The interactive relationship between ecological well-being performance and tourism economic development in major tourism cities in China

With the advent of mass tourism, the tourism industry has experienced unprecedented expansion in scale. The relationship between the tourism industry and ecology and society is a key issue in promoting sustainable development of tourist destinations. However, existing research has focused more on the relationship between ecological environment protection and tourism economic development, while neglecting the impact on human well-being in this process. Therefore, based on the concept of ecological welfare, this study explored the coupling coordination degree (CCD) and interaction relationship between the tourism economy development (TED) and ecological wel-being performance (EWP) of 58 major tourism cities in China, providing a more inclusive theoretical perspective and enriching the theory of sustainable tourism and ecological tourism. The results show that: (1) from 2004 to 2019, the EWP and TED of major tourism cities in China showed a steady upward trend. The improvement level of TED was more significant than that of EWP. (2) The CCD between EWP and TED of most tourism cities in China has been improved to varying degrees during the period from 2004 to 2019, especially in many inland tourism cities. (3) There is a dynamic interaction relationship between EWP and TED in major tourism cities in China during the research period, but a virtuous interaction has not yet been formed between the two. The results of this study can also provide practical insights for the sustainable development of urban tourism industry.


Introduction
Human society has transitioned from a relatively abundant 'empty world' of natural capital to a 'full world' constrained by ecological environments (Zhu and Zhang 2014). In a 'full world,' artificial capital has become relatively abundant, and the factors constraining human development have shifted to absolute scarce natural capital (Daly 2005). With the rapid development of global industrialization and urbanization, the economic system has continuously expanded relative to the ecological system, and the shortage of natural capital and its services has become increasingly severe on a global scale. The constraints of natural capital on economic and social development have become increasingly apparent, while economic growth is an important means of improving human welfare (Hu et al 2021). Therefore, how to balance the relationship between ecological environment, human welfare, and economic development has become the fundamental challenge in achieving sustainable development. The Millennium Ecosystem Assessment Synthesis Report (Millenium Ecosystem Assessment 2005), jointly published by the United Nations Environment Programme and other international organizations, marked the beginning of global attention to the relationship between ecosystems and human welfare. Currently, EWP serves as an important measurement tool for evaluating how a country or region converts natural consumption into welfare levels. It effectively integrates the relationship between sustainable development capacity, ecosystem services, and human welfare, providing a new research perspective for reflecting the level of sustainable development of a country or region (Zhu and Zhang 2014;Hu et al 2021).
In fact, reducing regional resource consumption, alleviating ecological environmental damage, and transforming economic development are usually regarded as key points for improving regional EWP. The tourism industry is often considered a 'smokeless industry' and 'green industry' with prominent characteristics such as environmental friendliness, low resource consumption, and ecological sharing (Liu and Yin 2022). Developing the tourism industry helps to reconcile the contradiction between regional social and economic development and ecological protection, and promotes the coordinated and symbiotic development of regional population, economy, resources, and environment from a low level to a high level (Tang et al 2018, Cong et al 2019. This development path effectively matches the requirements for improving EWP. However, with the rapid development of the tourism economy, the negative externalities of resource disorderly development in the tourism industry can easily cause ecological environmental problems such as excessive carbon dioxide emissions, air quality deterioration, and increased land carrying capacity pressure, leading to an imbalance in the human-land system and exacerbating the contradiction between humans and the environment (Boavida-Portugal et al 2016, Meng et al 2016). It can be said that the tourism industry is a 'double-edged sword' for regional social and economic development and ecological environment. While improving the happiness and quality of life of all residents, it also poses hidden dangers of increasing resource consumption and environmental pressure such as water, energy, and land in the regional ecosystem (Hu et al 2020). Therefore, how to scientifically measure the dynamic relationship between ecological protection, resident happiness, and tourism economic development has become an important issue in exploring the sustainable development of tourist destinations.
Unfortunately, current scholars have focused more on the relationship between ecological environment protection and tourism economic development in the limited existing research, while neglecting the issue of human development in this process. However, EWP effectively incorporates human happiness and quality of life into economic development and ecological protection, providing a good perspective for us to explore the relationship between ecological environment, resident welfare, and tourism economic development in tourist destinations. Based on the above research, this study established the EWP and TED evaluation index systems respectively to measure the ecological welfare performance and tourism economic development level of tourism cities, and analyzed the dynamic relationship between the two. The research results can provide practical insights for effectively solving the current population, economic, resource, and environmental problems in tourism cities and promoting the sustainable development of local tourism industry.
Tourism cities are urban forms that have been formed and accumulated over a certain period of time with tourism development as an important goal and with prominent tourism functions. They have stronger typicality and representativeness for exploring economic, ecological, and population issues in the process of tourism development. Therefore, we selected 58 major tourism cities in China as the research area. This paper contributes to the relevant literature on this important topic in three aspects. (1) Based on the concept of ecological welfare, this study explores the relationship between tourism, ecology, and welfare in major tourism cities from a more inclusive perspective, which is different from previous studies that only focus on a single ecological perspective and ignore the role of 'human development' in this relationship.
(2) This study analyzes the relationship between TED and EWP from both a static and dynamic perspective, which is different from previous research that mainly analyzes the relationship between tourism and ecology from a static perspective of coupling and coordination. (3) The research results can enrich the theory of sustainable tourism and ecological tourism from the perspective of ecological welfare, and provide theoretical guidance and policy inspiration for the sustainable development of tourism economy in tourist destinations.
2. Literature review 2.1. Ecological well-being performance EWP refers to the efficiency of converting natural resources and ecological inputs into human welfare levels (Zhu and Zhang 2014). At a certain level of ecological input or welfare level, EWP reflects the degree of sustainable development of a country, region, or city. The research on EWP can be traced back to 1974 when Daly proposed that the sustainable development level of each country could be evaluated by calculating the social welfare level generated by unit natural resource consumption (Daly 1974). However, Daly did not propose specific quantifiable indicators and calculation methods for natural resource consumption, which has hindered the widespread application of this theory (Zhu and Zhang 2014). It was not until the Human Development Index (HDI) released by the United Nations Development Programme in 1990(Programme U, UNDP 1990) and the concept of ecological footprint proposed by Rees in 1992(Rees 1992) that the concepts of natural resource consumption, ecological environmental impact, and human welfare involved in EWP began to be reliably measured. In the 21st century, Chinese scholar Zhu drew on Daly's theoretical ideas and first proposed the concept of 'EWP', defining it as the efficiency of converting ecological resource consumption into social welfare levels, and quantifying it by the ratio of the Human Development Index to the ecological footprint, which opened the prelude to EWP research (Zhu 2008). Currently, research on EWP mainly focuses on three aspects: Firstly, regarding the construction and measurement of the EWP index system. Zhang et al defined and constructed the EWP index from two aspects: the Human Development Index and the ecological footprint. They measured the efficiency of ecological consumption in 82 countries converted into human welfare and proposed relevant development suggestions (Zhang et al 2018). Bian et al established an indicator system from the input-output perspective to evaluate the EWP of cities and evaluated the EWP of 30 provincial capital cities in China. Subsequently, Bian et al further constructed the EWP indicator system from the perspectives of inputs (such as labor force and fixed asset investment) and outputs (such as per capita GDP and dust), and conducted empirical research on 278 prefecture-level cities in China . Li et al introduced methods such as the vertical and horizontal opening grade method, coupled coordination model, and spatial exploratory data analysis (ESDA) to measure and analyze EWP, providing new ideas for EWP evaluation research (Li 2022). Xia et al improved the Human Development Index and ecological consumption, constructed an EWP evaluation indicator system, and decomposed EWP into two stages: ecological-economic transformation and economic welfare transformation. They used a two-stage DEA model and Malmquist index for dynamic measurement (Xia and Li 2022). These studies provide a comprehensive evaluation of EWP from different perspectives and shed light on the development of sustainable cities and regions.
Secondly, regarding the relationship between EWP and economic factors. Dietz et al defined the environmental intensity of human well-being as the ratio of a country's per capita ecological footprint to its average life expectancy at birth, and used panel data from 58 countries to find that the relationship between per capita GDP and the environmental intensity of well-being is U-shaped, contrary to the Kuznets curve (Dietz et al 2021). Feng et al explored the impact path of industrial structure green adjustment and green total factor productivity on EWP growth from the perspective of green goals (Feng et al 2019). Jorgenson et al evaluated the time dynamic relationship between the energy intensity of human well-being and economic development in a sample of 12 countries and found the possibility of a relatively harmonious relationship between future development, human well-being, and the natural environment (Jorgenson et al 2014). Wang et al used the logarithmic mean Divisia index method (LMDI) and stochastic frontier analysis (SFA) to reveal the driving effects of EWP changes and the determinants of economic growth effects. They found that the overall change in China's EWP shows an economic growth-driven pattern, but the determinants between regions show significant heterogeneity (Wang and Duan 2023). These investigations provide valuable perspectives on the intricate interplay between economic factors and EWP, and furnish direction for policymakers aiming to foster sustainable economic advancement.
Thirdly, regarding the exploration of factors affecting EWP, Li et al found through research that the main influencing factors of inter-provincial EWP in China are technological progress, greening level, social expenditure, medical level, urbanization, industrial structure, and environmental regulation (Li et al 2019). Fang et al explored the impact of factors such as energy structure, urbanization level, industrial structure, technological progress level, environmental regulation, foreign direct investment, economic development level, and the square of per capita GDP on regional EWP (Fang and Xiao 2019). Wang et al attributed the influencing factors of EWP to scale effect, structural effect, and technological effect, and used a vector autoregressive model to analyze the dynamic response relationship between EWP and influencing factors . In conclusion, scholars have examined various factors that influence EWP using different approaches, including climate, political, economic, social, technological progress, greening degree, environmental regulation, medical level, foreign direct investment, and GDP per capita square. These studies provide valuable insights into the development of sustainable economic policies and promote the achievement of sustainable development goals.
Overall, EWP provides a unique analytical perspective and tool for sustainable development, and existing research continues to enrich the theoretical connotation and empirical application fields of EWP. However, existing research has mostly explored the development relationship between EWP and traditional, macroeconomic industries, and there is still a lot of research space for analyzing the relationship between emerging and segmented economic sectors.

