Coupling Tourism Carbon Emission Efficiency and Economic Resilience in China’s Yellow River Basin

2.1 Tourism Economic Resilience

The concept of “resilience” originated in physics and engineering, denoting an object’s capacity to recover to its original form following deformation (Manyena 2006). Through interdisciplinary integration and conceptual evolution, resilience has gradually been introduced into fields such as regional and tourism economics. It has also acquired broader connotations (Luthe and Wyss 2014; Yang et al. 2023). Some scholars have expanded the concept of economic resilience from an evolutionary perspective. They argue that economic systems are not static. Instead, they restore core functions through structural adjustments when facing shocks. This perspective has driven the development of economic resilience theory. It has prompted the academic community to gradually form a comprehensive and well-structured resilience framework. Economic resilience is widely recognized as the capacity to resist, adapt, and recover from shocks (Martin 2012; Zhou and Chen 2025). As an important component of regional economic resilience (Ulucay and Okumuş 2024), TER directly impacts the sustainable development of the tourism industry (Agayi and Gündüz 2020). TER refers to the ability of a tourism economic system to maintain relative stability, recover its original state, or even achieve a new equilibrium after encountering external shocks (Yang et al. 2022). Tourism economic systems are highly integrated and complex (Ma et al. 2019). They are affected by economic fluctuations, social changes, and natural disasters. Regions heavily reliant on tourism have been particularly vulnerable in times of crisis (Wut et al. 2021). However, tourism economic systems also demonstrate adaptability and dynamic adjustment capabilities. They build resilience through continuous responses to external pressures, thereby enhancing their future resilience.

Most of the existing research has adopted two measurement methods: the core variable approach and the index system method. The core variable approach measures an economy’s stress resistance and recovery capacity by analyzing the differences between the actual performance and expected values of key economic indicators such as employment, unemployment rate, and GDP. However, this method requires establishing a baseline state to determine the impact of the shock. This makes it challenging to fully capture the multi-dimensional characteristics of resilience. To overcome the limitations of single-index measurement, an increasing number of scholars are developing multi-dimensional comprehensive index systems. For example, Martin et al. (2016) conducted a systematic study on regional resilience by dividing the ability to cope with economic crises into four dimensions: resistance, recovery, re-orientation, and renewal. Faggian et al. (2018) focused on resistance, recovery, and renewal to empirically assess regional economic resilience. Wang and Guo (2023) developed an assessment framework, which evaluates the resilience of inbound tourism based on resistance, recovery, adjustment, and renewal.

2.2 Tourism Carbon Emission Efficiency

Research on tourism carbon emission efficiency (TCE) primarily focuses on measurement methods, influencing factors, and spatiotemporal relationships. Early studies on TCE mainly employed the single-ratio method for measurement (Wang et al. 2022). Although this method provided a basic understanding of TCE, it failed to comprehensively reflect the multidimensional characteristics of TCE. As research progresses, scholars have started to use the indicator system approach with multiple inputs and outputs to calculate the carbon emission efficiency of the tourism industry (Li et al. 2022; Wang et al. 2025c). These models incorporate inputs such as capital, labor, and energy consumption, alongside outputs like GDP and tourism revenue. With further methodological advancements, academic attention has shifted toward identifying the driving forces behind changes in TCE. The Kaya identity, LEAP model, LMDI model, and STIRPAT model are commonly used to analyze influencing factors (Jiang et al. 2025; Zeng and He 2023). Empirical studies consistently demonstrate that economic growth, technological innovation, regulatory intensity, and industrial structure adjustment play decisive roles in shaping TCE. The spatiotemporal dynamics of TCE have become a crucial research focus, employing techniques such as spatial autocorrelation analysis, Tobit regression, and social network models. These methods reveal spatial clustering patterns, regional disparities, and network structures within the tourism system (Wang et al. 2025c; Xia et al. 2022; Zhong et al. 2025). Research findings indicate that the carbon emission efficiency of inter-provincial tourism in China is high in the eastern regions and low in the western regions, demonstrating a pronounced spatial agglomeration pattern (Xu et al. 2023).

