Inovation

Threshold effect of technological innovation on carbon emission intensity based on multi-source heterogeneous data

Calculation of carbon emission intensity

In order to understand the change in carbon emission intensity in various provinces and cities in China, the analysis is carried out from 2010 to 2020 as an example. Read Table 2 for details.

Table 2 The carbon emission intensity of provinces and cities in China from 2010 to 2020.

It can be seen from Table 2 that the carbon emission intensity in most regions of China has declined by 2020. The carbon emission intensity of 11 provinces and cities, including Beijing, Shanghai, Zhejiang, and Jiangsu, has dropped to below 1. The carbon emission intensity of 10 provinces and cities, including Tianjin, Anhui, and Shandong, is between 1 and 2. The carbon emission intensity of Hebei, Liaoning, Jilin, Heilongjiang, and Jiangxi is between 2 and 3. The carbon emission intensity of Shanxi, Inner Mongolia, Ningxia, and Xinjiang is higher than 3. The above significant changes show that the carbon emission intensity of provinces and cities in China’s region generally presents a spatial pattern of low in the south and high in the north, which is basically consistent with the current level of economic development among regions in China. The carbon emission intensity of Jiangsu, Zhejiang, Shanghai, and Beijing is lower than that of most central and western regions such as Liaoning and Hebei, which indicates that China’s carbon emissions have gradually decoupled from the degree of economic development, and the carbon emissions of economically developed regions are showing a downward trend. In terms of the time trend of regional carbon emission intensity, all regions have decreased to different degrees in the sample period, and there is a trend of further development to a lower level. It may be influenced by the economic development momentum at home and abroad in recent years, as well as the optimization and upgrading of China’s energy structure and industrial structure and the innovation and progress of low-carbon technology, the energy consumption of most provinces and cities in China has reached a peak inflection point. In addition, the in-depth implementation of energy saving and carbon reduction policies in various regions has greatly reduced the carbon emission intensity of all provinces and cities in China. At present, China’s provinces, cities, and regions have gradually embarked on a low-carbon and green development path, laying a solid foundation for achieving a carbon peak in 2030 and carbon neutrality in 2060.

Exploratory spatial data analysis

Global spatial autocorrelation analysis

Firstly, the spatial autocorrelation between the level of scientific and technological progress and the intensity of carbon emission is tested by the overall Moran’I value. The results are listed in Table 3.

Table 3 Global spatial autocorrelation analysis results.

Table 3 shows that the overall Moran’I of the level of scientific and technological and the intensity of carbon emission are highly significant. The Moran’I value of carbon emission intensity is between 0.21 and 0.30, and the degree of correlation has increased in recent years, indicating that there is a positive spatial correlation between the carbon emission intensity of each province. In other words, the carbon emission intensity of a province will have a positive impact on the carbon emission intensity of neighboring provinces, and this impact is becoming stronger. The Moran’I value of the level of scientific and technological progress is between 0.17 and 0.26. Although the level of scientific and technological innovation in each province shows a positive spatial autocorrelation, its positive impact reveals a downward trend year by year.

Local spatial autocorrelation analysis

The spatial aggregation of the level of scientific and technological progress and the intensity of carbon emissions in various provinces and cities is further explored through the Moran scatter map, and the results are shown in Figs. 1 and 2 respectively.

Figure 1
figure 1

Moran scatter chart of scientific and technological progress in 2010 and 2020 (Note: The corresponding relationship between figures and provinces and cities is: 1 Beijing, 2 Tianjin, 3 Hebei, 4 Shanxi, 5 Inner Mongolia, 6 Liaoning, 7 Jilin, 8 Heilongjiang, 9 Shanghai, 10 Jiangsu, 11 Zhejiang, 12 Anhui, 13 Fujian, 14 Jiangxi, 15 Shandong, 16 Henan, 17 Hubei, 18 Hunan, 19 Guangdong, 20 Guangxi, 21 Hainan, 22 Chongqing, 23 Sichuan, 24 Guizhou, 25 Yunnan, 26 Shaanxi, 27 Gansu, 28 Qinghai, 29 Ningxia, 30 Xinjiang. Figure 2 is the same).