Relationship between ecological environment and tourism economic development
Tourism economic growth and ecological protection are a contradictory unity in the process of tourism destination economic development. How to coordinate the relationship between the two and achieve sustainable development of tourism destinations has always been a hot topic of concern among scholars. Currently, there is limited direct research on the relationship between EWP and TED, but there has been extensive research on the interactive relationship between ecological environment and tourism economy, which has mainly focused on the following three aspects in the past: Firstly, tourism development brings pressure to the ecological environment. With the rapid development of the global tourism industry, tourism activities based on ecological environment and natural resources inevitably generate negative impacts during the development process, causing huge environmental pressure on tourism destinations (Lin et al 2018, Ma et al 2021. The ecological risk coefficient of tourism destination ecosystems is beginning to rise, and the sustainable development of the tourism industry is being threatened. Therefore, a large amount of research is currently focused on exploring the adverse effects of tourism development on the ecological environment. Among them, ecological footprint analysis, as an emerging method of ecological economics, has been introduced by many scholars to explore tourism development and ecological construction evaluation (Ali et al 2021, Lee and. For example, Liu et al revealed the growth assumption dominated by the tourism industry and the inverted U-shaped relationship between the Pakistani tourism industry and ecological footprint by exploring the impact of tourism on ecological footprint . In addition, the ecological carrying capacity of tourism destinations is also an important indicator for measuring the impact of tourism on the ecological environment. For example, Adrianto et al evaluated the carrying capacity of tourism on small islands in Indonesia through a social-ecological system model, and the research results provided important references for improving sustainable tourism management on islands (Adrianto et al 2021). Ali et al used regression analysis to prove that the reduction in ecological footprints was caused by an increase in tourism, renewable energy, urbanization, and cultural globalization in high, upper-middle, and lower-middle-income countries (Ali et al 2021).
Secondly, the ecological environment on the tourism development of the negative effect. As tourism relies on natural resources, any changes in the ecological environment can significantly affect tourism activities, tourism attraction, and tourism growth. One of the main research areas in this regard is the influence of natural disasters on the tourism industry. Tourism is naturally fragile to natural disasters including tSunamis, wildfires, Earthquakes, floods, droughts, heat waves, avalanches, hurricanes, volcanic eruptions, and landslides. These disasters can cause significant damage to regional tourism and economic activities, leading to a sharp decrease in the number of tourists in affected areas (Ghaderi andHenderson 2013, Rosselló et al 2020). Scholars have undertaken various studies to gain a better understanding of how natural disasters impact tourism. For instance, using the tourism background trend line, Zhang et al examined the effect of the 2017 Jiuzhai valley Earthquake in China on tourism (Zhang et al 2021). The study found that the Earthquake disaster destroyed the tourism industry in Jiuzhai valley and altered the relative importance of Jiuzhai valley in the broader tourism industry in Sichuan Province. Chen et al calculated the economic influence of the 2017 hurricane in Florida on shared accommodations in coastal tourist destinations. The losses, both direct and indirect, caused by the disaster were significant, highlighting the severe effect of natural disasters on the tourism industry. Moreover, weather conditions also have a significant effect on daily sales at outdoor tourism locations (Chen et al 2021). Craig et al discussed the impact of weather conditions on daily sales at outdoor tourism locations, highlighting the need for tourism businesses to be flexible and adaptable to changing weather conditions (Craig and Feng 2018). The ecological environment's impact on tourism development is a multifaceted and complex phenomenon that requires further research to mitigate the negative effects of natural disasters on the tourism industry.
Thirdly, the interactive relationship between the ecological environment and tourism development. In order to more effectively explore the mutual suitability of ecology and tourism within a region, scholars have constructed different econometric models or theoretical analysis frameworks to comprehensively explore the dynamic relationship between ecological environment and tourism development from a comprehensive perspective. For example, Zeng et al combined landscape attractiveness and ecological sensitivity, and comprehensively considered land cover, topography, climate, and other conditions to draw the spatial changes in the conflict trend between natural tourism development and ecological protection in China (Zeng et al 2022). Liu et al revealed the circular and causal relationship between urban ecological land encroachment and tourism growth by quantitatively constructing a probability curve for urban expansion (Liu et al 2021). Liu et al used the 'Driver-Pressure-State-Impact-Response (DPSIR)' framework to construct a dynamic multi-scenario simulation model based on Chinese provincial regions, and found that economic factors are no longer the biggest driving force for tourism ecological security. Instead, service quality, technology, environmental governance and protection, and system response capacity have become important ways to improve tourism ecological security (Liu and Yin 2022). The interplay between the natural environment and the growth of tourism is a multifaceted and complex phenomenon that requires further exploration. To achieve sustainable development, scholars need to continue exploring this relationship from a comprehensive perspective, using various theoretical frameworks and econometric models.
Overall, whether examining the impact of tourism economy on the ecological environment, the effect of the ecological environment on the tourism industry, or the interaction between the two, has been confirmed. However, existing literature generally overlooks the role of 'human beings' in the interaction between tourism and the ecological environment. Existing research generally regards humans as the experience objects of tourism activities, protectors or destroyers of the ecological environment, but lacks research on the subjective quality of life and happiness of humans in the process of tourism development and ecological environment improvement (destruction). Therefore, this is also an important reason why this article introduces the concept of 'EWP' to better analyze the relationship between tourism economy and the ecological environment.

