Research on environmental performance generally focuses on two levels: regional and enterprise. At the regional level, factors affecting environmental performance mainly include national environmental audits (Li and Sun, 2019), government environmental audits (Zeng and Li, 2018), and coordinated efforts to promote pollution reduction and carbon emissions (Zheng and Zhang, 2024). These factors work together to enhance the environmental behavior of enterprises in the region and promote the implementation of green development policies. At the enterprise level, the main factors affecting environmental performance include the introduction of digital products (Shao and An, 2023), government environmental subsidies (Wang and Zheng, 2020), green human resource management (Zahid et al., 2020; Gill et al., 2021), and green innovation (Singh et al., 2020; Rehman et al., 2021). These factors improve environmental performance by optimizing production methods, enhancing resource utilization efficiency, and reducing pollution emissions.

Institutional innovation affects environmental performance through multiple channels. For example, it can enhance resource allocation, strengthen environmental regulations, and improve government governance capacity, thereby fostering proactive involvement and synergy in environmental governance between enterprises and local governments (Han et al., 2023; Bai and Ding, 2023). In this process, the quality of institutions directly impacts the ability of economic entities to coordinate in their production and business activities, which in turn influences the improvement of environmental performance (Musa et al., 2021). Institutional innovation can also enhance environmental performance by optimizing the market system. China has long leveraged institutional innovation to foster enterprise development and refine the market economy system (Sun et al., 2024). providing enterprises with clearer environmental requirements and market incentives. These innovations have not only strengthened corporate environmental responsibility but also facilitated the widespread adoption and application of green technologies. Institutional innovation is also reflected in the collaborative cooperation among various stakeholders within a region. In regional environmental governance, a cooperation mechanism involving the government, enterprises, and social organizations has gradually taken shape, driving the joint achievement of pollution reduction and green innovation (Zheng and Zhang, 2024).

According to the environmental spillover effect theory of Copeland and Taylor (2004), international trade may affect national environmental protection policies through the “environmental spillover effect”. The study of Assogbavi et al. (2023) shows that international trade not only directly affects the economic activities of various countries, but also indirectly affects the global distribution of carbon emissions. The open international market may push the country to strengthen environmental regulation to cater to the global market demand for green products. With the development of globalization and international trade, the role of service trade in promoting economic growth and environmental protection has received increasing attention. Zhao and Fang (2024) studied the practical effects of service trade innovation in China’s A-share listed companies, focusing on energy conservation and emission reduction. Their research shows that service trade not only promotes the structural transformation of the economy, but also contributes to the achievement of energy conservation and emission reduction targets to some extent. Therefore, the role of international trade in promoting global environmental cooperation, especially how to reduce negative impacts through international agreements and green trade policies, is a direction worthy of further exploration.

Although existing studies have extensively explored the role of institutional innovation in environmental protection, there are relatively few studies on the relationship between institutional innovation and environmental performance in the field of service trade. Especially in China, with the promotion of “carbon peak” and “carbon neutral” goals, green transformation and sustainable development in the field of service trade have become urgent issues to be solved. China launched the service trade innovative development pilot policy in 2016, which is seen as an institutional innovation aimed at promoting the green development and international competitiveness of the service industry. Existing studies have shown that institutional innovation can promote the transformation and innovation of the service industry in environmental governance by improving the efficiency of resource allocation, optimizing industrial structure, and promoting the introduction of green technology (Dai and Ma, 2024). This provides a new perspective and practical basis for studying the impact of institutional innovation on environmental performance in the field of service trade. By studying the impact of service trade innovative development pilot policies on environmental performance, this paper hopes to fill the research gap on the relationship between institutional innovation and environmental performance in the field of service trade, and provide policy inspiration for the green development of service trade in China and even the world. Through quantitative and qualitative analysis, we hope to reveal the mechanism and path of institutional innovation in promoting green transformation and improving environmental performance. This study not only expands the research on the relationship between institutional innovation and environmental performance, but also provides a new theoretical basis for the interaction between service trade and green development.

