How Do Investment Structure and Business Category Affect the Ecological Efficiency of Forest Parks?—A Case Study from Liaoning Province, China

Abstract

Accurately assessing ecological efficiency and illustrating its impact mechanism can facilitate the construction of an environmentally friendly forest park management system. In this study, we took 28 national forest parks in Liaoning from 2008 to 2017 as study objects, constructed an ecological efficiency framework with a stochastic cost function, and analyzed empirically how investment structures affect the efficiency of forest parks and their mechanism. We found that, firstly, the improvement potential of the ecological efficiency of forest parks in Liaoning was 48%. Secondly, there was a large gap in the ecological efficiency among forests, but it was decreasing over time. Thirdly, investment structure had a significant impact on the ecological efficiency of forest parks. The increase in the proportion of private capital decreased the ecological efficiency, with the marginal contribution ratio being 10.6%. Fourthly, the business category played a mediating role between the investment structure and ecological efficiency of forest parks. Investment structures negatively affected ecological efficiency by increasing the proportion of accommodation and amusement. Lastly, there was heterogeneity in how investment structures affected the ecological efficiency of forest parks, with high-class forest parks through accommodation and low-class forest parks through amusement activities.

1. Introduction

Forest parks incorporate tourism and leisure into nature, which promotes local development and improves employment and social welfare. However, due to the high opportunity cost associated with nature conservation and the pressure brought by economic benefits, many governments chose to reduce the regulations on nature conservation, leading to the weakening of the restriction on destructive human development activities [1]. Especially in China, forest tourism has experienced a period of rapid growth since the policy of accelerating the development of forest tourism was issued in 2011. By 2019, the number of visitors to forest parks reached 740 million, and the tourism revenue was 9.527 billion US dollars, which was 2.56 times and 2.26 times that of 2011, respectively. After a long period of tourism development, ecological problems such as air pollution, soil erosion, ecological degradation, and vegetation damage arose because of the excessive pursuit of economic benefits [2]. Thus, we need to re-examine the management of forest parks and continuously explore ways of sustainable development that can effectively coordinate forest resource protection and tourism development.

Ecological efficiency signifies how efficient economic activity is in consideration of the ecosystem’s resources and services, and environmental impact [3], which effectively assesses whether the tourism business activities follow the principles of sustainable development [4]. By comparing the differences in the ecological efficiency of forest parks, we can get an overall understanding of forest tourism development in a country or a region. The calculation of ecological efficiency involves the production process of multiple inputs and outputs, and the calculation methods mainly include Data Envelope Analysis (DEA) and Stochastic Frontier Analysis (SFA). DEA, as a non-parametric method that can avoid the influence of setting the function form subjectively, has become the most widely used method in the field of tourism research [5,6,7,8,9]. However, this method fails to separate the efficiency value from the error term, which might overestimate the value of the inefficiency, making the research results more sensitive to outliers and sample size [10,11,12]. SFA separates the efficiency value from the error term and makes up for the shortcomings of DEA to a certain extent; however, there may be problems related to the explanatory variable and the combined error term, and the distance function often does not meet the concave and quasi concave assumptions in economic theory, which affects the credibility of its results [13]. To address these problems, Herrala and Goel established a new connection between the goal of minimizing pollutant emissions and the setting of random frontiers based on the stochastic frontier cost function (SFC) [14], which has not yet been applied in tourism research. In addition, concerning the selection of pollution variables (undesirable outputs), it is difficult to accurately measure the actual impact of pollutant emissions on the environment or ecology. The common practice is to separately incorporate pollutants with different chemical components into the model [15,16]. However, the effects of pollutants with different chemical components on ecosystems may be the same, such as the superposition effect of SO₂ and NO_x in the atmosphere, which together form secondary pollution such as acid rain, haze, and PM2.5, directly affecting the local’s survival and vegetation growth. Therefore, the impact of economic activities on ecosystems can be reflected by assigning weights to emissions with the same pollution properties for normalization.

To improve the ecological efficiency of tourism, the influencing factors were further analyzed. Except for the regional economic level, the level of tourism development, the technical level of energy, the investment level, and environmental regulation on the eco-efficiency of tourism and other external business environments [6,8], and tourism investment, as the internal operation foundation of forest parks, are important factors. At present, there have been disagreements on the relationship between tourism investment and ecological efficiency. Some viewpoints believe that tourism investment can improve ecological efficiency [8,17] because greater investment in the tourism industry will stimulate long-term tourism revenue, innovation, and sustainable growth in the sector [18]. Others argue that tourism investment reduces ecological efficiency [19]. Peng et al. and Li et al. both found a negative correlation between tourism investment and ecological efficiency [20,21]. It can be seen that tourism investment plays a heterogeneous role in ecological efficiency, and the business category is a key factor in the relationship between the two. If the investment is in green tourism-related infrastructure, the negative impact of tourism development on ecology can be reduced [22], and if the investment is in high energy consumption entertainment, the contradiction between tourism development and ecology will be exacerbated [21].

The investment structure is essential to tourism operators to attract tourists with different traveling needs. From a practical perspective, there are diversified investment patterns including government investment, private investment, public-private partnerships (PPP), etc. In general, public investment mainly focuses on tourism-related infrastructure, while private investment is typically for commercial purposes and mostly driven by profit incentives, which are concentrated in such sectors as commercial accommodation and transportation services that generate high economic income, such as vacation homes, hotels, conference centers, airplanes, cruise ships, and tourist buses, etc. [22]. Although private investment is more conducive to improving tourism efficiency compared to other sources of investment, these sectors are also pointed out as high energy-consuming industries [23], especially in developing countries where the economic benefits brought by tourism are often more valued, while the excessive development of resources and the generation of environmental pollution are neglected [24,25]. Hence, the business category of tourism destinations is affected by the investment structure, which not only directly affects tourism revenue, but also has a close relationship with energy consumption types and pollution emissions [8,26]. That is, investment structure and business category do not independently affect ecological efficiency, but there is a mechanism instead. However, existing research has been limited to the overall effect of investment on ecological efficiency, which makes it impossible to determine and compare the heterogeneous effects of different tourism investment entities on ecological efficiency. It is necessary to clarify the relationship between tourism investment structure and ecological efficiency, in particular, to further verify how investment structures affect ecological efficiency through different business categories, which will help establish an environmentally friendly forest park management system.

