Research

INTRODUCTION

Conserving forest resources is an important policy objective in China, both as a means of wildlife conservation, but also for the valuable ecosystems services that forests provide, such as moderating extreme weather events, water purification, and soil retention (Li et al. 2022). China has implemented many national policies to address deforestation over recent decades, such as logging bans, afforestation programs, and programs aimed at converting cropland to forests, such as the Natural Forest Conservation Project (NFCP) and Grain to Green Program (GTGP). Some evidence suggests that these efforts have been moderately effective in controlling soil erosion and floods and increasing forest coverage (Liu et al. 2008). However, the successful restriction of logging and agriculture may have spillover effects in encouraging other economic activities, such as enhancing the role of livestock production as a prevalent source of income in many of China’s biodiversity hotspots. Evidence suggests that the growing number of livestock has seriously threatened the ecological environment, imperiling the wildlife in these areas (Li et al. 2017). Of particular ecological and cultural value is the giant panda (Ailuropoda melanoleuca), which lives only in six mountain ranges in China.

The giant panda depends on the integrity of forest ecosystems with abundant bamboo resources for its survival. The conservation of giant panda habitats also protects the centers of other endemic species in China, providing US$2.6-6.9 billion/year of ecosystem service values (Li and Pimm 2016, Wei et al. 2018). With increasing human disturbances such as land-use changes and climate change, giant panda habitats have become increasingly fragmented, resulting in an increasing number of isolated populations (State Forestry Administration of P.R. China 2015, Xu et al. 2017). Livestock grazing has emerged as one of the most prevalent human disturbances impacting panda habitats (Li et al. 2017). In the area most intensively used by free-ranging cattle and horses, livestock grazing impacts giant pandas by degrading bamboos and reducing available habitats by 34% (Li et al. 2017). In 2017, China announced the establishment of a pilot national park for giant pandas that spans parts of three provinces and incorporates several existing protected areas as well as some areas adjacent to these fragmented protected areas (Li 2020). The park was changed from the pilot stage to the official establishment in 2021. There are hopes that this national park would not only facilitate the conservation of the species and the re-connection of fragmented habitats but also provide sustainable livelihoods for local communities because of the greater potential for ecotourism and other income sources (Huang et al. 2020). Nonetheless, as more residential areas are included in the protected area, the potential for conflicts between natural resource use and conservation could be increasing. If livestock grazing remains a pressing issue in and around the national park, it will not only harm the giant panda habitat but also damage the potential economic benefits that could accrue to the local people in the long run due to the degradation and loss of natural capital. Thus, there is an important need to control the number of livestock grazing in forests to protect key giant panda habitats and to promote local sustainable development. This involves natural resource management, protected area management, and community-based conservation, in which cooperation and coordination play important roles (Meinzen-Dick 2009).

As a classic example of common property resources (CPRs), forests are generally subject to problems of illegal logging, rapid deforestation, and natural resource depletion (McKean 2000). These resources share the non-excludability feature of pure public goods, yet the consumption of these resources necessarily implies there is less available for the enjoyment of others. As is often discussed in the literature on common property resource management (e.g., Ostrom 1990), these resources frequently entail a social dilemma in which individual motives are contrasted with socially optimal actions. Classically, it has been postulated that it is irrational for an individual to voluntarily contribute to the conservation of a shared good or resource unless they could be excluded from future enjoyment from that good, thus leading to overexploitation of the resource and resulting in a “tragedy of the commons” (Hardin 1968).

A body of literature has evolved studying individuals’ behavior when confronted with such social dilemmas, often viewing the decision-making landscape through a game-theoretic lens (e.g., Dawes et al. 1977, Seabright 1993, Ostrom et al. 1994). Poteete et al. (2010) integrated various works in this field by different scholars, showing the values and limitations of each and the need for multiple approaches. In moving from theory to empirical observation, researchers have relied on experimental games (e.g., Isaac et al. 1984, Ostrom et al. 1994, Andreoni 1995, Cárdenas et al. 2017). The majority of these studies rely on experimental methods in sterile laboratory environments. The most commonly used experimental methods are referred to as “public goods games,” so named because they present participants with the social dilemma between pursuing a private optimum and achieving a social optimum. These games often take the form of repeated games and typically find initial voluntary contribution rates of 40-60 percent of participants’ initial endowments, declining over time as participants “learn” their dominant strategy. Researchers consider a range of possible determinants of behavior in these games, including age, income, social information (information about others’ decisions), the group size, the level of trust, and risk preferences (Smith et al. 1995, Gächter et al. 2004, Zhou and Song 2008, Shang and Croson 2009, Qin et al. 2011). Over the course of these previous studies, there have been several stylized facts that have emerged regarding individual behavior in these settings: (1) there is a negative effect on voluntary cooperation when the groups collectively face risks in managing shared resources, but not when the risks are limited to individual actions (Cárdenas et al. 2017, Ward et al. 2019); (2) each participant’s expectations about the behavior of others is a critical determinant of successful cooperation (Offerman et al. 1996, Fischbacher and Gächter 2010, Cárdenas et al. 2017); and (3) overall tendencies toward trusting behavior (whether that be trust in institutions more generally or specifically trust in one’s peers) are positively correlated with increased rates of voluntary cooperation (Qin et al. 2011). Although some results appear robust across a number of studies, such as the positive effect of increasing the benefits from cooperative outcomes, there is conflicting evidence regarding the influence of some other factors. For example, previous studies reported that risk preferences played a significant role in social dilemma settings. However, no consensus has been reached: some studies found a positive effect (Offerman et al. 1996), whereas others found a negative effect (Teyssier 2012), or no effect at all (Kocher et al. 2015).

The different findings across these assorted studies suggest that cooperation is a context-based phenomenon. The vast majority of these existing studies are laboratory-based studies, usually with college students or, more generally, urban residents, but there are legitimate concerns about the generalizability of many of the existing laboratory-based studies to real-world contexts of shared resource management, especially in rural areas. There have been a few studies conducted among rural residents in developing countries that are noteworthy. For example, Sirán et al. (2006) studied how changes in income level affect people’s hunting behavior in an Indigenous community in the Amazon. Janssen et al. (2013) performed experimental work in Thailand and Colombia with different types of communities and student participants. However, relatively few empirical studies using similar experimental methodologies have been conducted among rural residents in China. As Henrich et al. (2006) suggested, because the results of behavioral experiments vary substantially across populations, this is in many ways a missed opportunity because many of the most salient applications of these insights pertain to shared natural resources.

Also of relevance is the nature of the pro-social attitudes that are being studied, and different variations of public goods games reveal different insights based on design specifications. In this study, we compare results from two common types of public goods games: the voluntary contribution mechanism (VCM) and the provision point mechanism (PPM). Both variations enable participants to voluntarily contribute to the creation of a shared good but differ in the production function used to translate voluntary contributions into the shared good. In a general VCM, the amount of the resource that is provided and shared among all members of the community is a linear function of the voluntary contributions of the community members (Isaac and Walker 1988). In a PPM, on the other hand, the production function is a piecewise function such that community members can only obtain the benefit of the shared resource as long as the threshold level of voluntary contributions (a provision point) is achieved (Bagnoli and Lipman 1989). As such, there is a greater incentive for participants to coordinate, in addition to cooperating. Consequently, the VCM is frequently used to study patterns of cooperation, whereas the PPM is more frequently used to study coordination. In other words, the VCM introduces participants to a scenario in which they are confronted with the choice between pursuing purely self-motivated aims or cooperating to achieve an outcome that is more socially beneficial, while the PPM presents an additional complication in that participants must somehow coordinate their levels of cooperation such that the provision point is successfully attained. The PPM is generally more efficient (in terms of generating greater social welfare) than the VCM and can increase total contributions (Rondeau et al. 2005). However, participants in PPMs have frequently been observed to fail to attain the provision point when the potential gains from the shared public good were relatively small. Further, the risk of failing to achieve the social optimum decreased with the benefit-cost ratio (Rondeau et al. 2005).

