*3.1. Theoretical Basis*

According to the theory of economic development, in the initial stage, more consideration will be given to allocating limited resources to the industrial sectors with the most productive potential, that is, the greatest linkage effect. Priority is given to developing these sectors while overcoming the bottleneck of economic development. This drives the development of other industries and sectors, and the economic development at this time is unbalanced. However, when the economy enters an advanced stage, from the perspective of industrialization and speeding up economic development, various departments and industries should maintain a certain proportional relationship and development in coordination. The economic development at this time is balanced. The ultimate goal of short-term unbalanced growth is to achieve long-term balanced development. In economic decision-making, resource protection and sustainable resources utilization must be highly valued and considered comprehensively. Improving resource utilization efficiency and economic development are inseparable. For economic development, the excessive use and development of natural resources have caused problems such as resource abuse and

The theoretical basis of this paper is mainly analyzed from the perspective of supply and demand. In the early stage of economic development, people's goal was survival. Transportation was inconvenient in places with abundant forest resources, and the industrial structure was single. Farmers mainly lived on cultivated land, and people had more need for basic material life. At the same time, population increase has brought enormous pressure on agricultural land, and the quality of agricultural land is getting lower and lower. This change also gradually reduces the marginal product of labor and lowers incomes [48]. Economic development is inhibited. But with the economic development to a certain extent, the rapid development of industrialization and urbanization has provided many non-agricultural employment opportunities and higher wages [49]. People's lives have improved, and they have begun to pursue a better living environment. The economy has begun to seek green and low-carbon development. A place with abundant forest resources can promote high-quality economic development. The continuous development of science and technology makes the sustainable management of forest resources more efficient. In addition, the public health and environmental functions brought by forest resources also generates corresponding economic value [50].

In Figure 2, in the early stage of economic development, when the forest coverage rate increases from D1 to D2, economic growth will decrease from Y2 to Y1. Forest resources will affect the allocation of land resources, which may bring about poverty and ultimately affect economic development. Forests have the potential to encroach on agricultural land, and agricultural land change is often used as a proxy for forest cover loss. The increased forest resources have brought about a single industrial structure and inconvenient transportation. Initially, people could only rely on the natural forests or land converted from them for their livelihoods [29]. However, according to the elasticity of demand theory in economics, when the economy expands, people's income levels will increase, and economic prosperity will increase people's demand for green resources. With the advancement of technology, people's incomes are no longer limited to arable land. Abundant forest resources have spawned tourism, led to the gathering of people, and increased employment. Increasing green supply also meets the needs of low-carbon economic development. If the forest cover rises from D3 to D4, the level of economic development will rise from Y3 to Y4. The curve to the right of the dotted line DE reflects people's demand for forest resources with economic development. Figure 2 reflects the impact of increasing forest resources on economic growth.

Based on the above analysis, this paper proposes four hypotheses:

**Hypothesis 1:** The impact of forest resource abundance on the level of economic development in the YRD region has a U-shaped non-linear characteristic.

**Hypothesis 2:** The impact of forest resource abundance on economic development has spatial spillover effects in YRD.

**Hypothesis 3:** The abundance of forest resources in the YRD region promotes green total factor productivity improvement.

**Hypothesis 4:** The impact of economic development on the abundance of forest resources presents an environmental Kuznets curve in YRD.

12

**Figure 2.** Theoretical framework.

## *3.2. Spatial Econometric Model*

Spatial econometric models usually include the spatial autoregressive model (SAR), spatial error model (SEM), and spatial Durbin model (SDM). Spatial econometric models are gradually being used to verify the relationship between environmental economics studies such as carbon dioxide emissions and economic growth and to demonstrate spatial spillover effects [51].

The steps of regression analysis in this paper are as follows. First, determine whether there is a spatial correlation, and then determine whether a spatial measurement model is available. On the premise of determining the available spatial econometric model, the POLS regression is firstly performed, and the Lagrange multiplier (LM) test of the SAR and SEM models is performed simultaneously. The LM method is applied to test the spatial interaction of the data, specifically the spatial lag and spatial error autocorrelation [52]. The LM test is based on the residuals of non-spatial models with spatial fixed effects, temporal fixed effects, and spatial and temporal double fixed effects, and obeys a chi-square distribution with 1 degree of freedom. If the results reject POLS in favor of the SAR or SEM model, SDM should be selected because SDM model includes both spatially lagged explanatory variables (WY) and spatially lagged explanatory variables (WX). The spatial lag explained variable WY represents the interaction effect between the explained variable and adjacent spatial units. The SDM model can produce better fitting results [53]. The spatial benchmark regression model is constructed as follows.

