Introduction

China’s economy is transitioning from rapid growth to high-quality development, involving a reshaping of its economic model, refinement of its structure, and transformation of growth drivers. However, this shift has increased pressure on the natural environment, evident in the discharge of industrial waste that has damaged air, water, vegetation, and soil, leading to a significant decline in environmental quality. The World Health Organization (WHO)’s 2021 Global Air Quality Guideline Values (see Appendix A1 for the website) show that PM2.5 exposure levels remain high in Asia, Africa, and the Middle East, with China’s annual average concentration reaching 35 μg/m3, 3.5 times higher than the 2005 target. In 2020, China set ambitious goals to achieve a ‘carbon peak’ by 2030 and ‘carbon neutrality’ by 2060, aiming for sustainable, high-quality development1. In this context, enhancing ecological resilience (Based on Holling’s definition of resilience, scholars have interpreted ecological resilience from the perspective of adaptation or restoration. Among them, ecological resilience under the perspective of restoration refers to the ability of ecosystems to maintain themselves, regulate themselves, and resist various pressures and perturbations from the outside world after deviating from the equilibrium state under external disturbances. Ecological resilience under the perspective of adaptation refers to the degree of resistance of ecosystems to external changes. In short, ecological resilience refers to the ability of ecosystems to return to a stable state after being subjected to external shocks.) and the ability to withstand disruptions is crucial. If economic development and ecological conservation can be harmoniously integrated, following the principles of sustainable scale, efficient distribution, and equitable distribution2, we can create a more just and equitable world. This requires the efficient and equitable use of resources to ensure genuine and sustainable well-being for both humanity and the natural world3.

Technology is crucial in enhancing ER, serving as a key driver of social progress and environmental conservation. However, environmental innovations confront complex and interconnected challenges, such as climate change and resource depletion, which are highly uncertain and urgent. Effectively addressing these issues requires collaboration across multiple disciplines, sectors, and agencies4. Some scholars highlight that green technological innovation significantly aids in carbon reduction, energy conservation, and new renewable energy source development, thereby strengthening the governance of socioecological systems5,6. Moreover, environmental innovation does not completely discard traditional technologies but rather enhances and refines them. This incremental innovation approach gradually introduces new solutions while maintaining system stability7. As Pestre noted, TI and green technologies alone do not sufficiently address environmental challenges; researchers should focus on transforming production and lifestyle as well8. However, traditional productivity development approaches struggle to meet the urgent demand for ER. The emergence of the NQPF offers a new opportunity for improving ER. Compared with traditional productivity, the concept of the NQPF requires a deeper understanding of the natural world and human production activities, promoting harmonious symbiosis between humans and nature while considering production efficiency and quality9,10. The NQPF, which is characterized by high-tech, high-efficiency, and high-quality features while prioritizing innovation and deviating from conventional economic growth and productivity development paths10,11.

In 2023, General Secretary Xi Jinping first proposed the concept of the NQPF during an inspection in Heilongjiang. On January 31, 2024, at the 11th Collective Study of the Political Bureau of the Communist Party of China (CPC) Central Committee, he stressed the importance of accelerating the development of the NQPF for high-quality development10. As an important economic region in China, the Yangtze River Delta Economic Belt has experienced NQPF that drive economic restructuring and upgrading. The NQPF has accelerated the transformation of traditional industries and promoted high-tech and strategic emerging industries, forming a modern industrial system of high-tech, advanced manufacturing, and modern service industries. This process has enhanced the region’s economic competitiveness and sustainability. Moreover, the region has reformed innovation systems and mechanisms and optimized the innovation environment, providing strong support for TI. Amid growing global economic uncertainty, studying the impact of the NQPF on UER is crucial for enhancing a region’s ability to withstand external shocks. In this work, we choose 27 cities in the Yangtze River Delta Economic Belt as the research area. We explore the differential impact of the NQPF on economic resilience, analyse the factors influencing cities with different levels of economic resilience, and offer relevant policy suggestions.

We make three main contributions in this work. First, we focus on the heterogeneous relationship between NQPF and ER and explore the municipal level instead of the provincial level, as in previous studies, thereby enriching the existing academic research on ER. Second, on the basis of the theoretical framework of adaptability–resilience–resistance, we construct a new set of ER measurement indicator systems that enhances the understanding of ER, enriches the ER measurement system, and contribute to the existing knowledge of ER. Finally, from an academic perspective, we not only explore the direct effects of NQPFs on ER but also analyze the indirect effects of NQPFs on ER from the perspective of TI.

