How do trust and property security influence household contributions to public goods?

How do trust and property security influence household contributions topublic goods?The case of the sloping land conversion program in China

Qin TU a,b,⁎, Arthur P.J. MOL c,d, Lei ZHANG c,d, Ruerd RUBEN e

a Institute of World Economics & Politics, Chinese Academy of Social Sciences, Chinab Department of Development Economics, Wageningen University, The Netherlandsc Department of Environmental Policy, Wageningen University, The Netherlandsd School of Environment and Natural Resources, Renmin University, Chinae Centre for International Development Issues, Radboud University Nijmegen, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history:Received 18 March 2010Received in revised form 25 July 2011Accepted 26 July 2011Available online 3 August 2011

Trust and property rights are generally considered to influence farmers' behavior regardingresource use and environmental management. Previous studies show that higher trust levelsmay enhance contributions to public goods. This paper investigates how trust and (land)property rights security influence the provision of one concrete public good: land protectionthrough the Sloping Land Conservation Program in China. The analysis is based on householdsurvey data from Ningxia Autonomous Region in China. From our questionnaire two trustfactors are derived and distinguished, using factor analysis: general trust and kinship trust.Farm households are less likely to contribute to public goods when they perceive more secureland rights, but trust has mixed effects on public goods. The results show that general trust andkinship trust may rely on two opposite effects for influencing public goods provision. On theone hand, high levels of general trust may directly enhance people's willingness to providecontributions to public goods (by reduced likelihood to reconvert forest land) when farmersare aware of the positive environmental effects of the program, that's the public goods effect.On the other hand, general trust may also make it more likely that people invest more in theirown private goods to pursue their own welfare (a more likely reconversion of forest land toarable land), that's the private goods effect. The final outcome depends on the size anddirection of both effects. Compared to general trust, kinship trust is more inward-looking andself- or group-interested compared to more reciprocal general trust. Thus, unlike general trust,kinship trust may have no significant public goods effect on the provisioning of public goods.

© 2011 Elsevier Inc. All rights reserved.

Keywords:General trustKinship trustLand property rightPublic goodsLand conversionChina

1. Introduction

Chinese policy makers are increasingly focusing attention on the consequences of rapid economic growth on environmentaland natural resource quality. In the previous (2006–2010) and current (2011–2015) Five Year Plans the protection of the urbanand rural environment receive significant attention, with major measures to manage erosion, desertification and rural waterconservation. The institutions involved in the protection and provision of these public goods in China are to some degree differentfrom the institutions that became involved in public goods provision in western countries in the 1970s. In this paper we focus on

China Economic Review 22 (2011) 499–511

⁎ Corresponding author. Tel.: +86 010 85195354; fax: +86 010 65126105.E-mail addresses: [email protected], [email protected] (Q. Tu).

1043-951X/$ – see front matter © 2011 Elsevier Inc. All rights reserved.doi:10.1016/j.chieco.2011.07.011

Contents lists available at SciVerse ScienceDirect

China Economic Review

two institutions that are believed to contribute to public goods in China, and which are both notably different from equivalents inwestern countries: trust and property rights.

Trust has been studied both at the level of societies and institutions, and at the micro level of individuals. Sociologists such asGiddens (1984, 1990) and Beck (1986; Beck, Giddens, & Lash, 1994) have analyzed trust in relation to the changing character ofmodern societies, emphasizing how trust in persons has changed into trust in expert systems. Putnam (2000) and other socialscientists in the social capital tradition have focused on the role of trust as binding elements for communities and societies.Nooteboom (2002) provides a concise overview of the economic literature on trust. In economics, Knack and Keefer (1997),Desdoigts (1999), andWoolcock (1998) have examined the relationship between trust and economic performance at the nationallevel. Fukuyama (1995) emphasized the importance of non-kinship trust relations (which he distinguishes from family-based (orkinship) trust relations) on the ability of building economic corporations and businesses that can better compete in the worldmarket. At themicro level, trust is not only essential for cooperating within a group (as Messick et al., 1983; Parks & Hulbert, 1995;Putnam, 2000; and de Vos & Mol, 2010 illustrate), but it is also believed to contribute to public goods, even though there is scarcerobust empirical results in the literature (Anderson, Mellor, & Milyo, 2004; Bouma, Bulte, & van Soest, 2008; de Krom &Mol, 2010;Leonard, Croson, & de Oliveira, 2010). A well-known interdisciplinary volume, edited by Ostrom and Walker (2003) withcontributions from economics, sociology, political sciences and psychology, gives an illustrative overview of the trust literature atindividual level, applying behavioral experiments to explain how trust (and reciprocity) develops and is related to cooperation.The volume also helps us to make a link between trust and the provision of public goods.

The relationship between property right and the provision of public goods is not very conclusive in the literature (Starrett,2003). In special cases, if the free-rider problem can be solved and a public good becomes a private good, then a more securesystem of property rights may contribute to better public goods provision. But in most cases, secure property rights may increaseindividual welfare, because of the free-rider problem of public goods: conditional on the contribution of others to public goods, anindividual is maximizing her own utility if she contributes nothing to that particular public good (being a free-rider). Then, secureproperty rights may reduce an individual's contribution to public goods if she only values her own benefit and doesn't value socialbenefits. This follows the prediction in classical economic theory: everybody wants to become a free-rider with no contribution topublic goods.

In combating erosion and desertification (and thus to provide public goods), the Chinese government has set up the SlopingLand Conversion Program (SLCP) to give subsidies to farmers who convert their sloping arable land into grassland or forest. Butthis program and thus the subsidies are limited in time. In studying empirically the implementation of the SLCP in NingxiaAutonomous Region, China, we aim to investigate how trust and land property rights affect farmers' further contribution to thesepublic goods, when the subsidy program will have ended in the future.

The paper is organized as follows. Section 2 provides a brief overview of the SLCP in China and especially in NingxiaAutonomous Region. Section 3 develops hypotheses on how trust and property rights in China are related to public goods,especially in the framework of the SLCP. Section 4 reviews the survey and presents descriptive statistics for the variables used inthe model. Results of model estimations are discussed in Section 5, while Section 6 draws some overall policy conclusions.

2. Sloping Land Conversion Program in Ningxia Autonomous Region

In recent years, the Chinese central government introduced an ambitious land conservation program — the “Sloping LandConversion Program” (also called: grain-for-green project, or in amore direct translation “return cropland to forest land”). It is oneof the largest ecological recovery programs in the world. After a short pilot phase during the years 1999–2001, it has been fullyimplemented since 2002. The degraded ecological system in upper-stream areas are believed to be the cause of severe droughts inthe Yellow River in the 1990s and the devastating floods in 1998 in the Yangtze River Basin and Northeast China. The recovery ofthe degraded ecological system formed the main motivation for designing and implementing the SLCP. The greatly strengthenednational treasury and the surplus of grain supply at that time also provided a good support to implement such a large-scale landconversion program (Bennett, 2008; Hyde, Belcher, & Xu, 2003; SFA, 2003;WWF, 2003; Xu, Katsigris, &White, 2001; Xu, Tao, Xu, &Bennett, 2010; Zhang, Tu, & Mol, 2008; Uchida, Xu, & Rozelle, 2005). With the SLCP, the central government hoped to “kill twobirds with one stone”: a major ecological improvement and a significant contribution to poverty alleviation through economicdevelopment in the affected regions.

