Research on the Evaluation of Enterprises’ Green Growth Efficiency Based on DEMATEL-DEA

2.1 The Model of DEMATE^[13]

Bastille National Laboratory of American created a tool using graph theory and matrices to solve complex and difficult problems in the real world in 1971. The traditional DEMATEL method uses expert scoring to judge the effect of the relationship between the various influencing factors. However, if there are too many factors, the expert cannot easily find the relationship of various factors, and the scoring results might include error. On this basis, this paper firstly uses the G1 method to determine the weight of each index and then to determine the degree of influence index between the ratio of index weight. The specific application processes are as follows:

1) Determine the factors of the system

First, this paper analyses the probable input and output indexes of enterprise green growth efficiency according based on the research content, and the factors are as follows:

2) Establish the direct impact matrix

The direct impact matrix represents the relationship of direct influence of each factor and sets up the direct impact matrix X = (GGOI_ij)_n×n. It then calculates the direct impact matrix according to the weight of G1 of each factor,

GGOI_ii = 0 and GGOIij=wijwij′represent the ratio of the weight between one index and the other.

3) Establish the comprehensive impact matrix

After the direct impact matrix X is standardized, we obtain the impact matrix of standardization Y. The method of standardization is to take the maximum value of all elements in each row of the matrix and then divide the direct impact matrix by the maximum value. We obtain the direct impact matrix of standardization Y through

Matrix Y is standardized and dealt with in the following way

E indicates a unit matrix, and (E – Y)^-1is the inverse matrix of (E – Y).

4) Factor’s calculation results

Calculate the impact degree D_i, impacted degree R_i, centrality degree M_i and reason degree U_i by the comprehensive impact matrix,

D_i is the summation of the elements of each row of the comprehensive impact matrix, R_i is the summation of elements of each column of the comprehensive impact matrix, the centrality degree M_i= D_i + Ri, and the cause degree U_i = D_i – R_i.

2.2 The Model of Super Efficiency DEA

The DEA method, which is a CCR model, as an efficiency measurement method was first used by Chames et al^[14] under the assumption of fixed returns to scale. It obtained efficiency according to multiple sets of input and output dates using the principle of mathematical programming. Banker et al.^[15] proposed a more stringent correction model called BCC mode. It decomposed the technical efficiency of CCR model into the pure technical efficiency and the scale efficiency, namely, technical efficiency = scale efficiency × pure technical efficiency. Then, the BCC model separated two reasons that caused technical invalidity (not in the best size and production technical inefficiency). The pure technical efficiency is more accurately reflected by the efficiency of the management level of the inspection object than the technology efficiency of the CCR mode in the case of exclusion of scale factors^[16]; therefore, this study selects the BCC mode of the DEA model.

The DEA model can be divided into input oriented and output oriented. The former makes a minimum planning investment under a certain level of output; the latter makes a maximum planning investment under a certain level of input. In essence, they solve the same problem from a different perspective, and the final conclusion is identical^[17]. Because the input variable is the basic variable of implementing the process of green growth, and it is easier to control compared to the efficiency of output, this study is based on the input-oriented DEA model.

When using the traditional DEA model to measure the efficiency evaluation unit, the result may include multiple evaluation units at the same time in the production frontier situation and may be relatively effective, so that the evaluation unit that is relatively effective cannot distinguish the difference in the efficiency value and affect the comparative differences among the evaluation units. To solve this problem, Fare et al.^[18] established the super efficiency DEA on the basis of the traditional DEA model to perform the comparative analysis of the difference in efficiency values among the relatively low values of effective evaluation units. The basic idea is as follows: When measuring an evaluation unit, the first step is to exclude it from the evaluation unit set. In Figure 1 and Figure 2, when measuring the efficiency value of evaluation unit A, it should be excluded from the participation set of the evaluation unit. The original frontier of A turns from SABS’ to SCBS’, the efficiency value of A is θ = OA'/OA > 1, and the production frontier of inefficiency evaluation unit C’ is SABS’. To effectively distinguish the difference in the evaluation unit, the efficiency value is 1. Therefore, this paper evaluates the efficiency of the tourist industry of 31 regions in China based on the supper efficiency DEA model.

