Nonmarketization Bargaining and Actual Compensation Level for Land Requisition: A Qualitative Comparative Analysis of China’s Land Requisition Conﬂict Events

: In the land requisition market in China, two very di ﬀ erent compensation levels for land requisition can be seen in the real world: one is the highly rigid o ﬃ cial compensation level for land requisition and the other is a fuzzy actual compensation level for land requisition. In order to uncover the determinants of the actual compensation level for land requisition in China, this paper adopts Qualitative Comparative Analysis (QCA) to analyze the causal relationship between nonmarketization bargaining factors, like land-losing farmers’ bargaining ability, bargaining strategy, external intervention, etc., and the actual compensation level for land requisition by taking 70 land requisition conﬂict events occurring from 2002 to 2017 as the point of departure. The results of the empirical analysis show that if land-losing farmers have a relatively strong bargaining ability, forgo a radical bargaining strategy, and use a relatively gentle bargaining strategy instead, they can e ﬀ ectively force local governments to make concessions and compromises on the compensation level for land requisition. This paper not only enriches the existing research on the structure of social power, but it also has signiﬁcance for the ongoing reform of the land requisition system.


Introduction
In the field of land requisition in China there are many "Chinese stories" worthy of deep academic excavation, particularly in terms of the compensation for land requisition, since the compensation level for land requisition in official documents differs significantly from the compensation level awarded in reality. Under the existing land requisition system, local governments have the power to unilaterally set the compensation level for land requisition. If the land requisition policy at the formal institutional level is strictly enforced, land-losing farmers can only passively accept the official compensation level for land requisition and have no right to negotiate with local governments [1]. Under the land finance model of "producing wealth with lands, and maintaining lands with wealth," local governments have a strong motive to minimize the official compensation level for land requisition and increase revenues by means of "low-price land requisition in farmland conversion market and high-price land transfer in urban primary land market" [2]. Consequently, the official compensation level for land requisition is very low and land-losing farmers cannot share in the achievements of industrialization and urbanization. Coupled with the simple method of compensation and resettlement for land requisition and the lack of a social security system, many land-losing farmers cannot obtain effective guarantees for gaining a long-term livelihood and so become vulnerable. Guojun Chai et al. have carried out an empirical investigation into the compensation level for regional land requisition nationwide by using the Data of Financial Investigation on Chinese Families 2011 [3]. Their research results prove that the monetary compensation obtained by 75% of land-losing families only met the minimum standard set in the government's policies. Not only is this unfair, but it also goes against the efficiency principle to a certain degree [4]. At the same time, too low official compensation level for land requisition will also bring high transaction costs including negotiations, information collections and contracting for land development projects. Yani Lai et al. studied the transaction costs of urban village reconstruction projects in Shenzhen from 2004 to 2009, and the results showed that the official compensation level for land requisition resulted in many time-consuming transactions and hindered the redevelopment of urban villages [5].
The official compensation level for land requisition is obviously on the low side and most land-losing farmers cannot share in the dividends of urbanization [6]. However, some land-losing farmers do obtain higher compensation for their land than the official compensation level by various means, like negotiation, confrontation, and struggle. In urban villages and in some rural-urban fringe zones, the official compensation level for land requisition is not always executed effectively [7]. News media frequently share stories of land requisition leading some villagers to make a great fortune overnight. Therefore, some scholars worry that overcompensating land-losing farmers will make them "raise the compensation price in virtue of the immovable land position during urbanization process, and independently enjoy all benefits of urbanization," and so become "highly-organized, reactionary, and decadent rentier groups with strong action ability, and a nightmare in the process of China's modernization" [8].
In terms of the compensation level for land requisition, "under-compensation" and "over-compensation" exist simultaneously and have attracted the attention of numerous scholars. For example, Eddie Hui et al. and Qiuxiang Li et al. consider that the implementation of compensation for land requisition in reality is similar to the bargaining that takes place between governments and land-losing farmers over land value capture, and the actual compensation level for land requisition that land-losing farmers could obtain depends on the gaming and balancing of stakeholders [9,10]. Local governments, through the power of unilateral pricing, grasp "inherent advantages," while land-losing farmers, through their low bargaining position, can only obtain compensation according to official compensation levels for land requisition; few farmers with relatively strong bargaining abilities can obtain compensation for land requisition that is actually considered to be priced out of the market [11]. Among developing countries in the transitional period from a planned economy to a market economy, such as Laos and Ethiopia, similar phenomena are also ubiquitous [12,13]. Bargaining game theory could potentially completely explain the reasons for the huge difference between the official compensation level for land requisition and the actual compensation level for land requisition, but it lacks the data of practical scenarios to support its research conclusions. Relevant literature documents that analyze land requisition problems by using the game theory model do not describe the complexity and diversity of negotiation on land requisition. Compared with the existing research, this paper makes contributions in the following three areas: (1) This paper introduces the concept of nonmarketization bargaining and analyzes the influences of relevant factors like bargaining ability, bargaining strategy, external intervention, etc. on actual compensation levels for land requisition; (2) This paper carries out a meticulous decomposition and presents a panoramic display of land-losing farmers' nonmarketization bargaining behavior while discussing the social power structure problem hidden behind it in depth; (3) At present, the research on land requisition conflict mostly takes the form of single-case analyses or qualitative interpretations, which is helpful for exploitative research, but less suitable for explanatory research. This paper has collected and sorted 70 land requisition conflict events and hopes to draw more valuable research conclusions by using Qualitative Comparative Analysis (QCA). Researchers concerned with the reform of China's land requisition system can identify the influencing factors of land value increments distribution and the casual pathways among these factors from the research results of this paper. This paper is organized as follows. The second section is about Research Background; the third section is about Research Approach; the fourth section is about Research Design; the fifth section is about Empirical Analysis Results; the last section draws conclusions and puts forward the future research directions.

Research Background: Comparison of the Official Compensation Level for Land Requisition and the Actual Compensation Level for Land Requisition
Land requisition is called "exercise of supreme land right" in American law "forced purchase" in British law, and "land acquisition" or "land purchase" in Japanese law. In modern society, it is very common for countries to impose the requisition of private (collective) land within the scope authorized by the constitution and in accordance with the provisions of the constitution and the law due to the needs of public interests, regardless of whether the owner of the land agrees or not. The root of the land requisitions system is the legitimate deprivation of private property rights, thus becoming the breakthrough point for public power to intervene in the private sphere. In order to balance public interests and private interests, especially to prevent excessive infringement of public power upon private interests, countries have imposed a strict standard on land requisitions under the guidance of the principle of "legitimacy". Meanwhile, developed market economies also actively introduce market tools in land requisitions through institutional innovation, which as a result reduces the transaction cost of land requisitions. For example, the United States, Italy, Hong Kong protect private property rights and realize fair distribution of land value-added benefits by Transferable Development Rights (TDR) and Purchasable Development Rights (PDR) [14][15][16][17].