The SLCP is supposed to offer farmers in severely degraded areas an opportunity to break through the vicious circle aroundpoverty and ecological degradation. It offers financial subsidies to take arable land out of production and stimulates both otherfarm activities, such as forestry, fruit trees and livestock, and off-farm jobs and businesses. As one of the first national level“payment for environmental services” (PES) programs in China, the SLCP is notably different from China's other national forestryprograms, as it employs a public payment scheme that directly engages millions of rural households (in principle voluntarily) ascore agents of project implementation (Bennett, 2008; SFA, 2003). If de facto implementation of the program indeed involvesdecentralized, voluntary, grassroots participation of farmers, the SLCP would represent an important departure from “business asusual” in how China manages its forest resources (Bennett, 2008). A total budget of 337 billion Yuan (over US $ 40 billion) wasallocated for implementation of the program between 2001 and 2010. Compensations to farmer households are both in cash and inkind (e.g. grain). Considering the average regional grain yields, the annual grain subsidy is set at 2250 kg per converted hectare inthe Yangtze river basin and 1500 kg/ha in the Yellow River Basin. The cash subsidy reaches 300 Yuan/ha of eligible land per year.Both grain and cash subsidies will be given to farmers for 8 years when land is turned into ecological forests, and for 5 years or

500 Q. Tu et al. / China Economic Review 22 (2011) 499–511

2 years when land is turned into economic forests or grassland, respectively.1 Seedling provision for afforestation is subsidized at alevel of 750 Yuan/ha.

Most of the previous studies on the SLCP focus on the implementation capacity of the Chinese government (e.g. Dorward, Kidd,Morrison, & Urey, 2004; Ningxia Forestry Bureau, 2006; Zhang et al., 2008). Over the years, there has been a cloud of doubts aroundimplementation effectiveness of the program, given China's past records of tree planting and other large scale projects within andoutside the forestry sector. Several of the mentioned implementation concerns include issues regarding property rights of theconverted lands; failure to meet the target areas for conversion; a too narrow evaluation regime focusing on the direct indicators ofimplementation only; the danger to create a new landless and poor rural population due to the fast expansion of land conversion andthe absence of other income sources; and poor results due to insufficient compensation to local governments for implementation(Ningxia Forestry Bureau, 2006; Xu, Bennett, Tao, &Xu, 2004). Since 2004, the attention anddebateswith respect to this programhaveshifted to the sustainability of theprogram, because over-achievementof landconversion rates havebeen reported inmanyprovinces,while at the same time doubts emerged about what farmers would do after 8 years when the compensation scheme stops (Groom,Grosjean, Kontoleon, Swanson, & Zhang, 2010; Yeh, 2004). The sloping land conversion policywas originally based on the hypothesisthat after conversion farmers would be able to find a sustainable livelihood and income outside arable farming; either a direct incomefrom forest/grassland or indirect ones from livestock and off-farming. Bothwould not result in reconverting forest or grassland back toarable land after 8 years of compensation. However, this permanent transition of arable land will not take place automatically. Our2006 surveydata inNingxiaAutonomousRegion shows that themajority of householdsplan to reconvert landback to arable land after8 years. This fact may indicate that alternative off-farm opportunities are not developing as expected in a poor region like Ningxia. Itmay also be explained by poor levels of intrinsic factors (personal values, social concerns, trust etc.) that are behind themotivations toprovision of the public goods of forest and grassland.

Ningxia Hui Autonomous Region is located in northwest China, on the upper reaches of the Yellow River. Ningxia is one ofChina's five autonomous regions inhabited by minority groups. It is also the smallest provincial-level administrative unit in China,with only 66,000 km2 and 5.8 million people in 2004, of which about 35% are Hui minority. Ningxia was one of the regions that didnot share the large economic growth during the reform period after 1978 (Demurger, 2001). Among the 12 provinces,2 regionsand onemunicipality which form the big west, Ningxia ranked in 2004 at the bottom of most economic indicators: GDP per capita,GDP growth rate, urban and rural income per capita and their annual growth rates. It must be noted that within Ningxiadevelopments are not homogeneous; the rural net income per capita in the southernmountainous area is only about half of that inthe northern irrigated plain. The SLCP targets especially the poverty-stricken counties in the south.

The Ningxia deputy party secretary once argued that the root of poverty in the western part of China lays in the region'senvironmental fragility and degradation (Edmonds, 2005). For the relatively poor Ningxia counties targeted by the SLCP,subsidized land conversion was seen as a direct contribution to local economic and income growth. Over half of the total land andfour fifths of themountainous land that fall under the SLCP are found in the south; almost all sandy lands to be converted are foundin central Ningxia. Thus the project is heavily focused on the poorer areas in Ningxia (Edmonds, 2005). The SLCP, with its promiseto protect the environment and to alleviate poverty, seems to meet well the urgent needs of the poorer Ningxia farmers. Andindeed, the SLCP was welcomed by Ningxia leaders in 2000 with a slogan “complete the tasks of eight years within three years”.The regional government listed the SLCP as one of the key, high priority projects. By the end of 2005, Ningxia had accomplished theconversion of 11.04 million mu3; 4.56 million mu of crop land was converted into forest, 5.98 million mu wasteland wasconverted to forest, and 0.5 million mu of degraded mountainous land was closed off from any economic activity.

In order to gain an insight in how the SLCP has been implemented in Ningxia, how it has been received by farmers and to whatextent the (claimed) conversionswill endure after the finalization of the (financial support of the) program,we conducted a survey inthree counties inNingxia in early 2006. TheSLCP targeted area inNingxia is a semi-aridone,where theaverage yield of grainon slopingland is relatively low. Farmers are quite happy to participate in the SLCPwith the government subsidies, but our survey data show thatthe majority of households plan to reconvert back to arable land after 8 years if the government stops subsidies.

3. Hypothesizing trust and property rights in relation to public goods

3.1. Trust is the cornerstone of cooperation and public goods provision

Cooperation and reciprocity do not work without trust. By using both results from the experimental laboratory and the field,Walker and Ostrom (2009) illustrate the importance of trust and reciprocity for solving social dilemmas, like public goods andcommon pool resources (CPR). Furthermore, they demonstrate how trust and reciprocity affect the level and sustainability ofcooperation. Many other researches provide evidence of the importance of trust for other similar situations, Ostrom (2009) forCPR; Cox et al. (2009) for private and common property experiments; Anderson et al. (2004), Bouma et al. (2008) and Leonardet al. (2010) for public goods.

1 By the definition of the State Forestry Administration, “ecological forests” are timber-producing forests; “economic forests” are orchards or forests withmedicinal value.

2 The big west in China includes Yunnan, Guizhou, Sichuan, Chongqing, Shaanxi, Gansu, Ningxia, Qinghai, Xinjiang, Inner Mongolia, Tibet and Guangxi. In:http://www.chinawest.gov.cn/, July 29, 2006.

3 1 mu=1/15 hectare.

501Q. Tu et al. / China Economic Review 22 (2011) 499–511

If cooperation and reciprocity are limited to a small group, especially a kinship network, then, trust among kinship members maybe sufficient to guarantee this cooperation. But most public goods need cooperation from a larger group of people beyond the kinshipnetwork, where trust among general people (non-kinship members) rather than kinship members may play an important role. As aconsequence, general (non-kinship) trust would be more important to public goods protection and provision. Kinship trust, incontrast, is relatively more inward-looking, group-interested or “selfish” to small specific groups with kinship relations – excludingotherswho are not kinship related – and is less likely to be associatedwith common and publicwelfare provision and thuswith publicgoods. Therefore, general trust can be expected to have a positive relationship with public goods provisioning, but not kinship trust.

3.2. Trust, general trust and kinship trust, and their measurement

The trust question from theWorld Values Survey (WVS) – “Generally speaking, would you say thatmost people can be trusted, orthat you can't be too careful in dealing with people?” – is a widely recognized estimation for individual trust, and sociologists tend touse this question tomeasure and compare the trust level (percentage of people answering that they trust others) between countries atvarious points in time. In all surveys conducted over the last years, China got a suspiciously high trust score (Uslaner, 2002). It rankedas the third country in the1990survey among44countries, it got the samerankingwithmore countries in a1995 survey, and it rankedfourth in a 2001 survey among 82 countries. Inglehart (1999) explains this by referring to Confucian culture in China.

