Drawing on panel data from 284 Chinese cities spanning 2007 to 2022, this study analyzes how the Pilot Policy Promoting the Integration of Technology and Finance (PPITF) affects the collaborative development of digitalization and greenization (CDG) and investigates its underlying mechanisms using a multi-period difference-in-differences model. Digitalization is evaluated across three dimensions—digital economy, digital hard environment, and digital soft environment, while greenization is assessed through green production, green lifestyle, and green ecology, with a coupling coordination model employed to measure CDG. The empirical results indicate that PPITF significantly enhances CDG, and these findings remain robust across various tests. Mechanism analysis reveals that PPITF fosters CDG primarily through talent agglomeration and financial subsidies. Furthermore, the effects are more pronounced in central and western cities, low fiscal pressure cities and large market size cities. This study contributes new empirical evidence that PPITF facilitates both digital transformation and green development of Chinese cities.
The climate crisis and resource depletion are key issues that pose a severe threat to global sustainable development (Borck & Mulder, 2024; Wu et al., 2025). In response, the European Commission has proposed the concept of “digital and green transitions,” which means achieving a systematic transformation through digital technology to pursue a green future and using digital transformation to offset carbon footprints (Lenz, 2025). As the world’s largest developing country, in the context of escalating urban pollution, gradually weakening cost advantages, and the inapplicability of the factor-driven model, China regards the twin transition as a powerful driving force for the construction of Chinese-style modernization and a foundation for fostering harmonious coexistence between humans and nature, and China is serving as a globally significant case.
The collaborative development of digitalization and greenization (CDG) refers to a mode that integrates digital technology and greenization concepts to achieve sustainable development (Kang & Shang, 2025). Owing to high costs and potential risks, the motivation and willingness for collaborative development are insufficient, which requires innovation and support from the financial service system. In the context of digital economy development, the in-depth integration of finance and emerging technologies is an effective way to compensate for the shortcomings of traditional financial services and to accelerate green innovation and digital transformation (Singh et al., 2023). Nevertheless, as two relatively independent systems, technology and finance exhibit a certain degree of exclusion toward each other’s inclusive development (Moshirian et al., 2021). In this context, China launched two batches of the “Pilot Policy Promoting the Integration of Technology and Finance” (PPITF) in 2011 and 2016 to achieve effective connectivity between the two parties. This institutional arrangement optimizes the financing structure, environment, and methods, and strengthens risk management and avoidance. Moreover, it promotes corporate digital technological breakthroughs by improving digital infrastructure (Wu et al., 2025). Significantly, PPITF in corporate sustainable development has demonstrated green benefits by integrating digital tools such as big data and cloud platforms, which shows potential to enhance environmental information disclosure and green R&D investment, and generates heterogeneous effects in terms of ownership, industry type, and geographical location (Liang et al., 2025; Yang et al., 2025).
Sporadic literature confirms that PPITF has significant potential to promote corporate green innovation (Li et al., 2025; Yu & Zhang, 2025), while most existing studies concentrate on the unilateral impact of digital transformation on green innovation (Wang et al., 2024; Hoque & Lee, 2025; Xia & Chen, 2025). It should be noted that substantial research gaps remain in exploring the impact of PPITF on CDG. First, these gaps are reflected in the absence of a systematic framework integrating urban digitalization and greenization into a unified analytical model for synergistic measurement and research, whereas most existing studies primarily focus on the industrial or corporate level. Second, the functional mechanisms of PPITF remain vague and underexplored, particularly in empirical research. In conclusion, theoretical elaboration and empirical evaluation of how PPITF influences urban CDG and through which mechanisms remain insufficient, and the heterogeneity tests is still limited.
In view of this, this study constructs an integrated theoretical model to explore how PPITF can promote urban CDG through financial subsidies and talent agglomeration. Twin transition theory emphasizes that green and digital transitions are mutually reinforcing rather than separate paths, and this synergy heavily depends on the embedded support of digital technology for green development (Song et al., 2025), while the key obstacles in digital technology development are funding and talent (Wu et al., 2025; Hua et al., 2025). As a key driving force supporting technological innovation, PPITF explicitly requires pilot cities to continuously innovate fiscal science and technology investment mechanisms and to introduce supportive policies for talent recruitment, thereby providing financial guarantees and technical support for urban digital transformation. Hence, PPITF has great potential to drive the “digital–green synergy.” However, this potential impact may vary according to regional, governmental, and market conditions. The first factor is the significant disparity in financial service systems, talent agglomeration, and economic development levels among cities in the eastern, central, and western regions (Wang et al., 2025a, 2025b), which further affects the policy’s effectiveness. The second is that PPITF requires governments to innovate fiscal science and technology investment mechanisms, and fiscal pressure can substantially influence policy outcomes. Additionally, market size determines the returns on innovation investment (Ferraro & Peretto, 2020), and insufficient scale reduces the willingness of innovation entities to adopt digital technologies and pursue green transformation. By integrating these analyses, this study clarifies the mechanisms through which PPITF stimulates financial subsidies and talent agglomeration and reveals the multi-level heterogeneity effects, thereby elucidating the logic and contextual conditions of sustainable transformation driven by PPITF.
The contributions of this study are as follows. First, regarding research content, existing literature primarily examines the single effect of digitalization on green transformation in isolation (Hoque & Lee, 2025; Wang et al., 2025a, 2025b), whereas this study integrates digitalization and greenization within a unified framework, and explores the impact of PPITF on urban CDG, which is crucial for accelerating sustainable development. Furthermore, this study establishes a two-dimensional assessment system combing digitalization and greenization, yielding an empirical foundation for more accurate CDG assessment and policy analysis. This study also expands the application of classical theories such as innovation systems and signal transmission in CDG, providing theoretical support and contributions concerning how digitalization and greenization can synergistically promote China’s comprehensive sustainable development. Second, regarding research perspective, existing literature primarily focuses on the impact of PPITF on corporate green innovation (Yang et al., 2025) and the institutional policies affecting CDG and confirm the role of AI policies on urban CDG (Kang & Shang, 2025). However, these literatures insufficiently explore whether PPITF has a profound effect on urban CDG and its intrinsic mechanism of action. Consequently, this study explores the mediating mechanism of PPITF on urban CDG using talent aggregation and financial subsidies as technical and financial support factors, which comprehensively assesses the potential paths of PPITF on urban CDG and provides valuable references for effectively leveraging the policy effect of PPITF and promoting sustainable development. Third, regarding research scope, although existing studies have explored the heterogeneity characteristics of the impact of PPITF on green innovation from aspects such as resource endowment, ownership, and industrial characteristics, this study examines the heterogeneity of PPITF in promoting urban CDG from the perspectives of geographical location, fiscal pressure, and market size, providing theoretical support and practical guidance for the government to formulate differentiated policies and dynamic strategies.