Study area
This study examined 58 cities in China, as listed in the Tourism Statistical Yearbook (National Tourism Administration of China 2020), that are characterized by abundant tourism resources, high tourist activity, and a significant role of tourism in urban economic development. These cities are representative of exploring the relationship between urban EWP and TED. The list of cities (figure 1) includes both large and small cities, such as Beijing, Shanghai, and Chongqing, as well as smaller cities such as Beihai, Chengde, and Lianyungang. The cities are spread across various regions of China, including the north, south, east, and west, providing a diverse sample set for the study. The selected cities are of great importance to the tourism industry, with a substantial number of tourists visiting each year. Thus, the effect of tourism on the environment and the measures taken for environmental protection and improvement are of great significance in these cities. The study aims to explore the correlation between EWP and TED in these cities and provide insights into the sustainable development of the tourism industry.

Data source
This study covered the period from 2004 to 2019, and the data mainly came from two sources. For TED data, it mainly came from the statistical yearbooks and national economic and social development statistical bulletins published by 58 major tourism cities each year. For EWP data, it mainly came from the 'China City Statistical Yearbook', statistical yearbooks or national economic and social development statistical bulletins published by 58 major tourism cities each year. For some missing data, interpolation was used to fill in the gaps. In order to enhance comparability, some data were calculated for the second time (per capita). The data sources for each case city were detailed in table A1 (appendix).

Research framework
The implementation framework of this study, as shown in figure 2, mainly included several steps. Firstly, the indicator system of TED and EWP was constructed, and data collection work was carried out. Secondly, after completing the data collection of the two systems, the value of EWP was calculated using the super SBM model, while the value of TED was calculated using the entropy weight comprehensive evaluation model. Finally, after obtaining the values of EWP and TED respectively, on the one hand, the coupling coordination model was used to analyze the coupling coordination relationship between EWP and TED, and on the other hand, the Panel Vector Auto-regression (PVAR) model was used to analyze the interactive response between EWP and TED.
3.4. Construction of index system 3.4.1. EWP index system The primary objective of EWP is to maximize the output of well-being while minimizing resource input and environmental costs . Thus, when measuring the EWP of a region, it is typically evaluated from two dimensions: ecological input and well-being output (table 1).
For the ecological input dimension, this study selected resource consumption and environmental pollution as the ecological input indicators by referring to the indicator selection methods of previous studies (Zhu and Zhang 2014, Bian and Ren and et al 2020, Deng et al 2021, Xu et al 2021. Environmental pollution comprises waste water, waste gas, and soot discharge, which were respectively measured by per capita waste water discharge, per capita waste gas discharge, and per capita soot discharge. Resource consumption consists of power consumption, water resource consumption, and land resource consumption, which were respectively measured by per capita electricity consumption, per capita water consumption, and per capita construction land area. The selection of these ecological input indicators is critical to accurately evaluate the EWP of a region. The environmental pollution and resource consumption indicators can reflect the environmental impact of economic activities and the efficiency of resource utilization, respectively. By measuring these indicators, it is possible to identify the strengths and weaknesses of a region's EWP and provide insights into how to improve its sustainability.

