AccessMyLibrary provides FREE access to millions of articles from top publications available through your library.
In this article, I analyze how the structure of the Chinese state affects the probability that local cadres will comply with the directives of the center. Because the Chinese state consists of a five-level hierarchy of dyadic principal-agent relationships, the existence of even moderate levels of routine incompetence and noise ensures that compliance will be less than perfect due to simple error. Moreover, because the center cannot perfectly differentiate between simple incompetence and willful disobedience, the structure of the state enables cadres to engage in strategic disobedience. I thus conclude that the complexity of the linkages between center and locality are a major factor in the observed persistence of corruption and institutional malfeasance.
Chinese governments, both modern and historic, have been dogged by problems of local compliance. Even in the best of times, the leadership in the capital has never truly controlled the localities and local malfeasance has been a consistent theme in Chinese political history. Historically, the center sought to control local malfeasance by recruiting "honest and upright" officials and erecting parallel monitoring and supervisory structures (e.g., the Imperial Censorate, the Chinese Communist Party's Central Disciplinary Inspection Commission). Chronic noncompliance has persisted nevertheless. Since the advent of reform, to take one example, the center has repeatedly sought to control excessive taxation by local governments. Legally, rural taxes cannot exceed 5 percent of farmers' income. By 1990, however, taxes averaged 10 percent of farm income, with some localities reportedly levying taxes up to 20-40 percent of farm income. In all, the "farmers' burden" (nongmin fudan) imposed by the "three disorders" (san luan: arbitrary taxation, arbitrary levying of fines, and unauthorized expropriation) may have totaled between Y25 billion (US$6 billion) and Y100 billion (US$25 billion) per year in the 1990s (Wedeman 2000). Despite sporadic outbreaks of violent tax protests (e.g., Renshou County 1993) and repeated crackdowns by the central government, excessive taxation has remained a persistent problem in many areas for nearly two decades.
In this article I seek to explain persistent local noncompliance and malfeasance. Persistent local malfeasance reflects, obviously, some relatively significant level of dishonesty by local officials and hence willful disobedience. In this article, however, I argue that persistent noncompliance is not simply a function of dishonesty. Although dishonesty plays a role, I contend that the structure of the Chinese state is such that incompetence, random chance, and noise will create a significant "natural rate of noncompliance" absent willful disobedience. Some degree of incompetence is inherent in any bureaucratic hierarchy, with the result that a principal's orders may not be carried out simply because her agents lack the skill and ability to complete their assigned tasks. In other instances, well-intentioned, diligent, and competent agents may fail because "nature" turns against them. Noise--information that has been distorted in the process of collection and transmission--creates additional complications beca use well-intentioned agents may misunderstand the principal's orders and do the wrong thing. Noise also makes it difficult for the principal to determine if her orders have been correctly carried out and, if not, whether the failure results from incompetence, bad luck, or willful disobedience.
Uncertainty about her agent's reliability and the causes of noncompliance means that the principal may either fail to detect both inadvertent and willful noncompliance or, if she does, to incorrectly attribute willful disobedience to incompetence or accuse her agents of willful disobedience when in fact they failed to carry out her orders due to incompetence or bad luck. Because information about her agents' actions may be contaminated with falsehoods, the principal may also accuse them of disobedience when they have in fact faithfully carried out her orders. False accusations, however, can have a highly detrimental effect on agents' morale and their willingness to faithfully serve a principal with a reputation for arbitrariness. I thus posit that when faced with ambiguity, the principal is apt to be cautious and reticent.
The resulting combination of environmental ambiguity and caution creates conditions in which cadres can take advantage of the center's uncertainty to engage in willful disobedience. If cadres believe that any violation of orders will be immediately and inevitably detected and punished, then they would presumably never disobey, so long as the punishment they expect to suffer is greater than the anticipated benefit of disobedience. Because it is quite possible for the center to adopt a policy of automatic and harsh punishment (in game theory terms, a "boil in oil" punishment strategy), so long as cadres are convinced that there is a high probability of getting caught, they will be deterred from violating their roles as agents. Incompetence, random chance, and noise, however, create a situation in which automatic punishments may prove dysfunctional or impossible.