The remainder of this paper is organized as follows. In “Literature review” section, we review the relevant literature on ER; in “Methodology” section, we describe the methodology of the empirical model; in “Research data” section, we describe the data sources and model setup; in “Empirical research”, we present the empirical results; and in “Conclusions and recommendations” section, we present conclusions and recommendations.

Literature review

To explore the research in this paper, we conduct a literature review in three separate areas: the measurement of ER, the factors affecting it, and the progress of research on NQPFs.

Measurement of ecological resilience

The concept of resilience, derived from the Latin term ‘resilire’, signifies the ability of a system or individual to recover to its original state after a disturbance. While resilience was initially used in physics, it was introduced into ecology by Holling in 1973, who defined ecological resilience as the extent of disturbance that a system can absorb before changing to another state12. Alberti further described it as the capacity of an ecosystem to withstand risks, recover quickly, and ensure ecological security and sustainable development13. Folke added that ecosystems can adapt, learn, and self-organize to achieve stability14. Essentially, ecological resilience refers to the ability of ecosystems to return to a stable state after being subjected to external shocks.

Research on ecological resilience started with theoretical analysis, a single assessment indicator, and, eventually, a multidimensional comprehensive assessment system. Adger viewed resilience as a characterization of social and economic conditions and a key to sustainable development15. Wu and Wu emphasized that urban design should be based on ER to address internal and external changes16. Early ER measures used single indicators, such as those used by Alberti and Marzluff, to study the impacts of urban landscape patterns on ecological conditions through bird and aquatic macroinvertebrate diversity13. Adger examined ER in relation to mangrove forest changes in Vietnam15. Urban ecological resilience has become a key benchmark for measuring urban ecological progress and sustainability.

When a multidimensional assessment system is built, the ER and urban resilience indicators often draw on each other. Xiu et al. proposed an urban resilience research framework based on scale–density–morphology, linking urban planning and resilience research17. Wang et al. analysed the resilience of 139 cities from a socioeconomic–ecological perspective, enriching urban resilience research18. The traditional pressure–state–response (P–S–R) model has been expanded to include dimensions such as drivers and impacts for more comprehensive ER assessment19,20,21,22. Additionally, Wu and Huang used the ecological sensitivity, ecological resilience, and ecological pressure (SRP) model to identify factors affecting urban ecological vulnerability23. Li et al. constructed an ER evaluation system from three dimensions: environmental pollution generation (PPI), environment pollution management (PMI), and ecological safeguard (ESI)24. Liao and Zhang developed a more detailed ER evaluation index system from the resistance‒response‒renewal perspective25.

Factors influencing ecological resilience

The impacts of TI on ER can be divided into direct and indirect effects. Socioecological innovations are crucial for environmental governance, and they enhance the resilience of ecosystem services in urban areas26. Thus, TI and green technology innovation are often seen as key methods for increasing ER. For example, Li et al. noted that environmental improvement investment, sewage taxes, green technology innovation, and industrial upgrading are the main drivers of ER enhancement24. Elrick-Barr et al. highlighted innovation as vital for coastal communities to achieve socioecological resilience27. Fu et al. suggested that the coordinated development of green technologies and UER is important for sustainable urban development in China28. Indirectly, the TI influences the ER through other channels. Jiang et al. reported that financial reforms can significantly enhance urban resilience, with TI and industrial agglomeration as key pathways29. Zhao et al. discovered that digital finance can improve UER by promoting TI and reducing energy intensity30. Zhou et al. revealed that university scale positively affects urban and social resilience, with city size and TI effects being crucial factors31. Lee et al. concluded that government green investment positively impacts UER, whereas population density, economic development, and financial development negatively affect UER32. Although TI is recognized as a crucial factor in enhancing ER, its promotion and application may be limited by regional economic development levels and the intensity of environmental regulation. In areas with underdeveloped economies or lax environmental oversight, land resource misallocation can skew the allocation of innovative resources and reduce the efficiency of technological innovation33. This limitation shows that while TI advances sustainable development, it can result in negative impacts in the form of resource allocation problems, leading to resource consumption and environmental pressure, undermining investments in ecological protection34,35, and further intensifying the vulnerability of urban ecosystems36.

Additionally, some scholars have explored ER influence mechanisms from other perspectives. Liu et al. concluded that climate change is the main ER impact mechanism on the Tibetan Plateau37. Tao et al. reported that environmental regulation can significantly reduce ecological risk, particularly in high-risk areas38. Wang and Li examined the spatial spillover effect of ER and its heterogeneity from a sustainable development planning perspective and reported that such planning can increase adjacent cities’ ER levels via spillover effects while increasing regional UER39. Zhu et al. noted that population agglomeration does not hinder ER development; instead, reasonable agglomeration can promote ER progress40. Zhang et al. reported that green technology innovation and advanced industrial structure are important strategies for increasing UER on the basis of big data pilot zone policy studies41. Zhang et al. reported that social response mechanisms play a key role in ER studies in the Yangtze River Delta42.