2.3 Relationship Between TCE and TER

Although previous studies have examined the issues of TCE and TER independently, research on their coordinated development remains limited. Recently, some scholars have begun to focus on the interrelationship between carbon emissions and economic development. On one hand, scholars have analyzed the coupling relationships between carbon emissions and economic growth from various aspects. These include tourism-related carbon emissions, transportation, energy consumption, and carbon emission efficiency (Wang et al. 2025c,b). On the other hand, scholars explored the interactive coupling relationships between carbon emissions and economic development factors such as new urbanization, environmental protection, regional innovation, and industrial structure upgrading (Jiang et al. 2022; Lu et al. 2021; Wang et al. 2022a). Additionally, some scholars have employed spatial-geographic models to analyze the spatiotemporal evolution patterns of the CCD between environmental and economic systems. They found that the CCD exhibited significant temporal heterogeneity but relatively weak and random spatial variation.

In summary, research on TER and TCE has progressively advanced in both methodology and content, yielding significant findings. However, existing studies exhibit three main limitations. First, the research perspective remains relatively fragmented. Most studies focus on either TER or TCE as isolated dimensions. They lack systematic integration and a comprehensive assessment of the synergistic relationship between the two. Second, the analysis of underlying mechanisms is insufficiently thorough, and a comprehensive theoretical framework has yet to be established to explain the coupled development pathways, mechanisms, and influencing factors between TCE and TER. Third, the exploration of influencing factors primarily relies on linear analytical methods, with inadequate attention to nonlinear relationships and complex interactive effects. To address these gaps, this study constructs a comprehensive evaluation indicator system encompassing both TCE and TER. By employing the CCD model and the random forest model, this study reveals the spatiotemporal evolutionary characteristics and key drivers of their CCD. This study not only deepens the theoretical understanding of the synergistic development of TCE and TER but also provides a scientific basis and practical guidance for promoting low-carbon transformation and resilience in the tourism industry of the YRB.

3.1 The Impact of TCE on TER

TCE serves as a key internal driver in enhancing TER. First, high TCE establishes a solid foundation for TER by optimizing resource allocation. When the tourism industry generates the same revenue with lower energy use and fewer emissions, enterprises can reduce both energy and carbon management costs. Such an improvement in TCE subsequently enhances their market competitiveness. The dual advantage of cost efficiency and competitive strength boosts the system-wide capacity to withstand external shocks (Lu et al. 2019). Second, TCE strengthens TER’s adaptive capacity by encouraging the adoption of low-carbon practices. To meet carbon efficiency goals, the tourism sector promotes low-carbon transportation and upgrades energy-efficient infrastructure. These measures not only reduce emissions and costs but also help build a more flexible and efficient operational model. During market fluctuations or crises, businesses employing this model can swiftly adjust strategies and better manage risks. Finally, TCE safeguards the long-term sustainability of TER by speeding up the low-carbon transformation of the tourism sector. Global climate change and economic uncertainty are on the rise. Strong carbon management capabilities enable the sector to adapt more seamlessly to external disturbances. This adaptability turns into enduring resilience and offers continuous support for the sustainable development of tourism economies.

3.2 The Impact of TER on TCE

Enhanced TER exerts a significant positive feedback effect in improving TCE. Specifically, a higher level of TER stabilizes the tourism industry and strengthens its resilience to external shocks, thereby supporting continued low-carbon investment. A resilient tourism economy can buffer market volatility, maintain industrial vitality, and reduce firms’ exposure to short-term risks. This stability fosters confidence in long-term planning, encouraging enterprises to invest in low-carbon technologies and green practices. This investment lays a solid foundation for the tourism sector’s low-carbon transition (Zafar et al. 2020). Moreover, industrial upgrading and resource optimization constitute the primary pathways through which TER enhances TCE. Through resource integration and structural transformation, the tourism sector moves away from traditional energy-intensive growth patterns. Improved resource efficiency and a greener value chain reduce carbon emissions per unit of output. These changes also enhance carbon efficiency on both the supply and demand sides (Diaz et al. 2020; Guo et al. 2023). Finally, institutional innovation and policy coordination provide sustained momentum for the low-carbon transition driven by TER. Robust governance structures and effective risk management mechanisms ensure consistent implementation of green policies. The interaction between institutional frameworks and technological innovation amplifies the spillover effects of green investments. Through the virtuous cycle of policy incentives and market responses, TER reinforces the internal drivers of low-carbon development.