Fukuyama (1995) divides trust into two components: general (non-kinship or generalized) trust and kinship trust. WhenFukuyama regards trust as social capital (or virtue), he refers to non-kinship trust. His results show that not only mainland Chinabut also other areas with a Confucian culture (Chinese culture), like Hong Kong and Taiwan, have low levels of general trust, whichis inconsistent with the results of WVS. All three societies are mainly based on kinship networks. He separates countries into twocategories according to their non-kinship trust: high-trust countries, such as Japan, Germany, and the United States; and low-trustcountries such as China, Korea, Italy (especially the south), and France. Generally speaking, trust makes it easy for people tocooperate and work together. But high kinship trust may only make that people with kinship relations (a relatively small group)work easily together; and high general trust may make that people from the entire society cooperate easily. As a result, highkinship trust can make it easy for Chinese people to setup small family owned businesses, but low non-kinship trust will make itdifficult for them to further develop these small family-based businesses into large corporate companies.

Because of Confucian culture, people pay more attention to family (kinship) networks in China. Fei (1992[1948]) summarizesthat, unlike western societies, rural (or traditional) society in China is strongly based on kinship relations and networks, whichmay indicate that the structure of trust in China is strongly influenced by high levels of kinship trust. Fukuyama (1995) also showsthat family ties create kinship trust at the expense of general trust. Following his logic, trust levels in China may be especially highwithin kinship relations; but general trust levels – among people without kinship relations – can be much lower than in westerncountries and cultures. The WVS trust question might measure a mix of general and kinship trust in China.

Although there has been some debate whether attitudinal surveys are adequate instruments to measure social capital and trustfor predicting their contribution to economic performance (Glaeser, Laibson, Scheinkman, & Soutter, 2000), such surveys are stillwidely used, also in relation to public goods. Anderson et al. (2004) find that the most popular WVS trust question is stronglycorrelated with contribution in a public goods experiment. Their result shows that trust may improve individual contributions topublic goods, and it is good for social and economic development.

WVS trust is abinary choicequestionand is relatively simple. Itmaybedifficult to capture theessence of trust onlywith thisWVS trustquestion. Therefore, more comprehensive methods to measure the level of trust are becoming increasingly popular in the literature.

One approach is to use the result of trust games (amount of money sent by the first player) instead of a questionnaire as theproxy for the level of trust. Bouma et al. (2008) use a trust game to measure trust, and link trust with the provision of a local semi-public good (investments in soil and water conservation maintenance). Another approach is to use more questions about trustrather only the WVS trust question. Leonard et al. (2010) for instance use survey data to investigate the relationship betweensocial capital (trust and social networks) and contributions to public goods (self-reported voluntary contributions to localcharitable organizations). They employ a more specific operationalization for trust than in the WVS trust question: the totalnumber of groups trusted is divided into seven different groups, including people you know, people you meet for the first time,people of your religion, people of other religions, people of a different nationality, family members, and other neighborhoodresidents. They find a strong positive correlation between trust and self-reported charitable voluntary contributions in time ormoney (binary variable). Tu and Bulte (2010) combine the two approachesmentioned above to distinguish general trust and localtrust. They employ trust experiments to measure general trust of farmers, and questions (trust to friends and relatives) in aquestionnaire to estimate local trust. Their results show that general trust makes it more likely that people participate in labormarkets, and local trust makes informal (traditional) labor exchange arrangements more prevalent.

Consequently, the WVS trust question might not be specific enough to fully understand the relation between trust and publicgoods provision (see also Leonard et al., 2010), especially in China. Hence, in our 2006 survey we included twelve questions abouttrust, one of them being the WVS trust question. These twelve questions may help us to differentiate between general (non-kinship) trust and kinship trust, which is discussed in detail in Section 4.

3.3. Public good of the SLCP case in Ningxia

Our survey area, Ningxia Hui Autonomous Region, is mainly a semi-arid area. Generally, the economic benefit that famers getfrom using sloping land as cropland is less than the SLCP subsidies from government, but larger thanwhen they use it as forestland

502 Q. Tu et al. / China Economic Review 22 (2011) 499–511

or grassland. Therefore, farmers have a strong economic incentive to join the SLCP with the government subsidies, but also toreconvert the forestland or grassland back to cropland after the government stops subsidies.

The environmental benefit of farmers joining the SLCP is related to new forest and grassland. The benefit is for wider areas,especially when the SLCP area is located in upper stream areas. However, the results of our survey show that local farmers also getlocal environmental benefits through reduced land erosion, better water and air quality, and fewer sandstorms. As a result, theSLCP can be also considered as a local public good, not only a national one, especially when forest and grassland will be properlymaintained after extrinsic incentives have stopped (that is: after the government stops SLCP subsidies). In this case, those farmerswho joined the SLCP have two choices after the government ends the subsidies: one is not reconverting the forest or grasslandback to crop land, which can be considered as individual contributions to local public goods; the other is cutting trees andreconverting land back to arable land, which can be considered as an investment in private (family) goods.

In general, land is divided into several categories according to its soil quality, slope, location, etc. in rural China. Farmers in avillage get different plots of land of different categories randomly. The probability of relatives' land located on the same hill is verylow, evenwhen some relatives are living in the same village. So, local public goods in our case aremainly for local communities andnot so much for a kinship network.

3.4. How general trust and kinship trust affect contribution on public goods

We assume that households care about both their own economic benefit, which comes from private goods, and the localenvironmental quality, which is a local public good. When they make the decision of reconverting the land or not after the SLCPsubsidies are stopped, they are facing the trade-off between two alternatives:

• Reconvert land from forest/grassland to crop land. It is a costly activity. A household that invests in reconversion, which is aprivate good, gains the economic benefit of the reconverted crop land. That household does not invest in the local environmentalquality.

• Do nothing, just keep the forest/grassland unchanged. This can be considered as contributing to public goods — localenvironmental quality, because a household may lose the economic benefit of reconversion. The household invests in the localenvironmental quality.

The local environmental quality is affected by joint decisions of all households in the local community, and not from anindividual household. When farmers make their decisions, they may compare their own welfare changes that would result fromthese two alternatives. Therefore, when we analyze how trust and property rights affect the provision of public goods through theSLCP by rural households, we can assume that theremay exist two opposite effects, a public goods effect and a private goods effect,which together determine households' welfare when they make the decision on land reconverting. This is summarized in Table 1.

3.4.1. Public goods effectThere is a positive local environmental effect of the SLCP; therefore, keeping converted land of the SLCP as forest or grassland

can be regarded also as a continued contribution to local public goods. The public goods effect comes when farmers care aboutlocal environmental quality and want to invest in it. Then famers are less likely to reconvert the land and more likely to keep it asforest that will continue to contribute to public goods.

But not everybody believes in the local environmental effect of the SLCP. Our survey shows that around two thirds of thehouseholds recognize it. We assume that the public goods effect appears only when farmers believe that the SLCP has anenvironmental effect and that it can improve the local environmental quality, from which they can benefit.

As we discussed before, general trust may have a positive effect on the provision of public goods: farmers with high generaltrust are more likely to believe that others will do the same as they do and contribute to public goods, rather than becoming free-riders. As a consequence, for the SLCP case, we can expect that general trust has a significant public goods effect when farmers areaware of the improvement of the local environment by the SLCP. This effect may disappear when farmers are not aware of thepositive environmental effects of the SLCP. On the other hand, as we discussed before, kinship trust is mainly about small groupswith kinship relationships. It is not very likely that farmers with high kinship trust contribute to public goods, even when theyrecognize the positive environmental effects of the SLCP. It is more likely that they become free-riders. Hence, kinship trust willhave no significant public goods effect, regardless of farmers' awareness of environmental effects.