The National Development and Reform Commission of China, in its “Outline for the Innovative Development of the Services Sector (2017–2025),” advocates for the acceleration of the innovative advancement of the services sector. This initiative aims to bolster the emergence of new growth drivers for the service economy, serving as a pivotal strategy to comprehensively enhance the nation’s overall strength, international competitiveness, and capacity for sustainable development. In this context, to enhance China’s service trade promotion system and establish a hub for service trade system innovation, in February 2016, the State Council endorsed the Pilot Program for Innovative Development of Trade in Services. This initiative approved the launch of pilot programs for the innovative development of trade in services across 15 regions (Specifically, these include: Tianjin, Shanghai, Hainan, Shenzhen, Hangzhou, Wuhan, Guangzhou, Chengdu, Suzhou, Weihai and Harbin New District, Nanjing Jiangbei New District, Chongqing Liangjiang New District, Guizhou Guian New District, and Shaanxi Xixian New District), including Tianjin. It also stipulated the exploration of policies and institutions tailored to foster the innovative development of the service trade sector. In June 2018, the State Council endorsed the “Comprehensive Plan for the Deepening Innovative Development of the Trade in Services Pilot Program.” This initiative led to the inclusion of two new pilot zones in Beijing and the Xiong’an New Area, while also extending the scope of the existing pilot programs in Nanjing Jiangbei New Area and Harbin New Area to encompass the entire city. The directive aims to encourage these pilot zones to delve into the establishment of institutional mechanisms, policy measures, and opening strategies that are tailored to the innovative and developmental needs of the service trade sector. The ultimate goal is to foster the emergence of a new paradigm for comprehensive opening-up. The “Pilot Overall Program” has been expanded to encompass 28 regions, with a focus on fully leveraging the supportive role of the service trade sector in stabilizing foreign trade and foreign investment. It also aims to foster the transformation and upgrading of foreign trade, as well as to promote high-quality development.

After analyzing the policy documents, we found that once a city is selected as a pilot for the innovative development of service trade, a series of measures will be taken to promote green innovation and optimize the industrial structure. Specifically:

Combining financial innovation with policy support to ease financial pressure and incentivize green innovation: Pilot cities have implemented financial innovation measures and policy support to ease financing difficulties and incentivize green innovation. These measures include encouraging financial institutions to support service enterprises in pilot cities, particularly small and medium-sized enterprises that align with industrial policies, as well as providing financial support for key areas such as research and development, energy conservation, environmental protection, and environmental services. For example, after becoming a pilot city, Shanghai set up a service trade innovative development fund to attract private capital into key service trade areas. In 2018, Weihai relaxed the list of pilot enterprises for service trade branches, expanded the reach of cooperative banks, and increased financing channels for enterprises. Additionally, technologically advanced service enterprises and technology-based SMEs (small and medium-size enterprise) in the city can enjoy tax incentives that encourage the introduction of high-end service products and green technology R&D. After being selected, Chongqing Liangjiang New Area expanded the scope of technologically advanced service enterprises and technology-based SMEs, and provided tax incentives, implementing zero tax rates and tax exemptions for cross-border VAT (value-added tax) on services.

Fostering new industries and business models to promote industrial upgrading: The selected pilot cities actively adapt to emerging trends in service trade, strengthen cross-sectoral cooperation by deepening reforms and fostering continuous innovation, and cultivate new industries and business models to promote the upgrading of industrial structures. For example, Hainan leverages Haikou High-tech Zone and Hainan Ecological Software Park to build characteristic service outsourcing industrial clusters. Wuhan emphasizes the development of technology services, computer and information services, financial services, cultural services, and other emerging fields. Tianjin, by accelerating the construction of the “Double Innovation Special Zone,” promotes the agglomeration of high-tech services and vigorously develops new business models such as cloud computing and big data.

