3.1. Model Construction
- (1)
The establishment of a benchmark regression model
For Hypothesis 1, this paper tests the role of information infrastructure in promoting the flow of low-carbon technology knowledge by establishing a measurement model as follows:
In Formulas (1) and (2), indiffit and outdiffit are the explanatory variables, which represent the level of low-carbon technology knowledge inflow and low-carbon technology knowledge overflow of City i in t year, respectively. The coefficients β1 and β2 indicate the impact of information infrastructure construction on the flow of technological knowledge. If β1 and β2 are both significantly regular, it indicates that the development of information infrastructure construction has a positive impact on the inflow and outflow of technological knowledge. Controls are a collection of control variables. Furthermore, cityi is an urban fixed effect and yrart, a time fixed effect respectively. is a random perturbation term. The fixed effect model in this paper has passed the Hausman test and the p-value is significant.
- (2)
The mechanism of regression model building
For Hypothesis 2, this paper mainly uses the mediating effect model to verify the mechanism that factor flows promote the flow of low-carbon technology knowledge for information infrastructure. The specific model is as follows:
Among them, humit and fdiit represent the level of human capital of city i in t year and the utilization level of foreign direct investment; other variables are the same as Equation (1).
3.2. Data Description
- (1)
Low-carbon technology knowledge flows
When applying for a patent, the applicant must refer to all “prior art” related to the invention, while a new technology will also be referred to by subsequent “inventions”. This series of circular references can reflect the complete process of knowledge flow [
43,
44]. Although patent citation is an accurate but noisy indicator of the actual knowledge flow, it constitutes the most widely used measure of knowledge flow in economic, management, and policy literature [
45]. Therefore, this paper tracks the inflow of external low-carbon technology knowledge with the number of patents per ten thousand people and evaluates the spillover of local to external low-carbon technology knowledge with forward citations. According to Yu et al.’s [
46] method, the ratio of patent citations and citations to ten thousand permanent residents is used as an overall indicator to measure knowledge flow. Patent data used herein are derived from the incopat Patent Database (incoPat.com).
- (2)
Information infrastructure construction
In this paper, the information infrastructure is divided into two categories. One is the telecommunications network infrastructure, which includes mobile phones and network infrastructure. The other category is the information service level, which refers to the local postal and telecommunication service level. The regional information infrastructure level refers to the method of Li et al. [
19] to select four indicators. They are per capita internet users, per capita mobile phone users, per capita total telecommunications business, and per capita total postal business in the region. Per capita internet users and per capita mobile phone users are the reflection of the level of information infrastructure. Since entering the 21st century, mobile phones have produced an obvious substitution effect on fixed phones, so it is more appropriate to select mobile phones. The per capita telecommunication service and the per capita post and telecommunication service can represent the information service level. Bartlett’s sphere test and KMO test were carried out before the principal component analysis method was used. The Bartlett’s sphere test showed that the
p-value was significant at the level of 1% and the coefficient of KMO test was 0.758 greater than the critical value of 0.6, indicating that the correlation between explanatory variables was strong and suitable for principal component analysis. Each of the two types of indicator variables is derived from the wind database.
- (3)
Control variables and mechanism variables
Control variable: the level of economic development is measured using the logarithm of the city’s GDP per capita (logarithm of per capita GDP yuan). The population size is measured by the number of the resident population of the city (thousands of people per square kilometer take logarithm). The upgrading of industrial structure is measured by the ratio of the added value of the tertiary industry to that of the secondary industry. The level of financial development is measured by the ratio of the balance of deposits and loans of financial institutions to GDP at the end of the year. The degree of industrialization is measured by the proportion of the added value of the secondary industry to GDP.
Mechanism variable: human capital is measured by the ratio of the number of employees in urban units at the end of the period to the resident population in the city. Foreign direct investment is measured by converting the actual foreign capital utilized in the current year into RMB. Both the control variables and the mechanism variables are derived from the city statistical yearbook.
- (4)
Variable descriptive statistics
Table 1 shows the descriptive statistical results of the main variables in this paper. The standard deviation of the descriptive statistical results of low-carbon technology knowledge flow (Indiff, Outdiff) is 2.34, and the variance is small, indicating that there is no significant gap in the level of low-carbon technology knowledge flow in various regions. The minimum value of information infrastructure construction (ICT) is1.48819, and the maximum value is 20.63911. From this, we can see that some cities have better information infrastructure, and some cities are weaker.
3.3. Positive Results
- (1)
Benchmark Regression
Based on the above measurement models (1) and (2), this paper examines the direct impact of information infrastructure on the knowledge flow of low-carbon technologies. The results in
Table 1 show that for every additional unit of information infrastructure construction in cities, an average of 0.406 low-carbon technology patents will be cited per ten thousand people, i.e., the flow of low-carbon technology knowledge. Similarly, this also shows that the impact of information infrastructure on the inflow of low-carbon technology knowledge has a significant positive incentive effect, which verifies the conclusion of hypothesis 1 above. However, information infrastructure has a certain inhibitory effect on knowledge spillovers of low-carbon technologies, but the inhibitory effect is not significant.