Literature reviewExisting studies have shown that the establishment of PPITF positively influences corporate green innovation (Liang et al., 2025; Yang et al., 2025) and promotes corporate digital technological breakthroughs (Wu et al., 2025). Li et al. (2025) examined the effect of PPITF on strategic and substantive green innovation using a DID approach, finding that the effect is more pronounced in non-state-owned enterprises and non–high-tech sectors. Moreover, although the literature has not directly addressed urban greenization, the establishment of PPITF has been found to reduce urban carbon emissions, support low-carbon development, and exhibit significant variation across urban types and government attention (Yu & Zhang, 2025). Meanwhile, existing research has primarily investigated the impact of PPITF on firm entry (Liu & Liu, 2024), corporate innovation boundaries (Huang et al., 2025), and export green sophistication (Wang et al., 2025a, 2025b), and conducted heterogeneity analyses by ownership structure, industry type, and geographical location.
The CDG represents a core component in cultivating new quality productive forces and advancing China’s high-quality development. Scholars have emphasized that digitalization is a complex process involving the application of digital technologies, whereas greenization evolves from concepts such as sustainable development and green economy, highlighting the transformational goals of low-carbon and environmentally sustainable growth (Broccardo et al., 2023; Li et al., 2024). The White Paper on the Collaborative Development of Digitalization and Greenization (2022), issued by the China Academy of Information and Communications Technology, defines CDG as a framework for improving economic and social systems by enabling mutual reinforcement—digitalization empowering greenization and greenization driving digitalization—thereby achieving comprehensive high-quality development (Zhang et al., 2025). Digitalization serves as a key pathway for realizing greenization by using artificial intelligence, big data, and related technologies that leverage data resources to facilitate energy structure transformation, industrial upgrading, and promote green mobility and sustainable consumption (Guandalini, 2022). Furthermore, greenization drives the continuous advancement and iteration of digital technologies by establishing green development objectives, imposing new requirements on data acquisition and processing capacities, and promoting the deep integration of digital technologies across industries (Chen, 2022).
In empirical research examining the relationship between digitalization and greenization, most studies have discussed the positive effects of digitalization on green transformation at the provincial, municipal, or enterprise level, indicating that digital transformation serves as an important mechanism for promoting green innovation, reducing carbon emissions, and improving green total factor productivity (Hoque & Lee, 2025; Wang et al., 2025a, 2025b). Furthermore, scholars have noted that green development requires innovative entities to account for the “rebound effect” arising from the expansion of digital technologies and advocate for a moderate degree of digitalization (Péréa et al., 2023). Several studies have begun to explore the synergistic effects of digitalization and greenization in terms of trend evolution, influencing factors, and utility value, such as scholars evaluate the digital–green synergies in China’s regional advanced manufacturing industry to reveal its evolutionary characteristics (Tian et al., 2023), while other scholars have highlighted the role of AI policies in enhancing industrial digitalization and greening synergies (Kang & Shang, 2025). In addition, researchers have empirically examined the enabling effect of digital–green synergies on enterprise ESG performance (Zhang et al., 2025).
Policy background and research hypothesesPolicy backgroundSci-tech finance constitutes a crucial policy instrument implemented by the state to enhance independent innovation capabilities, promote the transformation and application of scientific and technological achievements, and guide the evolution of the economic development model. These policies aim to foster enterprise development and innovation projects through financial support and preferential conditions, thereby providing strategic momentum for innovation-driven urban growth. The implementation of PPITF in China can be delineated into three distinct phases, and its graphical timeline is illustrated in Fig. 1.
- (1)
Exploration and development phase (before 2011). In 1985, the Decision of the CPC Central Committee on the Reform of the Science and Technology System first proposed the establishment of venture capital to support high-tech development, marking the beginning of China’s sci-tech finance development. In 1995, the Central Committee of the Communist Party of China and the State Council issued the Decision on Accelerating the Progress of Science and Technology, which emphasized the need to broaden sci-tech finance channels and substantially increase the scale of technology loans. Furthermore, the National Medium and Long-Term Program for Science and Technology Development (2006–2020), implemented in 2006, elevated sci-tech finance from a supporting to a central role in innovation promotion. During this stage, the focus was primarily on credit and capital markets, and the government enhanced the cooperation between technology, and finance by providing loans, venture capital, and bonds to establish financial guarantees for technological innovation.
- (2)
Pilot launch and expansion phase (2011-2016). In December 2010, the Ministry of Science and Technology, together with four other governmental departments, jointly issued the Notice on Printing and Distributing the Implementation Plan for the Pilot Policy Promoting the Integration of Technology and Finance. The first batch of PPITF projects was launched in 2011 across 41 cities, including Shanghai, Tianjin, and Nanjing, to encourage these pilot cities to implement innovative practices and generate replicable policy outcomes. Subsequently, in June 2016, 9 cities—including Zhengzhou, Xiamen, and Ningbo were designated as the second batch of PPITF projects, marking a strategic step toward the program’s nationwide expansion. The distribution of these pilot cities is shown in Table 1. During this stage, the implementation of PPITF reshaped the financial sector, enhanced operational efficiency across industries, and enabled innovation entities to access greater financing opportunities and policy subsidies. Despite these advancements, tax and procurement policies during this period lacked specificity in addressing the unique requirements of this emerging field, constraining incentives for broader participation.
