Economic Impacts and Spatial Spillovers of the National Park Pilot Policy: Evidence from Yunnan, China

Abstract

National parks are a key institutional tool for coordinating ecological conservation and sustainable development. This paper takes the pilot national park program in Yunnan Province, China, as a case study. Using panel data from 127 counties between 2001 and 2023, we empirically examine the economic impact of the national park pilot program using a Time-varying difference-in-differences (DID) approach and a Spatial Durbin Model (SDM). The study finds that (1) the pilot policy significantly increased per capita GDP in the counties by approximately 5057 RMB, with a 4- to 5-year lag effect and a long-term marginally increasing trend; (2) the policy drives economic growth through three main channels: increased fiscal transfers from higher levels of government, induced industrial upgrading, and the stimulation of fixed-asset investment; (3) the policy’s impact is more significant in areas with low economic levels, high altitudes, and high ecological quality; (4) national parks not only stimulate local economic growth but also promote coordinated development in surrounding regions through significant spatial spillover effects. This paper confirms the feasibility of transforming ecological advantages into economic advantages and provides empirical evidence for optimizing spatial governance in “Global South” countries.

1. Introduction

As global awareness of ecological protection and sustainable development deepens, the national park system has become a pivotal arrangement for countries to construct high-level nature conservation systems, serving as a spatial governance tool that balances the realization of ecological values with regional development [1]. The International Union for Conservation of Nature (IUCN) classifies national parks as “Category II” protected areas, emphasizing the holistic protection of large-scale natural or near-natural ecosystems [2], while simultaneously mandating the accommodation of public education, scientific research, cultural experience, and recreational functions [3,4]. Since the Convention on Biological Diversity and the 2030 Agenda for Sustainable Development promoted “harmony between humanity and nature” as a global consensus, achieving a balance between ecological conservation and regional economic development has become a core concern in policy design and academic research [5]. Against this backdrop, scientifically evaluating the causal effects, transmission pathways, and spatial dimensions of national park establishment on regional economic development is of significant importance for optimizing institutional arrangements and fostering coordinated regional development.

Existing research on the economic effects of national parks can be categorized into two perspectives. On one hand, strict ecological regulations may constrain development activities and inhibit resource-dependent industries, potentially limiting economic growth in certain regions [6]. On the other hand, scholars argue that conservation and development are not mutually exclusive [7]. Through rational institutional design, national parks can serve as critical vehicles for realizing ecological value [4,8,9]. Specifically, ecological compensation can alleviate conservation costs, ecotourism can cultivate new growth points, and the national park brand effect can enhance regional competitiveness [10]. These mechanisms not only drive green transformation [11] but also incentivize communities to shift from “passive protection” to “active protection” [12], thereby achieving synergy between ecology and economy.

Regarding mechanisms, the majority of empirical studies indicate that national parks generate economic multiplier effects primarily through tourism [13,14]. For instance, the establishment of U.S. national parks increased local employment by an average of 4% and income by 6%, effects largely driven by visitor spending rather than government investment [15,16]. Austria’s Hohe Tauern National Park generates approximately €700 million in economic benefits annually, with each unit of tourist expenditure producing 1.7 times the local economic spillover [17]. Australia’s Dorrigo National Park contributes 8.4% to local employment, demonstrating significantly higher economic leverage per tourist compared to surrounding parks [18]. In South Africa, the Kruger National Park case shows that $53 million in direct tourism expenditure can be amplified into $95 million in total output, supporting over 11,000 jobs [19]. Even when facing a 40% leakage of tourism revenue, Zambia’s South Luangwa National Park still contributes 40% of local household income through employment and local procurement [20]. These studies consistently demonstrate that the economic value of national parks is reflected not only in direct revenues but also in the overall activation of regional economic systems.

However, the economic benefits of national parks exhibit distinct spatial and inter-group inequalities [21,22]. Research in Costa Rica indicates that while wages for tourism workers near park entrances are 6–8% higher than in distant areas, the agricultural sector benefits little [23]. Wages for migrant workers increased by 14–16%, far exceeding those of local residents [24], suggesting that national parks may restructure income distribution through factor mobility. In Italy, eight national parks promoted tourism and employment growth in the short term but had a negative impact on the number of agricultural operating units [25]. Although South Africa’s Kruger National Park created 4000 local jobs, wage disparities were significant: unskilled positions earned approximately $3000 annually, while skilled positions reached $10,000 [10]. Chinese cases show similar divergence: the income structure of farmers around Wuyishan has not changed significantly [6], whereas the Giant Panda National Park improved the income of low-income farmers through transfer payments [26]. In Qianjiangyuan National Park, while about 37% of residents saw income growth, 10% suffered losses due to the externalities of ecological protection [27]; similarly, the policy dividends of Nanshan National Park flowed more to wealthy households, exacerbating the income gap [28]. Furthermore, while ecotourism in Shennongjia National Park promoted economic growth, it led to vegetation degradation and water quality decline [29], highlighting the challenges of sustainable development.

Despite the rich findings in the existing literature regarding the ecological values and economic effects of national parks, several deficiencies remain. First, existing studies predominantly focus on micro-level case studies [26,28,30] or descriptive statistics [31] of single parks, lacking systematic quantitative assessments of policy net effects based on macro-level panel data. This perspective, limited to micro-cases, is often constrained by the specific resource endowments and development stages of the case sites, leading to a lack of generalizability and making it difficult to extrapolate case experiences into general policy evidence [16]. Second, even among the few quantitative studies, observation periods are often short [6]. Constrained by insufficient time spans, these studies fail to capture the long-term dynamic characteristics of policy effects, thereby missing the potential lagged release of institutional dividends from “initial investment” to “later output.” Finally, the existing literature generally overlooks the spatial dimension of national parks’ economic impacts. National parks are not isolated ecological islands; their establishment generates significant externalities for surrounding regions through ecosystem service flows, transportation improvements, and industrial linkages [32]. However, few studies have placed national parks within a regional economic system to examine their “growth pole” effects from the perspective of spatial spillovers.

Against this backdrop, this study focuses on Yunnan Province, a pioneer in China’s national park reform. Utilizing panel data from 127 counties spanning the period 2001–2023, we construct a Time-varying difference-in-differences (DID) model and a Spatial Durbin Model (SDM) to systematically evaluate the direct impacts and spatial spillover effects of national park establishment on county-level economies. Furthermore, we explore the intrinsic mechanisms from the perspectives of fiscal transfer payments, industrial structure upgrading, and fixed-asset investment.

The marginal contributions of this paper are primarily reflected in three aspects: First, in terms of the research sample, this study transcends the limitations of previous micro-case studies. By conducting a systematic assessment based on full-sample panel data from 127 counties in Yunnan, this approach effectively avoids the issue of limited generalizability caused by the specificity of single case sites, significantly enhancing the representativeness and external validity of the conclusions. Second, in terms of the temporal dimension, this study leverages the advantages of long-series data spanning 23 years. Combined with the Time-varying DID model, it accurately identifies the dynamic evolutionary laws of policy effects, effectively bridging the gap in the existing research where short observation periods fail to capture the lagged release characteristics of institutional dividends. Third, regarding the spatial dimension, this paper moves beyond the traditional perspective that neglects spatial correlations. By introducing spatial econometric models, we quantify the spillover effects of national parks on surrounding areas, revealing the spatial attenuation characteristics of these effects and identifying their effective radiation boundary, thereby providing a more comprehensive empirical basis for regional eco-economic governance.

2. Background and Theoretical Hypothesis

2.1. Regional Background

The establishment of Yellowstone National Park in the United States in 1872 marked the inception of the modern national park system. With a core philosophy of “ecological protection as the top priority,” the U.S. model underscores both the public welfare nature of national parks and their symbolic role as national representations. This paradigm has since influenced the development of national park systems in numerous countries [5]. By 2010, at least one national park had been formally established in 161 sovereign states and 23 non-sovereign territories, making national parks one of the most prevalent forms of protected areas globally [1].

Inspired by international experience, China began to explore localized models of national parks. Yunnan Province, one of the most biodiverse regions in China, was at the forefront of this process. As early as 1996, Yunnan collaborated with international organizations to initiate national park research and institutional experiments, notably through the case of Yulong Snow Mountain. These efforts marked an early shift toward ecosystem-based, integrative conservation models. In 2007, Yunnan established Pudacuo National Park—the first provincial-level national park in China—thus launching the locally driven pilot phase of national park development. In 2008, the former State Forestry Administration officially designated Yunnan as a pilot province for national park construction. Since then, the province has successively advanced pilot projects for 13 provincial-level national parks, including Laojunshan and Xishuangbanna National Parks. These initiatives were accompanied by the gradual development of a comprehensive institutional framework encompassing application, evaluation, monitoring, and management processes [33,34].

All counties involved in the national park pilot program in Yunnan Province and their corresponding approval years are reported in

Table 1. Counties Involved in the National Park Pilot Program in Yunnan Province and Their Approval Years.