In this context, we took 28 national forest parks in Liaoning Province as the research objects to identify the impact of different investment structures on ecological efficiency from 2008 to 2017. The marginal contributions of this article are: firstly, considering the superposition effect of pollutants in the environment, the environmental acidification substances generated in the process were calculated as ecological indicators, and a modified ecological efficiency model was constructed with a stochastic frontier cost function (SFC) to measure and analyze the ecological efficiency. Secondly, we divided the investment of forest parks into public investment and private investment and focused on analyzing the impact of different investment structures on ecological efficiency. Thirdly, we introduced business categories as mediating variables and constructed a mechanism analysis framework of “investment structure–business categories–ecological efficiency”. Through the horizontal comparison of forest parks, the reasons for and paths of the decline of ecological efficiency are revealed from the inside. This research is helpful for policymakers and park managers to adopt more environmentally friendly projects and accordingly achieve the goal of sustainable development of forest tourism.

The structure of this paper is as follows: the second part describes the research hypothesis, research methods, data sources, and data characteristics. The third is the empirical analysis. In the fourth part, the scientific nature of the calculation method and the research results and findings as well as the research deficiencies and prospects were discussed. The last part is the conclusion and policy enlightenment.

2. Methodology and Data

2.1. Research Hypothesis

Forests parks in China are transformed from state-owned forest farms to self-management and self-financing forest farms. The construction funds are mainly from investments from society and fundraising. Certain funds for the building of tourist facilities, such as ecological car parks, hiking trails, and signage systems, can also be applied by the government. Additionally, the government provides annual financial compensation to public welfare forest landowners based on the land area [27], and the compensation received can be arranged independently by the forest park. Consequently, the investment structure of China’s forest parks is dominated by private capital and supplemented by public capital, which is composed of three investment subjects: government financial investment, self-financing capital, and external investment. Due to differences in the profitability of investment entities, the business categories are different [28]. The focus of government investment is on social welfare and ecological protection, which is relatively less profit-seeking and mainly makes protective investments to restore forests. While private investment is more directed at the pursuit of economic benefits, mainly focusing on development investment, such as the investment in accommodation, amusement, transportation, and other reception projects. Due to the different purposes of investment entities, there are remarkable differences in investment destinations. As a result, the share between the private and government capital in the investment structure inevitably affects ecological efficiency.

Investors can choose different tourism products for investment based on market research results and tourism planning. These projects not only bring differences in factor inputs of production and income but also cause varying degrees of environmental pressure. The energy consumption per bed per night is 120 MJ in hotels, 90 MJ in resorts, and only 50 MJ in campgrounds. Among amusement activities [23], the energy consumption of horse riding is 0.6 MJ/tourist, museums (10 MJ/tourist), zoos (16 MJ/tourist), experience centers (29 MJ/tourist), rafting (36 MJ/tourist), adventure activities (57 MJ/tourist), cruises (215 MJ/tourist), and diving (800 MJ/tourist) [29]. Fossil fuels are the main energy source in China’s forest parks, and high energy consumption means high pollution emissions. Tourism products preferred by consumers are often high-energy consuming [30], and private capital is more likely to choose high-energy tourism products to attract tourists. Although these tourism products can attract tourists and bring higher revenue, they consume more energy and cause greater pressure on the ecological environment. Accordingly, the investment structure can inevitably affect ecological efficiency, and business categories play a mediating role between tourism investment and ecological efficiency.

Based on the above analysis, we constructed the analysis framework of “investment structure, business category, ecological efficiency” (Figure 1). Consequently, we assumed the following:

Hypothesis H1. 

Investment structure has a significant impact on the ecological efficiency of forest parks. Specifically, the increase in the share of private investment leads to the decline of ecological efficiency, while the increase in the share of public investment brings the improvement of ecological efficiency.

Hypothesis H2. 

The business category plays a mediating role between investment structure and ecological efficiency. Specifically, private investment can cause a decline in ecological efficiency through accommodation and amusement.

2.2. Models

2.2.1. Estimation of Ecological Efficiency of Forest Parks

Forest parks, with their unique resources, diversify tourism products to generate revenue but concurrently cause environmental pollution through emissions. This divergence in eco-efficiency is quantifiable through the ratio between the minimum and actual emissions along environmental frontiers. Pollutants as bad outputs should be minimized as much as possible, so the relationship between minimum and actual emissions cannot be obtained directly in the form of a production function. For this reason, we employed the stochastic frontier function of environmental cost proposed by Herrala and Goel to estimate the ecological efficiency of forest parks [14]. According to the definition, Formula (1) is used to express the ecological efficiency of forest parks:

$${EE}_i=\exp \left(-u_i\right)=\frac{\mathrm{m}\mathrm{i}\mathrm{n}\left(P\right)}{P_i}$$

(1)

where, $E{E}_i$ is the ecological efficiency of the i^th forest park. $P_i$ represents the actual pollution emission of the i^th forest park. $\mathit{min}(P)$ is the minimum emission on the environmental frontier. $u_i$ denotes ecological inefficiency.

Following the stochastic frontier function method of environmental cost defined by Herrala and Goel [14], Equation (1) can be expressed as Equation (2).

$$P_i=f(y_i,\sum x_{i,k}:\beta )\mathrm{e}\mathrm{x}\mathrm{p}\left(u_i\right)$$

(2)

where, $f(y_i,\sum x_{i,k}:\beta )$ is the minimum emission value; $y_i$ is the operating income of the forest park; $x_{i,k}$ is the set of inputs, which includes labor, capital inputs, parkland, and energy.

Since the parameter estimation requires the specific form of the function, the more inclusive transcendental logarithm function was employed. Then the specific form of ecological efficiency can be transformed as Equation (3).

$$\begin{gathered} ln{P}_{it}={\alpha }_0+{\alpha }_y\ln y+\frac{1}{2}{\alpha }_{yy}{\left(\ln y\right)}^2+\sum {\alpha }_i\ln x_i+\frac{1}{2}\sum \sum {\alpha }_{ij}\ln x_i\ln x_j \\ +\sum {\alpha }_{ij}\ln x_i\ln y+{\alpha }_{t}t+\frac{1}{2}{\alpha }_{ty}t\ln y+\sum {\alpha }_{ti}t\ln x_i+v_{it}+u_{it} \end{gathered}$$

(3)

where $v_{i,t}$ is a random disturbance term, which satisfies the classical econometric hypothesis, i.e., $v_{it}~i$ $dN(0~{\sigma }_u^2)$; the value of $u_{i,t}$, which follows a half-normal distribution, is between 0 and 1. The operation of the forest parks is ecologically efficient if $u_{i,t}=1$, and it is ecologically inefficient if $u_{i,t}<1$. Battese and Coelli assumed that $u_{i,t}=u_i\exp [-\eta \left(t-T\right)]$, where η indicates the effect of the time factor on ecological inefficiency term [31]. Meanwhile, they set ${\sigma }^2={\sigma }_u^2+{\sigma }_v^2$ and $\gamma =\frac{{\sigma }_u^2}{{\sigma }^2}$ to test the proportion of ecologically inefficient items in the composite disturbance terms, where the value of γ is from 0 to 1. It implies that the ecological deviation is mainly determined by the ecological inefficiency item if $\gamma =1$, while the ecological deviation is mainly determined by random error if $\gamma =0$.

