Farmland Dispute Prevention: The Role of Land Titling, Social Capital and Household Capability

Abstract

Disputes over farmland constitute an important challenge for tenure security, economic growth and social stability. Land titling is a theoretically promising policy instrument that can enhance tenure security and reduce the occurrence of farmland disputes in the developing world. However, the impact of land titling on the occurrence of disputes has been found to be highly conditional. Empirical evidence on this issue has been surprisingly limited and has often lacked the consideration of a specific context. In this study, whether land titling affects the incidence of farmland disputes in the context of China is investigated, focusing particularly on the interaction between land titling and social capital and household capability. Both the probit and CLL models are applied, using household data entailing a total of 3693 samples located in rural China. The results show that (1) land titling in China reduces the incidence of farmland disputes, and that (2) the analysis based on interaction terms indicates that households who are disadvantaged in social capital and household capability are more likely to experience fewer disputes. Implications for the government and organizations are as follows: (1) the titling programme should seriously consider the current tenure arrangement. Confirmation of current tenure is recommended over the establishment of a new system to avoid possible institutional shopping and overlapping claims; and (2) land titling, if well implemented, is recommended to redress local power asymmetry and to help vulnerable groups defend their property.

1. Introduction

Disputes over farmland in rural areas cause social and economic problems. Disputes are a common phenomenon in daily life, especially in developing countries which feature weak tenure and governance. Although disputes themselves are not always damaging in the long run, those without proper resolutions may eventually lead to violent conflicts, causing physical and mental damages, or even social turbulence [1,2,3,4]. Economists consider disputes to be the source or consequence of tenure insecurity and stress the efficiency loss that occurs once a piece of farmland falls into pending dispute because the land then becomes difficult to cultivate or trade [5,6]. Moreover, the resolution of farmland disputes also requires financial and labour input which could be devoted to more productive uses [7,8,9].

Fundamentally, farmland disputes represent conflicts arising over the control of land tenure [10]. Theoretically, the absence or a weak farmland tenure system has been blamed as the underlying reason for the incidence of farmland disputes [7]. If the tenure was well defined and enforced, there would be no disputes at all [11]. Land titling is an important policy instrument to establish a clear tenure regime and is expected to reduce farmland disputes. However, there has been a growing consensus that the dispute reduction effect of land titling, which is expected to produce better protection and higher tenure security, is highly condition dependent. One of the conditions lies in imperfect implementation. For instance, the high cost, non-inclusion and discriminative identification in programme implementation may exclude vulnerable groups and restrict the effects of such titling programmes [7,12]. Even when the implementation level is acceptable, land titling may still result in more disputes in countries where customary tenure prevails. For example, in Honduras, customary tenure has been properly defined and enforced at the community level. Titling can solve a problem that does not exist but complicates the tenure system [13]. A dual tenure system may lead to the institutional shopping and overlapping claims and result in more disputes [13,14,15]. Moreover, some governments have been unable to ensure subsequent legal and administrative enforcement, so that efforts at land titling have ultimately failed [16,17,18].

Compared to these compelling theoretical insights, the empirical evidence has surprisingly been limited. Adding to studies that found that land titling has direct and positive effect [19,20,21], Dower and Pfutz contributed that land titling decreased violent conflicts only when the municipal authorities’ discretionary power in land allocation was restricted [4]. These empirical works, while emphasizing the role of land titling, neglected the fact that the protection of land tenure can also be provided by alternative entities such as communities and households themselves. Informal approaches are preferred and applied in dispute resolution more often than formal approaches [22]. Research on social capital has already revealed its important role in protecting private property [23,24] and preventing deviant behaviours in water use [25]. When households themselves exist in a context of weak tenure and governance, it has been found that that are more active in tenure protection, for example by building salient boundaries, allocating defensive labour and planting trees to declare secure claims and prevent possible violations and disputes [26,27,28]. Considering all these alternative mechanisms, the effect of land titling on farmland disputes may therefore depend on the efficacy of protective measures by communities and households implement themselves.

The objective of this study is to investigate the effect of land titling on farmland disputes in the context of rural China, especially focusing on the interaction of titling with social capital and household capability. The land titling programme in China offers an appropriate opportunity for studying this issue for several reasons. First, this programme is the widest in the world in terms of the area and population it covers, concerning approximately one million km² of land farmed by 200 million rural households. Second, the features of the Chinese programme, such as the ways in which households are included, the fact that it is free, and its high issuance rate (96% of the beneficiaries have already received formal certificates), suggest that implementation failure is not an issue of concern. Third, this programme is not a redistribution programme but a confirmation of the current tenure arrangement based on existing documents. Thus, it is expected that no parallel tenure arrangement will be created in China, making the institutional shopping perspective less significant there than other places.