Name of National Park	Administrative Region(s)	Year of Approval
Shangri-La Pudacuo	Shangri-La City	2007
Lijiang Laojunshan	Yulong County	2009
Xishuangbanna	Jinghong City, Menghai County, Mengla County	2009
Meili Snow Mountain	Deqin County	2009
Pu’er	Simao District	2009
Gaoligongshan	Longyang District, Tengchong County, Longling County	2011
Nangunhe	Cangyuan County, Gengma County	2011
Daweishan	Hekou County, Pingbian County, Gejiu City, Mengzi City	2013
Ailaoshan	Chuxiong City, Nanhua County, Shuangbai County	2016
Baima Snow	Deqin County, Weixi County	2016
Dashanbao	Zhaoyang District	2016
Nujiang Grand Canyon	Lushui City, Fugong County, Gongshan County	2016
Dulongjiang	Gongshan County	2016

[35]. The counties listed in Table 1 constitute the treatment group in this study. In the model specification, we use each county’s first approval year as the policy shock timing and construct the DID variable accordingly.

Building on the pilot practices in Yunnan Province, China officially proposed the establishment of a national park system in 2013 and launched the first batch of ten pilot parks in 2015, including Pudacuo National Park in Yunnan, marking the province’s entry into institutionalized reform. In 2021, China formally established five national parks—Three-River-Source, Giant Panda, Northeast China Tiger and Leopard, Hainan Tropical Rainforest, and Wuyishan [32]. In 2022, the government released the National Park Spatial Planning Scheme, designating 49 candidate areas covering more than 10% of the country’s land territory, with the aim of building the world’s largest national park system [36]. Four national parks in Yunnan were included in this spatial plan.

As an important practice in the development of China’s national park system, the Yunnan pilot has undertaken multidimensional explorations through institutional innovation, strengthened ecological protection, and community integration.

In terms of industrial restructuring, the pilot regions, in accordance with the Regulations on the Administration of Yunnan National Parks, strictly prohibited the construction or expansion of mining, chemical, and other projects that are highly polluting, energy-intensive, and ecologically damaging. Existing resource-extraction facilities underwent comprehensive remediation: tailings ponds were closed and restored within a prescribed timeframe, while related enterprises were guided to gradually exit or undergo industrial upgrading. To mitigate the short-term shocks caused by industrial adjustments, the government established policy-based safety nets by channeling stable financial support through ecological compensation and fiscal transfers, thereby easing the fiscal and employment pressures on local governments [37,38].

Regarding green industry development, eco-tourism and green services rapidly emerged, gradually replacing the traditional resource-dependent economic structure. National Park administrations actively encouraged community residents to engage in diversified industries such as tour guiding, catering, and cultural and creative businesses [39]. For example, since the introduction of eco-tourism in 2012, Pudacuo National Park has generated a cumulative tourism income of 1.819 billion yuan, while the average household income increased from 20,000 yuan to 100,000 yuan [37], achieving synergistic outcomes in ecological protection, livelihood improvement, and community employment.

In terms of infrastructure development, national parks have emphasized the coordinated advancement of nature education and tourism functions. Between 2016 and 2020, Pudacuo, supported by central government funding and nearly 600 million yuan of corporate financing, built new patrol trails, eco-boardwalks, management stations, scenic viewpoints, and environmental education centers [38]. These projects significantly improved visitor experiences and strengthened environmental education. With the continued release of brand effects, the ecological image and cultural appeal of the pilot regions have been further enhanced, providing practical support for analyzing the mechanisms through which national parks contribute to regional economic development.

2.2. The Impact of National Parks on Regional Economic Growth

To gain a deeper understanding of how national park development influences regional economic performance, this section conducts a mechanism-based analysis grounded in the “Triple-Value Synergy Mechanism,” which encompasses three dimensions: fiscal transfer payments, industrial upgrading, and fixed-asset investment. These dimensions correspond to government resource allocation, market structure adjustment, and increased factor input, respectively, forming an integrated transmission pathway through which national parks promote regional economic development.

2.2.1. Fiscal Transfer Payment Mechanism

According to the theory of fiscal decentralization, local governments tasked with stringent ecological conservation duties within national parks often face structural imbalances in fiscal revenues and expenditures. On the one hand, strict controls over resource utilization limit traditional revenue sources such as mining and forest product exploitation [40]. On the other hand, constraints on land development intensity reduce land lease revenues, further narrowing fiscal space [41]. Within the institutional framework of national parks, higher-level governments provide targeted support through ecological compensation funds, national parks construction subsidies, and transfer payments to key ecological function zones [42]. These mechanisms significantly enhance the volume and stability of fiscal transfers received by pilot counties. Such transfers not only reflect the externality of ecological protection but also offer financial security to underdeveloped ecological regions via additional capital injections.

These fiscal transfers foster regional economic development through multiple pathways. Directly, earmarked funds prioritize the construction of national parks infrastructure, including visitor centers, ecological monitoring systems, and environmental education facilities. These projects create employment opportunities [41] and stimulate demand in construction, logistics, and service sectors, yielding substantial investment multiplier effects. Indirectly, improved infrastructure enhances regional accessibility and service capacity, laying the foundation for the development of green industries such as ecotourism, research and education, and cultural innovation. This attracts more tourists and investors, extends the industrial value chain, and diversifies local income sources. Over the long term, sustained fiscal transfers reduce local government debt pressure and release additional public resources for social welfare and industrial support. Simultaneously, the ecological compensation mechanism monetizes ecological value, incentivizing local governments to transform ecological assets into economic advantages, thus achieving a harmonious balance between conservation and development.

2.2.2. Industrial Upgrading Mechanism

Industrial structure optimization and upgrading is an important driver of sustainable regional economic development. The implementation of strict nature conservation policies in China’s national parks [43] guides production factors toward high value-added and low-pollution industries through ecological access standards and spatial regulations, thereby promoting industrial upgrading. From the perspective of industrial exit, environmental constraints force traditional industries to transform, with high-risk sectors such as mining and chemicals facing stricter environmental assessments and approval limits, compelling firms to upgrade technologies or exit the market [41]. The released factors of production are then reallocated to emerging industries. From the perspective of industrial cultivation, relying on the rich biodiversity and unique landscapes of national parks, ecotourism has gradually become the core driver of regional industrial upgrading [8]. As the construction of national parks progresses, their ecological branding and regional image have been significantly enhanced [15], strengthening their attractiveness to external tourists and investors [44]. In this process, green industries centered on ecotourism have rapidly developed [25,45], giving rise to diversified industrial clusters such as rural homestays, specialty agricultural product processing, and e-commerce, and forming a virtuous cycle of coordinated development involving ecological conservation, tourism services, and the local economy.

Industrial upgrading contributes to regional growth in several key ways. First, in terms of factor allocation, green industries typically exhibit higher labor productivity and capital returns than traditional resource-based sectors. Sectors such as ecotourism and wellness services combine low energy consumption with high employment potential, improving allocation efficiency and overall output. Second, regarding income distribution, the development of high-value-added sectors generates substantial wage premiums. Jobs in tourism and wellness care offer significantly higher incomes than those in traditional farming. The extension of industrial chains also generates entrepreneurship opportunities, fostering the rise in farmer cooperatives, family farms, and homestay businesses [45], thereby broadening household income streams. Third, from the perspective of regional competitiveness, green industrial clusters generate economies of scale and scope, reduce operating costs, enhance competitiveness, and improve local resilience and long-term development potential.

2.2.3. Fixed-Asset Investment Mechanism

National parks pilot programs are typically accompanied by improvements and new developments in physical infrastructure [46,47], entailing large-scale public investment [34] and generating substantial investment-driven effects on county-level economies. In terms of investment composition, infrastructure construction dominates. To fulfill the dual mandates of conservation and recreation, pilot areas undertake extensive infrastructure development in non-core park zones, including roads, science education centers, visitor facilities, and environmental monitoring systems [37,39], creating robust demand in construction, building materials, and equipment rental sectors.

On the public service side, the establishment of healthcare, education, cultural, and environmental management infrastructure in and around park areas receives policy priority. Expanded coverage and improved service quality enhance residents’ quality of life. On the commercial side, national parks catalyze investment in entrance communities and thematic towns. Some counties also build on national parks branding to establish ecological agriculture demonstration zones and industrial clusters, generating scale effects in eco-industrial parks.

Sustained investment in fixed assets promotes regional economic growth through multiple transmission mechanisms [48,49]. From the demand side, large-scale infrastructure projects create short-term employment during construction, increase household income, and stimulate consumption, which in turn fuels retail, food service, and hospitality sectors, forming a virtuous cycle of investment-driven consumption. On the supply side, infrastructure improvements lower logistics costs, enhance factor mobility, and provide a foundation for subsequent industrial development. In particular, increased tourism reception capacity and service quality directly expand the scale and potential of the tourism economy. From the perspective of multiplier effects, fixed-asset investment not only contributes directly to GDP but also induces upstream and downstream industrial linkages. Park infrastructure development stimulates demand in building materials, landscaping, and decoration, while associated commercial projects promote the growth of financial, legal, and consulting services, creating a compound economic stimulus.

2.2.4. Synergistic Effects of Fiscal Support, Investment, and Industrial Upgrading

Fiscal transfers, fixed-asset investment, and industrial upgrading are not independent channels. Rather, they follow a chain of transmission that moves from government guidance to capital formation and then to structural transformation. In the early stage of national park development, intergovernmental fiscal support plays a key kick-starting role. The incremental funds are first translated into fixed-asset investment such as transport networks, ecological monitoring facilities, and public service platforms, thereby alleviating geographic constraints and infrastructure bottlenecks. As infrastructure improves and the business environment is upgraded, the multiplier effects of public investment begin to materialize, reducing transaction costs and enhancing the region’s service capacity. This, in turn, generates a crowding-in effect that attracts private capital. Such capital tends to concentrate in areas including concession-based operations, eco-accommodation, and cultural and creative experiences, providing the physical facilities and consumption scenarios needed for the development of green industries such as ecotourism. The expansion of these sectors facilitates the reallocation of labor and technology from resource-intensive activities toward modern services, thereby accelerating industrial upgrading.