4.1 Data Sources

This study examines nine provinces in the YRB. Data for measuring TER are sourced from the China Tourism Statistical Yearbook, provincial yearbooks, statistical bulletins, and the National Bureau of Statistics website. The data on carbon emissions are obtained from the China Carbon Accounting Database (CEADs). Considering the availability of data on TCE and TER in the YRB, this study covers the period from 2013 to 2022. Data with inconsistent definitions were adjusted for comparability. Missing values were estimated through linear interpolation.

4.2 Index System Construction

This study employs the Super-SBM model to calculate the TCE. The input indicators selected for this analysis include the scale of fixed asset investment in the tourism sector, the size of the workforce within the industry, and the energy consumption associated with tourism. Specifically, fixed asset investment in tourism is estimated using the perpetual inventory method proposed by Young (2003), which calculates fixed capital stock for each province or region. A depreciation rate of 9.6 % is applied in this calculation. Labor input is represented by the average number of tourism employees at the end of the year, while energy consumption is quantified by the total energy utilized in the tourism industry. Regarding output indicators, desired outputs are measured by total tourism revenue. This reflects the industry’s economic contribution. Undesired outputs are measured by tourism-related carbon emissions. Carbon emissions are not a direct goal of tourism. They are negative by-products of energy use and significantly affect ecosystems and sustainable development. Therefore, they must be included in efficiency assessments. The calculation of tourism energy consumption and carbon emissions employs a “bottom-up” approach. This approach views tourism from the consumer’s perspective. It breaks the sector into three main parts: transportation, accommodation, and tourist activities. Each of these components is calculated independently, and the results are then aggregated to yield an overall outcome (Patterson and Mcdonald 2004).

According to Wang et al. (2020), this study constructs a comprehensive evaluation index for TER. The index is developed from three dimensions: resistance, resilience, and renewal capacity. Resistance measures the ability of the tourism economic system to withstand internal and external risks while maintaining structural and functional stability. Recovery assesses the speed and extent to which the tourism economic system can restore its core functions after experiencing disruptions. Renewal evaluates the ability of the tourism economic system to adapt to new environments through internal optimization and external support. Specific indicators are presented in Table 1.

4.3 Selection of Influential Factors

Based on a review of existing research and the specific characteristics of tourism in the YRB (Bakary et al. 2025; He et al. 2023), this study selects nine indicators across four dimensions: economic level, government policy, tourism industry development, and innovation capacity to analyze their impact on the CCD between TCE and TER in the YRB. Regarding the economic level, regional economic development (ED) is measured by per capita GDP; industrial structure (IS) is represented by the proportion of tertiary industry value added to GDP; and infrastructure levels (INF) are reflected by the ratio of highway mileage to the total population at the end of the year. These three indicators collectively represent the regional economic foundation and its supporting capacity. Regarding government policies, the intensity of environmental regulations (ER) is measured by the ratio of investment in industrial pollution control to industrial added value; the extent of government intervention (GI) is indicated by the proportion of environmental protection expenditures in local fiscal spending, highlighting the government’s guiding role in ecological governance and resource allocation. Regarding the development of the tourism industry, the scale of tourism development (TDS) is measured by the ratio of total tourism revenue to GDP; tourism resource endowment (TRE) is represented by the number of A-rated tourist attractions, which reflects the supply and attractiveness of the tourism sector. Regarding innovation capacity, technological innovation (TI) is measured by the number of approved patent applications; R&D intensity (RDI) is evaluated by the ratio of internal R&D expenditure to regional GDP, indicating the region’s ability to drive innovation through science and technology.