This study aims to fill this important research gap by reflecting on the results from previous studies and applying them to the specific context of rural China to address the following questions: (1) What is the propensity among rural residents for cooperating and coordinating in the management of common property resources in communities depending on livestock grazing in or around giant panda habitats? (2) Are individual attitudes toward cooperation and/or coordination influenced by an overall sense of social cohesion? And (3) how does variation in socioeconomic characteristics at the individual or societal level explain differences in cooperative or coordinating behavior among these rural residents? We attempt to answer question (1) by conducting a series of framed public goods games, including VCM and PPM, among rural communities in Baishuijiang and Wanglang nature reserves in Gansu and Sichuan provinces, respectively. We explored questions (2) and (3) by conducting a questionnaire and analyzing participants’ social attributes with the results of the game. Because many farmers with livestock continue to graze their animals inside the boundaries of the two nature reserves in our sample, there might be concerns that these individuals might be less likely to contribute to the collective management of forest resources. As the engagement of local communities in collective livestock management inside the national park is increasingly needed, observing such farmers’ behavior in these experimental games might shed some light on how they may behave in real-world situations and provide evidence to support the formulation of effective management plans and policies.

MATERIALS AND METHODS

Study sites

The experiments were conducted in Baishuijiang National Nature Reserve (hereafter, Baishuijiang) in Gansu province and Wanglang National Nature Reserve (hereafter, Wanglang) in Sichuan province in central China (Fig. 1). Despite being situated within two different provinces, the nature reserves are quite close. Both nature reserves are in the Min Mountains, which have the most abundant wild giant panda population in China (State Forestry Administration of P.R. China 2015). Baishuijiang was founded in 1978 with a total area of 223,671 hectares. In the Baishuijiang area, residents are composed of Han ethnic majority and several ethnic minorities, including Baima Tibetan and Hui (Ting et al. 2012). Wanglang was established in 1965 with an area of 32,300 hectares primarily to protect the giant panda and other endangered species as well as their habitats (Li et al. 2012). There are no residents in Wanglang, but it borders Baima Tibetan Township (Li et al. 2017). Baima Tibetans used to graze livestock inside Wanglang and in the surrounding panda habitat (Li et al. 2017). All nature reserves in China have three zones, the experimental zone in which all the human residence and subsistence activities are concentrated, the buffer zone in which most economic and productive activities are prohibited, and the core zone in which entry is not allowed except with official permission (Ting et al. 2012).

Data collection

In each of these nature reserves, we identified regions in which livestock production was a major income source. For Baishuijiang, we visited five communities near the Baimahe Region and eight communities near the Danpu Region. These areas were identified as the areas with the most critical grazing problem in Baishuijiang according to the pre-survey conducted in 2018. For Wanglang, we visited 12 communities across 4 administrative villages in the Baima Tibetan township.

The research team sought the assistance of local village cadres (leaders) to maximize the recruitment of available villagers. We preferred participants who were capable of understanding the rules of the game and of acting in a way consistent with their underlying preferences. Most of the participants from around the Wanglang National Nature Reserve were of the Baima Tibetan ethnic minority, some of whom could not speak Mandarin. For these participants, we had translators who were able to transcribe the rules of the game into the local dialect and verify that the participants could fully understand the rules of the game. We recruited a total of 287 participants, 191 from the 13 selected communities near Baishuijiang, and 96 from the 12 selected communities near Wanglang. Fewer participants were recruited in Wanglang because of the smaller population. The experimental games were conducted from July 5 to July 25, 2019. Because the cadres’ homes are usually used as the information exchange center for local communities, we conducted most of the experiments there. Although we forbade communication among participants during the experiment, participants in some groups were allowed to explain game mechanics to each other in their local dialects (because members of the research team did not speak the local dialect) to ensure that all participants were able to comprehend the game instructions.

To explore how variations in socioeconomic factors may explain differences in contribution preferences of players in rounds, each participant was also asked to complete a survey before the game. Data collected through this survey included demographic characteristics, livestock grazing information, the perception of giant panda conservation, trust among community members, household income, and consumption. Among individual demographic characteristics, each participant provided information on their age, gender, religion, ethnicity, household size (the number of adults and children living within the homestead and sharing household income), level of education, (specifically for how many years they attended school), and employment (including both on- and off-farm employment). Table 1 reports the summary statistics for the participants in our study.

About 56% of participants were males with an average age of 45.7. About 58% of participants were members of the Han ethnic group, with the remaining 42% being of the Baima Tibetan ethnic community. The ethnic composition of samples from the two nature reserves were markedly different, with the villages around Baishuijiang Nature Reserve consisting of mostly Han (86.4%), and villages around Wanglang consisting of mostly Baima Tibetan (97.9%). Most participants considered themselves to be non-religious, but there were small pockets of Buddhists, Christians, and those who practice nature worship (6.3%, 3.1%, and 19.2%, respectively). This latter group is especially prevalent in Wanglang Nature Reserve, representing 57.3% of the sample from that region. Most participants lived in a household of about four members, and about half of the participants had no more than primary education. On average, around 36.2% of participants raised livestock. Among the participants from near Wanglang Nature Reserve, nearly 70% reported raising livestock, compared to only about 20% among participants near Baishuijiang Nature Reserve. Among those who currently raised livestock, households near Wanglang tended to rely more heavily on cattle, with an average of roughly 26 head of cattle, compared with only 4 head of cattle among those near Baishuijiang. The latter tended to rely more heavily on sheep, with an average of about 36 sheep, compared to only 6 sheep among those near Wanglang. Across both nature reserves, people tended to view members of their communities as generally trustworthy, and about 30% of participants from around each nature reserve expressed that the protected area designation limits some of their production. People generally viewed pandas positively and thought the presence of pandas in their nearby nature reserves could potentially enhance their income through ecotourism. Figure 2 indicates a large amount of participants’ income was from tourism, grazing, working outside, and a stable job. There was a substantial difference in the composition of residents’ income between the two sites. For Baishuijiang, most of the residents’ income came from tourism, followed by working outside and stable jobs. For Wanglang, the off-farm job was the main source of income. Second was jobs in other counties and grazing, with a similar proportion.

To arrive at a reliable estimate for household annual income, we asked participants to list their annual income from the following sources: agriculture, tourism, livestock husbandry, herb collection, bee and honey, compensation from government policies, work outside, financial support by family members, stable jobs, and their total family income. We also asked participants to identify their top three sources of income and the economic activities their family intended to develop in the future. Figure 2 reports the breakdowns of income sources for the full sample as well as for the two nature reserves.

The differences in sources of household income between the two reserves were mainly in tourism, grazing, agriculture, and working outside due to variations in ethnic and traffic conditions. Wanglang was less accessible for nearby towns compared to Baishuijiang.

One indicator of the social fabric of the community is the degree of trust that an individual holds to other members and to the community as a whole. Castillo et al. (2011) explored how experience influenced participants’ decisions in Colombia and Thailand pertaining to the management of fishery resources. Interestingly, they did not find high levels of trust among local fishermen as a sufficient condition for resource sustainability. Janssen et al. (2013) expected that greater trust in other community members leads to greater cooperation based on the general findings of Poteete et al. (2010), however, their results showed that the higher the trust among the villagers, the less social pressure the villagers would abide by the external rules, and the greater their confidence was in breaking the rules.

To capture the essence of interpersonal trust and obtain an estimate of trust with other community members, we incorporated a short module in our survey that allowed participants to indicate the degree to which they agreed with some statements about their community (Cárdenas et al. 2017). Responses were coded via a Likert Score with scores ranging from (1) strongly disagree to (5) strongly agree. The three statements to which participants responded were:

1. Most people in this community are honest and can be trusted.
2. If a mother in this community has an emergency and needs to leave her baby for the day, she will easily find someone in this community she can trust with her baby.
3. If a neighbor in this community lends some money to another neighbor, it is very likely that the lender gets her money back.

Based on individuals’ responses to these three questions, we constructed an equal-weight index as a measure of trust. As seen in Table 1, the overall level of trust in the sample area was quite high, with an average index of 13 (out of a possible maximum of 15), indicating strong agreement with these three statements, with no difference in the level of trust between the two study regions.