To disentangle causality, a rational behavioural framework is developed to explain the incidence of disputes. The underlying idea is that disputes are possibly determined not only by the ways in which governments and courts define land tenure and enforce dispute resolution, but also by the efforts of the community and households themselves. This idea is inspired by Barzel’s insight of economic property right that “a person’s right over a commodity depend on factors such as his valuation of the commodity, his and the state’s expenditure to protect the commodity, and the expenditure made others to capture it” [29]. Therefore, the exogenous interventions leading to improvements in legal tenure may or may not explain the incidence of disputes, depending on the inherent level of protection that is provided through the other two approaches. The data used for empirical analysis are based on an integrative nationwide database1, and access has been authorized for academic purposes. This paper may contribute to the current discussion on land titling and farmland disputes from three perspective: First, current empirical work has been rather limited and has mainly investigated the direct effect of land titling on farmland disputes [4,19,20,21]. The inclusion of social capital and household capability in this research deepens the understanding of the indirect and heterogeneous effects of land titling. Second, as stressed above, the emergence of land titling in China, the fact that it is free, and its confirmatory features constitute prerequisites for the success of the programme, and hold important policy implications for future relevant programmes across the globe. Third, the issue of land titling has been rarely discussed in the context of rural China. An exception lies in the work by Krul [30], who investigated the disputes arising during programme implementation. Unlike Krul’s, our analysis focuses on the comparison between people who have received certificates and those who have not.

The rest of the paper is organized as follows: Section 2 describes the tenure structure, policy background and theoretical framework linking land titling and relevant disputes over land tenure; Section 3 addresses the dataset and estimation strategy; Section 4 provides the estimation results and the interpretation of the main findings; Section 5 draws conclusions and implications.

2. Institutional Background and Theoretical Framework

2.1. Farmland Tenure Reform in China

In the year 1978, China began to establish the household contract responsibility system, structuring the rural land tenure system nationwide around the idea of the village collective holding land ownership and the household holding “contract rights” to that land (a local term used to mean land use rights); under this system, households have been able to gain access to land use rights through contracting with the village committee at no cost. Since 1978, these contracts have experienced two rounds, the first one lasting 15 years (1983–1997) and the second 30 years (1998–20272). Recently, in 2019, the central government announced an automatic 30-year extension beyond the second round, and farmland tenure is now regarded as a quasi-permanent right.

Despite the increasing security of land tenure across time, another key feature of land contracting is that it has created binding relationships between specific households and specific plots. The initial land distribution in the 1980s was based on the egalitarian ideology that every household that qualified as a member of the village (hukou system) shared the collectively owned land. Moreover, the amount of share depended on the family size. The size of families being in constant flux because of births and deaths, farmland was administratively redistributed every 3 or 5 years at the village level in the first few years. While maintaining the egalitarianism, periodic redistribution, however, caused tenure insecurity and ineffectiveness in agricultural production and farmland markets [31,32]. Hence, the redistribution was legally forbidden under the Rural Land Contract Law (2003), and the allocation of farmland became dependent on the developing rental market.

While legal security increased under the law, a large number of rural households still lacked official tenure documents before the land titling programme, whether contracts or certificates. A survey conducted in six provinces in rural China showed that the proportion of households holding contracts hardly reached 20% in 2000 and that the proportion was even lower in 2008; the proportion of households holding certificates was the same as that of households holding contracts in 2000, and that proportion grew slightly between 2000 and 2008 but never reached more than 30% [33]. The fact that households did not have any documents that contained key information has been regarded as one of the main reasons why households’ land tenure has been violated and restricted to issues to justice [34].

2.2. Farmland Disputes and Land Titling in China

After the abolishment of agricultural taxes and fees in the early 2000s, farmland disputes arose at the heart of rural social and legal affairs. A survey conducted by China Central Television (CCTV) and Renmin University of China in 2004 showed that land disputes in rural areas accounted for approximately 60~70% of all the complaints, whether filed in writing or by phone [35]. Another study found that disputes over farmland constituted the largest portion of reported grievances from 2000 to 2009 [22]. By reviewing the legal verdicts delivered from 2016 to 2020, a more recent survey found that disputes over farmland still ranked first in the list of all rural land-related cases [36].

A snapshot of the land tenure reform and incidence of farmland disputes demonstrates that improvements in legal tenure regimes have not necessarily reduced the incidence of disputes. Although land tenure has become more secure under the law, disputes have remained a key issue in rural areas. There are still gaps between the legal enhancement and de facto protection, i.e., households still need legal proof to support their claim to land.