Assume that there are K evaluated areas; each region has N types of input factors of enterprises green growth and M types of outputs from the input of enterprise green growth. The input set and output set of each region are expressed as: x_k = (x_1k, x_2k, …, x_Nk) y_k = (y_1k, y_2k, …, y_Nk), and the variable return to scale of the super efficiency DEA model based on the oriented input is expressed as: x_k = (x_1k, x_2k, …, x_Nk) y_k = (y_1k, y_2k, …, y_Nk), and the variable return to scale of the super efficiency DEA model based on the oriented input is expressed as:

3.1 The Selection of Variables and Data Sources

Pollutants generated from industrial enterprises in the development process account for a large proportion in all categories of enterprises. The pollution generated from the above designated size industrial enterprises has brought extremely limited economic development. The research objects of this paper are large scale industrial enterprises in China, and the green growth efficiency of the large scale industrial enterprises is evaluated. Considering that there are differences in regional economic and geographic conditions and because the input and output efficiencies of the green growth mode that the enterprises adopt are various in different regions, this paper takes 31 regions as the decision making units of the efficiency evaluation.

According to the literature analysis and the related statistical data, the efficiency of the input indexes of enterprise green growth include the total investment in environmental pollution control, environmental pollution control investment share of GDP, the environmental investment of completing the environmental acceptance projects this year, the urban environment infrastructure investment, the number of environmental acceptance projects this year, completion of investment of industrial pollution control, sources of forestry investment, and completion of the forestry investment. The output indexes include the main business income, total assets, fixed assets, owners’ equity, total of current assets, main business taxes, total profits and other indexes. Because the DEA model limits the number of input and output indexes, this paper analyzes the input and output indexes on the basis of considering the impact degree, impacted degree, centrality degree and reasons degree of the various indexes by the DEMATEL method. Finally, it takes the total investment in environmental pollution control, the urban environment infrastructure investment, completion of investment of industrial pollution control, and completion of the forestry investment as input indexes of enterprise green growth efficiency and the main business income of the industrial enterprises and the total of the current assets as the output indexes, which are represented with the letters of EI, EII, IPI, FI, II, and CA. The unit of the six variables is billion yuan.

The policies and the degree of attention that enterprises paid to green growth are different in China in different periods of time, changing with the varying green growth efficiency. To determine the change in enterprises’ green growth from the perspective of time, this paper takes 2003, 2007 and 2012 as the research time points. Research data come from the “China Environmental Statistical Yearbook”, “China Statistical Yearbook”, the national economic and social development statistical bulletin and government website. The characteristics of the variable data are shown in Table 1.

3.2 The Result of Enterprise Green Growth Efficiency

On the basis of the super efficiency DEA model of investment and the variable returns to scale, this paper calculates the enterprises green growth efficiency in different regions in China with the DEAP software (Version 2.1) from the Coelli group design in 2003, 2007 and 2012, which takes the summation of the total amount of investment in environmental pollution control, investment in urban environmental infrastructure construction, the completion of the investment in industrial pollution control and the completion of forestry investment as the input variables and the main business income of standard-sized industrial enterprises and liquid assets as the output variables. The results are shown in Table 2.

3.3 Difference Analysis of the Overall Efficiency of Enterprise Green Growth

As seen from Table 2, nationally, the overall efficiency of enterprise green growth showed a downward trend, decreasing from 0.825 in 2003 to 0.712 in 2012. The average efficiency was 0.737 for 10 years. The level of the enterprise green growth efficiency needs to be further improved. The values of enterprise green growth efficiency changed significantly over the ten years. The highest efficiency reached 0.825 in 2003, whereas the lowest efficiency reached 0.674 in 2007. A possible reason for this trend was that in 2003, China was in the “Tenth Five-year” period of development. During this period, the speed of economic development was relatively slow and the negative impact of enterprise development was smaller, but in 2007, China was in the “Eleventh Five-year” period of development, which emphasized rapid economic growth and neglected the governance of environmental issues. On December 25, 2009, the State Council held an executive meeting and decided that the carbon dioxide emissions per unit of GDP should be decreased 40%~45% by 2020 compared to 2005. During the two sessions in 2010, the low-carbon economy was discussed as an important issue, and energy saving and green growth gradually became the core concerns of the Chinese government. In the planning of the “Twelfth Five-year”, the goal was set that energy consumption per unit of GDP should be reduced 16% and carbon dioxide emissions per unit of GDP should be reduced 17%. The economic growth rate by “Eleventh Five-year” of 7.5% was adjusted to 7%, and the development of strategic emerging industries was greatly advocated during the “Twelfth Five-year”. This series of national policy guidance had an effect on the variation of enterprise green growth efficiency.