$$Y\_{it} = \rho \mathcal{W}\_i Y\_t + \beta X\_{it} + \theta \mathcal{W}\_i X\_{it} + \mu\_i + \xi\_t + \varepsilon\_{it} \tag{1}$$

$$\mathcal{W}\_{\text{i}}Y\_{\text{f}} = \sum\_{j=1}^{n} \mathcal{W}\_{\text{ij}}Y\_{\text{j}\text{f}} \tag{2}$$

$$
\varepsilon\_{it} = \lambda M \varepsilon\_t + u\_{it} \tag{3}
$$

where *Y* and *X* denote the explained variable and explanatory variable, respectively. *Wi* is the *ith* row of the spatial weight matrix *W*. *μ<sup>i</sup>* and *ξ<sup>t</sup>* represent the optional individual effect and time effect, respectively. ε represents the disturbance term. *i* = 1, 2, ... , N. *t* = 1, 2 . . . , T. *W* represents the spatial weighting matrix of the dependent variable, and M is the disturbance term. *Wij* represents the spatial weight matrices of city *i* and city *j*. *ρ* is the coefficient of the spatial lag term of the explained variable. *θ* is the spatial autocorrelation coefficient of the explanatory variable. *β* and *θ* represent the parameter vector. *β* reflects the influence of explanatory variables on the explained variables. λ is the coefficient of the

error term. Here are two hypotheses: H0: *θ* = 0; H0: θ + *ρβ* = 0;

If *θ* = 0, SDM will transform into SAR,

$$Y\_{it} = \rho W\_i Y\_t + X\_{it} \beta + \varepsilon\_{it} \tag{4}$$

If *θ* = −*ρβ*, SDM will transform into SEM,

$$Y\_{it} = X\_{it}\boldsymbol{\beta} + \varepsilon\_{it} \tag{5}$$

If *ρ* = 0, λ = 0, the model will degenerate into POLS.

All the data are normalized to minimize the absolute difference and avoid the influence of extreme values. In order to prevent the bias and endogeneity problems caused by model estimation, this paper constructs SDM, SAR, and a non-spatial panel data model, respectively.

$$\text{GDPPC}\_{\text{it}} = \rho \sum\_{j=1}^{n} \mathcal{W}\_{\text{ij}} \text{GDPPC}\_{\text{j}t} + \beta \text{'CORE}\_{\text{it}} + \theta \sum\_{j=1}^{n} \mathcal{W}\_{\text{ij}} \text{CORE}\_{\text{ij}t} + \beta \text{''CONT}\_{\text{it}} + \theta \sum\_{j=1}^{n} \mathcal{W}\_{\text{ij}} \text{CONT}\_{\text{ij}t} + \mu\_{\text{i}} + \xi\_{\text{t}} + \varepsilon\_{\text{it}} \tag{6}$$

$$GDPPC\_{it} = \rho \sum\_{j=1}^{n} \mathcal{W}\_{ij} GDPPC\_{jt} + \beta' \mathcal{CORE}\_{it} + \beta'' \mathcal{CONT}\_{it} + \mu\_i + \xi\_t + \varepsilon\_{it} \tag{7}$$

$$\text{GDPPC}\_{it} = \beta^{\prime} \text{CORE}\_{it} + \beta^{\prime\prime} \text{CONT}\_{it} + \mu\_i + \tilde{\xi}\_t + \varepsilon\_{it} \tag{8}$$

where Equations (6)–(8) are SDM, SAR, and POLS models, respectively. *GDPPC* is the explained variable GDP per capita, CORE is the core explanatory variable, *CONT* represents the control variables, *β* and *β"* are the coefficients of the variables, respectively. Other variables and symbols are consistent with the base model.

#### *3.3. Spatial Autocorrelation*

This paper constructs the spatial panel data of 41 cities in the YRD region from 2007 to 2019. It uses the spatial panel econometric analysis method to test the impact of forest resource abundance on the economic development level of the YRD region. This paper constructs geospatial weights based on latitude and longitude. The variables are re-estimated based on the spatial distances of all local points in the sample set to the target analysis point. The economic geospatial weighting matrix was not used, mainly because economic conditions change yearly. The distance between two points based on the latitude and longitude distance formula is expressed as follows. The element calculation of the spatial weight matrix is determined by three factors: spatial bandwidth, kernel function, and distance calculation formula. The spatial weight matrix is expressed as follows.