Experimental Games

To study cooperation and coordination, we used a series of experimental games, specifically, variants of public goods games. Public goods games are a series of experiments expressly designed to test related aspects of the free-rider problem, a specific type of market failure that arises when individuals lack the proper private incentives to participate in collective action toward the provision of a good or service that benefits all members of a society without exclusion (Maxwell and Ames 1981). There are several variants of public goods games, and although they differ in some important ways, they all present individual decision makers with a social dilemma in which they must choose between pursuing a private dominant strategy or a social optimum.

In a traditional VCM, an individual must choose between voluntarily contributing all or part of their endowment toward the production of a public good that yields benefits shared by all players, or consuming their endowment and earning personal gratification. The public good production function is most often specified as a linear function of voluntary contributions from all group members. In our experiment, to simplify the choice situation and to better reflect the “all or nothing” nature of cooperation in resource management, each individual’s choice was limited to a binary contribution decision. In other words, each participant was endowed with a coupon that they could either keep or invest in a project, i.e., the public good. Under this structure, each individual’s payoff function is given as:

(1)

where each player has a choice of two options, x_i = {0,1}: keep their endowment for private consumption (x_i = 0) or invest their endowment in the community project (x_i = 1). Retaining the endowment (x_i = 0) yields a private return of p to player i only, whereas the total contributions to the community project yield a marginal per capita return (MPCR) of a to each player, regardless of whether a given player contributed to the project. That is, the total value of the public good is equal to the sum of the contributions of all group members, and this total value is then distributed evenly to all group members. The ratio of the MPCR to the return of the private option is a/p, so as long as MPCR < 1, it is irrational for a player to voluntarily contribute to the community project, and therefore the Nash (dominant) strategy will be x_i = 0 for all i, so that the Nash equilibrium is one in which each individual retains their endowment and and no public good is provided. Any deviation from this Nash equilibrium reflects a deviation from purely self-interested behavior and is thus consistent with cooperative behavior. Contrary to most previous research implementing these types of games, which assumed that participants in the game derived utility from private consumption and consumption of the community good, we allow for an altruism function (g), i.e., to also contribute to individual utility. This adjustment can effectively map others’ enjoyment of the benefits of the community good to individual utility. It is perhaps useful to think of g as reflecting individual i’s indirect enjoyment of the community good, or i’s indirect enjoyment of j’s (j not equal to i) direct enjoyment of the community good. This modification to the general model allows for individual payoffs to better reflect observations that individual engagement in conservation programs is rarely only determined by direct monetary reward from conservation, but also entails considerations of shared costs and benefits, including the distribution of risks within the community.

We conducted the VCM over three separate rounds, primarily varying in the degree of risk associated with the returns on private consumption or on contributions to the community project. In our particular design, the private return in the first (baseline) round was a certain RMB¥50, whereas the public return for each coupon (i.e., the MPCR) was a certain RMB¥10 (equivalent to US$1.45 at the time of the study). For our target group size of 8 participants, full cooperation would yield RMB¥80 for each participant, whereas universal defection would leave each participant with only RMB¥50. The ratio of the marginal return on the public good to the return of the private option was 1/5 = 0.2. Therefore, participants faced a clear social dilemma in which the dominant strategy was for all participants to keep the coupon and earn RMB¥50, but a socially optimal strategy would entail all participants contributing to the group project to each earn RMB¥80.

The second and third rounds were designed to test the roles of two specific risks faced by community members in cooperative behavior: private risk in payoffs associated with personal earnings (such as wage labor) and collective risk in payoffs associated with the shared resource (such as benefits from common property resources, shared infrastructure, etc.). In the round incorporating private risk, participants faced the choice of investing in the community project with a certain MPCR of RMB¥10 or retaining the coupon for personal consumption and facing a 50/50 probability of a double-or-nothing private return (RMB¥100/RMB¥0), depending on the outcome of a fair coin toss after all participants had made their decision. In the collective risk round, participants had to decide between retaining their coupon for private consumption and earning a certain RMB¥50 or investing the coupon in the group project and facing a 50/50 probability of a double-or-nothing MPCR (RMB¥20/RMB¥0) for each coupon invested by the group. Importantly, as Ward et al. (2019) pointed out, this latter scenario introduces two sources of uncertainty when contributing to the community project: one associated with the MPCR on contributions to the community project and one associated with the overall level of cooperation in the game.

The fourth round differed from the first three rounds in that it was structured as a PPM rather than a VCM. Structurally, it was like the baseline round in that there was no uncertainty with regard to either the returns on private consumption or in the MPCR to contributions to the group project. One fundamental difference in structure, however, was the presence of a provision point such that the community project would only yield shared benefits if contributions to the community project exceeded a specified threshold. This mechanism has clear real-world analogies, such as community-based payments for ecosystem services (PES) programs, for which conservation payments would only be made by the program if conservation objectives or program commitments were satisfied. Other real-world analogues include the construction of shared infrastructure with nontrivial fixed start-up costs, in which it would be unwise to begin construction until the community could be relatively assured they would eventually be able to fund construction to its completion. In this specific experiment, the community project would only produce shared benefits if more than half of the participants in a given group opted to make voluntary investments. With a group consisting of eight members, if fewer than four players chose to voluntarily contribute, the community project would be valued at RMB¥0. However, if at least four players chose to invest, total contributions to the community project would generate benefits with an MPCR of RMB¥20. Players who kept their coupons could always get a fixed private return of RMB¥50. The payoff function is given as:

(2)

where, as before, a is the MPCR on contributions to the group project (RMB¥20) and p is the return on private consumption of the coupon (RMB¥50). Here, as before, we allow for individual decisions to be at least partly influenced by their consideration of others’ enjoyment of the shared community project (denoted by the altruism function g). In addition to differences in the public good production function, the PPM also differs from the VCM in terms of the number and nature of equilibria. Whereas the VCM has only a single Nash equilibrium (with no participants contributing to the creation of the public good), the PPM has multiple efficient Nash equilibria, wherein the public good is exactly provided (i.e., the contributions to the creation of the public good are exactly sufficient to meet the provision point). Because of this feature of the PPM vis-à-vis the VCM, the challenge that participants face is less about cooperation and more a challenge of coordination: which of the many equilibria should they collectively pursue?

We also examined the extent to which voluntary cooperation was conditional, or the idea that an individual’s propensity to cooperate was conditioned on expectations of peers’ cooperation. A common finding in related studies (e.g., Cárdenas et al. 2017, Ward et al. 2019) is that an individual’s decision to voluntarily contribute to the community project is positively correlated with their expectations about the level of contributions of others in their group or community. Cárdenas et al. (2017), for example, found nearly a 1:1 relationship between the average fraction of community project investment and the expected level of contributions by others in the baseline (no risk) round. Ward et al. (2019) found a similar relationship, though of a slightly lower magnitude. Ward et al. (2019) also found that these expectations are an indicator of more broadly held attitudes of social cohesion, which can be at least partly influenced by external pro-social interventions, such as programming aimed at promoting socioeconomic improvements at the community level. We elicited participants’ beliefs about others’ contributions by asking each of them to predict the fraction of players in their group that they expected to invest in the community project in each round of the game, immediately following their own contribution decision. To avoid concerns about endogeneity, we first converted participants’ responses into a ratio of each individual’s expectations of peer cooperation by subtracting their contribution (if they contributed) and dividing this by the number of other participants in the group (in most cases, 7).

Data analysis

To isolate the effects of various game characteristics and socioeconomic factors on individuals’ voluntary binary contribution decisions in the VCM and PPM games, we estimated a series of linear probability models, in which the dependent variable was the individuals’ binary contribution decision. These models were estimated using least-squares methods, with standard errors adjusted for the clustered nature of the experiments. We ran regressions separately for the VCM rounds (rounds 1-3) and the PPM round (round 4), given the fundamental differences in the game structures and the implicit focus on cooperation or coordination. In our base model, we considered how the nature of risk would affect individuals’ voluntary contributions to the community project. The base model is given as:

(3)

with two primary explanatory variables x₁ (private risk) and x₂ (collective risk) as well as a binary dependent variable y_i = {0,1}:keep or invest the coupon in the community project, where

(4)

(5)

Then, individuals’ expectations of peer contributions and the level of trust in the community were added to the base model both separately and jointly to examine their effects on cooperative behaviors. After that, grazing activities and attitudes toward protected areas were added on top of the aforementioned factors after controlling for the influence of other potential confounding factors.