Table 1Effects of general trust, kinship trust and land property rights on public goods (PG) provision (through not reconverting land).

Variable Public good effect: contributeto PG

Private good effect: contributeto PG

Total effect on PG provision

General trust If aware of environmental effects + Unknown UnknownIf not aware of environmental effects 0 Unknown Unknown

Kinship trust If aware of environmental effects 0 Unknown UnknownIf not aware of environmental effects 0 Unknown Unknown

Land property rights 0 – –

503Q. Tu et al. / China Economic Review 22 (2011) 499–511

3.4.2. Private good effectIt is easy to understand that farmers may care more about their own economic benefit when they make the land reconverting

decision. If farmers can get higher income from land reconversion after subsidies stop (which is the case in the SLCP in Ningxia),economically, it is more likely that they reconvert their land, and less likely that they contribute to public goods (maintain the landas forest). Hence, we hypothesize that the potential increased income of reconverting land has a negative effect on public goodsprovision; farmers pay more attention to and invest in their private goods (reconverted land) because of economic incentives. Wecall this negative effect on the contribution to public goods the “private good effect”.

In the literature, some studies show (Knack, 2001; Knack & Keefer, 1997; Tu & Bulte, 2010) that trust levels are positivelycorrelated with economic performance. The trust in those studies is more like general trust rather than kinship trust. There is nodirect evidence on the relationship between general trust, kinship trust and private investment in the literature. Therefore, we donot knowwhether general trust and kinship trust have a private good effect or not; but we can test it with our data.We expect thatkinship trust may have a stronger private good effect than general trust, because kinship trust may make people pay moreattention to their own private economic benefit than general trust when people are facing a social dilemma like public goodprovision.

3.5. Land property rights and contribution to public good

People need secure property rights to protect their investments in their property. Property rights are also related to publicgoods provision, andmay interfere with trust only through the economic incentives of households. In general, one can expect thatless secure property rights will make people less willing to invest in their property, whether it involves public goods or not. Theirfocus will primarily be on economic exploitation for private gains.

Income from reconverted land is higher than that from forest or grassland after subsidies stop in Ningxia; investing in privategoods is economically profitable. So, land property rights have no public goods effect. They only have a private good effect, becausefor economic reasons, higher security of land property rights increases the likelihood that farmers reconvert land and invest inland reconversion. Hence more secure land property rights makes it less likely that farmers contribute to public goods in the SLCPcase in Ningxia.

3.6. Summary

The total effects of trust and land property rights on public goods provision are shown in the last column of Table 1. Whenhouseholds are aware of the positive effects of SLCP on the local environment, the total effect of general trust on public goodsprovision is not conclusive, as it depends on the size of two opposite effects: a positive public goods effect and an unknown privategood effect. In all other three cases of trust, because there are no public good effects, the total effects are determined by theunknown private good effects.

Farmers in one village face equal land property rights, but may differ in awareness of property right security. We use farmers'“safe feeling for land rights” as a measure of land security in our analysis. Land property rights have no public goods effect, but thefeeling of more secure property rights make it more likely that farmers opt for gaining profit from reconverting land, and thus ithas a negative effect on public goods provision. Trust may also influence our property right variable, and property rights might beendogenous in our analysis. Wewill check the correlation coefficients between trust and land property rights whenwe discuss thedata.

Now the question is how general and kinship trust levels, and secure land property right (or better: the perception of secureproperty rights), affect the willingness to contribute to public goods (by not reconverting forest or grassland after the SLCP ends).The dependent variable is then measured as the household's tendency or likeliness not to reconvert forest and grassland back tocropland after the end of the subsidies under the SLCP. Hence, our main hypotheses are: 1) compared to individuals with highkinship trust, it is more likely that individuals with high general trust contribute to public goods when farmers are aware of theenvironmental effects; 2) perceptions of secure property rights will make people contribute less to public goods as they are morewilling to invest in their own private goods, whichmeans reconversion of forest and grassland into crop-land. Building on previousstudies, this paper uses household survey data from Ningxia and estimates what the most important factors are that affect theprobability of households to reconvert forest and grassland into cultivated land after the subsidies stop.

4. Data

We conducted our survey in February 2006, in three counties in the south of Ningxia Autonomous Region — Tongxing,Pengyang and Xiji. Together the three are representative for the nine counties of the southern part of Ningxia involved in the SLCP,but they have different levels of farmer income per capita.4 Average per capita rural household net income in Ningxia is only 2509Yuan in the year 2005 (NBS, 2006), which is significantly lower than the national average of 3255 Yuan. In addition, there is a hugeregional gap between the northern irrigated area and the southernmountainous area: the average farmer net income per capita is

4 The mountainous area in the southern part of Ningxia is said to be “the main battle field for combating poverty”. This southern part occupies 58% of the totalland and is the home of 43.4% of the total population of Ningxia. Of its 2.56 million population, 1.35 million are Hui minority (65% of the total Hui people inNingxia), Ningxia Poverty Alleviation Office (2006).

504 Q. Tu et al. / China Economic Review 22 (2011) 499–511

3527 and 1715 Yuan in the north and south, respectively (Ningxia DRC, 2006), and 89% of GDP in 2005 is contributed by northernNingxia. Net farmer income per capita is used as the criterion to select the counties to represent the relatively richer, middle andpoorer counties of southern Ningxia, ranging from 1312 Yuan to 1762 Yuan. Our samples' average income level is 1280 Yuan in2005.5

In total, 349 households from 12 townships were interviewed. After cleaning the data, 316 households remained for regressionanalysis. Apart from basic household characteristics (age, education, family size), we included questions on household income,assets and consumption, membership of farmer's organizations, access to education, and market information, general trustquestions and trust to different groups of people, and perception of secure property rights (see Appendix A for an overview of thedescriptive statistics).

One of the key questions in the questionnaire focuses on the willingness of contributing to public goods. Because thegovernment is likely to end the subsidies of the SLCP in the next 3–4 years6 in Ningxia, we cannot yet observe the reconversionbehavior. We can only ask participating households about their willingness or tendency of not reconverting land. This is thus asubjective expected possibility of not reconverting forest and grassland back to crop land, which can be considered as a continuouscontribution to public goods. Interviewed households have to choose an answer from a five-point scale (see Table 2): 5) surely notreconvert; 4) probably not reconvert, 3) not clear, 2) probably reconvert, and 1) surely reconvert the forest land. Since the data areordered in five categories instead of on a continuous scale, we should use an ordered Probit model instead of a linear regression tocapture the characteristics of our data.

The results of Table 2 indicate that the sustainability of the SLCP is doubtful if the governmentwould not take any further actionafter the subsidies stop. Almost half of the respondents (47%) indicate it quite probable that they will re-cultivate forest orgrassland after the government subsidies stop. Only 18% of the respondents indicate they will not re-cultivate the land.Interestingly, the data also show that farmers are more likely to contribute to public goods (less likely to reconvert the forestland)if they are aware of the SLCP's environmental effects. Themeans of the likelihood of not reconverting are 2.62 and 2.33 respectivelyfor groups with and without awareness of the environmental effects; the difference is significant at 5% level.

Different groups of households answered differently on the probability to not reconvert the land. In our total sample, 94% of thehouseholds are participating in the SLCP. More than 2/3 (68.7%) of the total households confirmed the positive environmentaleffects of SLCP, either in terms of less sand storm, less water erosion following rain, or better air quality. This indicates that theenvironmental goals of the SLCP are achieved to a certain extent during the last years according to the farmer households. Table 3shows household characteristics of the respondent groups for willingness of reconverting land.