After years of pilot programmes in several provinces since 2008, the Chinese government launched an inclusive rural land titling programme in 2013. As of 2020, according to the minister of agriculture and rural affairs (MARA), official certificates have been issued for 96% of the country’s farmland. According to land titling rules and guidance book produced by MARA, the implementation of the titling has been managed at the county level and exercised at the village level by experts, trained technicians, villagers and cadres. The procedures have been as follows: (1) The first phase is programme preparation. Technicians first need to assemble farmland contracts, contract registers, and household information to create the initial database including tables and maps. (2) During the second phase, the technicians check the information included in the tables pertains to one household at a time to obtain confirmation after the resolution of overlapping claims (if there were any) and develop the database. (3) The third step entails technical measurements. Technicians measure the plots’ position and area in the field with professional equipment and inscribe these measurements onto the cadastral map3. (4) The fourth step involves publicising and revising. The tables and maps are publicised in a bulletin. When dissents and disputes require a resolution, a revised version of the bulletins are publicised when there is no longer any disagreement. The resolution of dissents and disputes requires that the parties of the disputes, village cadres, technicians and traditional authorities meet. The entire database is then delivered to the county government. (5) The fifth step entails the issuing of the official certificate. The entire database is then transformed into a standardised cadastre and the official certificate is printed based on this cadastral information and issued at the household level. The certificate includes the names of both household members and information about the plot (area, location, contours, and the four limits).

The land titling programme in China has several unique features that are rarely found anywhere else in the world. First, the programme is by no means a legal adjustment or redistribution scheme; it is a formal confirmation of the current tenure arrangement. The official guidance has stressed the fundamental principle that titling should be based on existing contracts or other historical records and that no redistribution or expropriation is allowed. This principle is quite important for this because abundant research in Africa has suggested the existence of multiple land tenure regimes after titling programmes have taken place created sources of disputes [6,10,18]. Second, the process of dispute resolution and the series of cadastral publication has guaranteed that registered land tenure does not lead to conflicting claims and is formally and informally recognised. Third, land titling in China is inclusive and free of charge at any stage of the process, which ensures the participation of all farmers and maintains the programme as an exogeneous intervention.

2.3. Theoretical Consideration

Framework to address tenure protection and farmland disputes. To determine the impact of land titling on the incidence of farmland disputes, it is necessary to explain how a dispute emerges in the first place. Usually, disputes arise after a violation has taken place [37,38], such as a tort, breach, expropriation, etc. Within a rational framework, people violate a rule only when it is profitable to them to do so [39]; that is, a dispute will not arise if the expected cost exceeds the expected benefits of the dispute.

The incidence of farmland disputes can be reduced by enhancing tenure protection because potential violators would then not benefit as much from the dispute. There are three possible approaches for a household facing a violation: legal action, social capital or household efforts (Figure 1). These three approaches are quite different in the ways in which they are delivered. At the country level, legal enforcement is a public good and is equally distributed among all citizens. In contrast, social capital only protects people included in a particular group, and the power of social capital varies across villages. Household members’ personal efforts represent an investment in protection and varies across households depending their wealth and capital. Hence, without an improvement in the legal procedure, the protection of land tenure is to remain unequally distributed across different levels of social capital and household capability. Therefore, it is possible that exogeneous improvements will be less effective for households that have enjoyed greater protection.

Land titling and farmland disputes. Land titling is important to prevent disputes over farmland because it increases the probability that households win in legal procedures. More specifically, according to Term 67, Article 6 of the Civil Procedure Law of China, a disputant undertakes the burden of submitting evidence related to his or her claims. Households holding formal certificates that can produce low-cost and conclusive evidence to meet the burden of proof in a dispute are likely to win a lawsuit, which in turn, increases the expected cost of potential violators and decreases the incidence of farmland disputes [19].

Social capital and farmland disputes. Farmland disputes are civil disputes, and their prevention depends not only on legal procedures, but also on a households’ access to informal procedures and their individual capacity to defend themselves. Social capital is defined as a collective asset in the form of shared norms, values, beliefs, networks and social relations that facilitate governance and collective action [40,41]. There are potentially two channels through which social capital can influence land disputes. First, members of the same village share and respect the knowledge of tenure arrangement. Violating someone else’s land tenure is disgraceful and the violator will be marginalized within the social network and lose future cooperative benefits [42]. Second, in the case of disputes between villagers and outsiders, villagers will stand together because of reciprocity principles, which can produce credible threats against violators and protect property from being disputed [23].

