In this section, we analyze the empirical results of the influencing factors of the environmental efficiency. First, the manufacturing industries are classified according to the pollution intensity. Second, we describe the data and propose the hypotheses. Third, we utilize a two-stage approach to analyze the efficiency. In the first stage, we use the SBM-DEA model to measure environmental efficiency and economic efficiency of manufacturing in China. The score falls between 0 and 1. In the second stage, we use the Tobit regression model to analyze the driving factors which influence the efficiency.
3.1. Manufacturing Classification
In this section, we establish a pollution intensity index to measure the pollution extent of manufacturing industries in China. Thus, 29 manufacturing industries are classified into three groups mainly: heavily polluted industries, moderately polluted industries, and lightly polluted industries. The pollution extent of manufacturing industries has a different effect on environment due to the diverse characteristics. Therefore, it is difficult to use for analyzing the effect of determinants on environmental efficiency from the perspective of overall manufacturing industries only. In order to make the research more concrete, it is necessary to understand the pollution condition of subdivided manufacturing industries.
Industrial pollution mainly includes the emission of wastewater, waste gas, and solid waste. However, Zaim [
37] pointed out that air pollution is a main byproduct of the manufacturing activity process. Thus, waste gas is used as a proxy for industrial wastes to participate in the following calculation. Higher pollution intensity index means that manufacturing pollutes the environment more seriously.
(1) We define the pollutant emission intensity as follows:
where
q is the serial number of the manufacturing industries,
PI is the pollution emission intensity of each manufacturing industry,
PEq denotes industrial wastes and we use total volume of industrial waste gas emission as proxy,
GVq denotes gross industrial output value.
(2) The indices mentioned above are converted into proper dimensionless ones [
38] whose value ranges from 0 to 1:
where Max
(PIq) and Min
(PIq) denote the maximum and minimum values of
PIq, and
PIIq is the pollution emission intensity index which is used to classify the manufacturing industries. The specific classification standard is shown in
Table 1.
3.3. Environmental Efficiency Measurement of Manufacturing in China
We analyze the efficiency in two steps in this section. First, we calculate the manufacturing environmental efficiency using SBM-DEA model, considering undesirable output (volume of industrial waste gas); Second, we compute only economic efficiency taking no account of undesirable output. According to the efficiency scores we have obtained, we analyze the effect of the environmental factors on the economic performance of manufacturing industries.
Table 3 presents environmental efficiency scores considering undesirable output. These scores are used as the independent variables in the Tobit regression model. In
Table 3, most of the environmental efficiency scores of manufacturing in China rise steadily from 2006 to 2011. It shows that in recent years, the development of manufacturing industry in China is in good condition.
As the results shown in
Table 3 cannot reflect the effects of environmental factors on environmental efficiency evaluation, we calculate the pure economic efficiency scores by the SBM-DEA model mentioned above, without considering the environmental factors. We employ total assets, industrial labor, energy consumption, and expenditure on R&D as inputs, and gross industrial output value as the output. All data covers manufacturing in China from 2006 to 2011. The new calculation results are shown in
Table 4.
Table 4 indicates that without considering environmental factors, the efficiency results calculated by the SBM-DEA model can also reflect the gradual increasing of manufacturing efficiency during 2006 and 2011. It shows that the environmental factors do not change the overall development trends of manufacturing environmental efficiency.
In
Table 3 and
Table 4, whether consider environmental factors or not, most of China’s manufacturing industries’ efficiencies rise steadily. Although some industries fluctuate in separate years, the overall condition goes well.
In
Table 3, the environmental efficiency scores of Printing, Reproduction of Recording Media and Manufacture of Articles for Culture, Education and Sport Activity decrease significantly in
Table 4, while they perform well without considering environmental factor. This indicates that environmental factors cause the efficiency to decline in the two industries mentioned above. Although these two industries are lightly polluted, the development still has a negative effect on the environment. According to the data from the China Statistical Yearbooks on Environment, the waste gas emission of Printing, Reproduction of Recording Media increases sharply from 11 billion m
3 to 25.1 billion m
3 during 2010 and 2011 which is 2.28 times higher than the previous year. In addition, the total volume of industrial waste gas emission of Manufacture of Articles for Culture, Education and Sport Activity increases sharply from 2009 to 2011. The emission in 2009 is 2.37 times higher than the year before. While in 2011, the emission is more than four times as high as in the previous year. Thus, the environmental efficiency of Manufacture of Articles for Culture, Education and Sport Activity drops constantly, and these two industries are high environmental risks.