Because our main objective is to gain the insight in how trust and property rights affect households' contribution to publicgoods, farmers' perception regarding security of land rights was included in the household survey. With respect to trust, twelvedifferent questions were included in our questionnaire. Besides the standard WVS trust question, respondents were also asked toindicate their trust level to eleven different groups (parents, brothers/sisters, children, other relatives, local officials,classmates/peers, religious staff, neighbors, people in the same village, people in nearby villages, strangers), using a scale from0 (totally distrust) to 10 (fully trust). Table 4 shows the descriptive statistics of all twelve trust questions.

It is very clear that trust levels to groups with kinship relations, like parents, brothers/sisters, children and relatives, are highestamong all groups; and trust level to strangers is much lower than all other eleven groups. In order to check whether our 12 trustquestions are consistent with Fukuyama's structure, factor analysis with themethod of principal-component factors is used. Factoranalysis is a good way to discover simple patterns among variables. Because the standard method of factor analysis only getsorthogonal factors, but general trust and kinship trust may be correlated, we should allow correlation between factors. Therefore,we used oblique rotation to get factors with correlation. Interestingly, two retained factors emerged from the analysis. Table 5gives the pattern matrix of factor analysis.

5 Xiji county is richer than the other two counties and has the highest SLCP participating rate. With respect to other variables, such as household size, years ofeducation, land per capita, the counties do not differ strongly.

6 Ningxia started the SLCP in 2000. Most farmers (also most of those included in our survey) started participation in the SLCP during 2001 to 2003, so theywere expected to see the end of subsidies in 2009–2011. Our data was collected in 2006, midway the program. In 2007, the central government announced a newpolicy to double the years of subsidies. So now ecological forests can have an extra 8 years of subsidies, but subsidies are reduced to half of the level they werebefore.

Table 2Likelihood of not reconverting the land (contribution to public goods).

Freq. Percent (%) Awareness of the environmentaleffects of SLCP

No Yes

5) Surely not reconvert 26 8.2 7.3 8.64) Probably not reconvert 32 10.1 7.4 10.93) Not clear 110 34.8 31.6 36.42) Probably reconvert 66 20.9 17.9 22.31) Surely reconvert 82 26.0 35.8 21.8Total 316 100.0 100.0 100.0

505Q. Tu et al. / China Economic Review 22 (2011) 499–511

From Table 5, we can see that all 12 variables contribute positively to the first factor and trust to non-kinship groups has muchhigher weights than trust to kinship groups, especially to family members. Trust to kinship groups, especially family groups,contributes with high weights to the second factor, while trust to other groups hardly contributes, having small positive ornegative weights. The structure of factor analysis is very clear, and wemay interpret the first factor as general trust and the secondone as kinship (or family) trust. The correlation coefficient between general trust and kinship trust is quite low, only 0.11.

Other independent variables included in the analysis refer to:

a) Demographic characteristics:Ningxia is a Hui Autonomous Region; around 34% of the total population belongs to the Hui ethnic group, which is Muslim. Lessthan 1% belongs to other minority groups. Around 65% are Han nationality. Our sample has a quite similar ethnic structure: 34%are Hui and the others are Han; no respondents from other small minority groups were included. Generally, men have morepower within the family than women in rural China, especially in the north-west part of the country. In our survey in Ningxia'srural areas, nearly 97% of the household heads are men and their average age is 42 years. Their education level is relative low:on average household heads have only 5.2 years of education. The average size of households is 4.5 persons, of which 66% arelaborers.7

b) IncomeOur research site is a poor area, the average net income per capita is only 1280 Yuan (the average of its logarithm is 6.6).

c) Land and its securityThere are three questions asked in this category. The first question is about security of land tenure rights: “how safe do you feelabout your land?” Actually, it is the general perception of land rights security. The answer ranges from 1, not safe at all, to 4,very safe. Larger numbers mean safer (subjective) land rights. The average score is 2.8. This variable may correlate with trust,because people with high (general) trust may trust the government more and have a higher perception of high land security. Itseems that our results do not support this hypothesis, because the correlation coefficients between perception of land securityand general trust and kinship trust are very low, 0.047 and 0.067 respectively.The second question is about the cropland per capita after converting sloping land into forest and grassland. On average,households in our sample have 3.8 mu cropland per capita left. The third question relates to grain self-provision of farmers:“Does your land produce enough grain after your participation in the SLCP?”Only 22% of the respondents answered affirmative.It seems that most households do not have enough grain after implementation of the SLCP. Traditionally, farmers care muchabout their grain self-provision in China, and this may have significant effects on land reconverting decisions.

d) Access to informationThe variable “Years of knowing the SCLP” comes from the question “in which year did you hear of the SLCP?” Some heard aboutit from the start of the pilot phase of the project (8 years ago); some just recently heard about it (in 2005). On average, farmersbecame familiar with the SCLP in 2001, 5 years before our survey was carried out.

7 A laborer is defined as a member between 18 and 60 years old in the household.

Table 4Descriptive statistics of trust questions.

Variable Mean Std. dev. Min Max

1) WVS trust question 0.455 0.485 0 12) Trust to parents 9.85 0.613 3 103) Trust to brothers/sisters 9.29 1.29 0 104) Trust to own children 9.09 1.74 0 105) Trust to relatives 7.36 2.65 0 106) Trust to local officials 4.43 3.08 0 107) Trust to classmates/peers 6.14 2.79 0 108) Trust to religious staff 4.93 3.76 0 109) Trust to neighbors 5.93 2.90 0 1010) Trust to villagers 5.62 2.89 0 1011) Trust to people in nearby village 4.31 2.92 0 1012) Trust to strangers 1.43 2.41 0 10

Table 3Household characteristics for respondent groups.

Variable Sure not reconvert Prob. not reconvert Not clear Prob. reconvert Sure to reconvert All (mean and [SD])

Sample (N) 26 32 110 66 82 316Net income per capita (Yuan) 1378 1236 1190 1267 1398 1280 [1465]Number of members in hh 4.3 4.6 4.3 4.9 4.5 4.5 [1.5]Number of laborers in hh 3.0 2.9 2.9 2.7 2.7 2.8 [1.0]Years of education of hh head 6.5 6.4 5.2 4.6 4.9 5.2 [3.4]Land per capita (mu) 4.4 3.7 4.1 3.8 3.4 3.8 [2.2]

506 Q. Tu et al. / China Economic Review 22 (2011) 499–511

5. Results

The dependent variable, the willingness of contributing to public goods, comes from the question of the likeliness of notreconverting the land after the governmental compensation stops. Ranging from 5 (highest likelihood of contributing to publicgoods) to 1 (lowest likelihood of contributing to public goods) on a five-point scale, the dependent variable is an ordered one andwe carried out both OLS and ordered Probit estimation. The results are presented in Table 6. Two county dummy variables and theconstant term are also included in the regression, but their estimated coefficients are insignificant; we do not present them inTable 6. As we discussed before, the property security variable is almost not correlated with trust variables; it seems thatinstrument variables are not necessary. Below, we discuss the most important independent variables with respect to their impacton farmers' willingness of contributing to public goods. The results of OLS are quite similar with those of the ordered Probit model.