RESULTS

Voluntary contributions

Table 2 presents the percentage of participants making voluntary contributions to the community project in each of the four rounds of the games, both in total and by location (i.e., the nature reserve to which participants are nearest). On average, our results suggest rather high levels of voluntary contributions, with between two-thirds and three-quarters of participants voluntarily contributing to the community project rather than consuming their endowment. Across the three rounds of the VCM, voluntary contribution levels were remarkably similar to those found by Cárdenas et al. (2017) in their Chinese sample (part of which included villages in Sichuan province), in which they found an overall contribution rate of nearly 73%. As expected, and consistent with related studies (e.g., Cárdenas et al. 2017, Ward et al. 2019), voluntary contribution rates were highest in the baseline round (nearly 74%t of participants contributed) and lowest in the collective risk round (only 63% contributed to the community project). This, too, is consistent with expectations because this round incorporated uncertainty in the MPCR to contributions to the community project, in addition to uncertainty around the total level of contributions to the community project.

Across all rounds of the game, the rate of voluntary contributions was higher among villagers near Baishuijiang Nature Reserve than those near Wanglang Nature Reserve, though the differences in contribution rates were not statistically significant. However, in both regions, the same general pattern emerged. Contributions were highest in the baseline round, followed by the round in which there was uncertainty in the returns to private consumption of the endowment, then the provision point round, and finally the round in which there was uncertainty in the MPCR to contributions to the community project (Table 2). At least anecdotally, these results suggest that the nature of risk would be an important consideration for individuals in deciding whether to cooperate with others in their community in the production of the community project (Cárdenas et al. 2017. Ward et al. 2019). In the baseline round, for example, the only source of risk was the level of contributions of the other group members. Given the high levels of trust in these communities (Table 1), it is reasonable that individuals are willing to make these investments in the community project. However, participants did not seem to view the community project as a less risky option in the second round, in which there was a significant risk associated with the returns to private consumption. However, there was also no convincing evidence that participants altered their behavior to any appreciable degree in this scenario because the contribution rates were statistically indistinguishable from the contribution rates in the baseline round.

Despite the average contribution rate being lower in the PPM than in either the baseline round or the private risk round, participants were quite successful in coordinating to ensure the provision of the community project, with 86% of groups overall having sufficient voluntary contributions to achieve the provision point. Again, we see differences between groups from villages near Baishuijiang and those near Wanglang. On average, our results suggest that the villages near Wanglang were more successful in achieving the provision point than those near Baishuijiang, by a margin of 92% to 83%, although the difference was not statistically significant. Because the observations were now at the group level rather than the individual level, these success rates were not measured very precisely, and thus the difference in achieving the provision point was not statistically significant.

Impacts of risks on cooperation

Table 3 reports the results of the linear probability model regression for the three rounds of the VCM. We present several alternative specifications, and in each of these specifications we control for individual and community characteristics that might confound the effect of our primary explanatory variables of interest. Column (1) shows our base model results. It indicates the fragility of voluntary cooperation to the presence of risk and specifically the collective risk shared by all members of the community. In particular, we found that the presence of collective risk lowered the probability of individuals making voluntary contributions to the community project by roughly 10% (p = 0.006). Although the presence of risk of private activities reduced voluntary contributions to the community project by 3%, this effect was only marginally statistically significant (p = 0.16). This is consistent with Cárdenas et al. (2017) in their Chinese sample, though contrary to Ward et al. (2019), who found that the presence of risk in private activities increased community contributions, with community investments likely perceived as a “flight to safety” vis-à-vis risky private activities. Importantly, these effects were consistent across various model specifications (i.e., across all columns in Table 3), both in the direction and magnitude. The collective risk reduced voluntary cooperation by 9.9% to 10.6% across all 4 specifications. Further, these results suggest that the combination of private and community risks could additively reduce cooperative efforts in managing community resources or projects. When there were risks in the returns to both private activities and community projects, which is likely in reality, the voluntary cooperation toward the community projects was reduced by more than 14% (p = 0.03).

Impacts of trust and expectation of others on cooperation

The observed voluntary contribution could depend on expectations about others’ behavior, or as Qin et al. (2011) suggested, this could reflect the overarching level of trust in the community. Nonetheless, there is a possibility that these factors could be correlated with each other. Thus, we tested the effect of trust and whether this effect is independent of expectations of others and report the results in columns (3) and (4) of Table 3. We find very little evidence to suggest that increasing levels of trust in the community and individual community members are significantly correlated with cooperative behavior, with a coefficient of only 0.003 (p = 0.80; Table 3, column 3). However, when examining the joint effects of both trust and expectations of peers’ cooperative behaviors, there was little evidence that these two were strongly correlated. The standard errors were consistent across columns (2) and (4) and columns (3) and (4) for expectations and trust, respectively, suggesting a trivial inflationary influence of any correlation between these terms on the coefficient estimates’ variances. Although trust and expectations were positively and significantly correlated (p-value = 0.03), this correlation was trivial with a correlation coefficient of 0.074. This could be caused by the limited variation in individuals’ reported levels of trust. Nearly 95% of the respondents in our sample had a high trust index score of at least 12 (out of a possible 15).

Impacts of grazing activities and opinions about protected areas on voluntary contributions

Table 4 shows how farmers’ grazing behavior and overall opinions about the protected areas could affect their propensity toward cooperative behavior after controlling for the influence of risk in the returns to private or collective actions, expectations of others’ actions, and other potential confounding factors. We do not find that farmers that are currently grazing livestock were any less likely on average to voluntarily contribute to the community project than individuals who were not currently grazing livestock, nor did there appear to be any effect of the intensity with which they cultivate livestock (i.e., the number of different types of livestock owned). We also gauged the extent to which the participants felt that the protected area status of the forests near their homestead impeded their production capabilities, and these results also suggest that such concerns might not impede cooperative management of shared forest resources, and these results also suggest that such concerns might not impede cooperative management of shared forest resources.

Impacts of expectations and trust on voluntary contributions under a provision point mechanism

As previously mentioned, the participant groups were largely quite successful in achieving the provision point (i.e., voluntary contributions from at least half of the group members). Only 4 out of 36 groups failed to meet the provision point, with 5 groups exactly meeting the provision point (i.e., exactly half of participants voluntarily contributing). The remaining 27 groups exceeded the Nash equilibria outcome (in which the provision point was exactly satisfied) and had a higher degree of coordination than was required. However, as mentioned (Table 2), the average contribution level decreased from about 74% in the baseline round to about 70% in this game. Without the existence of uncertainty regarding either the returns to private consumption or in the MPCR, once the threshold contribution level was attained, the principal factor driving the reduction in voluntary contributions (although rather small in magnitude) must be due to the increased uncertainty about the behavior of others, and in particular about whether enough of one’s peers would be willing to contribute toward the attainment of the provision point. Indeed, when we used a linear probability model to study the determinants of contributions in the PPM round of the game, we observed that individuals’ expectations of peers’ behavior became paramount (Table 5). Here, we saw that a 10% increase in the expected cooperation of one’s peers resulted in a 2.7% increase in the probability of one’s own contribution, and this effect was again independent of the overall level of trust that one placed in their community or individual members of the community. Because contributions above the provision point have less intrinsic value, if participants expected more of their peers to contribute, they might tend to free ride. Indeed, there was evidence of some nonlinearity (not reported in Table 5), but the turning point level of expectations, beyond which the probability of one’s own contribution started to decline, was well beyond 100% of one’s peers. So, as long as expectations were within the realistic range, increasing expectations of peers’ contributions to the PPM increased the probability of one’s own contributions. As before (i.e., in the VCM), the overall level of trust had virtually no effect on the likelihood of contribution, although this again may simply be caused by little variation in trust levels among the participants in our sample.