Figure 2
figure 2

Moran scatter chart of carbon emission intensity in 2010 and 2020.

Firstly, it analyzes the spatial aggregation of the level of scientific and technological progress in each province and city. As is shown in Fig. 1, in 2010, most provinces and cities were in a low–high and low–low concentration state, with a few in a high–high concentration state and some in a high–low concentration state. Specifically, Jiangsu, Zhejiang, and other provinces are located in high–high concentration areas, indicating that the scientific and technological progress levels of these provinces and neighboring provinces are relatively developed. Guangdong is located in a high–low concentration area, whose level of scientific and technological progress is relatively high while that of neighboring provinces is relatively low. The level of scientific and technological progress in Qinghai, Ningxia, Xinjiang, and other regions is in a low–low concentration area, revealing that the level of scientific and technological progress in these provinces and their neighboring provinces is low. Jiangxi, Guangxi, and other provinces are in low–high concentration areas, meaning that the level of scientific and technological progress of these provinces is low while the level of scientific and technological progress of their neighboring provinces is high. In general, the spatial agglomeration state of provinces and cities has changed greatly in 2020. Most provinces and cities are in low–low agglomeration areas, a few in high–high agglomeration areas, and some in high–low agglomeration areas. Specifically, Jiangsu, Zhejiang, Shanghai, and other places are in high–high concentration areas; Guangdong is in high–low concentration areas; Xinjiang, Ningxia, and other places are in low–low concentration areas. This may owe to the developed economic level of Jiangsu, Zhejiang, and Shanghai as well as its relatively-concentrated education resources and colleges, and the policies issued by local governments which are conducive to the gathering of talent. In addition, the radiation effect of big cities makes the urban scientific and technological levels of Jiangsu, Zhejiang, and Shanghai reach the forefront of China. Xinjiang and Ningxia, located in the west of China, are far behind the eastern coastal areas in terms of economic development and college education. At the same time, the local favorable policies for talent are also not as good as those in the east, resulting in a low level of science and technology in Xinjiang, Ningxia, and surrounding cities.

Then it analyzes the spatial concentration of carbon emission intensity in each province and city. As can be seen from Fig. 2, in 2010, most provinces and cities were in a low–low and low–high concentration state, with a few in a high–low concentration state and some in a high–high concentration state. Specifically, Shanxi, Inner Mongolia, and other provinces are located in high–high concentration areas, that is, the carbon emission intensity of these provinces and their neighboring provinces is relatively high. Guizhou and Yunnan are located in high–low concentration areas whose carbon emission intensity is relatively high while that of neighboring provinces is relatively low. The carbon emission intensity of Beijing and Tianjin is in the low–low concentration area, indicating that the carbon emission intensity of these provinces and their neighboring provinces are low. Liaoning, Jilin, and other provinces are in low–high concentration areas, indicating that these provinces have low carbon emission intensity while their neighboring provinces have high carbon emission intensity. There is no significant change in 2020 overall, and locally, Guizhou moves to a low–low concentration area.

Spatial metrological analysis

Model selection

The Moran’I test shows that the impact of technological innovation on carbon emission intensity presents significant spatial autocorrelation. However, the Moran’I test can only test whether there is spatial autocorrelation between variables. The LM test can further test the existence of spatial autocorrelation. Therefore, the LM test is carried out on the panel data of technological innovation affecting carbon emission intensity, including LM-error test, Robust LM-error test, LM-lag test Robust LM-lag test Among them, the alternative hypothesis model of LM-error test is the spatial error autocorrelation and the spatial error moving average model. The alternative hypothesis model of LM-lag test is the spatial lag model. The Robust LM-error test and the Robust LM-lag test are amendments to the LM-error test and the LM-lag test when the model error occurs.