For the reform of China's land requisition system, the core issue is also to make the compensation level for land requisition reasonable for land-losing farmers. The Land Administration Law, revised in 1998, changed the previous land requisition system of "administrative examination and approval and joint negotiation" into a compulsory land requisition system of "executing land requisition examination and approval first, and then making announcements to land-losing farmers." Under the "examination and approval first, and then announcement" system, local governments are endowed with a buyer's monopoly and land-losing farmers can only passively accept their unilateral pricing [18]. Concerning unilateral pricing, local governments set the compensation for land requisition according to its original use, and there are regulations on the upper limit of what constitutes maximum compensation. As a result, the compensation level for land requisition has been on the low side for a long time. It cannot sufficiently guarantee land-losing farmers' future working and living, and it cannot establish a long-term mechanism for farmers to share in a land increment benefit. This has inevitably led to the discontent, doubt, and confrontation of land-losing farmers and the rise of collective visit events and land requisition conflict events all over the country [19].
In the face of social conflict, local governments have increasingly begun to introduce negotiation mechanisms into land requisition procedures to jointly determine compensation and resettlement plans for land requisition with land-losing farmers [20]. Therefore, two very different compensation levels for land requisition can be seen in the real world: one is the highly rigid official compensation level for land requisition and the other is a fuzzy actual compensation level for land requisition. Under the actual compensation level for land requisition scenario, land-losing farmers can possibly obtain compensation for land requisition that is higher than the official compensation level through negotiation with local government. However, if they avoid direct confrontation or fail in their confrontation with local government, they might only obtain the level of compensation for land requisition that was previously determined by the local government.

Official Compensation Level for Land Requisition with Regional Integrated Land Prices Prevailing
In accordance with the provisions of Article 47 of the existing Land Administration Law, compensation for land requisition (such as cultivated lands, for example) mainly includes the following: (1) land compensation fee: 6 to 10 times the average annual output value in the three years before the requisition of the cultivated lands; (2) resettlement subsidy: calculated as per the cost of the agricultural population to resettle. The level of resettlement subsidy for every agricultural population to resettle is 4 to 6 times the average annual output value in the three years before the requisition of the cultivated lands; moreover, the resettlement subsidy for every hectare of expropriated cultivated lands is, at the most, no more than 15 times the average annual output value in the three years before the requisition of the cultivated lands; and (3) compensation for attachments to and young seedlings on the land: compensation is generally made according to the actually evaluated value. Article 47 of the Land Administration Law states: "If it still could not make resettled farmers maintain their original living standard, then the resettlement subsidy may be increased with the approval of the people's governments of provinces, autonomous regions, and cities directly under the central government. However, the sum of the land compensation fee and resettlement subsidy shall not be more than 30 times the average annual output value in the three years before the requisition of the lands." Nevertheless, since land-losing farmers' original standards of living are strongly heterogeneous, and during the concrete implementation of policies a uniform standard operation is lacking, this principle is difficult to truly embody in the land requisition policies of local governments.
Since the introduction of the concept of "annual output value" in the Regulations on Land Expropriation for State Construction promulgated by the State Council in 1982, China has always adopted an annual output value multiple method to determine land requisition compensation. It is a method to calculate a land compensation fee and resettlement subsidy by multiplying the cultivated lands' annual average output value in the three years before requisition. The quantity of land expropriated and the compensation multiple are the core elements determining the final compensation for land requisition. The annual output value multiple method was insufficient in the survey on land location factors, so the State Council issued the Decision of the State Council on Reform Deepening and Strict Land Management in October 2004. Since then, the annual output value multiple method has gradually been replaced by the integrated land price of the land requisition area method. The so-called integrated land price of the land requisition area method divides land requisition areas and measures and calculates the integrated compensation level for the land requisition based on the annual output value multiple method according to factors such as output value, land type, agricultural land level, per capita quantity of cultivated lands, land location, land supply-demand relationship, urban residents' minimum living security level, local economic development level, etc. Within a land requisition area, the local government will adopt the same compensation level for land requisition according to the integrated land price of the area, which prevents granting different compensation for similar lands, to a certain degree. Taking the city of Yueqing in Zhejiang Province as an example, Table 1 reflects the changes in recent years in the integrated land price (including land compensation fees and resettlement subsidies) of the land requisition areas. As shown in Table 1, the compensation level for land requisition has tended to rise year after year, and since January 2009 the compensation fee for land requisition in first-level areas has reached up to RMB140,000/mu. However, even within the same land requisition area, different parcels of land might have very distinctive differences in terms of factors like land type, location, water conservation facilities, traffic conditions, etc., and such differences directly affect their land revenues. The integrated land price of a land requisition area covers the abovementioned land parcel differences in their entirety with an integrated compensation price, and it nearly eliminates all the differences within a reasonable boundary and compensation limit of different land parcels in the implementation of concrete land requisition projects. Since the concept of compensation as per the land's original use has not changed, it is difficult to fundamentally reverse the pattern of the official compensation level for land requisition being far lower than the market value by implementing the "regional integrated land price" policy.
Through survey sampling in Suqian, Nantong, and Kunshan, Peixin Zhu et al. have discovered that the official compensation level for land requisition has been consistently rising, but that the compensation fee for land requisition actually obtained by land-losing farmers is only around 3-16% of the land transfer price in the primary market, while the net income obtained by local governments is up to around 75% [21]. (During the process of urban land first-level development, the local government carried out land development and supporting facility construction to increase land value. Some land-expropriated farmers in some areas were also provided with employment opportunities as compensation at the time of land acquisition. Because of difficulties in data acquisition, Peixin Zhu et al. did not deduct the land development costs of the local government and the costs of employment opportunities provided for land-expropriated farmers when calculating the net income of land supply, resulting in the greater net income from land supply calculated by Peixin Zhu et al. than the actual net income from land supply.) For example, in terms of residential lands, the compensation fee for land requisition accounts for only 4-10% of the residential land transfer price. In terms of commercial service lands, the compensation fee for land requisition accounts for only 2-15% of the commercial service land transfer price. The distribution of land increment revenue between local governments and land-losing farmers is severely unbalanced and this actually deprives farmers of their rights, like land development rights, etc., which makes them lose the opportunity to use land resources to increase their property income. If the low level of official compensation for land requisition is not adjusted at the proper time, this will surely spark the resistance of land-losing farmers, which will possibly incite social conflict.

