Impact of Forest Ecological Compensation Policy on Farmers’ Livelihood: A Case Study of Wuyi Mountain National Park

Abstract

Forest ecological compensation policies (FECPs) are a key institutional arrangement for balancing ecological conservation and farmers’ development needs in national parks. Existing research has often treated such policies as a homogeneous whole, failing to clearly reveal the mechanisms through which different policy types affect farmers’ livelihoods, while also paying insufficient attention to complex property-rights settings. This study takes Wuyi Mountain National Park—a typical representative of collective forest regions in southern China—as a case study. Based on 239 micro-survey datasets from farming households and employing the mprobit model and moderating effect models, it investigates the influence, mechanisms, and heterogeneity of farmers’ livelihood capital in terms of their livelihood strategy choices under the moderating roles of “blood-transfusion” and “blood-making” FECPs. The results show the following: (1) Among the sample farmers, livelihood strategies are distributed as follows: pure agricultural type (31.8%), out-migration for work type (20.5%), and commercial operation type (47.7%). (2) Farmers’ livelihood capital has a significant impact on their livelihood strategy choice, with different dimensions of capital playing distinct roles. (3) FECPs follow differentiated moderating pathways. “Blood-transfusion” policies emphasize compensation and buffering functions, reducing farmers’ livelihood transition pressure through direct cash transfers; “blood-making” policies reflect empowerment and restructuring characteristics, activating physical assets and reshaping the role of social capital through productive investment. Together, they constitute a complementary system of protective security and transformative empowerment. Accordingly, this study proposes policy insights such as building a targeted ecological compensation system that is categorized, dynamically linked, and precise; innovating compensation fund allocation mechanisms that integrate collective coordination with household-level benefits; optimizing policy design oriented toward enhancing productive capital; and establishing robust monitoring, evaluation, and adaptive management mechanisms for dynamic FECPs.

1. Introduction

The Earth is facing a global crisis of biodiversity loss, and the conservation of forest ecosystems has become a core issue for the international community. Since the 1970s, community-based natural resource management has gradually become a formal approach to biodiversity conservation [1], addressing the limitations of traditional conservation models that excluded local communities from the governance and management of protected natural areas [2]. However, the tension between conservation objectives and community livelihood needs remains a central dilemma in the management of biosphere reserves [3,4], a challenge also explicitly identified by the United Nations Convention on Biological Diversity as a key obstacle to achieving global biodiversity targets [5]. Against this backdrop, forest ecological compensation policies (FECPs), as an important institutional arrangement for reconciling the conflict between development and conservation in protected areas and safeguarding the livelihoods of community households [6], carry both significant academic value and practical relevance.

Ecological compensation falls under the core category of payments for ecosystem services. It can not only reduce the disturbance of human activities on ecosystems through economic incentives, but also improve the livelihoods of compensated groups through the optimization of resource allocation [7,8,9]. From the perspective of international research, studies on the relationship between forest ecological compensation policies (FECPs) and farmers’ livelihoods have formed a diverse landscape. In developed countries, FECPs are often integrated with market-based mechanisms to promote the transition of farmers’ livelihoods toward eco-friendly models. For instance, Yellowstone National Park in the United States extends the ecotourism industry chain to diversify farmers’ livelihoods [10]; the Bavarian Forest National Park in Germany employs a payment for ecosystem services mechanism to steer farmers toward sustainable livelihood activities such as forest carbon sink management [11]; and Uluru-Kata Tjuta National Park in Australia adopts a “joint Aboriginal management” model to enable tourism revenue sharing and cultural livelihood continuity [12]. These studies indicate that in a mature market environment with well-developed policy support, FECPs can act as a catalyst for upgrading farmers’ livelihoods. In contrast, the implementation of forest ecological compensation in developing countries is more nuanced and challenging. For example, Serengeti National Park in Tanzania links compensation funds with community public services [13], and the Manu Biosphere Reserve in Peru empowers indigenous communities to participate in forest management [14]; both cases have achieved, to some extent, synergy between ecological conservation and community development. On the other hand, conservation measures in Kenya’s Maasai Mara National Reserve that restrict traditional nomadic activities have led to a significant decline in household incomes [15], while insufficient infrastructure and benefit-sharing mechanisms in South Africa’s Kruger National Park have resulted in limited local community benefits from ecotourism revenues [16]. These differences reflect the multiple practical challenges that must be addressed when implementing FECPs in developing countries, including relatively weak economic foundations, incomplete policy systems, and the need for socio-cultural adaptation.

Forest ecological compensation policies (FECPs) in China are not only directly linked to farmers’ livelihoods but also involve complex issues such as collective forest ownership and the marketization of ecological products, reflecting distinct national characteristics. Since the 1990s, China’s FECPs have gradually emerged and evolved, transitioning from single-purpose compensation for ecological public welfare forests to diversified payments for ecosystem services, and from a government-dominated model to one combining government guidance with market-based supplements [17,18]. Following the explicit call in the 2017 Master Plan for Establishing a National Park System to “improve the ecological conservation compensation system”, the development of China’s national park system has advanced comprehensively, leading to increasingly rich research on FECPs. Scholars have demonstrated the necessity and urgency of establishing a forest ecological compensation system within national parks [19], analyzed key issues such as “who to compensate” and “how much to compensate” [20], evaluated the costs and benefits of ecological compensation [21], and constructed policy frameworks and content standards for FECPs [22]. At the micro level, Chinese scholars have also conducted extensive empirical studies on the impact of FECPs on farmers’ livelihoods. Early research primarily focused on the direct effects of policies such as the Grain for Green Program [23], livestock [24] and poultry breeding bans [25], and ecological relocation on farmers’ livelihoods and incomes, providing a preliminary basis for assessing FCEP outcomes. As research deepened, the academic community gradually recognized that policies often do not act directly on outcomes but rather exert influence by reshaping household resource endowments. Subsequent studies introduced the Sustainable Livelihoods Framework, with substantial empirical evidence indicating that livelihood capital serves as a key mediator linking FECPs to household behavioral decisions. Whether in wetland compensation [26], herder asset allocation [27], or the livelihood transition of households exiting livestock farming [28], livelihood capital plays a central transmission role. In recent years, micro-empirical research has further advanced toward greater refinement and contextual relevance, mainly reflected in two aspects: First, by deconstructing and differentiating FECPs themselves. For example, some studies specifically analyze the heterogeneous impacts of ecological public welfare forest compensation on forest farmers’ livelihoods [29], while others examine how diverse compensation methods guide households toward different development pathways [30]. Second, by conducting field investigations embedded in the specific ecological constraints and rural social contexts of national parks. Such research not only focuses on the overall improvement effects of national park FECPs on community livelihoods [31] but also sensitively identifies new problems arising from conservation actions, such as the impact of human–wildlife conflict on farmers’ livelihoods and their adaptation strategies [32].

Overall, the existing literature has accumulated a rich body of knowledge regarding policy effect evaluation, exploration of transmission mechanisms, differentiation of policy instruments, and embedding in complex contexts. However, two potential research gaps remain (Figure 1): On the one hand, the “black box” of forest ecological compensation policies (FECPs) has not been fully opened. Most existing studies treat diverse compensation methods homogenously, overlooking the potential differences in effects and the distinct mechanisms of action between “blood-transfusion” and “blood-making” compensation, which differ in their institutional logics. Furthermore, while studies often directly link FECPs with livelihood outcomes, there is a lack of clear empirical testing on the specific transmission pathway—namely, how FECPs differentially influence farmers’ strategy choices by altering their livelihood capital endowments. On the other hand, there is insufficient attention paid to real-world scenarios characterized by complex tenure systems and large-scale coverage. Existing empirical cases primarily focus on regions with relatively clear property rights and limited compensation scopes. For protected areas dominated by forests in southern China—such as Wuyi Mountain National Park, which features a high proportion of collective forest tenure and broad compensation coverage—the implementation logic and adaptation patterns of compensation policies have not been fully revealed. This may limit the generalizability and operational applicability of policy insights from existing research when addressing the typical yet complex realities of southern China’s collective forest regions.