TED index system
Regarding the measurement of TED, following previous studies ( Tang   importance of tourism in regional economy. In terms of absolute income of tourism economy, it was measured by per capita tourism total income, tourism output density and tourism total income compared with the growth rate of the previous year. In terms of the importance of tourism in regional economy, the proportion of tourism revenue in GDP, the elasticity coefficient of tourism to GDP growth, and the proportion of tourism revenue in the output value of the tertiary industry were respectively measured.

Super SBM model
The conventional DEA (Data Envelopment Analysis) models, including the CCR (Charnes-Cooper-Rhodes) and BBC (Banker-Charnes-Cooper) models, utilize radial angle-based measurement techniques that solely take into account either input or output perspectives. The drawback of this methodology is that it does not account for the relaxation of input and output, which can result in measurements that do not accurately reflect the actual situation (Deng et al 2021). To address this inadequacy, the non-radial Slack-Based Measure (SBM) model was suggested by Tone, which availably addresses the issue of input factors redundancy or deficiency (Tone 2001). However, like traditional DEA models, the SBM model is unable to discern the efficiency difference between effective Decision-Making Units (DMUs) for DMUs with a maximum efficiency score of 1. To overcome this limitation, Tone developed the super-efficiency SBM model by building on the SBM model. The super-efficiency SBM model combines the benefits of both the SBM model and the super-efficiency DEA model, enabling it to efficiently compare and evaluate advanced DMUs (Tone 2002). As the variable returns to scale assumption was considered, this study employed the non-radial and non-angular super SBM model to evaluate the EWP. The integration of the super SBM model enhances the accuracy of measuring environmental well-being performance by overcoming the shortcomings of traditional DEA models (Deng et al 2021). By adopting the super SBM model, this study aims to provide a more comprehensive and accurate evaluation of the environmental wellbeing performance of different regions. This model overcomes the limitations of traditional DEA models and provides a more nuanced understanding of the efficiency of DMUs.

Dimensions Indicators Units
Absolute income of tourism economy per capita tourism total income $10,000/person tourism output density $10,000/km 2 tourism total income compared with the growth rate of the previous year % Importance of tourism in regional economy the proportion of tourism revenue in GDP % the elasticity coefficient of tourism to GDP growth The non-radial and non-angular super SBM model was utilized as the basis for constructing the model for measuring EWP. The model is designed with several components, including δ which represents the DEA super efficiency value, λj as the weight vector, x and y as input and output variables, and m and S as the number of input and output variables, respectively. The slack variables of input and output are represented by S-and S+, respectively. If δ is greater than or equal to 1, the decision unit is deemed efficient, but if δ is less than 1, the decision unit is considered ineffective. The degree of EWP is determined by the value of δ, with higher values indicating higher levels of EWP.

Entropy weight-comprehensive model
(1) To weight the evaluation index, the entropy weighting method was employed. This method effectively addresses the issues of subjectivity and information overlap among multiple indicator variables that are inherent to the subjective weight method. The entropy weight method is highly objective, and its utility value is deeply reflected in the index information entropy value. It is widely applied in social economy and other research fields (Wang et al 2013, Wang andHe 2020). For this method, the underlying principle of assessment is that the larger the difference in the value of the object being assessed in a particular index, the more significant the object becomes, and thus, it is assigned a higher weight value. The equation for the entropy weighting method is presented as follows (Tang 2015):

/
The method of entropy weighting was utilized to assign weights to the index being evaluated, is calculated as follows: e j represents the index entropy value, and its value ranges between 0 and 1. y ij denotes the normalized value of the ith sample and the jth index, obtained by non-dimensionalization and homotropization of each index evaluation value using the min-max method. M and n correspond to the sample size and indices, correspondingly. k = 1 / lnm, and w j represents the weight of the indicator. This formula ensures that the weight value of the index is objective and accurate, reflecting the importance of each index in evaluating TED. By employing the entropy weighting method, this study guarantees a more comprehensive and accurate evaluation of TED, taking into account the significance of each index in the evaluation process.
(2) A comprehensive evaluation model was employed to measure TED system. The formula for the model is expressed as (Tang 2015): Where, S represents the comprehensive index of the system; Wj indicates the weight of each indicator in the system. Yk represents the evaluation value of each indicator.

Coupling coordination degree model
The notion of 'coupling' was first introduced in the realm of physics to describe the consistency and compatibility between two systems during their evolution. It reveals the process of how a discordant relationship can evolve into a harmonious one over time. In present times, scholars widely utilize the coupling coordination model to evaluate the association between multiple systems. As a result, the CCD has become a useful tool to determine the coupling relationship between EWP and TED. The formula for calculating the CCD is expressed as follows (Sun et al 2021): In the given equation, C represents the coupling degree between the two systems, where S1 and S2 denote the comprehensive indices of EWP and TED, respectively. Additionally, D stands for the CCD, and T denotes the comprehensive coordination index among the systems. The value of D ranges from 0 to 1, where a higher value of D indicates a higher level of coordinated development between subsystems. The coefficients α and β are undetermined, with their sum equalling to 1. As EWP and TED complement each other, both are assigned a value of 0.5. With reference to relevant studies , Ariken et al 2021, Huang et al 2021, the CCD between EWP and TED was divided into five levels: High Coordination (0.8 < D 1), Moderate coordination (0.6 < D 0.8), Reluctant Coordination (0.4 < D 0.6) and Moderate Imbalance (0.2 < D 0.4) and Serious imbalance (0 < D 0.2).

