The introduction of high-speed rail (HSR) is a significant development in many countries, aiming to improve interregional exchanges and economic growth. However, the environmental impact, particularly concerning carbon emissions, is a subject of ongoing debate. While some studies suggest HSR can increase carbon emissions due to energy consumption during construction and operation, others highlight its potential to reduce emissions by shifting travel from more carbon-intensive modes like air travel and private cars. The impact of HSR on carbon emissions varies significantly depending on factors such as energy sources used, operational efficiency, and the overall transportation system's sustainability. In the context of China, where the HSR network is the world's largest and fastest-growing, understanding this relationship is particularly crucial. This study aims to address the gap in existing research by investigating the mediating role of green innovation (GI) and the resilience of environmental investment (REI) in the relationship between HSR introduction and carbon emission reduction. The study leverages signaling theory, fault tolerance theory, and stakeholder theory to propose a theoretical framework that explains how HSRs can indirectly reduce carbon emissions by influencing GI and REI. The study focuses on China, given its extensive HSR network and commitment to sustainable development. Understanding this relationship in China, where rapid economic development has also led to increased environmental concerns, is vital for informed policymaking and sustainable transportation planning.

Existing literature presents contrasting views on the impact of HSR on carbon emissions. Some studies highlight the energy consumption and environmental impact of HSR construction and operation, particularly in countries like the US. However, other research emphasizes the potential for carbon emission reduction when HSR replaces more carbon-intensive modes of transportation and incorporates sustainable operational practices. Studies examining the experience of countries like Japan, Germany, and the UK show varied results depending on the specific context and implementation strategies. In China, the literature indicates a debate surrounding the overall effect of HSR on the environment. While the government's policy emphasizes regional development and economic growth through HSR, the literature also recognizes the energy consumption associated with the technology. Existing literature often focuses on the direct impact of HSR on carbon emissions, neglecting the mediating roles of green innovation and the resilience of environmental investments. This study aims to fill this gap by examining these mediating mechanisms in the context of China's rapidly expanding HSR network. The concepts of green innovation, emphasizing environmentally friendly technologies and practices, and environmental investment resilience, focusing on the capacity of environmental investments to withstand economic and environmental shocks, are central to the study's theoretical framework. The study will draw upon various theoretical lenses, including signaling theory, stakeholder theory, and fault tolerance theory, to explain the complex interplay between HSR introduction, green innovation, environmental investment resilience, and carbon emission reduction.

This study employs a difference-in-differences (DID) model to analyze panel data from 284 prefecture-level cities in China from 2000 to 2021, resulting in 4920 observations. The DID approach compares carbon emissions in cities with HSR (treatment group) to those without (control group) before and after HSR introduction, controlling for time-invariant city-specific factors and time-varying factors. The dependent variable is carbon emissions (InCO2), measured using CO2 emission data from the China Urban Statistical Yearbook. The key independent variable is the introduction of HSR (DID), a binary variable indicating whether a city has an operational HSR in a given year. The mediating variables are green innovation (InGI) and the resilience of environmental investment (REI), measured using multiple indicators and methods. Green innovation is measured by the number of environmental protection-related patents. REI is further broken down into EI stability (InEIS) and EI flexibility (InEIF), measured using a data envelopment analysis (DEA) model, specifically the Slack-Based Measure Data Envelopment Analysis (SBM-DEA), to assess the efficiency of environmental investments. EI stability reflects the maximum loss of environmental investment efficiency during shocks, such as the 2008 financial crisis, while EI flexibility represents the recovery time after such shocks. Several control variables are included to account for other factors influencing carbon emissions, such as population size (Inpeo), green area (Ingreenarea), number of hospitals (Inhos), revenue from telecommunication services (Ininfor), total industrial output (Inindu), employment in mining (Inmine) and manufacturing (Inmanu), household LPG use (InLPG), natural gas supply (Ingas), and electricity consumption (Inelec). The study also incorporates moderating variables, including the digital economy (InDigitalcity and DID_Digitalcity), urban-rural disparity (InU-Rdiff and InHDI), and whether a city is resource-based (Resourcecity and DID_Resourcecity). Robustness checks include parallel trend tests, propensity score matching (PSM)-DID, and placebo tests. Instrumental variable tests address potential endogeneity concerns. Heterogeneity analyses are conducted based on city size (population, area, GDP) and resource availability (public finance and water resources). The study utilizes elasticity analysis to quantify the effects of different variables and their interactions.