Uncertain Compensation Level for Land Requisition
If the compensation level for land requisition that land-losing farmers could actually obtain is deemed as the selection result after repeated managing of different stakeholders, then even under the system of "strong officials and weak people" land-losing farmers will possibly strive for an economic benefit greater than the official compensation level by resisting land requisition. Hong Zheng has analyzed the land requisition conflict between local governments and land-losing farmers by constructing a dynamic game model. The result of his game model shows that within the framework of the existing land requisition system the official compensation level for land requisition is not always the same and that land-losing farmers still have opportunities to strive for appropriate local government concessions in terms of their compensation level [22]. Hui Wang et al. have systematically analyzed the relationship between land requisition negotiation and the resulting compensation for land requisition by using survey sampling data on land requisition from 12 Chinese cities in 2009 [23]. Using this information, they provided direct empirical evidence for the result of this dynamic game model. Table 2 preliminarily summarizes the land requisition investigation data. Among a total of 773 copies of national effective land requisition samples, 364 farmer households (47.09%) negotiated with local governments during the land requisition process; moreover, 342 of them (44.24%) ultimately obtained actual compensation for land requisition that exceeded the official compensation level. That is to say, nearly 94% of the 364 farmer households participating in the negotiation on land requisition achieved greater economic benefits than if they had not negotiated. Therefore, although land-losing farmers do not have the right to participate in land requisition procedures and can often only passively accept the integrated land price of land requisition areas as determined by local governments, if they have the opportunity to negotiate with local governments then they can force local governments to make concessions. Negotiation on land requisition could effectively break the rigidity of official compensation levels for land requisition and make the actual compensation level for land requisition shift in a direction that is favorable to farmers. But how do land-losing farmers get the opportunity to negotiate with local governments? At what cost and sacrifice does the negotiation opportunity come? How many economic benefits could they obtain if they got the opportunity to negotiate on land requisition? The field investigation conducted by Lei Yang et al. helps us to begin to answer these questions. In the case study on L City and G Development Zone of Hubei Province, they discovered that when confronted with land requisition and demolition conflict, local governments tend to solve the conflict of interests arising from land requisition and demolition by means of strategic management in order to simultaneously consider the financial revenue from lands and the stability of the social order [24]. That is to say, by looking at different subjects of resistance, they implement different means of governance. For example, local governments may reason with some farmers, give proper warning to some key households, adequately raise the compensation level for the farmers continuing to resist, execute special governance under the leadership of organized visiting people, etc.
Wei Cheng has discovered similar phenomena in Chongqing's Yubei District. The Yubei District Demolition Office has five different compensation levels for land requisition and demolition, does not publicize the compensation levels to each demolished household, and "settles" different demolished households by strategic means of crushing them one by one [25]. Consequently, the actual compensation for land-losing farmers differs according to their different resistance behaviors (including whether they resist, the resistance strategies employed, the resistance organization upon which they rely for support, etc.). The farmers who do not have the opportunity to negotiate with local governments might only obtain compensation at the official level, or they might possibly obtain higher-level compensation by relying on other resisting farmers. The farmers who have the opportunity to negotiate with local governments could possibly be forced to accept compensation as per the original level for their failure to negotiate well or they might possibly succeed in resisting but have to pay considerable costs for doing so.
In sum, as the integrated land price of the land requisition area shows, the compensation available to land-losing farmers has been regulated in detail in the relevant documents of local governments at all levels. But in reality, the compensation level for land requisition is fuzzy. On the one hand, the farmers obtaining compensation for land requisition as per official compensation levels cannot fairly share the land increment revenue and this has led to the current problem of "under-compensation." On the other hand, in some regions of China, land-losing farmers can possibly obtain higher-level compensation for land requisition by means of negotiation, consultation, resistance, etc., and so the problem of "over-compensation" has been witnessed in some regions of China. For example, in individual regions of Guangdong Province, the old age security payment for the requisition of 1 mu of agricultural land is up to RMB1,000,000, while the average compensation level of Guangdong Province is around RMB10,000/mu [26]. In order to explore why the actual compensation level for land requisition differs so widely in reality, this paper has to discover what influential factors determine the actual compensation level for land requisition.

Research Approach
In a mature market economy society, transaction parties will negotiate as per marketization bargaining and in line with the principles of equality, free will, and fairness. If any transaction disputes occur, a judicial body will make public and fair judgments as an independent third party. However, in China's rural society, when land-losing farmers are unwilling to accept the compensation level for land requisition unilaterally priced by local governments and hope to share in the land increment revenue after rural collective lands enter into the urban construction land market, they usually choose to fight with local governments for land increment revenue by means of nonmarketization bargaining since they there is no space for bargaining under any formal system [27]. Nonmarketization bargaining is directly connected to informal governance based on human relations and the particularistic logic of a traditional acquaintance society, and it is concretely embodied as various non-standard bargaining methods which are legal but unreasonable, or are reasonable but illegal [28] Farmers who engage in such nonmarketization bargaining may be roughly divided into radical negotiants and non-radical bargainers. With respect to radical bargaining, land-losing farmers might choose relatively radical means of resistance, such as processions and demonstrations, collective petitions, laying siege to governments, assault, vandalism, looting, arson, blocking traffic, etc. Within China's management framework of "pressure-type stability maintenance," the central government subcontracts the responsibility of stability maintenance to local party and government institutions at all levels, layer by layer, in order to ensure the absolute stability of society. Under the system of high top-down political pressure, many local officials regard stability maintenance as an overwhelming political task [29]. If group incidents threatening social stability occur due to land requisition, local officials' political promotion will surely be severely and negatively affected [30]. For example, in the regions where "the one-vote veto system for petitions to Beijing" is implemented, local officials directly assuming responsibility will temporarily lose the opportunity to gain promotion. For local governments, this is a hidden cost of land requisition. With respect to non-radical bargaining, land-losing farmers will adopt circuitous, tolerable, and flexible weapons for resistance, such as availing themselves of every opportunity, obsession, closing ranks, risk sharing, benefit sharing, etc. [31], and this directly results in difficulties promoting land requisition projects, construction period delays, high development costs, etc., which is a kind of explicit land requisition cost [32].
