A Micro-Geography of State Extractive Power: the Case of Rural China

Abstract

This article argues that geography is essential to explain subnational variation in extractive power. To exercise power over the expanse of their realm, governments need an infrastructure to access even more peripheral communities. But as rulers overcome the problems of distance, they discover the challenges of proximity, because locals influence state agents and undermine state autonomy. Studying the uneven reach of Chinese local governments within their jurisdictions, we demonstrate that micro-variation of extraction in rural China, including monetary and non-monetary levies, provides a rare window onto the micro-geography of power. Physical distance still obstructs extractive power projection into extremely remote places, but physical proximity has become a greater constraint for the exercise of power. Local governments are pressured by their immediate neighborhood, at the expense of less fortunate citizens, who live at intermediate distances: still within the reach of the state, but already too far away to turn the state to their favor. The effect of organizational distance, which in China is mediated primarily through the Communist Party, is similarly dualistic: Party networks alleviate asymmetric information problems, which hamper rural taxation around the world, but party networks are sustained by patronage and therefore jeopardize state autonomy.

“Build roads, bridge distances!” is a universal battle cry of ambitious state-builders. This article argues theoretically and shows empirically the ambivalent effect of distance on the ability to exercise power. While bridging physical and organizational distances helps to consolidate state power, it also has unintended side effects affecting state authority. Shrinking distances empower not only the state but also its citizens. The same road that allows a state to provide for its citizens also allows citizens to seek out their state, in legitimate ways as a check on power and in less legitimate ways as influence peddling. On the one hand, roads are an effective device for a state to project power. But on the other hand, roads also pave the way for petitioners, rebels, and sycophants. Roads stand for a number of state-building devices, which face a similar dilemma. Organizational networks, which function as instruments of control, are often maintained on the basis of privileges. State-controlled media networks not only allow a state to spread its message but also provide citizens with a common language of protest. State ideology not only disciplines citizens but also puts pressure upon the government to act in accordance with orthodoxy. In general, state-builders who penetrate society at the same time expose themselves to societal pressures. A state finds it challenging not only to project power over a distance but also to exercise authority at close range.

While speaking to larger themes of geography and power, this article limits itself to one particular kind of power, namely state extractive power, and to one particular geographic variable, namely distance. In political science as in other social sciences, distance is a variable that is often evoked, yet seldom¹ systematically studied. The distance of a community from the political centers is a fascinating variable, because ordinary citizens themselves are acutely aware of the geographic place of their community within the larger polity; some citizens have implicit theories as to how their relative distance from the government has practical consequences for their lives. This is unlike other social scientific variables, which lie far beyond the experiential horizon of ordinary citizens. The distance of a community from the political centers is also fascinating, because state-builders everywhere act in geographic space, so the problem of distance and power is of universal relevance. The goal of this article is to formulate a parsimonious micro-geography of state power, mapping out and explaining the degree of power a state exercises over communities as a function of their physical and organizational distance from the political centers.

China, a country whose government must deal with daunting distances, provides an ideal case to explore mechanisms and test implications of a micro-geography of state extractive power. Local governments confront both the problems of distance and the perils of proximity. Since at the beginning of the twenty-first century, the country has a relatively advanced transportation infrastructure, one might wonder whether physical distance of a locality from the political center still matters. Challenging conventional wisdom, according to which governments find it easier to rule over nearby communities rather than distant places, our central claim is that, as one moves away from administrative centers, state extractive power varies like an inverse U: With increasing distance, state agents become more insulated from citizen influences and state extractive power increases. Yet beyond a certain point as distance increases further, remote places remain difficult to tax. Studying variation from household to household, the analysis demonstrates that a subnational approach to the state can be fruitful even at an extremely high resolution.

The theoretical scope is limited to rural areas, since state-society relations function differently in urban settings. Moreover, distances in urban geography are much shorter than in rural geography, so that the two cannot be analyzed on the same terms. For instance, one would need to decide on the urban distance that matters: The one to “People’s Square” as a potential site for protest? To the party committee as the pinnacle of power? Or to the tax office? In the countryside, these places are in relative proximity, compared to distances from rural households to the nearest township or county seat. Theoretical refinements for urban settings are beyond the scope of this article; the urban portion of the survey we use lacks information on distance needed for testing.

The article is organized as follows. “Toward a Geography of State Extractive Power,” provides the conceptual building blocks of this article. Next, the case under consideration is introduced in “State Extractive Power in Rural China in 2002.” “Physical Distance and Extractive Power” hypothesizes how physical distance affects state power across space. “Organizational Distance and State Strength in China” expands the theory, analyzing the role of Communist Party networks. The section “Empirical Analysis,” carries out tests at two levels of spatial resolution and confirms that extractive power is greatest at intermediate distances from local political centers. The last section concludes that in contexts where extractive power varies across space, proximity can pose greater challenges to a state than distances.

Toward a Geography of State Extractive Power

This section presents our interpretation of state extractive power, avoiding the flawed distinction between state capacity and state preferences. It describes how the article takes the subnational turn in the state-building literature to an extremely fine-grained level, by disaggregating state extractive power all the way down to the household level. The section also points to the promises and challenges that accompany geography-based explanations of subnational micro-variation.