Panel vector auto-regression model
The dynamic panel model, known as Panel Vector Auto-regression, incorporates fixed effects. Endogeneity can be attributed to all variables in the model, without the need for distinguishing between exogenous and internal variables. The association between EWP and TED is not a straightforward linear relationship that goes in one direction. The PVAR model can treat the target variable as an internal system, accounting for all variables and lag periods to reflect the interactive relationship of each variable. To investigate the dynamic relationship between EWP and TED, this study employs the PVAR model. The PVAR model can be expressed using the following equation (Charfeddine and Kahia 2019): Where, i representing different cities and t standing for the year. Yit constitutes two column vectors, incorporating EWP and TED. In addition to the time effect variable (β t ) and the random disturbance term (ε it ), the analysis is impacted by various components of the PVAR model, including the intercept term vector γ 0 , the lag order p, the parameter matrix of lag order j (γ j ) and the inclusion of a variable with individual fixed effects (α i ). This comprehensive formula allows for a thorough investigation into the dynamic relationship between EWP and TED, taking into account all relevant variables and individual effects across various cities and years.

Spatial-temporal characteristics analysis of EWP and TED
This study used the super efficiency SBM model and the entropy weight-comprehensive evaluation model to calculate the EWP and TED indices of 58 major tourism cities in China from 2004 to 2019. The entire study period was divided into four stages, and the data were aggregated and calculated as the mean value for each stage using a four-year interval to overcome errors caused by data fluctuations. Figure 3 visualized the evolution process of EWP of 58 major tourism cities in China from 2004 to 2019.

Spatial-temporal characteristics of EWP
In the first stage (2004)(2005)(2006)(2007), the overall level of EWP in each city showed a distribution pattern of lower levels in coastal cities and higher levels in some inland provincial capital cities. Specifically, the cities with higher levels of EWP during this period were mainly Changchun, Shenyang, Nanjing, Huangshan, Nanchang, Haikou, Sanya, Shantou, Chengdu, Datong, Taiyuan, and other cities. Qinhuangdao, Yantai, Shanghai, Ningbo, Fuzhou, Shenzhen, and other cities were at the lowest level. This is important because during this stage, coastal areas were in a faster stage of urbanization development, with large resource consumption and serious environmental pollution, and overall tended to focus on improving development speed while neglecting the improvement of development quality.
In the second stage (2008-2011), similar to the previous stage, cities with lower levels of EWP were still mostly distributed in coastal areas. However, a more obvious change during this stage was that the number of cities with higher levels of EWP decreased, and the number was small, with only a few cities such as Taiyuan, Nanjing, Zhongshan, Haikou, and Sanya. This is important because during this stage, China as a whole was still in a stage of extensive urbanization and rapid development.
In the third stage (2012-2015), a prominent feature of this stage was that the number of cities with the highest and lowest levels of EWP decreased, breaking the development pattern of polarization. It is worth noting that during this stage, Qinhuangdao's EWP level significantly improved and reached a higher level. In addition, Sanya's EWP remained at a high level as in the previous stage. Many coastal cities saw a significant improvement in their EWP, mainly because during this stage, China gradually began to pay attention to the quality of urbanization development and focused on improving basic public services and residents' satisfaction and happiness.
In the fourth stage (2016-2019), compared with the previous stage, the EWP level of many cities has significantly improved, and the number of cities with the highest level of development has increased significantly. Cities such as Harbin, Beijing, Jinan, Xi'an, Nanjing, Wuhan, Nanning, Guangzhou, Shenzhen, and Haikou all had the highest level of EWP. It can be seen that most of these cities with the highest level are provincial capital cities. This is mainly because provincial capital cities are generally more economically developed and have gradually begun to enter the transformation stage of urban development. During this stage, they not only focused on improving GDP but also made good practices in ecological environment protection and improving residents' living quality. Figure 4 visualized the evolution process of TED of 58 major tourism cities in China from 2004 to 2019.

Spatial-temporal characteristics of TED
In the first stage (2004)(2005)(2006)(2007), there were few cities with a high level of TED, only Beijing, Shanghai, Huangshan, Xiamen, Shenzhen, Sanya, and other cities. During this stage, China's tourism was still in its infancy, and tourism experiences were mainly focused on sightseeing tours of classic scenic spots. Cities such as Beijing, Huangshan, and Sanya had a high level of tourism economy because they had rich natural, cultural, and historical tourism resources, which had great tourism appeal. Shanghai, Xiamen, and Shenzhen were mainly due to their status as special economic zones or international metropolises, with a large flow of domestic and foreign visitors, resulting in a high level of tourism economy.
In the second stage (2008-2011), similar to the previous stage, cities such as Shanghai, Huangshan, Beijing, Xiamen, Shenzhen, and Sanya still had the highest level of TED among major tourism cities in the country. In this stage, cities with a high level of tourism economy were still concentrated in the Yangtze River Delta and Pearl River Delta regions. It is worth noting that Guangzhou, Zhuhai, and Guiyang also saw a significant improvement in their TED level during this stage, reaching a higher level of development.
In the third stage (2012-2015), cities with a high level of TED in the previous stage also had the highest level among all tourism cities in this stage, and the overall level of TED also improved. In addition, the TED level of cities such as Nanjing, Wuhan, Hangzhou, Wuxi, Suzhou, and Tianjin increased by one level compared to the previous stage. In the fourth stage (2016-2019), the TED level of most cities has been significantly improved. Cities such as Shanghai, Huangshan, Xiamen, Guangzhou, Shenzhen, Guiyang, and Sanya had the highest level of TED in this stage. Except for Chongqing, tourism cities along the Yangtze River were at a medium to high level of development. In addition, the tourism economy level of tourism cities in the Bohai Rim region has significantly improved compared to the previous stage, with most of them at a medium to high level.