Both the radical bargaining that brings about hidden land requisition costs and the non-radical bargaining that brings about explicit land requisition costs will possibly force local governments to make compromises and concessions and transfer some land increment revenues to land-losing farmers. The concrete proportion of land increment revenues to be transferred by local governments cannot be effectively defined in the form of a formal system, nor can it be said that a set of definite game rules to determine the concrete distribution proportion of land increment revenue is lacking. Practically, local governments will adopt means of strategic governance and determine how to distribute the land increment revenue according to land-losing farmers' bargaining ability and their degree of effort. Generally, the actual power structure between local governments and land-losing farmers, namely their balance of power, decides the actual proportion of land increment revenue distributed. Effectively, local governments create a public domain of power on purpose [33], and in this public domain, the actual compensation level for land requisition is not determined by a formal system, but it is determined by the two parties' bargaining ability in the public domain and their respective degrees of effort.
In order to verify this judgment, this paper needs to uncover the experiential facts that can be observed in the real world to carry out an empirical analysis. The land requisition conflict events occurring frequently all over the country have provided us with high-quality research materials. On the one hand, observing the processing results after the occurrence of land requisition conflict events enables us to know whether farmers' nonmarketization bargaining could loosen the rigidity of the official compensation level and give farmers the right to participate in the distribution of land increment revenues. On the other hand, land requisition conflict events result from centralized conflict breaking out between local governments and land-losing farmers. In the process of the conflict, the two parties attack and defend and completely reveal their respective action strategies and bargaining abilities. This could help us understand how the rigidity of official compensation levels is shaken and what factors could determine the proportion of land increment revenue distribution in the public domain.
If land-losing farmers hope to participate in the distribution of land increment revenue, they shall have sufficient power to force local governments to make concessions. This involves two problems. One is whether land-losing farmers can muster comparable bargaining power, and the other is how land-losing farmers use their own power to bargain with local governments. In addition, the occurrence of land requisition conflict events may very well attract the attention of external forces, such as the supportive reports of news media and the intervention of higher-level governments. Therefore, interventions from external forces may also significantly affect the actual land acquisition compensation level.
Comprehensively considering these factors, the concept of "nonmarketization bargaining" can be divided into three categories of influencing factors: "bargaining ability", "bargaining strategy", and "external intervention". Considering that almost no existing literature explains or analyzes China's land acquisition conflict events from the perspective of nonmarketization bargaining, this paper reveals the possible causal relationship between non-market bargaining factors and actual compensation level of land acquisition by comparing and summarizing land acquisition conflict events, rather than by directly proposing the specific research hypothesis of the influential effect of non-marketization bargaining factors.

Research Samples and Data Collection Methods
This paper examines 70 land requisition conflict events that occurred during 2002-2017. (This paper takes the information integrity of land requisition conflict events as the primary standard for sample selection. Therefore, the authors of this paper have excluded a large number of samples with missing conflict processes and conflict results, and only reserved research samples fully describing the causes and effects and complete process of land acquisition conflicts.) The 70 land requisition conflict events share the following basic features: (1) the expropriated lands are all lands collectively owned by farmers, including agricultural lands, rural collective construction lands, curtilages, urban village lands, etc.; (2) land-losing farmers' appeals emphasize economic benefits (this paper excludes appeals to social commonweal, public policy, codes of ethics, etc.); (3) a variety of areas are covered, including small towns, rural areas, medium-and large-scale cities, and each of the country's economic regions; and (4) the number of each conflict event reports varies from more than ten to dozens, so the conflict events are relatively representative and theoretically universal.
The research samples collected for this paper are extremely rich and diversified and were collected mainly through the following three channels. First, the China Core Newspapers Full-text Database at China National Knowledge Internet was used as a case database; a full-text retrieval with "requisition" or "land requisition" as the keyword(s) yielded 7867 news reports. Next, the authors of this paper read through these news reports and selected the land requisition conflict events reported by more than three media sources as research samples. However, when reporting on land requisition conflict events, news media usually focus on their concrete processes and do not carry out follow-up in-depth interviews or track the investigation, so the final processing results of all the land requisition conflict events cannot be known. Therefore, although large quantities of land requisition conflict events were sourced from media reports, the samples that could be used in this research were extremely limited. After comprehensive consideration, this paper finally selected 24 research samples sourced from media reports.
Second, this paper selected the China Court Judgments and Decisions Network as a case database, carried out a full-text retrieval with "requisition" or "land requisition" as the keyword(s), and obtained 11,384 first-instance or second-instance court judgment documents. When selecting the court judgment documents as research samples, the authors of this paper encountered a dilemma in that the event processing results were known, but the event occurrence processes were either not detailed or unknown. This paper could only select partial court judgment documents as basic materials and then supplement the occurrence process of the land requisition conflict events through texts found online, such as open letters, official documents, network interviews, pictures and video, forum posts on Blogger and Micro.Blog, etc. To meet the condition of guaranteeing the authenticity and credibility of relevant materials, this paper collected a total of 16 research samples sourced from the court judgment documents.
Third, this paper selected expert and scholarly case studies and field investigations on land requisition conflict events as a case database. The materials in this case database included published monographs, journal articles printed in domestic and international periodicals, as well as relevant doctoral and master's dissertations, etc., and they are generally characterized by their detailed contents, the abundance of their detail, definite results, and high credibility. Based on an in-depth reading of these materials, this paper collected a total of 30 research samples sourced from their case studies and field investigations.
Before encoding the attributes of land requisition conflict events, the authors of this paper compared and rechecked every case for more than five rounds in order to repeatedly confirm the details of the case variables. For the processing results of land requisition conflict events in particular, the authors of this paper used the materials obtained from different channels, to carry out cross validation. On the whole, the research data adopted in this paper demonstrate relatively high validity and reliability.

Research Strategy: Qualitative Comparative Analysis
This paper selects a qualitative comparative analysis method based on set theory and Boolean algebra instead of a statistical regression analysis based on the binary relation of independent and dependent variables. This decision was mainly based on the following considerations.
First, as was pointed out in Section 4.1, many influential factors leading to the final processing results of land requisition conflicts do not act on them independently but usually lead to the final results jointly by combined means. The QCA method does not pay attention to the net effect of a single explanatory variable on the explained variables but considers each explanatory variable as a combination condition in order to analyze different or similar results [34]. In light of the QCA method, this paper attempts to excavate the complex interactions among multiple influential factors from limited land requisition conflict events and to explore the methods and channels for different combination conditions to act on the final results. Comparatively, statistical regression analysis cannot deal with the interaction of multiple variables, which is of little help in exploratory research with unclear causality.
Second, in the process of collecting and sorting out data and materials in the early stages, the authors of this paper discovered the possibility of dissymmetry between the conditions (combinations) and the results, which cannot be treated by means of a statistical analysis based on correlativity. If it is assumed that a local government always considers condition Z when deciding whether to increase the compensation for land requisition and takes the existence of Z as the precondition for determining whether to increase the compensation for land requisition, then the inexistence of condition Z is not definitely a condition for the government to decide not to increase the compensation for land requisition. Comparatively, QCA carries out a logical inference based on necessary conditions and sufficient conditions and can treat the asymmetry between conditions (combinations) and results [35].