State Extractive Power

Subnational variation in taxation levels has received much less attention from political scientists than cross-national variation of taxation levels, which has served as a valuable clue about state-society relations (Lieberman 2002). While the ability to raise taxes is indicative of state capacity, actual levels of taxation are certainly not a perfect measure of state capacity; alternatives have been suggested (Soifer 2008). The link between taxation and state capacity is tenuous, because state preferences are a confounding factor: Some states may choose to punch below their weight. Low levels of taxation may indicate a lack of state capacity or state preferences for low taxes.

Operationalizing the distinction between preferences and capacity is hard, partly because none of the two concepts can be directly observed and partly because the distinction is conceptually flawed. In strategic situations, agents internally formulate their preferences while taking into account their options and the tradeoffs involved. At the time of formulating tax policies, the state already factors in its realistic options. Preferences are not independent of capacity. The metaphoric personification of states as possessing capacity and preferences is fundamentally ambiguous: If a regime refrains from raising taxes out of fear for its political survival or because it feels bound by customary law, we may refer to this as preferences for low taxes or a lack of capacity. In democratic contexts, when state preferences reflect citizen preferences, low taxation could be interpreted as reflecting public opinion or as the failure of the state to convince its citizens to trust in state projects.

This article conceptualizes state extractive power as the force resulting from state preferences in combination with state capabilities. State extractive power refers to the share of resources that a state extracts from its citizens, encompassing both monetary and non-monetary resources. In the case of rural China, this is an essential point, since labor services have become a major venue through which local governments extract resources from citizens. State extractive power is the result of complex policy-making and policy-implementation processes, each stage of which affects the resulting levels of taxation (compare Fig. 1). State extractive power depends on the preferences held by a dictatorial clique or a democratic population, with widespread societal preferences for high taxation favoring greater state extractive power. Tradition and experiences may lead to state-friendly preferences, which empower the state going forward. In the process of aggregating preferences to a state policy, institutions may be shaped in ways to favor statist over non-state solutions, thereby further increasing state power. The ease of resource collection depends not only on the efficiency of the state’s bureaucracy but also on the costs of extraction in a particular milieu. These factors will determine how much the state decides to extract and whether it will be able to achieve its goals. In short, the share of resources that a state extracts from households points to key aspects of state power, which depend on ideas and institutions as much as on taxation infrastructure.

Fig. 1

State extractive power—the ability and willingness to extract

The Subnational Turn in the State-Building Literature

Many states are much stronger in some areas of their territory than in others. State strength can vary from one province to the next, from one county to the next, and even from one neighborhood to the next. Variation within countries can be greater than variation between countries, and is of utmost consequence for the lives of ordinary citizens, especially in non-OECD countries. Beginning with Richard Snyder’s call for scaling down and making use of the “subnational comparative method” (Snyder 2001), interest in the subnational has surged among political scientists (Tsai 2009). Yet until recently, the state-building literature, committed to the idea of a unitary state, resisted the subnational turn. This changed, partly thanks to the steady stream of subnational studies of the state appearing in this journal. Subnational concerns were a recurrent theme in a special issue on state infrastructural power (Soifer and von Hau 2008), and the conceptual work on subnational democracy in Latin America is relevant far beyond that region (Giraudy 2013).

Many works that recognize subnational variation in the reach of the state stop short of a fully subnational analysis. For instance, one study recognizes that places with rough terrain experience state-building differently, but goes on to study the implication for cross-country variation (Fearon and Laitin 2003). Some authors who speak to the historical origins of subnational variation of state capacity ultimately remain interested in cross-national comparisons (Soifer 2015). Theories that would be highly relevant to explain subnational variation are tested cross-nationally (Nunn and Puga 2012). As measurement issues are being addressed (Gibson 2010), we also see fully subnational analyses, for instance explaining variation in state capacity within Germany, both through the lenses of election fraud (Ziblatt 2009) and public goods provision (Ziblatt 2008). A political scientist of Africa explains why the hold of the state over some rural communities is stronger than over others (Boone 2003), pointing out that this kind of variation may be the result of state-builders’ strategizing (Boone 2012). In China, where state-building outcomes can vary dramatically from one county to the next (Zhan 2013), there is an increasing number of subnational studies (Sheng 2009; Rithmire 2014).

Our article not only fully adopts the subnational mode of analysis but also turns to the smallest possible micro-level, zooming in all the way to the individual household level. The analytical key for subnational analysis is to disaggregate state-building outcomes in ways that meaningfully relate particular outcomes to a geographical location. Not all extractive activity can be disaggregated in this way: The extraction of customs duties, for instance, would only allow to compare state extractive power at different border posts, not throughout the expanse of the country’s territory. Direct resource extraction from individual rural households lends itself to subnational analysis in an almost ideal way, because households can be located in space and because they live throughout the expanse of the territory. Disaggregation to the household level promises to reveal the micro-geography of state extractive power in China.

Geography Variables

For political scientists, it is challenging to argue with geography, because the causal mechanisms involved bridge two very different domains: the physical domain of topography and the human domain of politics. But there are good reasons not to discard geographic variables. Geographic variables such as distance evoke a comparable calculus of power in different country contexts. In addition, explanations of political outcomes as a result of institutions invite questions as to where the consequential institutions came from in the first place. Geographic variables such as sheer physical distance of a locality from a political center do not lend themselves to similar questions—the only caveat being that people interact with geography, for instance by changing geographic features or migrating. At the end of the day, geographic hypotheses such as the one that rough mountains keep the state at a distance (Scott 2009), or that hydraulic systems predestine a society for authoritarianism (Wittfogel 1957), draw their appeal from their status as universal and deep reasons for political outcomes.

