5.1. Spatiotemporal Patterns
Figure 2 illustrates the temporal evolution of China’s coal power industry chain resilience from 2015 to 2022. At the national level, the resilience index exhibited a fluctuating upward trend during the study period, ranging from 0.219 to 0.266, with a cumulative increase of 21.5%. This trend is likely closely related to a series of policy measures in China aimed at strengthening the power industry chain in recent years, such as enhancing power infrastructure, promoting diversification of energy supply, improving smart grid capabilities, and establishing comprehensive energy emergency response mechanisms. These measures have effectively strengthened the industry chain’s capacity to withstand external shocks and recover from disruptions, thereby contributing to the steady annual increase in resilience observed between 2015 and 2022.
From a regional perspective, the coal–power industry chain resilience index in China’s eastern, central, western, and northeastern regions all exhibited a fluctuating upward trend. Specifically, the index in the central region fluctuated between 0.266 and 0.323, in the eastern region between 0.212 and 0.249, in the western region between 0.221 and 0.278, and in the northeastern region between 0.135 and 0.161. Overall, the resilience levels ranked from high to low as follows: central region > eastern region > western region > northeastern region.
This regional disparity can be largely attributed to differences in resource endowments, industrial structures, and policy support. The central region, serving as a crucial hub for both energy production and consumption, possesses strong infrastructural and industrial coordination capacity. The eastern region, although primarily a consumption area, benefits from advanced technology and efficient management, which contribute to higher resilience. The western region, while rich in resources, suffers from weaker infrastructure, leading to a relatively lower resilience level. In contrast, the northeastern region faces challenges such as industrial transformation pressures and population outflow, resulting in comparatively weaker resilience.
5.2. The Impact of Green Electricity Consumption on the Resilience of the Coal–Power Industry Chain
5.2.1. Results of the Baseline Regression Model
(1) Linear regression results
This study employed a fixed-effects panel regression model incorporating both province and year effects, while controlling for a series of influencing factors. The regression results demonstrated strong statistical significance. Specifically, green electricity consumption exerts a significant and positive impact on the resilience of the coal–power industry chain. This finding suggests that, to some extent, the increase in green electricity consumption contributes to enhancing the resilience of the coal–power industry chain.
Table 3.
(2) Quadratic regression
However, this positive effect is not indefinitely sustainable. Based on the theoretical framework, a quadratic term of green electricity consumption was introduced into the baseline regression, given the significance of the linear term (see
Table 4). The results indicate that both the linear and quadratic terms are statistically significant at the 1% level, with the quadratic coefficient being negative.
This finding suggests that the impact of green electricity consumption on the resilience of the coal–power industry chain follows an inverted U-shaped relationship—positive in the early stage but negative thereafter. In other words, green electricity consumption initially enhances overall resilience; however, once its level exceeds a certain threshold, the substitution pressure on traditional energy systems intensifies, leading to a decline in resilience. Thus, the relationship exhibits a characteristic inverted U-shaped effect.
The regression results show that green electricity consumption (green) enters the model with a significantly positive coefficient, while its squared term (green2) is significantly negative, indicating a clear inverted U-shaped relationship between green electricity consumption and the resilience of the coal power industry chain. Based on the standard turning-point formula for a quadratic specification, the estimated threshold is approximately 0.63. This implies that when the level of green electricity consumption is below this threshold, the expansion of green electricity contributes to enhancing the resilience of the coal power industry chain; however, once green electricity consumption exceeds this critical level, further increases begin to exert a suppressing effect on coal power industry chain resilience.
From a practical perspective, substantial heterogeneity exists in green electricity consumption across Chinese provinces during the sample period. Most provinces, particularly traditional energy-producing regions, remain on the left side of the turning point, whereas several provinces with relatively high levels of renewable energy installation and consumption have gradually approached or even surpassed this threshold. This pattern suggests that the impact of green electricity consumption on coal power industry chain resilience exhibits pronounced stage-specific characteristics.
Before reaching the turning point, green electricity consumption enhances coal power industry chain resilience by optimizing the power generation structure, alleviating constraints associated with fossil energy dependence, and facilitating the functional transformation of coal power toward flexible regulation and system support. Beyond the turning point, however, the crowding-out effect of renewable electricity on coal-fired generation capacity becomes increasingly evident, accompanied by declining utilization hours and rising cost pressures, which in turn weaken the stability of investment, operation, and upstream–downstream coordination within the coal power industry chain.
Therefore, the inverted U-shaped relationship indicates that green electricity consumption is not “the more, the better,” but rather needs to be aligned with the regulatory capacity and transformation pace of the coal power industry chain. From a policy perspective, greater attention should be paid to the pace and structural design of renewable electricity expansion, so as to avoid excessive negative impacts on coal power industry chain resilience during high-consumption stages.
5.2.2. Robustness Checks
Table 5 presents the results of robustness checks. Three approaches were employed: (i) using an alternative measurement method for the dependent variable, (ii) winsorizing the sample, and (iii) introducing a one-period lag of the explanatory variable.