Advances in new quality productive forces research

Current research on NQPF is primarily qualitative and focused on theoretical analysis. Cao and Cai developed a comprehensive evaluation system for NQPF, assessing it from three dimensions: information productivity, green productivity, and innovation productivity43. Empirical studies have emerged. Lin et al. reported that NQPF positively impacts agricultural quality development, with two identified thresholds9. Liu and He highlighted NQPF’s key mediating role in the mechanisms by which industrial agglomeration can affect the high-quality development of manufacturing44. With respect to NQPF development, Shi and Sun noted that creating national big data pilot zones aids NQPF growth45. Zhao and Chen stressed that the digital economy drives NQPF development46.

Although the impact mechanisms of ER have been well researched in academia, several shortcomings remain. In particular, the mechanisms of the emerging dynamics of NQPFs on ER have not been empirically investigated. Traditional research methods, such as the panel fixed-effects regression, mean reversion, and autoregressive distributed lag (ARDL) approaches, have provided insights into the influence mechanisms of ER. However, these methods fall short in fully capturing the heterogeneous effects of NQPFs on the ER. In this context, we innovatively combine panel quantile regression, panel moderated regression and panel threshold regression to comprehensively explore the linear and nonlinear effects of the NQPF on ER. Therefore, the empirical findings of this research can better explain the heterogeneity of the effects of NQPF on ER.

Methodology

Panel quantile regression

The quantile regression method originated from Koenker and Bassett’s research in 1978 and is an extension and improvement of traditional mean regression47. The conventional mean regression approach may yield biased parameter estimates owing to inherent limitations, and its application typically necessitates adherence to more rigorous prerequisites. These prerequisites often entail conditions such as the random error term conforming to a normal distribution, which can be challenging to satisfy in practical applications48. In contrast, quantile regression alleviates the distributional constraints imposed on the error term and transcends the limitation to a normal distribution. This process demonstrates remarkable robustness and resilience to outliers in the dependent variable49. Furthermore, quantile regression provides a comprehensive depiction of the conditional distributions of the explanatory variables, capturing the nuanced characteristics across various quantile points. This process is modelled as follows:

$$Y_{it,\tau } = X_{it} \beta_{\tau } + \varepsilon_{it}$$
(1)

where \(Y_{it,\tau }\) denotes different quantile levels of the explained variables, \(\tau (0 < \tau < 1)\) denotes different quantile levels, \(X_{it}\) denotes explanatory variables, \(X_{it} \beta_{\tau }\) denotes different quantile levels of the explanatory variables, and \(\varepsilon_{it}\) denotes the random error term.

Panel threshold regression

Tong first introduced threshold regression modelling in 197850, which was later extensively explored and developed by Hansen in 200051 (see Appendix A2 for the process). The core idea of the model is to identify the critical points or critical regions where jumps in threshold variables, that is, specific points or regions where the model coefficients will change significantly, may occur to more accurately identify and explain the complex relationships in real economic phenomena. The threshold regression model is further subdivided into two types: single threshold regression and multiple threshold regression. The specific expression for single threshold regression can be expressed as follows:

$$Y_{it} = \mu_{i} + X_{i} {{\varvec{\upbeta}}}_{1} I\left( {Q_{it} \le \gamma } \right) + X_{i} {{\varvec{\upbeta}}}_{2} I\left( {Q_{it} > \gamma } \right) + \varepsilon_{it}$$
(2)

where \(Q_{it}\) denotes the threshold variable, \(\gamma\) denotes the threshold parameter, and \(I( \cdot )\) denotes the indicator function.

Research design

Variable measurement

Core explanatory variable: new quality productive forces

To date, a unified evaluation index system for assessing NQPFs does not exist. Among existing studies, Yu52 extracted the connotative characteristic keywords of new quality productivity by calculating the ratio of the total text word frequency to the exact keyword frequency in works related to new quality productive forces. Guo53 and Bueno54 counted the frequency of keywords in the literature to identify the key themes and areas in urban resilience research. Therefore, we measure new quality productive forces on the basis of the total text word frequency and the frequency of keywords related to new quality productive forces in government work reports. In this work, we first crawl raw documents of work reports from local government websites and process the text into panel data. Then, we remove stop words and use the Jieba library in Python to perform exact and full-mode lexical segmentation and word frequency computation. In this section, we retain Chinese, English, numbers, and punctuation and calculate word frequencies for 46 words using regular expressions. For full-text word frequency, we remove stop words, calculate exact and extended words, keep Chinese, English, and numerical text, and calculated length of the full text. Finally, we save detailed information about the total word frequency of the text, the frequencies of 46 words, and the length of the text. The NQPF calculation formula is as follows:

$$NQPF_{it} = \frac{{PK_{it} }}{{TWF_{it} }}$$
(3)

where \(NQPF_{it}\) denotes the level of NQPF in city \(i\) in period \(t\), \(PK_{it}\) denotes the frequency of keyword terms related to NQPF in the government work report of city \(i\) in year \(t\), and \(TWF_{it}\) denotes the total word frequency (TWF) of the text in the government work report of city \(i\) in year \(t\). In the calculation of NQPF, the exact keyword frequency is obtained by using the keyword search related to NQPF in the government work report. In this study, the main keywords chosen are NQPF, artificial intelligence, scientific and technology innovation, green economy, scientific development, innovative momentum, disruptive technologies, circular economy, and new economy, among others.

Figure 1 shows the trends in the average NQPFs across the cities. Notably, Changzhou has the highest average NQPF level among the 27 cities in the Yangtze River Delta Economic Belt, whereas Chuzhou has the lowest level. The top five cities, namely, Changzhou, Hefei, Nanjing, Jinhua, and Wuhu, owe their high NQPF levels to the emphasis placed on them by their local governments, who have implemented corresponding policies to foster their development. Furthermore, the significant disparities observed in average NQPF levels among certain cities highlight the urgent need for a balanced approach to promoting NQPFs across the region.

Fig. 1
Fig. 1
Full size image

Average new quality productive forces by municipality (2006–2022).

Explained variable: ecological resilience

Adaptability, resilience and resistance are the three core dimensions recognized in ER research, which together constitute a complete framework of ER and can comprehensively reflect the ecosystem’s ability to manage disturbances. Therefore, we build upon the work of Lee et al.32 and Wang et al.39, embracing the adaptability–resistance–resilience framework to develop a comprehensive evaluation index system tailored for the Yangtze River Delta Economic Belt (see Table 1 for details). To maintain the objectivity of our empirical research and circumvent subjective biases in weight assignment, we have adopted the entropy weight method, as outlined by Yin et al.55, for the precise measurement of relevant indicators (see Appendix A3 for the process).

Table 1 Explained variable definitions.
Full size table

Figure 2 shows how ER changed in the Yangtze River Delta Economic Zone from 2006 to 2022. In 2006, most areas had low ER, with low-resilience cities clustered in groups: Anqing, Chizhou, and Tongling in the west; Yangzhou, Zhenjiang, Nanjing, Changzhou, and Maanshan in the north; and Shanghai, Suzhou, and Nantong in the east. In the central region, Hangzhou, Shaoxing, Ningbo, and Zhoushan were closely clustered, whereas Wenzhou in the southern region was the only city with low ER.

Fig. 2
Fig. 2
Full size image

Source GS(2023)1822).

Spatial and temporal distribution of ecological resilience (This map was created by authors using ArcGIS 10.8.

By 2011, the ER of the Yangtze River Delta Economic Zone had significantly improved. Taizhou in Jiangsu Province remained the only city with a low ER of 0.1858, indicating a slight decrease from 0.2108 in 2006. Moreover, Taizhou, Chizhou, and Xuancheng in Zhejiang became leaders, with high ER levels of 0.6898, 0.6161, and 0.6030, respectively. The previously low-resilience patchwork clusters improved significantly, moving into the low- to medium-resilience category.

By 2016, the region ERs improved significantly and are now mostly in the medium–high range. Notably, the cities of Hangzhou, Huzhou, Wuxi, Changzhou, Zhenjiang, and Yangzhou form a banded cluster of high resilience. Other high-resilience cities are more dispersed and include Chizhou, Maanshan, Zhoushan, Nantong, Wenzhou and Taizhou (Zhejiang).

By 2022, most cities in the region had high levels of emergency response. Four cities had higher resilience: Hefei (0.6749), Tongling (0.7565), Wuxi (0.7895), and Shanghai (0.7658).

Threshold variable: technology innovation

Drawing upon the studies by Yang et al.56 and Hu et al.57, we utilized the number of patents granted in three categories, namely, invention patents, utility model patents, and design patents, as indicators to assess the level of technological innovation.