There is more to geography than topography. In its most physiographic interpretation, distance can be understood as the crow flies. Since political scientists are not studying crows, but people, they might instead count the miles, or the minutes, one needs to travel to get from A to B or the transportation costs for such a trip. Moving further and further from physical geography to human geography, one may conceptualize distance in more sociological or institutional terms. How close are the linkages between two communities? Compared to topography, human geography tends to be more specific to the country context. Recognizing that physical and organizational distances have much in common, this article studies them jointly.

State Extractive Power in Rural China in 2002

Rural China provides an ideal case for a micro-geography of state extractive power, since the large size of its territory and the diversity of its citizens make it impossible even for a highly effective government to achieve uniformity of control. This section introduces particularities of the Chinese case. It places rural resource extraction into the larger context of fiscal affairs, describes the agents and the institutions that exercise state extractive power in rural China, and addresses the significance of the pivotal year 2002.

Subaltern Fiscal Extraction in Rural China

Beijing’s fiscal policy toward rural areas is characterized by a lack of interest in the countryside as a direct source of revenue. The resources studied in this article are insignificant to sustain the health of the national budget. In fact, while we refer to these extracted resources as taxes, they are not officially called so and are largely ignored by fiscal budgets. If these resources were included, they would account for less than 2 % of total annual government revenues. For the Ministry of Finance, the question is how to limit fiscal transfers to rural areas. In combination with ambitious socio-economic programs designed by the central government to maintain stability in the countryside, the result is a proliferation of “unfunded mandates,” despite fiscal transfers (Wong 2010). To fulfill the many programmatic expectations of higher-level government, local officials are under pressure to raise income on their own.

The informal nature of fiscal extraction in rural China creates conditions where the fiscal burden of households varies dramatically across space. Effective levels of resource extraction depend on the result of a state-societal bargain. Local officials who are under formidable pressure to finance local, socio-economic programs face rural citizens who need to pay the price and do not necessarily want the programs (Takeuchi 2014). The stakes are high, since the lives of rural citizens are greatly affected by fiscal activity in the countryside. In the last decade, rural land expropriations on a massive scale were one of the most disruptive mechanisms for raising fiscal revenues. They are intimately linked to corruption (Chen and Kung 2014) and rural protest (Cai 2010). Apparently, localized political dynamics result in very different levels of informal taxation. After controlling for legitimate factors that determine the fiscal burden, there remains much variation which might be called tax discrimination. Since the state-societal bargain differs across space, the micro-geography of state extractive power is highly uneven across space.

The Extractive Power of a Multi-layered State

For decades the central government has declared its intention to reduce the peasant burden, publicly deploring abusive behavior by local officials (Bernstein and Lü 2003). On the one hand, such statements are part of a blame game between central and local authorities (So 2007). Since the central government creates incentives to carry out costly projects, denying responsibility for the peasant burden has been called an “evasive strategy” (Yep 2004). On the other hand, since peasant discontent is perceived as a threat to stability, central government admonishments are more than rhetoric. The tax-for-fee reform in 2002, which commutes fees into a tax to stop abuse, as well as the abolition of the agricultural tax in 2006 indicates genuine commitment to lower taxes on peasant households (Kennedy 2007). In short, a lack of bureaucratic control in combination with a lack of political will for a long time prevented Beijing from alleviating the peasant burden (Edin 2003).

The gap between centrally declared intentions and locally implemented policies by no means invalidates usual interpretations of state extractive power. High levels of resource extraction indicate both that the center lacks power vis-à-vis local authorities and that local authorities are powerful vis-à-vis citizens. In a context where the state is taxing in spite of itself, thinking of China as a unitary state is a dead end. Only by taking into account the multi-layered nature of the Chinese state, with the central government as well as local governments being powerful entities in their own right, can we arrive at a reasonable interpretation of state extractive capacity. Local governments possess significant autonomy to implement tax policies. The most relevant agents for local tax collection are the many thousand county and township seats scattered throughout the country. In a narrowly legalistic sense, extraction is carried out by self-governing bodies that are even outside the state. Yet since these village authorities are imbued with state-like authority, and since they are usually firmly under state control, they must be considered part of the state. These local sections of the state apparatus compel or convince citizens to give up part of their income and turn it over to local state agents. Even if different sections of the state work at cross purposes, resources that flow from citizens to state agencies indicate extractive power.

The Pivotal Year 2002

In the late 1990s and early 2000s, the peasant burden, not counting obligatory labor contributions, stood at an estimated 10 % of peasant net income (Aubert and Li 2002), maybe much higher (Li 2003). Since 2003, the central government moved decisively to alleviate the peasant burden. The abolition of the “2600-year-old” agricultural tax came with propagandistic fanfare, as if it meant the end of peasant exploitation. While the measures taken between 2003 and 2006 reduced the peasant burden (Li 2012), they failed to fundamentally alter the incentives of the agents involved. Local governments continue to be hard pressed for money (Fewsmith 2011) and have little choice but to seek less visible ways to extract resources. In the last 10 years, land development, sometimes under the “New Socialist Countryside” label, has created substantial revenues. Taxation in rural China remains a highly informal and contested political bargain.