Third, in statistical regression analysis, endogeneity is a very difficult problem. By using QCA to clarify the necessary conditions and sufficient conditions between the conditions (combinations) and the results, this method solves the endogeneity to a certain extent. At the same time, it is not easy to address multiple negative influences of collinearity and autocorrelation negative effects.
Fourth, with regard to the limitations of the completeness and authenticity of the data and materials, although this paper collected and sorted 70 land requisition conflict events, this does not amount to a large sample level in terms of quantity, and therefore this paper could not obtain highly robust results through regression analysis. If case study method suitable for a small sample were adopted, it would be unable to treat 70-plus samples precisely and conveniently, especially when surveying multiple variables. The QCA method integrates a traditional case comparison method into a quantitative statistical method and integrates quantification logic into a qualitative inference system. The robustness of QCA results only depends on whether the samples include representative individuals, which is unrelated to sample quantity. The 70 land requisition conflict cases selected for this paper are of appropriate sample quantity, highlighting the applicability of the QCA method. More importantly, by virtue of the QCA method, this paper could deeply survey the representative individuals among the samples that were in turn based on a systematic and quantitative analysis of land requisition conflict events.
Fifth, QCA is characterized by its integration and transcendence of traditional research methods. The research of quantitative orientation is more concerned with the degree of "revealing" common problems. It pursues larger sample size, better statistical significance, larger population, objectivity and repeatability of research. However, the research of qualitative orientation emphasizes case study and a comprehensive and in-depth understanding of the research object. It attaches great importance to the role of individual researchers in the research process. QCA has absorbed the respective advantages of both quantitative research and qualitative research, attempting to open up an "intermediate manner" [34]. On the one hand, QCA's core logic is Set Theory, which focuses on how variables combine with each other and jointly influence a certain result, thus distinguishing it from traditional quantitative research. On the other hand, although QCA emphasizes individual cases, it still uses multiple methods falling within the scope of quantitative research in case coding and calculation. Therefore, QCA cannot be regarded as a pure qualitative research. To some extent, QCA is more like the embodiment of "mixed-methods research" advocated by social science circles in recent years.
QCA aims at establishing relationships of necessity and sufficiency between explanatory (influential) conditions and results. If X contains Y, then X will be regarded as a necessary condition of Y. However, if X is contained in Y, then X will be regarded as a sufficient condition of Y. In the QCA process, the operators commonly used include: "+", which refers to connecting different causal paths, and "*", which refers to simultaneous occurrence. Upper-case letters refer to conditions that occur and lower-case letters refer to conditions that do not occur. For example, "X 1 * X 2 + X 2 * X 3 = Y" means that two causal paths may lead to the occurrence of Y, wherein causal path X 1 * X 2 means that X 1 occurs and X 2 does not occur; while causal path X 2 * X 3 indicates that X 2 occurs and X 3 occurs simultaneously.
In order to measure the probability of necessity and sufficiency, QCA mainly adopts a consistency index to judge the relationship between explanatory conditions (combinations) and results. The calculation method is as shown in Equation (1): In Equation (1), indicates the membership of sample i in condition X, and indicates the membership of sample i in condition Y. The range of values in the consistency index is 0-1. If the consistency index is 1, X is completely a member of Y and a perfect asymmetric relationship occurs between them.
To some extent, the consistency degree is similar to the confidence level in regression analysis. As per the research of Carsten Schneider et al., if the necessity index is greater than or equal to 0.9, X may be regarded as a necessary condition of Y; and if the sufficient consistency index is greater than or equal to 0.8, X may be regarded as a sufficient condition of Y [36]. If the consistency index is satisfied, it will be possible to describe the explanatory power of the condition (combination) X for the result Y by measuring the coverage index, the calculation method for which is shown in Equation (2). The value scope of the coverage index is similarly 0-1, and it may be regarded as similar to the goodness of fit in regression analysis. If the coverage index is closer to 1, the condition (combination) X will have a greater explanatory power for the result Y.
Following concrete process analysis, this paper first uses the consistency and coverage indices to evaluate the necessity and sufficiency between a single condition and result. Next, this paper uses the consistency index to construct a truth table to exhibit the different combination relations between conditions (combinations) and results, uses a Boolean minimization algorithm to simplify the truth table, and finally obtains many causal paths leading to the occurrence of results as well as the common effects of different factors [37].

Variable Design and Assignment
This paper selects explanatory variables according to the theoretical analysis set out in the previous section and mainly surveys the influences of explanatory conditions like bargaining ability, bargaining strategy, external intervention, etc. on result variables. This paper adopts a 6-value assignment plan: the value of a variable score is determined as per the sequence of 0, 0.2, 0.4, 0.6, 0.8, and 1 in turn, wherein 0 indicates that the condition does not occur, 1 indicates that the condition occurs, and the other values are between the two circumstances. Table 3 reports the attribute codes and classified assignments of the explanatory variables and result variables selected in this paper. (According to the initial research plan, all the result variables and explanatory conditions in this paper adopt a 6-value assignment plan. However, limited by the research information in the land requisition conflict events, for many explanation conditions, the authors of this paper can only make a judgment on whether the condition occurs or not. For these explanatory conditions, this paper can only assign them to 0 or 1 respectively, and cannot describe various intermediate states in between. For those explanation conditions with more abundant research information, the authors of this paper uses the 6-value assignment plan to assign values according to the specific research information collected under different explanatory conditions. For example, for the mobilization network, only three situations can be distinguished, namely, "forming an obvious protester alliance", "Obtain the support of multiple lineal relatives", "Protesters fight a lone battle" according to the specific information provided by the land requisition conflict events. After repeated discussions and tradeoffs by the four authors of this paper, the above three situations are respectively assigned to 1, 0.4 and 0.)  (1) Result variable. The result variable of this paper is the "event processing result." In the land requisition conflict events selected for this paper, the resistance appeal of land-losing farmers is the economic benefit appeal, so this paper uses the degree of realizing the economic benefit appeal to quantify the "event processing result." Ronggui Huang et al. researched the influences of the central government's intervention and the national media's supportive reporting on demolition resistance results using the QCA method, and the assignment method they adopted for the variable "demolition resistance success" provided a direct empirical reference for this paper [38].