4.4 Research Methods

5.1 Spatiotemporal Characteristics of TCE and TER

5.1.1 Temporal Characteristics

As shown in Figure 1, both TCE and TER exhibit fluctuating upward trends, but the magnitudes of their fluctuations and their stage-specific characteristics differ. Specifically, TCE follows a pattern of “initial decline, mid-term stability, and rapid increase in the later stage”. It decreased from 0.642 in 2013 to 0.433 in 2016. Afterwards, it gradually recovered, reaching a peak of 0.969 in 2020, and then slightly declined to 0.835 in 2022. This indicates a strong rebound of TCE in the YRB after the pandemic and reflects steady progress in the tourism industry’s low-carbon transition. In contrast, TER follows a trajectory of “steady growth – short-term decline – gradual recovery”, rising steadily from 0.254 in 2013 to 0.372 in 2019, without any significant downturn during this period. In 2020, due to the impact of the pandemic, TER temporarily dropped to 0.294. From 2021 to 2022, it showed a slow recovery trend, reaching 0.312 and 0.304, respectively. Although TER has not fully recovered to pre-pandemic levels, its resilience remains strong. This shows that the tourism economy has a clear capacity for self-recovery.

Figure 1:

Trends of TCE and TER in the YRB from 2013 to 2022.

5.1.2 Spatial Characteristics

There are significant regional differences in TCE within the YRB, displaying an overall unbalanced pattern characterized by high values being concentrated and low values dispersed (see Figure 2). Sichuan, Qinghai, and Ningxia have remained in the high-value zone. Sichuan’s average TCE exceeded 1.0 and has stayed above 1.3 since 2018. This is mainly due to its low-carbon tourism model based on world heritage resources. Policies restricting high-energy-consuming projects and promoting eco-tourism have greatly reduced carbon emissions per unit of tourism output. The high TCE in Qinghai and Ningxia reflects their unique resource-based advantages. These provinces have effectively advanced low-carbon tourism pathways. For example, the implementation of ecological protection red line policies and the promotion of new energy transportation has effectively reduced carbon emissions per unit of tourism. In contrast, provinces like Shanxi and Inner Mongolia have maintained low TCE levels over the long term. Gansu also had relatively low TCE between 2013 and 2018. However, it has shown a rapid improvement trend since 2020, reaching 1.398 in 2022. This rebound may stem from Gansu’s “comprehensive tourism demonstration province” program and eco-attraction upgrades. However, large fluctuations suggest its green transition foundation remains unstable.

Figure 2:

Heat maps of TCE across nine provinces in the YRB from 2013 to 2022.

As shown in Figure 3, TER exhibits a distinct spatial hierarchy, with Shandong and Sichuan demonstrating relatively superior resilience and emerging as leaders in resilience enhancement within the YRB. From 2013 to 2022, Shandong’s average TER reached 0.55, ranking first among all provinces and showing a stable growth trend. This indicates that its tourism industry possesses a strong ability to resist risks and effective recovery mechanisms. For example, Shandong has vigorously promoted the “cultural tourism integration” strategy in recent years, establishing a high-quality tourism product supply system that provides a solid foundation for responding to market shocks. Sichuan ranked second with an average TER of 0.46. Leveraging its resource endowments and broad tourism market, Sichuan maintained relative resilience despite the pandemic’s impact; however, its recovery pace lagged behind Shandong after 2020, indicating room for improvement in its mechanisms for responding to external shocks. In contrast, Gansu, Qinghai, and Ningxia have consistently maintained low TER values, with averages below 0.2. These regions have weak tourism economic foundations, high external dependency, and highly seasonal tourism markets, making them more susceptible to prolonged recovery difficulties when facing systemic risks such as the pandemic.

Figure 3:

Heat maps of TER across nine provinces in the YRB from 2013 to 2022.

5.2 Analysis of Coupling Coordination Results

5.2.1 Temporal Characteristics

This study quantifies the changes in the CCD between TCE and TER across nine provinces and regions in the YRB, as illustrated by the temporal trends shown in Figure 4. Overall, the CCD between TCE and TER in the YRB has exhibited an upward trend, characterized by specific evolutionary patterns.

Figure 4:

Trends of the CCD in the YRB from 2013 to 2022.

From 2013 to 2016, the average CCD in the YRB was 0.573, indicating a state of reluctant coordination. During this period, significant fluctuations occurred for two main reasons. On the one hand, the tourism industry in the YRB was large in scale but structurally rigid. The slow transition from heavy industry to services, combined with stricter environmental policies, hindered structural optimization and coordinated development. On the other hand, tourism development in the YRB was uneven. In some provinces, the tourism sector lagged behind, and environmental protection measures were insufficient. The lack of coordination between ecological protection and tourism development reduced the overall CCD.