Household capability and farmland disputes. Developing the capacity of households to protect their tenure is another way to prevent land from being in disputes. Researchers found that individuals may invest in trees (often less profitable) and fences or allocate defensive labours to defend the land and guarantee that they occupy the land; there strategies may also keep their tenure more secure when formal tenure and governance are weak [13,43]. In this case, a wealthier household may be able to invest more to protect farmland from potential violations and disputes.

The heterogeneous effect of land titling. Land titling, as an exogenous policy instrument, interacts with households’ inherent social capital and household capability, producing a heterogeneous effect on farmland disputes. As mentioned above, the strength of social capital varies across villages, which suggests that some villages can offer adequate protection from disputes to their members while others cannot. Therefore, the dispute reduction from land titling may be more effective in those villages that cannot offer protection from disputes. Similar to household capability, the ability for households to invest in protection may have already led to fewer disputes; thus, these households find themselves less favoured by land titling than those households that have not been able to shield themselves from land disputes.

3. Empirical Model and Estimation Strategy

3.1. Model Setting

To explain the incidence of farmland disputes, a binary empirical model is set, applying standard probit to estimate the effect of land titling on farmland disputes. The specification is as follows:

$$\mathit{d}\mathit{s}\mathit{p}{\mathit{t}}_{\mathit{i}\mathit{j}}={\mathsf{\beta }}_{\mathbf{0}}+{\mathsf{\beta }}_{\mathbf{1}}\mathit{C}\mathit{e}\mathit{r}\mathit{t}{\mathit{i}}_{\mathit{i}\mathit{j}}+{\mathsf{\beta }}_{\mathbf{2}}\mathit{S}\mathit{o}{\mathit{c}}_{\mathit{j}}+{\mathsf{\beta }}_{\mathbf{3}}\mathit{C}\mathit{o}{\mathit{n}}_{\mathit{i}\mathit{j}}+{\mathsf{\beta }}_{\mathbf{4}}{\mathit{X}}_{\mathit{i}\mathit{j}}+{\mathsf{\mu }}_{\mathit{i}\mathit{j}}$$

(1)

$\mathit{d}\mathit{s}\mathit{p}{\mathit{t}}_{\mathit{i}\mathit{j}}$ on the left is the experience of farmland disputes for household i in village j. $\mathit{C}\mathit{e}\mathit{r}\mathit{t}{\mathit{i}}_{\mathit{i}\mathit{j}}$ and $\mathit{C}\mathit{o}{\mathit{n}}_{\mathit{i}\mathit{j}}$ respectively denote the status of land titling and the household defensive capability of household i in village j. $\mathit{S}\mathit{o}{\mathit{c}}_{\mathit{j}}$ denotes the village social capital at the collective level. ${\mathit{X}}_{\mathit{i}\mathit{j}}$ denotes the other control variables, including other attributes of the village, households, and farmland that potentially affect the incidence of disputes, and ${\mathsf{\mu }}_{\mathit{i}\mathit{j}}$ is the residual in the equation.

To further test the interaction between land titling, social capital and households’ capability to defend themselves, cross terms are introduced into Equation (1), and the specification is as follows:

$$\mathit{d}\mathit{s}\mathit{p}{\mathit{t}}_{\mathit{i}\mathit{j}}={\mathsf{\delta }}_{\mathbf{0}}+{\mathsf{\delta }}_{\mathbf{1}}\mathit{C}\mathit{e}\mathit{r}\mathit{t}{\mathit{i}}_{\mathit{i}\mathit{j}}+{\mathsf{\delta }}_{\mathbf{2}}\mathit{S}\mathit{o}{\mathit{c}}_{\mathit{j}}+{\mathsf{\delta }}_{\mathbf{3}}\mathit{C}\mathit{e}\mathit{r}\mathit{t}{\mathit{i}}_{\mathit{i}\mathit{j}}\ast \mathit{S}\mathit{o}{\mathit{c}}_{\mathit{j}}+{\mathsf{\delta }}_{\mathbf{4}}\mathit{C}\mathit{o}{\mathit{n}}_{\mathit{i}\mathit{j}}+{\mathsf{\delta }}_{\mathbf{5}}{\mathit{X}}_{\mathit{i}\mathit{j}}+{\mathsf{\epsilon }}_{\mathit{i}\mathit{j}}$$