Through these interactions, the three channels form a synergistic mechanism. Fiscal support provides basic guarantees and guiding funds for regional development; fixed-asset investment strengthens the material foundation by improving factor allocation; and industrial upgrading transforms the growth engine from resource dependence to greener and more efficient development. Together, they constitute a closed-loop pathway through which national parks promote regional development, helping county economies shift from reliance on external “transfusions” toward endogenous, self-sustaining growth, ultimately achieving a win–win outcome for ecological conservation and economic development.

Based on the above theoretical analysis, the following hypotheses are proposed:

H1: 

National park development has a significant positive effect on regional economic growth.

H2a: 

National park development promotes regional economic growth by increasing fiscal transfer payments from higher-level governments.

H2b: 

National park development facilitates regional economic growth by driving industrial upgrading.

H2c: 

National park development stimulates regional economic growth by boosting fixed asset investment.

2.3. The Economic Spillover Effects of National Parks

The construction of national parks, as a regional policy with spatial organizational characteristics, exerts impacts that extend beyond their immediate boundaries [15,25]. Drawing on growth pole theory, diffusion effect theory, and the principle of spatial interdependence in regional economics, major ecological projects and iconic spatial governance institutions often generate spillover effects on surrounding areas through mechanisms such as infrastructure development, industrial linkages, brand effects, and factor flows. Whether the establishment of national parks influences economic development on a broader spatial scale therefore warrants further investigation.

From the perspective of infrastructure spillovers, national park development is typically accompanied by large-scale investments in transportation, communications, and public services. These facilities transcend administrative boundaries, enhance regional accessibility, and facilitate the circulation of factors and population mobility, thereby creating development opportunities for neighboring counties [32,50]. From the perspective of industrial linkages, because core protected areas are subject to strict restrictions on development, certain functions such as tourism reception, product processing, and logistics tend to shift toward adjacent regions. This creates opportunities for local residents to engage in businesses such as homestay operations and tourism product sales, further promoting the coordinated development of services, agricultural product processing, and cultural tourism-related industries [51,52]. From the perspective of brand diffusion, national parks represent a national-level brand resource whose visibility and credibility embody strong positive externalities. Neighboring regions can leverage geographic proximity to share brand value, enhance their attractiveness at relatively low cost, and attract tourists, investment, and policy support, thereby benefiting from the “brand dividend” and its radiating effects.

Based on the above analysis, the following hypothesis is proposed:

H3: 

National park development fosters economic development in surrounding regions through spatial spillover mechanisms.

3. Methodology

3.1. Study Area

This study selects Yunnan Province as a case to evaluate the economic effects of national parks, primarily because it represents both the policy frontier and a complete sample in China’s national park institutional reform. Yunnan Province is one of the first regions in the country to systematically explore and implement the national park system. Since the establishment of China’s first provincial-level national park—Pudacuo National Park—in 2007, the province has continuously deepened its pilot efforts, promoting the establishment of 13 provincial-level national park pilots in batches between 2007 and 2016. This process clearly demonstrates an institutional evolution path: beginning with locally initiated exploration in 2007 and gradually transitioning to a nationally led, integrated system reform after 2016. Such a multi-level and multi-stage policy practice provides a favorable quasi-natural experimental setting for accurately identifying the dynamic effects of national park policies using a time-varying DID model.

Against this background, this study employs balanced panel data from 127 counties in Yunnan Province covering the period from 2001 to 2023 to systematically assess the impact of national park establishment on county-level economic development. In the sample structure, 24 counties involved in the 13 national parks are classified as the treatment group, while the remaining 103 counties without national parks serve as the control group.

3.2. Model Specification

3.2.1. Benchmark Regression Model

Taking the approval of the national park pilot program as a quasi-natural experiment, this study employs a time-varying difference-in-differences (Time-varying DID) model to evaluate whether the establishment of national parks promotes county-level economic development. The baseline specification is given by

$${pgdp}_{it}={\beta }_0+{\beta }_1{did}_{it}+{{\beta }_{2}X}_{it}+{\gamma }_t+{\lambda }_i+{\epsilon }_{it}$$

(1)

where the dependent variable ${pgdp}_{it}$ represents the per capita GDP of county $i$ in year $t$. The key explanatory variable ${did}_{it}$ indicates whether county $i$ has entered the implementation period of the national park pilot policy by year $t$. Specifically, let $T_i$ denote the first approval year of county $i$ (see Table 1). We define ${did}_{it}=1$ if ${t\ge T}_i$, and ${did}_{it}=0$ otherwise. Counties that are never approved remain ${did}_{it}=0$ throughout the sample period. ${\beta }_1$ is the coefficient of the key explanatory variable ${did}_{it}$, showing the effect of national park establishment on per capita GDP. $X_{it}$ represents the control variables, including the share of fiscal expenditure in GDP, the share of household deposits in GDP, population density, infrastructure, human capital level, and urbanization rate. ${\gamma }_t$ is the year fixed effect, controlling for macroeconomic and time effects faced by all counties in different years. ${\lambda }_i$ is the county fixed effect, controlling for the impact of time-invariant characteristics of each county on its economic development during the study period. ${\epsilon }_{it}$ is the random disturbance term.

3.2.2. Mechanism Testing Model

To examine the mechanism through which the establishment of national parks influences regional economic development, this study follows the modeling approach proposed by Baron and Kenny [53] and specifies the following models for mechanism testing:

$${channel}_{it}={\beta }_0+{\beta }_1{did}_{it}+{{\beta }_{2}X}_{it}+{\gamma }_t+{\lambda }_i+{\epsilon }_{it}$$

(2)

$${pgdp}_{it}={\eta }_0+{\eta }_1{did}_{it}+{{\eta }_{2}channel}_{it}+{{\eta }_{3}X}_{it}+{\gamma }_t+{\lambda }_i+{\epsilon }_{it}$$

(3)

In Equation (2), the effect of national parks establishment on the mechanism variable is tested. Compared to Equation (1), Equation (3) incorporates the mechanism variable as a control variable. ${channel}_{it}$ denotes the mechanism variable for county $i$ in year $t$, and the other notations carry the same meanings as in Equation (1).

3.2.3. Spatial Durbin Model

To further examine the spatial spillover effects of national park establishment on regional economic development, we build on the baseline regression and estimate a SDM. The model is specified as

$${pgdp}_{it}={\alpha }_0+\rho W{pgdp}_{it}{+\alpha }_1{did}_{it}+{\theta }_{1}W{did}_{it}+{{\alpha }_{2}X}_{it}+{\theta }_{2}W{X}_{it}+{\gamma }_t+{\lambda }_i+{\epsilon }_{it}$$

(4)

Model (4) builds on Model (1) by introducing spatial lags of both the dependent and independent variables. The parameter $\rho$ captures spatial dependence in the dependent variable; ${\theta }_1$ identifies the spatial spillover effects of the national park policy on neighboring areas; and ${\theta }_2$ measures the spatial externalities associated with the control variables. The spatial weight matrix $W$ is constructed based on the principle of global inverse geographic distance. The specific construction steps are as follows: First, we extract the geometric centroid coordinates of each county administrative division and calculate the spherical distance $d_{ij}$ between county $i$ and county $j$. Second, considering that the ecological and economic radiation effects of national parks may possess characteristics of long-distance transmission, we did not apply a distance truncation threshold. Specifically, for any two distinct counties ($i\ne j$) within the sample, we set $W_{ij}=1/d_{ij}.$ Finally, the constructed spatial weight matrix is row-standardized, satisfying the condition $\sum _j{W}_{ij}=1$.

3.3. Variables

(1)

Dependent Variable

This paper selects GDP per capita (pgdp) as the core dependent variable to measure the level of county-level economic development. This indicator reflects the quality and efficiency of regional economic growth and serves as a crucial benchmark for assessing the economic impact of national park construction on both local and surrounding areas. It is important to note that while per capita GDP is a core indicator of regional economic output, it emphasizes aggregate growth and cannot be fully equated with broad-based social welfare improvement or inclusive development. Therefore, the selection of this indicator in this study is primarily intended to quantify the policy’s growth effect on macroeconomic output, rather than to provide a comprehensive evaluation of overall well-being.

(2)

Core Explanatory Variable

The core explanatory variable is whether a national park is established. Treating the establishment of a national park as an exogenous policy shock, we construct a DID dummy variable, denoted as did. Let $T_i$ denote the first approval year of county $i$ (see Table 1). We define ${did}_{it}=1$ if ${t\ge T}_i$, and ${did}_{it}=0$ otherwise. Counties that are never approved remain ${did}_{it}=0$ throughout the sample period.