Table 1.List of pilot cities.
- (3)
Deepening implementation phase (2017-present). This stage primarily focused on innovating financial products, enhancing financial efficiency, and addressing the shortcomings identified in the first batch of PPITF projects. Its objective is to promote scientific and technological innovation while advancing digital and intelligent financial services, thereby fostering deeper integration between technology and finance.
The CDG refers to a mode of integration of digital technologies and greenization concepts to achieve sustainable development (Kang & Shang, 2025). Digitization can break through geographical and spatial limitations, improve knowledge and data dissemination, enhance energy efficiency, and stimulate green technological innovation (Wu et al., 2024), while greenization cannot be separated from the support of environmentally friendly technologies (Huang et al., 2021), and the demand for green production also drives the construction and improvement of digital technologies (Zhang et al., 2025). First, according to the innovation systems theory, the effectiveness of innovation policies is closely related to their ability to promote interaction among different stakeholders in the innovation ecosystem. The PPITF can significantly optimize the innovation system for digital and green technologies by fostering close integration between innovation entities and financial institutions. This policy-driven coordination helps overcome obstacles in digital and green transformations by improving innovation resource allocation, accelerating knowledge transfer, and establishing collaborative mechanisms for problem-solving (Li et al., 2025; Wang et al., 2025a, 2025b). Second, under the ecological modernization framework, fintech platforms can incorporate non-financial indicators such as environmental pollution and carbon emissions—into credit evaluation systems (Liang et al., 2025). This approach enables differentiated investment allocation and compels high-emission industries to transition toward greener operations. Finally, from an integration perspective, the digital enablement of green transformation represents a process of systematic data resource optimization. Leveraging digital technologies such as big data analytics and artificial intelligence, fintech platforms can access extensive datasets, enhance information filtering accuracy, strengthen risk management capabilities, and foster the bidirectional and integrated development of urban digitalization and greenization, promoting their integrated and mutually reinforcing development. Therefore, this study proposes the following hypothesis:
Hypothesis 1 PPITF can significantly promote CDG.
Talent represents the core driving force behind knowledge creation and technological innovation. Urban CDG depends on professionals in related disciplines, while the PPITF fosters emerging industries and implements complementary talent policies to accelerate talent agglomeration. The PPITF transforms traditional employment patterns by applying information technologies and optimizing resource allocation, which accelerates the restructuring of regional development models and provides improved employment and research opportunities for skilled professionals. Furthermore, the implementation of the PPITF facilitates the development of comprehensive talent policies and associated support systems that offer competitive remuneration, financial incentives, and intellectual-property protection for innovators (Ayyagari et al., 2011). In addition, there are two primary pathways through which talent agglomeration promotes regional transformation. First, the concentration of professional talent enhances both formal and informal communication networks, accelerates the diffusion of professional knowledge and technical skills (Zhang & Guo, 2025), and reduces cost barriers to digital development. Second, talent agglomeration intensifies market competition and improves the efficiency of urban resource allocation. Consequently, enterprises characterized by high carbon intensity, low efficiency, and limited adaptability to transformation are gradually eliminated (Huang et al., 2023), thereby further promoting urban green development. Therefore, the following hypothesis is proposed:
Hypothesis 2 PPITF can promote CDG by exerting talent agglomeration.
The positive externalities and inherent uncertainties of innovation activities often cause actual R&D investment of innovative entities to deviate from the socially optimal level, information asymmetry between capital suppliers and demanders further exacerbates financing constraints. Consequently, digital and green innovation activities frequently encounter severe funding barriers (Liang et al., 2025; Wu et al., 2025). To promote technological innovation, the PPITF explicitly requires pilot cities to explore new mechanisms for fiscal science and technology investment and increase the intensity of financial subsidies. According to signal transmission theory, financial subsidies act as carriers for signaling. Through the implementation of the PPITF, pilot cities adopt the philosophy of digital and green development, continuously channeling financial subsidies toward digitalization and low-carbon technologies (Wei et al., 2023), and this mechanism enhances firms’ motivation to pursue green and digital transformations. Furthermore, the additional financial subsidies stimulated by PPITF exert a leveraging effect, mobilizing the “large capital” of the financial market through the “small funds” of government investment, and encouraging innovation entities within the jurisdiction to expand their investments in digital and green innovation. Therefore, the following hypothesis is proposed:
Hypothesis 3 PPITF can promote CDG by facilitating financial subsidies.
Given China’s vast territory and complex socioeconomic structure, the effects of the same policy may differ considerably across geographical region, fiscal pressure, and market size (Wang & Sun, 2025). First, the eastern region features a well-developed digital infrastructure, abundant digital talent and reasonable industrial structure (Ma et al., 2023), providing a favorable environment for implementing the PPITF and advancing digital or green transformation initiatives. Furthermore, with the continued promotion of the Western Development Strategy and the Belt and Road Initiative, the economic foundation and technological environment of the central and western regions have been significantly improved. Compared with the eastern region, financial service systems in the central and western regions remain less developed, while the implementation of PPITF can mitigate the high costs associated with recruiting top-tier talent and constructing innovation-related infrastructure (Wu et al., 2025), thus accelerating the industrial structural transformation in these areas. Therefore, the effect of the PPITF is expected to be more pronounced in the central and western regions than in the eastern region.
Second, the effectiveness of policy implementation is closely related to regional government capacity (Wang & Shao, 2025). The PPITF requires governments to continuously innovate fiscal investment mechanisms in science and technology and to expand the scale of financial subsidies, which the realization of this requirement is constrained by fiscal pressure. In cities with lower fiscal pressure, local governments can flexibly adjust the structure of fiscal expenditures and increase fiscal investment in science and technology, thereby enhancing regional innovation capacity. However, excessive fiscal pressure may restrict the scale of government spending, leading local authorities to prioritize self-interested budgets that emphasize production over innovation, while this budgets typically favor productive investments with immediate and high returns rather than long-term, high-risk digital or green projects (Kong & Zhu, 2022). Furthermore, excessive fiscal pressure weakens the funding capacity of the PPITF and bias support toward large, state-owned, or politically connected enterprises, thereby deviating from the policy’s intended objectives and weakening its overall effect. Therefore, the influence of the PPITF is expected to be weaker in regions experiencing greater fiscal pressure.