In combination with the land requisition conflict samples, this paper gives the following assignment to the variable "event processing result": "1" indicates that land-losing farmers succeed in bargaining and obtain compensation for land requisition that is higher than the official compensation level without incurring high costs and their economic benefit appeal is basically realized; "0.6" indicates that land-losing farmers have their economic benefit appeal basically satisfied but they also incur heavy costs, which is defined in this paper as "Pyrrhic victory I"; "0.4" indicates that in the resistance process the costs incurred by land-losing farmers are greater than the economic benefits attained and the result is similar to failure, which is defined as "Pyrrhic victory II" in this paper; "0" indicates a bargaining failure; specifically, land-losing farmers incur heavy costs and their economic benefit appeal is not satisfied.
(2) Bargaining ability. Measurement of the farmers' bargaining ability includes variables such as their mobilization network, organization level, village cadres' alliance [39], grassroots leaders [40], and clan power [41]. A mobilization network represents the social networking relationship used by land-losing farmers, and the value of this variable score is determined as per the sequence of "0", "0.4", and "1" in turn. With respect to organization level variable, "1" indicates high organization level, "0.6" indicates moderate organization level, and "0" indicates loose organization or basically no effective organization.
Village cadres' alliance indicates that the cadres of the Communist Party of China village's branch committee and villagers' autonomous administration committee support the villagers and negotiate with local governments and land use units on their behalf, wherein "1" indicates that village cadres form an alliance relation with land-losing farmers in the land requisition process, while "0" indicates that villagers' cadres support the local governments or do not form an obvious alliance with land-losing farmers.
The grassroots leaders variable indicates that there are activists and action leaders engaging in spontaneous organization and mobilization among land-losing farmers or stakeholders. In this paper, "1" indicates the land requisition resistance has the obvious participation of grassroots leaders, while "0" indicates no grassroots leaders are participating. Clan power indicates that in the land requisition resistance land-losing farmers organize and mobilize by relying on their clan relationships or seek support through their clan network. In this paper, "1" indicates that the land requisition resistance has the obvious intervention of clan power, while "0" indicates no clan power intervention.
(3) Bargaining strategy. Bargaining strategy represents various bargaining approaches taken by land-losing farmers in the concrete bargaining process after obtaining the corresponding bargaining ability, including a destructive strategy, violent resistance strategy, power balance strategy [42], tragic resistance strategy [43], and legal resistance strategy [44]. A destructive strategy indicates that the resistance behaviors disturb public order and endanger public security, such as fighting, property damage, looting, rioting, blocking public traffic, laying siege to governments, impacting party and government institutions, etc. Because they are difficult to measure, this paper does not distinguish between the degree of destruction caused by different resistance behaviors and simply determines that the value is "1" if a destructive strategy is used and the value is "0" if no destructive strategy is used.
A violent resistance strategy indicates whether land-losing farmers use violent behaviors in the land requisition conflict, and the value of this variable score is determined as per the sequence of "0", "0.6", "0.8", and "1" in turn.
The power balance strategy refers to land-losing farmers seeking the intervention and positive support of higher-level governments through visits to them, wherein "1" indicates a "visit to the central party and government institutions" in Beijing, like the State Council, the National.
People's Congress, the Chinese People's Political Consultative Conference, etc.; "0.6" indicates a "visit to provincial party and government institutions or same-level central ministries and commissions"; "0.2" indicates a "visit to prefecture city party and government institutions"; and "0" indicates a "visit to county-level party and government institutions" or no visits at all.
A tragic resistance strategy refers to land-losing farmers highlighting their weak status during the resistance process by means of collective complaint, collective kneeling, kneeling in front of officials, self-immolation, self-mutilation, etc. In this paper, "1" indicates that a tragic resistance strategy has been adopted in the resistance process and "0" indicates that a tragic resistance strategy has not been adopted. The legal resistance strategy refers to land-losing farmers seeking redress by judicial means. In order to avoid generalization while using the concept of "legal resistance," this paper makes it clear that whether there is a formal intervention of judicial organs (including circumstances in which the cases are not pursued after the cases are accepted by the courts) will be taken as the criterion for selection. In this paper, "1" indicates adopting legal resistance strategy and "0" indicates a legal resistance strategy has not been adopted.
(4) External intervention. There are two main explanatory variables for measuring external intervention: media support [45] and higher-level government intervention [46]. The assignment of media support simultaneously considers the source's attributes, attitudinal tendencies, and social influences. Concretely, "1" indicates the supportive reporting of central-level party newspapers and party periodicals like the CCTV, People's Daily, Xinhua News Agency (local branches), Xinhua Daily Telegraph, China Youth Daily, The Procuratorate Daily, Legal Daily, etc.; "0.8" indicates the supportive reporting of local media such as Jiefang Daily, Oriental Morning Post, Beijing News, Southern Metropolis Daily, Yangcheng Evening News, provincial satellite TV stations, etc.; "0.6" indicates supportive reporting provided by emerging media such as Blogger, Micro.blog, community forums, portal websites, etc., which could attract national network attention; and "0" indicates relevant reports sourced from other media or only attracting local network attention. Considering that a land requisition conflict event could possibly attract the simultaneous reporting of traditional media and emerging media, this paper executes assignment in line with the principle of giving priority to higher values.
With respect to the variable higher-level government intervention, this paper mainly surveys whether higher-level governments adopt intervention behaviors that are beneficial to land-losing farmers and executes assignment as per the level of government. Therefore, "1" indicates that the central party and government institutions like the State Council, the National People's Congress, the Chinese People's Political Consultative Conference, etc. intervene in conflict events by means of issuing documents, giving instructions, dispatching working groups, etc.; "0.6" indicates that provincial party and government institutions or same-level central ministries and commissions intervene in conflict events; "0.2" indicates that prefecture city party and government institutions intervene in conflict events; and "0" indicates that county-level party and government institutions intervene in conflict events or that higher-level government does not intervene in conflict events.

Explanation of Single Explanatory Conditions on Event Processing Results
This paper first analyzes the relationship between single explanatory conditions and result variables, the results of which are shown in Table 4. The necessary consistency of all 12 explanatory conditions in the three analysis dimensions, like bargaining ability, bargaining strategy, and external intervention, is less than 0.9, indicating that these explanatory conditions are not the necessary conditions that are sufficient to constitute event processing results. What is worthy of notice is that the necessary consistency index of the explanatory condition "mobilization network" is 0.815 and 0.840 respectively, which is slightly lower than the critical value of 0.9, so it could be regarded as an approximately necessary condition. The sufficiency analysis results of single explanatory conditions show that only the sufficient consistency index of clan power is greater than 0.8, indicating that with the exception of the explanatory condition clan power, all the other explanatory conditions are not sufficient conditions of the event processing results. Moreover, the sufficient coverage of clan power is only 0.031 and it has extremely weak explanatory power for event processing events. Therefore, with respect to single explanatory conditions, no explanatory condition could constitute both the necessary condition and the sufficient condition for event processing results. Here, this paper needs to transfer our research to the combination of each explanatory conditions. Note: Under the circumstance that the consistency index is less than initial critical value, the coverage index has no computational significance which is indicated by a line.