(3)

Control Variables

To mitigate the influence of potential confounding factors on the estimated effect of national parks on regional economic growth, and following related studies [54,55,56], this paper includes six county-level control variables in the regression models. Specifically, population density (density), measured by the ratio of permanent population to administrative area, is included to reflect population agglomeration and the associated resource-environmental pressures. Human capital (human) is proxied by the proportion of students in general primary and secondary schools to the total population, capturing the educational attainment of the labor force and its support for long-term growth. Urbanization level (urban), denoted by the share of the urban population, characterizes the transformation of the regional economic structure and modernization trends. Government intervention (gov), measured by the ratio of local fiscal expenditure to GDP, reflects the extent of government participation in economic activities and the supply of public services. Infrastructure (infra) is measured by road network density (total road length within the county divided by administrative area) [57], which serves as a representative indicator of transport infrastructure and reflects physical accessibility within and outside the county; higher road density implies lower transport costs and higher factor mobility. Finally, financial development level (deposit), expressed as the ratio of household deposit balances to GDP, is controlled to reflect the level of regional capital accumulation and financial deepening.

(4)

Mechanism Variables

Based on the theoretical discussion above, this paper investigates three channels through which national parks may affect regional economic outcomes: fiscal support, industrial upgrading, and fixed-asset investment.

For the fiscal channel, county-level data on upper-level fiscal transfer payments are missing after 2009. Jiang and Wang (2025) [58] use the “gap between general budgetary expenditure and revenue” as a proxy for upper-level transfer payments and show that the two are highly correlated (R² = 0.85). Following their approach, we construct two measures based on the general public budget balance (expenditure minus revenue):per capita fiscal net subsidy (per_transfer), measured as (general public budget expenditure−general public budget revenue) divided by county year-end population; and fiscal net subsidy intensity (transfer_int), measured as (general public budget expenditure−general public budget revenue) divided by GDP, which captures the strength of fiscal support relative to the size of the local economy.

For the industrial channel, we use two indicators to measure changes in industrial structure. The first is industrial structure upgrading (ts_ratio), measured as the ratio of tertiary-sector value added to secondary-sector value added. The second is an overall industrial transformation index (ind_index), constructed following the weighted structural index in Wang and Li (2024) [59]: share of primary-sector value added × 1 + share of secondary-sector value added × 2 + share of tertiary-sector value added × 3, which more comprehensively reflects the overall evolution of the industrial structure from lower- to higher-end activities.

For the investment channel, we also construct two measures: per capita fixed-asset investment (per_inv), measured as per capita total fixed-asset investment, reflecting the level of capital input; and fixed-asset investment intensity (inv_int), measured as fixed-asset investment divided by GDP, capturing the intensity of investment relative to economic scale.

Descriptions and assignment methods for the main variables are provided in

Table 2. Definitions of Key Variables and Descriptive Statistics.

Category	Variable Description	Meaning and Assignment	Obs	Mean	Std. Dev.	Min	Max
Dependent Variable	Per Capita Gross Domestic Product (pgdp)	Per capita GDP of each county (ten thousand RMB per person)	2921	2.437	2.492	0.102	18.534
Core Explanatory Variable	Establishment of National Park (did)	Whether the county is a national park pilot area (equals 1 in the approval year and thereafter; 0 otherwise).	2921	0.091	0.287	0	1
Mechanism Variables	Per capita fiscal net subsidy (per_transfer)	(General public budget expenditure − General public budget revenue)/County year-end total population (10,000 yuan/person)	2921	0.456	0.542	−0.149	9.665
	Fiscal net subsidy intensity (transfer_int)	(General public budget expenditure − General public budget revenue)/GDP	2921	0.240	0.188	−0.044	1.566
	Industrial structure upgrading (tert_sec_ratio)	Ratio of tertiary to secondary industry value added (no unit)	2921	1.231	1.081	0.055	10.333
	Overall industrial transformation (ind_upgrade)	Weighted sum: (Primary share × 1) + (Secondary share × 2) + (Tertiary share × 3)	2921	2.139	0.191	1.614	3.132
	Per capita fixed asset investment (per_inv)	Per capita total fixed asset investment (10,000 yuan/person)	2921	2.445	3.315	0.015	44.945
	Fixed asset investment intensity (inv_int)	Total fixed asset investment/GDP (no unit)	2921	0.868	0.767	0.016	24.607
Control Variables	Population Density (density)	Permanent population per square kilometer (10,000 persons/km2)	2921	0.020	0.036	0.001	0.341
	Human Capital Level (human)	Proportion of primary and secondary school students to total population (no unit)	2921	0.144	0.029	0.058	0.275
	Urbanization Level (urban)	Share of the urban population (no unit)	2921	0.215	0.194	0	0.986
	Government Intervention (gov)	Proportion of local government fiscal expenditure to GDP (no unit)	2921	0.299	0.190	0.022	1.619
	Infrastructure (infra)	Total length of county road network/Administrative area (km/km2)	2921	0.246	0.408	0.003	5.340
	Financial Development (deposit)	Proportion of household deposit balances to GDP (no unit)	2921	0.775	1.297	0.213	31.968

.

3.4. Data

This study uses county-level balanced panel data from 127 counties in Yunnan Province covering the period from 2001 to 2023 to examine the impact of national parks on county-level economic development. The ecological environment quality (EI) data is sourced from the National Earth System Science Data Center of China. Road network density is constructed using road vector data from the open-source geographic information platform OpenStreetMap (OSM); we process the data through map projection, rasterization, and length/area calculations in ArcGIS (v10.8). Other data are obtained from the China County Statistical Yearbook, China Regional Economic Statistical Yearbook, county-level statistical yearbooks, the statistical bureaus of various counties, and official government websites.

In December 2002, with approval from the State Council, the original Lijiang region was restructured into Lijiang City, and Lijiang County was split into the Gucheng District and Yulong Autonomous County. To ensure data comparability, the sample excludes the Gucheng District and Yulong Autonomous County.

All empirical estimations were performed in Stata SE (v18).

4. Empirical Results and Analysis

4.1. Baseline Regression Results

Table 3. Baseline Regression Results.

	(1)	(2)	(3)	(4)
did	1.9363 ***	0.4237 ***	1.3405 ***	0.5057 ***
	(0.1500)	(0.0872)	(0.1171)	(0.0784)
Controls	No	No	Yes	Yes
County FE	No	Yes	No	Yes
Year FE	No	Yes	No	Yes
N	2921	2921	2921	2921
adj. R2	0.0495	0.8644	0.5959	0.8947

Note: Standard errors in parentheses; *** p < 0.01.

reports the baseline regression results on the impact of national park establishment on county-level economic development. To assess the robustness of the core variable estimates, the analysis sequentially introduces control variables and fixed effects. The results show that across all model specifications, the coefficient of the core explanatory variable did remains significantly positive.

In the full model with all control variables and two-way fixed effects, the coefficient of did is 0.5057 and significant at the 1% level. This estimate suggests that, on average, the establishment of national parks increases county-level per capita GDP by about 5057 RMB, showing clear economic and statistical significance. The baseline results consistently indicate that national park establishment promotes local economic development and supports H1.

4.2. Pre-Treatment Test and Dynamic Effect Analysis

To verify the validity of the DID model, this paper employs the event study method to conduct a pre-treatment trend test and examine the dynamic effects of the policy. To avoid multicollinearity, the year prior to the policy implementation (Relative Year −1) is set as the benchmark period. The results are shown in Figure 1.

First, the analytical results are consistent with the parallel trend assumption. Before the policy implementation, the regression coefficients fluctuate around zero and are statistically insignificant, indicating that there were no systematic trend differences between the treatment and control groups beforehand.

After the policy implementation, the regression coefficients show an upward trend year by year. However, they are not statistically significant in the first three years, suggesting a lag in the release of economic dividends from national park construction. Starting from the fourth year, the coefficients become significantly positive and continue to grow, demonstrating a marginal increasing effect. This dynamic trend reflects the evolutionary pattern of the regional economy, transitioning from initial infrastructure consolidation to intermediate eco-tourism development, and ultimately to long-term industrial structure upgrading: specifically, the early phases of infrastructure construction and resource integration entail an objective gestation period; however, as ecological brand effects materialize and industrial chains extend, the policy’s driving effect on the regional economy exhibits a trend of continuous strengthening in the medium-to-long term.

4.3. Robustness Check

To further verify the validity of the causal identification regarding the impact of national park establishment on regional economic development, this paper employs a placebo test for robustness analysis. Specifically, while keeping the control variables and fixed-effects specifications unchanged, the timing of policy implementation and the treatment group assignments were randomly reshuffled. This virtual regression process was repeated 1000 times to generate a distribution of simulated intervention coefficients. The results are presented in Figure 2.

The results show that the kernel density curve of the simulated coefficients centers around zero, and the vast majority of p-value scatter points lie above the 0.1 significance level line, indicating that the randomly constructed dummy policies yielded no significant impact. To facilitate a direct visual comparison with the placebo distribution, Figure 2 marks the true policy estimate (0.5057) with a vertical dashed line on the right side of the x-axis. As shown, the true coefficient clearly deviates from the main range of the simulated distribution and lies in its extreme right tail. This confirms that the economic promotion effect of national parks is not derived from chance factors or model bias, demonstrating the high robustness of the baseline regression results.

To establish the reliability of the causal inference, this paper conducts a systematic robustness check from dimensions including model specification, variable measurement, sample outliers, and policy exclusivity. The results are presented in

Table 4. Robustness Test Results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	Alternative Models		Alternative Data			Excluding Policy Interference
	csdid	psm	GDP	Winsorization	Log Transformation	Overlapping Policies	Subsample Period
did		0.2683 ***	18.4062 ***	0.5310 ***	0.0637 ***	0.5692 ***	0.4832 ***
		(0.0713)	(4.8490)	(0.0699)	(0.0079)	(0.0795)	(0.0714)
ATT	0.4536 **
	(0.2250)
Key Ecological Function Zone Policy						−0.4907 ***
						(0.0561)
controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes
County FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	2921	2799	2921	2921	2921	2921	2540
adj. R2		0.9094	0.8834	0.9082	0.9973	0.8969	0.9054

Note: Standard errors in parentheses; ** p < 0.05, *** p < 0.01.