For this part of the empirical evidence we use frontier4.1.

2.2.2. Mediating Effect Model

In line with our hypothesis, the investment structure affects the ecological efficiency of forest parks through business categories such as accommodation and amusement, which can be tested through the mediating effect model. At present, there are three common methods: the Sobel test [32], the causal steps approach [33], and the bootstrap method which directly tests the significance of coefficient products [34]. Among them, the causal steps approach is simple and easy to understand and explain, and most rigorous in testing making it the most commonly used method. According to the test method of Baron and Kenny [33], the following causal steps approach is constructed to test the mediating effect.

$$E_{i,t}^U=a_{ot}+a_{1}K{C}_{i,t-1}+a_2{X}_{i,t}+{\xi }_{i,t}$$

(4)

$${MED}_{i,t}=b_{ot}+b_1{KC}_{i,t-1}+b_2{X}_{i,t}+{\xi }_{i,t}$$

(5)

$$E_{i,t}^U=c_{0t}+c_{1}K{C}_{i,t-1}+c_{2}ME{D}_{i,t}+c_3{X}_{i,t}+{\xi }_{i,t}$$

(6)

where, $E_{i,t}^U$ is the ecological efficiency of the i^th forest park; $K{C}_{i,t-1}$ denotes the investment structure of the i^th forest park in the previous year; $ME{D}_i$ represents business categories; and $X_{i,t}$ refers to the other control variables that influence forest parks. $a_1$ in Equation (4) is the total effect of investment structure on ecological efficiency; $b_1$ in Equation (5) is the impact of investment structure on the mediating variables; and $c_1$ in Equation (6) represents the direct impact of investment structure on the ecological efficiency of i^th forest park after considering the mechanism of business category. By observing the significance of $b_1{c}_2$, we can get the indirect impact of the investment structure on the ecological efficiency by the mediating variables. Considering the value of the dependent variable ranks from 0 to 1 [35], a truncated regression model (Tobit model) is applied to this scenario.

For this part of the empirical evidence we use Eviews8.

2.3. Variables

2.3.1. Input Variables

Labor, capital, land natural resources, and energy are the basic inputs for the operation of forest parks [35,36,37,38]. Labor is expressed by the number of employees. The capital stock was calculated by the perpetual inventory method [39,40], and the actual value was deflated according to the price level in 2011. The natural resource input of forest parks is land, and the footpath is used as the proxy variable of land input since the forest park does not use all of the approved acreages. The energy consumption of forest parks was converted into standard coal based on the relevant coefficients (shown in

Table 1. Conversion factors of Standard Coal.

Fuel Type	Coal (Ton)	Diesel Oil (Ton)	Gasoline (Ton)	Electricity (10,000 KWH)
Coefficient (ton)	0.714	1.4571	1.4714	1.229

) provided in [41], and then aggregated.

2.3.2. Output Variables

Output variables contain economic output and pollutant emission. The economic output was represented by the income of the forest parks, and the actual value was deflated according to the price level in 2011. The pollutants were described by the atmospheric acidification index, which is composed of SO₂ and NO_x. The pollutants were normalized on the basis of the environmental impact assessment method obtained from Equations (7)–(9).

$$Q_{s{o}_2}={\sum }_{k=1}^K{S}_k{F}_k{P}_k,k=\mathrm{1,2},\cdots ,K$$

(7)

$$Q_{n{o}_x}={\sum }_{k=1}^K{S}_k{F}_k{P}_k,k=\mathrm{1,2},\cdots ,K$$

(8)

$$Q_i=Q_{s{o}_2}+Q_{n{o}_x}\times 0.7,k=\mathrm{1,2},\cdots ,K$$

(9)

where, $S_k$ denotes the intensity of the k^th emission source; $F_k$ is the k^th emission factor; and $P_k$ refers to the unit conversion coefficient which ensures that the unit of Q is t/a. The equivalent coefficient of the conversion of environmentally acidified substances into corresponding environmental impact potential is 0.7 [42]. The SO₂ and NO_x emission factors are affected by combustion equipment and fuel types. Based on the relevant studies [43,44,45,46], we specified the emission factors of SO₂ and NO_x in Liaoning national forest parks (shown in

Table 2. SO₂ and NO_x Emission Factors of Fuel Types in Forest Parks.

Fuel Type	NOx				SO2
	Coal	Diesel Oil	Gasoline	Electricity	Coal	Diesel Oil	Gasoline	Electricity
Coefficient	4 kg/t	9.62 kg/t	16.7 kg/t	1.36 g/kWh	10 kg/t	2.24 kg/t	1.6 kg/t	0.39 g/kWh

).

2.3.3. Independent Variable

Investment structure was employed as the core independent variable. Due to the different nature of pursuing profit for different investors, there are great differences in the use and flow of funds. It is difficult for private capital to coordinate enterprise income and ecological problems, so the ratio of private capital to all capital investment was adopted as the variable of investment structure.

2.3.4. Mediating Variables

The business category was adopted as the mediating variable. The proportion of accommodation income and amusement income was used as the proxy variables of business categories. Accommodation and amusement activities produce economic benefits, and consume different types of energy, bringing varying degrees of ecological problems, which eventually exert different effects on ecological efficiency.

2.3.5. Control Variables

To avoid the endogenous problems caused by missing variables as much as possible, we input the factors that may affect the ecological efficiency of forest parks, containing business categories and investment structure, in this model. These control variables involve the external business environment and self-condition. The external business environment includes the per capita GDP, population, industrial agglomeration, traffic accessibility, scenic spot density, scenic spot grade, and energy structure [6,9,20]. Among them, (1) the per capita GDP was obtained by deflating 2011 as the base period. (2) Population size was represented by the population of the city where the forest park is located. (3) Scenic spot density was characterized by the number of scenic spots above A level in the city where the forest park is located. (4) Traffic accessibility refers to the distance between the city centers where the scenic spot is located and the scenic spot. (5) The scenic spot class follows the quality class standard of scenic spots in China. Forest parks are divided into five classes, 5A, 4A, 3A, 2A, and 1A from high to low. The higher the level, the higher the score of scenic spots in service quality, environmental quality, landscape quality, and tourist satisfaction. (6) The energy structure of the forest park was the ratio of electric energy to total energy because electricity is less polluting than other sources of energy.

