Digital infrastructure construction drives green economic transformation: evidence from Chinese cities

The paper addresses how digital infrastructure construction influences urban green economic transformation in China. Against the backdrop of resource constraints, environmental degradation, and China’s rapid urbanization, the study argues that cities must shift from high resource consumption and emissions toward greener growth. With the rapid rise of the digital economy, the authors posit that digital infrastructure (e.g., broadband networks, 5G, data centers) could transform production and consumption patterns, potentially improving energy efficiency and reducing environmental impacts. The research aims to empirically identify the causal effect of digital infrastructure on green economic development and clarify mechanisms, providing evidence for China and insights for other developing countries.

Prior work has largely focused on transportation infrastructure’s impacts on economic and social outcomes, with less attention to digital infrastructure’s environmental performance. Existing studies suggest digitalization can affect energy use, industrial upgrading, innovation, and consumer behavior. However, most research on new digital infrastructure has been qualitative, lacking rigorous causal identification and empirical evaluation of environmental outcomes. This paper contributes by: (1) shifting focus from transport to digital infrastructure to examine its role in green economic development; (2) employing a non-radial directional distance function (NDDF) to measure cities’ green economic performance and a staggered DID leveraging the Broadband China policy; and (3) offering policy suggestions from the perspective of digital infrastructure construction.

Design: The Broadband China policy (pilot cities designated in staggered years starting in 2015) is used as a quasi-natural experiment. A staggered difference-in-differences (DID) with two-way fixed effects estimates the causal impact on green economic performance (GEPI) using panel data for 271 prefecture-level cities, 2003–2019. Model: Y_it = alpha0 + alpha1*DID_it + X_it + mu_i + nu_t + eps_it, where DID_it=1 for pilot cities post-implementation; mu_i and nu_t are city and year fixed effects. An event-study specification tests the parallel trends and dynamic effects. Outcome measurement: Green economic performance index (GEPI) is computed using the non-radial directional distance function (NDDF) with a global DEA framework, allowing asymmetric adjustments of inputs, desirable outputs, and undesirable outputs. Inputs: capital stock, labor (employment), energy (electricity consumption). Desirable output: real GDP. Undesirable outputs: industrial wastewater discharge and SO2 emissions. Weights follow prior literature (equal importance of energy input and outputs; capital and labor weights set to zero to eliminate their inefficiencies). Prices and quantities are converted to 2003 constant prices. Core treatment: DID_it based on inclusion in the Broadband China pilot list; treatment group comprises 100 pilot cities after data cleaning. Controls: Per capita GDP (PGDP), government fiscal expenditure share (GFE), environmental regulation (ER; comprehensive utilization rate of industrial solid waste), fixed-asset investment share (FAI), and foreign direct investment share (FDI). Robustness and identification checks: (1) Time-trend plots and event-study parallel trends; (2) heterogeneous treatment effect bias assessment for TWFE weights (two-wayfeweights in Stata, weights positive); (3) placebo test with 2000 random assignments; (4) exclusion of confounding policies by removing low-carbon city pilots and by narrowing the sample window (2011–2019). Mechanism analysis: Test channels via regressions of DID on energy intensity (EI; electricity consumption/GDP), digital industrialization (DI; employment share in information transmission, computer services, and software), and green technological innovation (GREEN; number of green patent applications).

- Parallel trends: Pre-treatment trends of GEPI for treated and control cities are similar; post-2015 effects grow over time, consistent with infrastructure build-out lags. - Baseline DID: Digital infrastructure significantly improves GEPI. Reported DID coefficients include 0.1446 (SE 0.0071), 0.0477 (SE 0.0171), 0.0918 (SE 0.0071), and 0.0474 (SE 0.0162) across specifications; significance at 1% level where indicated. - Robustness: - Heterogeneous TWFE weights: All treatment effects and weights positive; robustness indicator 0.3405, suggesting minimal bias from heterogeneity. - Placebo tests (2000 draws): Null effects centered near zero with p-values >0.1; the actual estimate is an outlier, supporting causal interpretation. - Excluding other policies: After removing low-carbon pilot cities (N=4046), DID=0.0321*; restricting to 2011–2019 (N=2439), DID=0.0362***; effects remain positive and significant. - Mechanisms (Table 4): - Energy intensity (EI): DID coefficient about −0.0205 (significant), indicating reduced energy intensity. - Digital industrialization (DI): DID ≈ 0.0039 (significant), indicating greater employment share in digital sectors. - Green innovation (GREEN): DID ≈ 0.1282 (significant), indicating more green patent applications. - Heterogeneity (Table 5): - By region: East 0.0688*; Central 0.0603*; West 0.0001 (ns). Benefits concentrated in eastern and central cities. - By development level: Economically developed 0.0492*; economically undeveloped 0.0137 (ns). Gains larger where digital infrastructure is earlier and more complete.

The findings support the hypothesis that digital infrastructure construction fosters urban green economic transformation. Broadband expansion catalyzes diffusion of digital technologies across sectors, improving energy efficiency, promoting digital industrialization, and stimulating green innovation—key mechanisms for greener growth. Event-study dynamics suggest effects materialize with lags, consistent with infrastructure deployment and adoption cycles. Robustness checks, including placebo tests and exclusion of confounding policies, strengthen causal claims. The heterogeneity patterns imply that regions with stronger initial digital bases and economic capacity (eastern and developed cities) can better leverage digital infrastructure to achieve green gains, highlighting complementarities between digital readiness and environmental outcomes and suggesting potential equity and capacity-building needs for lagging regions.

The study shows that digital infrastructure, proxied by the Broadband China policy, significantly promotes green economic transformation in Chinese cities. It does so indirectly by lowering energy intensity, expanding digital industrialization, and enhancing green technological innovation. Effects are more pronounced in eastern and economically developed cities. Policy implications include: scaling investment in core digital infrastructure (broadband, 5G, data centers); accelerating enterprise digital transformation—especially in energy-intensive industries—to unlock efficiency gains; and planning digital infrastructure with attention to energy consumption by siting data centers where renewable resources are abundant and advancing energy-saving technologies, thereby greening digital infrastructure itself. Overall, as digital networks expand and network effects grow, long-run benefits outweigh initial costs, enabling high-quality, low-carbon growth.

- Scope: City-level panel analysis; lacks micro-level (firm) evidence. - Mechanisms: Empirically tests only three channels (energy efficiency, digital industrialization, green innovation) due to data availability; other channels such as consumer behavior and environmental governance remain for future research. - Treatment proxy: Uses Broadband China as the measure of digital infrastructure; other components (e.g., data centers, industrial internet) are not directly measured. Future work could adopt broader and more granular digital infrastructure indicators.