It can be seen from Table 4 that the p-value of the spatial error model is 0.60, which is not significant. The modified spatial error model has passed the significance level test of 5%, indicating that the residual estimated by the model has spatial autocorrelation. The p-value of the spatial lag model is 0.00, which is significant at the 1% level, so the spatial lag model is better than the spatial error model. Then Wald and LR tests were carried out, and both passed the 1% significance level test. The test results rejected that the spatial Durbin model would degenerate into a spatial error and the lag model, so the spatial Durbin model needs to be selected for quantitative analysis. In addition, the Hausman test results show that the statistical value of t is 24.35 and the p-value is 0.01, which is significant at the level of 5%, indicating that the original hypothesis that the random effect is superior to the fixed effect should be rejected, and the fixed effect model should be selected. With the estimation results of the spatial Durbin model compared and analyzed under the conditions of fixed space, fixed time, and double fixed time and space, as shown in Table 5, the R2 value and the log-likelihood of fit in the three fixed effects are both spatiotemporal double fixed coefficients are relatively large.

Table 4 Model inspection.
Table 5 Fixed effect coefficient of spatial Durbin model.

To sum up, the SDM model with the fixed effects is selected for the regression analysis, and the regression results of SEM model and SAR model are used as a reference for comparison. Read Table 6 for the results.

Table 6 .Regression results of the spatial econometric model.

It can be seen from Table 6 that under the SDM model with fixed effects, the coefficient of the technological innovation level is significantly negative at the level of 1%, which indicates that the technological innovation level has a strong inhibitory effect on the growth of carbon emission intensity. But will this relationship change due to different model choices? Therefore, the regression results of the SEM model and SAR model introduced in this paper are compared with the SDM model under the double-fixed effect. From Table 6, we can find that the coefficient of technological innovation level is also significantly negative under the SEM model and SAR model. In addition, through comparison, it is found that the positive and negative regression coefficients of each model are relatively consistent and the significant number difference is not large, so it is more reasonable to select the SDM model with a double fixed effect model.

Analysis of spatial spillover effect

The analysis of the spatial impact of the level of technological innovation on carbon emission intensity requires a partial differential decomposition of the SDM model to examine the direct and indirect effects of the variables. The specific direct and indirect effects under the Fixed effect SDM model are listed in Table 7.

Table 7 Direct and indirect effects under the Fixed effect SDM model.

It can be seen from Table 7 that the direct effect of technological innovation level on carbon emission intensity is significantly negative, indicating that the improvement of technological innovation level in a province has a significant inhibitory effect on local carbon emission intensity; the indirect effect is positive and the result is not significant, which indicates that the improvement of local technological innovation level has little impact on the carbon emission intensity of neighboring provinces and cities, but the overall effect is negative. Despite a rebound effect on carbon emissions concerning the level of technological innovation in the short term40, the intensity of carbon emissions is not only related to carbon emissions but also related to a certain level of economic development. Therefore, in the long run, the improvement of technological innovation level will still inhibit the increase of carbon emission intensity. The direct effect, indirect effect, and total effect of a national cultural level are significantly negative at the level of 1%, which indicates that the improvement of a national cultural level not only has a significant inhibition effect on local carbon emission intensity but also inhibits the carbon emission intensity of neighboring provinces and cities. In October 2022, the Chinese government proposed that promoting green and low-carbon economic and social development is the key link to achieving high-quality development. The improvement of the national cultural level can promote national comprehensive quality and environmental awareness, and better implement the initiatives put forward by the Chinese government in life, so as to consciously advocate low-carbon life41. The direct effect of R&D investment intensity on carbon emission intensity is significantly negative, indicating that the improvement of local R&D investment intensity will have a restraining effect on carbon emission intensity to a certain extent. The indirect effect and total effect are not significant enough, indicating that the intensity of local R&D investment only has an impact on the intensity of carbon emissions in the province and city. According to the report on China’s investment in science and technology in 2021, China invested a total of 2795.63 billion yuan in research and experimental development, an increase of 356.32 billion yuan over the previous year, with an annual increase of 14.6%42. The intensity of R&D investment is the key factor to improve the level of scientific and technological innovation, so the increase of R&D investment intensity will indirectly inhibit the intensity of carbon emissions. The direct and total effects of urbanization level are significantly negative at the level of 1%, while the indirect effects are not significant enough, which shows that the improvement of urbanization level in a province has a significant inhibitory effect on the carbon emission intensity of the province, but has little impact on the neighboring provinces and cities. From 2010 to 2020, China’s urbanization rate rose from 49.68 to 63.89%. The acceleration of urbanization will lead to the expansion of the existing city scale and the emergence of a number of new cities. Accordingly, the demand for energy-intensive industries such as steel and cement will also be more vigorous, which will inevitably lead to a significant increase in China’s carbon emissions43,44. However, with the increase in urban population and the explosion of energy-intensive industries, the huge economic benefits brought by the increase in the urbanization rate cannot be denied.