In response to the aforementioned research gaps, this study takes Wuyi Mountain National Park as a case area. Based on 239 micro-survey datasets from farming households and employing mprobit and moderating effect models, it investigates the impact, mechanisms, and heterogeneity of farmers’ livelihood capital on their livelihood strategy choices under the moderating effect of forest ecological compensation policies (FECPs). Correspondingly, the novelty of this study is reflected in three aspects: First, a differentiated research perspective: breaking through the existing limitation of homogenizing compensation policies, it takes policy type differences as an entry point to comparatively analyze the effect variations between “blood-transfusion” and “blood-making” FECPs. Second, deepened research mechanisms: it elucidates the transmission pathway of “forest ecological compensation → livelihood capital → livelihood strategy” from a micro-level perspective, clarifying the internal logic through which FECPs influence farmers’ livelihoods. Third, a typical research setting: focusing on Wuyi Mountain National Park—a forest-dominated protected area characterized by complex collective forest tenure and broad compensation coverage—the study integrates local practical features, enhancing the generalizability and operability of policy optimization recommendations and addressing the current lack of attention to complex, context-specific national settings in existing research.

2. Theoretical Analysis and Research Hypotheses

2.1. Analysis of the Impact Mechanism of Farmers’ Livelihood Capital on Their Livelihood Strategy Choices in National Parks

Farmers’ livelihoods constitute a complex issue influenced by multiple factors, including natural conditions, individual capabilities, policy institutions, and the external environment [33,34,35]. Drawing upon the Sustainable Livelihoods Framework (SLA) and integrating ecological compensation theory, this paper examines the impact mechanism of farmers’ livelihood capital on their livelihood strategy choices under the influence of forest ecological compensation policies (FECPs) within the context of Wuyi Mountain National Park. The SLA reveals the interrelationships between livelihood capital and livelihood strategies under the constraints of policies, institutions, and the external environment: Based on the characteristics of the local regional environment, farmers combine their owned livelihood capitals, choose livelihood methods according to their own capabilities, demonstrate corresponding livelihood strategy choices, and consequently produce corresponding livelihood outcomes. Simultaneously, when institutional policies undergo changes, farmers tend to adjust their livelihood strategies to adapt to new human–land relationships [36] (Figure 2). Therefore, farmers’ choice of livelihood strategies is a comprehensive consideration based on their own livelihood capital status. The specific impact pathways are as follows:

First, human capital. Human capital is often reflected in factors such as the number of household laborers, age structure, and education level. Within the Wuyi Mountain National Park area, households with labor shortages and lower education levels often find it difficult to acquire new knowledge related to ecotourism services, concession operations, or green industry skills, thereby limiting their capacity for livelihood transition and making them more inclined to maintain traditional pure agricultural livelihood strategies. Conversely, households with abundant human capital possess stronger learning and adaptive abilities, making it easier for them to shift toward non-agricultural livelihood strategies.

Second, natural capital. Natural capital comprises the ecological resource endowments on which farmers rely for survival, such as cultivated land, garden plots, forest land, and water resources. Under the strict ecological protection requirements of Wuyi Mountain National Park, the use of certain natural resources may be restricted or prohibited. When natural capital is scarce or its use is constrained, traditional livelihood strategies dependent on intensive natural resource exploitation—such as cultivation and breeding—become unsustainable. This situation prompts farmers to shift toward alternative strategies with lower dependence on natural resources, such as out-migration for work or commercial operations.

Third, social capital. Human economic activities are embedded within social relationship networks. Social capital manifests as farmers’ family networks, community ties, participation in cooperative organizations, and their ability to access external information and support. Rich social capital can broaden farmers’ information channels, enhance their capacity to cope with risks, and provide cooperation opportunities, thereby significantly promoting the diversification of their livelihood strategies.

Fourth, physical capital. Physical capital includes production tools, housing facilities, transportation equipment, etc. For instance, farmers who possess tea processing equipment or vacant houses can upgrade their houses into tea processing and sales points or renovate them into homestays for ecotourism business operations. In contrast, households with only low levels of physical capital lack the foundational conditions for developing high-value-added industries, and their livelihood strategy choices are often confined to out-migration for work or maintaining simple pure agricultural production.

Fifth, financial capital. Sufficient financial capital provides farmers with a risk buffer and investment capacity, enabling them to venture into livelihood strategies that require upfront investment, such as tea business operations or tourist homestays. Conversely, households with scarce financial capital often allocate most of their funds to daily consumption, making it difficult to accumulate the initial capital needed for livelihood transition. Consequently, their strategy choices tend to be more conservative.

H1.

Farmers’ livelihood capital has a significant impact on their livelihood strategy choices, and the direction of influence varies across different dimensions of capital.

2.2. “Blood-Transfusion” vs. “Blood-Making”: Differences in the Institutional Logic of Forest Ecological Compensation in National Parks

Ecological compensation is essentially an environmental and economic tool that internalizes external costs. As a special ecological and environmental area, national parks provide various essential ecological and environmental resources for human development. They perform important ecosystem service functions such as water conservation, carbon sequestration, and oxygen release, and thus exhibit typical external economies. At the same time, the implementation of strict protection in core zones and buffer zones comes at the cost of sacrificing local residents’ original dependence on resource environments and their normal development opportunities, showing external diseconomies. Therefore, to bridge the contradiction between “external economies” and “external diseconomies”, it is necessary for the government to carry out “compensation” work and internalize the externality of ecosystem services in national parks.

At the current stage, forest ecological compensation policies (FECPs) in national parks can be categorized into two types: “blood-transfusion” and “blood-making”. These two types differ in terms of compensation methods, target groups, and functional purposes, which in turn affect farmers’ livelihoods in practice [26] (Figure 3). “Blood-transfusion” FECPs mainly refer to one-time compensation in the form of monetary funds or in-kind materials, including compensation for ecological public welfare forests, buyouts of commercial forests, and compensation for forest right owners. After participating in “blood-transfusion” compensation, farmers can receive one-time, short-term financial subsidies. However, their original opportunities for planting, harvesting, and developing forest resources may be restricted as a result. In contrast to the one-time, short-term nature of the former, “blood-making” FECPs aim to enhance the economic self-sufficiency capacity of compensated areas, with a focus on long-term economic and social development. This approach involves measures such as subsidizing initial investments for national park community residents engaged in organic forestry production, supporting the development of green forestry industries, and providing skills training and necessary technical assistance. These efforts help farmers improve their labor skills and production efficiency, thereby achieving economic self-improvement. Although the initial economic support provided by “blood-making” FECPs may be lower than that of “blood-transfusion” FECPs, through long-term investment in human capital, farmers can achieve stable income growth and economic self-sustainability, forming a more sustainable livelihood situation [37,38,39,40].

2.3. Analysis of the Moderating Effect of Forest Ecological Compensation Policies in National Parks

Both “blood-transfusion” and “blood-making” forest ecological compensation policies (FECPs) exert a moderating effect on the process through which farmers’ livelihood capital influences their livelihood strategies (Figure 4).

Theoretical analysis of the moderating effect of “Blood-Transfusion” forest ecological compensation policies. The core of “blood-transfusion” forest ecological compensation policies (FECPs) lies in providing short-term financial subsidies. Their moderating logic is to alter the association between livelihood capital and livelihood strategy choices by filling the income gap resulting from ecological protection. This can be understood from two perspectives: moderating natural capital and moderating financial capital.

(1)

Dimension of Natural Capital and the Out-Migration for Work Livelihood Strategy. Natural capital serves as the foundation for farmers engaged in traditional agriculture. If its endowment is insufficient, compounded by usage restrictions under the ecological protection constraints of national parks, income from traditional agriculture may fail to cover livelihood expenses, forcing farmers to choose out-migration for work. However, “blood-transfusion” FECPs can directly provide funds to fill the income gap caused by restricted use of natural capital. Even with insufficient natural capital, subsidies can support basic living needs, reducing the necessity to rely on out-migration to supplement income. Consequently, this weakens the association between insufficient natural capital and the choice of out-migration for work. Based on this, H2a is proposed: “Blood-transfusion” FECPs weaken the positive driving effect of insufficient natural capital on the out-migration for work livelihood strategy.

(2)

Dimension of Financial Capital and the Out-Migration for Work Livelihood Strategy. Financial capital acts as a livelihood buffer for farmers. If financial capital is scarce, farmers may struggle to cope with short-term fluctuations in local livelihoods and are more inclined to choose out-migration for work to secure stable income. Conversely, if financial capital is ample, farmers have greater buffer space, reducing their willingness to migrate for work. In this context, if farmers receive “blood-transfusion” compensation, those originally lacking financial capital gain additional cash through subsidies, diminishing their need to rely on out-migration to maintain income stability. This weakens the association between financial capital scarcity and the choice of out-migration for work. Accordingly, H2b is proposed: “Blood-transfusion” FECPs weaken the positive driving effect of financial capital scarcity on the out-migration for work livelihood strategy.