First, In the baseline model, the resilience index constructed using entropy weighting is employed as the dependent variable. To examine the sensitivity of the resilience measure to the choice of weighting scheme, we replace the dependent variable with a composite index calculated using the CRITIC weighting method (denoted as cri), and the corresponding regression results are reported in Column (2). The results show that both the sign and the statistical significance of the core explanatory variable remain consistent, indicating that the main conclusions are not driven by a specific weighting method and are therefore robust.
Second, to address the influence of extreme values, the core variables were winsorized at the 1% level. The regression results (Column 3) show that the coefficients of key variables remain significant and retain their original direction, further confirming that the results are not driven by outliers.
In further robustness checks, this study re-estimates the baseline model using province-level clustered standard errors to account for potential intra-group correlation. As reported in column (4), after applying clustered standard errors, the coefficients of green electricity consumption and its squared term remain positive and negative, respectively, consistent with the baseline results, indicating that the direction of the inverted U-shaped relationship remains unchanged. Although the statistical significance of the coefficients declines under this more stringent inference approach, the overall model structure and the estimated effects of the core explanatory variables remain stable. This suggests that the identified nonlinear relationship between green electricity consumption and coal power industry chain resilience is not driven by a specific standard error specification; while robustness is somewhat weakened, the results remain persuasive.
Overall, through multiple robustness checks—including alternative index construction methods, winsorization of the data, and the use of province-level clustered standard errors—the core conclusions of this paper are consistently supported, indicating that the findings are robust. Overall, through alternative variable computation methods, winsorization, and lagged explanatory variables, the core conclusions of this study are consistently supported, demonstrating robustness of the results.
5.2.3. Empirical Analysis
In the empirical analysis, although a set of relevant control variables is included, potential endogeneity concerns may still arise. On the one hand, omitted variable bias may exist, as factors such as regional policy implementation capacity in promoting green transition, the technological innovation environment, or strategic adjustments in the energy structure may simultaneously affect both green electricity consumption and the adaptability and transformation capacity of the coal-fired power industry chain, yet are difficult to fully quantify and incorporate into the model. On the other hand, there may be a bidirectional causal relationship between green electricity consumption and the resilience of the coal-fired power industry chain. Specifically, regions with higher coal power industry chain resilience tend to possess stronger system regulation and support capacity, which may in turn facilitate a higher level of green electricity consumption. In addition, measurement errors in relevant variables may further contribute to endogeneity issues.
Under these circumstances, relying solely on fixed effects regressions may lead to biased estimates of the true causal effect. Therefore, this study further employs an instrumental variable approach and adopts the two-stage least squares (2SLS) method to address potential endogeneity.
To ensure the validity of the 2SLS estimation, the selected instrumental variable must satisfy two key conditions: relevance and exogeneity. This study constructs an instrumental variable based on the interaction between wind speed and the share of wind power installed capacity in total installed capacity to instrument green electricity consumption. Regarding relevance, wind speed, as a typical natural geographic and meteorological factor, directly determines the exploitable potential of wind energy. Its interaction with the wind power installation coefficient captures the potential of wind power that can be effectively converted into electricity and integrated into the grid under given installation conditions, thereby exerting a significant impact on regional green electricity consumption. With respect to exogeneity, wind speed is a natural endowment whose short- and medium-term fluctuations are not directly influenced by regional economic development, industrial structure adjustment, or changes in the resilience of the coal-fired power industry chain. Based on the above considerations, the 2SLS estimation is conducted, and the corresponding regression results are reported below (
Table 6).
In the instrumental variable regression, both green electricity consumption and its squared term are treated as endogenous variables and are jointly identified using the constructed instrumental variable and its squared term. The first-stage regression results indicate that the instrumental variables exhibit strong explanatory power for green electricity consumption and its squared term, and the corresponding statistics suggest that there is no apparent weak-instrument problem, supporting the validity of the instrumental variable strategy. The second-stage estimation results show that although the statistical significance of the core coefficients becomes weaker after introducing instrumental variables, their signs remain consistent with those obtained from the fixed effects regressions and remain statistically significant. Overall, these findings indicate that after addressing potential endogeneity concerns, the direction of the effect of green electricity consumption on the resilience of the coal-fired power industry chain remains fundamentally unchanged, thereby further strengthening the credibility of the baseline conclusions.
5.2.4. Mediation Effect Analysis
To examine the specific mechanisms through which green electricity consumption affects coal power industry chain resilience, this study conducts a mediation analysis based on the baseline regressions and explicitly aligns the identified mediation pathways with the previously proposed Hypotheses H2 and H3.
Given that the fixed-effects model indicates a nonlinear impact of green electricity consumption on the resilience of the coal–power industry chain, the mediation analysis explicitly controls for the nonlinear (quadratic) term of green electricity consumption. This ensures that the identification of the mediation effect fully accounts for the inverted U-shaped main effect.
First, power source diversity was introduced as a mediating variable. Hypothesis H2 posits that, in the early stage of renewable electricity expansion, green electricity consumption enhances coal power industry chain resilience by increasing energy structure diversity and strengthening the industry chain’s resistance capacity at the onset of external shocks. The underlying logic is that a more diversified power supply reduces reliance on a single energy source, enhances redundancy and substitutability within the supply structure, and thereby facilitates the maintenance of basic operational functions when facing external disturbances. The regression results are presented in
Table 7.