(2)

$$\mathit{d}\mathit{s}\mathit{p}{\mathit{t}}_{\mathit{i}\mathit{j}}={\mathsf{\gamma }}_{\mathbf{0}}+{\mathsf{\gamma }}_{\mathbf{1}}\mathit{C}\mathit{e}\mathit{r}\mathit{t}{\mathit{i}}_{\mathit{i}\mathit{j}}+{\mathsf{\gamma }}_{\mathbf{2}}\mathit{C}\mathit{o}{\mathit{n}}_{\mathit{i}\mathit{j}}+{\mathsf{\gamma }}_{\mathbf{3}}\mathit{C}\mathit{e}\mathit{r}\mathit{t}{\mathit{i}}_{\mathit{i}\mathit{j}}\ast \mathit{C}\mathit{o}{\mathit{n}}_{\mathit{i}\mathit{j}}+{\mathsf{\gamma }}_{\mathbf{4}}\mathit{S}\mathit{o}{\mathit{c}}_{\mathit{j}}+{\mathsf{\gamma }}_{\mathbf{5}}{\mathit{X}}_{\mathit{i}\mathit{j}}+{\mathsf{\sigma }}_{\mathit{i}\mathit{j}}$$

(3)

Noting that the frequency of farmland disputes in our sample is 3.3%, the estimation may thus face a rare event bias: the effect of independent variables may be underestimated. This is because the nature of MLE in estimators as probit or logit will systematically overestimate the left tail of the distribution of independent variable equaled to 1 when rare events exist [44]. A possible solution to this brought by Chen suggests that the bias can be partly eliminated by applying the complementary log-log model (CLL) [45]. CLL assumes an extreme value distribution rather than a normal distribution, as in the probit model, and the cumulative distribution is left-skewed because negative (dspt = 0) samples are significantly larger than positive (dspt = 1) samples. Therefore, in CLL, the estimation of the probability of positive and negative events is asymmetric and the former will be overestimated, which corresponds to the underestimation of rare event.

3.2. Data and Variable Definition

The data used for empirical analysis are derived from an integrated database on the remote platform created by Zhejiang University. Household level data are based on the 2017 China Family Database (CFD) of Zhejiang University and the 2017 China Household Finance Survey (CHFS) produced by Southwestern University of Finance and Economics, while the data at the community level are based on the CFD, China Community Governance Survey (CCGS)produced by Southwestern University of Finance and Economics and China Grassroots Governance Survey (CGGS) produced by Nanjing Audit University. Stratified sampling and PPS (probability-proportional-to-size) methods were used to select representative counties, villages and households. The sampling considered the demographic, economic and geographic status to guarantee representativeness. A total of 5440 households were asked about their experience of farmland disputes. Out of these households, 1747 were asked only about disputes over their rented plots. Titling and other information pertaining to these plots were unknown and could not be connected to information about the households operating these plots. Therefore, the total sample used in empirical analysis included 3693 households, covering 29 provinces and 493 villages (Figure 2).

Table 1. Definitions and descriptions of variables.

Variables	Definition	Means	Std. Dev
Dependent variable
Farmland disputes	=1 if the household experienced farmland dispute at least once in 2016; =0 otherwise	0.033	0.179
Tenure protection
Land titling	=1 if the household holds an official farmland certificate; =0 otherwise	0.639	0.480
Moderate variables
Social capital	Average amount of gift money in the village (10 thousand yuan)	0.182	0.155
Household capability	Household expenditure per capita in 2016 (10 thousand yuan)	1.870	2.560
Land characteristics
Land fragmentation	Number of plots of the contracted farmland	4.637	4.931
Land value (WTA)	Willingness to accept farmland sales (ten thousand/mu)	3.064	11.588
Household characteristics
Age	Age of household head	55.251	13.615
Gender	=1 if gender of household head is male; =0 otherwise	0.850	0.357
Household size	Numbers of family members	3.263	1.638
Education	Education level of household head	8.243	1.879
Village Characteristics
Political stability	Number of years current cadres have been in office (party secretary or committee head)	7.722	6.872
Land value (rent)	Average land rent in the village (ten thousand/mu)	0.061	0.157
Village economics	Disposable income per capita in 2016 (ten thousand)	0.845	0.593
Wealth inequality	The deviation of household expenditure in village	1.871	1.774
Land redistribution experience	Lapse of time since the last land redistribution	18.529	6.403
Regional dummy
East China	=1 if the household/village locates in east China; =0 otherwise	0.391	0.488
Middle China	=1 if the household/village locates in middle China; =0 otherwise	0.384	0.487
West China	=1 if the household/village locates in west China; =0 otherwise	0.224	0.417

displays the definition and descriptive statistics about the dependent variables and the four groups of control variables. The variables used for empirical analysis included several types of interests. In addition to key information about the experience of disputes, land titling, social capital and household capability, control variables drawn from of households’ characteristics, land plots, communities and regional dummies were also included.