5.1. Demographic variables

The result shows that the minority (Hui) group is less willing to contribute to public goods than the majority (Han) group. Themain reason might be that people from the Hui group find more difficulties in getting jobs in cities, especially in southeast China,where most inhabitants are Han, than those from the Han group. Most Hui people are Muslims, who mainly live in thenorthwestern part of China. The cultural (religious) differences between the two groupsmay play a role. Hui farmers are less likelyto do off-farm jobs in the coastal area than Han farmers. Because Hui people rely more on farming and land than Han people, theyare more likely to reconvert land to crop land after the SLCP subsidies stop. Household heads that are male and have highereducation are more likely to contribute to public goods, as education can enhance people's understanding of the advantage of

Table 5Pattern matrix (rotated) of factor analysis.

Variable Factor 1 (general trust) Factor 2 (kinship trust)

1) WVS trust question 0.39 −0.0502) Trust to parents 0.042 0.603) Trust to brothers/sisters 0.018 0.824) Trust to own children 0.12 0.755) Trust to relatives 0.58 0.326) Trust to local officials 0.52 −0.0917) Trust to classmates/peers 0.74 −0.118) Trust to religious staffs 0.50 −0.349) Trust to neighbors 0.76 0.03210) Trust to villagers 0.83 0.05211) Trust to people in nearby village 0.79 0.09712) Trust to strangers 0.64 0.015

Table 6Results for likeliness to contribute to public goods. Dependent variable: likelihood of not-reconverting land.

Variable OLS Ordered probit

Coef. Robust std. err. Coef. Robust std. err.

Minority: 1 yes, 0 no −0.495⁎⁎⁎ 0.169 −0.534⁎⁎⁎ 0.175Gender of household head: 1 male, 0 female 0.621⁎⁎⁎ 0.225 0.590⁎⁎ 0.287Age of household head / 10 −0.0251 0.0625 −0.0312 0.0636Years of education of household head 0.0483⁎⁎ 0.0214 0.0436⁎⁎ 0.0211Number of members −0.0177 0.0604 −0.0120 0.0580Labor / member ratio 0.613⁎ 0.330 0.617⁎ 0.328Log (income per capita) −0.149⁎⁎ 0.0634 −0.152⁎⁎ 0.0600Dummy for enough grain 0.569⁎⁎⁎ 0.175 0.555⁎⁎⁎ 0.171Non-converted land per capita (mu) 0.0602⁎⁎ 0.0291 0.0630⁎⁎ 0.0283Years of knowing the SLCP 0.108⁎ 0.0582 0.116⁎⁎ 0.0580Public goods effect of general trust: (General trust * Environmental awareness) 0.429⁎⁎⁎ 0.131 0.490⁎⁎⁎ 0.140Public goods effect of kinship trust: (Kinship trust * Environmental awareness) 0.0480 0.132 0.0523 0.134Private goods effect of general trust: (General trust) −0.397⁎⁎⁎ 0.0987 −0.450⁎⁎⁎ 0.114Private goods effect of kinship trust: (Kinship trust) −0.209⁎⁎ 0.101 −0.214⁎⁎ 0.106Property security: land safe feeling −0.155⁎⁎ 0.0691 −0.165⁎⁎ 0.0723Environmental awareness 0.152 0.146 0.174 0.148(Pseudo) R2 0.225 0.088

Numbers of obs: 316.⁎⁎⁎ pb0.01.⁎⁎ pb0.05.⁎ pb0.1.

507Q. Tu et al. / China Economic Review 22 (2011) 499–511

public goods. Household heads that are male and have higher education probably also have more off-farm opportunities and thusare less dependent on income from crops. The variable labor/member ratio is significant at 10% level, meaning that householdswith a high proportion of labor are more likely to contribute to public goods; this may also relate to off-farm opportunities. A lowlabor/member ratio means that there may be more elderly or children at home who need caring. Hence, family laborers have lesschance to go out for off-farm jobs, and are thus more dependent on land and agricultural income. It is thus more likely that thesehouseholds reconvert land and less likely that they contribute to public goods. The estimated coefficients for age of householdhead and number of household members are insignificant.

5.2. Income

We expected that rich households would pay more attention to the local environment, and therefore would be more likely tocontribute to this specific public good and not reconvert land. Beyond our expectation, households with high income have asignificantly lower likelihood to contribute to public goods. The reason might be that high income households may have moreability to invest more in reconverting land, and have thus higher profits when producing crops, compared to poor farmers. Anotherexplanation is also possible. The rural area in Ningxia, especially the southern part, is very poor compared to the east coast area inChina. In our sample, net income per capita in 2005 is only 2875 Yuan for the household at 90% income percentile. Hence,households with such low incomesmay not pay attention to the local environment. Results might have been different if the surveywas carried out in a rich area.

5.3. Grain security and land per capita

Traditional farmers in China are autarkic. They produce their own grain, and pay significant attention to their grain security.Even when they are rich enough to buy grain from the market, they still prefer to produce their own grain. Against thisbackground, it is easy to understand that households are less likely to reconvert land and thereby contribute more to public goodsif they think that they can produce enough grain. Farmers may get more income from planting if they have more land per capitaleft after the SLCP. Not surprisingly, farmers are more likely to contribute to public goods and not to reconvert the land if they havemore arable land left after the SLCP.

5.4. Information

Our results show that the more years farmers know about the SLCP, the less likely farmers reconvert land. This may be asignal that the reconverting rate may reduce if we wait the other 3–4 years, just before subsidies stop. It may also relate tothe fact that environmental effects become more obvious when the SLCP has been implemented for a longer period, andfarmers may be more willing to continuously contribute to public goods if they get more benefit from local environmentalimprovements.

5.5. Trust

As we hypothesized in Section 3, results of trust in Table 6 show that the situations are quite different when respondents dorecognize the positive environmental effects of the SLCP (that is: realizing the SLCP in terms of a public goods) compared to whenthey do not recognize these effects. We assume that the public good effect may only appear when the environmental effects arerecognized. As a consequence, we use the interaction term of trust and recognition of the environmental effects (dummy variable)to capture the public good effects of general trust and kinship trust. The private good effect exists regardless of recognition of theenvironmental effects, so we can just use trust variables alone to represent the private goods effect. We try to use the framework ofSection 3 (see summary in Table 1), distinguishing between a public good effect and a private good effect, to explain how trustmayimpact on public goods provision. All results of trust are presented in Table 7.

5.5.1. Public goods effectWe assume that the public good effect only exists when farmers have an awareness of the environmental effects of the SLCP.

The result shows that general trust has a very significant and strong public good effect. As we predicted before, unlike general trust,kinship trust has no significant public good effect, we cannot reject the hypothesis that the public good effect of kinship trust isequal to zero. All these results regarding the public good effect are consistent with our hypothesis in Section 3.

As we discussed before, kinship trust originates from kin-relationships, and it may mainly work inside a kinship network.Compared to general trust, kinship trust may be not so useful to promote the cooperation in an extended network beyond kinshiprelationships. It sheds a light on solving social dilemmas in China. As Fukuyama (1995) illustrates, China has a strong kinship trustand low general trust, whichmakes it difficult for farmers to cooperatewhen they are facing social dilemmas, like problems of localpublic good provision and the management of common pool resource.

5.5.2. Private goods effectBoth general trust and kinship trust have significant private good effects. In other words, people with either higher general

trust or higher kinship trust are all more likely to pursue their own economic benefit, and are thus more likely to invest in private

508 Q. Tu et al. / China Economic Review 22 (2011) 499–511

goods (land reconversion after government subsidies stop). A larger private good effect may arise when the potential economicbenefit is bigger. The strong private good effects may prevent farmers with high trust from contributing to the public good in ourspecial case. Surprisingly, general trust has an even stronger private good effect than kinship trust; this is different from ourexpectation. This result is inconsistent with the literature that trust promotes cooperation and hence is good for public goodprovision.