Drawing on new institutional economics theory and green innovation theory, and considering the dual mechanisms of institutional innovation and industrial upgrading, this study explores how service trade innovative development pilot policies can enhance the environmental performance of pilot areas through green innovation and industrial upgrading. New institutional economics theory emphasizes the guiding role of institutions in economic behavior, arguing that institutional innovation is a key driver of economic growth and industrial upgrading. Particularly, service trade innovation policy, as a form of institutional innovation, can effectively change enterprise behavior and promote green innovation and industrial transformation by optimizing market rules, policy frameworks, and incentive mechanisms. These policies support green technology innovation by fostering technological advancements, capital flows, and market openness through institutional arrangements. Green innovation theory posits that technological innovation, especially green technological innovation, plays a crucial role in improving environmental performance and achieving sustainable development. Institutional innovation further supports the development and application of green technologies through knowledge spillover, technology diffusion, and policy incentives. Under the service trade innovative development pilot policies, regions can promote the innovation and application of green technologies by lowering technical barriers, accelerating technology transfer, and strengthening cross-border collaboration.

Combining these two theoretical perspectives, this study views service trade innovative development pilot policies as institutional innovation, which ultimately improves the environmental performance of pilot areas through two mechanisms: enhancing green innovation capabilities and promoting industrial upgrading.

This study examines the institutional innovation of pilot policies for the innovative development of trade in services as a quasi-natural experiment. It designates regions with established pilot policies for the innovative development of trade in services as the treatment group, and regions without such policies as the control group. The study employs a staggered difference-in-differences methodology to assess the impact of these institutional innovations on environmental performance. Given the delayed initiation of the third batch of pilot regions, this paper primarily concentrates on the initial two groups, comprising a total of 17 regions. Taking into account the regulations aimed at reducing sulfur dioxide emissions and preventing and controlling air pollution during the “11th Five-Year” plan period, along with the significant amount of missing data for certain variables, and to mitigate the potential impact of the COVID-19 pandemic on various aspects of service trade and empirical outcomes (Mao and Guan, 2022; Li and Ma, 2023), the research period has been established from 2011 to 2019. Among them, the list of pilot regions and their respective implementation timelines are derived from the documents released by the Ministry of Commerce, as well as from each pilot region; the carbon emission data were obtained from the China Carbon Accounting Databases (CEADs); and other data were obtained from the database of the China Research Data Service Platform (CNRDS), the Statistical Yearbook of Chinese Cities, the Statistical Yearbook of each province, and the Statistical Yearbook of each city.

Before the empirical analysis, the samples were processed as follows: First, in order to ensure the comparability between the treatment group and the control group, 70 large and medium-sized cities and 12 port cities in China are used as the base samples. One reason is that the selection of pilot sites for the innovative development of trade in services across the nation is not entirely random. It primarily takes into account regions with a well-developed service industry and service trade, predominantly opting for regional central cities. Therefore, it is difficult for most cities to obtain pilot services trade innovation development (Chen, 2021). China’s 70 large and medium-sized cities are basically national or regional center cities, with strong homogeneity among cities (Cao, 2020). The second reason is that the pilot zones are likely to be urban areas characterized by superior technological infrastructure and a higher degree of international integration. ¹ Therefore, referring to the study conducted by Wang and Meng (2013), 14 new inland river ports and 17 coastal port cities are taken into consideration; however, only 12 port cities are actually added to the list, as some of them already fall within the category of the 70 large and medium-sized cities mentioned previously. ² The final reason is that due to insufficient data, both Sanya and Dali are excluded from the list of 70 large and medium-sized cities. Consequently, the foundational research sample for this paper comprises 80 cities. Second, considering the availability of pilot data for state-level new areas and the heterogeneity of control policies, this paper excludes Harbin, Nanjing, Chongqing, Guiyang, and Xi’an from the original list of 80 cities. After screening, the final study sample was 75 cities, of which the treatment group was 11 cities: Tianjin, Shanghai, Haikou, Shenzhen, Hangzhou, Wuhan, Guangzhou, Weihai, Chengdu, Suzhou and Beijing.