This article focuses on the year 2002, the pivotal year before Beijing took the decisive reform measures. Previous reform initiatives were half hearted and largely ineffective (Bernstein and Lü 2003). The year 2002 is ideal for our analysis, because in the wake of the fiscal reforms to alleviate the peasant burden, extractive activity has been pushed further into informality. Local governments had even more reason to disguise their fundraising activities. It may not be a coincidence that the year 2002 is the last and only year for which we could find the necessary data. By not including the agricultural tax, as the most standardized extractive institution, our analysis concentrates on the kinds of fees, levies, and labor contributions that have continued to exist after 2002. After the reform shock, local state-societal bargains may have been in flux. Yet the informal tax activities of the year 2002 have been ongoing at least for one decade, representing an equilibrium.

Physical Distance and Extractive Power

This section develops the main hypotheses of this article, suggesting a curvilinear relationship between extractive power and the physical distance from the government. Distance affects state power in two distinct ways: First, the state experiences difficulties to reach communities in remote localities—a problem of infrastructural power. Second, it is difficult to insulate state agents from undue influences from their surroundings—a problem of state autonomy. We put both together to predict the micro-geography of state extractive power.

State Infrastructure and Distance

The null-hypothesis that extractive power does not vary as a result of distance might be plausible for highly developed countries. Yet even in countries that experience highly uneven extraction, state leaders deny such variation, because it implies a failure to achieve full territorial control. In China, the party line implies that levels of extraction do not depend on location. After all, the legitimacy of the party draws on its advertised commitment to fairness. While in fact, China is ripe with discrimination, which citizens may be willing to tolerate as a transitional phenomenon, cadres have told the authors the following, official stance on the micro-geography of taxation:

Hypothesis H₀ (party line): Households pay identical tax rates, whether they live close to local government or further away from it. Effective tax rates vary only for good reasons, such as income differentials or the preferential treatment of poor areas.

Contrary to this utopian vision of a geographically uniform state presence, political scientists are aware of the difficulty to project power over a distance, sometimes modeling it as transportation costs. African geography generates unusually high transportation costs, hindering development (Sachs 2003). Disadvantageously shaped countries are hard to govern, because power needs to travel long distances (Herbst 2000). Distance poses a formidable obstacle to state-builders when it comes in combination with rough terrain; “friction of terrain” helps people to escape state-building (Scott 2009). Instead of using sophisticated concepts of distance (e.g., Skinner 1977), the following analysis works with the parsimonious concept of linear distance. What needs refinement are not the measures of distance, but our understanding of state-society relations close to the center as opposed to the periphery, and the implications for extraction.

Across domains of governance, one can expect more state activities in places close to the local centers of power. A state’s infrastructural power, ranging from transportation networks to its bureaucratic presence, tends to be denser in the vicinity of county capitals and township seats, presumably even if controlling for economic factors. As for the provision of government services, nearby communities can hope for more attention from the government. When it comes to fiscal extraction and ensuring compliance, physical enforcement along with information collection will be easiest in proximity to the state, at least as a good first approximation. There is a point beyond which the grip of the government slackens, although the current transportation infrastructure has pushed this point further away than in the past.

Hypothesis H_1a (infrastructural power effect): Since it is both least costly and most rewarding for local governments to be active in nearby communities, households pay less taxes the further they live away from local administrative centers, controlling for other factors.

For the Chinese case, physical distance is also routinely cited as an obstacle to projecting power (Zhong 2003, p.85). Given the authoritarian-bureaucratic nature of the state, inspection tours by higher-level officials are at the core of mechanisms that results in a bias of tax extractive activity in communities close to the government. Visible achievements such as public works are the key for career advancement. As the literature shows, performance-based promotion systems result in temporal patterns of investment, local political budget cycles (Guo 2009).

They also result in spatial patterns, as local officials place projects in sight of inspection tours. Newly planted trees lining country roads are striking features of China’s countryside. They are part of China’s wasteful road economy—also called inspection economy—that serves to impress leaders on hasty visits, by placing imposing projects along their paths (Cai 2004). Beyond roads, more generally, some localities are more likely to be visited by higher-level leaders than others, as a result of physical geography. Local officials focus on nearby areas to showcase achievements. After all, even with low transportation costs, their superiors will find it more convenient to inspect villages in the immediate environs of the county seat, rather than remote mountain villages. Even if occasionally a higher-level leader travels further afield, the destination of such a trip is harder to predict, since the number of potential targets of long-range inspections is large. In short, the local official will focus on nearby places.

Since these government projects come at a price, the location of government projects, in turn, is bound to shape geographic patterns of taxation. One would expect informal taxation to occur where the money is spent: Labor contributions make up a large part of the resources under consideration here and presuppose that people are physically carried to the construction site. Monetary contributions are more flexible, but must also be in vicinity of the project for reasons of legitimacy. After all, it is hard to ask people to pay for a school that is too far away to benefit their children. The geography of taxation follows the geography of government spending, though this is not to deny that there is room for deviating from this pattern, e.g., by transporting people over a certain distance to contribute their labor.