As the ratio of carbon emissions to GDP, the carbon emission intensity has a certain inhibitory effect on the carbon emission intensity when the economic benefits brought by the urbanization process exceed the carbon emissions generated by itself. The direct effect of total energy consumption on carbon emission intensity is positive and highly significant, which shows that the total energy consumption mainly based on fossil energy plays a huge role in promoting the improvement of local carbon emission intensity. As a typical country with coal consumption as its main source, China’s coal consumption accounted for about 64% of China’s total energy consumption in 2015. It can be said that China’s rapid economic growth is at the cost of huge coal energy consumption45,46. In recent years, the proportion of coal energy consumption in China’s total energy consumption has declined, but it still exceeds 50%. In addition to the proportion of other fossil fuels, it is still a huge source of continuous growth in carbon emissions. Therefore, when the total energy consumption increases year by year, China’s GDP growth slows, which will inevitably lead to an increase in the intensity of carbon emissions. The direct and indirect effects of China’s foreign trade level on carbon emission intensity are significantly positive in the model regression results, indicating that the level of foreign trade not only promotes the improvement of local carbon emission intensity but also promotes the improvement of carbon emission intensity in neighboring provinces and cities. China has already realized the transformation of its export product structure from primary agricultural products such as agricultural and sideline products to major industrial products, with steel and cement as major47. In recent years, the increase in China’s export volume has also indirectly promoted the production of these energy-intensive industries, followed by the breeding of a large amount of carbon dioxide emissions, leading to an increase in carbon emission intensity.

Robustness test based on different regions

Because the regression results of spatial econometric models are largely affected by regional factors, this paper divides China into eastern, central, and western regions, and tests the robustness of the models for each regional regression. The results are shown in Table 8.

Table 8 Results of SDM model robustness test based on different regions.

It can be seen from the results in Table 8 that the direct effect of technological innovation level in the eastern, central, and western regions is significantly negative, consistent with the above conclusions, and the control variables have not changed significantly. This proves that the conclusion of this paper is reliable.

Threshold effect analysis

Threshold effect test

This paper takes the level of economic development as the threshold variable to explore the nonlinear impact of technological innovation level on carbon emission intensity under different economic development levels. Firstly, the number of threshold variables needs to be determined. The test results are listed in Table 9.

Table 9 Threshold effect test.

It can be seen from Table 9 that at different levels of economic development, both the single threshold and the double threshold have passed the 1% significance level test, while the triple threshold has not passed the significance level test. Therefore, there are single and double thresholds for the impact of technological innovation level on carbon emission intensity. The estimated value of the single threshold of economic development level is 2.37 and the estimated value of the double threshold of economic development level is 2.54 and 1.71.