2.4. Data

The forest park system in China is primarily composed of national-level forest parks, supplemented by provincial-level forest parks. As of the end of 2020, the total area of forest parks nationwide reached 18.5755 million hectares, with national-level parks accounting for 68.76% of the total area. For our study, we selected 28 national-level forest parks in Liaoning Province as the research subjects, covering the time period from 2008 to 2017. Liaoning Province boasts diverse and abundant forest resources, with a variety of operational categories for its forest parks, all under the unified management of the Liaoning Provincial Forestry Bureau. Choosing national-level forest parks in this region as our research focus provides a robust foundation for our argumentation. Regarding the variables, data such as capital stock, labor force, tourism income, investment sources, operational revenue, and hiking trail statistics were obtained from “Statistics on Forest Parks in Liaoning”. Information on energy types, energy consumption, and scenic area levels was collected through on-site investigations of the forest parks. Metrics related to the forest park’s region, including per capita GRDP, population size, and scenic area density, were sourced from “Liaoning Statistical Yearbook” and “Statistical Yearbooks of Various Cities.” The distance between the forest park and the city center was determined using Baidu Maps. The statistical characteristics of the variables are shown in

Table 3. Descriptive Statistical Analysis of Various Variables.

Variable Type	Variable	Unit	Variable Description	Mean	Std. Dev.
Input variables	capital	10 thousand RMB	calculated by the perpetual inventory method	13,249.83	27,087.31
	labor	people	the number of employees	90.01	94.96
	energy consumption	ton	converted into standard coal	222.75	221.08
	land	km	the length of the footpath	45.58	31.97
Output variable	tourism income	10 thousand RMB	the income from forest parks	2441.09	6055.84
Pollution variable	environmental acidity	ton	calculated by Equations (7)–(9)	2038.67	1995.11
Independent variables	the proportion of private capital	%	the ratio of private capital to all capital investment	84.43	28.85
	the proportion of self-financed capital	%	the ratio of self-financed capital to all capital investment	65.36	39.06
	the proportion of external capital	%	the ratio of external capital to all capital investment	19.07	33.32
Mediating variables	amusement	%	the ratio of amusement income to the total income	10.50	15.26
	accommodation	%	the ratio of accommodation income to the total income	28.65	24.40
Control variables	GDP per capita	10 thousand yuan	the ratio of GDP to population	5.98	2.92
	population	10 thousand people	population	392.28	200.71
	traffic accessibility	km	the distance from the scenic spot to the local railway station	84.14	62.10
	scenic spot density		the number of scenic spots above Class A	27.56	18.43
	scenic spot class		5A = 5, 4A = 4, 3A = 3, 2A = 2, A = 1, or else is 0	2.86	1.60
	energy structure	%	the ratio of electricity to energy	33.82	23.25

.

3. Results

3.1. Ecological Efficiency of Forest Park

We mapped the distribution of ecological efficiency of Liaoning National Forest parks according to the Kernel Density Curve (presented in Figure 2). Firstly, the peak of the kernel density lies on the left and concentrates between 0.4–0.5. This indicates that the ecological efficiency of most forest parks is low. The average value of ecological efficiency was 0.52 during the period from 2008 to 2017. Secondly, the kernel density curve of ecological efficiency was the single peak, suggesting that there is no phenomenon of bipolarity or multipolarity. Thirdly, the range of the kernel density curve was wide, demonstrating that the efficiency differences between forest parks were very large. The maximum value (0.98) was 3.85 times the minimum (0.26) from 2008 to 2017. Fourthly, the main peak moved slightly to the upper right over time, representing that the gap in ecological efficiency among forest parks decreased gradually. The average ecological efficiency of forest parks increased by 15% from 0.485 in 2008 to 0.558 in 2017.

3.2. The Mechanism of the Impact of Investment Structure on Ecological Efficiency

In consideration of heteroscedasticity, the Tobit model of panel data was adopted to analyze the impact of investment structure on the ecological efficiency of forest parks. As illustrated in

Table 4. Mechanisms of the Impact of Investment Structure on Ecological Efficiency.

		Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency
		Model 1	Model 2	Model 3	Model 4
Independent variable	investment structure	−0.106 ***	0.171 ***	0.142 **	−0.070 *
		(0.038)	(0.047)	(0.067)	(0.039)
Mediating variables	amusement activities				−0.150 **
					(0.055)
	accommodation				−0.219 ***
					(0.043)
Control variables	scenic spot class	−0.011 *	−0.033 ***	−0.027 ***	−0.021 ***
		(0.006)	(0.008)	(0.010)	(0.005)
	log(transportation)	−0.030 ***	0.022	0.075 ***	−0.019 *
		(0.011)	(0.015)	(0.019)	(0.011)
	scenic spot density	0.002 **	0.005 ***	0.002	0.003 ***
		(0.001)	(0.001)	(0.001)	(0.001)
	GDP per capita	0.018 ***	−0.009	0.002	0.017 ***
		(0.004)	(0.007)	(0.007)	(0.004)
	energy structure	0.087 ***	−0.068	−0.384 ***	0.005
		(0.032)	(0.048)	(0.059)	(0.033)
	population	−0.057 ***	−0.033 **	−0.030 **	−0.064 ***
		(0.009)	(0.014)	(0.015)	(0.000)
	constant	0.694 ***	0.146 *	0.236 **	0.797 ***
		(0.049)	(0.082)	(0.095)	(0.052)
	log likelihood	111.6	−26.2	−58.2	127.7

Note: * Significant at 0.1 level. ** Significant at 0.05 level. *** Significant at 0.01 level.

, after introducing the control variables fully, the proportion of private capital showed a negative relationship with the ecological efficiency of forest parks (β= −0.106, p < 0.01) in model 1, a 1% increase in the share of private capital in the investment structure is associated with a 10.6% decrease in the eco-efficiency of forest parks, which confirms the existence of the total effect of investment structure on the ecological efficiency of forest parks. Regarding the investment structure of forest parks, the higher the share of private capital, the lower the ecological efficiency of forest parks is, while the opposite is true for state-owned capital (H1 is supported). The results of model 2 and model 3 show that the investment structure affected the mediating variables. The increase of 1% in the share of private capital caused an increase in the share of amusement income and the share of accommodation income by 17.1% and 14.2%, respectively (β = 0.171, p < 0.01; β = 0.142, p < 0.05). This shows that forest parks increase tourism revenue mainly through private capital investment in amusement activities and accommodation operations, while state capital is mainly invested in non- or low-energy-consuming tourism products. As stated by the results of model 4, both the share of amusement income and the share of accommodation income had a significant negative relationship (β = −0.15, p < 0.05; β = −0.219, p < 0.01) under the control of investment structure, supporting the mediating role of business categories between investment structure and ecological efficiency (H2 were supported). The indirect effect of amusement and accommodation was −0.025 and −0.031, respectively, indicating that accommodation in forest parks has a greater negative impact on ecological efficiency compared to amusement.