$$\mathcal{W}\_{\{i\}} = \begin{bmatrix} \mathcal{W}\_{\{1\}\_{h \to i}} & \cdots & 0 \\ \vdots & \ddots & \vdots \\ 0 & \cdots & \mathcal{W}\_{\{i\}\_{h \to i}} \end{bmatrix} \tag{9}$$

The calculation formula is: *<sup>W</sup>*{*i*}*h*→*<sup>i</sup>* <sup>=</sup> *<sup>f</sup> <sup>D</sup>*{*i*}*h*→*i*, *Bandwidth* . Where *<sup>D</sup>*{*i*}*h*→*<sup>i</sup>* is the distance from all data in the sample set to the local point *i*. *f(*·*)* is the kernel function. Bandwidth is the spatial bandwidth. One of the commonly used functions is the Gaussian function [54]. The specific representation is as follows.

$$\mathcal{W}\_{ij} = \exp\left[ -\frac{1}{2} \left( d\_{ij} / b \right)^2 \right] \tag{10}$$

where *Dij* represents the distance between the centroids of region *i* and region *j*. B represents bandwidth. The formula for calculating the distance between two points based on the latitude and longitude distance formula is as follows.

$$D\_{\{i\}\_{h\to i}} = r\_{\varepsilon} \times \arccos\left[\sin(v\_i \theta)\sin\left(v\_{\{i\}\_h}\theta\right) + \cos(v\_i \theta)\cos\left(v\_{\{i\}\_h}\theta\right)\cos\left(u\_i \theta - u\_{\{i\}\_h}\theta\right)\right] \tag{11}$$

where *ϑ* represents an empirical constant and *ϑ* = π/180. *re* represents the radius of the earth. *re* = 6378.1km.

Therefore, this paper uses the latitude and longitude data of 41 cities to generate a geospatial weight matrix calculated based on the Gaussian kernel function. According to the distance weight matrix, the first judgment is the spatial autocorrelation of the explained variables. Table 1 provides Moran's Index and Geary's C value of the explained variables from 2007 to 2019. The Moran indices were all significantly greater than 0 and significant at the 1% level. Geary's C is less than 1, proving a positive spatial autocorrelation relationship.


**Table 1.** Moran's I and Geary's C based on the distance weighting matrix.

#### *3.4. Data Sources*

This paper constructs an index system from the perspectives of resources, environment, society, economy, and system to verify the influence of the abundance of forest resources in the YRD on economic development (More details in Table 2). The indicators that involve price, such as GDP, per capita consumption level, etc., are deducted from the impact of price.

The raw data are from 2005–2019. Some missing data were filled up by interpolation. There are two main methods for dealing with missing data. One is to use the mean method, which is to take the average value of two adjacent years, such as the CONPC values of Jiaxing in 2014 and 2016; the other is to use the linear interpolation method, that is, the ARIMA imputation method. This method is mainly for the lack of forest resource coverage data in 2019. We then estimate the forest area data and divide it by the land area to calculate the forest cover rate. Missing data is less than 5%. In order to investigate the economic convergence effect and time lag effect of the research city, this paper uses the real GDP per capita data from 2005 to 2019 to generate the first-order difference lag variable of the real GDP per capita. Therefore, this paper uses the 2007–2019 data as the research sample. The descriptive statistics of the data are shown in Table 3.


**Table 2.** Data source and variables.

Note: SYB: Statistical yearbooks of provinces and cities over the years; JFB: Jiangsu Forestry Bureau; ZFRMC: Zhejiang Forest Resources Monitoring Center; CEADS: Carbon Emission Accounts & Datasets; CNKI: Platform of China National Knowledge Infrastructure; CUSY: China Urban Statistical Yearbook; DEEA: Department of Ecology and Environment of Anhui Province; WRB: Water Resources Bulletin; IFIND: Financial Data Center; SB: Statistical Bulletins of National Economic and Social Development of All Cities.

**Table 3.** Descriptive statistics of variables.


This paper uses the methods of variance inflation factor (VIF) and tolerance (TOL) to verify the multicollinearity of independent variables. The larger the VIF, the more severe the multicollinearity. Serious multicollinearity exists if the VIF is larger than 10. Tolerance is the inverse of VIF. A collinearity problem exists if TOL is less than 0.1. The VIF of the variables selected in this paper are below 5, and the average is 1.99 (Please see Table 4). In addition, the TOL of each indicator is above 0.1, indicating no multicollinearity.