From 2017 to 2020, the average CCD in the YRB increased to 0.604, shifting from reluctant coordination to basic coordination. This indicates an improved relationship between tourism and ecological protection within the basin. This transformation can largely be attributed to the ongoing enhancement of national policies and regulations. In 2017, the National Tourism Administration issued the Guidelines for the Creation of All-for-One Tourism Demonstration Zones and related policies. These measures created a favorable environment for tourism development in the YRB. These policies improved market mechanisms, encouraged innovation, and boosted the vitality of the tourism economy. The ripple effects of these policies facilitated the coordinated advancement of TCE and TER in the tourism industry.

From 2021 to 2022, the average CCD in the YRB significantly increased to 0.623, with most provinces in the basin entering the basic coordination stage. However, the overall CCD experienced a sudden decline to 0.579 in 2021. This drop can be attributed to the ongoing global pandemic that year, which severely impacted the tourism industry in the YRB, leading to a sharp decline in tourism revenue and a slowdown in the improvement of TCE. Additionally, several tourism projects were suspended or restricted due to pandemic prevention measures, hindering the normal development of the tourism industry. After the pandemic eased, the 2022 Yellow River Protection Law emphasized pollution control, ecological restoration, and green tourism promotion. Consequently, the tourism industry in the YRB gradually recovered, and provinces within the basin established a more closely coordinated relationship between tourism development and ecological protection.

5.2.2 Spatial Characteristics

To further investigate the spatial changes in the CCD, this study selected the years 2013, 2016, 2019, and 2022, considering the limited scope and research focus of this study. This study used ArcGIS 10.8 software to create a distribution map illustrating the CCD across nine provinces and regions in the YRB. The natural breaks classification method was employed to enhance the visual clarity of spatial variations, as shown in Figure 5.

Figure 5:

Spatial pattern of the CCD between TCE and TER in the YRB. Note: Based on the standard map produced in accordance with the standards of the Ministry of Natural Resources’ Standard Map Service Website, with the review number GS(2023)2767, the boundaries of the base map have not been modified.

From the perspective of spatial evolution trends, between 2013 and 2019, the CCD between TCE and TER in the nine provinces and regions along the YRB exhibited an overall spatial pattern of decreasing values from east to west. Eastern provinces such as Shandong and Henan demonstrated higher coordination levels. Shandong has been in a high-level coupling phase since 2013, while Henan has maintained a basic coordination within the adjustment phase. Central provinces such as Shanxi and Shaanxi primarily remained in the adjustment phase, with relatively stable coordination levels. Western provinces like Qinghai and Gansu were stuck in the antagonistic phase for an extended period during the early stages, only entering the basic coordination range in 2019. A deeper exploration of the reasons behind this gradient distribution reveals three main factors. First, Eastern provinces have rich tourism resources, convenient transport, and strong reception capacity. Their mature market mechanisms give them greater resilience and adaptability to ecological constraints. Second, these regions have demonstrated more proactive institutional responses in ecological civilization construction, having established policy tools such as ecological red lines and carbon emission constraints at an early stage, thereby providing institutional safeguards for the coordination between tourism and low-carbon transformation. Third, fiscal decentralization differences cause variation in local policy execution. This affects how governments coordinate tourism and environmental governance. Eastern regions, with ample fiscal resources and high policy autonomy, can advance green tourism projects tailored to local conditions. In contrast, western regions, constrained by fiscal pressures, often rely on transfers from higher-level governments, which weakens their initiative and resource allocation capacity in policy implementation. Therefore, spatial CCD differences reflect both economic development gaps and differences in governance and institutional capacity for green tourism transformation.