Additionally, the environmental efficiency scores of the Manufacture of Measuring Instrument, Machinery for Cultural and Office Work increase sharply from 0.42 to 1.00 during 2010 and 2011, while the economy efficiency scores drop less. According to the original data from China Statistical Yearbooks on Environment, the waste gas emission drops sharply from 55.5 billion m3 to 10.1 billion m3 during 2010 and 2011, with 81.8% down from the previous year. In 2011, the environment of Manufacture of Measuring Instrument, Machinery for Cultural and Office Work has improved significantly and environmental factors have great influence on that industry. Thus, its environmental efficiency enhanced greatly.
Table 3 and
Table 4 demonstrate that as an undesirable output, environmental factors have little influence on overall evaluation of environmental efficiency, but have great influence on some industries’ economic efficiencies.
However, from
Table 3 and
Table 4, we still cannot analyze the difference between environmental efficiency and economic efficiency of heavily polluted, moderately polluted, and lightly polluted industries. Thus, we will make a comparative study on different industries according to classification by pollution intensity. First, the heavily polluted industries are analyzed as follows:
Figure 1 demonstrates the comparison of average efficiency of manufacturing in six heavily polluted industries from 2006 to 2011. Industries which are classified as heavily polluted industries are mainly composed of pollution-intensive industries and traditional heavy chemical industries such as Manufacture and Processing of Ferrous Metals; Manufacture and Processing of Non-ferrous Metals and Paper Products and Manufacture of Chemical Raw Material and Chemical Products. This classification is basically in accordance with the conclusion of Copeland and Taylor [
56]. The main characteristics of such industries are with greater resource consumption and higher pollution intensity. The efficiency scores of heavily polluted industries are rather low and there is little difference between the scores with or without considering undesirable outputs. Additionally, the environmental efficiency scores are slightly higher than that without considering environmental factors, since environmental factors brings down efficiency scores as undesirable outputs. This result indicates that heavily polluted industries are basically inefficient ones. Input and output efficiency of heavily polluted industries needs to be improved as the coordination between economy and environment is poor and the environmental risk is still high.
Figure 2 demonstrates the comparison of average efficiency of manufacturing in 15 moderately polluted industries from 2006 to 2011. The industries which are classified as moderately polluted industries mainly consist of two parts: (1) traditional labor-intensive industries and light industries, such as Processing of Food from Agricultural Products; Manufacture of Foods and Manufacture of Textile; and (2) some heavy industries whose pollutant emission intensity are second only to heavy chemical industries because of low environmental technology content such as the Manufacture of Rubber; Manufacture of Plastic, Manufacture of Metal Products. The efficiency scores of most moderately polluted industries are rather low and there is little difference between the scores with or without considering undesirable output. Additionally, the environmental efficiency scores are slightly higher than that without considering environmental factors. This is because the environmental factors can also bring down the economic efficiency scores as an undesirable output. However, two special industries are Processing of Petroleum, Coking, Processing of Nucleus Fuel and Processing of Food from Agricultural Products. The efficiency scores of these two industries are apparently higher than other moderately polluted industries. This implies these two industries are of good environmental benignity and the environment governance of other moderately polluted industries still needs to be improved.
Figure 3 demonstrates the comparison of average efficiency of manufacturing in eight lightly polluted industries from 2006 to 2011. The industries which are classified as lightly polluted industries are mainly composed of high technology industries (i.e., Manufacture of Electrical Machinery and Equipment; Manufacture of Communication, Computer, Other Electronic Equipment and Manufacture of Artwork, Other Manufacture) and clean industries (i.e., Manufacture of Tobacco; Printing, Reproduction of Recording Media and Manufacture of Articles for Culture, Education and Sport Activity). The main characteristics of such industries are less resource consumption and pollutant emission. This classification is basically in accordance with Li and Tao [
57]. The efficiency scores of most lightly polluted industries are generally high. Only the efficiency of Printing, Reproduction of Recording Media industry is rather low, which implies its production characteristics bring about serious environmental problem and its environmental condition needs to be improved. Additionally,
Figure 3 shows that different from heavily polluted and moderately polluted industries, more than half of the lightly polluted industries’ environmental efficiency scores are higher than the economic ones which implies these industries’ environmental conditions have been greatly improved during the production process. Thus, the environmental and the economic condition of these industries can coordinate well.