(1) Farmland disputes. The dependent variable of interest, farmland disputes, is a dummy variable representing the households’ actual experience of disputes over their contracted land in self-reported terms. Self-reported disputes consist of both manifested and latent, low-intensity disputes. The latter rarely appears in administrative statistics. Therefore, the measurement for self-reported disputes includes more information than the administrative data. Types of disputes include those over land boundaries, land transactions and land between neighbours, villagers, village committees or trade partners. The value equals 1 if the sample household has experienced any of the aforementioned dispute types, or 0 if otherwise.

(2) Land titling. The independent variable is “the holding of official farmland certificates”. In previous empirical studies, land certificates have been a credible and robust indicator of land titling and have been proven to have a significant impact on many economic outcomes [27,33,46]. In our case, the certificate is qualified as the indicator of land titling because it is the final outcome of this land titling that should be issued at the very end of the programme, suggesting that households holding certificates have gone through all procedural steps in the programme.

(3) Social capital. Social capital measures the strength of individual and community social networks. Indicators for social capital vary across studies, and their validity differs across social contexts [47]. In rural China, giving out money on certain social occasions, such as births, marriages, funerals and festivals, is an important social activity and the amount of money given out is highly related to the density of social networks. Thus, the gift money represents a popular indicator in the context of rural China [48,49]. To consider the collective level of social capital, we used the money gifted per household as a proxy.

(4) Household capability. A household’s capacity to invest in tenure protection largely depends on the status of family wealth. The indicator for this capacity is thus represented by the household annual per capita consumption in a year5 in that the consumption can identify the household’s liquidity available for investment.

(5) Control variables. To further eliminate the possible effects of other factors, four groups of control variables were included: land characteristics, household characteristics, village characteristics and regional dummies. Land characteristics are used to capture farmland’s physical and economic attributes. Land fragmentation captures the density of boundaries and complexity of adjacency, which result in a higher risk of boundary disputes [50]. The willingness to accept (WTA) for land sale prices6 constitutes a proxy for land value and it is expected that land with a higher value causes the probability of disputes to increase [8].

Household characteristics mainly capture attributes relating to the response to and management of disputes. Elders may be more cared for and respected and thus experience fewer disputes, but it is possible that they are vulnerable to the risk of violations. Therefore, the effect of age is uncertain. Likewise, the experience of disputes by female heads of household cannot be clearly predicted. Households gathering more members imply that they are credible threats of potential violation; thus, these households may be less likely to experience disputes. Households in which members are educated members are accustomed to facing policies and laws and thus are more apt at handling potential violations through formal approaches, and are less likely to experience disputes.

Village characteristics consist of political stability, land value (at the village level), village economics and wealth inequality. The more politically stable a village is, as indicated by cadres holding office for years, more peaceful the governance environment will be; moreover, there will be fewer grievances, leading to a lower probability of disputes [19]. In contrast, wealth inequality and the absence of land redistribution may trigger demand for equalitarianism and lead to more grievances [51]. The village’s economic status may have positive or negative effects. On the one hand, people in poorer villages may be more intensively pressured in their livelihoods and experience more conflicts over resources. On the other hand, people in poorer villages have fewer resources to compete over, thus the incidence of disputes may be reduced [8].

Three regional dummies were introduced based on geographic location and economic development to control for other unobservable factors that may potentially affect the incidence of farmland disputes.

To initially assess the effect of land titling, a t test was applied to check the significant difference in dispute experience between the titled and untitled groups. The results are shown in

Table 2. T test of the dispute probability between groups with and without titles.

	Total Sample		Low Social Capital		Middle Social Capital		High Social Capital
	Not Titled	Titled	Not Titled	Titled	Not Titled	Titled	Not Titled	Titled
Pro(Dispute)	4.20	2.80	4.35	2.40	4.31	3.04	3.95	2.96
Difference	1.40 **		1.95 **		1.27		0.99
	Total sample		Low consumption		Low consumption		High consumption
	Not titled	titled	Not titled	titled	Not titled	titled	Not titled	titled
Pro(Dispute)	4.20	2.80	3.45	1.96	5.47	3.36	3.63	3.05
Difference	1.40 **		1.49 *		2.11 **		0.58

Note: *, ** represent the significance at 10%, 5% respectively.

. Across the entire sample, the ratios of households that had experienced at least one dispute in the two groups was 4.2% and 2.8%. The difference was 1.4% and statistically significant which suggests that the untitled group faced a higher risk of farmland disputes than the titled group. Moreover, the sample was divided by trisection according to the scale of social capital and household capability to test the effect of land titling on farmland disputes in each subgroup. The results show that the difference in farmland dispute frequency was significant in groups with low social capital, and low- to middle-household capability, while no significant difference was found in the other three subgroups.