Third, demand-induced innovation theory posits that market size is a key determinant of technological innovation, as innovation based on a large market typically yields higher returns (Desmet & Parente, 2010). Armand & Mend (2018) found that the decline in aggregate demand during the Great Depression reduced corporate innovation activities, providing reverse support for demand-induced innovation theory. As one of the core drivers of urban CDG, the capacity of greenization to stimulate digitalization is closely linked to its return on investment. In regions with smaller market sizes, the potential gains from the green transformation of traditional industries are limited, thereby weakening their incentive to adopt digital technologies for green development. Under such conditions, the diffusion and application of digital technologies remain constrained, making it difficult to strengthen CDG. In summary, the effect of the PPITF is expected to be more pronounced in regions with larger market size.
Hypothesis 4 The effect of PPITF on CDG shows significant heterogeneity across geographical location, fiscal pressure, and market size.
As a widely used econometric framework for assessing policy effects, the basic principle of the DID method is to treat policy implementation as an exogenous event. This approach evaluates changes in outcomes between the treatment and control groups before and after policy implementation, thereby isolating the effects that can be attributed to the policy intervention. The PPITF displays the characteristics of a quasi-natural experiment. Considering that the PPITF was introduced in multiple stages, this study adopts a multi-period DID model for empirical estimation. The model is specified as follows:
Where CDGitindicates the collaborative development of digitalization and greenization in city iin year t. PPITFitrepresents if city i belongs to the pilot cites, its value is set to 1, and 0 otherwise. Controlitdenotes the set of control variables. λi and φt represent city-fixed effect and year-fixed effect, respectively. εitis the random disturbance term.VariableDependent variable
The two interrelated systems of digitalization and greenization jointly constitute the CDG, and these systems are inherently complex and multifaceted, making it difficult to measure CDG accurately using a single indicator (Kang & Shang, 2025). Therefore, the essential task is to establish and refine a comprehensive indicator system for digitalization and greenization before quantifying urban CDG.
Urban digitalization (DG) refers to the application of digital technologies, digital infrastructure, and digital talent in various fields to enhance urban intelligence, promote greenization, and achieve sustainable development. Drawing on existing research, this study constructs an evaluation index system for digitalization based on three dimensions: digital economy, digital hard environment, and digital soft environment (Ma & Lin, 2023; Hao et al., 2023; Borjigin et al., 2025). First, digital economy functions as the driving force of urban development. Numerous new industries and business models have emerged from digital technologies, injecting momentum into digital transformation, and is primarily measured by telecommunication service revenue and digital inclusive finance. Second, digital hard environment serves as a critical foundation for urban digital transformation and is measured by the number of Internet service and mobile telephone subscribers. Third, digital soft environment represents the institutional and human capital basis of urban digitalization and is evaluated through digital policy and digital talent. Digital policy is proxied by the total word frequency of digital economy policies and the ratio of digital economy policy word frequency, while digital talent is measured by the number of employees in the information transmission, computer services, and software industries. Finally, the panel entropy weight method is employed to calculate the comprehensive DG score.
Urban greenization (GRE) takes green production as the principal means, green lifestyle as the developmental goal, and green ecology as the foundational basis, which fully integrates the concept of “green development” into the processes of production, daily life, and ecological construction (Zhou & Qiao, 2022). Accordingly, this study constructs an evaluation index system for greenization across three dimensions: green production, green lifestyle, and green ecology, and employs the panel entropy weight method for quantification. First, green production is evaluated from the dual aspects of pollution emission and pollution control. The indicators include wastewater discharge per unit GDP, sulfur dioxide emissions per unit GDP, the comprehensive utilization ratio of industrial solid waste, and the centralized treatment ratio of wastewater at sewage plants (Ma et al., 2023). Second, green lifestyle is measured through green life management and green public services. Key indicators include the harmless treatment rate of domestic waste, annual per capita rides on buses and trolleybuses, and the number of buses and trolleybuses in operation at year-end. Finally, green ecology is assessed based on the area and degree of greenization, represented by the total area of green land, the area of park green space, and the proportion of green coverage in built-up areas (Wang & Liu, 2024). The comprehensive indicator system of CDG is constructed according to the above theoretical framework (see Table 2).
CDG’s indicator evaluation system.
The coupling coordination model is an effective analytical tool for measuring the degree of interaction and coordination among elements within a composite system (Zhang et al., 2025). Based on the established indicators of urban digitalization and urban greenization, this study applies the coupling coordination model to evaluate the level of collaborative development between the two subsystems (CDG). The specific formula is expressed as follows:
Cshows the coupling degree ranges from 0 to 1, which cannot indicate the individual development level of each system and only reflects their mutual influence. Therefore, a coupling coordination model is constructed to evaluate the coordination relationship between systems. The formula for the coupling coordination model between two systems is as follows:
CDGrepresents the coupling coordination degree. Cdenotes the coupling degree, and Trepresents the comprehensive coordination index of urban digitalization and urban greenization, indicating the contribution degree of their overall development of the coupling coordination degree. αand βdenote undetermined coefficients; since urban digitalization and urban greenization are equally important in high-quality development, so the undetermined coefficients are set to α=0.5 and β=0.5 (Zhang et al., 2025).
Independent variablePolicy data on the PPITF include the list of pilot cities and policy implementation timelines, reflecting the implementation of the Pilot Policy Promoting the Integration of Technology and Finance across different regions in China. The specific implementation timelines for PPITF in each region is determined by official government documents and announcements. The effect of PPITF is identified through the interaction between the treatment dummy variable and the policy timing variable. If a city becomes a pilot city, the value is 1 as the treatment group; otherwise, the value is 0 as the control group. Meanwhile, the policy timing variable takes the value of 1 for the policy year and subsequent years and 0 otherwise.