(3)

The Relationship Between Natural Capital and the Commercial Operation Strategy. In Wuyi Mountain National Park, natural capital is a crucial asset for commercial operations (e.g., tea cultivation for business). Since ecological protection constraints limit traditional income from natural capital, farmers may shift towards commercial activities. “Blood-transfusion” FECPs, by providing cash subsidies, compensate for the income loss from restricted natural capital use. This support reduces the immediate pressure to rely heavily on natural capital for commercial income, thereby potentially weakening the direct, positive driving effect of natural capital on choosing commercial operations. Based on this, H2c is proposed: “Blood-transfusion” FECPs weaken the positive effect of natural capital on the choice of the commercial operation livelihood strategy.

(4)

The Relationship Between Financial Capital and the Commercial Operation Strategy. Launching a commercial operation requires initial capital. Farmers within the national park with low savings and limited credit access may find it difficult to afford such investments. “Blood-transfusion” FECPs can directly increase household cash holdings, slightly boosting financial capital and alleviating this constraint to some degree. However, as the primary purpose of these subsidies is to offset livelihood losses, funds are often used for daily consumption. They may only enable low-barrier commercial ventures (e.g., small stalls selling local products) rather than supporting large-scale investments. Therefore, these policies are expected to mildly strengthen the supportive role of financial capital for commercial operations. Accordingly, H2d is proposed: “Blood-transfusion” FECPs strengthen the positive effect of financial capital on the choice of the commercial operation livelihood strategy.

Theoretical analysis of the moderating effect of “Blood-Making” forest ecological compensation policies. The core of “blood-making” Forest Ecological Compensation Policies (FECPs) lies in empowerment. By providing productive investments through measures such as skills training, industrial support, and facility improvements, these policies tangibly enhance farmers’ capabilities to utilize their own resources for green transformation and income generation. This, in turn, moderates the relationship between livelihood capital and the choice of livelihood strategies. The mechanism can be understood from two perspectives: its moderating effect on social capital and on physical capital.

(1)

The Relationship Between Social Capital and the Out-Migration for Work Strategy. Out-migration for work often relies on social networks and trust. Farmers participating in “blood-making” FECPs frequently gain opportunities to join organized platforms like cooperatives or industry associations within Wuyi Mountain National Park. These platforms effectively facilitate production collaboration and information sharing among farmers, which can, to some extent, diminish the pull of traditional social networks towards out-migration for work. Therefore, H3a is proposed: “Blood-making” FECPs weaken the positive effect of social capital on the choice of the out-migration for work livelihood strategy.

(2)

The Relationship Between Physical Capital and the Out-Migration for Work Strategy. Scarcity of physical capital is a significant factor forcing labor migration. The production facility improvements and industrial support embedded in “blood-making” FECPs can elevate the level of fixed assets available to farmers for forestry production and primary processing. This enhances the attractiveness and feasibility of staying locally to engage in productive activities. Consequently, farmers receiving such support experience substantially reduced pressure to migrate for work due to insufficient physical capital. Accordingly, H3b is proposed: “Blood-making” FECPs weaken the positive driving effect of insufficient physical capital on the choice of the out-migration for work livelihood strategy.

(3)

The Relationship Between Social Capital and the Commercial Operation Strategy. Commercial operations require stable supply chains and sales channels. “Blood-making” FECPs, through industrial subsidies and support, guide farmers to join cooperative organizations or establish links with leading enterprises. This lowers the transaction costs and risks for farmers entering the market independently and provides them with operational resources that bring business opportunities and economic returns. Based on this, H3c is proposed: “Blood-making” FECPs strengthen the positive effect of social capital on the choice of the commercial operation livelihood strategy.

(4)

The Relationship Between Physical Capital and the Commercial Operation Strategy. The initiation and expansion of commercial operations are highly dependent on specialized fixed assets. Support measures such as subsidies for forestry production equipment and industrial development assistance can, to some degree, offset the operational costs for farmers. Furthermore, these resources can be directly used to enhance product value-added, expand reproduction, or improve service conditions, thereby strengthening the impetus towards choosing a commercial operation strategy. Therefore, H3d is proposed: “Blood-making” FECPs strengthen the positive effect of physical capital on the choice of the commercial operation livelihood strategy.

3. Data Sources, Variable Description, and Model Specification

3.1. Data Sources

This study selects Wuyi Mountain National Park in Fujian Province as the casestudy area (Figure 5). Located in northern Fujian Province and bordering the southern part of Yanshan County, Jiangxi Province, the pilot area of Wuyi Mountain National Park encompasses multiple types of protected natural areas, including a World Cultural and Natural Heritage site, a national nature reserve, a national scenic area, and a national forest park, with a total planned area of 1001.41 square kilometers. It involves 4 counties (cities, districts)—Wuyishan, Jianyang, Shaowu, and Guangze in Fujian—9 townships (towns, sub-districts), and 29 administrative villages.

As a typical representative of protected natural areas within China’s southern collective forest regions, Wuyi Mountain National Park is characterized by a relatively large resident population, a high proportion of collectively owned forests, and industries with strong dependence on natural resources. Conducting research here on the impact of farmers’ livelihood capital on their livelihood strategies under the moderating effect of Forest Ecological Compensation Policies (FECPs) holds significant typicality and practical relevance, specifically manifested in three aspects:

First, representativeness in terms of the property rights system. Unlike protected areas in northern China dominated by state-owned forests, collective forests account for approximately two-thirds of the land within Wuyi Mountain National Park. Conservation practices here must directly confront the core tension between the persistence of collective forest rights, farmers’ operational rights, and ecological protection restrictions, which represents the most prominent institutional challenge for protected areas in southern China’s collective forest regions.

Second, representativeness in terms of industrial characteristics. Wuyi Mountain National Park centers its economy on the tea industry and ecotourism. Tea cultivation involves the majority of local farming households, while ecotourism has also engaged a large number of farmers in commercial operations. This development model, which is highly dependent on natural resources, is common across ecologically protected regions in southern China.

Third, representativeness in terms of the policy system. As one of China’s first pilot areas for the national park system, Wuyi Mountain National Park has established a comprehensive FECP framework comprising 11 measures, including compensation for ecological public welfare forests, compensation for forest tenure, and subsidies for the forestry industry. The breadth and depth of its policy coverage hold exemplary value for protected areas across southern China.

To ensure the scientific rigor and representativeness of the sample while considering the feasibility of field surveys, this study adopted a two-stage mixed sampling scheme for data collection. The first stage involved village-level probability sampling. From all 29 administrative villages within the general control zone of Wuyi Mountain National Park and a 2-km buffer zone beyond its boundary, 16 villages were selected as primary sampling units using simple random sampling. This approach ensured unbiased spatial distribution and representativeness of the sample. The second stage consisted of household-level non-probability sampling. Within the selected villages, as a real-time and complete roster of resident households was unavailable for use as a sampling frame, and considering practical factors such as building community trust and ensuring interview success rates, this study employed a field interview strategy combining community-based convenience and purposive sampling to identify respondent households. Specific procedures included: (1) With support from village committees, local cadres familiar with village conditions acted as coordinators, guiding the research team to contact typical households that exhibited diversity and richness of information in terms of livelihood types, degree of resource dependence, and policy participation. (2) Simultaneously, the research team conducted systematic mobile visits within each village to broaden contact and conduct supplementary interviews with accessible and willing households. The research team carried out multiple rounds of field surveys between September 2022 and June 2025, employing structured questionnaires for one-on-one household interviews. The survey period spanned the complete annual cycle of the tea industry and the annual disbursement nodes of forest ecological compensation policies. This design captured seasonal variations in farmers’ livelihood strategies as well as the direct effects of policy implementation, ensuring the data reflected real livelihood decision-making processes. The questionnaire comprehensively covered key variables including basic household information, livelihood capital, livelihood strategy choices, and participation in forest ecological compensation policies. After rigorous logical checks and data cleaning, questionnaires with missing critical information or low response quality were excluded, resulting in a final valid sample of 239 questionnaires.

It is important to note that although this study employed a two-stage mixed sampling design and strove to maximize the representativeness of the sample, potential sampling biases remain and warrant careful consideration. On one hand, there is potential coverage bias at the village sampling stage. Although simple random sampling was used in the first stage, the selection of 16 sample villages may, by chance, not fully cover all types of administrative villages within the study area. This could impose certain limitations when generalizing the research findings to all villages. On the other hand, the more significant potential source of bias lies in the non-probability bias at the household sampling stage. This manifests specifically in several ways: when village cadres guided the interviews, they might have unconsciously tended to recommend households with whom they had good relations, strong willingness to communicate, or positive attitudes towards the policies, potentially leading to a sample skewed in terms of attitudes, social capital, or policy experience. Furthermore, household surveys inherently favor contact with members who reside in the village long-term, possibly missing households that are entirely engaged in out-migration work, have relocated as families, or are seldom present in the village. Additionally, households willing and available for in-depth interviews may differ from non-respondents in personality traits, interest in the research, or current availability of free time, which could introduce a degree of unobservable selection bias.