Spatio-temporal analysis of the CCD between EWP and TED
To explore the coupling and coordination relationship between EWP and TED, the CCD of 58 major tourism cities in China from 2008 to 2018 was calculated and analyzed (figure 5). According to the coupling coordination judgment standard, the CCD value was divided into five intervals, and the spatiotemporal evolution process was understood by the four time periods.
The cities of high coordination. Among the 58 tourism cities in China in the first stage (2004)(2005)(2006)(2007) and the second stage (2008)(2009)(2010)(2011), there was no highly coordinated city, mainly because China's tourism industry was still in the rapid development stage, and environmentally friendly and sustainable economic and social development was still in the exploratory stage. In the third stage (2012)(2013)(2014)(2015) and the fourth stage (2016-2019), only Sanya's CCD reached a highly coordinated level, maintaining a high level of steady growth. The average CCD values in the two stages were 0.8232 and 0.8970, respectively. The reason why Sanya has developed a higher level of coordination is mainly due to its good ecological environment and tourism development conditions. Sanya is located at the southernmost tip of Hainan Island in China and is known as the 'Oriental Hawaii'. It is the only tropical coastal international tourist city in China that can be visited and traveled all year round. In 2016, the Chinese government positioned the island of Hainan, where Sanya is located, as an 'international tourism island' for construction and development, aiming to build it into a world-class island leisure and vacation tourism destination and a national ecological civilization construction demonstration zone. The government's series of support measures have greatly promoted the improvement of Sanya's TED and EWP.
The cities of moderate coordination. In the first stage (2004)(2005)(2006)(2007), there were only two moderately coordinated cities among the 58 cities, Sanya and Huangshan. The CCD values of the two cities were 0.7545 and 0.6429, respectively. In the second stage (2008-2011), Huangshan's CCD decreased, and only Sanya remained a moderately coordinated city. In the third stage (2012)(2013)(2014)(2015), the number of moderately coordinated cities increased to five, including Nanjing, Beijing, Guangzhou, Shenzhen, and Huangshan. Except for Huangshan, the other four cities are economically developed large cities with high population mobility, developed tourism industry, high urbanization level, and complete public service facilities. This is also an important reason why their CCD is at a good level. Huangshan, as a classic tourist city in China, has abundant tourism resources, high tourism visibility, and good ecological environment, which contributes to its higher CCD. In the fourth stage (2016-2019), the number of moderately coordinated cities increased rapidly from five in the previous stage to 25. This is mainly because the tourism consumption demand of Chinese residents gradually increased in this stage, and the tourism economy in various regions has been greatly improved. On the other hand, China's ecological civilization construction has gradually shown results in this stage, and cities have moved towards high-quality development that takes into account both economic and ecological benefits.
The cities of reluctant coordination. In the first stage (2004)(2005)(2006)(2007), there were 21 basically coordinated cities, including not only international metropolises such as Beijing, Shanghai and Guangzhou, but also provincial capitals such as Zhengzhou, Taiyuan, Urumqi, Xi'an, Wuhan and Chengdu, as well as cities with relatively good tourism resources such as Datong, Beihai and Xiamen. In the second stage (2008)(2009)(2010)(2011), the number of cities in reluctant coordination was 25, which was increased compared with the number of cities in the previous stage. In the third stage (2012)(2013)(2014)(2015), the number of cities in reluctant coordination increased to 38. During this period, the largest number of cities fell under the category of basic coordination among the five coupling coordination types. However, in the fourth stage (2016-2019), the number of cities classified under reluctant coordination declined to 28. This was primarily due to the fact that more cities had progressed towards a higher level of coupling coordination, such as moderate coordination.
The cities of moderate imbalance. In the first stage (2004)(2005)(2006)(2007) and the second stage (2008-2011), there were many cities with moderate imbalance, with 33 and 31 cities, respectively. This is closely related to the low level of TED at that time and the generally extensive mode of urban economic growth. In the third stage (2012)(2013)(2014)(2015), the number of cities with moderate imbalance significantly decreased to 14. Except for Yinchuan and Shijiazhuang, these cities are mostly China's second-and third-tier cities. In the fourth stage (2016-2019), the number of cities with moderate imbalance further decreased to only four cities, Zhangzhou, Quanzhou, Nantong, and Yinchuan. This is mainly because most cities in this stage not only focus on developing the tourism economy but also pay attention to improving the quality of the ecological environment and people's living standards.
The cities of serious imbalance. There are few cities where CCD are in a state of severe imbalance. In the first stage (2004)(2005)(2006)(2007), only Zhangzhou and Nantong were in a state of severe imbalance, with CCD values of 0.1986 and 0.1995, respectively. In the second stage (2008-2011), only Zhangzhou was in a state of severe imbalance, with a CCD of 0.1956, which was slightly lower than the previous stage. It is worth noting that Zhangzhou's CCD has always been at a low level, with an evolutionary feature of 'severe imbalance-severe imbalance-moderate imbalance-moderate imbalance'. This is closely related to the weak position of Zhangzhou's TED among the 58 tourism cities.
Overall, the coupling coordination degree of EWP and TED in the 58 tourism cities has been improving to varying degrees, but there is significant spatial differentiation in CCD. In the first stage (2004)(2005)(2006)(2007) and the second stage (2008-2011), cities with higher CCD were mostly cities with high-quality and abundant tourism resources or economically developed cities. In the third stage (2012-2015) and the fourth stage (2016-2019), the CCD of the vast majority of tourism cities has significantly improved, especially in many inland tourism cities.

Data test
(1) To prevent spurious regression in panel data, a unit root test was deemed necessary. To ensure the test's accuracy, both the LLC and IPS tests were employed simultaneously. As shown in table 3, the original EWP and TED series did not completely reject the null hypothesis of the existence of unit root of variables. However, the first difference series of dEWP and dTED both refuted the hypothesis of unit root for the variables. As a result, we can conclude that the PVAR model can be implemented to estimate dEWP and dTED. Ultimately, this approach avoided spurious regression and ensured the accuracy of the analysis.
(2) To ensure the validity of parameter model estimation, the optimal lag order was determined using AIC, BIC and HQIC criteria prior to PVAR estimation. According to the results in table 4, the analysis identified the optimal lag order as being third-order, which led to the construction of a third-order PVAR model. Ultimately, this method ensured that the selected lag period was appropriate for the analysis.