Control variablesTo eliminate problems caused by omitted factors and enhance the reliability and interpretability of the model, this study controls urban characteristic factors that may affect CDG to more accurately evaluate the effect of PPITF.
The level of economic development (GDP) is measured by per capita gross domestic product (Jiang et al., 2024). Industrial structure (INDS) is measured by the proportion of the tertiary industry’s output to GDP (Yu et al., 2025). Fiscal revenue (FR) is measured by the ratio of fiscal revenue to GDP. Innovation level (INN) is measured by the natural logarithm of patent authorizations plus 1. Financial development (FIN) is measured by the natural logarithm of total deposits and loans of financial institutions (Fu et al., 2024). Foreign investment (FDI) is measured by the logarithm of actual foreign investment utilized in that year.
Data sourcesSince PPITF has been implemented for more than 10 years, the sample was rationalized to ensure systematic assessment of its effects over a longer time span. Four steps were employed to process the sample data to enhance representativeness and scientificity. (1) Districts or counties were removed from the sample. (2) Exclude the sample with severely missing data. (3) Missing values were completed using the linear interpolation method. (4) All continuous variables were winsorized at the 1% level for both upper and lower bounds to reduce the influence of outliers. This process yielded panel data for 284 cities from 2007 to 2022. Among them, 50 pilot cities constituted the treatment group, and the remaining 234 cities formed the control group. Basic data were mainly obtained from the China City Statistical Yearbook, Government Work Report, National Economic and Social Development Statistical Bulletin, and statistical yearbooks of provinces, cities, and counties. The number of patent authorizations was derived from the CNRDS database. Table 3 presents the descriptive statistics of key variables.
Descriptive statistics.
Table 4 reports the regression results on the impact of PPITF on CDG. Column (1) presents the benchmark estimates without control variables but with city and year fixed effects. Column (2) includes both fixed effects and control variables. The results remain consistent: the coefficient of PPITF is positive and statistically significant at least at the 5% level. To enhance the robustness of the benchmark regression, Columns (3)-(6) show the estimation results of PPITF on DG and GRE before and after including control variables, indicating that PPITF has a promoting effect. Based on the economic interpretation of the coefficient in Column (2), PPITF promotes CDG growth by 2.4% compared with non-pilot cities. According to the descriptive statistics, the average value of CDG before the policy shock is 0.246, suggesting that PPITF improves CDG by 9.7% (0.0239/0.246 × 100%). Therefore, Hypothesis 1 is supported.
Benchmark regression.
Note: ***, **and *indicate significant at 1%, 5% and 10% significant levels. Robust standard errors are in parentheses. This notation applies to subsequent tables.
Satisfying the assumption of parallel trends is a prerequisite for the multi-period DID. In other words, the CDG values of the control and treatment groups should exhibit similar time trends prior to PPITF implementation, and this similarity ensures that any observed differences after the policy implementation can be attributed to the policy itself. Meanwhile, owing to the phased rollout of the policy intervention, the composition of city groups may vary in different phases. To address this issue, this study adopts the event-study method (Roth, 2022) and the year preceding the implementation of PPITF is taken as the benchmark year to account for heterogeneous timing of policy shocks.
As shown in Fig. 2, in the period before the PPITF takes effect (9 years pre-policy), the regression coefficients are close to 0 and statistically insignificant, confirming that the control (non-pilot cities) and treatment (pilot cities) groups exhibited consistent CDG trends before the policy implementation, satisfying the parallel trends assumption and supporting the applicability of the empirical model. The post-implementation period (11 years post-policy) reveals statistically significant positive coefficients, indicating that the pilot cities’ CDG significantly improved compared to non-pilot cities after policy implementation. This finding further confirms the positive effect of PPITF on urban CDG, with the effect emerging over time.
Propensity score matching
Although the multi-period DID method can identify the net effect of PPITF on urban CDG, while the PPITF is not allocated in a completely random manner, sample selection bias may arise from differences between pilot and non-pilot cities that extend beyond the policy itself. Following to the approach of Song et al. (2025), this study uses propensity score matching method to mitigate selection bias, because it can select samples closest to the features of the control and treatment groups based on observable covariates. The results are shown in Column (1) of Table 5, which show that the coefficients remain significantly positive, highly consistent with the previous empirical analysis, indicating that policy effect is not caused by pre-existing observable differences between groups. In conclusion, these results confirm the authenticity and reliability of the positive effect of the PPITF, and strengthen the credibility of the conclusion.
Robustness test.
In the multi-period DID framework employing traditional two-way fixed effect (TWFE), heterogenous treatment effects and biased estimation can be observed due to the “bad comparisons” problem (Baker et al., 2022). Referring to Goodman-Bacon (2021), this study initially identified the existence of “bad treatment groups” in Table 6. The results show that the negative weight is 0.003, which is approximately zero. The overall regression results are obtained through counterfactual testing by using the never-treated group as the control group, with a weight of 95.5%. This further confirms the methodological rigor of the benchmark regression analysis, with the results statistically confirming that results are unlikely to be driven by model errors or omitted variable bias. This rigorous research framework confirms the conclusion that PPITF has a significant and quantifiable effect on urban CDG.
Although the negative weights are relatively small, the non-zero result indicates that some comparisons in the multi-period DID model may include “bad controls”. Therefore, this study followed Cengiz et al. (2021) to re-estimate the benchmark results. Fig. 3 confirms that the parallel trends assumption holds before PPITF implementation. After the policy shock, CDG in treated cities significantly increased, verifying the robustness of the benchmark findings.