3.2. Variable Description

(1)

Dependent Variable: The dependent variable is farmers’ livelihood strategies. Farmers’ livelihood strategies refer to the behavioral strategies adopted by households based on their endowment of livelihood capital. Depending on research purposes, contexts, and methodologies, scholars have adopted different typologies for classifying livelihood strategies. Based on participation in non-agricultural livelihood activities, the presence of non-agricultural income, and the proportion of non-agricultural income, livelihood strategies can be categorized into three types: pure agricultural, agriculture-dominant with non-agricultural sidelines, and non-agriculture-dominant with agricultural sidelines. Based on the composition of total household income, strategies can be classified into traditional livelihood specialization, non-agricultural specialization, diversification, and subsidy dependence. According to participation in livelihood activities, farmers’ livelihood strategies can be divided into four categories: crop and forestry cultivation, livestock breeding, non-agricultural business operations, and out-migration for work. Drawing on existing research and considering the specific context of Wuyi Mountain National Park, this paper selects the pure agricultural type, out-migration for work type, and commercial operation type to represent livelihood strategies.

The pure agricultural type refers to households whose livelihood activities center on traditional crop and forestry production, with their main household income derived from agricultural activities directly related to forest resources, such as tea cultivation, moso bamboo management, and seedling cultivation. The out-migration for work type refers to households whose livelihood activities focus on non-agricultural employment. Their main household income comes from wage work performed by household members outside their registered residence or in local non-agricultural sectors. This includes employment in distant factories, construction work, domestic services, and forestry-related non-agricultural labor. The commercial operation type refers to households engaged in self-employed business activities. Their main household income stems from business operations leveraging the ecological resources of Wuyi Mountain National Park or local specialty industries, including ecotourism services, deep processing of forest products, and retail sales.

(2)

Explanatory Variables: Livelihood capital serves as the explanatory variable in this study, encompassing five dimensions: human, natural, social, physical, and financial capital. First, human capital refers to the intangible resources possessed by households, such as knowledge, skills, and health status, that support production and livelihood. This study uses education level and physical health status to measure human capital. Second, natural capital denotes the tangible natural resources directly usable by households and related to the ecological environment. This study employs tea garden area and moso bamboo forest area to measure natural capital. Third, social capital refers to the ability of households to obtain resource support through social networks, trust relationships, and reputation. It serves as a crucial social foundation for farmers to access information, seek opportunities, and expand livelihood options. This study uses social networks and social trust to measure social capital. Fourth, physical capital comprises the material infrastructure and production tools used by households for production and daily life. This study selects house area and the value of household physical assets to measure physical capital. Fifth, financial capital refers to the financial resources at the disposal of and accessible to households, including cash, savings, and credit. This study uses access to loans and annual household income to measure financial capital.

(3)

Moderating Variable: Forest ecological compensation policies (FECPs) serve as the moderating variable in this study. Classifying FECPs into “blood-transfusion” and “blood-making” types for discussion holds greater practical significance and facilitates a more in-depth analysis of the policies. This paper defines the variables based on the specific policy items outlined in the Implementation Measures for Establishing an Ecological Compensation Mechanism in Wuyi Mountain National Park. Accordingly, “blood-transfusion” FECPs specifically include compensation for ecological public welfare forests, subsidies for logging bans and forest management, compensation for forest right owners, and resettlement compensation for ecological migrants. “Blood-making” FECPs primarily refer to subsidies for forestry industry development, subsidies for forestry production equipment, subsidies for the purchase of physical seeds (e.g., seedlings), and similar measures. To obtain specific information on farmers’ participation in FECPs, this study used the survey questions: “How many types of ecological compensation policy subsidies have you received? What are these subsidies? And how are the subsidies specifically distributed?”.

(4)

Control Variables: To control for the potential influence of basic household characteristics on livelihood strategy choices, this study selects the following variables as control variables: First, gender. This refers to the gender of the household head, reflecting gender differences in the primary decision-maker of the household. In rural contexts, gender is often associated with the division of labor, risk preferences, and access to resources. For example, men may be more likely to engage in out-migration for work or commercial operations, while women may focus more on agricultural production. It is necessary to control for its potential effect on livelihood strategy choice. Second, age. This refers to the actual age of the household head, reflecting labor capacity, willingness to adopt new things, and risk tolerance. Older farmers often experience declining labor capacity and have more fixed livelihood pathways, making them more inclined to maintain traditional pure agriculture or rely on subsidies. Younger farmers are more likely to venture into out-migration for work or commercial operations. The influence of this variable needs to be controlled. Third, the number of household laborers. This refers to the total number of family members with the capacity to work who can participate in production and business activities, reflecting the total labor supply of the household. The number of laborers directly determines the scale of human resources a household can allocate to agricultural cultivation, commercial operations, or out-migration for work. For instance, households with sufficient labor are more capable of engaging in commercial operations, while those with labor shortages may rely more on out-migration or subsidies. This constitutes a key foundational variable affecting livelihood strategy choice. The names, definitions, measurements, and descriptive statistics of the variables are presented in

Table 1. Descriptive Statistics of Variables.

Variable Type	Variable Name		Definition and Assignment	Mean	Standard Deviation
Dependent Variable	Livelihood Strategy	Pure agricultural	Traditional farming/forestry as main livelihood activity
		Outmigration for work	Non-farm work as main livelihood activity
		Commercial Operation	Self-run business as main livelihood activity
Explanatory Variable	Human Capital	Education Level	Years of formal education received by household head (1 = illiterate/primary school, 2 = junior high school, 3 = senior high/vocational secondary school, 4 = college, 5 = undergraduate and above)	2.36	0.85
		Health Status	Average health status of family members (1 = very poor, 2 = poor, 3 = general, 4 = good, 5 = excellent)	3.81	1.03
	Natural Capital	Tea Garden Area	Contracted tea garden area of the household (mu)	15.63	21.16
		Bamboo Forest Area	Contracted moso bamboo forest area of the household (mu)	20.04	142.78
	Social Capital	Social Network	Abundance of social relations available for support (1 = very poor, 2 = poor, 3 = general, 4 = good, 5 = excellent)	4.13	0.65
		Social Trust	Trust in village collectives, neighbors and markets (1 = very untrustworthy, 2 = untrustworthy, 3 = general, 4 = trustworthy, 5 = very trustworthy)	3.88	0.92
	Physical Capital	Housing Area	Owned residential and production building area (1 ≤ 50 m2, 2 = 50–100 m2, 3 = 100–150 m2, 4 ≥ 150 m2)	3.47	0.73
		Value of Physical Assets	Value of production fixed assets such as agricultural machinery, processing equipment, and transportation tools (10,000 yuan)	23.00	33.19
	Financial Capital	Loan Convenience	Difficulty of obtaining loans from banks, credit unions and other financial institutions (1 = inconvenient, 2 = general, 3 = convenient)	2.62	0.86
		Household Income	Total income from various sources of the household in the past year (10,000 yuan)	27.82	60.82
Moderating Variable	Forest Ecological Compensation Policy	“Blood-transfusion” Forest Ecological Compensation Policy	Participation in the “blood-transfusion” forest ecological compensation policy (0 = no, 1 = yes)	0.40	0.49
		“Blood-making” Forest Ecological Compensation Policy	Participation in the “blood-making” forest ecological compensation policy (0 = no, 1 = yes)	0.44	0.50
Control Variables	Individual Characteristics	Age	Actual age of the household head (years)	50.38	13.10
		Gender	Gender of the household head (0 = female, 1 = male)	0.79	0.41
	Household Characteristics	Number of Laborers	Number of family members with labor capacity (person)	3.53	1.33

.

3.3. Model Specification

(1)

Multinomial Probit Model. Given that the livelihood strategies of farmers in Wuyi Mountain National Park exhibit the characteristic of multiple category choices, this paper employs the multinomial probit (mprobit) model to conduct an empirical analysis of farmers’ livelihood strategy selection behavior, including three types: “pure agricultural livelihood strategy”, “outmigration for work livelihood strategy”, and “commercial operation livelihood strategy”. Compared with the multinomial logit model, the multinomial probit model does not rely on the “Independence of Irrelevant Alternatives” assumption. It can more flexibly depict the correlation structure between various livelihood strategies, thus being more suitable for the research context of this paper. The general form of the model is as follows (Model 1):

$$P(Y_k=j)={\mathsf{\Phi }}_j({\alpha }_j+X_k{\beta }_j),\hspace{1em}\hspace{1em}j=1,\dots ,J$$

(1)

Among them, $X_k$ is the livelihood capital vector of farmer k, ${\beta }_j$ is the coefficient vector corresponding to the j-th strategy, and ${\mathsf{\Phi }}_j$ is the cumulative distribution function of the multivariate normal distribution.