For the control variables, the effects of GDP per capita and industrial agglomeration on ecological efficiency were significantly positive (β = 0.02, p < 0.01; β = 0.002, p < 0.1). The effects of population size and transportation distance on ecological efficiency were significantly negative (β = −0.06, p < 0.01; β = −0.03, p < 0.01).

3.3. Mechanisms of the Impact of Private Investment on Ecological Efficiency

The private investment in China’s forest parks comes from self-financing capital and external capital, and the differences in the scale of capital and operating capacity between them lead to some differences in the selection of business category and the impact on ecological efficiency. The findings obtained from model 5 and model 9 in

Table 5. Mechanisms of the Impact of Private Investment on Ecological Efficiency.

		Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency	Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency
		Model 5	Model 6	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Core variables	self-financed capital	−0.059 **	0.097 ***	0.079 *	−0.037
		(0.027)	(0.037)	(0.045)	(0.026)
	external capital					−0.016	0.006	0.025	−0.011
						(0.030)	(0.046)	(0.045)	(0.028)
Mediating variables	amusement activities				−0.162 ***				−0.174 ***
					(0.056)				(0.056)
	accommodation				−0.224 ***				−0.230 ***
					(0.043)				(0.043)
Control variables	scenic spot class	−0.012 *	−0.031 ***	−0.025 **	−0.022 ***	−0.013 ** (0.006)	−0.029 ***	−0.023 **	−0.023 ***
		(0.006)	(0.008)	(0.010)	(0.005)		(0.008)	(0.010)	(0.005)
	log(transportation)	−0.033 ***	0.026 *	0.078 ***	−0.020 *	−0.031 ***	0.023	0.075 ***	−0.018 *
		(0.010)	(0.015)	(0.019)	(0.011)	(0.010)	(0.015)	(0.019)	(0.011)
	scenic spot density	0.002 *	0.006 ***	0.002 *	0.002 ***	0.002 **	0.005 ***	0.002	0.003 ***
		(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.002)	(0.001)	(0.001)
	GDP per capita	0.021 *** (0.005)	−0.015 **	−0.003	0.019 ***	0.017 ***	−0.009	0.004	0.016 ***
			(0.007)	(0.007)	(0.004)	(0.005)	(0.007)	(0.008)	(0.005)
	energy structure	0.090 *** (0.033)	−0.069	−0.388 ***	0.004	0.080 **	−0.056	−0.374 ***	−0.004
			(0.047)	(0.060)	(0.034)	(0.033)	(0.049)	(0.061)	(0.034)
	population	−0.057 *** (0.009)	−0.032 **	−0.029 **	−0.064 ***	−0.059 ***	−0.030 **	−0.028 *	−0.065 ***
			(0.014)	(0.015)	(0.000)	(0.009)	(0.015)	(0.015)	(0.000)
	constant	0.754 *** (0.052)	0.054	0.160 *	0.839 ***	0.736 ***	0.089	0.181*	0.832 ***
			(0.084)	(0.096)	(0.051)	(0.052)	(0.085)	(0.099)	(0.052)
	log likelihood	109.3	−29.2	−59.7	126.5	107.1	−33.2	−61.3	125.5

* Significant at 0.1 level. ** Significant at 0.05 level. *** Significant at 0.01 level.

exhibit that the share of self-financing capital is substantially negatively related to ecological efficiency (β = −0.059, p < 0.01), but the share of the external capital is insignificant. It indicates that in the investment structure of national forest parks in Liaoning Province, self-financed capital is the capital that really significantly affects the ecological efficiency of forest parks. The outcomes of model 6 and model 7 show that an increase of 1% in the share of self-financing capital led to an increase in the share of amusement revenue and the share of accommodation revenue by 9.7% and 7.9%, respectively (β = 0.097, p < 0.01; β = 0.079, p < 0.1), revealing that self-financing capital increases tourism revenue by investing in amusement activities and accommodation. The results of model 8 show that under the control of investment structure, both the share of amusement revenue and the share of accommodation revenue had a significant negative relationship (β = −0.162, p < 0.01; β = −0.224, p < 0.001), proving that the business category plays a mediating role between self-financing capital and ecological efficiency. The indirect effect of amusement activities was −0.016 and the indirect effect of accommodation operation was −0.018, which implies that accommodation still has a more negative effect on ecological efficiency than amusement activities.

3.4. Robustness Analysis

In conformity with the policies of accelerating the development of forest tourism issued by the State Forestry Administration and the National Tourism Administration in 2011, we tested the robustness of the estimation results of models 1–12 by analyzing the samples from 2011 to 2017. As presented in

Table 6. Robustness Analysis (1).

		Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency
		Model 1	Model 2	Model 3	Model 4
	investment structure	−0.129 ***	0.187 ***	0.224 ***	−0.075 *
		(0.040)	(0.052)	(0.069)	(0.041)
Mediating variables	amusement activities				−0.115 *
					(0.067)
	accommodation				−0.242 ***
					(0.049)
Control variables	scenic spot class	−0.009	−0.037 ***	−0.038 ***	−0.021 ***
		(0.007)	(0.009)	(0.011)	(0.006)
	log(transportation)	−0.032 **	0.017	0.054 ***	−0.023 *
		(0.012)	(0.017)	(0.020)	(0012)
	scenic spot density	0.003 ***	0.005 **	0.001	0.003 ***
		(0.001)	(0.002)	(0.002)	(0.001)
	GDP per capita	0.019 ***	−0.015 *	−0.005	0.017 ***
		(0.005)	(0.008)	(0.008)	(0.005)
	energy structure	0.087 **	−0.083	−0.392 ***	−0.005
		(0.039)	(0.055)	(0.067)	(0.041)
	population	−0.069 ***	−0.027	−0.007	−0.071 ***
		(0.012)	(0.021)	(0.019)	(0.012)
	constant	0.691 ***	0.220 **	0.618 ***	0.818 ***
		(0.054)	(0.098)	(0.094)	(0.057)
	log likelihood	84.1	−4.6	−28.1	95.6

* Significant at 0.1 level. ** Significant at 0.05 level. *** Significant at 0.01 level.

and

Table 7. Robustness Analysis (2).

		Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency	Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency
		Model 5	Model 6	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Core variables	self-financed capital	−0.070 **	0.092 **	0.119 **	−0.039
		(0.030)	(0.046)	(0.053)	(0.030)
	external capital					−0.042	0.057	0.064	−0.022
						(0.032)	(0.059)	(0.056)	(0.028)
Mediating variables	amusement activities				−0.138 **				−0.147 **
					(0.068)				(0.068)
	accommodation				−0.258 ***				−0.264 ***
					(0.049)				(0.048)
Control variables	scenic spot class	−0.010	−0.034 ***	−0.036 ***	−0.023 ***	−0.015 **	−0.028 ***	−0.028 **	−0.026 ***
		(0.007)	(0.010)	(0.011)	(0.006)	(0.007)	(0.009)	(0.012)	(0.006)
	log(transportation)	−0.034 **	0.018	0.056 ***	−0.023 *	−0.030 **	0.014	0.051 **	−0.021 *
		(0.01)	(0.017)	(0.020)	(0.012)	(0.012)	(0.017)	(0.020)	(0.012)
	scenic spot density	0.002 **	0.006 ***	0.002	0.003 ***	0.003 ***	0.005 **	0.001	0.003 ***
		(0.001)	(0.002)	(0.002)	(0001)	(0.001)	(0.002)	(0.002)	(0.001)
	GDP per capita	0.022 ***	−0.019 **	−0.010	0.019 ***	0.015 **	−0.008	0.003	0.014 **
		(0.005)	(0.009)	(0.009)	(0.005)	(0.006)	(0.009)	(0.009)	(0.005)
	energy structure	0.089 **	−0.08	−0.398 ***	−0.01	0.064	−0.048	−0.352 ***	−0.027
		(0.040)	(0.054)	(0.070)	(0.042)	(0.039)	(0.056)	(0.071)	(0.041)
	population	−0.067 ***	−0.030 *	−0.011	−0.070 ***	−0.066 ***	−0.030	−0.012	−0.070 ***
		(0.012)	(0.021)	(0.020)	(0.012)	(0.012)	(0.022)	(0.020)	(0.012)
	constant	0.760 ***	0.128	0.512 **	0.868 ***	0.759 ***	0.126	0.489 **	0.871 ***
		(0.059)	(0.100)	(0.105)	(0.057)	(0.062)	(0.104)	(0.111)	(0.059)
	log likelihood	80.6	−9.5	−32.4	94.4	79	−11.3	−34.4	93.9

* Significant at 0.1 level. ** Significant at 0.05 level. *** Significant at 0.01 level.

, the estimation results were consistent with those in Table 4 and Table 5 except for some of the individual control variables. That is, the significance, correlation, and coefficient values were similar, indicating that the estimation results in the present work are robust.

3.5. Heterogeneity Analysis

The findings above verified that investment structure affects tourism efficiency by business categories. It is worth noting that the class of scenic spots reflects the comprehensive competitiveness of forest parks, such as service quality, management ability, and environmental quality. The high-class scenic spot with high market attractiveness can attract more capital, thus providing more abundant tourism products. In other words, the influencing mechanism of different grades of forest parks on ecological efficiency may be different. Therefore, we divided the forest parks into two groups: high-class (5A and 4A) and low-class (3A and below) based on the Classification and Evaluation of Tourist Scenic Spot Quality Class in China. The regression results are shown in

Table 8. Heterogeneity Analysis (1).

		Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency
		Model 1	Model 2	Model 3	Model 4
High-class scenic spots	investment structure	−0.067	0.076	0.149	−0.048
		(0.061)	(0.053)	(0.109)	(0.064)
	amusement activities				−0.165
					(0.191)
	accommodation				−0.152 **
					(0.078)
	control variables	Include	Include	Include	Include
	log likelihood	46.6	24.1	−15.3	49.7
Low-class scenic spots	investment structure	−0.133 ***	0.240 ***	0.096	−0.095 ***
		(0.040)	(0.069)	(0.084)	(0.042)
	amusement activities				−0.201 ***
					(0.055)
	accommodation				−0.137 ***
					(0.042)
	control variables	Include	Include	Include	Include
	log likelihood	97.735	−30	−36.8	108.6

** Significant at 0.05 level. *** Significant at 0.01 level.

.

For high-class scenic spots, there was no significant effect of investment structure on the ecological efficiency of forest parks, and the total effect does not exist. For low-class scenic spots, the proportion of private capital had a significant positive correlation with ecological efficiency (β = 0.133, p < 0.01), and the proportion of private capital had a significant positive relationship with the proportion of amusement income (β = −0.24, p < 0.01), but it had no significant effect on the proportion of accommodation income. After the mediating variables were included in the regression equation, the mediating effect of amusement activities was still significant, demonstrating that private capital mainly reduces ecological efficiency by the amusement activities of low-class scenic spots.

We further tested models 1 to 12 by distinguishing the private capital into self-financed capital and external capital, and the results are shown in

Table 9. Heterogeneity Analysis (2).

		Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency	Ecological Efficiency	Amusement	Accommodation	Ecological Efficiency
		Model 5	Model 6	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
High-class scenic spots	self-financed capital	−0.110 ***	−0.012	0.122 *	−0.101 **
		(0.041)	(0.037)	(0.067)	(0.041)
	external capital					0.079 *	0.058	−0.07	0.080 *
						(0.044)	(0.050)	(0.071)	(0.044)
	Amusement activities				−0.208				−0.219
					(0.185)				(0.188)
	accommodation				−0.136 *				−0.150 *
					(0.075)				(0.077)
	control variables	Include	Include	Include	Include	Include	Include	Include	Include
	log likelihood	49.2	23	−14.9	52.2	47.4	24	−16.1	50.8
Low-class scenic spots	self-financed capital	−0.079 ***	0.123 **	−0.006	−0.064 **
		(0.031)	(0.059)	(0.060)	(0.030)
	external capital					−0.026	0.077	0.151 **	0.002
						(0.044)	(0.069)	(0.065)	(0.035)
	Amusement activities				−0.219 ***				−0.242 ***
					(0.056)				(0.058)
	accommodation				−0.147 ***				−0.141 ***
					(0.043)				(0.044)
	control variables	Include	Include	Include	Include	Include	Include	Include	Include
	log likelihood	94.46	−33.9	−37.7	107.4	91.36	−35.8	−35.7	104.9

* Significant at 0.1 level. ** Significant at 0.05 level. *** Significant at 0.01 level.