One possible explanation is that trust may promote market participation (Tu & Bulte, 2010). If famers are more likely to tradeon markets, they may want to produce more than their subsistence needs, and hence they are more likely to cut trees. This mayexplain why general trust has an even larger private good effect than kinship trust, because formal markets are more related togeneral trust than kinship trust. Another explanation might be that the land reconversion is a costly activity after the forest hasbeen growing there for more than 8 years; it is a big investment in terms of labor cost. Land reconversion may be difficult done byonly one household member (or head), and may need cooperation between household members. Therefore, it can be alsoconsidered as a family public good, and then both general trust and kinship trust may work here.

5.5.3. Total effectThe total effects of general trust and kinship trust on public goods provision are determined by the sum of two opposite effects,

public good effect and private good effect, as discussed above.For farmers with environmental awareness, two opposite effects of general trust cancel out. The size of the public goods effect

of general trust is slightly larger than the private goods effect (0.49 vs. 0.45). Hence, the total effect of these twomaymake that it isslightly more likely that people with high general trust contribute to public goods than that they do not contribute to public goods,but the total effect is not significant. The positive public goods effect of kinship trust is insignificant and much smaller than theprivate goods effect (0.05 vs. 0.21), hence the net effect is that people with high kinship trust are less likely to contribute to publicgoods. This net effect is significant at 10% level.

For farmers without environmental awareness, the public goods effect disappears and only the private good effect plays a role.Because both kinds of trust have significant private good effects, high general trust and kinship trust make farmers in this grouppursue their own economic benefit and less likely to contribute to public goods.

5.6. Land property security

Consistent with our previous hypotheses, a high perception of secure land rights makes it more likely that farmers reconvertland and less likely that they contribute to public goods. Since the late 1990s, the central government has been stimulating thirty-year-contracts for farming land. Therefore, in general land rights are quite secure in China, but some villages still make smalladjustments according to population changes. These land adjustments affect farmers' perception of land security. Reconvertingforest land after 5–8 years into crop land is quite time- and money-consuming, and farmers perceiving low land security levelsmay therefore prefer not to reconvert the land. Farmers need a feeling of safety regarding their land property rights before theyinvest in reconverting land.

Our result confirms that security of land rights makes farmers more likely to reconvert land, and this will cause problemsof sustainability of the SLCP. In some other provinces, profitable fruit trees were planted during the SLCP. If the expectedincome of these fruit trees is higher than the income from grain or other crops they planted before, secure land rights wouldactually make farmers less likely to reconvert land, and it would contribute to the sustainability of the SLCP after governmentsubsidies stop.

As we showed in Section 4, farmers are more likely to contribute to public goods when they are aware of the environmentaleffects (significant at 5% level). But the estimated coefficient of environmental awareness is insignificant after controlling for allother variables in our regression analysis.

6. Discussion and conclusions

With the help of a unique dataset based on a survey amongst 349 farm households in Ningxia related to the Sloping LandConversion Program, we identified the factors influencing the decisions for (not) reconverting the conserved forest andgrassland back to crop land when the government finishes subsidies of the SLCP. The non-conversion can be seen as a

Table 7Effects of general trust and kinship trust on public goods (PG) provision (through not reconverting land).

Variable Public good effect: contribute to PG Private good effect: contribute to PG Total effect on PG provision

Generaltrust

If aware of environmental effects 0.490⁎⁎⁎ −0.450⁎⁎⁎ 0.040If not aware of environmental effects 0 −0.450⁎⁎⁎ −0.450⁎⁎⁎

Total sample 0.342⁎⁎⁎ −0.450⁎⁎⁎ −0.108Kinshiptrust

If aware of environmental effects 0.0523 −0.214⁎⁎ −0.162⁎

If not aware of environmental effects 0 −0.214⁎⁎ −0.214⁎⁎

Total sample 0.0365 −0.214⁎⁎ −0.178⁎

⁎⁎⁎ Significant at pb0.01.

509Q. Tu et al. / China Economic Review 22 (2011) 499–511

continuous contribution to a public good. Subsequently, we investigated the relationships between the willingness tocontribute to public goods provision, the influence of general and kinship trust, and the role of the perception of security ofland property rights.

Trust has mixed effects on public goods provision. Farmers care about both their economic and environmental benefits. As aconsequence, both general trust and kinship trust may have two opposite effects on public goods provision.

On the one hand, farmers may be more likely to contribute to public goods if they care more about these environmental goods;we labeled that the public goods effect. This effect will appear only when people recognize the existence of public goods. Ourresults show that general trust has a strong significant public goods effect: high general trust makes that people care more aboutand invest in local public goods, and believe others will do the same. Such trust thus is more likely to contribute to public goods (inthe special case of the SLCP: not reconverting the land). Compared to general trust, kinship trust is more inward-looking and self-or group-interested (group with kinship relationship), as it refers to smaller groups and their – limited – interests. Unlike generaltrust, kinship trust has no public goods effect.

On the other hand, farmers are more likely to invest in their own private goods (or family goods) if they care more about theirown economic welfare; we labeled that the private goods effect. Both general trust and kinship trust have significant private goodseffects. Contrary to our expectation, general trust even has a stronger private good effect than kinship trust. In pursuing their owneconomic benefit, farmers with either high general trust or high kinship trust are more likely to invest in private goods (thatmeans: reconverting the land), and less likely to contribute to public goods.

The final effect of these contrasting mechanisms depends on the size and direction of these two effects. As the results show inTable 7, contrary to what the literature suggests, we find that trust, especially kinship trust, tends to be detrimental to public goodprovision in the SLCP case of Ningxia in China. A possible explanation is that the economic benefit of land reconversion is quiteessential for farmers in a poor area like Ningxia after government subsidies stop. The underlying mechanism is left for futureresearch.

Because investing in a private good (land reconversion) is more economically profitable in our special case (the SLCP inNingxia). Feelings of secure land property rights are expected to stimulate people to invest in their own private goods. As a result,higher property right security has no public goods effect, but only a private goods effect. In the SLCP case, the private good effectmeans reconverting the land. So, secure land rights make farmers more likely to reconvert land and less likely to contribute topublic goods.

Acknowledgement

This research was carried out within the research project SURE (SUstainable Natural REsource Use in Rural China, http://sure.ernasia.org/), which is funded by the Royal Netherlands Academy of Arts and Sciences (KNAW), grant 08-PSA-E-02, and theChinese Ministry of Science and Technology (MoST), grant 2008DFA90630, as part of the Programme Strategic Scientific Alliances.We thank Nico Heerink, Erwin Bulte, Michael Bennett, Wei Qu, all participants of the 2008 SURE Conference, and especially twoanonymous referees for their useful comments and suggestions that helped us to improve the paper. The authors remainresponsible for the remaining errors.

Appendix A

Descriptive statistics.

Variable Mean Std. dev. Min Max

Likelihood of not reconverting land 2.54 1.21 1 5Participation of SLCP 0.941 0.236 0 1Dummy for environmental awareness 0.687 0.455 0 1Minority 0.340 0.467 0 1Gender of household head 0.967 0.175 0 1Age of household head 41.8 10. 8 18 80Years of education of household head 5.22 3.38 0 16Number of members 4.52 1.55 1 9Labor/member ratio 0.662 0.229 0.1 1Log(income per capita) 6.62 1.14 2.53 9.69Property security: land safe feeling 2.83 0.920 1 4Non-converted land per capita (mu) 3.83 2.17 0.3 20Dummy for enough grain 0.220 0.407 0 1Years of knowing the SLCP 5.07 1.18 1 8General trust 0.00 1.00 −2.52 2.54Kinship trust 0.00 1.00 −7.11 1.18Dummy for county 1 0.347 0.477 0 1Dummy for county 2 0.338 0.474 0 1

510 Q. Tu et al. / China Economic Review 22 (2011) 499–511

References

Anderson, L., Mellor, J., & Milyo, J. (2004). Social capital and contributions in a public-goods experiment. The American Economic Review, 94(2), 373–376 [Papersand Proceedings].