5.3 Identification and Analysis of Key Influencing Factors

To further identify the key factors influencing the CCD between TCE and TER, this study employs the random forest model. The random forest model developed in this study demonstrates high accuracy, with an R ² of 0.8214, an RMSE of 0.0501, and an MAE of 0.0376. These metrics indicate that the model fits the data well and produces minimal prediction errors. To further interpret the model results, the nine influencing factors are ranked by importance. Figure 6 shows that RDI and INF rank highest, followed by TRE and TI. The influence of ED, GI, ER, TDS, and IS decreases in that order. This analysis suggests that the coupled and coordinated development of carbon emission efficiency and economic resilience in the YRB tourism industry primarily depends on RDI and INF. Greater R&D investment enhances technological capacity and supports the green transition of tourism. It also strengthens economic resilience. Improving infrastructure helps reduce resource consumption and carbon emissions, thereby enhancing the stability and accessibility of the tourism economy. The importance of TRE exceeds that of TDS, indicating that superior resource endowment is more effective than mere scale expansion in promoting low-carbon tourism development. Among innovation indicators, TI ranks fourth, highlighting that outcomes such as patent applications remain a key driver for enhancing CCD. In contrast, the roles of ED, GI, and ER are relatively minor, suggesting that policy-driven efforts have not yet been fully leveraged. IS and TDS rank lower, indicating that in the YRB, the marginal effects of structural optimization and scale expansion have diminished. Overall, RDI and INF are the core drivers for coordinating improvements in TCE and TER in the YRB, while policy and structural adjustments require further strengthening to achieve greater effectiveness.

Figure 6:

Ranking of influencing factors by importance.

The partial dependence diagram illustrates the relationship between influencing factors and CCD, further revealing the direction and strength of their effects (see Figure 7).

Figure 7:

Partial dependence plots of the influencing factors.

In terms of economic level, the effect of ED on CCD first rises briefly within 100–102.5. It then declines to a trough near 105 and rebounds sharply around 110. This indicates that the positive effect of ED on CCD in the YRB exhibits a threshold effect similar to a “middle-income trap”. In the early stage of economic development, tourism expansion in the region relied heavily on energy-intensive industries and extensive growth models, which led to increased carbon emissions and insufficient industrial resilience. Once the middle-income threshold was crossed, the region accumulated sufficient financial capacity for green investment and technological innovation, enabling tourism to benefit simultaneously from low-carbon transformation and risk resistance, thereby significantly enhancing the CCD. The effect of IS on CCD rises slowly until around 55 and then increases rapidly. This suggests the dominance of the service sector is the key turning point for releasing structural dividends. Modern services not only reduce emission intensity but also enhance industrial diversity, which strengthens the adaptability and resilience of the tourism economy, thus driving a leap in CCD. The effect of INF on CCD follows an inverted U-shaped curve, peaking around 50 before gradually declining. This pattern reflects the long-standing imbalance in infrastructure investment in the YRB, with a dense concentration in the east and relative scarcity in the west. In the upstream areas, which remain in a stage of “filling gaps”, traditional infrastructure investment can still significantly improve tourism accessibility and emergency capacity. In the downstream areas, however, infrastructure has reached near saturation, and excessive investment may lead to diminishing marginal returns, which in turn hinders coordinated development.

In terms of tourism industry development, the effect of TDS on CCD shows a rapid decline followed by a stabilization trend. In some areas of the YRB, the early stage of tourism development was characterized by resource-oriented expansion, where tourist growth outpaced governance capacity, resulting in greater ecological pressure, higher carbon emissions, and a decline in CCD. As the industry continued to expand, a reverse pressure mechanism promoted improvements in management efficiency and the adoption of green technologies, which gradually stabilized CCD. This highlights the necessity of shifting from a speed-oriented to a quality-oriented development model. The effect of TRE on CCD first decreases and then increases, indicating that resource advantages do not automatically translate into coordination advantages. Tourism resources in the YRB are unevenly distributed, and in some areas, the early stages of resource exploitation were accompanied by lagging management and insufficient supporting facilities, which created ecological pressures and reduced CCD. As resource development moved toward greater standardization and ecological orientation, CCD began to recover. This suggests that resource endowments must be matched with governance capacity in order to achieve a sustained positive response of CCD.