Exogenous policy shocks
To accurately evaluate the effect of PPITF, excluding the potential influence of other similar policies on urban CDG within the sample range is crucial. This study conducts robustness checks by incorporating two policies that are similar to the PPITF in promoting digital transformation and green innovation. The first is Made in China 2025 (MIC), which emphasizes the establishment of intelligent and green manufacturing systems, addressing technical barriers in industrial production and energy consumption, thereby facilitating green development. The second is the Data Trading Platforms (DTP), which plays a major role in promoting regional digital development and productivity, while supporting urban innovation and reducing pollution emissions. If these policies affect the urban CDG, they could potentially confuse the accuracy of the original policy’s effectiveness. Therefore, by incorporating both policies into the analytical framework to examine whether the effect of PPITF remains significant. According to the Columns (2)-(4) of Table 5, the coefficients of PPITF remain significantly positive even after controlling for these two parallel policies. This further confirms the authenticity of the policy effect and the reliability of the findings.
Endogeneity testAlthough this study uses the multi-period DID method and controls for relevant variables that may affect urban CDG, endogeneity problems cannot be completely ruled out, thus this study conducts robustness checks by reverse causality and instrumental variable methods. On the one hand, a city’s CDG level may influence its likelihood of being selected as a PPITF pilot city, namely the primary threat to the accuracy of the policy estimation lies in the potential endogeneity of the PPITF variable. To address this issue, this study adopts the approach of Beck et al. (2010), constructs the Weibull hazard mode and conducts a reverse causality test. As shown in Column (1) of Table 7, the coefficient of CDG is statistically insignificant, which indicates that there is no evidence to support the endogeneity issue caused by reverse causality.
Endogeneity test.
Note: Cities are dropped from the sample once they become the pilot cities for PPITF policy in the hazard model. Therefore, there are 3521 observations.
On the other hand, due to the selection of the pilot cities is not random, factors such as the economic foundation, resource endowments, and financial status are considered. Meanwhile, many factors can affect urban CDG, and it is challenging to exhaustive by controlling variables, leading to the problem of missing variables. Referring to Zhou et al. (2025), this study adopts the spherical distance (Distance) between the city and Hangzhou as an instrumental variable (IV) for the PPITF, this instrumental variable is closely related to the fact that the city has become a pilot city for the PPITF. Given that e-commerce enterprises in Hangzhou have demonstrated a leading position in financial development, which indicates that the closer to Hangzhou, the better development of tech-finance should be (Zhou et al., 2025). From an independence analysis, the smaller distance from Hangzhou does not mean the better digitalization and greening synergies, thus this instrumental variable satisfies the exogeneity assumption. Column (2) of Table 7 shows the coefficient of instrumental variables is significantly negative in first stage, indicating that the smaller distance from financial development center is associated with a higher possibility of a city being selected as a pilot city, which is consistent with the theoretical expectations. The Cragg-Donald Wald F-value is 153.228 more than 10, indicating that the weak instrumental variable problem is not present. Column (3) of Table 7 indicates the second stage result is significantly positive, which is consistence with the previous benchmark regression finding. These results reinforce the fact that even after addressing potential endogeneity issues with the instrumental variables, urban CDG can still benefit from the PPITF.
Dual machine learningPrevious tests have verified the authenticity and reliability of the benchmark regression findings. However, traditional regression models are plagued by the “curse of dimensionality” and multicollinearity in variable section and model estimation, which may lead to biased estimation results. Dual machine learning (DML) model can effectively address these issues and has the advantage of avoiding model setting bias in dealing with nonlinear data (Bodory et al., 2022). Considering that the control variables may have a non-linear impact on urban CDG, following Kang & Shang (2025), the DML model is constructed using the Random Forest algorithm, with the sample split ratio set to 1:4, and gradually incorporate the linear, quadratic, and cubic terms of the control variables into regression, the results are shown in Columns (5)-(7) of Table 5. The coefficients of PPITF remain significantly positive, in align with the previous empirical analysis, further verifying the reliability of the benchmark regression conclusions.
Mechanism testPrevious studies have extensively employed structural equation models to examine mediating effects. However, relying solely on the traditional exogeneity assumption is insufficient for determining causal mechanisms, as average-effect estimates become biased when the mediator and explanatory variable interact, while the causal mediation analysis can overcome these limitations. Therefore, following Imai et al. (2010), this study constructs a causal mediation model to identify the transmission mechanisms by decomposing the total exposure effect into random and indirect effects. Mi(e) represents the value of the mediating variable (M) when PPITF=e, and CDGi(e,m) represents the value when the treatment variable (PPITF) equals e and the mediating variable (M) equals m. The effect of PPITF on CDG is defined as the average treatment effect (ATE), calculated as:
If ATE exists, the M has two potential values: Mi(1)and Mi(0).Correspondingly, the values of the outcome variable are CDGi(e,Mi(1)) and CDGi(e,Mi(0)). Meanwhile, if e=1 or e=0, then the average causal mediation effect (ACME) can be defined as:
Given that e=1 or e=0, the average direct effect (ADE) can be defined as:
Generally, ATE can be obtained using formula (1), and the identification procedures for ACME and ADE are as follows:
Where Mrepresents the mediating variable, only CDGi(PPITFi,Mi(PPITFi))can be observed, while CDGi(PPITFi,Mi(1−PPITFi))cannot be observed. Therefore, an estimation method based on quasi-Bayesian Monte Carlo approximation is applied. The procedure is as follows: fit Eqs. (5) and (6), then simulate the potential values of the mediating variable based on the sampling distribution of the model parameters. Next, calculate the causal mediating effect; the valid estimate of ACME is φ1χ2, and ADE is χ1.
Furthermore, this study adopts the traditional mediation analysis to examine the influence of PPITF on the two mediating variables, further verifying the robustness of the causal mediation analysis. Meanwhile, the Bootstrap and Sobel tests are applied for robustness testing to address the limitations of the causal step method. Table 8 presents the results of the mediation effect test.
Mechanism test.
Note: ACME indicates the average causal mediating effect. ADE represents the average direct effect. ATE indicates the average total effect. Bootstrap has 1000 repetitions.