.

First, we employ alternative estimation models to mitigate potential specification bias. Given that traditional difference-in-differences (DID) models may yield biased estimates due to treatment effect heterogeneity in the context of staggered policy implementation, Column (1) reports the results using the CS-DID model proposed by Callaway and Sant’Anna [60]. Furthermore, we must acknowledge that the placement of national park pilots is not a completely random “natural experiment” but exhibits significant characteristics of “selection into treatment.” A review of the pilot distribution in Yunnan reveals that early pilots (e.g., Pudacuo and Laojunshan) were mostly concentrated in regions characterized by “superior ecological endowments but relatively lagging economic development.” This non-random selection logic, which simultaneously “selects for ecological superiority” and “targets economic weakness,” may lead to systematic ex-ante differences in resource endowments between pilot and non-pilot counties. If left uncontrolled, these initial differences could confound the identification of policy effects. To address this, Column (2) employs the Propensity Score Matching difference-in-differences (PSM-DID) method. We select control variables as covariates for nearest-neighbor matching to identify non-pilot counties with the most similar characteristics to the pilot counties as a control group, thereby effectively mitigating selection bias driven by location preferences. The results indicate that, even under these stricter identification assumptions, the policy effect remains significantly positive, and the coefficient estimates remain robust.

Second, the dependent variable is replaced, and the influence of extreme values is excluded. To test the sensitivity of the conclusions to the form of indicator construction, Column (3) replaces the dependent variable with total GDP. Columns (4) and (5) apply a two-sided 2.5% winsorization and logarithmic transformation to per capita GDP, respectively, to eliminate interference from outliers and mitigate heteroscedasticity. All regression results are statistically significant, demonstrating that the core conclusions do not rely on specific indicator forms or individual extreme samples.

Finally, the interference of competing policies and specific period shocks is excluded. On the one hand, to purify the policy evaluation environment, considering that the spatial overlap between the “Key Ecological Function Zones” policy and national park construction may induce confounding effects, Column (6) introduces a dummy variable for this policy as a control. On the other hand, to exclude potential interference from the formal establishment of national parks in 2021 and subsequent macro-policy adjustments, Column (7) excludes samples from 2021 and beyond, retaining only data from the pilot period for regression. The results show that the DID coefficient remains significant at the 1% level, effectively ruling out alternative explanations driven by competing policies or specific time windows.

In summary, through rigorous testing across multiple dimensions, the research conclusions of this paper demonstrate a high degree of consistency and robustness.

4.4. Heterogeneity Analysis

To further examine the heterogeneity of the effects of the national park pilot policy, this study conducts subgroup regressions along four dimensions: economic development level, ecological environmental quality, geographic location, and policy stage. Economic development level and ecological environmental quality are divided into high and low groups based on the median values of the average per capita GDP and the average Ecological Environment Quality Index (EI) over 2001–2006, respectively. Geographic location is proxied by county-level average elevation, and the sample is split into high- and low-elevation groups using the median elevation. Following Ren Zhizhong (2018) [61], policy stages are defined according to the first approval year of the pilot and are classified into the exploratory stage (1996–2007), the construction and promotion stage (2008–2014), and the institutional deepening stage (2015 and thereafter). In all subgroup regressions, counties that have never implemented a national park pilot serve as the control group. The results are reported in

Table 5. Heterogeneity Test Results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
	Economic Development		Ecological Environment		Geographic Location		Policy-Stage Heterogeneity
	Low-Income	High-Income	Eco-Poor	Eco-Rich	Low-Altitude	High-Altitude	Exploratory	Construction & Promotion	Institutional Deepening
did	0.5452 ***	0.1830 *	0.3044 ***	0.6458 ***	0.2208 **	0.5674 ***	2.0236 ***	0.5855 ***	0.3135 ***
	(0.0784)	(0.1089)	(0.0859)	(0.1529)	(0.1118)	(0.1047)	(0.3479)	(0.1012)	(0.1123)
Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
County FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	1449	1472	1449	1472	1472	1449	2415	2714	2576
adj. R2	0.9368	0.9088	0.9067	0.9026	0.9002	0.9020	0.8970	0.8940	0.8977

Note: Standard errors in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01.

.

First, regarding heterogeneity by economic development level, Columns (1) and (2) show that the national park pilot policy significantly promotes per capita GDP in both low- and high-income groups, but the estimated coefficient is substantially larger for the low-income group. This indicates a stronger marginal effect of the policy in economically less developed areas. A plausible explanation is that such areas tend to have more pronounced deficiencies in infrastructure and public services, so policy-induced investment improvements are more likely to translate into tangible gains. Moreover, fiscal support from higher-level governments and related project investments may generate stronger stimulus effects in these areas. By contrast, in high-income areas with a larger economic base, the incremental contribution brought by the policy is relatively limited, resulting in a smaller estimated effect.

Second, in terms of ecological environmental quality, Columns (3) and (4) compare policy effects under different ecological endowments. The results indicate significantly positive effects in both groups, while the estimated coefficient is larger in areas with better ecological quality. This suggests that the economic conversion efficiency of the national park policy depends closely on local natural resource endowments. Regions with higher-quality ecological conditions are more likely to leverage the national park brand and landscape resources to enhance attractiveness, promote ecotourism, nature education, and related activities, and therefore more fully translate ecological advantages into economic advantages.

Third, with respect to geographic location, Columns (5) and (6) show that the coefficient is significant and larger in high-elevation areas, whereas the effect in low-elevation areas is positive but smaller in magnitude. This difference may stem from the fact that high-elevation areas often possess scarcer mountain landscapes and richer biodiversity, which better align with the conservation and recreation functions of national parks. In addition, improvements in transportation and public service facilities during the pilot implementation are more likely to alleviate locational constraints in high-elevation areas, thereby facilitating the release of stronger latecomer growth potential.

Finally, regarding heterogeneity across policy stages, Columns (7) through (9) show that the policy effect is strongest in the exploratory stage, followed by the construction and promotion stage, while the institutional deepening stage remains significantly positive but exhibits a smaller effect. Overall, the results display a “faster early impact and slower later impact” pattern, consistent with the logic of policy evolution. During the exploratory and demonstration stage, locally driven pilots tended to place greater emphasis on branding and recreation-oriented development, with relatively weaker institutional constraints, and were often concentrated in areas with strong ecological endowments and high tourism potential, making it easier to generate a rapid “start-up effect.” In the construction and promotion stage, as pilots expanded and gradually became more standardized, infrastructure, tourism carrying capacity, and management systems continued to improve, and policy impacts were more likely to accumulate gradually and stabilize. In the institutional deepening stage, stronger institutional constraints—such as ecological red lines, land-use controls, and restrictions on resource development—may impose tighter limits on some traditional industries and offset part of the short-term gains. Meanwhile, fiscal support mechanisms are strengthened and governance standards become more standardized, but the cultivation of green industries, ecological value realization, and structural upgrading typically require more time to fully materialize. As a result, the overall effect remains positive but is more likely to manifest as slower yet more sustainable gains.

In sum, the heterogeneity analysis indicates that the economic promotion effect of the national park pilot policy varies substantially across contexts. The policy generates stronger impacts in less developed areas, high-elevation areas, and regions with better ecological endowments, where ecological resource advantages can be more readily transformed into economic growth drivers. Moreover, as the policy evolves from exploration toward institutionalization, its effects shift from rapid early gains to a more stable long-term influence, reflecting stage-specific characteristics consistent with institutional development.

4.5. Mechanism Testing

To examine the transmission channels through which the national park pilot policy affects county economic growth, this paper conducts channel tests along three dimensions—fiscal support, industrial structure adjustment, and fixed-asset investment—based on the baseline regressions. The results are reported in

Table 6. Fiscal Support Channel Estimation Results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	per_transfer	Per Capita GDP			transfer_int	Per Capita GDP
		Full Sample	Low-Income	High-Income		Full Sample	Low-Income	High-Income
did	0.3321 ***	0.4476 ***	0.2005 **	0.1094	0.0072 ***	0.5013 ***	0.6509 ***	0.1338
	(0.0433)	(0.0916)	(0.0862)	(0.1357)	(0.0015)	(0.0778)	(0.0931)	(0.1419)
per_transfer		0.2508 ***	0.7523 ***	0.3064
		(0.0939)	(0.0621)	(0.2748)
transfer_int						2.6481 ***	2.9363 ***	1.1527
						(1.0249)	(0.6022)	(1.7624)
controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
County FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	2921	2921	1449	1472	2921	2921	1449	1472
adj. R2	0.7220	0.8079	0.9157	0.8182	0.9893	0.8949	0.8953	0.8173

Note: Standard errors in parentheses; ** p < 0.05, *** p < 0.01.