To explore the changes in TI levels in the Yangtze River Delta Economic Zone over a 17-year period, the mean TI values for 27 cities were plotted. Figure 3 shows that Suzhou had the highest TI level, whereas Chizhou had the lowest level. The top five cities, in descending order, were Suzhou, Shanghai, Hangzhou, Ningbo, and Wuxi. These cities were likely to have high TI levels because of their large economies, advanced infrastructure and technology, strong education and research capabilities, and local government policies that attract high-tech talent. Additionally, significant differences in mean TI values between cities could be attributed to disparities in infrastructure and technology and to the siphoning effect of large cities attracting high-tech talent, thereby increasing the TI gap between them.

Fig. 3
Fig. 3
Full size image

Average level of TI by municipality (2006–2022).

Control variables

Since the relationship between the NQPF and ER may be affected by other factors, we refer to Zhang et al.42, Zhao et al.30, and Hu et al.57. The following control variables were used: population density (PD), level of urban openness to the outside world (OTW), urban industrial structure (UIS), level of urban financial development (LUFD), urban income level (UIL), and percentage of fiscal expenditure on education (PFEE). PD is measured as the ratio of the city’s year-end population to the built-up area of the administrative district. OTW is measured by the logarithm of the actual use of foreign capital. UIS is measured by the ratio of tertiary value added to secondary value added. LUFD is measured by the ratio of year-end financial institution loan balances to GDP. UIL is measured using the average wage of local workers. PFEE is measured by the ratio of local education expenditures to GDP. Table 2 shows detailed variable information.

Table 2 Variable definitions.
Full size table

Research data

In this work, panel data for 27 cities in the Yangtze River Delta economic belt from 2006 to 2022 are selected as a sample. The data come from the China Urban Statistical Yearbook, China Science and Technology Statistical Yearbook, local urban statistical yearbooks, national economic statistical bulletins and annual government work reports. For the missing values and outliers, we use the mean method with exponential prediction. Table 3 below shows the descriptive statistics between the variables.

Table 3 Descriptive statistics.
Full size table

Table 3 shows that the mean value of the ER is 0.5250, the standard deviation is 0.2150, and the minimum and maximum values are 0.0679 and 0.9700, respectively. This finding suggests that the ER in the Yangtze River Delta region has considerable dispersion and significant imbalance. The mean value of the NQPF is 0.0749, the standard deviation is 0.1130, and the minimum and maximum values are 0.0100 and 0.7670, respectively, indicating that the NQPF level in the Yangtze River Delta Economic Belt region shows significant variability.

Model construction

To explore the heterogeneity effect of NQPF on ER, we construct the following conditional quantile function:

$$ER_{it,\tau } = \alpha_{0} + \alpha_{1} NQPC_{it,\tau } + \alpha_{2} Col_{it,\tau } + \varepsilon_{it}$$
(4)

where \(ER_{it,\tau }\) denotes the level of ER for city \(i\) in period \(t\), \(\tau\) denotes the quartile level, \(NQPF_{it,\tau }\) denotes the level of NQPF for city \(i\) in period \(t\), \(Col_{it,\tau }\) denotes a set of control variables for city \(i\) in period \(t\), and \(\varepsilon it\) denotes the random error term. The setting is \(0 < \tau < 1\) in the model.

Since NQPFs are characterized by TI, technological advances accelerate the generation of NQPFs, which in turn contribute to higher ER levels43. The following model is constructed:

$$ER_{it} = \gamma_{0} + \gamma_{1} NQPC_{it} + \gamma_{2} Col_{it} + \gamma_{3} TI * NQPC_{it} + \varepsilon_{it} \cdot ER_{it}$$
(5)

where \(TI*NQPF_{it}\) denotes the cross-multiplier term between the TI level of city \(i\) in period \(t\) and the NQPF.

We construct a single-threshold regression model with TI as the threshold variable:

$$ER_{it} = \psi_{0} + \psi_{1} NQPC_{it} I\left( {TI_{it} \le \theta_{1} } \right) + \psi_{2} NQPC_{it} I\left( {TI_{it} > \theta_{1} } \right) + \psi_{3} Col_{it} + \varepsilon_{it}$$
(6)

where \(I( \cdot )\) denotes the indicator function, \(\theta_{1}\) denotes thresholds, and \(TI_{it}\) denotes the threshold variable and the level of TI in city \(i\) in period \(t\).

Empirical research

Correlation and multicollinearity tests

According to Table 4, NQPF and PD show a significantly negative correlation at the 1% significance level, whereas the other control variables exhibit a notably positive correlation at the same level. Specifically, UIL and LUFD have a highly significant positive correlation, with a coefficient of 0.853 at the 1% level. To improve the accuracy of the empirical analysis, UIL is excluded as a control variable. Table 5 shows the multicollinearity test results, with each variable’s variance inflation factor (VIF) being less than 10, indicating that there are no multicollinearity issues.