. Unlike the results in Table 7, the self-financed capital showed a significant negative relationship with ecological efficiency in both high- and low-class scenic spots (β = −0.11, p < 0.01; β = −0.079, p < 0.01), while it showed a significant positive relationship with the share of the revenue from accommodation in high-class forest parks (β = 0.122, p < 0.1), and a significant positive relationship with the share of the revenue of amusement activities in low-class forest parks (β = 0.123, p < 0.05). The mediating effect of accommodation was recognized in high-class scenic spots (β = −0.136, p < 0.01), and the mediating effect of amusement was observed in low-class scenic spots (β = −0.219, p < 0.01). The mediating effect of the external capital was untenable. This means that more funds were raised in high-class forests to be in hotels, hot springs, and other projects; whereas, in low-class forest parks, funds were invested in amusement activities with a relatively small amount of capital owing to the limitation of various factors such as the size and level of the scenic spot. Consequently, in high-class forest parks, the self-financed capital had a negative impact on ecological efficiency by accommodation, while in low-class forest parks, the self-financed capital had a negative impact on ecological efficiency mainly by amusement activities.

4. Discussions

4.1. Ecological Efficiency Characteristics of Forest Parks

It is indicated in this research that ecological efficiency could be more scientifically measured through the environmental cost stochastic frontier function and the new pollutant emission indicators based on environmental impact assessment. Due to the difficulty in obtaining data on environmental pollutants, previous studies included SO₂ or NO_x in ecological efficiency models [16,47] or only included SO₂ as an environmental acidifier [21,30], which cannot accurately reflect the actual impact of pollutant emissions generated by tourism operations on the environmental system. The weights provided by the environmental impact assessment were added in our study to normalize SO₂ and NO_x with different chemical compositions, but similar pollution properties on the total amount of real environmental acidification substances emitted by the forest park during its operation were re-estimated. In terms of measurement methods, we adopted the environmental cost stochastic frontier function proposed by [14] to establish a revised ecological efficiency model, which avoids overestimating the ineffective rate value and the sensitivity of research results to outliers in the previous literature using DEA.

The mean value of the eco-efficiency of forest parks in Liaoning Province during the study period is 0.52, which means that the potential for improving the eco-efficiency of forest parks is 48%, and the forest parks with the lowest ecological efficiency can improve by 72% potentially compared with the optimal forest parks. This indicates that the gap in ecological efficiency was very large, but was gradually narrowing over time. Similarly, the ecological efficiency of Huangshan Forest Park from 1981–2014, measured by [20], was 0.53; [48] and [21] calculated the ecological efficiency of inter-provincial forest parks in China as 0.773 and 0.502, respectively. These results confirm that the ecological efficiency gap of China’s forest parks is very large both vertically and horizontally. Our findings indicate that a higher proportion of clean energy, higher per capita GDP, and industrial agglomeration contribute to enhancing the ecological efficiency of forest parks. Regions with higher economic development levels and concentrations of the tourism industry can provide a better business environment for forest parks, attracting more visitors and enabling the adoption of more advanced management techniques [49]. A high share of clean energy means that fewer pollutants are emitted while creating the same economic value. Conversely, higher park grades, larger local populations, and further distance from transportation hubs tend to lower the ecological efficiency of forest parks. These results suggest that forest parks often attract visitors through energy-intensive operational activities, especially those located at a considerable distance, which typically offer accommodation services. The hotel industry has been confirmed as a low ecological efficiency sector in tourism [6,8,20,21], corroborating our research findings.

In addition to these factors, the investment structure and operational categories of the forest park are worthy of scholarly exploration as influential factors in ecological efficiency. Although the 28 national forest parks in Liaoning Province are of the same class and are under the control of the same government department, there were obvious differences in their investment structure and business category. For example, from 2008 to 2017, the average private investment took up 100% in Shenyang forest parks, with skiing as the main operating project, and private investment took up 92.4% in Dalian Xijiao Forest Park, with hot springs and hotels as the main operating projects, while national investment accounted for 20% in Benxi Forest Park with natural landscapes as the main operating project. The business category not only affects the energy structure of forest parks but also is an important reason for deciding whether to carry out perennial operations. In northern China, where coal is a major heating source and source of air pollution, forest parks that run perennially consume more coal than those that run seasonally, all of which can lead to differences in the ecological efficiency of forest parks.

We also found that, over time, the ecological efficiency gap of forest parks gradually decreases, which is consistent with [48]. Firstly, the Chinese government has been intensifying efforts in environmental governance. In 2009, the government issued the “Opinions of the State Council on Accelerating the Development of the Tourism Industry”, aiming to reduce electricity consumption by 20% in five years for star-rated hotels and A-grade scenic areas. In 2013, the State Council introduced the “Notice on Issuing the Action Plan for Air Pollution Prevention and Control”, targeting a more than 10% reduction in the concentration of inhalable particulate matter in national cities at or above the prefectural level by 2017 compared to 2012. In response to increasingly stringent environmental regulations, forest parks have implemented proactive measures, such as transitioning from coal to electric heating, as observed in Anshan National Forest Park. Secondly, technology exhibits a spillover effect, enabling mutual emulation and learning among park managers in terms of management experience and energy technologies, which led to a gradual improvement in the overall utilization of forest park resources, narrowing the gap in ecological efficiency.

4.2. The Relation between Investment Structure, Business Category, and Ecological Efficiency

The results above are new in that there has never been a study on the relationship between investment structure, business category, and ecological efficiency. The existing research has analyzed the impact of tourism investment on pollutant emissions in tourist destinations [8,19,22] or ecological efficiency [20,21]. However, there is no agreement, as these studies overlooked the complexity of investment structures in tourist destinations or scenic spots. However, both forest parks and tourist destinations may be composed of multiple capital types. Distinguishing capital types and investigating their impact mechanisms on ecological efficiency can contribute to identifying the types of capital that have a negative impact on the environment and their degree of impact.

Our results indicate that investment structure significantly influences the ecological efficiency of forest parks, and business category plays a mediating role between the two. The results, consistent with Mascia et al. and Paramati et al. [1,22], demonstrate that the higher the proportion of private capital in the investment structure, the lower the ecological efficiency of forest parks, and vice versa. This can be explained by the fact that a significant portion of national capital is used for the ecological construction of forest parks, such as afforestation and forest transformation, as well as the construction of parking lots, signage systems, and restroom renovations for tourism-related facilities. However, these activities do not significantly contribute to increased tourism revenue or visitor numbers [50]. In developing countries with low levels of urbanization, consumers prefer energy-intensive tourism products such as luxury hotels, spa resorts, and amusement parks to rustic and back-to-nature ones [30]. Private capital aims to maximize profits as much as possible during the contract period, and in order to meet the market demand mostly focuses on developmental investment. Both energy-consuming amusement programs and accommodation programs will aggravate the conflict between tourism development and ecology [21,23]. As our empirical results indicate, an increase in the proportion of private capital corresponds to a higher share of forest park revenues coming from recreation and accommodation, both of which negatively affect eco-efficiency. As forest recreation gradually gains popularity, hotels and restaurants, spa treatment, and rural village accommodations are increasingly favored by large-scale private capital. The energy consumption intensity of lodging facilities located on the periphery is two to three times higher than that of urban centers [51]. To maximize profits, private capital is unlikely to willingly embrace cleaner energy sources or invest in more energy-efficient equipment, as these would raise their costs, a finding in line with [30]. Effective forest park management in many developing countries may be achieved by decentralizing the allocation and control of investment funds to relevant interest groups [52,53]. However, our results find that forest parks with a significant private capital share experience a decrease in eco-efficiency. This is because the profit-seeking nature of the investment body is different, leading to differences in the destination of capital investment, so investment structure is the direct cause of the differences in the ecological efficiency of forest parks, and its path is the operation category.