,

Table 7. Fixed-Asset Investment Channel Estimation Results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	ts_ratio	Per Capita GDP			ind_index	Per Capita GDP
		Full Sample	Low-Income	High-Income		Full Sample	Low-Income	High-Income
did	0.2294 ***	0.6161 ***	0.5151 ***	0.3298 ***	0.0469 ***	0.4713 ***	0.7634 ***	0.0608
	(0.0583)	(0.0774)	(0.0760)	(0.1068)	(0.0113)	(0.0948)	(0.1003)	(0.1284)
ts_ratio		0.4273 ***	0.1327 **	0.4579 ***
		(0.0971)	(0.0661)	(0.1463)
ind_index						1.8047 ***	1.0233 ***	1.5601 ***
						(0.1869)	(0.1642)	(0.2913)
controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
County FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	2921	2921	1449	1472	2921	2921	1449	1472
adj. R2	0.4955	0.8965	0.9374	0.9094	0.5907	0.7754	0.8830	0.7812

Note: Standard errors in parentheses; ** p < 0.05, *** p < 0.01.

and

Table 8. Fixed-Asset Investment Channel Test Results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	per_inv	Per Capita GDP			inv_int	Per Capita GDP
		Full Sample	Low-Income	High-Income		Full Sample	Low-Income	High-Income
did	1.1979 ***	0.3492 ***	0.4363 ***	0.1070	0.0436 ***	0.5083 ***	0.5403 ***	0.1825 *
	(0.1877)	(0.0735)	(0.0749)	(0.0972)	(0.0141)	(0.0782)	(0.0792)	(0.1089)
per_inv		0.1306 ***	0.1317 ***	0.0749 ***
		(0.0220)	(0.0450)	(0.0117)
inv_int						0.0584 ***	0.0731 ***	−0.0578
						(0.0205)	(0.0266)	(0.0448)
controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
County FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	2921	2921	1449	1472	2921	2921	1449	1472
adj. R2	0.6499	0.9052	0.9473	0.9152	0.3631	0.8949	0.9374	0.9089

Note: Standard errors in parentheses; * p < 0.1, *** p < 0.01.

. It should be noted that our mechanism analysis follows the classic Baron & Kenny [53] framework; therefore, the results are intended to provide empirical evidence consistent with these channels as transmission mechanisms, rather than to deliver definitive causal identification of mediation effects.

Table 6 presents the results for the fiscal-support channel. Column (1) shows that the national park pilot significantly increases counties’per capita fiscal net subsidy, indicating an overall rise in fiscal support after policy implementation. After including per capita fiscal net subsidy in Column (2), the mediator is significantly and positively associated with per capita GDP, while the policy coefficient remains significant. This suggests a stable positive relationship between increased fiscal support and economic growth, providing evidence consistent with fiscal support as a transmission channel. The heterogeneity results indicate that this channel is more pronounced in low-income counties: in the low-income group, per capita fiscal net subsidy has a stronger and significant positive association with per capita GDP, whereas the relationship is not significant in the high-income group (Columns (3)–(4)). This implies that fiscal support is more likely to generate larger marginal gains in economically weaker counties by easing fiscal constraints, strengthening the provision of basic public services, and alleviating infrastructure bottlenecks. In more economically advanced counties, where fiscal systems and public service provision are relatively mature, additional fiscal inflows may be diluted by the larger economic base, and the marginal contribution to per capita output may be more limited; policy effects may therefore operate more through market-based allocation and industrial adjustment. As a robustness check, Columns (5)–(8) replace the per capita measure with fiscal net subsidy intensity; the direction and statistical significance are broadly consistent. The policy significantly increases fiscal net subsidy intensity, and this intensity measure is positively associated with per capita GDP, with stronger effects in the low-income group, indicating that the fiscal-channel evidence is not driven by a single measure.

Table 7 reports the results for the industrial upgrading channel. Column (1) shows that the policy significantly increases industrial structure upgrading, suggesting that national park pilots promote a shift toward a more service-oriented and higher value-added industrial structure. After including industrial structure upgrading in Column (2), the mediator enters with a significantly positive coefficient, and the policy coefficient remains significant. This provides evidence consistent with industrial upgrading as a transmission channel. Heterogeneity results further show positive relationships in both income groups, but the effect is stronger in the high-income group (Columns (3)–(4)). This may reflect that economically better-developed counties have more developed counties tend to have more complete market institutions, factor endowments, and service-sector capacity, enabling them to absorb brand spillovers and tourism demand more quickly and translate upgrading into employment expansion, firm agglomeration, and productivity gains. By contrast, although low-income counties also benefit, limitations in market size and complementary conditions may reduce conversion efficiency. Using the overall industrial transformation index as an alternative measure (Columns (5)–(8)) yields consistent results: the policy significantly promotes structural transformation, and the transformation index is positively associated with per capita GDP. The subgroup patterns are in line with the baseline measure, reinforcing the robustness of the industrial-channel evidence.

Table 8 presents the results for the fixed-asset investment channel. Column (1) shows that the national park pilot significantly increases per capita fixed-asset investment, indicating that policy implementation is accompanied by expanded investment in infrastructure and public services. After including per capita fixed-asset investment in Column (2), the mediator is significantly positive and the policy coefficient remains significant, providing evidence consistent with investment-driven transmission. Heterogeneity results show that this channel is more concentrated in low-income counties: the policy effect on per capita GDP is stronger in the low-income group, while it is not significant in the high-income group (Columns (3)–(4)); per capita fixed-asset investment is positive and significant in both groups, but the coefficient is larger in the low-income group. This suggests that in less developed counties, national-park-related investment is more likely to address binding constraints—such as accessibility, tourism-service facilities, ecological monitoring, and public service gaps—thus generating “from-zero-to-one” improvements and supporting subsequent industrial development by lowering transaction costs and raising carrying capacity. In more developed counties, with higher existing infrastructure and investment levels, the marginal contribution of additional investment to per capita output may be smaller. As a robustness check, Columns (5)–(8) use fixed-asset investment intensity instead of per capita investment; the results largely maintain the same pattern. Investment intensity is significantly associated with per capita GDP in the full sample and in the low-income group, but it is not significant and even turns negative in the high-income group. This may indicate that in high-income counties, investment expansion is more prone to structural misallocation or crowding-out effects, leading to lower marginal returns to per capita output. Overall, the investment channel shows consistent directional evidence across alternative measures and exhibits a stronger transmission pattern in low-income counties.

In summary, the promoting effect of the national park policy on the regional economy is not the result of a single pathway but the outcome of the synergistic action of fiscal support, industrial upgrading, and capital investment. Hypotheses H2a, H2b, and H2c are all strongly supported by empirical evidence.

5. Spatial Spillover Effect Analysis

5.1. Spatial Autocorrelation Test

Before constructing a spatial econometric model, testing for spatial autocorrelation among variables is a crucial prerequisite for assessing the rationality of the model specification. To this end, based on a spatial geographic weight matrix, this paper calculated the Global Moran’s I for per capita GDP across 127 counties in Yunnan Province from 2001 to 2023. The results are presented in

Table 9. Global Moran’s I and Significance Test Results.

Year	Moran’s I	p-Value	Year	Moran’s I	p-Value
Global	0.585	0.000	2012	0.095	0.000
2001	0.080	0.000	2013	0.100	0.000
2002	0.092	0.000	2014	0.100	0.000
2003	0.091	0.000	2015	0.105	0.000
2004	0.080	0.000	2016	0.110	0.000
2005	0.085	0.000	2017	0.115	0.000
2006	0.091	0.000	2018	0.130	0.000
2007	0.088	0.000	2019	0.127	0.000
2008	0.095	0.000	2020	0.117	0.000
2009	0.093	0.000	2021	0.111	0.000
2010	0.094	0.000	2022	0.105	0.000
2011	0.089	0.000	2023	0.104	0.000

.

The overall distribution shows that the Moran’s I values for both the full sample and individual years are positive and statistically significant at the 1% level. The Global Moran’s I for the full sample is 0.585, indicating a significant positive spatial autocorrelation in county-level economic development. A comparison reveals that the Global index calculated from pooled panel data (0.585) is notably higher than the cross-sectional indices for individual years (0.080–0.130). This statistical characteristic reflects the dual dependence of regional economic development in both spatial and temporal dimensions: the pooled sample captures not only lateral spatial clustering but also longitudinal temporal serial correlation (i.e., path dependence), the superposition of which makes the overall correlation more significant.

In terms of temporal trends, the degree of spatial clustering generally exhibits a fluctuating upward trend during the sample period. Between 2001 and 2011, the Moran’s I fluctuated stably within the range of 0.080 to 0.095. Since 2012, spatial correlation has gradually strengthened, reaching a peak of 0.130 in 2018. Although it declined slightly thereafter, it remained above 0.100. This indicates that with the development of regional economic integration, the economic interdependence between counties has continuously deepened. To minimize model misspecification and estimation bias, a spatial econometric model is introduced to accurately identify policy effects and spatial spillover impacts.

To further characterize the micro-morphology and temporal evolution of spatial agglomeration in Yunnan’s county-level economy, we plotted Moran scatterplots for six key time nodes within the sample period (see Figure 3). The selection of these time nodes is closely aligned with the evolutionary trajectory of the national park policy: 2001 serves as the baseline prior to policy implementation; 2007 marks the establishment of Pudacuo National Park and the initiation of the “local pilot” phase; 2012 corresponds to the substantive expansion of pilot programs across the province; 2015 marks the elevation of the reform to a national strategy with the issuance of the Pilot Plan for Establishing the National Park System; 2018 represents the peak of spatial agglomeration effects; and 2023 reflects the most recent spatial pattern.