Taking a long historical perspective, distances between centers of power and their population are not a physiographic given, but result from strategic choices. In the Song Dynasty (960–1279), the Chinese state placed its agents to facilitate extraction (Mostern 2011). Today, the government moves rural citizens to concentrated villages (Ong 2014). Internal migrants take into account the presence of the state, such as Han settlers accompanying bureaucrats to the imperial frontier. Yet in the short run, local administrative centers tend to remain in place. Even Mao’s revolutionaries, after failed experiments in the 1950s, reverted to the erstwhile county seats. Moreover, economic opportunities are the main driver of contemporary migration patterns, not taxation patterns. Thus, the analysis treats distances as fixed in the short run.

State Autonomy

It is hard to project power over a distance, but it is also hard to exercise power at close range. In China, where traditional relationship networks remain extremely consequential (Yang 2002), households living at the centers of local power are likely to have relatives or friends working for one of the many local agencies related to the party state. The reshuffling of leading cadres does not break the linkages that bind local officials to their surrounding communities. Securing tax privileges does not involve the higher echelons of power, where reshuffling policies are effective. Therefore, thanks to their personal relationships, households in the neighborhood of government authority can count on being lightly treated when it comes to extraction.

In addition, residents living close to the local government seat are also in a better position to put pressure on local officials. Comparing the effectiveness of different protest tactics, we know that a substantial government response is most likely to occur when protests are public and disruptive (Chen 2012, ch.6). Since a protest in a remote village could hardly attract the same attention as a protest in vicinity to the county seat, physical proximity is key. Protests at a county seat not only physically beleaguer the leaders but also allow for the protests to grow in size, thanks to higher population density. Citizens not living at the government seat would hesitate to make the trip, since for them, transportation costs are substantial: In the example of one county, most petitioners use the bike or walk, half of them traveling at least an hour (Zhang and Zhang 2009, p.147). Maybe most importantly, citizens in the neighborhood to the state know better how to effectively apply pressure and where to find the relevant offices. People from the countryside are less successful protesters than people from the urbanized center (Chen 2012, p. 183).

Citizens close to local centers of power may even notice visits by higher-level officials, whom they can turn to, taking advantage of central government efforts at alleviating the peasant burden. This is a particularly acute threat to local officials after suggestions by the central government to turn the local per-capita peasant burden into a promotion criterion.² This proposal may or may not have been implemented, yet it is clear that being caught charging excessive levies and fees poses a career risk. Citizens in places which receive attention from upper-level governments are in an advantageous position: Pressures from below and pressures from above work in tandem, so that local governments look for funding elsewhere.

Hypothesis H_1b (state autonomy): As local officials are pressured by and collude with citizens who live in the immediate environs of power, households living in close vicinity to the political centers pay less in taxes, ceteris paribus.

Synthesis

Hypotheses H1a and H1b are non-exclusive and complementary. As one moves away from the government seat, state infrastructure becomes sparse, while state autonomy increases. These two effects would cancel each other out only if the rate at which infrastructure weakens was identical to the rate at which state autonomy increases, an unlikely outcome. As we know from both natural sciences and social sciences, effects of distance tend to be non-linear. In physics, the inverse square law applies to fields ranging from electrostatics to acoustics, as well as to gravity. In international trade theory, the gravity model of trade predicts that the volume of trade between two countries is inversely proportional to the square of their distance. Similarly, one would predict infrastructural power to dissipate fastest in remote areas and state autonomy to dissipate fastest in the areas closest to power. It follows that we should expect an inverse U shape, with state extractive power reaching its maximum at an intermediate distance from power, although we are agnostic as to where this maximum might be.

Hypothesis H₁: The relationship between power and distance is curvilinear. Moving away from the centers of power, at first, state power increases, thanks to strong increases of state autonomy. Beyond a certain point, infrastructure begins to dissipate so quickly that it outweighs gains in state autonomy, and state power begins to decline.

Organizational Distance and State Strength in China

The Communist Party stands out as the most important organizational apparatus through which the Chinese government exercises power across its vast territory. Even in a party state, where the party and the state are interwoven, the two hierarchies remain distinct with separate command lines. This distinction is more obvious at the local level than at the central level. For instance, the state bureaucracy formally ends at the township level, whereas the party reaches down into the villages. The party’s rank-and-file members, who are not government officials, bridge the social distance between the rulers and the ruled. They also receive privileges, which are the perils of organizational proximity.

The Party as State Infrastructure

Since its takeover in 1949, the party has turned into a power-wielding device for transmitting instructions top-down and information bottom-up. One priority of the rural party apparatus is to “promote rural economic development.” ³ As a result, local party organizations are entangled with money, helping to raise the funds for a variety of activities. Village party members are not tax collectors, but mediators between the government and the villagers, involved in extraction. Until the early 2000s, village party branches received a cut from the money they collected, and the amount of money passing through the party’s village organization will only increase with additional service functions to be taken over (Ding 2014, p.83). Conceptually speaking, party members transmit information needed to assess taxes and monitor payments, thereby solving a thorny asymmetric information problem encountered throughout the world. In other countries, the transaction costs of rural finance can be prohibitive, but in China village party members provide a critical link. They are well connected in the village and also take instructions from the township government. The more party members there are in a village, the greater the local government’s extractive capacity, one would expect.