Table 4 Correlation test.
Full size table
Table 5 Multicollinearity test.
Full size table

Quantile regression

To explore the heterogeneity of the NQPF on the ERs in the 27 Yangtze River Delta Economic Zone, we set \(\tau \in \left\{ {0.05,0.25,0.5,0.75,0.95} \right\}\). Table 6 presents the results of the panel quantile regression.

Table 6 The quantiles regression results in the Yangtze River Delta Economic Zone.
Full size table

Table 6 shows that the estimated coefficient values of NQPF are positive at different quantiles of ER, which indicates that NQPF significantly contributes to ER. Moreover, it is not difficult to find significant heterogeneity in the coefficient estimates of the NQPFs. Specifically, the coefficient estimate for the NQPF is the largest at the 0.05 quantile at Q(0.05) = 0.0600. This finding highlights that the advancement of the NQPF has the most pronounced impact on ER in regions with exceptionally low resilience levels. This significant influence can be attributed to the vulnerability of the ecosystem and the severe environmental pollution prevalent in these areas, where the introduction of green technology can swiftly enhance the ecological environment. The coefficient estimates for NQPFs are significant at the 10% and 5% levels at the 0.25 and 0.5 quantiles, respectively. The results reveal that, in situations where the ER is low, an increase in NQPF positively influences the enhancement of the ER. Specifically, for every unit increase in NQPF, the ER experiences corresponding increases of 4.9% and 4.05%, respectively. The coefficient estimates for NQPFs at the 0.75 and 0.95 quantiles, although remaining positive, exhibit decreased significance. This result suggests that as the ER increases, the enhancement provided by the NQPF gradually diminishes, exhibiting the phenomenon of diminishing the marginal effect. Consequently, at higher ER levels, every incremental unit of the NQPF results in mere 3.31% and 2.42% gains in the ER.

The control variables in Table 6 reveal that the impact of each variable on ER varies across different quantiles. Population density (PD) has a consistently negative effect on ER, with the most significant impact at the 5% quantile (Q(0.05) = 0.0914). This finding indicates that a single unit increase in PD leads to a 9.14% decrease in ER at this quantile. At lower levels of ER, an increase in PD can severely limit ecological space and degrade the environment. However, as ER increases, the negative impact of PD diminishes due to the strengthened resilience of the ecosystem. In contrast, the coefficients for openness to the outside world (OTW), urban industrial structure (UIS), and urban financial development (LUFD) are significantly positive at the 1% level across all quantiles. The maximum coefficients are Q(0.05) = 0.1970 for OTW, Q(0.95) = 0.2130 for UIS, and Q(0.05) = 0.3710 for LUFD. These results highlight the positive role of economic development in enhancing ER. As economic prosperity increases, the relationship between economic and ecological systems tends to become more harmonious, leading to a positive impact on the ecosystem. Additionally, the coefficient for the percentage of fiscal expenditure on education (PFEE) is negative at the 5% quantile (Q(0.05) =  − 0.0288) but positive at other quantiles, with a maximum value of Q(0.25) = 0.0771. This finding suggests that while the effect of PFEE on the ecological environment has lagged and is not immediately apparent, increased investment in education can enhance public awareness of environmental protection in the long term. This phenomenon boosts the level of TI and ultimately improves the ecological environment.

Moderating effect

To further explore the moderating influence of the NQPF on the ER of the Yangtze River Delta Economic Belt, we utilize Jenks’ natural discontinuity point method to categorize the 27 cities within this region on the basis of their mean ER values55. Table 7 presents the specific city classifications, serving as the foundation for the subsequent panel moderation effects test and panel threshold effects test.

Table 7 Classification of ecological resilience.
Full size table

Table 8 presents the results of the moderating effects analysis, with Columns (1), (2), (3), and (4) displaying the outcomes for the overall, low-, medium-, and high-resilience cities, respectively, among the 27 cities in the Yangtze River Delta Economic Zone.