(2)

Moderating Effect Model. To examine the differentiated moderating effects of the two types of forest ecological compensation policies—“blood-transfusion” and “blood-making”—this study adopts a grouped regression comparison strategy. Specifically, based on the baseline mprobit model, the overall sample is first divided into a “blood-transfusion” policy sample group and a “blood-making” policy sample group according to policy type. Subsequently, within each policy type, the sample is further split into two subsamples based on whether farmers actually participated in the policy: a “participant group” and a “non-participant group”.

The test of grouped moderating effects is implemented by comparing the coefficient differences between the “participant group” and the “non-participant group” under the same policy type. Taking the “blood-transfusion type” policy as an example, let D^TT = 1 denote participation in this policy and D^TT = 0 denote non-participation (Model 2) and (Model 3). Estimate the model separately in the two subsamples:

Model for the Participant Group (D^TT = 1):

$$P(Y_k=j|D^{TT}=1)={\Phi }_j({\alpha }_j^{TT,1}+X_k{\beta }_j^{TT,1})$$

(2)

Model for the Participant Group (D^TT = 0):

$$P(Y_k=j|D^{TT}=0)={\Phi }_j({\alpha }_j^{TT,0}+X_k{\beta }_j^{TT,0})$$

(3)

The criterion for judging the moderating effect lies in comparing the coefficients βj^TT,1 and βj^TT,0 of the key livelihood capital variables (e.g., natural capital and financial capital) in terms of their statistical significance, sign, and magnitude. If there is a significant difference between the two coefficients, it indicates that the “blood-transfusion type” policy exerts a moderating effect on the relationship between this capital dimension and livelihood strategy selection. The path and intensity of this effect can be explained by the direction and magnitude of changes in the coefficients.

The model specification for testing the moderating effect of the “blood-making type” policy is similar to the above. Models (4) and (5) are estimated separately for its participant group (D^HC = 1) and non-participant group (D^HC = 0), and the moderating effect is identified by comparing βj^HC,1 and βj^HC,0.

$$P(Y_k=j|D^{HC}=1)={\Phi }_j({\alpha }_j^{HC,1}+X_k{\beta }_j^{HC,1})$$

(4)

$$P(Y_k=j|D^{HC}=0)={\Phi }_j({\alpha }_j^{HC,0}+X_k{\beta }_j^{HC,0})$$

(5)

4. Result

4.1. Analysis of Farmers’ Livelihood Strategy Choices in National Parks

Based on the classification standards of livelihood strategies from existing research and combined with the field survey results in Wuyi Mountain National Park, this study characterizes the livelihood strategy choices of farmers in Wuyi Mountain National Park into three types: pure agricultural, outmigration for work, and commercial operation. The data statistics are shown in

Table 2. Livelihood Strategy Choices of Farmers in National Parks.

Livelihood Strategy	Frequency	Percent	Cumulative Percentage
Pure Agricultural Type	76	31.80	31.80
Outmigration for Work Type	49	20.50	52.30
Commercial Operation Type	114	47.70	100.00

.

Among the 239 samples, 76 farmers adopted a pure agricultural livelihood strategy, accounting for 31.80% of the total sample. Wuyi Mountain National Park possesses favorable natural environmental and climatic conditions, providing local farmers with basic resource endowments for agricultural product production such as tea planting, tree cultivation, beekeeping, and crop farming. This ensures relatively diverse and stable sources of income from agricultural cultivation. Furthermore, given that tea is the characteristic industry of Wuyi Mountain National Park, tea planting, production, and sales can yield good economic profits. Therefore, farmers within the national park can meet their daily living needs by choosing a pure agricultural livelihood strategy and earning economic returns from it.

There are 49 farmers (20.50% of the total samples) adopting the outmigration for work livelihood strategy. On one hand, with the continuous development and expansion of the tea industry in Wuyi Mountain National Park, the labor demand for tea planting and picking has increased accordingly. When local farmers lack sufficient family labor to undertake the work, they will hire people from surrounding areas (mostly with Jiangxi household registration) to assist in tea planting, with a daily wage of 200–300 yuan per person. On the other hand, the outmigration for work livelihood strategy also includes workers in local towns and subdistricts. Besides some obtaining stable regular wage income, they also participate in tea production and trading to earn income from agricultural planting.

Farmers adopting the commercial operation livelihood strategy account for the largest proportion of the total, with 114 samples (47.70%). As mentioned earlier, the tea industry is the core industry of Wuyi Mountain National Park. Due to its high economic added value, many local farmers choose to operate tea shops, establish tea production workshops, etc., in addition to basic agricultural planting. This is most typical among farmers in Xingcun Town. Furthermore, Wuyi Mountain National Park has excellent cultural and tourism resources, attracting a large number of tourists every year. Thus, many farmers choose to open restaurants and hotels locally. Notably, Nanyuanling Village is an ecological resettlement community of Wuyi Mountain National Park. Guided by the government, the entire village was relocated to an area adjacent to the Wuyi Mountain National Tourist Resort, connected to the Xiandian Ecological Entrepreneurship Park.

4.2. Impact of Farmers’ Livelihood Capital on Their Livelihood Strategy Choices in National Parks

To verify the research hypothesis proposed earlier—that farmers’ livelihood capital significantly influences their choice of livelihood strategies, and that the direction of effects varies across different dimensions of capital (H1)—this chapter employs the mprobit model using Stata 16 for empirical testing. Using the pure agricultural livelihood strategy as the reference group, the analysis examines the impact of different livelihood capitals on the two livelihood strategy types: out-migration for work and commercial operation. The results of the multicollinearity test show that the maximum variance inflation factor (VIF) for all variables is 1.56, with a mean VIF of only 1.33, both well below the critical threshold of 10. This indicates that the model does not suffer from multicollinearity issues, and the estimation results are reliable. The regression results are presented in

Table 3. Results of Benchmark Regression Analysis.

Livelihood Capital	Indicator	Outmigration for Work (Pure Agricultural Type as Reference)	Commercial Operation (Pure Agricultural Type as Reference)
Human Capital	Education Level	0.069 (0.224)	0.100 (0.191)
	Health Status	0.269 (0.168)	0.240 * (0.142)
Natural Capital	Tea Garden Area	−0.047 ** (0.019)	−0.008 (0.008)
	Bamboo Forest Area	0.002 (0.003)	0.001 (0.003)
Social Capital	Social Network	−0.598 ** (0.288)	0.333 (0.222)
	Social Trust	0.383 * (0.223)	−0.285 * (0.170)
Physical Capital	Housing Area	0.279 (0.217)	0.153 (0.181)
	Value of Physical Assets	0.008 (0.009)	0.012 * (0.006)
Financial Capital	Loan Convenience	0.204 (0.237)	0.140 (0.202)
	Household Income	−0.023 (0.015)	0.004 (0.005)
Prob > chi2	0.0103
Wald chi2	45.51
Log likelihood	−212.838

Note: * and ** respectively indicate that the coefficient is significant at the 10% and 5% levels.

. The overall model significance test yields a Prob > chi² value of 0.0103, indicating that livelihood capital has statistically significant explanatory power for farmers’ livelihood strategy choices. The specific analysis is as follows:

First, the effects of human capital on farmers’ livelihood strategies vary. Health status has a significantly positive effect on the choice of the commercial operation strategy (coefficient = 0.24, p < 0.1). This indicates that farmers in better health are more inclined to choose a commercial operation livelihood. This is because business activities such as tea sales and homestay services require considerable physical effort and on-site involvement; good health serves as a fundamental condition for engaging in such activities, which aligns with theoretical expectations. However, education level does not show a significant effect on either the out-migration for work or the commercial operation strategy. This is primarily because the dominant industries in Wuyi Mountain National Park—tea and tourism services—rely more on practical experience and local resource knowledge than on formal education, thereby weakening the measurable effect of education level.