As reported in Appendix 1 (Tables A1-A3), individual or household characteristics added little explanatory power in determining individual behavior in these experimental games. Males tended to voluntarily contribute more often than females (by about 10%), though this was only marginally statistically significant (p = 0.15). This is contrary to the results of Seguino et al. (1996), who suggested that women are more likely to cooperate in contributing to public goods than males, but similar to the results of Peshkovskaya et al. (2019), who demonstrated that men have a higher level of trust and gratitude than women. Contradictory results regarding this factor have been obtained.

Similarly, members of the Baima Tibetan ethnic minority were about 10% more likely to voluntarily contribute to the community project than members of the Han ethnic majority, and individuals from communities near Baishuijiang were about 14.5% more likely to voluntarily contribute to the community project than individuals from communities near Wanglang National Nature Reserve.

Because some differences emerged when comparing the results in the VCM and PPM, we next sought to understand apparent changes in behaviors between the baseline VCM round and the PPM round, and whether these behavioral changes could be attributed to individual characteristics. Participants were divided into three groups based on behavior changes: (1) the 29 participants who did “not” contribute in the baseline VCM round, but did contribute in the PPM; (2) the 40 participants who did contribute in the baseline round, but did “not” contribute in the PPM; (3) the other 218 participants did “not” change their decision (contributing in both the baseline VCM and the PPM or not contributing in either the baseline VCM or the PPM). We conducted a Mann-Whitney rank-sum test to compare some of the individual characteristics of these different independent samples (Appendix 1, Table A4). In breaking the total sample down into these smaller sub-groups, it was relatively more difficult to make statistically robust observations about systematic differences. However, we observed that Group (1)’s annual income was distributed differently than Group (3)’s, with a p-value = 0.076. Members of Group (1), who chose to invest in the community project in the PPM after not contributing to the community project in the VCM, had an average annual income of RMB¥26,000, compared with an average of RMB¥39,000 among members of Group (3), who did not change their behavior between the 2 rounds of the game. This suggests that poorer households may be more inclined to contribute under a PPM when there is an incentive to coordinate and ensure that group contributions are sufficient to meet the provision point, whereas wealthier households may be more inclined to contribute regardless of whether there is a threshold that the community should be targeting. There was a marginally significant difference in the distribution of religions between Group (1) and Group (3), with Group (3) having a higher concentration of Buddhists (p-value of 0.15). We also observed a marginally significant difference in the age distribution between Group (2) and Group (3). In this regard, we see that individuals who chose to consume their endowment in the PPM after contributing to the community project in the VCM (Group 2) tended to be younger, with a mean age of 43.5 years compared to a mean age of 48 years among members of Group 3 (p-value = 0.12).

DISCUSSION

Uncertainty reduces cooperation among community member peers

When exposed to risk in the enjoyment of the shared benefits of community projects, individual behavior changed compared to a no-risk baseline scenario. Specifically, collective risk led to a roughly 10% decline in the probability of cooperation compared to the baseline. This was consistent with our expectations due to the compounding of uncertainty, i.e., uncertainty pertaining to both the MPCR and the overall level of cooperation. In the baseline round, the only source of uncertainty was the contributions of others. This uncertainty carried over to the scenario in which there was also a risk in the returns to private consumption. In the private risk round, therefore, participants had to weigh two risky alternatives, and largely still viewed investments in the community project as more appealing because uncertainty in returns to private consumption did not significantly crowd out contributions to the community project. When there was uncertainty in the MPCR, the total amount of uncertainty in investments to the community project was compounded, enhancing the negative impact that such uncertainty may have on the willingness to cooperate by each player. When there was risk in the MPCR to investments in the community project, private consumption emerged as a much safer alternative.

This effect was also observed by Burger and Kolstad (2009), Cárdenas et al. (2017), and Ward et al. (2019) across multiple settings (among university students at UC-Santa Barbara in Burger and Kolstad 2009; China, Colombia, Nepal, and Thailand in Cárdenas et al. 2017; and Odisha, India in Ward et al. 2019). Indeed, it is precisely this compounded uncertainty that leads to cooperation being so fragile, especially in a reality with rapidly changing environmental conditions or with increasing variability in an interconnected world. The results in the present study highlight that increasing variability around the benefits of cooperative management of forest resources may lead to a reduction in cooperation among community members.

Alternative hypotheses to explain the observed phenomena include the “learning hypothesis” and the “strategies hypothesis,” which explain why free riding is seldom observed in single-shot games but often approximated in infinitely repeated games (Andreoni 1988). The learning hypothesis holds that participants may not fully understand the incentives of the game the first time they play it, but that they may learn their dominant strategy to free ride in subsequent rounds of a repeated game (Roth and Erev 1995). The strategies hypothesis holds that repeated games allows participants to signal future moves to other players (Sonnemans et al. 1999). Thus, participants may choose a strategy of investing at first to conceal that the free-riding incentives have been learned. However, in our case, we would argue that these two hypotheses may play little or no role in explaining the phenomenon of decay. First, each round of our framed games was not a repetition of other rounds, but a whole new decision-making experience altogether. There is no possibility of learning because there is no repetition that could, over time, reveal a dominant strategy. Second, the earlier behavior of the other individual participants was never made known to the participants, so they were not able to update their expectations about others’ behavior to choose the best strategy. Third, the benefits of each round were not announced until all games were completed. Participants could not reinforce learning based on their past payoffs to repeat their decision in the future.

The expectation is a major determinant of cooperative behavior

We also found that the expectation of cooperation from peers was highly correlated with the decisions of the participants. Given the lack of information about the behavior of the others, participants relied more on the perceived probability of contribution in the group, which was based on their prior knowledge of each other in their respective communities (Offerman et al. 1996). As a result, peer expectation, as a prior, will affect participants’ willingness to cooperate and the subsequent outcome at the community level. In light of our results, the perception of social cohesion was similar to the actual levels of cooperation and was positively correlated with voluntary cooperation, which is consistent with the earlier results of Wiener and Doescher (1994). Although the results were not strong enough to support the consistent behavior hypothesis that individuals only contribute when the expected returns of contributing are higher than the expected returns of not contributing, we can conclude that peer expectation was a major determinant of cooperative behavior in our study.

As reported in Table 1, most participants from around Wanglang were of the Baima Tibetan ethnic minority, but there was also a nontrivial share of the Baima Tibetan ethnic minority in our sample near Baishuijiang National Nature Reserve. Taking these two results together would suggest that members of the Baima Tibetan minority in villages near Baishuijiang were about 10% more likely to voluntarily contribute than their Han counterparts, and about 15% more likely to voluntarily contribute than their fellow Baima Tibetans living in villages near Wanglang. It is also true that residents who live in Baishuijiang have lower annual household incomes (RMB¥38,104.74, equivalent to roughly US$5525 at the time of the study) than those who live in Wanglang (RMB¥63,166.67, equivalent to roughly US$9150 at the time of the study). Relative poverty could cause participants to be more risk-averse and less willing to choose such investments even with higher expected returns (Lipton 1968, Rosenzweig and Binswanger 1993).

Trust levels do not have a significant impact on cooperative behavior

Trust affects whether a participant is willing to cooperate in the expectation that others will rely on norms of reciprocity. Based on the positively reinforcing relationship among trust, reciprocity, and reputation proposed by Ostrom (1998), if participants’ trust in the reliability of others is moderately high in the first place, then they themselves may reciprocate this expected behavior, with the ultimate result being a higher level of cooperation. Previous studies conducted in other communities have supported this theory. For example, an experiment conducted in Shanghai, China among university students, middle school students, and community residents showed that the level of trust has a positive correlation with voluntary cooperation (Qin et al. 2011). The research developed by Gächter et al. (2011) in rural and urban Russia among 630 non-student and student participants yielded similar results. Given that our participants are residents of the same community in rural China, they know each other well, so they have a high level of trust, and we do observe an overall high willingness to cooperate. However, the lack of variation among different groups made it difficult to detect any systematic effect of trust on the propensity toward voluntary cooperation. Thus, we did not observe a significant impact of trust levels on cooperative behavior.