Cadres can, I argue, play on the center's assumptions about incompetence, random chance, and noise in ways that force us to consider detection and punishment as probabilistic, particularly for cadres operating at the bottom of the administrative hierarchy and hence at the greatest administrative distance from the center. That is, if they willfully disobey orders, there will be some not insignificant chance that the center will: (a) not detect their violation, (b) will assume that the violation was due to either incompetence or random chance, or (c) will not take any action because it is not sure that the violation was a result of willful disobedience. Moreover, even if the center does detect a violation, concludes it was due to willful disobedience, and orders disciplinary action, the multilayered structure of the Chinese state creates conditions in which those orders may not be carried out, for essentially the same reasons that the center's original orders were not: incompetence, random chance, and noise--p lus, of course, possible willful disobedience.
In making this argument, I do not assume that Chinese cadres are bunglers or that the Chinese state is dogged by bad luck. Instead, I assert that the complex, multilevel structure of the Chinese state apparatus means that even low levels of incompetence, bad luck, and noise will result in a high natural rate of noncompliance. The presence of a high rate of natural noncompliance, in turn, encourages high rates of willful disobedience. Persistence noncompliance problems such as the san luan thus have their roots in the structure of the Chinese state and are likely to continue so long as the state remains a unitary hierarchy. The structure of the Chinese state also helps explain the regime's heavy reliance on periodic crackdowns and campaigns. Because routine monitoring and "police patrolling" are likely to prove insufficient in controlling local malfeasance, the center is forced to resort to periodic "terror campaigns," often launched after local protests set off "fire alarms," whose overt function is to root out malefactors, but whose primary function is to instill fear among cadres and to deter malfeasance.
As typically employed, principal-agent models apply to situations where a principal employs an agent to perform some task, or set of tasks, on her behalf. In certain cases, the principal may employ a number of agents, either because the magnitude of the task is such that a team effort is required or because the principal relies on the agents to monitor each other and hence ensure that all exert a maximal effort. The model assumes conflicts between the goals of the principal and her agent such that unless the principal effectively monitors the behavior of the agent, or designs contracts that align the interests of the agent with her, the agent will not perform the tasks assigned to him but will instead either shirk (do no work) or usurp the authority entrusted in him by the principal to advance his interests rather than those of the principal (see Brehm and Gates 1997).
Although it is clearly possible to borrow elements of principal-agent theory and apply them to the study of central-local relations in China, it is first necessary to examine the structure of principal-agent linkages within the Chinese state (see Granick 1990; Shirk 1993; and Harding 1981). The Chinese state consists of five distinct territorial-administrative levels. At the top, directly below the center are the provincial-level units.  Beneath the provinces, the second level consists of municipalities and prefectures. The third level consists of counties, below which the fourth level consists of townships and villages. At the bottom of the administrative hierarchy is the hamlet. Technically, the township, village, and hamlet fall outside the formal state structure, and cadres employed at this level are not considered state officials. In practice, authority relationships between grassroots cadres and local governments at the township level are such that for my purposes these low-level cadres can be consi dered agents of the state, albeit ones whose links to the formal state are somewhat more tenuous than state cadres employed at other levels.
Because the current one-level down nomenklatura system defines a principal-agent interface at each of these levels, the Chinese state consists of a five-level hierarchy of dyadic principal-agent relationships. Thus, the province is the agent of the center, the municipality/prefecture is the agent of the province, the county is the agent of the municipality/prefecture, the township/village is the agent of the county, and the hamlet is the agent of the township/village. It is perhaps, however, more accurate to think of the province, prefecture, county, and township as agents-cum-principals because they serve concurrently in both capacities, acting as agents of their superiors and principals vis-a-vis their subordinates.
The relationship between levels is heavily influenced by the system of cadre contracts wherein superior levels set performance goals for their subordinates and withhold a percentage of their salaries as a form of performance bond (Li and O'Brien 1997; Whiting 2000; and Manion 1991). If cadres fulfill the targets spelled out in their contracts, they receive their salaries in full. If they overfulfill the targets, they may be granted bonuses. Those that fail to meet their targets are docked. As predicted by the idea of moral hazard, cadres thus strive to fulfill the targets set for them by their immediate superiors, particularly those that are "hard" (i.e., relatively easily measured and hence frequently those that can be quantified) but are less diligent in fulfilling "soft" targets and are relatively indifferent to the demands of their superior's superiors or other bureaucratic actors within the administrative hierarchy. Thus, for example, township and village cadres are apt to be …