GMM parameter estimation
According to the selected optimal lag order, the system GMM was used to estimate the PVAR model of panel data, and the Helmert Procedure method was used to eliminate individual effects. The estimation results are shown in table 5. Where h_dEWP and h_dTED respectively represent the series of EWP and TED using forward mean difference method to remove fixed effects. Table 5 illustrates that TED had a positive impact on itself, with impact coefficients of 0.420, 0.436, and 0.206 in phase 1, 2, 3 lag, respectively. All of these coefficients passed the 1% significance level test, indicating that TED had a certain degree of inertia. The impact of EWP in the first and second stages on TED was not significant, while the impact of EWP in the third stage was significant at the 10% significance level. On the other hand, when EWP was the explained variable, the impact coefficients of the lag stages 1, 2, and 3 on EWP were −0.426, −0.190, and −0.096, respectively. All of these coefficients passed the 1% significance level test. Although the  Note: * represents the optimal lag order of the model.
impact of EWP on itself was negative, the extent of its impact decreased with the extension of the period, indicating the progressive effect of its development. Moreover, the impact coefficients of TED lagged in the first and second stages on EWP were statistically significant at the 1% significance level, with values of 7.112 and 4.865, respectively. However, the impact of TED in the third lag period on EWP failed to pass the significance test. Overall, these findings reveal the complex interplay between TED and EWP, with TED exhibiting inertia while EWP displays a progressive effect on its development.
Overall, there was a significant mutual promotion mechanism between EWP and TED. However, it should be noted that the generalized moment estimation of the parameters of PVAR model can only reflect the dynamic simulation process between variables in a macroscopic way, and cannot specifically describe the relationship of causal logic between the variables, the dynamic transmission mechanism and the contribution of impact variables, which should be further investigated by impulse response functions and variance decomposition tools.

Impulse response
Impulse response analysis is a method of studying the dynamic impact of a variable on each variable in the system by using impulse response functions. In order to intuitively characterize the dynamic interaction between EWP and TED, this study conducted a one-standard-deviation shock to each variable, set the duration of the shock to 10 periods, and obtained the results of orthogonal impulse response functions (figure 6). The horizontal axis in the figure represents the number of lags, and the vertical axis represents the response polarity and strength. The red line in the middle represents the impulse response value of the response variable after giving a one-standard-deviation shock to a certain shock variable, and the upper and lower curves represent the boundary of the 95% confidence interval. Based on figure 6, the following conclusions can be drawn: When TED is given a one-standard-deviation shock to itself, the overall effect is positive. It shows a downward trend in the first period, and begins to rise in the second period. As the number of lags increases, the effect gradually tends to be flat. When EWP is given a one-standard-deviation shock to itself, it has a negative effect in the first period and gradually tends to zero after the second period. When TED is given a one-standarddeviation shock to EWP, it shows a slight negative effect in the first six periods, iterating around −0.001, and tends to be flat after the sixth period. This may be mainly related to the relatively extensive development of the tourism industry in the past. When EWP is given a one-standard-deviation shock to TED, it shows an overall positive effect, reaching the highest value in the first period, and then gradually decreasing, tending to be flat at 0.05. Therefore, it can be concluded that when EWP improves and brings about regional ecological environment improvement and public service level improvement, it can significantly promote the development of regional tourism economy.

Variance decomposition
Variance decomposition allows for the decomposition of the variance of forecast errors of each endogenous variable into components related to each endogenous variable based on its causes. This enables the assessment of the impact of each shock on changes in the system's endogenous variables. The outcomes of the variance decomposition are presented in table 6.
According to the variance decomposition of TED (table 6), it was predicted that the contribution rate of TED to itself is 1.000 in phase 1, then gradually decreased and tended to stop in phase 4, with a contribution rate of 0.992. The contribution rate of EWP to TED started to appear in the second period and stopped after the first four periods, with the contribution rate remaining at 0.008. The above decomposition results showed that the tourism industry mainly relied on its inertia to develop continuously, and TED can only promote the improvement of EWP in the short term due to the emergence of new tourism formats and diversified tourism demands.
According to the variance decomposition of EWP (table 6), it was predicted that the contribution rate of EWP to its own was 0.984 in the first period and will decrease to 0.782 in the 10th period. Regarding the variance decomposition of EWP, it is worth noting that while the variance contribution rate throughout the period was primarily due to EWP itself, the significant contribution of TED to EWP cannot be disregarded. The contribution rate of TED to EWP increased from 0.016 in phase 1 to 0.120 in phase 5 and then to 0.218 in phase 10, accounting for a large proportion of the total contribution rate and showing an increasing trend. This indicates to some extent that TED has an important effect on EWP. The above conclusion further verifies the interactive relationship between EWP and TED.

The spatiotemporal evolution logic of EWP and TED
This study explores the relationship between TED and EWP in major tourism cities based on the concept of ecological welfare, which provides a more inclusive perspective than the narrow viewpoints used in previous  research. In the Anthropocene era, people not only focus on the global impact of human activities on ecosystems but also value the harmony and sustainable development between humans and nature. Based on this concept, we found that the 'development of humans' in the relationship between ecology and tourism has been overlooked to some extent. Considering that ecological welfare effectively explains the relationship between ecological conservation and human well-being, we introduce the concept of ecological welfare to explore the relationship between tourism and ecology in tourism cities. This indicates that our research perspective is innovative compared to previous literature. In the sustainable development process of different regions in China, there are significant regional differences in EWP due to the driving forces of economic growth and welfare improvement (Fang and Xiao 2019). By exploring the spatial and temporal characteristics of EWP, we found that the EWP of tourism cities in the southern region is generally higher than that in the northern region, and the EWP in the central and western regions is weaker than that in the eastern region, which is consistent with the research results of Liu et al (Wang and Duan 2023). This suggests that the economic transformation and upgrading in the eastern and southern regions have achieved significant results, and ecological environment governance and quality of life improvement play a demonstrative role nationwide, while the central and western regions may be caused by the lack of economic growth momentum and excessive consumption of natural resources (Fang and Xiao 2019). At the same time, the EWP of provincial capital cities is generally higher, as Li's research has shown, and the fundamental reason for this difference lies in the snowball effect of provincial capital cities themselves and their suction effect on surrounding cities (Li 2022).
In terms of the spatial and temporal characteristics of the tourism economic level, we found that the TED level exhibits significant regional heterogeneity, with high TED cities widely distributed in the eastern coastal city cluster and less distributed in the central and western regions, which is similar to Li's research results. The reason for this is that the eastern region is in a leading position in terms of geographical location, economic strength, and innovative research capabilities compared to other regions . In Fang's research, the tourism economy in the eastern region belongs to the 'endogenous-driven type,' and TED has a strong foundation. It can support the stable development of the industry in the short term by relying on 'internal circulation.' The central and western regions belong to the 'exogenous-driven type,' which is closely related to external factors such as transportation, market, regional economy, and advanced industrial structure. It is difficult to fully activate these factors only by the industry and local society. Therefore, communication and cooperation between regions, especially with the eastern region, are crucial for the healthy and stable development of the tourism industry in the central and western regions, and 'external circulation' is more important (Fang et al 2023).