This study follows Wang & Liu (2024) in calculating the ratio of total employment in information transmission, computer services and software, scientific research, technical services and geological surveys, education, culture, sports and entertainment, leasing and business services, and finance to the total number of urban employees to measure talent agglomeration (TA). The results in Column (1) of Table 8 indicate that the ACME of PPITF influencing CDG through TA is 0.0154, and the 95% confidence interval lies above 0, confirming the existence of a mediating effect. This finding suggests that PPITF enhances CDG by promoting TA. Furthermore, according to the Column (3) of Table 8, the coefficient of PPITF on TA based on the traditional mediation effect test is 0.0094. Columns (5) and (7) present the results of the Bootstrap and Sobel tests, respectively, confirming that the indirect effect is significant, the confidence interval does not include 0, and the Sobel test is significant at the 1% level, again confirming the mediating effect of TA. The PPITF cultivates more highly educated talents by promoting joint talent programs between financial institutions and universities. It also promotes balanced development of related industries and stimulates demand for innovative talents, which facilitates the flow of digital and green knowledge and the diffusion of technologies. Hypothesis 2 is verified. These results are consistent with previous findings indicating that PPITF promotes digital technological breakthroughs by enhancing human capital (Wu et al., 2025). Additionally, this study confirms the impact of PPITF on the synergies between digitalization and greenization, and expands the research scope to prefecture-level cities.
Following Wang & Guo (2023), this study measures financial subsidies (FS) as the proportion of local government science and technology expenditure in total government expenditure. Column (2) of Table 8 shows that the ACME is positive, while ADE and ATE are 0.0503 and 0.0554, respectively; their 95% confidence intervals are all above 0, indicating that the mediating effect of FS is established. Furthermore, the coefficient of PPITF on FS from the traditional mediation test is 0.0052 and statistically significant in Column (4), confirming that the enhancement of FS is a practical path for PPITF to promote CDG. The results in Columns (6) and (8) indicate that FS passes both the Bootstrap and Sobel tests, confirming that the mediating effect is robust. As a policy supporting technological innovation, the PPITF requires pilot cities to continuously innovate in science and technology investment and increase the intensity of FS, which plays a guiding role under regulatory supervision. Hypothesis 3 is verified. Furthermore, previous studies have demonstrated the effectiveness of FS in promoting green total factor productivity and digital innovation (Wang & Guo, 2023; Tao et al., 2025). Therefore, this study concludes that PPITF enhances CDG by stimulating FS.
Heterogeneity testMunicipalities located in different geographical regions exhibit significant variations in economic development and resource endowment, which may influence the implementation and effectiveness of the PPITF. This study divides cities into two regions—eastern and central-western, based on the classification standards of the China National Economic and Social Development Statistics Bulletin. Columns (1) and (2) of Table 9 show that PPITF exerts a significantly positive effect in both regions, with a greater impact observed in the central and western regions. This is attributed to the more reasonable industrial layout and well-developed digital infrastructure in eastern region (Ma et al., 2023), which provides strong support for PPITF implementation. In contrast, the central and western regions lag in innovation resources, production efficiency, and economic growth, while the PPITF fosters a favorable environment and financial support for urban digitalization and low-carbonization, making its enabling effects more pronounced in these regions.
Heterogeneity analysis.
Following Wei et al. (2023), fiscal pressure is measured by the ratio of the difference between fiscal expenditure and fiscal revenue to fiscal revenue, and the sample is divided into high and low groups based on the median of city-level annual fiscal pressure. The results in Columns (3) and (4) of Table 9 show that PPITF significantly promotes CDG in both groups, with a more pronounced effect in cities experiencing lower fiscal pressure. This outcome may be attributed to the fact that cities with lower fiscal pressure possess relatively abundant financial resources compared with those under higher fiscal pressure, laying a solid foundation for government-driven innovation and providing stronger material support for PPITF-induced technological progress.
Following Liu (2013), this study uses the total wages of employed staffs in urban areas to measure market size and divides the sample into groups above and below the median of city-level annual market size. The results in Columns (5) and (6) of Table 9 show that the coefficients for both groups are significantly positive, but the effect is more pronounced in cities with larger market size. This finding supports the view that the investment return level influences innovation activities (Armand & Mend, 2018), as cities with smaller market size yield insufficient potential returns from green development, thereby weakening the willingness of innovators to advance green development through digital technology and limiting further urban digital progress. Furthermore, this study extends the theoretical perspective to the synergy between digitalization and greenization in emerging economies. In conclusion, Hypothesis 4 is verified.
Conclusions, policy recommendations, and research limitationsResearch conclusions and theoretical comparisonThis study regards the PPITF as a quasi-natural experiment and employs a multi-period DID model with panel data from 284 Chinese cities spanning 2007 to 2022 to comprehensively assess the impact of PPITF on the CDG. The empirical findings are statistically robust and carry both theoretical and practical implications. The main conclusions are as follows.
First, the benchmark regression results indicate that the coefficient of PPITF is significantly positive, providing preliminary evidence that the policy significantly promotes urban CDG. Furthermore, this study conducts multiple robustness and validity tests, including the parallel trend test, heterogeneity treatment effect, and endogeneity test, to ensure that the findings possess causal interpretability and strong explanatory power. The results verify and extend existing research demonstrating that PPITF promotes corporate green innovation, facilitates digital technological breakthroughs, and reduces carbon dioxide emissions (Yang et al., 2025; Wu et al., 2025; Yu & Zhang, 2025). Compared with previous studies, this research reveals the effect of PPITF on urban CDG and explores its regional heterogeneity, providing a valuable complement to existing policy analyses, and offering reference for more comprehensive empirical strategies through rigorous robustness verification.
Second, the mechanism analysis identifies two distinct pathways through which the PPITF enhances urban CDG. The first operates through talent agglomeration, which reduces the cost of skilled labor for digital innovation and provides intellectual support for digital-driven green transformation. The second functions through financial subsidies guide investment toward low-carbon technologies, promote environmentally responsible production practices, and facilitates digital transformation, thereby strengthening the innovation incentives of participating entities. These findings extend the understanding of the green innovation effects of PPITF (Li et al., 2025; Liang et al., 2025; Yang et al., 2025) and through the perspective of signal transmission theory, clarify the mediating role of financial subsidies, enriching the theoretical explanation of the policy’s influence mechanism.