4. Estimation Results and Robustness Check

4.1. Results

The estimation results of both probit and CLL models are given in

Table 3. Empirical regression results of both probit and CLL models.

Independent Variables	Probit1	CLL1	Probit2	CLL2	Probit3	CLL3
Land titling (T)	−0.260 **	−0.579 **	−0.697 ***	−1.528 ***	−0.451 ***	−0.901 ***
	0.126	0.281	0.269	0.574	0.158	0.330
Moderate variables
Social capital (S)	−1.054	−2.406	−4.020 *	−9.134 *	−1.037	−2.325
	0.868	2.058	2.156	4.931	0.865	2.037
Household capability (H)	−0.007	−0.016	−0.003	−0.004	−0.103*	−0.172
	0.020	0.042	0.020	0.041	0.054	0.110
Cross Term
T*S	/	/	3.896 *	8.873 *	/	/
			2.273	5.188
T*H	/	/	/	/	0.122 **	0.204 *
					0.059	0.120
Land Characteristics
Land fragmentation	0.017	0.036 *	0.015	0.031	0.017	0.036 *
	0.011	0.021	0.011	0.020	0.011	0.021
Land value (WTA)	0.017 *	0.030 *	0.017*	0.031 *	0.017 *	0.031 *
	0.010	0.018	0.010	0.018	0.010	0.018
Household Characteristics
Age	−0.001	−0.004	−0.001	−0.004	−0.002	−0.005
	0.006	0.013	0.006	0.013	0.006	0.013
Gender	0.435 *	0.917	0.450 *	0.958	0.436 *	0.894
	0.259	0.638	0.262	0.637	0.258	0.638
Household size	0.036	0.062	0.037	0.059	0.034	0.058
	0.039	0.086	0.040	0.087	0.039	0.086
Education	0.009	0.015	0.007	0.009	0.008	0.015
	0.021	0.050	0.021	0.052	0.021	0.050
Village Characteristics
Political stability	−0.014	−0.030	−0.014	−0.029	−0.014	−0.030
	0.010	0.023	0.010	0.023	0.010	0.023
Land value (rent)	0.723	1.491	0.655	1.308	0.761	1.604
	0.924	1.958	0.878	1.782	0.922	1.929
Village economics	0.021	0.056	0.030	0.070	0.026	0.067
	0.108	0.259	0.108	0.257	0.109	0.261
Wealth inequality	0.063 **	0.118 **	0.060 **	0.108 **	0.065 **	0.117 **
	0.028	0.049	0.028	0.048	0.029	0.050
Land redistribution experience	0.046 ***	0.095 ***	0.042 **	0.086 ***	0.046 ***	0.095 ***
	0.017	0.032	0.017	0.033	0.017	0.032
Regional dummy	YES	YES	YES	YES	YES	YES
Constant	−3.168 ***	−6.007 ***	−2.786 ***	−5.100 ***	−3.008 ***	−5.717 ***
	0.571	1.213	0.638	1.347	0.567	1.202

Note: *, **, *** represent the significance at 10%, 5%, 1%, respectively.

. The results in column 1 show that the holding of land certificates had a negative and significant effect on the experience of households’ farmland disputes. Likewise, the results from CLL (column 2) were consistent with those from Probit model, indicating that rare event bias did not disturb the estimation of the significance.

Our results confirmed that land titling reduces the occurrence of farmland disputes, which is consistent with the results found in previous empirical studies [4,19,20,21]. However, some critics argued that the effect of land titling on disputes and tenure security is strongly affected by the local governance environment and the level of investment on the part of households [6]; thus, further testing was conducted to investigate whether local social capital and households’ capability to defend themselves can change the negative effect of land titling on the occurrence of farmland disputes.

Columns 3 and 5 display the results of specifications (2) and (3) with the cross terms. After the introduction of cross terms, the effect land titling remained significant and negative. Parameters of cross term (T*S) and (T*H) were positive and significant, which restricts the effect of land titling, suggesting that the effect of land titling on farmland disputes is not equally distributed among heterogenous groups and more likely to affect households in villages where social capital is lower and that are less wealthy.

Regarding the control variables, WTA had a significant positive effect on farmland disputes, indicating that the occurrence of disputes increases with land value. The same trend was observed in the average land rent in village variable, though this effect was not significant. In terms of village characteristics, households in villages where wealth inequality was greater and where farmland was not redistributed for a longer duration were exposed to a higher probability of farmland disputes, which suggests that the erosion of equalitarian environment increases the incidence of farmland disputes.