As shown in Figure 3, the vast majority of county samples are concentrated in the first quadrant (High–High agglomeration) and the third quadrant (Low–Low agglomeration), exhibiting a distinct “bi-polar” polarization. This distribution reveals a significant “club convergence” phenomenon in county-level economic development, where economically developed counties tend to “cluster for development” through positive spillover effects, while underdeveloped counties exhibit characteristics of contiguous poverty. From a dynamic perspective, the slope of the regression fit line (corresponding to the magnitude of Moran’s I) has undergone a process of “stable start—accelerated rise—high-level stabilization.” Between 2001 and 2007, the slope growth was relatively moderate, suggesting loose inter-county economic linkages in the early policy stage. However, after 2012, with the comprehensive rollout of the national park system pilots, the slope steepened significantly, reaching a peak in 2018. This profoundly reflects that national park construction, as a tool for spatial governance, has effectively broken down administrative barriers and accelerated the cross-regional flow and agglomeration of factors by improving transportation infrastructure and building ecological industrial chains, thereby reinforcing the spatial connectivity and coordinated development of the regional economy at the macro level.

5.2. Selection and Testing of Spatial Econometric Models

To accurately identify the spatial spillover effects of regional economic growth, this paper conducts systematic statistical tests on the spatial dependence of the model setting, model form, and fixed effects. The results are shown in

Table 10. Spatial Econometric Model Testing and Selection.

Test Type	Test Method	Statistic	p-Value
Spatial Dependence Diagnosis	LM_err	11408.65	0.000
	R_LM_err	9157.62	0.000
	LM_lag	2927.97	0.000
	R_LM_lag	676.94	0.000
Comparing SDM and SAR	LR Test	129.04	0.000
	Wald Test	123.53	0.000
Comparing SDM and SEM	LR Test	189.60	0.000
	Wald Test	192.32	0.000
Fixed-Effects Selection	Hausman Test	60.45	0.000
	LR Test (Individual vs. Double Fixed)	60.37	0.000
	LR Test (Time vs. Double Fixed)	2594.28	0.000

.

First, spatial dependence diagnosis. Based on the LM test and its robust form (Robust-LM) for the spatial lag term and spatial error term, the results show that the statistics for LM_err, LM_lag, and their robust versions are all significant at the 1% level. This indicates that the sample data simultaneously exhibit significant spatial error autocorrelation and spatial lag correlation, preliminarily verifying the necessity of introducing a spatial econometric model.

Second, model form identification. To determine the optimal spatial model form, this paper uses the LR test and Wald test to examine whether the Spatial Durbin Model (SDM) degenerates into a Spatial Autoregressive Model (SAR) or a Spatial Error Model (SEM). The test results show that the LR and Wald statistics against the SAR model are 129.04 and 123.53, respectively, while those against the SEM are 189.60 and 192.32. All tests reject the null hypothesis that the SDM degenerates into SAR or SEM at the 1% significance level. This demonstrates that the SDM model, as a more inclusive general form, can more comprehensively capture the spatial interaction effects between variables, offering a superior fit compared to simplified models.

Third, fixed-effects testing. The Hausman test result is 60.45 (p < 0.0), rejecting the random-effects hypothesis and supporting the selection of a fixed-effects model. Furthermore, the LR test is used to discriminate the form of fixed effects. The results show that the LR statistics for individual fixed effects and time fixed effects are 60.37 and 2594.28, respectively, both significantly rejecting the null hypothesis of single fixed effects. This indicates that the model should simultaneously control for individual heterogeneity and time shocks.

In summary, this paper ultimately selects the two-way fixed-effects SDM as the baseline regression model. This specification not only effectively addresses omitted variable bias but also simultaneously captures the dual spatial spillover effects of both explanatory and dependent variables.

5.3. Testing for Spatial Spillover Effects

Table 11. SDM Regression Results.

	(1)	(2)
Variables	Inverse-Distance Weight Matrix	Adjacency Weight Matrix
Main	0.415 ***	0.387 ***
	(0.088)	(0.084)
Wx	0.279 ***	0.206 **
	(0.046)	(0.083)
LR_Direct	0.459 ***	0.405 ***
	(0.090)	(0.083)
LR_Indirect	0.305 ***	0.202 ***
	(0.120)	(0.011)
LR_Total	0.764 ***	0.607 ***
	(0.037)	(0.216)
rho	0.851 ***	0.466 ***
	(0.030)	(0.022)
sigma2_e	0.527 ***	0.485 ***
	(0.014)	(0.013)
controls	Yes	Yes
County FE	Yes	Yes
Year FE	Yes	Yes
Log-likelihood	−3229.785	−3160.864
Observations	2921	2921
R-squared	0.534	0.262
Number of id	127	127

Note: Standard errors in parentheses; ** p < 0.05, *** p < 0.01.

reports the estimation results of the Spatial Durbin Model (SDM) based on different spatial weight matrices. Column (1) and Column (2) employ the inverse-distance weight matrix and the adjacency weight matrix, respectively, aiming to verify the reliability of the empirical results by comparing different spatial dimensions.

First, regarding the spatial correlation characteristics and preliminary regression coefficients, the spatial autocorrelation coefficients $\mathrm{\rho }$ under both matrices are significantly positive at the 1% level (0.851 and 0.466, respectively), indicating strong spatial dependence and regional stickiness in the county-level economic development of Yunnan Province. In terms of coefficient estimation, the coefficients of the core explanatory variable (Main) and the spatial lag term (Wx) are both positive and significant. This preliminarily suggests that the construction of national parks has a positive impact on the local economy and that spatial spillover effects are emerging. However, since the SDM includes global spillover effects, point estimate coefficients cannot directly reflect true marginal effects. Therefore, the following analysis focuses on the direct, indirect, and total effects decomposed using the partial differentiation method.

Second, regarding the direct impact of the policy on the local area, under the inverse-distance matrix, the direct effect coefficient is 0.459 ($p<0.01$). This value is slightly larger than the main regression coefficient (Main = 0.415), with the difference capturing the “spatial feedback effect”—that is, the positive externalities spilling over to neighboring areas after the national park drives local growth feed back into the local economy through regional economic circulation. The result under the adjacency matrix (0.405) is consistent with this finding. This robust conclusion indicates that, regardless of the definition of spatial adjacency, national parks significantly enhance the economic development level of the pilot counties themselves by optimizing resource allocation and establishing ecological brand advantages.

Third, further examining spatial spillover effects, the indirect effect under the inverse-distance matrix is 0.305 ($p<0.01$), and under the adjacency matrix, it is 0.202 ($p<0.01$). Both matrices confirm that the construction of national parks possesses significant positive spatial externalities. This means that after a county establishes a national park, it effectively drives the increase in per capita GDP of neighboring areas through the interconnection of transportation infrastructure, the diffusion of tourist flows, and the extension of industrial chains. It is worth noting that the indirect effects under both matrices are smaller than the direct effects, exhibiting reasonable “distance decay” characteristics. This indicates that while the policy dividends have significant diffusivity, the core benefits primarily remain within the pilot counties. There is no “hollowing out” caused by excessive diffusion; rather, this pattern of “internal strengthening with external spillovers” aligns with the objective laws of regional economic development.

Fourth, comprehensively examining the overall impact of the policy, the total effect coefficients under the inverse-distance and adjacency matrices are 0.764 and 0.607, respectively, both significant at the 1% level. This result reflects the net contribution of national park establishment to the entire regional sector. Empirical data demonstrate that the policy has not induced a “siphon effect” at the expense of surrounding interests. Instead, it presents a benign development trend of “driving regional development through pilot hubs” (point-to-area development) and achieving regional co-prosperity, effectively validating the empirical value of national park construction as a new engine for promoting coordinated regional economic development.

5.4. Spatial Spillover Boundaries and Distance Decay Characteristics

To rigorously verify the geographical boundaries and attenuation laws of the spatial spillover effects generated by national parks, this study plotted the spatial attenuation trend curve as shown in Figure 4. In terms of the logic for parameter setting, the distance threshold was set to range from 100 km to 700 km, with a fine-grained step size of 50 km. The selection of this range is primarily based on the geographical scale characteristics of Yunnan Province, which ensures the coverage of the vast majority of inter-county economic interactions within the province while effectively filtering out the heterogeneity noise caused by ultra-long-distance spatial units. Furthermore, the 50 km step size helps to capture the marginal changes in spillover intensity within the “Golden Tourism Radius.”

The visualization results indicate that the driving effect of national parks on the regional economy does not follow a simple linear decay law but instead exhibits a significant inverted U-shaped non-linear characteristic. Specifically, within the distance interval of 100 km to 250 km, the spatial spillover effect does not weaken with increasing distance; on the contrary, it shows a significant strengthening trend, reaching its peak intensity at 250 km. This phenomenon suggests that the optimal radiation radius of a national park is not limited to the immediate adjacent areas but reaches its optimum within a transportation circle of approximately 2 to 3 h. This distance interval is conducive to the formation of regional tourism loops and multi-destination travel routes, facilitating the sharing of tourist flows and industrial complementarity, thereby maximizing positive spillovers.

However, after crossing the inflection point of 250 km, the mechanism of distance decay begins to dominate. The empirical curve shows that the regression coefficient exhibits a sharp downward trend as the geographical distance extends further. Specifically, when the distance extends to 350 km, the spillover effect is no longer statistically significant, and it turns negative at 400 km. This finding clearly defines the effective geographical boundary of the positive externalities of national parks at approximately 350 km. Beyond this range, constrained by rising transportation costs and the weakening of industrial linkages, policy dividends are difficult to transmit effectively. In summary, the core economic driving force of national park construction is primarily concentrated within a radius of 350 km. This finding provides an important quantitative basis for constructing appropriately scaled ecological economic circles.