Since the pilot policy on innovative development of trade in services has been implemented in stages, this study employs a staggered difference-in-differences approach and utilizes a two-way fixed-effects model to assess its impact on environmental performance. The econometric model is set as Equation 1: ln ⁡ E P i t = α + β d i d i t + γ C o n t r o l s i t + μ i + λ t + ε i t (1) where i and t represent region and year, respectively. The explanatory variable ln ⁡ E P i t is regional environmental performance. The core explanatory variable d i d i t indicates whether or not institutional innovation was implemented in region i in year t. The core explanatory variable is a control variable representing the series of other factors that may affect environmental performance. C o n t r o l s i t is a control variable representing a series of other factors that may affect environmental performance. μ i represents the region fixed effect, which is used to control the factors of regional characteristics that do not change over time; λ t represents the year fixed effect, which is used to control the factors such as macroeconomic fluctuations that do not change over time; ε i t is the error term. The focus of this paper is on the coefficient β , β should be significantly positive if institutional innovation improves environmental performance.

Table 2 reports the baseline results on the impact of institutional innovation in trade in services on regional environmental performance. Column (1) of Table 2 controls only for the core explanatory variables did and fixed effects, while column (2) further incorporates all control variables. From the regression results, it can be found that the coefficient of the institutional innovation in trade in services variable (did) is significantly positive at the 1% level regardless of the inclusion of control variables, which suggests that institutional innovation contributes to the environmental performance of the pilot regions. For instance, in column (2), it is observed that institutional innovation results in a 0.4635 percentage point enhancement in environmental performance within the pilot region relative to the control group. Consequently, this demonstrates that institutional innovation in the trade of services positively influences the environmental performance in the pilot region, thereby corroborating the research hypothesis.

The application of staggered DID requires the fulfillment of the parallel trend assumption, and there may be dynamic effects on the impact of institutional innovation in trade in services on environmental performance. Therefore, this paper draws on Liu and Mao (2019) to construct a model to test the parallel trend assumption and identify the differences in the impacts of different years of policy implementation on environmental performance by taking the previous year of institutional innovation implementation as the base period, and the model is set as Equation 2: l n C i i t = α + β k ∑ k = − 8 k = 3 d i d i t k + γ C o n t r o l s i t + μ i + λ t + ε i t (2)

Where k = 0 in the year of institutional innovation implementation, β k indicates the difference in environmental performance between the treatment and control groups in the kth year of institutional innovation implementation compared to the previous year of institutional innovation implementation. When k < 0, β k is not significantly different from 0, it means that the parallel trend assumption is satisfied; when k ≥ 0, β k portrays the dynamic effect of institutional innovation. Due to the sample size problem, periods 8 and 7 before the implementation of institutional innovation are combined into period 6 before the implementation, and period 3 after the implementation of institutional innovation is combined into period 2. As illustrated in Figure 1, there is no significant difference between the environmental performance of the treatment group and the control group before the implementation of institutional innovation, which satisfies the parallel trend hypothesis. Due to the lag of policy effect, environmental performance was significantly improved only in the second year after the implementation of institutional innovation, which verified the existence of environmental effects of institutional innovation in trade in services ³ .

For the staggered DID model, the traditional two-way fixed effects estimator has been mostly used in the previous literature. The validity of these traditional two-way fixed effects estimators is based on the assumption of homogeneity of treatment effects. However, in the presence of heterogeneous treatment effects, the traditional two-way fixed effects estimators can be biased (De Chaisemartin and D’Haultfoeuille, 2020; Goodman-Bacon, 2021). To solve this problem, the existing literature puts forward three solutions (group-period average treatment effect estimator, interpolation estimator, and stacked regression estimator), each of which has its own advantages and disadvantages. The core idea, however, is to find a reasonable control group or use the control group to calculate a reasonable counterfactual result.