Second, the constraining effect of natural capital on livelihood strategies is evident. Tea garden area has a significantly negative effect on the out-migration for work strategy (coefficient = −0.047, p < 0.05), meaning that larger tea garden areas are associated with a lower probability of farmers choosing out-migration. This is because tea gardens require continuous daily management, which ties up household labor and makes farmers more inclined to stay locally for agricultural production. In contrast, moso bamboo forest area does not significantly affect either strategy. This is likely due to the highly seasonal nature of bamboo forest management, where labor inputs are highly flexible and impose weaker constraints on farmers’ daily livelihood arrangements, thus showing no significant impact on strategy choice.

Third, the impact of social capital on livelihood strategies exhibits complex heterogeneity. Social networks have a significantly negative effect on the out-migration for work strategy (coefficient = −0.598, p < 0.05), indicating that farmers with denser social networks within Wuyi Mountain National Park are less willing to migrate for work. This is because local networks provide access to local livelihood opportunities such as agricultural cooperation and neighborhood odd jobs, reducing the necessity to seek work elsewhere. Notably, social trust influences the two strategies in opposite directions: it has a significantly positive effect on out-migration for work (coefficient = 0.383, p < 0.1) but a significantly negative effect on commercial operation (coefficient = −0.285, p < 0.1). This suggests that social trust plays different roles across strategies. Higher social trust may facilitate riskier out-migration for employment, whereas its effect on commercial operations—which require long-term cooperation and reputation building—is more complex.

Fourth, the influence of physical capital on livelihood strategies is relatively limited. Only the value of physical assets shows a weak positive effect on the commercial operation strategy (coefficient = 0.012, p < 0.1). This suggests that farmers possessing a certain scale of housing facilities, transportation equipment, and production tools are better positioned to engage in small-scale local business operations, consistent with theoretical expectations. However, housing area does not significantly affect either strategy, possibly due to ecological protection planning in Wuyi Mountain National Park that restricts housing modifications, thereby limiting the functionality of houses as business premises.

Fifth, financial capital does not show a significant effect on livelihood strategies. Neither access to loans nor annual household income significantly influences the choice of either livelihood strategy. This may be attributed to two reasons. First, farmers generally face credit constraints; nominal loan availability often does not translate into actual financing capacity. Second, household income primarily comes from agricultural operations, with limited accumulation scale that mostly covers daily expenses and is insufficient to support the investments required for transitioning to alternative livelihood strategies.

4.3. Analysis of the Moderating Effect of Forest Ecological Compensation on the Impact of Farmers’ Livelihood Capital on Their Livelihood Strategies

This chapter employs the Multinomial Probit model to test the moderating effect of forest ecological compensation policies (FECPs) on the relationship between farmers’ livelihood capital and their livelihood strategies through group regression. The total sample is divided into two groups: “participated in FECPs” and “did not participate in FECPs”, with separate model estimations conducted for each. The key findings are as follows:

(1)

Analysis of the Moderating Effect of “Blood-Transfusion” Forest Ecological Compensation Policies on the Relationship between Farmers’ Livelihood Capital and Their Livelihood Strategies (

Table 4. The Moderating Effect of “Blood-Transfusion” Forest Ecological Compensation on the Impact of Livelihood Capital on Livelihood Strategies.

		Outmigration for Work		Commercial Operation
		Participated in Blood-Transfusion	Non-Participated in Blood-Transfusion	Participated in Blood-Transfusion	Non-Participated in Blood-Transfusion
Natural Capital	Tea Garden Area	−0.004 [−0.016, 0.007]	−0.008 ** [−0.015, 0.001]	−0.009 [−0.021, 0.003]	0.002 [−0.003, 0.007]
	Bamboo Forest Area	0.001 [−0.000, 0.001]	−0.006 [−0.015, 0.002]	−0.001 [−0.001, 0.001]	0.009 ** [0.001, 0.019]
Financial Capital	Loan Convenience	0.042 [−0.072, 0.155]	−0.016 [−0.111, 0.078]	−0.064 [−0.179, 0.052]	0.128 * [−0.001, 0.257]
	Household Income	0.002 [−0.006, 0.010]	−0.009 ** [−0.016, −0.001]	0.024 *** [0.015, 0.033]	0.005 ** [0.001, 0.009]
Sample		N = 86	N = 153	N = 86	N = 153
Prob > chi2		0.237	0.069	0.237	0.069
Wald chi2		30.78	37.41	30.78	37.41
Log likelihood		−54.298	−124.651	−54.298	−124.651

Note: *, **, *** respectively indicate that the coefficient is significant at the 10%, 5%, and 1% levels.

).

First, the moderating effect of “blood-transfusion” forest ecological compensation policies on the relationship between natural capital and the out-migration for work livelihood strategy (Figure 6). Among farmers who did not participate in the “blood-transfusion” compensation policy, each additional mu of tea garden area reduced the probability of choosing the out-migration strategy by an average of 0.8 percentage points, and the effect was significant at the 5% level. This confirms the “locking-in effect” of natural capital: the larger the tea garden area, the less willing farmers are to migrate for work. However, in the group that participated in the “blood-transfusion” policy, the effect of tea garden area became statistically insignificant (marginal effect = –0.004, p > 0.1), and its magnitude was halved. This change indicates that the direct income support provided by “blood-transfusion” compensation effectively alleviates the pressure that keeps farmers from migrating due to tea-garden maintenance needs, partially offsetting the “locking-in effect” of natural capital. Consequently, it weakens the constraint that natural capital endowment imposes on labor outflow, supporting Hypothesis H2a.

Second, the moderating effect of “blood-transfusion” forest ecological compensation policies on the relationship between financial capital and the out-migration for work livelihood strategy (Figure 6). In the group that did not participate in any forest ecological compensation policy, each 10,000-yuan increase in annual household income reduced the probability of choosing out-migration by an average of 0.9 percentage points (significant at the 5% level). This suggests that in the absence of “blood-transfusion” compensation, the scarcer a household’s own financial capital, the stronger the motivation to migrate for work in search of stable cash flow. In the policy-participation group, however, this relationship became statistically insignificant (marginal effect = 0.002, p > 0.1). This shift confirms that “blood-transfusion” compensation, as an exogenous and stable cash inflow, provides a direct livelihood buffer for low-income households, reducing their economic vulnerability to local income fluctuations or shortfalls that would otherwise force them to migrate. Thus, it weakens the driving effect of household financial-capital scarcity on out-migration decisions, validating Hypothesis H2b.

Third, the moderating effect of “blood-transfusion” forest ecological compensation policies on the relationship between natural capital and the commercial operation livelihood strategy (Figure 7). In the group that did not participate in “blood-transfusion” compensation, each additional mu of moso-bamboo forest area raised the probability of choosing commercial operation by an average of 0.9 percentage points, significant at the 5% level. This indicates that, in the absence of such policies, greater bamboo-forest resources constitute both a resource base and a motive for farmers to shift toward commercial operations, compensating for traditional income losses due to conservation restrictions. However, in the policy-participation group, this driving effect disappeared (marginal effect = –0.001, insignificant). This suggests that “blood-transfusion” compensation directly fills the income gap caused by restricted resource use, so that farmers no longer need to rely on developing their own bamboo-forest resources for commercial conversion to maintain income. Thereby, it weakens the positive driving force of natural-capital stock on the commercial-operation strategy, supporting Hypothesis H2c.

Fourth, the moderating effect of “blood-transfusion” forest ecological compensation policies on the relationship between financial capital and the commercial operation livelihood strategy (Figure 7). Regarding annual household income, the marginal effect in the non-participation group was 0.005 (significant at the 5% level), indicating that higher household income enhances the tendency toward commercial operation. In the participation group, the marginal effect increased to 0.024 (significant at the 1% level), with both significance and magnitude rising. This shows that the cash injection from “blood-transfusion” compensation supplements farmers’ financial-capital stock and strengthens the supportive capacity of household income for commercial operations. Regarding access to loans, loan convenience had a positive effect in the non-participation group (marginal effect = 0.128, significant at the 10% level), whereas in the participation group the effect was insignificant and even reversed in direction (–0.064). This is likely because direct cash compensation partially substitutes for the need for external financing, especially for small-scale operations. Overall, the “blood-transfusion” policy mainly strengthens the support of financial capital for commercial operation by boosting household cash flow rather than improving credit channels, validating Hypothesis H2d.

(2)

Analysis of the Moderating Effect of “Blood-Transfusion” Forest Ecological Compensation Policies on the Relationship between Farmers’ Livelihood Capital and Their Livelihood Strategies (

Table 5. The Moderating Effect of “Blood-Making” Forest Ecological Compensation on the Impact of Livelihood Capital on Livelihood Strategies.