Explanation of a Single Analysis Dimension on Event Processing Results
In the next section, this paper analyzes the relation between a single analysis dimension and explanatory condition, and the results are shown in Table 5 Under the analysis dimension of bargaining ability, there are a total of three causal paths that could effectively explain event processing results; they are respectively: (A1) "narrow mobilization network, relatively low organization level, forming alliance with village cadres, and no intervention of grassroots leaders and clan power"; (A2) "broad mobilization network, relatively high organization level, not forming an alliance with village cadres, and having the intervention of grassroots leaders and clan power"; and (A3) "broad mobilization network, relatively high organization level, forming an alliance with village cadres, and no intervention of grassroots leaders and clan power." All three causal paths meet the sufficient consistency standard and the corresponding total coverage is 0.12. That is to say, the analysis dimension of bargaining ability alone could explain around 12% of the successful bargaining cases. The primary coverage of all three causal paths is 0.04, indicating that the three causal paths could explain 4% of the successful bargaining cases, respectively; their unique coverage is also 0.04, indicating that around 4% of the successful bargaining cases could only be explained independently by any one of the causal paths A1, A2, or A3. (Primary coverage indicates the number of cases that a certain causal path could explain. However, some cases may be jointly explained by multiple causal paths, so the unique coverage reflects the number of cases that could only be explained separately by a certain causal path.) The comparison of the three causal paths shows that mobilization network and organization level play important roles in the bargaining process: forming an alliance with village cadres helps successful bargaining, but the intervention of grassroots leaders and clan power has an obvious mutex relationship with the village cadres' alliance. Generally speaking, in a concrete land requisition conflict event, forming an alliance with village cadres will basically negate the functions of grassroots leaders and clan power. When village cadres oppose the farmers' resistance, grassroots leaders and clan power have the opportunity to lead the resistance.
Under the analysis dimension of bargaining strategy there are also three causal paths, which are: (B1) "Not using a destructive strategy, violent strategy, or tragic resistance strategy, but using a legal resistance strategy"; (B2) "Not using a destructive strategy or violent strategy, but using a power balance strategy and a legal resistance strategy"; (B3) "Using a destructive strategy, violent strategy, and power balance strategy, but not using a tragic resistance strategy or legal resistance strategy." Among the three causal paths, path B1 could explain around 32% of the successful bargaining cases and has the greatest explanatory power, whereas path B3 could explain only 4% of successful bargaining cases and has the weakest explanatory power. This indicates that it is easier for land-losing farmers to succeed in their resistance if they forgo a destructive and violent strategy in favor of a power balance and legal resistance strategy when selecting their bargaining strategy. Meanwhile, the functions of a tragic resistance strategy are not obvious and resistance behaviors like collective kneeling and collective complaint will possibly play only a miniscule role in reality.
Among the three analysis dimensions, external intervention cannot effectively explain the event processing results, while bargaining ability and bargaining strategy could only explain around 12% and 40% of the successful bargaining cases, respectively. Therefore, the following presents the condition combination with the strongest explanatory power from the interaction of different analysis dimensions.

Explanation of Different Analysis Dimension Interactions on Event Processing Results
Under a single explanatory condition and single analysis dimension, all possible causal paths have extremely limited explanatory power, so this section will mainly survey the combination effect of the three analysis dimensions-bargaining ability, bargaining strategy, and external intervention-and the results as shown in Table 6. Considering that the three analysis dimensions include 12 explanatory conditions in total, if parsimonious hypotheses are not introduced then the results obtained will be extremely complex and it will be impossible to carry out an effective theoretical extraction. Therefore, the results reported in Table 6 are parsimonious solution results and the information presented has certain drawbacks.
With respect to the combination effect of bargaining ability and bargaining strategy (the coverage is 0.48 and it could explain around 48% of successful resistance cases), it is possible to obtain eight causal paths. Due to space constraints, this paper will mainly report five causal paths with primary coverage greater than 0.1 (C1 to C5). Among the five paths, mobilization network occurs four times, indicating that if the resistance network mobilized by protesters is broad, it will be easier for them to succeed in their bargaining. However, in paths C2 and C5, the explanatory condition occurring simultaneously with broad mobilization network is relatively low levels of organization. Generally, in the resistance process, because of the difficulty of collective action, if the mobilization network is broad, it becomes more difficult to effectively organize action. When the resistance group is expanded, the organization of the resistance becomes loose. In path C3, the combination of the intervention of grassroots leaders and the non-destructive, non-violent, and non-visitation resistance strategy also leads to the successful bargaining. In terms of the resistance strategy, four paths require adopting a non-destructive strategy: two paths require not adopting a violent strategy and the other two require not adopting a tragic resistance strategy. In path C5, broad mobilization network, relatively low organization levels, no intervention of grassroots leaders, and adopting a power balance strategy could also lead to successful bargaining.
With respect to the combination effect of bargaining ability and external intervention (the coverage is 0.16 and it could explain around 16% of successful resistance cases), it is possible to obtain three causal paths: (D1) "Relatively high organization level, with the intervention of grassroots leaders and higher-level government"; (D2) "Relatively low organization level, with the intervention of grassroots leaders, media support, and higher-level government intervention"; (D3) "broad mobilization network, relatively low organization level, no intervention of grassroots leaders, and higher-level government intervention." Among the three causal paths, two explanatory conditions-media support and higher-level government intervention-do not play their roles simultaneously, but if one of the two conditions exists and cooperates with relatively strong bargaining ability, then land-losing farmers will have a higher possibility of succeeding in their bargaining. With path D1 as example, under the leadership of grassroots leaders, if a relatively high organization level is formed and further attracts the supportive intervention of high-level governments, then land-losing farmers will have a higher possibility of succeeding in bargaining.
