Green digital finance and technology diffusion

Achieving sustainable development depends on the development and diffusion of innovative and environmentally sound technologies. Technology diffusion promotes cumulative innovation, avoids wasted R&D from duplication, and can contribute more to productivity growth than internal R&D. Yet inventions are geographically concentrated and spillover benefits decay with distance, with knowledge often being tacit and spatially sticky. Promoting sustainable development thus requires not only stimulating internal R&D but also breaking the geographic localisation of technology diffusion. Green technology diffusion is shaped by regulations, geography, firm attributes, expected returns, complexity, and demand. Local financial markets are critical for diffusion, but traditional finance selects clients by assets and profitability, excluding diffusion with development potential. Green digital finance has emerged globally under carbon-neutrality goals, potentially reducing financing costs, easing constraints, guiding funds to green projects, enhancing information transfer, and encouraging inclusion and innovation. It may also require environmental information disclosure, alleviating information asymmetry and fostering spillovers. However, existing work focuses on innovation rather than diffusion, often ignores simultaneous effects of administrative boundaries and distances, and the net spatial effect of green digital finance remains uncertain. This study asks whether green digital finance can overcome localisation barriers and promote interregional technology diffusion, extends a spatial Durbin model to disentangle boundary and distance effects at the city-pair level, and tests the mechanisms of digital economy and market integration.

Literature establishes that technology diffusion is often localised, aided by geographical, cognitive, and institutional proximity, and especially so for complex and green technologies. Some evidence suggests renewable energy knowledge flows can span large distances. Green digital finance could break localisation barriers through digital technologies (big data, AI, mobile, blockchain) that reduce coordination costs and distance frictions, policy-driven focus that broadens and targets financing toward green projects, and mandated environmental information disclosure that reduces asymmetry and improves cooperation. Hypothesis 1: Green digital finance overcomes localisation barriers and promotes technology diffusion. Regarding the digital economy, green digital finance supports digitisation by expanding financial service scope and efficiency, fostering new entrepreneurial ecosystems, and promoting knowledge sharing. Mechanisms include: (a) level of digitisation (industry digitisation intensity) affecting interregional learning and technology adoption choices; (b) digitisation innovation reconfiguring distributed innovation networks and radiating spillovers; (c) reduced communication costs and information asymmetry via digital infrastructure and data. Hypothesis 2: Green digital finance promotes technology diffusion by fostering a digital economy through digitisation level, innovative capability, and reduced communication costs. Market integration is another channel: green digital finance can scale sustainable finance, lower barriers, and support greener trade, enabling access to high-tech inputs, reducing costs via intermediate trade, and spurring innovation through competition. Hypothesis 3: Green digital finance facilitates technology diffusion by promoting market integration.

Design: Panel dataset of 189 Chinese cities (2002–2015) creating 35,532 citing city–cited city pairs per year. Data sources include CSMAR (Annual Financial Reports of Chinese Listed Companies), the China Industry Business Performance Database, the Chinese Industrial Enterprise Database (National Bureau of Statistics), and the China City Statistical Yearbook. Variables were deflated where applicable and log-transformed.

Empirical strategy: An extended spatial Durbin model is implemented at the city-pair level with dual spatial weights to capture distinct spillovers from administrative adjacency and geographic distance. W1 is a within-contiguity (queen contiguity) matrix; W2 is an outside-distance matrix based on inverse geographic distance. The specification includes spatial lags of both the dependent variable (technology diffusion) and the key regressor (green digital finance) under both weight matrices, multi-way fixed effects (citing city, cited city, year), and clustered standard errors at the city-pair level. The focus parameter is the effect of green digital finance on technology diffusion after converting estimates to total spillover effects according to spatial matrices.

Dependent variable (Technology diffusion): Constructed from a city-level patent citation network by matching patent-citing and patent-cited information (SIPO data integrated from multiple platforms). Two measures: (1) TD1, a dummy indicating existence of any citation from city j to city i in year t; (2) TD2, the number of citations (cumulative over the past three years). Robustness uses 5-year and 10-year cumulative windows.

Independent variable (Green digital finance, GDF): A composite index capturing green transformation integrated with digital application. Original city-level green finance indicators (green credit, green investment, green insurance, green bonds, green support, green funds) from the City Statistical Yearbook are adjusted by matching industry-level digitalisation intensity to obtain corresponding green digital indicators; principal component analysis aggregates into the GDF index.

Controls: Single-city variables for both citing and cited cities: GDP, secondary industry output (STR), fixed-asset investment (FA), foreign direct investment (FDI), R&D investment (RD), financial deposits (FIN), Internet population (INT), and patent outputs (PAT). Pair-level controls include a province-level boundary dummy (BOR) and a high-speed rail connection dummy (HSR). All in logs where applicable.