In terms of innovation capacity, RDI emerges as the most important influencing factor. Its partial dependence curve shows a strong stimulating effect on CCD at the low stage, peaking near 2.5 % before marginal effects begin to decline. Given that overall R&D investment in the YRB is relatively low and the foundation for green technologies remains weak, small but targeted inputs at the early stage can trigger significant green transformation effects, to which CCD responds sensitively. However, once investment exceeds a certain threshold, the marginal effect diminishes, which may be related to low efficiency of technology transfer and poor adaptability of research outcomes. Therefore, the region should optimize its R&D structure, strengthen the alignment of green technologies with tourism applications, and enhance the sustained driving force of RDI on CCD. The effect of TI on CCD declines sharply at the starting point and remains at a low level for a prolonged period, only beginning to recover once the number of patent applications surpasses 50,000. Since the total number of green patents in the YRB is relatively limited and most are concentrated outside the tourism sector, the initial contribution of TI to CCD is weak. Only when green technologies accumulate to a sufficient scale does CCD begin to rebound. This implies that green technology innovation in tourism requires long-term accumulation and systematic planning, and rapid gains cannot be expected in the short run.

In terms of government policy, the effect of ER on CCD follows a U-shaped pattern, confirming the applicability of the “Porter Hypothesis” in the YRB. When regulation is too weak, enterprises lack the incentive for green transformation, leading to a decline in CCD. Moderate regulation can stimulate green investment and innovation, thereby enhancing CCD. However, overly stringent regulation may increase compliance costs and suppress market vitality. Therefore, in the YRB, particularly in upstream areas, regulatory intensity should be appropriately reduced while incentive measures are strengthened. In contrast, environmental standards in the middle and downstream regions can be gradually raised to promote green transformation. The effect of GI on CCD shows a monotonically decreasing trend, indicating that higher fiscal expenditure does not necessarily generate better outcomes. In some parts of the YRB, the proportion of fiscal expenditure on environmental protection is high but is primarily directed toward end-of-pipe treatment, with insufficient investment in the early stages of green technologies and resilience capacity. This has led to a decline in CCD. Accordingly, it is necessary to optimize the structure of fiscal expenditure by leveraging green finance, innovation subsidies, and tax incentives to guide the participation of private capital, thereby enhancing the positive effect of GI on CCD.

6.1 Conclusions

This study uses panel data from nine provinces in the Yellow River Basin spanning 2013 to 2022. It employs the Super-SBM model, entropy weight method, coupling coordination degree model, and random forest model to systematically evaluate the spatiotemporal characteristics and key driving factors of the CCD between TCE and TER. The following conclusions are drawn:

1. Both TCE and TER exhibit fluctuating upward trends. TCE displays a phased pattern characterized by an initial decline, mid-term stability, and a rapid increase in the later stage. In contrast, TER follows a trajectory of steady growth, a short-term decline, and gradual recovery.

2. Significant regional disparities in TCE exist across the YRB. It shows an overall unbalanced spatial distribution: high-efficiency zones are clustered while low-efficiency zones are widespread. TCE is relatively high in Sichuan, Qinghai, and Ningxia, whereas it is relatively low in Shanxi and Inner Mongolia. TER also exhibits distinct spatial hierarchical patterns, with Shandong and Henan performing well, while Gansu, Qinghai, and Ningxia consistently remain in low-value zones.

3. The CCD between TCE and TER in the YRB exhibits a fluctuating upward trend, evolving from reluctant coordination to basic coordination. However, significant regional development disparities persist, with the spatial pattern exhibiting a decreasing trend from east to west.

4. The effects of various influencing factors on CCD exhibit significant nonlinear characteristics. Among these, RDI, INF, and TRE are key factors affecting CCD. GI has an inhibitory effect on CCD, whereas IS plays a promoting role. Additionally, ED, INF, TRE, TI, and ER exert positive effects on CCD within specific ranges.

6.2 Policy Implications

Based on these findings, we propose the following policy recommendations:

First, implement differentiated policies tailored to regional conditions. Given the wide variations in TCE and TER across YRB provinces, policy interventions must align with local realities. For example, western provinces such as Gansu and Qinghai have relatively low TER. These regions face severe ecological degradation and inadequate infrastructure development. These regions should prioritize ecological protection and infrastructure investment. They should also promote low-carbon tourism projects and encourage the wider use of green technologies. Governments can speed up the green transition with fiscal subsidies and tax incentives. For example, they can support enterprises adopting low-carbon tourism technologies. Meanwhile, provinces such as Shandong and Sichuan should focus on innovation-driven development and technological progress. Strengthening R&D and the application of low-carbon technologies will help promote the coordinated development of TCE and TER.