The interactive relationship between EWP and TED
This study differs from previous research in that it analyzes the coupling and coordination relationship between TEF and EWP from a static perspective while exploring the long-term interactive relationship between TED and EWP from a dynamic perspective, resulting in more comprehensive conclusions. Previous research has shown that there is a certain convergence in the spatial and temporal evolution of tourism economic level and EWP, as demonstrated in Wang's research (Wang and Liu 2019). The TED and EWP are interdependent, interconnected, and inseparable, which is the result of many complex factors. The development of the tourism industry can stimulate emission reduction activities and economic growth incentives, and is considered a trigger for destination green growth (Marsiglio 2015). The development and growth of the tourism industry can help to ensure the investment of ecological protection funds, the introduction of advanced concepts and technologies, and the enhancement of environmental awareness of tourism stakeholders, thereby promoting regional ecological environment improvement and sustainable development (Wang and Liu 2019). On the other hand, the ecological environment is a passive carrier and recipient of tourism economy, and its changes in turn affect human tourism economic activities and EWP, and then respond to these changes through environmental protection policies, economic regulation policies, and changes in awareness and behavior, as explained by Guo et al (2020). This viewpoint explains the dynamic interactive relationship between TED and EWP, as well as the impact of EWP on itself.

Policy implications
The research results can enrich the theory of sustainable tourism and ecological tourism from the perspective of ecological welfare, and provide theoretical guidance and practical inspiration for the healthy development of tourism economy in tourist destinations. Firstly, attention should be paid to the regional differences in the ecological environment, and reasonable environmental policies should be formulated to promote a winwin situation for human welfare improvement and resource conservation. Each tourist city should actively respond to the government's ecological and sustainable development policies, and based on the specific economic development situation of each city, formulate reasonable and strict environmental protection measures. Promote the transformation and upgrading of the industrial structure according to local conditions, cultivate new tourism formats to promote local economic green development, and improve the efficiency of human welfare output caused by natural resource consumption. Secondly, the economic growth mode of the tourism industry should be transformed to effectively improve the utilization and output rate of tourism resources. Each tourist city should strengthen the structural reform of tourism supply-side, carry out deep experience tours, eliminate disorderly development and blind construction, abandon the extensive tourism development mode with high consumption and pollution, and actively transform to the ecological tourism development mode. In addition, when developing tourism resources in cities with relatively fragile ecology, the sensitivity of the local environment should be considered, and moderate and targeted tourism development should be carried out to minimize the negative environmental effects. Thirdly, while improving the level of local economic development, attention should also be paid to improving the quality of life of local residents. Each tourist city should increase the construction of basic public services and promote the improvement of social welfare level. While creating a good tourism environment, it can also bring a better living environment for local residents. Continuously guarantee and improve the living standards of the people to meet the growing needs of a better life.

Conclusion
With the advent of the mass tourism era, the development scale of the tourism industry has unprecedentedly expanded. How to scientifically measure and evaluate the regional economic, social, and ecological effects of tourism development and their pros and cons has become a key issue in achieving regional sustainable development. Therefore, this study takes 58 major tourism cities in China as the research object, based on panel data from 2004 to 2019, and constructs a measurement index system for EWP and TED. Through methods such as entropy-weighted comprehensive evaluation model, Super-SBM model, coupling coordination model, PVAR model, and spatial visualization, this study explores the spatiotemporal evolution of EWP and TED's CCD in major tourism cities in China and analyzes their dynamic interactive relationship. The main findings are as follows: In terms of the spatiotemporal changes of EWP, from 2004 to 2019, the EWP of major tourism cities in China has improved overall, although there is some spatial unevenness. While many coastal tourism cities have seen significant improvements in EWP, many inland cities still lag behind. In addition, provincial capital cities have performed outstandingly, especially from 2016 to 2019, where most cities with the highest EWP were provincial capitals.
Regarding the spatiotemporal changes of TED, from 2004 to 2019, the TED level of major tourism cities in China has significantly improved, with remarkable achievements in tourism development. Although the gap in TED levels between tourism cities is narrowing, there are still some spatial differences. The TED level of tourism cities in the Bohai Rim and the Yangtze River Basin is relatively good and has progressed significantly.
In terms of the CCD between EWP and TED, the coupling coordination degree of most tourism cities in China has improved to varying degrees from 2004 to 2019, especially in many inland tourism cities where CCD has greatly improved. From 2004 to 2019, only Sanya's CCD has consistently remained at a moderate to high level and has steadily increased. Most cities were in a state of basic coordination or moderate imbalance from 2004 to 2015. However, from 2016 to 2019, many cities have significantly improved their CCD, rising to a moderate coordination state, but the majority still remain in a state of basic coordination.
In terms of the dynamic interaction between EWP and TED, TED has a positive impact on itself and shows a gradually decreasing trend, while EWP has experienced a transition from positive to negative impact, gradually flattening out. Among them, EWP has a more significant impact on itself. TED has a weak negative impact on EWP, while EWP has a positive promoting effect on TED, but the positive impact weakens as the period increases. Overall, there is a dynamic interaction between EWP and TED in major tourism cities in China, but a virtuous interaction has not yet been formed.
This study is not without limitations.
(1) Although the measurement indicator system constructed in this study has good representativeness for measuring EWP and TED, with the increasing diversity of data statistics and acquisition channels in the future, more evaluation indicators can be incorporated to further optimize the evaluation results.
(2) This study mainly explores the mechanism of the interaction between the two macro systems of EWP and TED, but further refinement research is needed for the specific interaction between the dimensions within the two systems.
(3) This study focuses on the CCD between EWP and TED, and further testing and exploration are needed in the future for the factors that affect the CCD of different tourism cities.