Differences in Tourism Ecological Resilience and Its Asymmetric Driving Mechanisms in the Chengdu-Chongqing Economic Circle, China

Abstract

In response to frequent disruptions such as public health incidents and natural disasters, enhancing tourism ecological resilience (TER) has become crucial for achieving sustainable tourism development. This study constructs an evaluation index system for TER from three dimensions: resistance, recovery, and innovation. Employing the entropy weight method and fuzzy-set Qualitative Comparative Analysis (fsQCA), an empirical analysis was conducted on the spatiotemporal evolution and formation mechanisms of TER in the Chengdu-Chongqing Economic Circle (CCEC) from 2013 to 2023. The results indicate that: First, although an overall upward trend in TER was observed, significant regional disparities existed. Chongqing (0.634) and Chengdu (0.491) consistently led, while the average values for the other cities were generally below 0.155, revealing a pattern characterized by “dual-core prominence and peripheral lag”. Second, the impact of the pandemic exacerbated imbalances among subsystems, with resistance, recovery, and innovation capabilities all exhibiting core–periphery differentiation. Third, fsQCA results demonstrated that high resilience was driven by a “technology-service” core coupled with auxiliary conditions such as transportation or consumption, while low resilience stemmed from multiple systemic deficiencies, including insufficient government support, underdeveloped transportation, and weak technological innovation. This study provides configurational pathways and policy implications for building regional tourism resilience.

1. Introduction

With the continuous advancement of global transportation networks, information networks, and technology, the tourism industry has developed rapidly and is widely regarded as an important sector for promoting economic prosperity and social well-being [1]. However, if only the economic attributes of tourism development are emphasized, the unregulated expansion of mass tourism is likely to lead to the over-exploitation of tourism resources within a short period, further highlighting the vulnerability of the industry. This may eventually result in irreversible decline due to the depletion of destination resources and attractions [2]. Given the complexity and comprehensive nature of tourism, internal coordination and balance among its various components are often difficult to achieve, while external factors such as environmental disruptions also significantly affect the industry. Owing to varying sensitivities to different types of crises, tourism systems exhibit differentiated levels of impact. Over the past two decades, a wide array of major incidents—such as public health crises (e.g., various infectious diseases), economic downturns, natural disasters, and social unrest—have profoundly disrupted the trajectory of global tourism development. For instance, the terrorist attacks in the United States in September 2001 had long-term negative effects on its tourism industry [3]. In February 2014, the Ebola virus first broke out in West Africa and spread to multiple regions such as Liberia, Nigeria, Guinea, and Sierra Leone, causing severe losses in tourism revenue, a sharp decline in tourist visits, and significant challenges for recovery [4]. Furthermore, the impact of the COVID-19 pandemic on the global tourism sector has fully demonstrated the industry’s vulnerability and sensitivity—in 2021, both global tourist arrivals and total tourism revenue were only 53.7% and 55.9% of the 2019 levels, respectively (World Tourism Organization, 2021) [5]. Therefore, the impact of such crises on tourism development is profoundly significant, making the construction of a resilient tourism ecosystem a key focus of academic research.