6. Discussion

Taking the national park institutional reform in Yunnan Province, China, as an example, this paper empirically examines the economic effects of national park construction based on long-term panel data from 127 counties between 2001 and 2023.

The study finds that the national park policy has not only significantly increased the per capita GDP of pilot counties but also generated positive spatial spillover effects. This empirical result primarily provides compelling macroeconomic evidence for re-examining the relationship between “conservation and development.” There has been a long-standing controversy in academia regarding whether establishing protected areas inhibits economic growth. Some studies supporting the “restriction hypothesis” worry that strict ecological regulations will crowd out production factors by limiting resource exploitation and increasing compliance costs, thereby suppressing local economic growth [6,62]. However, our empirical results do not align with this concern; instead, they provide empirical evidence from a macro-regional level supporting the view that “conservation and development are not mutually exclusive” [4,7]. Our research indicates that under reasonable institutional arrangements, national parks do not become obstacles to development but may realize the conversion of ecological value into economic value by forcing a green transformation [11]. This suggests that the practice in Yunnan has, to a certain extent, avoided the “green poverty” trap feared by some scholars and supports the argument that ecological conservation can be compatible with, or even promote, regional economic development.

Regarding the specific driving mechanisms of “how to promote growth,” the findings of this paper not only corroborate some international experiences but also reveal the uniqueness of the Chinese context. Consistent with most studies focusing on developed countries such as the United States and Austria [15,17], we confirm that industrial structure upgrading driven by ecotourism is an important engine for economic growth, which reaffirms the significant economic multiplier effect of the tourism industry [13]. However, unlike the Western model, which relies mainly on market-oriented tourist consumption [16], this paper further discovers that “fiscal transfer payments” play a more critical role in the construction of China’s national parks. This finding echoes conclusions regarding the Giant Panda National Park [26], namely that vertical compensation from the central government effectively alleviates local protection costs. This implies that compared to a purely market-driven model, China’s “two-wheel drive” model of “government-led fiscal incentives + market-oriented industrial upgrading” may be better suited for developing regions to cope with the dual challenges of high population density and lagging development.

Furthermore, this study expands the horizon from the micro “island effect” to the macro “regional radiation.” Previous studies on the economic impacts of national parks have mostly focused on micro-effects within the park or its immediate communities [15,23,29], and some have pointed out the spatial inequality of benefit distribution [21,24]. In contrast, this study attempts to shift the perspective from “points” to “areas.” Our spatial econometric results show that the economic dividends of national parks are not confined “within the walls” but have produced significant positive spillovers to surrounding areas. This result quantitatively responds to the view of Cao et al. (2025) [32] regarding the externalities of ecological service flows from the perspective of economic geography. It implies that national parks are not isolated ecological islands but can become regional growth poles driving coordinated development through transportation improvements and industrial linkages.

Naturally, this study has certain limitations. First, regarding conceptual definitions and indicator selection, this paper uses the growth of per capita GDP as the primary proxy for “economic development,” which may be too absolute. GDP mainly measures the total growth of economic output and is not equivalent to broad-based social welfare improvement or inclusive development; it cannot fully reflect micro-level changes in income distribution, subjective well-being, and livelihood diversity. Although the data show that national parks significantly improved output levels, this does not directly mean that all groups benefit equally. Second, regarding the causal identification strategy, although we employed the PSM-DID method to mitigate differences in observable characteristics, the placement of national parks is not a completely random experiment. They tend to be located in regions with rich ecological resources but relatively lagging economic development (i.e., a “selection into treatment” characteristic). This selection bias, caused by unobservable factors such as local government intent or historical and cultural accumulation, may still interfere with the estimation results. Third, regarding the spatial econometric specification, although we used both inverse-distance and adjacency weight matrices for robustness checks, the measurement of spatial spillover effects is often sensitive to matrix settings. Real-world regional economic connections are influenced by multiple factors such as transportation networks and economic embeddedness; a weight matrix based solely on geographic information may not fully capture complex economic spillover networks. Finally, regarding external environmental impacts, the sample period of this study spans a long duration, and external shocks such as macroeconomic fluctuations may potentially affect the model estimates. This study mainly focuses on the average net effect of the policy itself and has not specifically isolated the non-linear interferences potentially caused by external macro-environmental changes. This points out the direction for further in-depth research.

7. Conclusions and Policy Implications

Based on panel data from 127 counties in Yunnan Province from 2001 to 2023, this paper constructs a time-varying difference-in-differences (DID) model and a Spatial Durbin Model (SDM) to systematically evaluate the impact of national park establishment on regional economic development. It explores the mechanisms from three dimensions—fiscal transfer payments, industrial structure upgrading, and fixed asset investment—while identifying potential spatial spillover effects. The main conclusions are as follows:

First, the establishment of national parks has a significant positive impact on the per capita GDP of pilot areas. This conclusion remains robust across different model specifications and multiple robustness checks, verifying the economic promotion effect of the national park policy. Dynamic effect analysis further shows that this impact possesses distinct lagged and persistent characteristics; policy effects generally emerge gradually 4 to 5 years after implementation and continue to strengthen in subsequent stages.

Second, heterogeneity analysis indicates that the economic driving effect of national parks exhibits significant contextual differences and temporal characteristics. In the cross-section, policy effects are stronger in regions with low economic development levels, high altitudes, and superior ecological environment quality. This suggests that the national park system has significant “pro-poor” characteristics, and regions with better ecological endowments and unique geographical environments are more likely to realize the economic conversion of ecological values. In the longitudinal time series, policy effects present an evolutionary pattern of “early explosion, later stability.” Specifically, the economic driving role is strongest during the “research and exploration stage” due to the rapid initiation of branding and recreational development. In the “institutional deepening stage,” although restricted by stricter ecological red lines and industrial transformation cycles, the marginal effect converges but remains significantly positive, reflecting a transition from “project-driven” to “structure-reshaping” policy dividends.

Third, mechanism tests reveal that national parks synergistically promote regional economic growth through three paths: (1) significantly increasing the scale of upper-level fiscal transfer payments, thereby alleviating the fiscal pressure on local governments; (2) increasing the share of the tertiary industry, promoting the optimization and upgrading of the industrial structure; (3) driving the growth of fixed-asset investment, providing critical capital support for regional economic development.

Finally, spatial analysis results show that national parks not only drive local growth but also generate significant positive spillover effects on surrounding areas through geographical proximity mechanisms. Spatial attenuation trend analysis further reveals that this spillover effect does not follow a simple linear decay but exhibits a significant “inverted U-shaped” characteristic: positive spillovers peak at 250 km, then decay, and lose significance at approximately 350 km. This finding clearly defines the effective geographical boundary of the national park’s economic radiation, indicating that it drives regional coordinated development within a radius of approximately 350 km through a “core-periphery” mechanism.

Based on the findings of this paper, the following policy recommendations are proposed:

First, increase financial and policy inclination towards underdeveloped areas to solidify the foundation for development. The study finds that the driving effect of the national park policy is more pronounced in areas with weaker economic foundations. Therefore, it is recommended to implement a differentiated strategy in fund allocation, prioritizing underdeveloped pilot counties. Central and provincial finances should focus on supporting the improvement of transportation roads and basic public services in these areas to help address infrastructure deficiencies. By improving the “hard environment” for investment, these regions can more quickly accommodate green industries such as ecotourism, thereby gaining tangible growth momentum in the early stages of the policy.

Second, promote the conversion of ecological advantages into industrial advantages according to local conditions to avoid homogeneous competition. Given the finding that policy effects are stronger in high-altitude and ecologically superior regions, it is recommended to fully utilize the unique landscape resources of these areas to pursue a path of distinctive and high-quality development. In development and utilization, low-level mass tourism development should be strictly controlled to avoid the “one park, one look” problem. Encouraging the development of high-value-added service formats such as nature education and ecological wellness, as well as building top-tier national park ecological brands, will improve the output efficiency per unit of resource and ensure sustainable economic returns.

Third, dynamically adjust the policy focus according to the construction stage to maintain long-term growth vitality. Given that policy effects slow down somewhat during the “institutional deepening stage,” it is suggested that policy support methods should be optimized according to the development stage. In the early stage of construction, the focus should be on relying on project investment and infrastructure construction to pull the economy. As the system matures and controls tighten, the focus should timely shift to cultivating endogenous drivers. For example, increasing support for community green industries and exploring diversified value realization paths such as franchising and carbon sink trading will prevent late-stage growth stagnation caused by reliance solely on external investment, ensuring the sustainability of policy dividends.

Fourth, establish cross-regional coordinated development mechanisms to allow surrounding areas to share development dividends. Based on the discovery of a radiation radius of approximately 350 km for national parks, the government should break administrative barriers and coordinate the planning of national parks and their surrounding areas. It is recommended to establish a cross-county coordination and cooperation mechanism to implement an integrated layout in tourism route design, transportation network connection, and industrial support. For instance, guiding surrounding areas to undertake supporting service functions such as catering, accommodation, and tourism distribution will form a dislocated complementarity with the core protected areas, allowing surrounding areas to gain economic benefits through serving the national park and achieving regional common development.