Party presence matters for tax collection also in a less direct, but by no means less important way, since county and township governments rely on rural party networks to govern their jurisdictions, including when implementing projects (Xiao 2008). Since ambitious projects come at a cost and are usually accompanied by substantial fees, upper-level governments rely on village party cells to communicate the goals of these projects, so that the projects are well understood and hopefully endorsed by the citizens (Zhou 2012). Leaders need an early warning system when projects create discontent, and if protests erupt, they need channels to mediate the situation before it escalates to a level that would be noted by the media and higher-up government. The sub-governmental party system provides these tools. Formal and informal contacts, party meetings, and consultation across hierarchical levels provide a regular, steady flow of information. The more information there is, the more confidently can local governments carry out projects. It is not only about the government squeezing out money but also about understanding better what kind of projects would be genuinely welcome.

Hypothesis H_2a: Party networks alleviate severe asymmetric information problems inherent to any rural taxation system. Communities with greater party presence carry a heavier tax burden than communities where party members are few and far between.

The Party and State Autonomy

Local governments have to pay a price for the indispensable services from the party members in their jurisdictions. Ever since the party evolved from the pre-1949 conquest organization to the post-1949 government constituency, party members have been demanding privileges. This trend has intensified as the ideological basis of party membership has given way to more transactional motives for joining the party (Li et al. 2008). After all, granting material benefits to the rank-and-file is a more predictable political strategy for party leaders, compared to granting political influence. If the township level government loses the loyalty of local party members, the kind of principle-agent problems inherent to most tax systems appear, as party members can collude with each other and with the villagers. In short, local governments have an overwhelming interest in maintaining good relations with its party constituency, on which their power rests. Spoiling the relationship with the party members in its constituency would cut the local government off from the rural population. Even if there were no corruption or pork-barrel politics, party members have easier access to the authorities and will be better able to plead for exemptions from levies.

Hypothesis H_2b: The party functions as a patronage network, its members receive tax rebates.

Synthesis

On balance, we would expect the party to strengthen the extractive power of the Chinese state. A structural-functionalist argument suggests that local party members would be cut out of local governance in general and tax collection in particular, if they had outlived their usefulness. More practically speaking, the number of grassroots party members is limited: Normally, no more than 25 % of all village households include at least one party member. Therefore, as long as tax exemptions per party household are no more than four times the gains in village revenues that come with an additional party member, party members on the ground improve extractive power.

Hypothesis H₂: On balance, the presence of the Chinese Communist Party in the countryside increases the extractive power of the state.

Empirical Analysis

To assess the hypotheses on the micro-geography of extractive power, this section conducts hard tests: We test the two hypotheses with regard to two hierarchical levels, the county level and the township level. If the theory is correct, we should find the hypothesized curvilinear relationship not only for distances between households and townships but also for distances between households and counties. We would expect the theory to hold at the prefectural or provincial level, but the survey data used here does not allow to test this: The surveyed households are spread out throughout the counties included in the survey, but at higher levels, they are too clustered in space to be of any use. Fixed effects, a set of powerful control variables, and alternative outcome variables further demonstrate the results’ robustness.

Specification

The empirical specification for testing hypotheses H₁ and H₂ predicts extraction from a rural household as a non-linear function of physical distance from the county and township seats, as a linear function of the Communist Party’s presence in the village, and a function of a dummy indicating whether any household member belongs to the party. We expect per hypothesis 1 that levies first increase and then decrease as one moves away from the county and township center. Households in villages with greater party presence should pay higher levies. But households containing at least one party member pay fewer levies.

$$ Extraction=a+{b}_1*{distance}_{county}+{b}_2*{distance}_{county}^2+{b}_3*{distance}_{town}+{b}_4*{distance}_{town}^2+{b}_5* CCPpresence+{b}_6* CCPmembership+{b}_7* household\ controls+{b}_8* village\ controls+ county $$

To probe robustness, the outcome variable extraction is defined in four alternative ways.

(1) The sum of levies a household pays per year, logarithmic (calculation method below).

(2) Same as (1), but not logarithmic.

(3) Same as (1), but excluding labor contributions. This variable allows to test whether non-monetary contributions alone drive the results.

(4) A dummy variable whether a household reports paying anything at all.

The main variables of interest are as follows:

• Linear distance from the county seat, also as a squared term.

• Distance from the township seat.

• CCP _presence is the percentage of party members in the village population.

• CCP _membership is a dummy that equals1 if a household comprises at least one party member.

Compared to the cities, rural economic activities are relatively homogenous, but households near government centers may have different income sources. Household controls account for this:

• The area of the land cultivated by a household. This is a likely confounding factor, since village leaders use cultivated land as the official base for tax assessments.

• Household size, which can also be used as a criterion for assessing levies.

• Household income, even if it is not normally used for tax assessment in rural China.

We include the following as village controls:

• Population: In larger villages, economies of scale may lower costs of communal projects.

• Physical geography: Is the village located in a mountainous or hilly area?

• Village finances: Does the village have income from collective enterprises? Is it in debt? Did the village have any industries during the Mao period? Did it experience any kind of calamity (including bad weather)?

• Official designations: Is the village located in a county recognized as poor by Beijing? Or by the province? Is it located in a township recognized as poor by the province? Has the village been designated as an experimental spot?

• Connections: Is there any official in the county government (or above) who originates from this village? (About half the villages have such connections.)

• Political interest: Are people checking the documents at the village blackboard?