Table 8 Moderating effect results.
Full size table

The results in Column (1) indicate that the coefficient of TI*NQPF is 0.273 and is significant at the 1% level. This finding indicates that within the overall scope of the Yangtze River Delta Economic Belt, TI significantly enhances the promoting effect of NQPF on ER. Specifically, as the level of TI increases, the positive impact of NQPF on ER becomes increasingly pronounced. These findings confirm the pivotal role of TI in catalysing the conversion of NQPF into the increased ER. The findings in Column (2) indicate that the coefficient for the interaction term TI*NQPF is 0.197, and it is statistically significant at the 5% level, suggesting a positive but relatively limited moderating effect due to their low initial ER levels. The findings from Column (3) indicate that the coefficient of TI*NQPF is 0.223, achieving significance at the 1% level, reflecting the more pronounced role of TI in boosting ER. This finding is attributed to these cities’ solid foundations in terms of economic development, industrial structure, and technological innovation, enabling them to leverage TI more effectively. The results from Column (4) reveal that the coefficient of TI*NQPF is 0.317, achieving significance at the 1% level, which is the highest among the subgroups. This result underscores TI’s significant role in enhancing the impact of NQPF on ER in these cities, which possess advanced TI systems and sophisticated industrial bases, allowing them to swiftly adapt to technological advancements and implement green technologies, thereby significantly boosting ER.

In summary, TI notably amplifies the positive impact of NQPF on ER, not only across the Yangtze River Delta Economic Zone overall but also within cities characterized by varying degrees of resilience. However, this facilitation shows some variability across cities with different levels of resilience, with the most significant facilitation in high-resilience cities and relatively weaker facilitation in low-resilience cities. This result suggests that policy-makers should fully consider the actual situation and differentiated needs of each region when promoting regional ER enhancement and formulate targeted TI policies to maximize the contribution of the NQPF to ER.

Threshold effect

Threshold effect test

Table 9 shows that the F statistic for the single threshold regression is significant across regions and that the F statistic for the double threshold regression is not significant. Therefore, we choose a single-threshold regression for further study.

Table 9 Threshold effect tests.
Full size table

Threshold regression results

Table 10 presents the findings of the single-threshold regression analysis with fixed effects. The table consists of four columns, presenting the regression outcomes for the entire Yangtze River Delta Economic Zone, along with the results for each of the three resilience categories. Overall, there is a significant TI threshold (0.3054) in the Yangtze River Delta Economic Zone, and the impact of TI on the relationship between NQPF and ER shows a ‘U’ shape with an initial decreasing and subsequent increasing effect. When the TI levels fall below this threshold, the NQPF has a constrained contribution to the ER, potentially even displaying a detrimental effect. This outcome arises because insufficient TI levels are ineffective at mitigating pollutant emissions during the production process and may inadvertently trigger fresh environmental challenges, such as the generation of e-waste. However, once the TI level surpasses this threshold, the contribution of NQPF to ER becomes markedly more pronounced, indicating that elevated TI levels can foster the adoption of green technologies and cleaner production methods, thereby bolstering ER.

Table 10 Threshold regression.
Full size table

The threshold effect test reveals that the TI threshold for low-resilience cities is 0.3214, which is slightly greater than the overall Yangtze River Delta Economic Zone threshold. This finding indicates that low-resilience cities require higher TI levels to overcome existing barriers to improving ER. Once the TI level surpasses this threshold, the contribution of NQPF to ER becomes significant (at the 5% level), indicating that these cities can quickly increase their ER through advanced technologies and increased TI. For medium-resilience cities, the TI threshold is 0.2303. After exceeding this threshold, NQPF significantly contributes to ER at the 10% level, but the effect size is small (coefficient = 0.0870). This finding suggests that medium-resilience cities face challenges in ER improvement and need more policy support, financial investment, and advanced technologies to overcome these hurdles. High-resilience cities have the highest TI threshold at 0.4353. Once this threshold is crossed, NQPF significantly increases ER at the 5% level, with a substantial effect size (coefficient = 0.140). This result highlights the potential of high-resilience cities to further enhance ER through consistent high-level TI inputs. They should continue to strengthen their TI systems and focus on developing green technologies and cleaner production methods for deeper ER improvements. Other factors, such as industrial structure and population density, should be considered to ensure comprehensive and sustainable development.

Robustness test

To ensure the robustness of the empirical results, tests are conducted by adjusting the sample period, adding control variables, and lagging the core explanatory variables. Specifically, the sample period was shortened from 2006–2022 to 2012–2022. The urban traffic level (UTL), measured as the logarithm of the volume of goods transported by road, was added as a control variable. The core explanatory variable NQPF was lagged by 5 periods. The 5-year lag serves two main purposes: (1) to address endogeneity issues, making the estimates more robust, and (2) to reflect the time lag effect between policy implementation and market response, as the Chinese government typically follows a five-year cycle for economic development planning. This approach better captures the actual impact of the NQPF on the ER.

Table 11 presents the robustness test results, with Column 1 showing the regression after reducing the sample period, Column 2 showing the results after adding the urban traffic level (UTL) as a control variable, and Column 3 showing the results after lagging the NQPF variable by five periods. The NQPF variable is significantly positive at the 5% level in Column 1, the 10% level in Column 2, and the 5% level in Column 3, indicating that the positive impact of NQPF on ER is robust.