4.2. Robustness Check

In this section, two robustness tests were applied to check the reliability of the results obtained above (the results are given in

Table 4. The estimation results in the group regression.

Variables	Social Capital			Household Consumption
Groups	High	Middle	Low	High	Middle	Low
Land titling	−0.054	−0.133	−0.382 **	−0.020	−0.336 **	−0.287 *
	(0. 153)	(0.148)	(0.161)	(0.160)	(0.146)	(0.170)
Other variables	Controlled

Note: *, ** represent the significance at 10%, 5% respectively.

and

Table 5. The effect of land titling on farmland disputes with alternative social capital indicators.

Variables	Probit (Vote Rate)	Cloglog	Probit (Norms)	Cloglog
Land titling	−2.979 ***	−6.446 ***	−0.996 **	−2.443 **
	(0.854)	(2.044)	(0.397)	(1.060)
Social capital	−0.014 ***	−0.026 ***	−0.438 **	−1.000 **
	(0.004)	(0.007)	(0.210)	(0.453)
Cross terms (C*S)	0.032 ***	0.068 ***	0.914 **	2.272 **
	(0.009)	(0.022)	(0.451)	(1.114)
Other variables	Controlled

Note: **, *** represent the significance at 5%, 1% respectively.

7). In addition to the introduction of the cross term, group regression is another common strategy to test heterogeneous effects [52]. Specifically, the total sample was divided by trisection according to the scale of social capital household consumption and a probit was run for each subgroup. The results in Table 4 show that land titling had a significant effect on subgroups with low social capital and low- and middle-household consumption, while it had no effect on the subgroups. The results are consistent with those obtained in Section 4.1.

In the second test, the complexity of social capital was noticed and two alternative indicators were used: the voting rate and the existence of written norms. Specifically, the indicator for social capital was replaced by the voting rate and the existence of written village norms. The results are displayed in Table 5. The results are basically consistent with those obtained above.

5. Conclusions and Policy Implications

Disputes over farmland may result in economic inefficiency and social instability [4,6]. Studies have revealed that the incidence of disputes can be reduced by defining and enforcing land tenure [11]. Land titling is an important policy instrument to establish a clear tenure regime and is expected to reduce farmland disputes [8]. However, the programme has been seriously criticised because of its imperfect implementation and it may complicate the inherent order supported by customary tenure and create more disputes [7,12,18]. This paper empirically investigated the incidence of farmland disputes in the context of China, especially focusing on the role of land titling, social capital and household capability. The main results and their implications for governments and international organizations are as follows:

(1) Land titling reduces the probability of farmland disputes through the implementation of well-defined land tenure and certificates through legal procedures. This finding is consistent with some recent studies conducted in Ethiopia, the Republic of Congo, Mexico and Bengal, where the incidence of farmland disputes has been reduced after land titling was implemented [4,19,20,21]. However, this finding is inconsistent with the results of Benjaminsen et al. [14] and Ege [15], who found that more disputes were observed after land titling was implemented in Niger and Ethiopia. These scholars attributed the disputes to the competition between formal land titles and customary tenure. Therefore, an implication of this comparison for governments and international organisations is that the titling programme should seriously consider current tenure arrangements. It is recommended that the current tenure system be confirmed rather than a new system established to avoid possible institutional shopping and overlapping claims.

(2) The reducing effect is heterogeneously distributed across different groups. The results of the cross terms indicate that households who are less advantaged in social capital and household capability are more likely to benefit from land titling in terms of a lesser occurrence of disputes. This finding has rarely been addressed empirically despite the consensus that land titling is specific to informal social structure and individual attributes [6,18]. Implications for governments are that land titling, if well implemented, is recommended to redress local power asymmetry and to assist the vulnerable groups in defending their properties. More studies are needed to investigate how formal protections such as land titling interact with inherent social forces and how the interaction affects social stability and individual welfare.

(3) In addition, we find that the environment of equalitarianism is important to the incidence of farmland disputes and the maintenance of social stability. This result can be explained by the political theory that grievances over social and economic status motivate one’s participation in disputes [19]. The equalitarianism issue should be carefully addressed by governments in the face of today’s challenges over economic growth and resource inequality.

The study has limitations. Unfortunately, we are not able to obtain the data related to the violators. A dispute is considered to be a specific relationship between at least two parties, the offender and the offended, thus, some the offender’s characteristics (e.g., risk preference, livelihood etc.) and the endogenous relationship between the two parties (e.g., power asymmetry, social relations etc.) may impact the incidence of farmland disputes. The nature of this impact is a point that can be examined in future studies.