Effects of land urbanization and internet penetration on environmental sustainability: a cross-regional study of China

Abstract

China has pledged to peak its carbon dioxide emissions (CEM) before 2030 and achieve carbon neutrality before 2060. The energy consumption related to land urbanization (LURB) and internet penetration (INT) may have a significant impact on CEM. It is critical for China to obtain a better understanding of the interaction among them in order to tackle challenges of climate change in the digital era. Initially, the multivariate panel bootstrap Granger causality test is conducted to estimate the causal effects of LURB and INT on CEM in China. The results reveal that the interaction patterns among them varied across different provinces, and particularly LURB and INT are two determinants of CEM mainly in Eastern China, which is consistent with the stochastic differential model. Then, CEM is not a Granger-cause of LURB and INT in the majority of Chinese provinces, as LURB and INT largely depend on government policies and industrial development. In addition, there are no significant interactions between LURB and CEM in over two-thirds of the provinces and there are also none between INT and CEM in over half of the Chinese provinces. Therefore, Chinese policymakers should further adopt differentiated and efficient policies targeting the weak links and achieve environmental sustainability through low carbon transition.

Appendices

Appendix 1

This paper uses the traditional measure of urbanization (URB) as a robustness test, and the findings are nearly the same with the measure of land urbanization (LURB). It indicates that the development patterns of land-centered urbanization and population-centered urbanization are coordinated on the whole in spite of some differences. Therefore, it makes this article’s estimation results convincing.

Table 7 Results of panel Granger causality between URB and CEM based on bootstrapped Wald statistics

The results differ in several ways. First, we do not find evidences of one-way Granger causality from URB to CEM in Shandong, Tianjin, and Shanxi. The possible reason is that there are still some restrictions on population moving from rural to urban areas in these provinces. As a result, people do not possess enough income to consume and invest. Thus, restrictions such as Hukou policies in these provinces should be further lifted. And the remaining provinces show that there may be other factors determining the CEM. Moreover, we find significant evidences of causality from URB to CEM in Hubei, which may be subject to the implementation of many attracting talent policies. For instance, in January 2017, Wuhan in Hubei province has proposed a graduate program aiming to keep one million college students in the city after they graduate. Population growth and migration induce URB, and higher population increases energy consumption, which resulted in an increase in CEM. Finally, we do not find evidences of one-way Granger causality from CEM to URB in Inner Mongolia and Shaanxi, as these two provinces are in the Western region with a small population and little internal migration.

Appendix 2

To further illustrate the long run effects of LURB and INT on CEM, this paper employs the panel autoregressive distributed lag (ARDL) model to test their association. The panel ARDL approach takes advantage of a single equation that makes it simple to perform, and it can be used when the studied factors are combination of I (0) and I (1) (Mert et al. 2019). There are two stages in the panel ARDL model when testing the long-term relationship among the variables. In the first step, the t statistics is calculated by the error correction model to determine whether the investigated variables have a long-term cointegration association. In the second step, the long-run and short-run coefficients can be obtained if the variables are cointegrated. Thus, referring to the studies by Mert et al. (2019), Munir and Riaz (2019), and N’dri et al. (2021), the panel ARDL model for the study is constructed as follows:

$$ \Delta \ln {CEM}_{it}={\mu}_i+{\varphi}_i\left[\ln {CEM}_{it-1}-{\lambda}_i^{\hbox{'}}{X}_{it}\right]+\sum \limits_{j=1}^{p-1}{\pi}_{ij}\Delta \ln {CEM}_{it-j}+\sum \limits_{j=0}^{q-1}{\omega}_{ij}\Delta {X}_{it-j}+{\varepsilon}_{it} $$

(10)

where i is the province, t is the time, μ_i is the cross-section effects,${\lambda}_i^{\hbox{'}}$ is the long-term coefficients, π and ω are the short-term coefficients, p and q are the optimal lags, and X includes LURB, INT, and ECG.

Initially, we perform panel unit root tests for the series in Table 8, the results show that the first differenced series of CEM, LURB, and ECG are stationary, and the INT series is stationary. Secondly, we perform the Pedroni panel cointegration test for the series after getting the integration orders in Table 9, and it confirms the cointegrating relationship among the variables. Therefore, we proceed with ARDL model to estimate the long-run and short-run coefficients along with the error correction coefficient, which are reported in Table 10.

Table 8 Results of IPS and ADF panel unit root tests

Table 9 Results of Pedroni panel cointegration test

Table 10 Results of the long-run and short-run estimations (dependent variable: Δ ln CEM; ARDL (1,0,0,1))

From an overall perspective, the long-run and short-run estimated results are presented in Table 10. The finding reveals that both urbanization and internet development have a significant positive impact on carbon emissions in the short run and a significant negative impact in the long run. The results further indicate that a 1% increase in LURB and INT could result in the reduction of CEM by 0.447% and 3.694%, respectively, in the long run. It corroborated that of Ali et al. (2019) in Pakistan and Shahnazi and Shabani (2019) in Iran, while it contradicted that of Dogan and Turkekul (2016) in USA and Altinoz et al. (2020) in top 10 emerging counties. The likely reason might be the local institutional and policy environment. For example, developed counties are known for their government strict energy regulations and measure taken for reducing energy consumption in the pursuit of environmental sustainability. However, developing counties are still accelerating economic development for their primary goals. China has now paid more attention to environmental sustainability after the achievement of rapid urbanization and digitization. The coefficient value of the error correction term (ECM) clarifies that the determinants of CEM from the long-run equilibrium will be corrected and converged to the equilibrium level at 35.1% annually.