Table 11 Robustness test.
Full size table

Conclusions and recommendations

Conclusions

We assessed the ER levels of 27 cities in the Yangtze River Delta Economic Zone using the entropy weight method. We further analyse the impacts of NQPF on ER and its mechanisms through panel quantile regression, panel moderating effect tests, and panel threshold effect tests. The main findings are as follows:

  1. (1)

    The study reveals that the ER in the Yangtze River Delta Economic Zone consistently improves over the study period. From the initial low-resilience regions in 2006, a significant transformation occurred by 2022, with most cities now being classified as high-resilience regions, indicating an overall enhancement of the regional ecological environment.

  2. (2)

    The panel quantile regression results reveal significant heterogeneity in the contribution of NQPF to ER. Notably, at the lower quantiles of ER, the promotional effect of NQPF is particularly pronounced. However, as the ER level continues to increase, this promotional effect gradually diminishes, showing a diminishing marginal effect.

  3. (3)

    The findings from the panel moderating effect analysis reveal that TI notably magnifies the positive influence of NQPF on ER within the Yangtze River Delta Economic Zone. Nonetheless, this reinforcement effect is not uniformly distributed; it results in considerable variations in the resilience levels. Specifically, the catalytic role of TI is particularly potent in high-resilience cities, whereas low-resilience cities appear less adept at leveraging TI to augment the ER-promoting effects of the NQPFs.

  4. (4)

    According to the results of the panel threshold effect test, the impact of TI on the relationship between NQPF and ER clearly exhibits a “U”-shaped pattern, characterized by an initial decrease followed by a subsequent increase. When the level of TI surpasses a certain threshold, the impact of NQPFs on ER soars dramatically. At lower levels of TI, while it can enhance resource utilization efficiency and mitigate pollution emissions to some extent, this effect often fails to reach statistical significance due to technological limitations. However, as the TI level exceeds the threshold, its contribution to the ER increases exponentially.

Recommendations

  1. (1)

    Enhancing the technology transformation ecosystem by accelerating the development of a modern industrial system by optimizing the service system for technology transformation to efficiently convert technological advancements into productivity. Specific measures include enhancing the industrialization of scientific achievements, adjusting industrial policies, and promoting future industries such as 5G technology, new infrastructure, artificial intelligence, and advanced materials. Policies should be tailored to empower cities on the basis of their resilience levels. Low-resilience cities should prioritize technological innovation, increase investment, strengthen local research and development (R&D) capabilities, and introduce eco-friendly production methods and innovative governance models. Medium-resilience cities should optimize the innovation ecosystem, foster an inclusive innovation environment, attract high-tech enterprises and talent, and facilitate the conversion of scientific achievements. High-resilience cities should leverage existing technological advantages, increase scientific research investment, promote the integration of technology with the economy and society, and lead industrial upgrading.

  2. (2)

    Accelerating the transformation and construction of new quality productive forces. Cities of different resilience levels within the Yangtze River Delta Economic Belt should adopt differentiated strategies and measures according to their own actual situation to expedite the cultivation of green new quality productive forces, thereby fostering harmonious development between the regional economy and the ecological environment. Low-resilience cities should initiate the journey towards green transformation by embracing proven green technologies and clean energy solutions, such as energy-efficient appliances and rainwater harvesting systems. Medium-resilience cities should merge practicality with innovation to advance renewable energy adoption, execute clean energy replacement initiatives, facilitate the green transition of traditional industries, nurture burgeoning green sectors such as energy conservation, environmental protection, and renewable energy, and propel economic structure optimization and upgrading. High-resilience cities should concentrate on avant-garde green technologies, such as carbon capture and storage and hydrogen energy technology, by fostering research, development, and pilot applications. The government should establish green industrial ecosystems, encourage synergistic green supply chain development, and cultivate internationally competitive green industry clusters.

  3. (3)

    Promoting high-quality economic development and achieving a win‒win scenario between new quality productive forces and environmental protection are key to ensuring sustainable development. For new quality productive forces to continue to be generated and for environmental protection to be sustained, the government needs to adopt a more comprehensive and long-term strategy in formulating economic development policies. Clear objectives should be defined. Policy-makers should establish precise, quantifiable short-, medium-, and long-term goals for both economic development and environmental protection to ensure policy continuity, stability, and a balanced coexistence between economic expansion and environmental sustainability. Comprehensive policies should be implemented. Policy-makers should integrate green metrics, such as green GDP and resource utilization efficiency, into evaluations. Measures such as green finance and environmental taxes should be adopted to incentivize business transformation and foster the growth of green industries.