The baseline specification controls for county fixed effects. An alternative specification is non-logarithmic, to allow for linear interpretation, and without fixed effects, which could potentially introduce an imbalance into the analysis, as certain far distances occur only in large counties.

Data

We can take advantage of an exceptional data set, which contains detailed information on the micro-variation of state extractive capacity, as well as on physical and organizational distances. In regular intervals since 1988, an international consortium of researchers carries out the Chinese Household Income Project (CHIP) with the goal to assess income inequality.⁴ The survey is primarily a household survey, but its edition for rural areas is complemented by interviews with a village focus group of community leaders, which provide all the information needed for the village-level control variables. Due to political constraints, the questionnaires include a changing set of variables. For the purposes of testing our hypotheses on distance and state power, only the 2002 edition of the survey contains all the necessary information.

Questions about taxes, levies, fees, and especially labor services are politically sensitive at a time when the peasant burden is politically salient. There are narrow limits as to the information that enumerators are able to elicit on such an issue. Yet CHIP achieves just that and on a large scale from over 38,000 individuals. The survey records private household spending in exceptional detail, so that the questions about state extraction are inconspicuously embedded among many questions on household expenditures. Moreover, the survey matches income and expenditures, so that inconsistencies would be noted and corrected. The information on state extractive activity seems unusually accurate (Table 1).

Table 1 Descriptive statistics

From the household survey, we calculate the levy in three steps.⁵ First, we add up the following categories of household expenditures: payments to the village, payments to the village and township, fees for collective contracts, and money collected for rural education. Each category comprises cash payments as well as in-kind contributions, which have been turned into monetary values by the survey team. Second, we multiply the days of unpaid labor, as reported by the household, with the daily wage for labor, as reported by a village official (typically the party branch secretary, head of the village committee, or village accountant). Third, we add payments a household has made to buy its members out of labor contributions. Penalties, which would amount only to a very small fraction (1.2 %) of the peasant burden anyway, are excluded.

It is noteworthy that labor contributions have made their comeback in rural China. Although the data from the 2002 round of the CHIP have been widely used, this phenomenon has largely escaped scholarly attention. While 77 % of all household have not contributed unpaid labor, the remaining 23 % households have. Each household on average contributed 14.4 days of work, amounting to 47 % of the peasant burden. Additionally, 11 % of the peasant burden consists of payments made to buy one’s household out of labor contributions. If the discreet advocacy for labor contributions by a think tank of the Ministry of Water Resources (Li 2005) or examples of labor contributions in Hebei (Zhou 2012) are a good guide, projects where the rural population is drafted for public works are usually large-scale initiatives by county or higher-level authorities.

Since the distances involved in governing the extremely spacious province of Xinjiang are much longer than in ordinary provinces and risk to confound the coefficient estimates, we exclude the data of the 400 Xinjiang households. We exclude another 170 households from two counties that were rolling out the tax-for-fee reform at the time when the questionnaire was fielded. We are concerned that the questionnaire was perceived as an integral part of the tax-for-fee reform, prompting strategic answers. After these two adjustments, the data set contains observations for 8630 households in 436 villages in 112 counties in 21 provinces. Finally, to prevent drastic outliers in the ratio of party members in the population from unduly influencing the result, we set to “missing” 20 observations that are more than four standard deviations above the sample mean (2 villages). Taking into account the various problems of missing data, we ultimately can work with 7390 observations.

Results

As the regression coefficients and their very high statistical significance show, there is a strong and systematic relationship between levy extraction and physical distance (compare Table 2). Inspecting the coefficient estimates in the logarithmic baseline specification, controlling for a large number of household and village characteristics, and including county fixed effects, we see that the four coefficients on physical distance and the two variables on organizational distance all point in the predicted direction and are significant. From the coefficients, it follows that households at intermediate distances from the county seat and at intermediate distances from the township seats carry the heaviest burden.

Table 2 Coefficient estimates

Due to the non-linear model specification, the absolute size of the distance effects is not constant. Ignoring the quadratic term and holding all else constant, moving further away from the county seat by one kilometer is associated with an increase in levies by one percent. In other words, to compare with the effects of other variables, moving roughly 3 miles away from the county seat could offset the extra levies associated with cultivating an extra mu of land. Yet taking into account the quadratic term, the distance effect varies and is smallest close to the predicted maximum point of extraction at 30 miles from the county seat. ⁶

To illustrate that differences in extraction across small areas of territory are not merely ripples on the ocean but substantively important, we simulate what the coefficients imply for households in Yanggu, an unexceptional county in Shandong province that is only half the size of Luxembourg. Specifically, we use the equation in “Specification” to calculate the expected amount of levies that a household has to pay, depending on its geographic location (see Fig. 2). The calculation takes the coefficient estimates of the baseline model in Table 2 and combines them with the observed average values (over the entire sample) for all control variables, as well as physical distances as they occur in Yanggu. The horizontal layer at the bottom of Fig. 2 is a simple map of Yanggu, showing the jurisdictional borders of the county and townships as black dots. The mountains and valleys represent different levels of extraction. Extraction is about 50 % higher at the peaks than at the nadirs. Moreover, maximums and minimums are in close neighborhood to each other, with many steep slopes throughout the jurisdiction. In short, differences in extraction due to physical location are of utmost practical relevance.