Uncertainties in experimental design

Cooperative behavior may be influenced by whether participants understand the game well. We measured participants’ level of understanding by doing a rehearsal in which we asked every participant about their investment willingness and the corresponding income. The game began after everyone had stated their income correctly. In addition, the game was designed to be simple and easily understood by rural residents with limited formal education, limited literacy, and limited numeracy. To lend credence to the reliability of our results, the game background was also in line with participants’ real life, and the regulations simulated the incentives they faced in the village. For example, if a participant keeps the coupon, his or her individual return will increase, and he or she can still enjoy the benefit of the group. In actual life, a villager who has been working outside for a long time and did not participate in local environmental protection could also enjoy the benefits of the natural environment via ecotourism and grazing. With experimental studies, there is always a question as to whether there is a correspondence between behavior observed in an experiment and behavior observed in the real world (Levitt and List 2007). To date, the empirical evidence has been mixed. Hill and Gurven (2004) and Gurven and Winking (2008) finding no evidence of a correspondence between behavior in experiments and behavior in the real world among their samples of Indigenous persons in Paraguay and Bolivia, respectively. Carpenter and Seki (2011) and Fehr and Leibbrandt (2011), on the other hand, found consistent behavior between the experimental situations and the real world in their studies among Japanese and Brazilian shrimpers, respectively. Gelcich et al. (2013) also found evidence that groups that demonstrated a greater degree of cooperation in their common property resource (CPR) games also exhibited superior performance in real world management of artisanal fishers in central coastal Chile. Although we cannot state unequivocally that our results are externally valid, the closer alignment of our experimental design with these latter studies, which suggest consistency with real-world behavior, coupled with the steps we took to frame participation in the context of real-world livestock management in and around these protected areas, lends credence to the reliability of our methods (internal consistency) as well as our results (external validity).

An additional, and unanticipated, threat to the validity of our results pertains to the presence of village cadres in some of the group meetings. Chinese village cadres are staff members who hold certain positions, exercise public power, manage public affairs, provide public services in grass-roots party organizations, villagers’ committees, and other organizations. In general, villagers are expected to listen to cadres’ opinions when it comes to making big decisions. Thus, cooperative behavior may be influenced by village cadres if, for example, people would view cooperation as a strategy to curry favor with the cadres. In some of the game settings, cadres were present, either because the only opportunity to conduct the game was in the cadre’s home, or because the cadre was interested in observing the research activities. As a robustness check, we re-ran the regressions dropping those observations in which the village cadre was present, and the coefficients in these regressions were not statistically different from the regression coefficients reported in Tables 3-5 (see Appendix 1,Tables A1-A3). Thus, although the presence of the cadre increased the rate of voluntary contributions, these deviations from the experimental protocol did not result in statistically different estimates for the effects of either collective risk or expectations of peers’ cooperation on voluntary contributions. That is, although the presence of these cadres may have led to higher levels of cooperation than if they were not present, their presence did not augment or diminish the effect of expectations of peers’ behavior in increasing voluntary cooperation, nor did it augment or diminish the effect of risk in reducing voluntary cooperation.

CONCLUSIONS AND POLICY IMPLICATIONS

Our results have several interesting policy implications. Although we do not have census data on these specific communities with which to compare our sample to determine whether, or to what extent, the members of our sample are representative of the broader community, and consequently we must be relatively modest in interpreting these results and making policy pronouncements, the results are at least suggestive of some general approaches to conservation policy. Free-ranging livestock is prevalent throughout China’s forests (Melick et al. 2006). The incongruence of private motivations and social welfare results in a policy-making dilemma with no readily apparent solution. On the one hand, local people rely on livestock as an important livelihood diversification strategy, maintaining livestock as important assets that they can change for money when needed to see a doctor, construct houses, or meet any of several other urgent needs that may arise. This has been especially true following the enactment of conservation policies that restricted activities and income in the logging and agriculture sectors. Moreover, income from tourism was also cut off because many households were forcibly relocated due to hydropower projects or road construction, or otherwise relocated due to the massive earthquake that struck Sichuan in 2009 (Xu et al. 2017, Li et al. 2021). These events intensified livestock grazing as an alternative source of income (Li et al. 2017). Weak regulation and law enforcement leave room for local people to use forests for their own enrichment, rather than incentivizing the social gains that could accrue due to sustainable forest management (Li et al. 2017). To make matters worse, overgrazing and severe winter storms have contributed to high annual death rates for livestock in these areas (Li et al. 2017). With limited resources and more severe weather events, local communities face an uncertain and precarious future (Li et al. 2017).

In this study, we explored herdsmen’s propensity for cooperation in the management of common property resources by conducting a framed public goods game among rural communities in China’s biodiversity hotspots. We also explored how variation in socioeconomic factors may explain differences in contribution preferences of participants by asking them to fill out a personal survey upon completion of the experimental procedure. Our results suggest that increasing variability or uncertainty on personal earnings in grazing communities leads to a reduction in cooperation with community members, and uncertainty on public return leads to a greater reduction. Our findings also show that expected cooperation among peers is a major determinant of voluntary cooperation. This is especially true in cooperative engagements in which there is a threshold to reach before the benefits could be shared among community members. Given the high cooperation rate among participants in our study and its fragility to collective risk, conservation policies should emphasize reducing the risks faced by the population. Collective management approaches such as community cooperatives have the potential to help residents improve their livelihoods sustainably. Village leaders can also play a positive role in setting up and supervising cooperatives.

The results from our experimental games may provide valuable insights for the design of forestry conservation policies and the collective management of natural resources. For example, there was a high rate of success in groups attaining the provision point, and in most cases substantially exceeding the provision point. This pattern suggests that policies requiring collective actions with a minimum target of participation and compliance to be met for subsidy or other forms of incentives could be a successful approach for forestry management in this area.

Overall, our findings suggest that collective risk can play a critical role in how rural residents make decisions about cooperative management of natural resources. Certainly, climate change is likely to increase the variability in the returns to collectively managed resources. However, in a society in which individuals and communities are increasingly interconnected, privatized management seems unrealistic. Further, prior research has demonstrated that risk-sharing reduced the cost of failed management (Fafchamps 2003), and strong and well-functioning collective management reduces the elite capture and may enhance community well-being (Mwangi and Markelova 2009).

China’s first giant panda national park is over 22,000 km², nearly 2.5 times the size of Yellowstone National Park (National Forestry and Grassland Administration and National Park Administration 2019). There are 120,838 people currently living within the boundaries of the new park (5553 reside inside the core protection zone, 115,285 reside surrounding the park; National Forestry and Grassland Administration and National Park Administration 2019). Residents will need to adapt to live under new restrictions. Considering the difficulties and efficiency of governing multiple uses and users of these local commons, there is an opportunity for local governance and collective management. Much of the previous literature has emphasized that, where appropriate, the relatively informal collective management of collective property resources can avoid the severe resource degradation predicted by the “tragedy of the commons” (Seabright 1993). For rural communities in or near protected areas such as Wanglang and Baishuijiang, in which individuals’ livelihoods inextricably depend on local common-pool resources from the forest, there is a need for grassroots-level cooperation. Given the high cooperation rate among participants in our study and its fragility to collective risk, we suggest that cooperatives have the potential to help residents improve their livelihoods sustainably. In a typical cooperative, members from local communities join voluntarily. They democratically elect committees to run the group. Apart from helping the members to manage local resources, cooperatives also provide a contact point among the local community, government, NGOs, and other stakeholders (Aminaiee and Ahmadabadi 2007). By this cooperative management, local communities can achieve standardized grazing, refined management, reducing collective risks, increase added value and competitiveness the market. China has explored community-based approaches for conservation and ecosystem management over the years, such as the first attempt in the Sanjiangyuan region in Qinghai, where farmers and herders play a key role in ecological protection at public welfare positions (Shen and Tan 2012). More active innovation and demonstration of policies that engage the local community into natural resources management and conservation should be made.