Further observation of the influence coefficient of its control variables shows that the coefficients of industrial structure upgrading, population size, and industrialization degree are all in line with the actual situation, while the economic development level and financial development level do not have a positive incentive effect on the flow of low-carbon technology knowledge as expected, and inhibit the flow of low-carbon technology knowledge at a significance level of 10%. The reason is that, for the level of economic development and financial development, although the investment from the city government with a higher economic development level will give the enterprises corresponding research and development subsidies to improve the enthusiasm of the low-carbon technology knowledge flow to a certain extent, at present, due to the lack of government supervision over the use of investment funds, capital abuse and improper use are often caused. At the same time, government intervention will also increase the demand for research and development resources of enterprises. In the short-term lack of flexibility, research and development costs will be further increased, which will force enterprises to transfer research and development projects, thus inhibiting the development of low-carbon technology knowledge flow to a certain extent. In cities with higher levels of financial development, it is more convenient for enterprises to obtain loan support, which may result in the mismatch of resources and the decrease in the utilization rate of funds, thus resulting in similar results to the level of economic development in inhibiting the flow of low-carbon technology knowledge.
- (2)
Mechanism testing
The regression results of the mediating effect are shown in
Table 2. For human capital and foreign direct investment, the regression results in the first step of the mechanism test show that information infrastructure can significantly promote the regional human capital flow and foreign direct investment increase. The regression results in the second step of the mechanism test show that both information infrastructure, human capital and foreign direct investment can significantly promote the flow of low-carbon technology knowledge. When the coefficients in the model are significant, the Sobel test is not required. Therefore, this paper argues that the mechanism test effect model is significant. Information infrastructure can not only directly promote the flow of low-carbon technology knowledge, but also indirectly promote the flow of low-carbon technology knowledge by promoting the flow of human capital and increasing foreign direct investment. Therefore, Hypothesis 2 holds.
- (3)
Endogeneity Test
Considering the possible reverse causal relationship between information infrastructure and the diffusion of low-carbon technologies and the omission of variables, this paper intends to use the instrumental variable method to mitigate the endogenous problems. In this paper, the topographic relief of prefecture-level cities is selected as the tool variable. Firstly, the geographical gradient of cities meets the correlation condition of the tool variable. The greater the geographical gradient of cities, the higher the cost of information infrastructure construction will be, and the quality of the broadband network will be affected, thus affecting the operation efficiency of information infrastructure. On the other hand, geographical gradient as a tool variable satisfies exogenous conditions, because it belongs to natural geographical factors and will not be affected by socio-economic factors and will not affect the flow level of low-carbon technology knowledge. Column (1) of
Table 3 shows the effect of stage 1 instrument variable (IV-Undulation) on independent variable (ICT), with a coefficient of 0.381, which is significantly positive at 1%, while stage 1 regression F statistic is greater than 10, rejecting the assumption of “weak instrument variable”. Column (2) is listed as the second stage regression of the instrumental variable, and the regression coefficient of the core explanatory variable is 0.557, which is significant at the level of 1%. After the possible endogenous problems are solved, the information infrastructure construction still significantly promotes the flow of low-carbon technology knowledge, which is consistent with the benchmark regression results. Hypothesis 1 has been verified again, indicating the robustness of the conclusions in this paper.
- (4)
Heterogeneity test
Location heterogeneity test
According to the geographical location of the cities, the samples are divided into eastern and central-western cities and grouped for regression as shown in
Table 4. The results show that the construction of information infrastructure in the eastern region does not show a good effect in promoting the flow of low-carbon technology knowledge. On the contrary, the construction of information infrastructure in the central and western regions has a significant effect on the flow of low-carbon technology knowledge. This is different from the conclusion that the low-carbon city pilot will play a stronger role in promoting the flow of low-carbon technology knowledge in the eastern region [
47]. The possible reason is that in the central and western regions where the infrastructure is not perfect enough, the improvement of the information infrastructure can drive the regional linkage with other mature low-carbon technology development, strengthen the exchange of talents and capital, and thus play a role in promoting the flow of low-carbon technology knowledge.
In accordance with the “Circular of the State Council on Printing and Issuing the National Sustainable Development Plan for Resource-based Cities (2013–2020)”, the sample cities are divided into resource-based cities and non-resource-based cities, and then grouped for regression. The results in
Table 5 and
Table 6 show that the promotion effect of information infrastructure construction on low-carbon technology knowledge flow in non-resource-based cities is significantly positive, while the promotion effect on resource-based cities is not obvious. It indicates that the “resource curse” effect of resource-based cities may exist, which will cause difficulties in the transformation of resource-based cities, and the improvement of information infrastructure construction will not be able to promote the flow of low-carbon technology and knowledge.