Threshold regression results

Table 10 shows the regression results with the level of economic development as the threshold variable.

Table 10 Threshold regression results.

With the level of economic development as the threshold variable, there is a double threshold between the level of technological innovation and the intensity of carbon emissions, and the impact coefficients are both negative but slightly different. When the economic development level is lower than the first threshold value of 2.37, the impact coefficient is − 0.10; after crossing the first threshold, the influence coefficient is − 0.15, indicating that the inhibition effect is enhanced at this time; after crossing the second threshold of 2.54, the influence coefficient becomes − 0.21, and the inhibition effect is further strengthened. It indicates that when the level of economic development is taken as the threshold variable, there is an obvious nonlinear relationship between the level of technological innovation and the intensity of carbon emissions, and the inhibition effect increases with the improvement of the level of economic development. It can be seen from Table10 that under different levels of economic development, both the single threshold and the double threshold have passed the test of 1% significance level, while the triple threshold has not passed the significance level test. Therefore, there are single and double thresholds for the impact of technological innovation level on carbon emission intensity, respectively. The single “Economic Development Level Threshold” threshold is estimated at 2.37 and the double “Economic Development Level Threshold” threshold is estimated at 2.54 and 1.7066. The threshold variable economic development level in this paper is measured by GDP per capita, and the two thresholds shown in the threshold regression results indicate that when China’s per capita GDP level reaches the two critical values of 2.37 and 1.71, the impact of technological innovation on carbon emission intensity will change. From the point of view of time, after 2012, although China’s economy is developing rapidly, the economic form is promising, but too rapid industrialization, industrialization has also brought ecological impact, at this time China’s overall economic level reached the first threshold, technological innovation began to show the inhibition effect on carbon emissions, the concept of “scientific and technological carbon reduction” became popular, and the Chinese government and society paid more attention to the protection of the ecological environment. Around 2017, after the 19th Congress of the Communist Party of China, the environmental problems caused by China’s rapid economic growth have been paid more and more attention, and the concepts of “green” and “low-carbon” have been put forward. At the same time, China has entered a stage of relative economic prosperity, the government proposes to develop a high-quality economy, at this time the Chinese government and all sectors of society have a certain amount of “pockets”, so more vigorously strengthen scientific and technological investment. At this time, China’s economic development level has reached around the second threshold, and the level of scientific and technological innovation has been continuously improved at this stage, which has brought about the corresponding impact of increasing carbon emission intensity inhibition. When the level of economic development is in the third range, the inhibition effect is the largest. Economic development provides a material basis for scientific and technological innovation, promotes the output of scientific and technological innovation, improves the efficiency of innovation transformation, and creates a good innovation environment. It is the high-quality development of China’s economic level in recent years that has led to major breakthroughs in green and low-carbon technologies. Green and low-carbon technological innovation is not only an important part of the new round of global industrial revolution and technological change, but also an important support for China to cope with climate change, promote high-quality development and achieve the goal of building a beautiful China48. Therefore, the economic level plays an important role in the level of technological innovation and the intensity of carbon emissions. With the growth of the economic level year by year, the inhibition of the level of technological innovation on the intensity of carbon emissions is becoming more and more significant.

Threshold interval classification analysis

From the threshold regression results, it can be seen that the level of technological innovation has an obvious nonlinear impact on the carbon emission intensity under the level of economic development. This paper takes 2020 as an example to analyze the corresponding threshold range of each province, city, and region, in order to provide targeted suggestions for each province and city.

It can be seen from Table 11 that when the level of economic development is taken as the threshold variable, the optimal range for reducing carbon emission intensity of technological innovation level is the third range. In 2020, Beijing, Tianjin, Shanghai, Jiangsu, Zhejiang, Guangdong, and other places are in the optimal range, while Hebei, Shanxi, Jilin, and other places are not in the optimal range.

Table 11 Threshold range of provinces, regions and cities in 2020.

Related Articles

Back to top button