Figure 4 shows that the average level of total environmental efficiency of manufacturing in China from 2006 to 2011. In
Figure 4, we can see that the efficiency scores of three types of industries are rising in the same trend. Among them, the economic efficiency scores of heavily polluted industries and the moderately polluted industries are higher than the economic-environmental ones. This implies that the environmental condition of these two types of industries still needs to be improved. Only the lightly polluted industries’ environmental efficiency scores are higher than the pure economic ones. It implies that a balanced development of economy and environment has been achieved in these industries. However, the overall economic efficiency scores are still higher than the economic-environmental ones which implies that the environmental condition of manufacturing in China still needs to be improved.
3.4. Tobit Regressions Analysis
Since the environmental efficiency scores of manufacturing in China are calculated by help of SBM-DEA model, we can assess the effects of different determinants on these efficiency scores. To this aim, a Tobit model is specified.
Table 5 presents the estimation results.
Table 5 reports the Tobit regression results of the overall environmental efficiency. At the industrial level, the variable of the openness degree defined as percentage of industrial export value (EV) is a significant positive determinant of environmental efficiency at the 10% significance level, as theoretically expected. The variable of the industry scale defined as percentage of industrial investment in fixed assets (IFA) has a statistically significant, but negative coefficient at the 1% significance level, which does not conform to the hypothesis. We consider the reason for the results mentioned above is that industries with large-sized, low-efficiency, highly polluted, and high-energy consumption account for a large proportion of manufacturing in China. The variable of the energy structure defined as the percentage of industrial coal consumption (CC) is insignificant which does not conform to the hypothesis. This anomaly likely arises from the improvement of energy over-exploitation in recent years. Meanwhile, manufacturing industries strengthen the environmental governance to reduce the polluting emission, but with little influence. Thus, energy consumption has an insignificant negative impact on the environment. This implies that the energy structure and environmental governance of manufacturing in China still need to be improved. However, the variable of the technological development level defined as a percentage of industrial invention patents (IP) has a negative and insignificant coefficient on environmental efficiency. In order to find out the reason for the phenomenon, we tested the correlation of the variables. It turns out that only the technology development level and the openness degree are highly correlated with the coefficient of 0.97. This implies that the technology development level is substituted by the openness degree to a great extent. The variable of profitability defined as the ratio of total profits to revenue from principle business is insignificant for overall industries and other three types of industries. The main reason for this condition is that good pollution control cannot bring the company a high stock valuation [
58]. Some scholars support our empirical findings. They believe there is no significant link between profitability and environment performance [
58,
59].
For the lightly polluted industries, the industry scale has a positive effect on the economic-environmental efficiencies as expected since most lightly polluted industries are high technology and clean industries. Additionally, the energy structure has a negative effect on the efficiency scores, which also conforms to the hypothesis. It implies that the energy structure is unreasonable, which needs to be improved at this level.
As for the moderately polluted industries, the coefficients of the openness degree and the industry scale are insignificant since most of the moderately polluted industries are traditional labor-intensive industries with poor economic and environmental benefits generally. The energy structure has a significant positivity effect on the efficiency scores. It implies that the energy structure is reasonable and it can improve the environment at this level. Additionally, the technological development level has a significant negative effect on environmental efficiency. This demonstrates that technology development decreases the economic-environmental efficiencies at this level.
For the heavily polluted industries, only industry scale has a significant negative effect on environmental efficiency, as well as for all industries mentioned above. This implies that most highly polluted industries belong to industries with large-sized, low-efficiency, and high energy consumption. As for other influence factors, they are insignificant.
According to the Tobit regression model, the openness degree has a positive significant impact on environmental efficiency of manufacturing industries, overall. The industry scale is a significant determinant for all kinds of industries, except for the moderately polluted industries which influence the overall and heavily polluted industries negatively, and influence the lightly polluted industries positively. The energy structure is a significant determinant of environmental efficiency for lightly polluted and moderately polluted industries, which lose its explanatory power for overall industries and heavily polluted industries. It influences the efficiency scores of moderately polluted industries positively, while influencing lightly polluted industries negatively.