Second, strengthen policy guidance on innovation and infrastructure development. Research indicates that R&D intensity and infrastructure investment significantly influence the CCD between TCE and TER. Therefore, the government should provide stronger support for science, technology, and green infrastructure. In the YRB, particularly the central and western regions, policies should direct investment to R&D and green technology. Governments should also promote green tourism product innovation. Additionally, raising R&D investment beyond a critical threshold (e.g. over 2.5 % of GDP) can stimulate energy-saving technology development. Additionally, infrastructure should undergo green transformation. Transport networks and environmental facilities need optimization to support sustainability. Optimizing infrastructure density prevents diminishing returns and improves both TCE and TER. This strategy boosts TCE and strengthens tourism’s economic resilience. It supports green and high-quality regional development.

Third, establish a dynamic balance mechanism between resource endowment and scale control. In YRB provinces such as Shanxi and Inner Mongolia, where tourism resource endowments are abundant but the CCD remains low, it is essential to shift from an extensive development model that prioritizes quantity over quality. Specifically, a “red line” system can regulate tourism development. Annual visitor limits based on ecological capacity can help prevent overexploitation. For example, reservation systems and visitor quotas in ecologically sensitive areas, such as Hukou Waterfall in Shanxi, can help ensure ecological sustainability. A resource value accounting system should also be established. It should internalize ecological costs in ticket prices and add fees for high-carbon transport such as fuel-powered sightseeing vehicles. These measures will promote low-carbon travel and ensure that tourism grows within ecological limits.

Finally, strengthen the implementation of low-carbon tourism policies. Because TCE and TER are closely linked, central and local governments should strengthen low-carbon tourism policies and promote comprehensive demonstration zones. In less developed regions such as Gansu and Ningxia, governments should work with market actors to promote ecological tourism projects. These initiatives will protect the ecological environment. They will also stimulate the green growth of the local economy. Therefore, policies should offer tax and financial incentives for low-carbon tourism to motivate businesses and local governments to invest in green technologies. At the same time, it is essential to strengthen the synergy between these incentives and environmental protection investments. Local governments should raise industrial pollution control investments above 0.5 % of industrial added value. This would activate the Porter effect and motivate enterprises to pursue green technological upgrades. For example, the Tengger Desert Wetland Tourism Area in Zhongwei has launched a pilot project that combines photovoltaic power with low-carbon sightseeing vehicles. The clean energy transport system aims to reduce carbon emissions per tourism revenue unit by 15 % within three years. It provides a replicable model for arid regions in Northwest China.

6.3 Limitations and Future Research

This study reveals the coupling coordination characteristics and driving mechanisms of TCE and TER in the YRB. However, several limitations remain. First, although the entropy weight method enhances objectivity, it may still introduce measurement errors, potentially affecting the accuracy and stability of the results. Second, while this study characterizes the spatiotemporal evolution of the CCD, the provincial-level sample size limits its ability to uncover more refined spatial correlation mechanisms. City-level data offer larger samples and finer spatial units. They allow more accurate analysis of local patterns, spatial correlations, and heterogeneity. Third, the study primarily focuses on the spatiotemporal evolution and identification of key influencing factors at the macro level, without fully incorporating microeconomic variables such as tourist preferences and price elasticity. This limitation may underestimate how microeconomic behavior affects changes in TCE and TER. To address these shortcomings, future research should focus on the following areas. First, combining subjective and objective evaluation methods could overcome the limitations of relying solely on the entropy weighting method and enhance the scientific rigor of data processing. Second, incorporating city-level data and utilizing ArcGIS for multi-scale spatial analysis would help reveal the spatial linkage mechanisms driving the CCD between TCE and TER across different cities. Finally, future research should incorporate more micro-level factors such as tourist behavior and market demand into a multidimensional and cross-scale framework.