From the results of Model (3), both the linear and quadratic terms of green electricity consumption remain significant at the 1% level. Model (1) indicates that green electricity consumption has a significant positive effect on power source diversity, demonstrating that the development of green electricity in its early stage markedly promotes the diversification of the power generation structure. After controlling for the mediating variable, Model (2) shows that green electricity consumption still significantly affects resilience, and power source diversity exhibits a significant positive impact on resilience, indicating that a diversified power structure enhances the system’s ability to adapt and recover in the face of external shocks. These results suggest that power source diversity partially mediates the effect of green electricity consumption on the resilience of the power industry chain.
The above results indicate that green electricity consumption enhances coal power industry chain resilience by optimizing the energy structure and increasing system redundancy and substitutability, thereby strengthening the industry chain’s ability to maintain key functions at the initial stage of external shocks. This finding provides empirical support for the positive mediation pathway proposed in Hypothesis H2.
To further explore the mechanism through which green electricity consumption influences the resilience of the power industry chain, the annual utilization hours of thermal power plants above 6000 kW were introduced as a mediating variable. Hypothesis H3 posits that green electricity consumption weakens coal power industry chain resilience by reducing coal-fired unit utilization hours, thereby adversely affecting the industry chain’s recovery capacity. The underlying mechanism is that, as the share of renewable electricity increases, coal-fired units transition from stable baseload operation to low-utilization, frequently ramped regulating units. This shift undermines equipment efficiency, operational stability, and investment incentives, which in turn hampers functional restoration and operational recovery following external shocks. The results are presented in
Table 8.
Model (1) shows that green electricity consumption has a significant negative effect on the mediating variable (coefficient = −1.310, p < 0.01), indicating that as the level of green electricity consumption increases, the annual utilization hours of traditional thermal power plants decline significantly, reflecting the substitution effect of renewable energy on conventional thermal power. Meanwhile, Model (2) shows that after controlling for the mediating variable, green electricity consumption still has a significant effect on resilience, and the mediating variable exhibits a significant positive impact on resilience, suggesting that the annual utilization hours of thermal power plants partially mediate the effect of green electricity consumption on the resilience of the power industry chain.
This result indicates that green electricity consumption suppresses overall coal power industry chain resilience by reducing the operating intensity and stability of coal-fired units, thereby weakening the industry chain’s recovery capacity after external shocks. This finding is consistent with the expectation proposed in Hypothesis H3.
To further explore the internal mechanisms through which green electricity consumption affects the resilience of the coal power industry chain, this study employs a bias-corrected Bootstrap method to test the potential mediation paths. The results are reported in
Table 9.
The Bootstrap mediation test results indicate that the confidence intervals of the indirect effects for both theoretical pathways do not include zero, suggesting statistical significance. Specifically, for the first pathway (the energy diversity mechanism), the indirect effect is positive, indicating that green electricity enhances industry chain resilience by improving system diversity. For the second pathway (the substitution effect mechanism), the indirect effect is negative, reflecting the transition pressure imposed on the traditional coal power industry chain due to the substitution effect of green electricity.
Moreover, the confidence intervals of the direct effects for both pathways also exclude zero, indicating that green electricity consumption exerts a statistically significant direct impact on the resilience of the coal power industry chain.
In summary, green electricity consumption affects the resilience of the power industry chain through multiple pathways. On one hand, in the early stages, the development of green electricity significantly enhances power source diversity, thereby improving the system’s adaptive and recovery capabilities in response to external shocks. On the other hand, as green electricity continues to grow, increased green electricity consumption reduces the utilization hours of traditional thermal power plants, demonstrating the substitution effect of renewable energy on conventional power and affecting the operational characteristics of the power industry chain.
5.2.5. Heterogeneity Analysis
To further examine regional differences in the impact of green electricity consumption on the resilience of the coal–power industry chain, the national sample was divided into four regions: Northeast, East, Central, and West, and separate regressions were conducted, as shown in
Table 10.
The results indicate that the regressions for the Central and Western regions are the most significant, both passing the 1% significance level and exhibiting a typical inverted U-shaped relationship. In contrast, the regression coefficients for the Eastern and Northeastern regions maintain the same direction as the full sample but do not reach statistical significance, suggesting that the effect of green electricity consumption on coal–power industry chain resilience in these regions is not yet substantial. This may be because the Eastern region, with a more mature electricity market and stronger grid regulation capacity, can smooth out the marginal effects of green electricity consumption, whereas in the Northeastern region, issues such as unstable green energy resource quality and insufficient power transmission capacity limit the systemic impact of green electricity.
Overall, the nonlinear relationship between green electricity consumption and coal–power industry chain resilience is more pronounced in the Central and Western regions, highlighting that policies promoting green electricity should also consider its structural impacts on regional energy systems, particularly by strengthening grid adaptability and enhancing the integration of green electricity with conventional power systems.