**Table 4.** Multicollinearity test of variables.


### **4. Results**

#### *4.1. Direct Effects*

Table 5 shows the forest resource abundance has a U-shaped non-linear effect on regional economic development. The impact of forest resource abundance on the level of economic development is negative and significant at the 10% level. Still, the relationship between the square of forest coverage and per capita GDP is positive and significant at the 1% level. It shows that in the initial stage of economic development, the abundance of forest resources negatively impacts economic growth. The main reason is that in the early stage of economic development, cities with rich forest resources often have inconvenient transportation and a single industry. The more abundant forest resources, the greater the impact on transportation and arable land. Therefore, the abundance of forest resources will inhibit economic development. However, when the economy develops to a certain extent, abundant forest resources can promote economic level improvement. This is mainly because with technology advancement, improving living standards, and people's changing ideologies, developing the advanced service industry will be faster. Forest resources can provide better spiritual, cultural, and related products and services and improve the ecological environment. The impact of forest resources on economic development mainly includes direct, indirect, and induced effects. The direct effect is reflected in the jobs created by the forest sector, which produces indirect and induced economic added value in other sectors [14].

The first-order lag term of the explained variable is significantly negative, indicating that the phenomenon of economic convergence exists in the YRD region. If the initial state of the economy is relatively underdeveloped, there will be more room for improvement in economic development [55]. This aspect is particularly evident in Anhui Province, which is relatively underdeveloped in the YRD region. Moreover, with the Nanjing metropolitan area's development strategy, Anhui's economic development is relatively fast.


**Table 5.** Spatial econometric estimation results.

Note: \*\*\* significant at 1% level; \*\* significant at 5% level; \* significant at 10% level; T statistic in brackets.

In terms of carbon sequestration, the impact of carbon sequestration per GDP on the level of economic development is negative and significant. Moreover, carbon sequestration has a certain social cost [56]. Economic development needs to consume a lot of resources and energy, and the production process will cause more carbon dioxide emissions. However, carbon dioxide can be absorbed through carbon sinks such as vegetation and land. The more carbon emissions, the more pollution it brings, which may ultimately inhibit regional economic development. Especially in the Anhui area, it is necessary to do a good job in the strategic layout of the low-carbon economy and circular economy development while undertaking the industrial transfer. In addition, in terms of carbon sequestration methods, there are generally two ways to increase carbon sequestration, one is by technology, and the other is by afforestation. Presently, carbon sequestration through scientific and technological means is costly. However, carbon sequestration through afforestation is currently the most economical and reliable method. Notwithstanding, large-scale afforestation may generate a lot of opportunity costs, such as land use and forestry industry development.

In terms of resources and the environment, the impact of industrial sulfur dioxide emissions per GDP on the level of economic development is positive, but the results are not significant. In addition, the impact of water resource carrying capacity on the level of economic development is negative. The more developed the economic level is, the greater the demand for resources will be and the weaker the resource-carrying capacity will be.

The impact of electricity consumption per GDP on economic development is significantly negative. Electricity consumption can reflect the economic development level of a region. However, electricity consumption and industrial structure are closely related. Generally, the demand for electricity consumption in primary and tertiary industries is not as large as that of the secondary industry. Although Jiangsu is a large manufacturing province, other industries such as finance, technology, and services are also very developed. In 2020, Shanghai's GDP ranked among the top ten in the country, although Shanghai's economy is relatively developed. Electricity consumption in Shanghai ranks low due to the developed tertiary industry, especially the financial industry. Industrial structure often determines electricity use, but high-tech industries typically generate much more GDP with less energy than low-end manufacturing. Especially after the implementation of power integration development, cities in the YRD gradually developed a mode of sharing and interoperability. The efficiency of energy utilization in economic development is getting higher and higher.

The YRD region is committed to industrial restructuring, optimization, and upgrading. The industry shows a trend of cluster development and can realize the integrated development of modern service and advanced manufacturing industries.

The impact of urbanization on economic development is significantly positive. Improving the urbanization level will help to improve the level of economic development. Urbanization is essentially a process of agglomeration of manpower, capital, and resources. It is also one of the important driving forces for upgrading and transforming the economic structure. The first step is the transformation of the population; that is, the rural population is transformed into an urban population and participates in non-agricultural production activities. The spatial transformation of the population residence was gradually realized after the population transformation. The transfer of rural surplus labor and the construction of cities and towns will gradually promote the development of local enterprises, thereby promoting the continuous improvement of the local economy. However, the way that accompanies economic growth is often extensive. With the continuous improvement of urbanization, higher requirements are put forward on land, culture, society, and other aspects. Ultimately, urbanization will gradually achieve coordinated development with the economy, and economic growth will gradually transit from an extensive development model to an intensive growth model.

