4.2.3. Model Selection Test

There are two methods for model regression, i.e., the fixed-effects model and the random-effects model; Hausman's test is required to determine the choice of model. The original hypothesis model was selected as the random effects model, and the Hausman test results (in Table 5) were obtained through Eview 6.0.

#### **Table 5.** Hausman test results.


From the analysis results, the *p* value is 0, so the original hypothesis is rejected, i.e., the original random effects model hypothesis is rejected, and the fixed effects model is adopted

#### 4.2.4. Model Regression Results

According to the regression model and test method, the regressions were conducted in China as a whole, and the east, central and west to explore the effects of the same influencing factors on different regions; this is because regions face various differences in their development environment, their geography and their development level which are heterogeneous. The results can be seen in Table 6.


**Table 6.** Regression results.

Note: \*, \*\*, \*\*\* means *p* < 0.1, *p* < 0.05 and *p* < 0.01, respectively.

The regression coefficients of market-based environmental regulation are significantly positive in the country, in the eastern and central regions and significantly negative in the west. Due to the significant differences in the marketization process between regions, some developed provinces in the east and central regions, most of which have entered the post-industrialization period, have more sound market mechanisms and a higher degree of marketization; they rely mainly on market incentives to solve environmental problems. The western region has relatively low environmental costs, and in order to promote regional economic development, it takes over some of the low-end forestry industries transferred from home and abroad. Some forestry enterprises take the payment of sewage charges as an excuse to continue polluting, and the expenditure on pollution control follows the marginal decreasing cost: the sewage charges paid are far from enough to cover the costs of environmental treatment and ecological restoration.

As a whole, investment in scientific research will promote FECO, but from the empirical perspective, the current impact coefficient of research funding input intensity on the eco-efficiency of the forestry industry in China is negative and the correlation is significant. This indicates that current investment in scientific research in China fails to meet the needs of the forestry industry to improve eco-efficiency. To a certain extent this hinders the development of forestry eco-safety, as well as the subsequent need to further improve the technological content and technological level of the forestry industry to make the development of modern forestry go in the direction of the circular economy and eco-efficiency improvement. The intensity of scientific research investment in the eastern region shows a significant promotion effect, which indicates that forestry industry development in the eastern region tends to be intensive; the eastern economy is developed, and sufficient investment in scientific research can meet the needs of regional forestry development, thus showing a promotion effect. Forestry industry development in the central and western regions is still based on the scale of rough growth, rather than high-tech orientation, which in the long run is important for forestry industry development and the improvement of FECO. However, the elasticity coefficient of central and western China is negative but not significant, indicating that the negative impact is not significant.

Economic development has a positive but insignificant effect on FECO. Forestry plantations and the forest industry itself are a more complete industrial chain. The higher the level of economic development, the more likely it is to make use of industrial policies, extend the industrial chain, acquire advanced resources such as technology and increase productivity. Forestry cannot be separated from forests, which have long regeneration cycles and extremely uneven distribution, and are also constrained by specific factors such as land. The higher the level of economic development, the stronger the demand for the multifunctional use of forests, i.e., the more wood, carbon sink and tourism use. The more prominent the multifunctional problems are, the more obvious the problem of wood resource supply constraints faced by forestry development. Therefore, even though the degree of economic development is high, it is difficult to obtain an effective allocation of capital and technology in the case of shortage of timber resources, which in turn restricts the improvement of total factor productivity. However, there are large differences among regions: the eastern region shows a positive correlation, indicating that the eastern region is more mature in the allocation of resources and technology and is qualitatively better than

the central and western regions; hence, in the east it is significant to enhance eco-efficiency, while it is not significant in the central and western regions.

The industrial structure inhibits FECO. This means that the FECO of the regions with a high proportion of secondary forestry industry in China is relatively low at present, i.e., the ecological and environmental costs of increased output value in the forestry industry are large. Specifically, the negative effect of the eastern region is more significant, which may be due to the fact that the paper industry, as the main component of the forest industry, is mostly concentrated in the eastern region. The central region is the main concentration of China's wood processing industry, probably because it is a lightly polluting industry, so the impact of the industrial structure is less than in the east; the impact of industrial structure on FECO in the western region is not significant. However, with further optimization of the three industrial structures, it is expected that FECO will be gradually improved in the future as the proportion of tertiary industries in forestry increases.

The impact of external openness on regional FECOs varies greatly among regions. The elasticity coefficients of foreign direct investment in the eastern, central and western regions were 0.0345, −0.1235 and −0.1897, respectively, and all regions passed the significance level test except for the central region, which did not pass the significance test. It can be seen that the environmental access threshold of the forestry industry in the eastern region is higher than other regions, and the quality requirements of foreign businessmen are higher, thus reflecting a positive effect. Foreign enterprises have a "pollution transfer effect" on the forestry industry in the central and western regions, i.e., the pollution-intensive forestry industry undertaken by these regions aggravates the regional environmental pollution problem. However, the negative correlation at the national level indicates that foreign direct investment is still dominated by the pollution transfer effect, but it does not pass the significance test, indicating that this negative effect is tending to be insignificant, and that FECO can be improved by improving the quality of foreign investment introduced.

#### *4.3. Discussion*

This paper adopts the super-efficient DEA model to measure the forestry eco-efficiency of 30 Chinese provinces and cities (except Hong Kong, Macao, Taiwan and Tibet) for 14 years from 2008 to 2021, and then introduces the Tobit model to analyze the influencing factors of forestry eco-efficiency in order to better understand the sustainable development level of forestry. Existing studies mainly focus on studying forestry production efficiency. For example, Xu et al. [49], Wei [50] and Tan et al. [51] all measured the regional forestry production efficiency in China based on the Malmquist-DEA model without considering the impact of forestry industry development on the environment, thus this paper incorporates environmental factors to study forestry eco-efficiency; this is complementary to existing studies. Different regions showed significant heterogeneity with large regional differences in this study, which is consistent with the conclusions reached by most scholars. For example, Zheng and Yin [20] concluded that the eastern region has significantly higher eco-efficiency values than other regions and has been at a high efficiency level of about 0.9, while the western and central regions have slowly increased eco-efficiency values to approximately the 0.6 level. Wu and Zhang [52] also found the same trend for forestry eco-efficiency.

However, there are some scholars who came to different conclusions. Chen et al. [18] and Hong et al. [19] concluded that the western region is higher than the central region, and the opposite conclusion exists with this paper. This is probably because the above scholars used the traditional DEA model in evaluating eco-efficiency and came to the result that the maximum is one, and the ones greater than the data are all one. This paper uses the super-efficient DEA model to break through the efficiency boundary of one and can be greater than one, which more accurately reflects the actual value of the results and provides a more accurate depiction of the problem. Meanwhile, this paper uses the Tobit model to verify the influence of some economic variables on forestry eco-efficiency; it analyzes the influence on forestry eco-efficiency from environmental regulation, marketization, research

funding, industrial structure and openness to the outside world, respectively, which are more variables than previous scholars had considered. The data used are also up-to-date, which can more accurately illustrate the influence relationship between variables. However, compared with other scholars, the research in this paper is dominated by linear relationship research, and does not explore the nonlinear relationship between variables; for example, Jiang et al. [53] verified the study of the threshold effect of environmental regulation on forestry eco-efficiency, which is the direction of future research for this paper, i.e., exploring nonlinear relationships between variables.

### **5. Conclusions and Implications**