LITERATURE CITED

Aminaiee, A. H., and M. N. Ahmadabadi. 2007. Learning cooperative object pushing with variable contact point. Pages 1-6 in 2007 IEEE/ASME international conference on advanced intelligent mechatronics. Institute of Electrical and Electronics Engineers, Washington, D.C., USA. https://doi.org/10.1109/AIM.2007.4412554

Andreoni, J. 1988. Why free ride?: strategies and learning in public goods experiments. Journal of Public Economics 37(3):291-304. https://doi.org/10.1016/0047-2727(88)90043-6

Andreoni, J. 1995. Cooperation in public-goods experiments: kindness or confusion? American Economic Review 85(4):891-904.

Bagnoli, M., and B. L. Lipman. 1989. Provision of public goods: fully implementing the core through private contributions. Review of Economic Studies 56(4):583-601. https://doi.org/10.2307/2297502

Burger, N. E., and C. D. Kolstad. 2009. Voluntary public goods provision, coalition formation, and uncertainty. Working paper 15543. National Bureau of Economic Research, Cambridge, Massachusetts, USA. https://doi.org/10.3386/w15543

Cárdenas, J.-C., M. A. Janssen, M. Ale, R. Bastakoti, A. Bernal, J. Chalermphol, Y. Gong, H. Shin, G. Shivakoti, Y. Wang, and J. M. Anderies. 2017. Fragility of the provision of local public goods to private and collective risks. Proceedings of the National Academy of Sciences 114(5):921-925. https://doi.org/10.1073/pnas.1614892114

Carpenter J., and E. Seki. 2011. Do social preferences increase productivity? Field experimental evidence from fishermen in Toyama Bay. Economic Inquiry 49:612-630. https://doi.org/10.1111/j.1465-7295.2009.00268.x

Castillo, D., F. Bousquet, M. A. Janssen, K. Worrapimphong, and J.-C. Cárdenas. 2011. Context matters to explain field experiments: results from Thai and Colombian fishing villages. Ecological Economics. 70:1609-1620. https://doi.org/10.1016/j.ecolecon.2011.05.011

Dawes, R. M., J. McTavish, and H. Shaklee. 1977. Behavior, communication, and assumptions about other people’s behavior in a commons dilemma situation. Journal of Personality and Social Psychology 35(1):1-11. https://doi.org/10.1037/0022-3514.35.1.1

Fafchamps, M. 2003. Rural poverty, risk, and development. Edward Elgar, Cheltenham, UK. https://doi.org/10.4337/9781781950685

Fehr, E., and A. Leibbrandt. 2011. A field study on cooperativeness and impatience in the tragedy of the commons. Journal of Public Economics 95:1144-1155. https://doi.org/10.1016/j.jpubeco.2011.05.013

Fischbacher, U., and S. Gächter. 2010. Social preferences, beliefs, and the dynamics of free riding in public goods experiments. American Economic Review 100(1):541-556. https://doi.org/10.1257/aer.100.1.541

Gächter, S., and B. Herrmann. 2011. The limits of self-governance when cooperators get punished: experimental evidence from urban and rural Russia. European Economic Review 55(2):193-2010. https://doi.org/10.1016/j.euroecorev.2010.04.003a

Gächter, S., B. Herrmann, and C. Thöni. 2004. Trust, voluntary cooperation, and socio-economic background: survey and experimental evidence. Journal of Economic Behavior and Organization 55(4):505-531. https://doi.org/10.1016/j.jebo.2003.11.006

Gelcich, S., R. Guzman, C. Rodriguez-Sickert, J. C. Castilla, and J. C. Cárdenas. 2013. Exploring external validity of common pool resource experiments: insights from artisanal benthic fisheries in Chile. Ecology and Society 18(3):2. https://doi.org/10.5751/ES-05598-180302

Gurven, M., and J. Winking. 2008. Collective action in action: prosocial behavior in and out of the laboratory. American Anthropologist 110:179-190. https://doi.org/10.1111/j.1548-1433.2008.00024.x

Hardin, G. 1968. The tragedy of the commons: the population problem has no technical solution; it requires a fundamental extension in morality. Science 162(3859):1243-1248. https://doi.org/10.1126/science.162.3859.1243

Henrich, J., R. McElreath, A. Barr, J. Ensminger, C. Barrett, A. Bolyanatz, J. C. Cárdenas, M. Gurven, E. Gwako, N. Henrich, C. Lesorogol, F. Marlowe, D. Tracer, and J. Ziker. 2006. Costly punishment across human societies. Science 312(5781):1767-1770. https://doi.org/10.1126/science.1127333

Hill, K., and M. Gurven. 2004. Economic experiments to examine fairness and cooperation among the Ache Indians of Paraguay. Pages 382-412 in J. Henrich, R. Boyd, S. Bowles, C. Camerer, E. Fehr, and H. Gintis, editors. Foundations of human sociality: economic experiments and ethnographic evidence from fifteen small-scale societies. Oxford University Press, Oxford, UK. https://doi.org/10.1093/0199262055.003.0013

Huang, Q., Y. Fei, H. Yang, X. Gu, and M. Songer. 2020. Giant Panda National Park, a step towards streamlining protected areas and cohesive conservation management in China. Global Ecology and Conservation 22:e00947. https://doi.org/10.1016/j.gecco.2020.e00947

Isaac, R. M., and J. M. Walker. 1988. Communication and free‐riding behavior: the voluntary contribution mechanism. Economic Inquiry 26(4):585-608. https://doi.org/10.1111/j.1465-7295.1988.tb01519.x

Isaac, R. M., J. M. Walker, and S. H. Thomas. 1984. Divergent evidence on free riding: an experimental examination of possible explanations. Public Choice 43(2):113-149. https://doi.org/10.1007/BF00140829

Janssen, M. A., F. Bousquet, J.-C. Cárdenas, D. Castillo, and K. Worrapimphong. 2013. Breaking the elected rules in a field experiment on forestry resources. Ecological Economics 90:132-139. https://doi.org/10.1016/j.ecolecon.2013.03.012

Kocher, M. G., P. Martinsson, D. Matzat, and C. Wollbrant. 2015. The role of beliefs, trust, and risk in contributions to a public good. Journal of Economic Psychology 51:236-244. https://doi.org/10.1016/j.joep.2015.10.001

Levitt, S. D., and J. A. List. 2007. What do laboratory experiments measuring social preferences reveal about the real world? Journal of Economic Perspectives 21:153-174. https://doi.org/10.1257/jep.21.2.153

Li, B. V. 2020. Conservation: guarding panda land. Current Biology 30(7):R302-R304. https://doi.org/10.1016/j.cub.2020.02.089

Li, B. V., M. J. Kim, W. Xu, S. Jiang, and L. Yu. 2021. Increasing livestock grazing, the unintended consequence of community development funding on giant panda habitat. Biological Conservation 257:109074. https://doi.org/10.1016/j.biocon.2021.109074

Li, B. V., C, N. Jenkins, W. and Xu. 2022. Strategic protection of landslide vulnerable mountains for biodiversity conservation under land-cover and climate change impacts. Proceedings of the National Academy of Sciences 119(2):e2113416118. https://doi.org/10.1073/pnas.2113416118

Li, B. V., and S. L. Pimm. 2016. China’s endemic vertebrates sheltering under the protective umbrella of the giant panda. Conservation Biology 30(2):329-339. https://doi.org/10.1111/cobi.12618

Li, B. V., S. L. Pimm, S. Li, L. Zhao, and C. Luo. 2017. Free-ranging livestock threaten the long-term survival of giant pandas. Biological Conservation 216:18-25. https://doi.org/10.1016/j.biocon.2017.09.019

Li, S., W. J. McShea, D. Wang, J. Huang, and L. Shao. 2012. A direct comparison of camera‐trapping and sign transects for monitoring wildlife in the Wanglang National Nature Reserve, China. Wildlife Society Bulletin 36(3):538-545. https://doi.org/10.1002/wsb.161

Liu, J., S. Li, Z. Ouyang, C. Tam, and X. Chen. 2008. Ecological and socioeconomic effects of China’s policies for ecosystem services. Proceedings of the National Academy of Sciences 105(28):9477-9482. https://doi.org/10.1073/pnas.0706436105

Lipton, M. 1968. The theory of optimising peasant. Journal of Development Studies 4(3):327-351. https://doi.org/10.1080/00220386808421262

Maxwell, G., and R. E. Ames. 1981. Economists free ride, does anyone else?: experiments on the provision of public goods, IV. Journal of Public Economics 15(3):295-310. https://doi.org/10.1016/0047-2727(81)90013-X

McKean, M. A. 2000. Common property: what is it, what is it good for, and what makes it work. Pages 27-55 in C. C. Gibson, M. A. McKean, and E. Ostrom, editors. People and forests: communities, institutions, and governance. MIT Press, Cambridge, Massachusetts, USA.

