrative and financial transparency of such local organs and citizens’ supervision of local cadres. Because Jiang was reportedly responsible for inserting an article into the village election law that affirms the village party branch rather than the elected village committee chairperson as the leadership core in villages, most observers and local officials perceived the center as encouraging administrative reform and organizational streamlining rather than promoting grassroots elections to higher levels (Cheng 2001; L. Li 2002). Nevertheless, some local governments carried out township-level elections, and the chief architect of the influential Buyun election, Zhang Jinming, has subsequently even been promoted to a higher party position and is widely regarded as a reform pioneer in China. This example shows that even in the more sensitive political sphere, local governments can initiate reforms in the face of unfavorable central rhetoric. At the same time, while the center’s attitude has been adverse to the spread of township-level elections, it did not annul the original reform or punish the local reformer.
An Example of the Separating Equilibrium
China’s experience in the wake of the 1989 Tiananmen Incident was consistent with the separating equilibrium in the model. Following the crackdown on the student movement, the reformers were mostly pushed out of the central leadership. Economically, the conservative premier Li Peng carried out a retrenchment program. Politically the party’s General Secretary Jiang Zemin decided to err on the “left” rather than “right.” More than the usual rhetoric, ideological preaching of antiquated slogans such as “class struggle” and “struggle with peaceful evolution” were reemphasized in official news media, and ideological adherence and Marxist morality became the leading criteria again in evaluating cadres (Zhao 1993). With the central rhetoric being a strong “socialism,” and with a > b, local governments opted to play low. This was the period when China’s economy hit record low growth rates in the reform period.
Frustrated by the retreat of reform and the stagnation of economic development, Deng Xiaoping made a celebrated “southern tour,” in which he rallied local reformist forces and pushed for bolder reforms and faster growth and threatened that anyone who obstructed reform would be removed from power (Yang 1997). He also publicly declared that there should be no debate about whether the reform was socialism or capitalism in nature but rather whether the reform can strengthen the country and improve people’s livelihoods. After a blizzard of media reports of Deng’s speeches and his pithy proclamations such as “be a little bolder, go a little faster” (following an initial period of silence), and the relay of his remarks to party cadres at various levels, local governments received a strong signal that the reformist force had come to dominate the center and reacted emphatically even when central bureaucrats’ responses had been more muted (Zhao 1993). Economic activities surged and a fever of development zones swept China, with many localities ignoring relevant central regulations in their haste to reform and to attract investments (Yang 1997).
This experience is consistent with the signaling game’s separating equilibrium, in which local governments play low when the conservatives are controlling the center, sending clear antireform signals, and then the reformist force breaks out by explicitly urging reform, leading local governments to rush to reform. That Deng was not able to push for reform in Beijing but had to rally local forces also illuminates the point that central reformers use local initiatives to circumvent conservative opposition to reform.
Concluding Remarks
China’s reform process has often been described as mysterious and confusing, and scholars have called for a general model of reform, Chinese style (Cao, Qian, and Weingast 1999). Although a general model will be difficult to construct because of the complexity of macro processes and the multitude of different circumstances under which new policies can and do emerge, the model here has provided a theoretical framework for understanding an important recurring phenomenon in China, that is, the center’s rhetoric is often unfavorable to reform, but local governments nevertheless have initiated or protected unauthorized reforms, with the associated covertness, jerkiness, and incoherence of various local policies (as well as the many problems this lack of coordination and overarching design often leads to). Sometimes, however, the central leadership has departed from usual practices and sent clear instructions urging reform, which this framework has also accounted for.
This characterization helps explain the nature of the gradualist reform in China, which has distinguished China’s reform from that of the former Soviet bloc and sparked debates about the relative virtues of “gradualism” versus “big bang” (Naughton 2007; Roland 2000). As of now, the reform era of the People’s Republic of China (1978–2012) has persisted longer than its pre-reform period (1949–78), and the reform is not over. An important but rarely explicitly asked question is why China has spent more time departing from the socialist system than being governed under that system.
A standard answer is that China’s reform is heavily experimental and exploratory, as the widely cited saying “crossing the river by feeling the stones” indicates. While the policy exploration explanation has many merits, it is more suited to explain when China first began reform and its eventual goal was vague, but after the early 1990s the goal of economic reform has arguably become very clear: establishing a competitive market economy (Naughton 2007). But China’s reform can still be characterized as “gradualism” in general, and so demands an explanation beyond “crossing the river by feeling the stones.”
This article provides a complementary answer: China’s gradualism has as much to do with politics as with policy exploration. When local governments are unsure of the center’s type, and when central reformers are wary of local officials’ ambitions and potential attacks from the conservatives, they adopt a particular style of “muddling through,” with the central reformers engaging in pro–status quo rhetoric and local reformers making guarded reform. Judging from this perspective, while China’s reform appears to be “crossing the river by feeling the stones” from a policy perspective, in the political sense central and local reformers have been using the time-honored stratagem of deception—“crossing the ocean by deceiving the sky” (man tian guo hai). This is what has made China’s reform a long, perplexing, and sometimes contradictory process.
The central–local interactions of rhetoric and reform modeled in this article are not unique to China but can be found in other countries too. For example, Malesky’s (2008) empirical analysis of how foreign direct investments affect subnational governments’ behavior in Vietnam shows that there are similar dynamics in China’s southern neighbor. There provincial governments have analogous incentives to those of Chinese local governments and often enact unsanctioned reforms beneficial to foreign investors. Vietnam’s central government, which contains both reformist and conservative elements, uses state-owned newspapers to censure such breaches to limit their spread, without actually stopping the provincial initiatives. More explicit or systematic applications of the present model to other countries, or to settings other than government reform, are beyond the scope of this article but are valuable future research topics.
Footnotes
Acknowledgements
I am grateful to John Aldrich, Thomas Hansford, Bahar Leventoglu, Edmund Malesky, Steven Nicholson, Emerson Niou, David Soskice, Michael Munger, Lynn White, and especially Karen Remmer for their advice and encouragement at various stages of the article. I would also like to thank all the anonymous reviewers who have provided helpful comments. All errors are of course mine.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: A version of this article was included in my dissertation at Duke University, for which I received the Stern Dissertation Fellowship.