Fig. 2

Uneven extraction in Yanggu county (simulation)

The results also mean that proximity poses a greater practical challenge than remoteness. Figure 2 highlights that remoteness is not an apparent obstacle to extraction in Yanggu, where most households live close to one of the townships. In most cases, including in the areas shown at the forefront of the graph, extraction is greatest at the mid-points between two townships. Yanggu is by no means unusual. According to the preferred specification of the model, a household would experience most extraction if located 30 miles from the county seat and 10 miles from the township seat. Only 14 % of households in the sample live in an area that is further away from the two centers of local power than the place of maximum extraction. These households, though disadvantaged for not having access to power, still experience less extraction, because the state’s infrastructural reach is fading in their areas.

The role of the Communist Party in the process of levy extraction is characterized by an implicit quid pro quo. Other things being equal, households including at least one party member on average contribute 15 % less in levies than households without a party member. But the existence of party networks also facilitates levy extraction: An increase in party membership by one percentage point is associated with a 1.8 % increase of levies from all villagers. Within normal parameters,⁷ this certainly is a good deal for the government. The bridge that party organizations build into the village communities improves extractive capacity, but comes at a price.

Conclusion

The reach of a state throughout its territory, measured in terms of extractive power, can be extremely uneven. The case of China suggests that this variation has a distinct spatial quality, related to the physical and organizational distance of a locality from local centers of government. In contrast to simple center-periphery approaches, which expect power to dissipate as one moves away from the centers, this paper finds that distance has a dualistic effect on power. State power depends not only on the ability of a state to exert pressure through its infrastructure but also on the vulnerability of the state to societal pressures. Moving away from local political centers, state infrastructure dissipates, but state autonomy increases. The state is most amenable to pressures that arise in its immediate neighborhood.

Therefore, a fundamental characteristic of the geography of state power is that the reach of the state is at its best at intermediate distances, where its infrastructure reaches well, yet where its autonomy is already great. Places close to the center and very remote places both would avoid resource extraction: one by co-opting the state and the other by escaping from it. The central challenge of state-builders, who try to project power across their territory, consists in penetrating society while remaining insulated from particularistic interests. State-builders seek to expand infrastructural power and state autonomy simultaneously. A powerful state not only permeates every corner of the realm but also preserves independence from the local society, which hosts it. Reversely, the strength of a state in a particular locality depends on the presence of state agents, as well as on theses agents’ autonomy.

The evidence dates from 2002, but the results indicate where rural China is headed in the future. According to optimists, the tax reform of 2006 may have solved the peasant burden problem for good, although there is no conclusive evidence for this yet. To the contrary, evidence emerges that imaginative local governments have found innovative ways to engage in fundraising within their jurisdictions.⁸ Meanwhile, the central government continues its routine to issue instructions urging cadres to solve the peasant burden problem.⁹ Political economists have documented Beijing’s puzzling inability to reduce the peasants’ burden in a sustainable way (Bernstein and Lü 2003). In some places, real estate deals are fundraising exercises. In other places, the new name of the old game is “single-issue/single-discussion funding,” which provides ample room for resource extraction. The seemingly innocuous “direct household subsidy” is also turning into a potent fundraising device, because it is coupled with contributions from below. Since this article has focused on the kind of informal non-tax revenues, which appear likely to remain intact after the reform, it can shed light on the most likely new forms of extraction mechanics, now that formal taxes have been abolished. In particular, the very voluminous resource extraction through labor contributions has not been the focus of recent reform efforts and is therefore likely to persist. The resource extraction landscape that we have described here seems durable and not so easily overcome.

The Chinese case demonstrates the usefulness of the subnational approach in order to illuminate fundamental questions of state power, especially when taken to an extremely fine-grained level. The analysis also juxtaposes the advantages of physical geography versus human geography. While both varieties of the geographic approach provide parsimonious explanations, the physical, topographic explanation has an appeal of universal relevance. Topography is an issue in states across world regions and is a deep cause affecting state-building outcomes. By contrast, human geography is closely tied to country-specific culture, history, and institutions. Today, at a time when only few countries are ruled by Leninist-style parties, it is an empirical question whether other organizations step in to bridge the distance between the rulers and the ruled.

Painting the contours of state power in the countryside goes straight to the heart of contemporary state-building challenges and reveals strengths and weaknesses of the Chinese state. Thanks to rapid improvements in transportation infrastructure, physical distance limits the reach of the state in all but the most remote corners of its realm. For the sake of extractive activity, China has practically solved the problem of projecting physical power throughout its territory. Instead, a lack of state autonomy has now become the Achilles heel of local government. The reluctance to extract resources in the immediate environs of local authorities, which arbitrarily aggravates fiscal pressures on people living further afield, reflects that local governments are unduly influenced by their surroundings. Similarly, private interests at the grassroots of the ruling party capture local governments. As organizational infrastructure, the party successfully bridges social distances, solves asymmetric information problems, and boosts extraction. But the party at least in part relies on patronage to buy loyalty. This may be a good deal for increasing tax extraction, but a lack of even-handedness undermines legitimacy and by implication can weaken the state. In short, the Chinese state has successfully bridged physical and organizational distances, but lost its “proper” distance. Not a lack of infrastructure but a lack of autonomy from particularistic interests is the central challenge facing local Chinese state-builders today.