		Outmigration for Work		Commercial Operation
		Participated in Blood-Making	Non-Participated in Blood-Making	Participated in Blood-Making	Non-Participated in Blood-Making
Social Capital	Social Network	−0.162 *** [−0.284, −0.040]	−0.129 ** [−0.250, −0.009]	0.154 * [−0.008, 0.317]	0.197 *** [0.068, 0.327]
	Social Trust	0.072 * [−0.006, 0.150]	0.122 ** [0.022, 0.222]	−0.099 * [−0.211, 0.120]	−0.181 *** [−0.284, 0.078]
Physical Capital	Housing Area	−0.003 [−0.096, 0.091]	0.086 * [−0.011, 0.183]	−0.012 [−0.147, 0.124]	−0.025 [−0.134, 0.084]
	Value of Physical Assets	0.001 [−0.003, 0.003]	−0.001 [−0.004, 0.004]	0.004 * [0.001, 0.007]	−0.001 [−0.006, 0.003]
Sample		N = 106	N = 133	N = 106	N = 133
Prob > chi2		0.4493	0.1166	0.4493	0.1166
Wald chi2		26.25	34.78	26.25	34.78
Log likelihood		−79.659	−118.942	−79.659	−118.942

Note: *, **, *** respectively indicate that the coefficient is significant at the 10%, 5%, and 1% levels.

).

First, the moderating effect of “blood-making” forest ecological compensation policies on the relationship between social capital and the out-migration for work livelihood strategy (Figure 8). The two dimensions of social capital show markedly different patterns. On the one hand, social networks consistently exhibit an “anchoring” effect: regardless of policy participation, a one-unit increase in social network level significantly reduces the probability of out-migration (by 12.9 percentage points in the non-participation group and 16.2 percentage points in the participation group). After participating in the “blood-making” policy, this constraining effect strengthened by about 25.6%, indicating that the policy enhances the cohesive force of local social networks, further retaining labor within Wuyi Mountain National Park. On the other hand, social trust shows a “driving” effect, but its potency is weakened by the policy: in the absence of the policy, a one-unit increase in social trust raises the probability of out-migration by 12.2%; after policy participation, this driving effect diminishes to 7.2%, a reduction of about 41%. This suggests that the formal cooperation and information channels provided by the “blood-making” policy partially substitute for the traditional reliance on generalized social trust to obtain external employment information. Therefore, Hypothesis H3a receives only partial support: while the policy indeed weakens the driving effect of social trust, it does not alter the direction of the anchoring effect of social networks.

Second, the moderating effect of “blood-making” forest ecological compensation policies on the relationship between physical capital and the out-migration for work livelihood strategy (Figure 8). The influence of physical capital varies by indicator. Housing area, as a stock-based wealth indicator, shows a significant pull effect on labor outflow when farmers do not participate in the policy (marginal effect = 0.086, significant at the 10% level), suggesting that households with better housing conditions may have more economic slack to support members seeking opportunities elsewhere. However, after participating in the “blood-making” policy, this pull effect disappears entirely and slightly reverses (–0.003, insignificant). A possible explanation is that the policy enhances the attractiveness and feasibility of local livelihoods through industrial support, so that households no longer need to “crowd out” labor through migration to optimize resource allocation, thereby dissolving the link between housing-area-represented wealth stock and migration decisions. The value of physical assets does not affect out-migration decisions in either group, indicating that the quantity of production tools and equipment per se does not directly promote or inhibit labor mobility, regardless of policy intervention. Overall, Hypothesis H3b receives partial support: the “blood-making” policy weakens the driving effect of “non-productive” physical capital (represented by household wealth stock) on out-migration, but its effect on productive physical assets remains insignificant.

Third, the moderating effect of “blood-making” forest ecological compensation policies on the relationship between social capital and the commercial operation livelihood strategy (Figure 9). Social networks have a significantly positive effect on the commercial operation strategy in both groups, consistent with the theoretical expectation that social networks provide market information and business opportunities (as posited in H3c). However, the positive effect in the participation group (0.154) is smaller than in the non-participation group (0.197), and the significance level is lower. This indicates that participation in the “blood-making” policy does not strengthen the promotional effect of social networks on commercial operation; rather, it slightly attenuates it. Social trust shows a significantly negative effect in both groups, and the negative effect is weaker in the participation group (–0.099) than in the non-participation group (–0.181). This suggests that, regardless of policy participation, a higher level of social trust is associated with a lower probability of choosing commercial operation. A possible reason is that, in Wuyi Mountain National Park, social trust is more embedded in traditional, non-market mutual-aid relationships rather than oriented towards riskier commercial activities; alternatively, households with high trust may prefer stable production modes and hold conservative attitudes toward market risks.

Fourth, the moderating effect of “blood-making” forest ecological compensation policies on the relationship between physical capital and the commercial operation livelihood strategy (Figure 9). Regarding housing area, both the participation and non-participation groups show an insignificant negative effect on the commercial operation strategy, with confidence intervals including zero. This indicates that, irrespective of policy participation, housing size has no statistically significant promotional effect on choosing the commercial operation strategy. This does not align with the theoretical expectation that housing, as a fixed asset, could support business operations—likely because housing area reflects living conditions and residential assets rather than directly productive fixed assets for commercial activities. However, the role of the value of physical assets changes under policy intervention: in the non-participation group its effect is weak and insignificant (–0.001); after policy participation, it shows a significant positive driving effect, with each 10,000-yuan increase in asset value raising the probability of commercial operation by an average of 0.4 percentage points (significant at the 10% level). This shift suggests that the core of the “blood-making” policy lies in its accompanying industrial support and market linkages, which can transform farmers’ dormant, general-purpose physical assets into specialized productive capital suitable for specific green industries, thereby significantly enhancing the marginal productivity and commercial conversion rate of physical assets. Consequently, Hypothesis H3d is supported in the dimension of physical asset value but not in the dimension of housing area. All research hypotheses and their corresponding test results are presented in

Table 6. Summary of Hypothesis Testing Results.

Hypothesis No.	Hypothesis Content	Verification Result
H2a	“Blood-transfusion” FECPs will weaken the positive driving effect of insufficient natural capital on the out-migration for work strategy.	Supported
H2b	“Blood-transfusion” FECPs will weaken the positive driving effect of scarce financial capital on the out-migration for work strategy.	Supported
H2c	“Blood-transfusion” FECPs will weaken the positive driving effect of natural capital on the commercial operation strategy.	Supported
H2d	“Blood-transfusion” FECPs will strengthen the positive supporting effect of financial capital on the commercial operation strategy.	Supported
H3a	“Blood-making” FECPs will weaken the positive driving effect of social capital on the out-migration for work strategy.	Partially Supported
H3b	“Blood-making” FECPs will weaken the positive driving effect of insufficient physical capital on the out-migration for work strategy.	Partially Supported
H3c	“Blood-making” FECPs will strengthen the positive driving effect of social capital on the commercial operation strategy.	Not Supported
H3d	“Blood-making” FECPs will strengthen the positive driving effect of physical capital on the commercial operation strategy.	Supported

.

4.4. Robustness Checks

To ensure the reliability of the research findings, further robustness checks were conducted based on econometric methods and data processing. (1) Changing the model estimation method. The baseline regression employed the mprobit model. To test for potential bias due to model specification, the estimation method was replaced with the mlogit model for re-estimation. Although these two models are based on different assumptions regarding the distribution of random errors, both are suitable for unordered multinomial dependent variables. The direction and statistical significance level of the marginal effects for each variable were highly consistent with the baseline regression results. This indicates that the core conclusions are not sensitive to the specification of different discrete choice models, demonstrating good robustness. (2) Robustness checks related to sample and data processing. First, a random subsample test was performed. A random 90% of observations were drawn from the total sample (N = 239) to re-estimate the baseline model and the moderating effect models. This process was repeated 500 times to observe the stability of the coefficients. The results showed that the signs and significance of the core explanatory variables remained stable in the vast majority of the samplings, indicating that the conclusions are not unduly influenced by individual observations. Second, all continuous variables were winsorized at the top and bottom 1% to control for potential interference from outliers. Regression results based on the winsorized data showed that, while the estimated coefficients of the main variables were slightly adjusted in magnitude, their significance and direction of effect were completely consistent with the main regression, confirming that the conclusions are not sensitive to extreme values. In summary, by changing the model estimation method, performing repeated random subsampling, and winsorizing continuous variables, the main empirical conclusions of this study remain stable, further enhancing the credibility of the findings.