3.4 Difference Analysis of the Regional Efficiency of Enterprise Green Growth Efficiency

1) The eastern region: The region includes ten provinces, and its mean enterprise green growth efficiency was 1.064, higher than the nationwide average level. This level was the highest nationwide. For enterprise green growth efficiency, the overall trend is downward, but in Beijing, Tianjin, Hebei and Jiangsu, it increased, especially in Beijing. There was a clear increasing trend in Liaoning, Beijing, Shanghai, Zhejiang and Guangdong, with values higher than the mean efficiency in the east. The average value of green growth efficiency for other enterprises was lower than the mean of the east. In Hebei and Shandong, it was also lower than the national average. These phenomena indicate that the enterprise green growth efficiency in the east had a higher status due to the contribution of Beijing, Shanghai, Zhejiang and Guangdong. From the point of view of pure technical efficiency, in the east, enterprise green growth was increasing. Namely, the technical capabilities of eastern enterprises were gradual increasing, which will help the enterprises to enhance the green growth efficiency in the future. However, the eastern enterprises green growth scale efficiency was declining, especially in Jiangsu, Zhejiang and Shandong, where the scale efficiency significantly decreased, indicating that the three areas should have properly controlled enterprise operating scales to improve the green growth efficiency.

2) The central region: The region includes six provinces, and its mean enterprise green growth efficiency was 0.658, lower than the nationwide average level. This level was the third highest nationwide. Enterprise green growth efficiency was in decline. In Jiangxi and Hunan, it was rising and reached greater than 1 in 2012. In Anhui, it declined slightly, and in Shanxi, Hubei and Hunan, it declined significantly. However, the efficiency of Hunan was in an effective state in 2003. The pure technical efficiency and scale efficiency of the central enterprise green growth both tended to decrease, but the scale efficiency declining trend was smaller. From the perspective of returns to scale, the scale benefit of Shanxi, Anhui, Hubei and Hunan were all in incremental stages; namely, enterprises should enhance the green growth efficiency through appropriately expanding the operation scale.

3) The western region: The region includes twelve provinces, its mean enterprise green growth efficiency was 0.517, lower than the nationwide average level, and this level was the lowest nationwide. Enterprise green growth efficiency was in obvious decline overall. Chongqing, Sichuan, Ningxia and Xinjiang were increasing, but the other provinces showed a downward trend. Guizhou, Yunnan, Tibet, Gansu and Qinghai declined significantly, but the enterprise green growth efficiency of Tibet and Qinghai were in an effective state in 2003. The pure technical efficiency and scale efficiency of enterprise green growth were both in decline in the western region, indicating that the western enterprises had a high potential for efficiency gains from a technical aspect. As seen from the returns to scale, the western enterprises scale benefit was in incremental stages, namely, enterprises should enhance the green growth efficiency through appropriate upgrading of the operating scale.

4) The Northeastern region: The region includes three provinces, its mean enterprise green growth efficiency was 0.681, lower than the nationwide average, and this level was the second highest nationwide. It had a distinct gap with the eastern region. Enterprise green growth efficiency was on the rise overall, but in Jilin, it declined significantly. In Jilin, the mean enterprise green growth efficiency was the highest of the northeast region. In Liaoning and Heilongjiang, it was lower than the mean efficiency in the northeast. In the Northeast, the pure technical efficiency of enterprise green growth showed a clear upward trend, but the scale efficiency was in obvious decline. Jilin and Heilongjiang had the potential to expand the scale of operation to improve green growth efficiency.