Robustness and identification: Multiple spatial weighting schemes (city-level and province-level distances), alternative dependent variables (existence vs. counts; alternative accumulation windows), sub-dimension analyses (six GDF components), and an instrumental variable approach using spherical distance to Hangzhou (origin of China’s internet economy) as an instrument for GDF. First-stage strength and overidentification tests are reported, supporting instrument validity.

Mechanism tests: Interaction-term analyses examine digital economy channels—industry digitisation intensity (DIT), digitisation innovation (DIN), and communication cost proxy via optical cables laid (OCL)—and market integration channels—intercity trading flows (RT) and trading intensity (RT). Heterogeneity: Directional diffusion is assessed via interactions indicating East–Central/West citing-cited patterns.

• Benchmark effects: A 1% increase in green digital finance increases the overall patent citation probability (TD1) by 16.89% and the number of patent citations (TD2) by 28.06%, indicating substantial spatial stimulus effects on technology diffusion across cities.
• Duration: Using 5-year and 10-year cumulative diffusion windows yields persistently positive, significant effects, with slight cumulative phenomena over longer horizons.
• Dimensions of GDF: Among six sub-indices, green digital credit, green digital investment, and green digital support show the strongest and more evident positive effects on diffusion (both TD1 and TD2 specifications).
• Spatial factors: Province-level boundaries (BOR) negatively associate with diffusion, while high-speed rail links (HSR) positively associate, consistent with boundary frictions and transport connectivity facilitating diffusion.
• Instrumental variables: Using spherical distance to Hangzhou as an instrument confirms a positive causal effect of GDF on both TD1 and TD2; first-stage F-statistics exceed conventional thresholds and overidentification tests support validity.
• Mechanisms—digital economy: Interaction terms show that higher industry digitisation intensity (DIT) strengthens GDF’s diffusion effect; digitisation innovation (DIN) exhibits an even stronger mechanism effect than DIT and optical cable density; more optical cables laid (OCL) reduce communication costs and amplify diffusion.
• Mechanisms—market integration: Both higher intercity trading flows and trading intensity enhance the positive effect of GDF on technology diffusion.
• Directionality: Diffusion from eastern-region cities is stronger than from central-west regions. East-to-east and central-west-to-east citation patterns exhibit larger coefficients than east-to-central-west or central-west-to-central-west, indicating the east as a primary source of diffusing technologies.

The findings directly address whether green digital finance can overcome geographic localisation barriers in technology diffusion. The strong positive effects on both the likelihood and intensity of patent citations across cities indicate that GDF facilitates intercity knowledge flows beyond administrative borders and distance frictions. This supports Hypothesis 1 and underscores the role of digitalised, environmentally oriented finance in lowering financing and information frictions that traditionally constrain diffusion. Mechanism analyses show that GDF fosters a digital economy—through digitisation intensity and especially digitisation innovation—thereby reducing communication costs and information asymmetry and enabling more efficient matching and cooperation among innovators. Market integration via trade further enhances diffusion by easing access to intermediate and final goods, reducing costs, and increasing competitive pressure to adopt and adapt technologies. The directional results reveal that the technologically advanced eastern region serves as a key source of diffusing technologies, highlighting regional disparities in innovation capacity and absorptive capabilities. Overall, the study demonstrates that integrating green objectives with digital financial tools can be a powerful policy lever to scale technology diffusion crucial for sustainable development.

This paper contributes by: (1) establishing a robust causal link between green digital finance and intercity technology diffusion using a novel dual-weighted spatial Durbin model that separately captures administrative contiguity and geographic distance; (2) documenting substantial effect sizes—16.89% for citation probability and 28.06% for citation counts per 1% increase in GDF; (3) uncovering digital economy (digitisation intensity, digitisation innovation, communication costs) and market integration (trade flows, intensity) as effective mechanisms; (4) identifying directional heterogeneity with stronger diffusion from the eastern region. Policy implications include aligning green finance and innovation policies, building a unified national market with reduced trade costs, investing in transport and information infrastructure, and enhancing R&D and absorptive capacity in technologically lagging regions to benefit from diffusion. Future research could refine measures to isolate green patent diffusion, leverage micro-level firm or inventor networks, examine policy heterogeneity, and test generalisability across countries and time periods with alternative identification strategies and datasets.

The study measures overall patent technology diffusion at the city level but does not precisely identify diffusion specifically for green patents. Data limitations (availability only from 2002–2015 and reliance on manual matching and multiple databases) constrain granularity. Future work aims to obtain new data to assess green patent diffusion more accurately and to expand temporal and sectoral coverage.