5. Discussion

Against the backdrop of the global biodiversity crisis and the transformation of natural protected area governance, forest ecological compensation policies (FECPs) have become a key institutional instrument for reconciling the conflict between conservation and development. Based on the Sustainable Livelihoods Framework and taking Wuyi Mountain National Park as a case study, this research empirically demonstrates that both “blood-transfusion” and “blood-making” FECPs moderate the relationship between farmers’ livelihood capital and their strategy choices through differentiated pathways. The following section will synthesize the main findings, clarify the research implications, acknowledge limitations, and outline directions for future work.

5.1. Main Findings

The core mechanism of “blood-transfusion” forest ecological compensation policies (FECPs) can be summarized as “compensation” and “buffering”. Through direct cash transfers, they fill the income gaps or liquidity shortages faced by farmers due to ecological protection restrictions. This exogenous resource injection primarily serves to alleviate livelihood pressure and reduce vulnerability, thereby weakening the push factors that force households to adopt passive strategies due to resource or capital shortages. Even when promoting commercial operation to some extent, the channel of influence is not through large-scale productive investment, but rather by supplementing household liquidity to support low-barrier business activities, and partially substituting the reliance on formal credit. Therefore, “blood-transfusion” FECPs essentially function as a livelihood safety net, effectively preventing livelihood deterioration but rarely triggering sustainable endogenous development.

In contrast, “blood-making” FECPs embody a deeper logic of “empowerment and restructuring”. Their moderating effect is not simply to strengthen or weaken existing relationships, but to transform the nature and outcomes of farmers’ capital through productive investment and organizational development. This is highlighted in two key aspects: First, activating and transforming capital, for instance, converting general physical assets into specialized productive capital suitable for local green industries, thereby stimulating their driving force toward commercial operations. Second, reshaping the function of social capital: the formal cooperative platforms embedded in “blood-making” FECPs not only strengthen the “anchoring effect” of social networks but also partially replace the role of traditional social trust in accessing external employment information, and may even substitute informal networks in transmitting business information through institutionalized channels. This further demonstrates that “blood-making” FECPs can intervene in the micro-level process of how farmers utilize their capital, structurally adjusting their livelihood patterns with more profound and lasting impacts.

Additionally, the results reveal heterogeneous responses across different dimensions of livelihood capital. Within social capital, social networks and social trust respond differently in direction and intensity to the same policy; similarly, within physical capital, housing area and the value of physical assets show distinct patterns. Therefore, in both academic analysis and policy design, livelihood capital should not be treated as a homogeneous whole—instead, a more refined dimensional deconstruction is necessary.

5.2. Research Implications

The findings of this study carry both theoretical and practical significance. Theoretically, by constructing an analytical model of “FECPs → livelihood capital → livelihood strategies”, the research deepens the understanding of the micro-level mechanisms through which FECPs operate. Differentiating FECPs into “blood-transfusion” and “blood-making” types and validating their distinct moderating pathways enriches the theoretical knowledge of the heterogeneous effects of ecological compensation policies. Furthermore, insights gained in the context of Wuyi Mountain National Park—a setting characterized by complex property rights and high livelihood dependence on natural resources—extend the application of the Sustainable Livelihoods Framework to the governance of protected areas, highlighting the crucial moderating roles of institutional environments and resource characteristics.

Practically, this study offers concrete, actionable insights grounded in Chinese local experience for optimizing the ecological compensation mechanisms of protected areas centered on national parks. The research points out that in southern collective-forest national parks like Wuyi Mountain, relying solely on “blood-transfusion” compensation is insufficient to achieve sustainable livelihood transformation. The effectiveness of “blood-making” policies, on the other hand, depends on their precise alignment with local industrial features, social networks, and capital structures. This provides a basis for managers to design more targeted and differentiated FECPs.

5.3. Limitations

This study has three main limitations:

• First, the use of cross-sectional data can only reflect policy effects at a specific point in time; it cannot capture the dynamic evolution of farmers’ livelihood capital or the long-term impacts of policies, making it difficult to reveal the temporal adjustment patterns of livelihood strategies. Future research could employ longitudinal panel data to track the long-term evolution of policy effects and enhance data reliability by cross-verifying policy participation with administrative records.
• Second, information on participation in FECPs relied on farmers’ self-reporting, which may be subject to recall or perception biases. Meanwhile, the measurement of social capital and financial capital still has room for improvement—for example, the structural characteristics of social networks and variations in credit-constraint intensity were not considered, which may affect the precision of the results.
• Third, the sample focused solely on Wuyi Mountain National Park, and the generalizability of the conclusions needs to be further tested in protected areas with different property-rights regimes and industrial structures. Future studies could expand the scope of research and compare policy implementation effects across regions with different tenure systems and industrial types to clarify the applicable boundaries and optimization directions of FECPs.

6. Conclusions and Policy Recommendations

6.1. Conclusions

Taking Wuyi Mountain National Park as a case study and working based on micro-survey data from 239 farming households, this research employed the mprobit model and moderating effect models to explore the impact of farmers’ livelihood capital on their livelihood strategy choices under the moderating effect of forest ecological compensation policies (FECPs). The core conclusions are as follows:

(1)

The configuration of farmers’ livelihood capital significantly influences their choice of livelihood strategy, with different dimensions of capital playing distinct roles.

(2)

“Blood-transfusion” and “blood-making” FECPs exhibit differentiated moderating effects on the relationship between livelihood capital and livelihood strategies. “Blood-transfusion” policies, relying primarily on direct cash compensation, function to “fill gaps and stabilize livelihoods”. They significantly mitigate the driving force of insufficient natural and financial capital on out-migration for work and provide initial financial support for low-barrier commercial activities.

(3)

“Blood-making” policies, through productive investment and organizational empowerment, achieve “activation and transformation”. They effectively activate the commercial production attributes of physical assets, driving local commercial operations. Simultaneously, they reshape the functions of social capital, enhancing its local cohesion while partially substituting its traditional role in external connections.

(4)

Overall, the two policy types play complementary roles in sustainable livelihood transition: “Blood-transfusion” FECPs focus on protective security, safeguarding the livelihood baseline and preventing poverty induced by conservation; “Blood-making” policies emphasize transformative empowerment, fostering endogenous drivers and promoting the development of eco-friendly industries.

6.2. Policy Recommendations

Based on the above conclusions, to enhance the precision and effectiveness of FECPs in national parks and promote sustainable community development, the following recommendations are proposed:

• First, implement a targeted compensation system characterized by “categorized interventions and orderly linkage”. The design of FECPs should be differentiated based on the livelihood capital status of farming households. For vulnerable groups, ensure the timely and full disbursement of “blood-transfusion” funds to solidify the livelihood foundation underlying ecological conservation. For households with certain resources and development potential, increase “blood-making” investments and ensure effective linkage, guiding their transition toward ecological industrialization through industrial support, skills training, and market connections, thereby cultivating an endogenous mechanism that mutually reinforces conservation and development.
• Second, innovate and establish a “blood-transfusion” FECP allocation mechanism combining “collective coordination with household-level benefits”. Given the property rights feature of Wuyi Mountain National Park, where collective forests account for about two-thirds of the area, village collectives should coordinate a portion of the compensation funds for improving shared infrastructure such as tea processing facilities and ecotourism trails within the forest area. The remaining funds should then be distributed to households according to their forest tenure shares. This approach avoids the problems of low per-household compensation amounts and weak policy effects while balancing collective interests with individual rights.
• Third, optimize the design of “blood-making” policies to strengthen the formation of productive capital. Industrial subsidies and equipment support should focus on areas that can generate specialized ecological production capital, such as smart forestry facilities and standardized renovations for eco-homestays. Concurrently, enhance the cultivation of new types of business entities like cooperatives and family forest farms, emphasizing their organic integration with local social networks. These entities should serve as comprehensive platforms for resource integration, market linkage, and risk diversification, rather than simply replacing traditional forms of cooperation.
• Fourth, establish and improve dynamic monitoring, evaluation, and adaptive management mechanisms for FECPs. Given the long-term and complex nature of FECP effects, a comprehensive monitoring network covering livelihood capital, strategy choices, ecological outcomes, and community perceptions should be developed. Regular policy evaluations should be conducted, paying particular attention to the heterogeneous impacts of policies on different capital dimensions and social groups. Based on evaluation results, dynamically optimize compensation standards, target groups, and methods to achieve adaptive policy management, ensuring ongoing alignment with the dual objectives of community development and ecological conservation in national parks.

In summary, successful FECPs require not only financial investment but also precise design and synergistic efforts based on regional resource endowments, industrial structures, and the characteristics of farmers’ capital. This study can provide a reference for optimizing FECPs in similar protected areas, contributing to the long-term synergy between ecological conservation and community development in natural protected areas, and offering a Chinese approach to aligning global sustainable development goals with local governance practices.