With respect to the combination effect of bargaining strategy and external intervention (the coverage is 0.16 and it could explain around 16% of successful resistance cases), it is possible to obtain three causal paths: (E1) "Not using a destructive strategy, but using a power balance strategy, and no media support"; (E2) "Not using a destructive strategy, but using a legal resistance strategy, and higher-level government intervention"; (E3) "Not using a violent resistance strategy, not using a power balance strategy, not using a tragic resistance strategy, and having media support." Among the three causal paths, only path E3 contains the support of the media, while path E1 clearly does not require the media's support. It appears that relative to higher-level government intervention, media support could play an extremely limited substantial role. All three causal paths simultaneously require not using a destructive strategy, violent resistance strategy, or other relatively radical bargaining strategies. To summarize, (1) as seen from the analysis dimension of bargaining ability, having a broad mobilization network, high organization level, forming an alliance with village cadres, and benefitting from the intervention of grassroots leaders or clan power can all boost the success of bargaining to a certain degree. If land-losing farmers have a stronger bargaining ability, it will be easier for them to succeed in bargaining. (2) In the condition of not using bargaining strategies, like destructive and violent resistance, etc., it will be easier to succeed in bargaining. Using relatively gentle bargaining strategies, like power balance and legal resistance, etc., could obviously raise the probability of success in bargaining. (3) The combination effect of bargaining ability and bargaining strategy could effectively explain the success of nonmarketization bargaining. Under the circumstance that land-losing farmers have a relatively strong bargaining ability, forgoing radial bargaining strategies in favor of gentle bargaining strategies could effectively help them succeed in bargaining with local governments. (4) The single explanatory condition and single analysis dimension of external intervention could not effectively explain bargaining success. But if external intervention creates appropriate combination conditions with bargaining ability and bargaining strategy, it can elevate the possibility of successful bargaining to a certain degree. Xuefeng He observes that over the past decade, nearly all areas in China have raised their compensation level for land requisition year after year, but the number of conflicts surrounding land requisition have not dropped, and have instead increased [47]. The fundamental reason for this phenomenon is that the official compensation level for land requisition increases annually, but the adjustment is made by taking a piecemeal approach based on the existing land requisition system. Land-losing farmers still can only passively accept the unilateral pricing of local governments and their basic rights can never be effectively guaranteed, like the right to full disclosure about negotiations on land requisition, the right to object, the right to negotiate, etc. If they have disputes over land requisition with local governments, land-losing farmers cannot easily obtain redress from the judicial system. When the gates of the formal system cannot be forced open, land-losing farmers are forced to look for nonmarketization bargaining strategies to guarantee their right to participate in pricing in the process of land requisition. Therefore, to reform the land requisition system, an effort to guarantee the basic rights of land-losing farmers to participate in formal negotiation and consultation is necessary. That is to say, China should promote the marketization negotiation and bargaining between local governments and land-losing farmers by adopting many measures, like guaranteeing public participation, establishing a system of voluntary agreement-based transfer, developing independent third-party intermediary evaluation institutions, strengthening the validity of judicial supervision, etc.

Conclusions and Avenues for Future Research
This paper analyzes in detail 70 land requisition conflict events occurring from 2002-2017 by means of qualitative comparative analysis and thus discusses the causal relationship between land-losing farmers' nonmarketization bargaining behavior and the actual compensation level for land requisition. The basic research conclusions prove that local governments' unilateral pricing of the compensation level for land requisition is equivalent to depriving farmers of the opportunity to participate in the distribution of land increment revenue and induces a sizeable difference between the true market value of rural lands and the official compensation level for land requisition. Dissatisfied with the current distribution pattern of land increment revenue, land-losing farmers have successfully bargained with local governments through various behavioral methods of nonmarketization bargaining and loosened the rigidity of the official compensation level to a certain degree. If land-losing farmers have a relatively strong bargaining ability, forego radical bargaining strategies, and use relatively gentle bargaining strategies instead, they are able to effectively force local governments to make concessions and compromises in terms of the compensation level for land requisition. Accompanying nonmarketization bargaining behaviors are frequently occurring group incidents and relatively high operation costs, so an important direction for the reform of the land requisition system to take in future is to promote marketization negotiation and bargaining between local governments and land-losing farmers.
More concrete research conclusions include the following: (1) if bargaining ability is used to measure the ability of land-losing farmers to bargain with local governments, then broader mobilization networks, higher levels of organization, forming alliances with village cadres, and the intervention of grassroots leaders or clan power could all force local governments to make concessions and compromises in terms of the level of compensation granted for land requisition.
(2) If bargaining strategy is used to measure how land-losing farmers use their own bargaining ability, then if land-losing farmers adopt bargaining strategies that constitute serious opposition and conflict with local governments, it will be very difficult for them to succeed in bargaining. Instead, adopting relatively gentle resistance strategies within the scope allowed by law will increase their probability of successful resistance. (3) If land-losing farmers' bargaining ability and bargaining strategy are reasonably combined, then the combination of a relatively strong bargaining ability with a relatively gentle bargaining strategy could effectively increase the probability of success in bargaining with local governments. (4) If external intervention (including media support and higher-level government intervention) could create appropriate combination conditions with bargaining ability and bargaining strategy, it would be possible to increase the possibility of successful bargaining to a certain degree.
But if land-losing farmers only depend on external intervention, it will be very difficult for them to change the existing land increment revenue distribution pattern.
Although this paper has many limitations, they offer several possibilities for future research. First, this paper describes in-depth the nonmarketization bargaining behaviors of land-losing farmers, but due to a lack of research materials, it was unable to describe and analyze local governments. In order to analyze the land requisition benefit between land-losing farmers and local governments, the authors of this paper strive to take relevant explanatory conditions into the expanded analysis framework in the future like local government administrative officials, budgets, governance ability, etc. Second, when selecting the samples, this paper emphasizes the integrity of the sample materials and excludes numerous research samples on the basis of insufficient data, so this paper does not achieve a random sampling. This inevitably leads to errors of bias in the sample selection, so in future, the authors of this paper will further expand the source base from which this paper gather the research samples in order to strengthen the representativeness of the research conclusions. Another limitation of this study is the time span of the research. The samples cover 2002-2017 and during this period China's land requisition system did not change much, but new trends did arise, like the official compensation level for land requisition consistently increasing; the scope of land requisition began to decrease; the procedures for negotiation on land requisition were fixed in the form of a formal system in some regions, etc. Restricted by the research methods employed, our research design does not contain the time dimension of each explanatory condition, so it is impossible to deeply excavate the dynamic mechanism between explanatory conditions and result variables. In future research, it will be feasible to use larger sample data to carry out time measurements and analysis and then execute mutual corroboration with the results obtained from the case comparison. Finally, the description of the external intervention dimension in this paper is relatively superficial. Distinguishing between the support of traditional state media and that of emerging Internet-based media and distinguishing between the methods and channels of the intervention behaviors adopted by all levels of government could help us to understand Chinese stories of conflict governance more profoundly.