Investment in scientific research and education has an obvious role in promoting economic development, and the estimated coefficient of the model is above 0.2. It shows that scientific and technological innovation is very important to realize the economic development of the YRD region.

Government intervention has suppressed the economic development in YRD, but the results are not significant. China's economy has adopted a "government-led market economy" for a long time. The government intervenes deeply in the process of regional economic growth. It suppresses the role of the regional market mechanism, which may lead to imbalances in the economic structure and cannot be adjusted in time. In addition, government intervention may also bring about mistakes in decision-making and improper resources allocation. This will make the economic micro-subjects lose the initiative and vitality of economic activities. Moreover, excessive government intervention can easily lead to rent-seeking. Under modern economic conditions, the government's main role should carry out reasonable macro-intervention and moderate guidance on economic activities through laws and regulations. However, excessive intervention may inhibit economic development.

Foreign direct investment positively impacts economic development. The more foreign direct investment, the higher the level of economic development. First, foreign investment has brought about the application and promotion of advanced green technologies and improved the production efficiency of enterprises. Second, foreign investment will also relatively impact the overall corporate environment positively through production scale expansion, industrial structure adjustment, and talent introduction. Third, by absorbing FDI, the YRD region can accelerate the accumulation of regional capital, accelerate capital formation, and further improve investment level. Meanwhile, it can promote the level of employment in the region.

In terms of consumption, the model results show that the per capita total retail sales of consumer goods positively impact economic development. The average retail sales of social consumption are generally affected by income level, price level, and consumption environment. Only with economic development, continuous improvement of residents' income level, stable price level, and the good consumption environment can the growth of total retail sales of social consumer goods be stimulated. Conversely, the growth of consumer demand will also play a direct and final decisive role in economic growth. Investment demand and aggregate demand depend on consumption demand to some extent. From a medium and long-term perspective, only investment supported by consumer demand is effective, and effective investment and consumption make a greater contribution to economic growth.

#### *4.2. Effects of Decomposition*

The effects of spatial econometric models can be divided into direct effects, indirect effects (spatial effects), total effects, and feedback effects. The direct effect is the impact of an independent variable in a certain region on the dependent variable. The feedback effect is the direct effect coefficient minus the regression coefficient of the estimated result. The feedback effect means that the explanatory variables in a certain area will impact the explained variables in the surrounding area, which will affect the explained variables in the local area. Indirect effects are the effects of an explanatory variable on other regions, which is the influence of an explanatory variable in a neighboring area on the explained variable in the local area. The total effect is the sum of the direct and indirect effects.

Table 6 is calculated according to the spatial Durbin model of GDP per capita as the explained variable. The results show that the magnitude and significance of the coefficients of the direct effect are almost consistent with the coefficients and significance of the model estimates. The relationship between forest resource abundance and economic growth always maintains a U-shaped non-linear relationship. Forest resources can inhibit the initial stage of economic development. However, when the economy develops to a certain level, the abundance of forest resources and economic growth will develop together. Moreover, it has a positive pulling effect on the economic development of the YRD region.

Table 6 shows that the spatial effect exists. The impact of the abundance of forest resources in neighboring cities on the region's economic development presents a nonlinear inverted U-shaped trend. In the initial state, the more abundant forest resources are in neighboring cities, the more the level of economic development in the region will be promoted. When it reaches a certain level, the excessively abundant forest resources in neighboring regions will inhibit the level of economic development in the region. One reason is that the growth of trees requires material conditions such as land. Efforts to massively increase vegetation cover are increasing to alleviate climate conditions and achieve the grand vision of carbon neutrality. Excessive expansion of forest areas may seriously damage biodiversity. Cutting down old forests and planting new ones may break the original ecological balance. This will eventually crowd out the production and living resources of the region and affect economic development.

From the perspective of the feedback effect, the relationship between the abundance of forest resources of the surrounding areas and the region's economic development again shows a non-linear U-shaped trend. It means that the direct consumption of forest resources has different impacts at different stages of economic development. With the continuous economic improvement, the sustainable development of forest resources has gradually been paid attention to. The ecological, economic, and social values of the forest have been continuously excavated. For example, after the Three Plenary Sessions of the Eleventh Session, Zhejiang entered a period of revitalization and restoration of forestry, which has been vigorously developed. While striving to develop the economy, efforts should be made to realize the sustainable utilization of forest resources. The YRD region gradually realizes the coordinated development of the environment and economy and finally promotes the sustainable development of the economy.


**Table 6.** Decomposition of direct effects, indirect effects, and total effects.

Note: \*\*\* significant at the 1% level; \*\* significant at the 5% level; \* significant at the 10% level; T statistics in parentheses.