Meinzen-Dick, R. S. 2009. Coordination in natural resource management. Pages 319-342 in J. F. Kirsten, A. Dorward, C. Poulton, and N. Vink, editors.Institutional economics perspectives on African Agricultural Development. International Food Policy Research Institute (IFPRI), Washington, D.C., USA. https://ebrary.ifpri.org/utils/getfile/collection/p15738coll2/id/129491/filename/129702.pdf

Melick, D., X. Yang, and J. Xu. 2006. Seeing the wood for the trees: how conservation policies can place greater pressure on village forests in southwest China. Pages 385-397 in D. L. Hawksworth and A. T. Bull, editors. Plant conservation and biodiversity: topics in biodiversity and conservation. Springer, Dordrecht, The Netherlands. https://doi.org/10.1007/978-1-4020-6444-9_24

Mwangi, E., and H. Markelova. 2009. Collective action and property rights for poverty reduction: a review of methods and approaches. Development Policy Review 27(3):307-331. https://doi.org/10.1111/j.1467-7679.2009.00448.x

National Forestry and Grassland Administration and National Park Administration. 2019. The overall development plan of Giant Panda National Park. National Forestry and Grassland Administration and the National Park Administration, Bejing, China.

Offerman, T., J. Sonnemans, and A. Schram. 1996. Value orientations, expectations, and voluntary contributions in public goods. Economic Journal 106(437):817-845. https://doi.org/10.2307/2235360

Ostrom, E. 1990. Governing the commons: the evolution of institutions for collective action. Cambridge University Press, Cambridge, UK. https://doi.org/10.1017/CBO9780511807763

Ostrom, E. 1998. A behavioral approach to the rational choice theory of collective action: presidential address, American political science association, 1997. American Political Science Review 92(1):1-22. https://doi.org/10.2307/2585925

Ostrom, E., R. Gardner, and J. Walker. 1994. Rules, games, and common-pool resources. University of Michigan Press, Ann Arbor, Michigan, USA. https://doi.org/10.3998/mpub.9739

Peshkovskaya, A., T. Babkina, and M. Myagkov. 2019. Gender effects and cooperation in collective action: a laboratory experiment. Rationality and Society 31(3):337-353. https://doi.org/10.1177/1043463119858788

Poteete, A. R., M. A. Janssen, and E. Ostrom. 2010. Working together: collective action, the commons, and multiple methods in practice. Princeton University Press, Princeton, New Jersey, USA. https://doi.org/10.1515/9781400835157

Qin, X., J. Shen, and X. Meng. 2011. Group-based trust, trustworthiness, and voluntary cooperation: evidence from experimental and survey data in China. Journal of Socioeconomics 40(4):356-363. https://doi.org/10.1016/j.socec.2011.04.007

Roth, A. E., and I. Erev. 1995. Learning in extensive-form games: experimental data and simple dynamic models in the intermediate term. Games and Economic Behavior 8(1):164-212. https://doi.org/10.1016/S0899-8256(05)80020-X

Rondeau, D., G. L. Poe, and W. D. Schulze. 2005. VCM or PPM? A comparison of the performance of two voluntary public goods mechanisms. Journal of Public Economics 89(8):1581-1592. https://doi.org/10.1016/j.jpubeco.2004.06.014

Rosenzweig, M. R., and H. P. Binswanger. 1993. Wealth, wealth risk and the composition and profitability of agricultural investments. Economic Journal 103(416):56-78. https://doi.org/10.2307/2234337

Seabright, P. 1993. Managing local commons: theoretical issues in incentive design. Journal of Economic Perspectives 7(4):113-134. https://doi.org/10.1257/jep.7.4.113

Seguino, S., T. Stevens, and M. Lutz. 1996. Gender and cooperative behavior: economic man rides alone. Feminist Economics 2(1):1-21. https://doi.org/10.1080/738552683

Shang, J., and R. Croson. 2009. A field experiment in charitable contribution: the impact of social information on the voluntary provision of public goods. Economic Journal 119(540):1422-1439. https://doi.org/10.1111/j.1468-0297.2009.02267.x

Shen, X., and J. Tan. 2012. Ecological conservation, cultural preservation, and a bridge between: the journey of Shanshui Conservation Center in the Sanjiangyuan region, Qinghai-Tibetan Plateau, China. Ecology and Society 17(4):38. https://doi.org/10.5751/ES-05345-170438

Sirán, A. H., J. C. Cárdenas, and J. D. Machoa. 2006. The relation between income and hunting in tropical forests: an economic experiment in the field. Ecology and Society 11(1):44. https://doi.org/10.5751/ES-01640-110144

Smith, V. H., M. R. Kehoe, and M. E. Cremer. 1995. The private provision of public goods: altruism and voluntary giving. Journal of Public Economics 58(1):107-126. https://doi.org/10.1016/0047-2727(94)01455-W

State Forestry Administration of P.R. China. 2015. The giant pandas of China: status quo - major findings of the fourth national survey on giant panda. State Forestry Administration of P.R. China, Beijing, China. https://www.forestry.gov.cn/main/65/20150302/758158.html

Sonnemans, J., A. Schram, and T. Offerman. 1999. Strategic behavior in public good games: when partners drift apart. Economics Letters 62(1):35-41. https://doi.org/10.1016/S0165-1765(98)00203-1

Teyssier, S. 2012. Inequity and risk aversion in sequential public good games. Public Choice 151(1-2): 91-119. https://doi.org/10.1007/s11127-010-9735-1

Ting, Z., G. P. Shivakoti, C. Haiyun, and D. Maddox. 2012. A survey-based evaluation of community-based co-management of forest resources: a case study of Baishuijiang National Natural Reserve in China. Environment, Development and Sustainability 14(2):197-220. https://doi.org/10.1007/s10668-011-9316-6

Ward, P. S., M. Alvi, S. Makhija, and D. J. Spielman. 2019. Cooperation and the management of local common resources in remote rural communities: evidence from Odisha, India. Selected Paper presented at the 2019 Agricultural and Applied Economics Association Annual Meeting, Atlanta, Georgia, USA, July 21-24, 2019. nternational Food Policy Research Institute (IFPRI), Washington, D.C., USA. https://doi.org/10.2499/p15738coll2.133790

Wei, F., R. Costanza, Q. Dai, N. Stoeckl, X. Gu, S. Farber, Y. Nie, I. Kubiszewski, Y. Hu, R. Swaisgood, X. Yang, M. Bruford, Y. Chen, A. Voinov, D. Qi, M. Owen, L. Yan, D. C. Kenny, Z. Zhang, R. Hou, S. Jiang, H. Liu, X. Zhan, L. Zhang, B. Yang, L. Zhao, X. Zheng, W. Zhou, Y. Wen, H. Gao, and W. Zhang. 2018. The value of ecosystem services from giant panda reserves. Current Biology 28(13):2174-2180. https://doi.org/10.1016/j.cub.2018.05.046

Wiener, J. L., and T. A. Doescher. 1994. Cooperation and expectations of cooperation. Journal of Public Policy and Marketing. 13(2):259-270. https://doi.org/10.1177/074391569401300207

Xu, W., A. Viña, L. Kong, S. L. Pimm, J. Zhang, W. Yang, Y. Xiao, L. Zhang, X. Chen, J. Liu, and Z. Ouyang. 2017. Reassessing the conservation status of the giant panda using remote sensing. Nature Ecology and Evolution 1(11):1635-1638. https://doi.org/10.1038/s41559-017-0317-1

Zhou, Y., and Z. Song. 2008. The voluntary contribution mechanism of public goods: an experimental research. Economic Research Journal 7:90-104.