Finally, the impact of the PPITF on urban CDG varies according to geographical location, fiscal pressure, and market size. The policy effect is more pronounced in central and western cities, as well as in cities with lower fiscal pressure and larger market size. Although previous studies have examined the heterogeneity effects of similar policies on firms’ green innovation and digital technology breakthrough from the perspectives of region, industry, and ownership, few have explored the heterogeneity effects of policy implementation under government-level (fiscal pressure) and market-level (market size) conditions. By incorporating these contextual variables, this study systematically analyzes the heterogeneity characteristics of policy effects and enriches the literature on enhancing policy effectiveness under varying regional and institutional environments.
Policy recommendations- (1)
The central government should summarize the successful experiences of existing pilot cities, develop a replicable, scalable, and operational model, and encourage additional cities to participate in the pilot program. Meanwhile, the digitalization and greening synergies should be incorporated into policy-effect evaluations, and a more scientific and verifiable dynamic evaluation system should be established to regularly monitor and adjust implementation, thereby ensuring the simultaneous realization of policy effectiveness and twin transition objectives. Additionally, fintech companies should seize the policy window by actively engaging in pilot programs, integrating into policy chains, developing specialized financial products, and establishing risk-control alliances. Through these measures, they can transition from being mere “fund providers” to becoming “policy partners,” achieving leapfrog development amid the wave of digitalization and green synergy. Meanwhile, fintech companies should align their strategies with government-led technology investment, leverage technologies such as artificial intelligence to build project selection platforms, incorporate non-financial indicators such as environmental pollution and carbon emissions into credit assessment models, strengthen market-based screening mechanisms, and prevent short-term, low-level projects from obtaining policy support, thereby ensuring the precise allocation of financial resources.
- (2)
The policy pilot’s national promotions should be intensified, with the focus on strengthening its crucial role in exerting talent agglomeration and facilitating financial subsidies. On the one hand, continuously implement the talent development strategy and optimize talent recruitment and training. Pilot areas should formulate targeted special policies and provide preferential treatment in areas such as housing subsidies, competitive salaries, children's education, research funding and other benefits. Relying on local universities, vocational colleges, and leading enterprises to co-build training bases and industry-university-research platforms, fostering customized talent aligned with regional development needs. Meanwhile, optimize the business environment and public services, improve talent evaluation and incentive mechanisms, facilitate talent development pathways, and promote talent agglomeration. On the other hand, local governments should focus on key areas such as green development and digital innovation, establish differentiated subsidy standards, and build a one-stop service platform. Furthermore, local governments should fully leverage the effectiveness of by employing government-guided funds and financial instruments to drive digital and green innovation, maximizing the leveraging effect of fiscal funds to attract greater financial resources and private investment in major digital and green energy projects, thereby facilitating twin transition synergies.
- (3)
Policy implementation should be tailored to regional characteristics, considering geographical location, fiscal pressure, and market size. First, the eastern cities should leverage their strengths in talent and capital to bolster support for the digitalization and greening synergies. Simultaneously, eastern cities can enhance experience exchange and technological collaboration with central and western cities, capitalizing on their comparative advantages to promote coordinated regional development. The central and western cities should seize institutional innovation opportunities, draw lessons from the policy experiences of the eastern cities, further release policy dividends, and capitalize on their late-developing advantage in policy empowerment. Second, the central government should attempt to reform the current financial and taxation system, appropriately alleviating the fiscal pressure on local governments, rationally planning existing fiscal expenditures through laws and regulations, and encouraging competition among local governments in digitalization and greening synergies. Meanwhile, the central government should incorporate indicators such as green development, environmental protection, and digitalization into the promotion assessment system for officials, thereby preventing local governments from merely pursuing GDP growth while neglecting coordinated development. Furthermore, local governments should optimize fiscal revenue and expenditure structures, enhance fiscal fund utilization efficiency by improving intergovernmental fiscal relations, refining the tax system, and perfecting transfer payment mechanisms, thereby reducing the distortionary impact of fiscal pressure on government actions. Finally, cities with insufficient market consumption potential need to promote the construction of digital infrastructure, improve the income distribution system and social security system, increase the income level of residents, and create diversified business models guided by application scenarios, fully stimulate consumption potential, and create necessary market space for regional digitalization and greening synergies. Furthermore, with the support of relevant finance and taxation policies, local governments should guide digital enterprises to pay attention to residents’ green low-carbon consumption demands, and leverage their own advantages to explore green low-carbon digital technologies, thereby achieving sustainable digital development.
Although this study provides empirical evidence highlighting the influence of PPITF on city-level CDG, it represents only one analytical dimension. First, while valuable insights are derived from city-level data, the findings may lack generalizability. Future research should incorporate more granular datasets, covering firms, industries, and cross-country comparisons, and adopt diverse empirical methodologies to validate and enrich the conclusions. Second, this study employs a multi-period DID model to examine the effects of PPITF and uses reverse-causality tests, double machine learning, and other approaches to mitigate estimation bias, while potential selection bias may persist. Future research could adopt multiple-case study designs or expand the dataset scope to address this concern, thereby enhancing causal inference and result robustness. Third, the variables were primarily measured using statistical data, which may not fully capture the multidimensional characteristics of CDG. Future research should integrate multi-source datasets and employ emerging data-collection technologies to obtain more comprehensive and precise information.
FundingThis work was supported by the Ministry of Education Humanities and Social Science Fund Youth Project (24YJC630035); Youth Project of Hebei Natural Science Foundation (G2023203013; G202403010); Scientific Research Project of Colleges and Universities of Hebei Province (SQ2024215).
Ethics approvalNot applicable
Consent for participateNot applicable
Consent for publicationNot applicable
CRediT authorship contribution statementPanpan Li: Writing – original draft, Supervision, Software, Methodology. Xiaozhou Ding: Writing – original draft, Supervision. Tao Guo: Supervision.
We declared that we have no conflicts of interest in this work.
We are very grateful to editors and anonymous reviews for reviewing this paper.