Finally, the negative effect of investment structure on the eco-efficiency of forest parks was 3.1% higher in high-level forest parks because of the heterogeneity effect of the operation category as a mediating effect in different levels of forest parks. High-level forest parks have the capacity to secure more substantial funds, enabling investments not only in amusement projects but also in the development of hotels, spas, accommodations, and other hospitality-related ventures. Conversely, low-scenic-level forest parks typically have access to fewer funds, leading to investments in lower-cost, faster-return amusement projects. For example, KUANDIAN Forest Park self-funded $900,000 for the construction of tourism projects such as children’s playgrounds, a skating rink, and a water park. Therefore, the high-leveled scenic area has a negative impact on eco-efficiency mainly through food and lodging operations; while low-leveled forest parks have a negative impact on eco-efficiency mainly through amusement operations. Since the negative impact of accommodation on eco-efficiency is higher than that of amusement, the negative impact of investment in forest parks is more pronounced in high-level forest parks. Specifically, it is more difficult to attract investment in rural areas of Liaoning Province, China, and this phenomenon is more prominent in forest parks with a high proportion of self-financing, and these findings can help the government and forest park operators to be more targeted in their environmental management and pollution control.

4.3. Limitation and Future Research

One of the shortcomings of this research is that we only consider the emissions of environmental acidification substances during the operation of forest parks. Coal, gasoline, diesel, gas, and other energy sources are consumed in tourist attractions, which emit not only environmental acidification substances but also greenhouse gases and PM2.5. For the reason that the forest system can absorb greenhouse gases and fine particles, we have ignored these problems. Another challenge is that obtaining data from non-listed companies is so difficult that we only obtained 10-year data for 28 forest parks in one province. Fortunately, these samples almost cover the number of local national forest parks and are representative of the region.

Considering the specific research content, future researchers can adopt the environmental impact assessment to give weight to more pollutants with different properties as an environmental indicator, accordingly including in the ecological economic model for relevant research on sustainable development. This method is not limited to the tourism field but can also be applied in other industries such as agriculture or industry. For forest parks or other tourism fields, future research can further reveal the impact mechanism of ecological efficiency, which is helpful in achieving the initial goal of sustainable tourism.

5. Conclusions and Implications

Based on the input and output data of 28 national forest parks in Liaoning province from 2008 to 2017, we measured the ecological efficiency with the atmospheric acidification index and the stochastic cost function, constructed the analysis framework of “investment structure, business category, ecological efficiency”, and analyzed the impact of investment structure on the ecological efficiency of forest park and its mechanism. The conclusions are as follows:

Firstly, the average improvement potential of the ecological efficiency of national forest parks in Liaoning province was 48%. There was a large gap in the ecological efficiency among forest parks, but these differences were decreasing over time. Secondly, the investment structure had a significant impact on the ecological efficiency of the forest parks. The increase in the proportion of private capital in the investment structure significantly decreased the ecological efficiency of the forest parks, and its overall impact on the ecological efficiency of the forest parks was more than 0.106. Thirdly, business categories played a mediating role between the investment structure and the ecological efficiency of the forest park. Investment structures negatively affected the ecological efficiency of the forest parks by increasing the proportion of accommodation and amusement activities. Finally, there is heterogeneity in the way that investment structures affect the ecological efficiency of forest parks, with high-class forest parks through accommodation and low-class forest parks through amusement.

Although our evidence and arguments are derived from a specific province in China, they bear relevance for other regions and countries, especially in developing nations. We proffer a quartet of strategic paradigms poised to augment the ecological efficiency of forest parks. Firstly, atmospheric pollution is primarily caused by using fossil fuels, so adjusting the fuel composition and improving energy efficiency can help improve eco-efficiency. Echoing the sentiments articulated by [17], policymakers can incentivize forest park operators to replace antiquated machinery and facilities with more efficient alternatives and increase the proportion of clean energy through financial subsidy policies, such as installing solar streetlights and solar-powered buildings, promoting the use of new energy tour vehicles, and encouraging visitors to explore on foot. These devices not only align better with the ecological tourism philosophy of forest parks but also, from a long-term perspective, reduce fossil fuel consumption and economize tourism operation costs. Secondly, corporate innovation can contribute to improved energy efficiency and reduced environmental pollution [54,55]. The government can stimulate businesses to invest in research and development for low-energy engines in the tourism sector through policies such as tax reductions or subsidies as well as the development of building insulation materials to minimize heat loss from heating or air conditioning systems. Concurrently, public-private partnerships (PPPs) provide a strategic avenue to address government fiscal constraints, reducing business risks and alleviating project financing challenges. The government should supervise through meticulous investment selection mechanisms and contractual frameworks to mitigate investment risks and prevent disputes. Thirdly, forest park operators can enhance ecological efficiency through foreign direct investment (FDI) to introduce low-energy tourism products. An array of scholarly investigations indicates that FDI inflows play a significant role in reducing energy consumption across sectors [22,56,57,58]. Therefore, the infusion of foreign capital emerges as a potent lever for operators to elevate their ecological sustainability by assimilating advanced technologies into their operational environment. Fourthly, effective management of tourism product development is imperative [59,60]. Policymakers can guide private capital to develop the experiential value of forest parks through special appropriations or project subsidy policies. For instance, based on the unique resource advantages of each scenic area, developing and constructing creative hotels such as camping tents, crystal houses, and log cabins with lower energy consumption and construction costs, instead of constructing energy-intensive star-rated hotels commonly found in urban areas. Encouraging forest parks to increase the proportion of low-energy recreational products is essential. The government can engage in collaborative endeavors with forest parks to orchestrate forest tourism festivals, allowing visitors to experience joy in activities like forest yoga, fishing, bamboo rafting, nature education, forest picking, forest bathing, and jungle trekking, with an emphasis on experience, participation, health, and individual development, so as to achieve development and protection in tandem and to place equal emphasis on economy and ecology.