Beck, U. (1986). Risikogesellschaft: Auf dem Weg in eine andere Moderne (Risk Society: On the Way Toward Another Modernity). Frankfurt am Main: Suhrkamp.Beck, U., Giddens, A., & Lash, S. (1994). Reflexive modernization: Politics, tradition and aesthetics in the modern social order. Cambridge: Polity Press.Bennett, M. (2008). China's Sloping Land Conversion Program: Institutional innovation or business as usual? Ecological Economics, 65(4), 699–711.Bouma, J., Bulte, E., & van Soest, D. (2008). Trust and cooperation: Social capital and community resource management. Journal of Environmental Economics and

Management, 56(2), 155–166.Cox, J., Ostrom, E., Walker, J., Castillo, J., Coleman, E., Holahan, R., et al. (2009). Trust in private and common property experiments. Southern Economic Journal, 75

(4), 957–975.Demurger, S. (2001). Infrastructure development and economic growth: An explanation for regional disparities in China? Journal of Comparative Economics, 29(1),

95–117.Desdoigts, A. (1999). Patterns of economic development and the formation of clubs. Journal of Economic Growth, 4(3), 305–330.Dorward, A., Kidd, J., Morrison, J., & Urey, I. (2004). A policy agenda for pro poor agricultural growth. World Development, 32(1), 73–89.Edmonds, R. (2005). Great western development in the Ningxia Hui Autonomous Region: A cultural remake of a cultural remake? Provincial China, 8(2), 144–163.Fei, X. (1992[1948]). From the soil: The foundations of Chinese society. Berkeley: University of California Press [Chinese version of his book was first published in

1948].Fukuyama, F. (1995). Trust: The social virtues and the creation of property. London: Penguin.Giddens, A. (1984). The constitution of society. Cambridge: Polity Press.Giddens, A. (1990). The consequences of modernity. Cambridge: Polity Press.Glaeser, E., Laibson, D., Scheinkman, J., & Soutter, C. (2000). Measuring trust. Quarterly Journal of Economics, 115(3), 811–846.Groom, B., Grosjean, P., Kontoleon, A., Swanson, T., & Zhang, S. (2010). Relaxing rural constraints: a ‘win–win’ policy for poverty and environment in China? Oxford

Economic Papers, 62(1), 132–156.Hyde, W., Belcher, B., & Xu, J. (Eds.). (2003). China's forests: Global lessons from market reforms. Resources for the Future: Washington D.C.Inglehart, R. (1999). Trust, well-being and democracy. In M. Warren (Ed.), Democracy and trust. Cambridge: Cambridge University Press.Knack, S. (2001). Trust, associational life and economic performance. In J. Helliwell (Ed.), The contribution of human and social capital to sustained economic growth

and well-being: International symposium report (pp. 172–202). Quebec: Human Resources Development Canada (HRDC) [2001].Knack, S., & Keefer, P. (1997). Does social capital have an economic payoff? A cross-country investigation. Quarterly Journal of Economics, 112(4), 1251–1288.de Krom, M. P. M. M., & Mol, A. P. J. (2010). Food risks and consumer trust: Avian influenza and the knowing and non-knowing on UK shopping floors. Appetite, 55

(3), 671–678.Leonard, T., Croson, R. T. A., & de Oliveira, A. C. M. (2010). Social capital and public goods. The Journal of Socio-Economics, 39(4), 474–481.Nooteboom, B. (2002). Trust: Forms, foundations, functions, failures and figures. Cheltenham: Edward Elgar.Messick, D., Wilke, H., Brewer, M., Kramer, R., Zemke, P., & Liu, L. (1983). Individual adaptations and structural change as solutions to social dilemmas. Journal of

Personality and Social Psychology, 44(2), 294–309.NBS (2006). China's statistical yearbook 2006. Chinese National Bureau of Statistics. Beijing: China Statistics Press.Ningxia Forestry Bureau (2006). The achievements of SLCP in Ningxia 2000–2005 (in Chinese). .Ningxia DRC (2006). Implementation of ‘Wujiehe’ measures and development of off-farm industries. Ningxia Development and Reform Commission, June, 2006 (in

Chinese).Ningxia Poverty Alleviation Office (2006). Introduction on poverty alleviation in Ningxia, April 20, 2006 (in Chinese). .Ostrom, E. (2009). Building trust to solve commons dilemmas: Taking small steps to test an evolving theory of collective action. In S. Levin (Ed.), Games, groups, and

the global good (pp. 207–228). New York: Springer.Ostrom, E., & Walker, J. (Eds.). (2003). Trust and reciprocity: interdisciplinary lessons from experimental research. New York: Russell Sage Foundation.Parks, C., & Hulbert, L. (1995). High- and low-trusters' responses to fear in a payoff matrix. Journal of Conflict Resolution, 39(4), 718–730.Putnam, R. (2000). Bowling alone: The collapse and revival of American community. New York: Simon and Schuster.SFA (2003). Guidance and practices of SLCP. State forestry administration. Beijing: Chinese Agricultural Science and Technology Press [in Chinese].Starrett, D. (2003). In K. Mäler, & J. Vincent (Eds.), Property rights, public goods and the environment. The handbook of environmental economics, Volume I. (pp.

97–126) The Netherlands: Elsevier.Tu, Q., & Bulte, E. H. (2010). Market participation and economic outcomes: Evidence from rural China. World Development, 38(8), 1179–1190.Uchida, E., Xu, J., & Rozelle, S. (2005). Grain for green: Cost-effectiveness and sustainability of China's conservation set-aside program. Land Economics, 81(2),

247–264.Uslaner, E. (2002). The moral foundations of trust. Cambridge: Cambridge University Press.de Vos, B., & Mol, A. P. J. (2010). Changing trust relations within the Dutch fishing industry: the case of national study groups. Marine Policy, 34(5), 887–895.Walker, J., & Ostrom, E. (2009). Trust and reciprocity as foundations for cooperation. In K. Cook, M. Levi, & R. Hardin (Eds.), Whom can we trust?: How groups,

networks, and institutions make trust possible (pp. 91–124). New York: Russell Sage Foundation.Woolcock, M. (1998). Social capital and economic development: Towards a theoretical synthesis and policy framework. Theory and Society, 27(2), 151–208.WWF (World Wildlife Fund) (2003). Report suggests China's ‘grain-to-green’ plan is fundamental to managing water and soil erosion. cited on May 31, 2005, in:.

http://www.wwfchina.org [March 31, 2003]Xu, J., Katsigris, E., & White, T. A. (Eds.). (2001). Implementing the natural forest protection program and the sloping land conversion program: lessons and policy

implications. CCICED-WCFGTF (China Council for International cooperation on environment and development, western China forests and grassland task force).Beijing: China Forestry Publishing House.

Xu, J., Tao, R., Xu, Z., & Bennett, M. (2010). China's sloping land conversion program: Does expansion equal success? Land Economics, 86(2), 219–244.Xu, Z., Bennett, M., Tao, R., & Xu, J. (2004). China's sloping land conversion program four years on: Current situation. International Forestry Review, 6(3–4), 317–327.Yeh, E. T. (2004). Thinking about globalization and China's environment: The case of western China. 2004 Mansfield Conference. Missoula: The University of

Montana.Zhang, L., Tu, Q., & Mol, A. P. J. (2008). Payment for environmental services: The sloping land conversion program in Ningxia Autonomous Region of China. China &

World Economy, 16(2), 66–81.

511Q. Tu et al. / China Economic Review 22 (2011) 499–511