The study's key findings support the hypothesis that HSR introduction reduces carbon emissions. The DID model shows a statistically significant negative relationship between HSR introduction and carbon emissions (Model 2, Table 3), confirming Hypothesis 1. The mediating effects analysis reveals that both green innovation (InGI) and the resilience of environmental investment (REI) significantly mediate the relationship between HSR introduction and carbon emission reduction (Models 3, 5, and 7, Table 3), supporting Hypotheses 2 and 3a, 3b. Elasticity analysis indicates that while the direct effect of HSR on reducing CO2 is relatively weak (-0.001), its indirect effect through GI and REI is significant. The moderating effects analysis reveals that the digital economy strengthens the negative relationship between HSR and carbon emissions and the positive relationship between HSR and GI, EI stability, and EI flexibility, confirming Hypotheses 4a, 4b, 4c, and 4d. Conversely, the urban-rural disparity weakens the negative relationship between HSR and carbon emissions (Model 5, Table 12), supporting Hypothesis 5, while resource-based cities weaken the negative relationship (Model 6, Table 12), supporting Hypothesis 6. Robustness tests, including parallel trend tests, PSM-DID, placebo tests, instrumental variable tests, and heterogeneity analyses based on city size and resource availability, consistently support the main findings. The heterogeneity analysis indicates that the effect of HSR on carbon emission reduction is more pronounced in larger cities and those with greater financial resources, whereas in medium-sized cities with abundant water resources, the effect can be less pronounced or even reversed.

The findings of this study provide strong evidence supporting the notion that HSR introduction can contribute to carbon emission reduction in China. The mediating role of green innovation and the resilience of environmental investments underscores the importance of considering these factors when planning and implementing HSR projects. The study's results highlight the importance of integrating sustainable practices and technologies into HSR development and operation. The significant moderating effects of the digital economy suggest that leveraging digital technologies can further enhance the environmental benefits of HSR. The negative moderating effects of urban-rural disparity and resource dependence emphasize the need for targeted policies to address regional inequalities and promote sustainable development in all types of cities. The findings contribute significantly to the understanding of the complex interplay between transportation infrastructure, innovation, investment, and environmental sustainability. The study's focus on China, with its unique context of rapid economic development and large-scale HSR deployment, offers valuable insights for other countries considering similar infrastructure development projects.

This study provides comprehensive evidence that HSR introduction in China contributes to carbon emission reduction, mainly through mediating effects of green innovation and the resilience of environmental investment. The digital economy strengthens this positive relationship, while urban-rural disparities and resource dependence weaken it. The findings offer valuable implications for policymakers and city managers in promoting sustainable transportation and development. Future research could expand the geographical scope, explore diverse data sources, and delve deeper into the specific mechanisms of green innovation and environmental investment resilience. Further investigation into the long-term effects and potential unintended consequences of HSR deployment is also warranted.

The study's limitations include reliance on a single data source (China Urban Statistical Yearbook), potential sampling bias due to the focus on Chinese cities, and unexplained variance in some models, as indicated by low R-squared values. While robustness tests mitigate some concerns, future studies could address these limitations by incorporating multiple data sources, expanding the geographic scope, and refining the econometric models to account for additional variables and interactions. The study's reliance on existing definitions and measurements of green innovation, environmental investment resilience, and the digital economy also represents a limitation. Future research should explore more nuanced and context-specific measures of these concepts.