In this paper, the potential heterogeneous treatment effects may come from “bad control group” and “dynamic policy effect”. First, due to the inconsistent implementation times of the service trade innovation development pilot, cities that set up the pilot earlier (such as Chengdu and Suzhou) may serve as the control group for cities that established the pilot later, which could lead to estimation bias. Second, in a dynamic context, the estimated coefficients for each phase may be cross-contaminated across periods, making them difficult to explain, which indicates that the pilot policy for the innovative development of trade in services exhibits significant temporal dynamics. Considering the small sample size of this paper, a large number of samples will be lost when calculating the group-period treatment effect, which may affect the estimation efficiency, and the stacked regression estimator may cause the problem of data reuse (Liu et al., 2022). Therefore, this paper uses the two-stage difference-difference method in the interpolated estimator to evaluate the impact of institutional innovation in trade in services on environmental performance (Gardner, 2022). Specifically, in the first stage, the city and year fixed effects are estimated using the unprocessed observation data ( d i d = 0), as shown in Equation 3, and the parameters γ ∧ , μ ∧ i and λ ∧ t are estimated. In the second stage, the dependent variable with the fixed effects removed and the policy treatment variable are both used in regression to identify the average treatment effect, as shown in Equation 4, where θ is the robust estimator after correcting the estimation bias of the traditional bidirectional fixed effect difference-in-differences model. l n C i i t = α + γ C o n t r o l s i t + μ i + λ t + ε i t (3) ln ⁡ C i ∧ i t = l n C i i t − γ ∧ C o n t r o l s i t − μ ∧ i − λ ∧ t = θ d i d i t + ε i t (4)

Based on Formulas 3, 4, the impact of institutional innovation in trade in services on environmental performance is evaluated. The regression results are shown in Table 3. It can be found that in the case of using the staggered differential robust estimator, regardless of whether the control variables are added, the institutional innovation of trade in services still significantly improves the environmental performance of the pilot area. It indicates that the heterogeneity treatment effect has limited impact on the estimation results, which verifies the reliability of the findings.

Considering that the selection of cities for the implementation of institutional innovations in trade in services may not be completely random, this part uses two-stage least squares (2SLS) and PSM-DID to test the possible endogeneity of the policy variables. Firstly, combining with existing research, the interaction term of geographic centrality and the lagged one-period value added of the tertiary industry is adopted as an instrumental variable to conduct the 2SLS estimation (Guo and Lan, 2021) ⁴ . In general, cities with higher geographical centrality have more accessible regional networks, better urban openness, and are more likely to implement institutional innovation. Moreover, the geographical index is strictly exogenous, does not affect regional environmental performance, and meets the requirements of relevance and exclusivity. The results in column (1) of Table 4 show that the estimation results of the first stage. The K-P LM and K-P Wald F tests indicate that the model is identified and the instrument variables are strong. The instrumental variable (DistanceStr3) is a strong and significant predictor of the policy variable. Column (2) shows the estimation results of the second stage. The coefficient of the core explanatory variable remains significantly positive. Secondly, in order to make the characteristics of the pilot and non-pilot areas as consistent as possible, there is put-back sampling according to 1:2 caliper nearest-neighbor matching, which matches the pilot areas on a year-by-year basis. The reason for using 1:2 matching is that 1:2 matching strikes a balance between bias, efficiency, and robustness while ensuring that the sample size remains sufficient for meaningful analysis. The regression results using the PSM-DID method are presented in Table 4, Column (3), and show that the coefficients on the core explanatory variables remain significantly positive. It can be seen that institutional innovation can still significantly improve the environmental performance of the pilot districts after further mitigating potential endogenous problems.

The results of the previous study show that the innovation of trade in services system can significantly improve the environmental performance of the pilot region, while what is the acting mechanism behind it needs to be further examined. According to the research hypothesis, the mechanism of action may include both green innovation capacity enhancement and industrial structure upgrading, and the mechanism testing model is set as Equation 5: M i t = β 0 + β 1 d i d i t + γ C o n t r o l s i t + μ i + λ t + ε i t (5) M i t denotes the mechanism variable. Green innovation capacity is measured by the number of green patent applications per 10,000 people (patapp) and the number of green patents obtained per 10,000 people (pataut), and the data are obtained from the database of China Research Data Service Platform (CNRDS). Industrial structure upgrading is measured by the ratio of value added of tertiary industry to value added of secondary industry ( I n d s t r u c ) (Gan et al., 2011). The coefficient β 1 indicates the impact of institutional innovation on the improvement of regional green innovation capacity and industrial structure upgrading.