Research on tourism ecosystems primarily revolves around such logical threads as conceptual definition, systemic structural characteristics, measurement from specific perspectives, and optimization pathways. It is proposed in academia that the tourism ecosystem is a composite system jointly constituted by economic, tourism, social, and ecological dimensions. A growing number of studies have attempted to evaluate regional tourism ecology from specific perspectives, with relevant research focusing on tourism eco-efficiency [6,7], tourism environmental carrying capacity [8,9], and tourism ecological security [10,11]. Furthermore, research scales are diverse, predominantly macro and meso, yet increasingly trending toward micro-level analysis. Major strategic regions have become a key focus of recent studies, resulting in a coexisting multiplicity of scales including provincial [12,13], key regional [11], municipal [14], and county levels [15]. Research content continues to deepen, expanding from single-dimensional measurement of tourism ecology to associated studies with environmental regulation [16], tourism economy [17,18], and ecological security [19]. Regarding research methods, interdisciplinary integration is prominent. Methods for measuring the comprehensive level of tourism ecology mainly include the DEA model [20], ecological footprint approach [21], and spatial correlation network analysis [22], while methods such as fuzzy-set Qualitative Comparative Analysis (fsQCA) [23], Tobit model [24], and geographical detector [10] are widely applied to analyze influencing factors and pathways of tourism ecology. As research progresses, the integration of multiple disciplines such as tourism studies, ecology, geography, and management science has enriched research content and perspectives. It is important to note that with the advent of the “era of mass tourism,” negative ecological and environmental issues related to tourism continue to emerge. Coupled with external shocks such as rapid urbanization and the COVID-19 pandemic, the tourism ecosystem urgently needs to enhance its resilience level to strengthen its defensive and adaptive capacities. However, research on TER remains limited. Previous studies have mostly approached the vulnerability of the tourism industry from the perspective of tourism economic resilience [25,26,27], focusing on the adaptive development process of resisting shocks and disturbances through model renewal during tourism economic operations. These studies often adopt a “unidimensional” perspective, such as input–output, carrying capacity, or ecological footprint, lacking in-depth discussion on TER. This approach fails to reflect the growth and transformation trajectory of complex tourism ecosystems when facing impacts and challenges and does not align with the comprehensive characterization of tourism ecosystems that integrate economic benefits, environmental benefits, and social carrying benefits. This limitation hinders a scientific understanding of the connotations of tourism ecosystems. Therefore, evaluating tourism ecosystems from a “resilience” perspective and deeply analyzing the spatial correlation network structural characteristics and pathways of TER can not only intuitively reflect the comprehensive capacity of tourism ecosystems to resist, recover, and innovate, as well as the differences in TER, but also help reveal the correlation effects of tourism ecosystems under different dimensions such as geographical proximity and industrial linkage. This has profound practical and theoretical significance for promoting the sustainable development of the tourism industry.

The tourism system is recognized as a complex giant system that is subject to disturbances from numerous factors, thereby influencing the level of TER. Consequently, under internal and external environmental shocks, the tourism resilience of different regions reflects complex dynamic interactions among specific key elements, demonstrating a need for “nonlinear management”. However, current research on influencing factors of resilience predominantly explores linear relationships among these factors, focusing on simple symmetric linear correlations between individual antecedent conditions and outcomes [28]. These studies emphasize the independent effects of different factors while lacking analysis of the configurational effects of multiple influencing factors on TER, as well as investigations into the impact mechanisms within complex systems, which presents certain limitations. Qualitative Comparative Analysis (QCA), based on a configurational perspective, serves as a holistic paradigm for analyzing the effects of multi-element configurations [29]. Although it has been applied in economic and management research in China for a relatively short period, its use has been concentrated in fields such as enterprise management, leaving a gap in studies examining the configurational effects of influencing factors on TER. In terms of selecting antecedent conditions, existing research on TER determinants has primarily focused on the supply side, including tourism destinations, government, and communities, while neglecting demand-side aspects such as tourist consumption. This oversight makes it difficult to effectively reflect the demand-oriented characteristics of the tourism system. Therefore, to scientifically explain the causal logic underlying the complex tourism system, this study investigates the intricate set relationships between influencing factors and outcomes of TER from a configurational perspective. Furthermore, it proposes selecting antecedent conditions from both the “demand side + supply side” dimensions to examine the causal mechanisms between these factors and TER.

Resilience, a concept originating from physics and materials science, was introduced into ecology by Canadian ecologist Holling in the 1980s [30]. Since then, resilience theory has undergone three major conceptual shifts: from engineering resilience to ecological resilience, and finally to evolutionary resilience. Engineering resilience describes the ability of a system to return to its original equilibrium after disturbance. Ecological resilience, first proposed by Holling, emphasizes the capacity of an ecosystem to absorb changes, persist, and reorganize while maintaining essential functions, thereby achieving a new steady state. Emerging in the 1990s, evolutionary resilience conceptualizes resilience as an adaptive, learning, and transformative capacity of complex socio-ecological systems to respond to pressures and constraints through change, adaptation, and transformation. Throughout these shifts, academic perspectives have moved away from equilibrium-centered views toward a dynamic understanding of spiral-like development. Resilience theory focuses on the ability of regional systems to flexibly cope with various environmental shocks, highlighting continuous dynamic processes and contributing to a deeper understanding of tourism systems’ resilient functioning [31].

Given the leading role and distinctive status of the CCEC in the international development of China’s tourism industry—where tourism contributes nearly 20% of the national income—16 cities within the CCEC were selected as cases for in-depth analysis in this study [32]. By comprehensively applying an evaluation index system for TER and fsQCA, the temporal evolution and spatial differentiation of TER levels, as well as the complex set relationships between influencing factors and outcomes of TER, were investigated. This study aims to address the following core questions:

Q1. How does the temporal evolution and spatial differentiation of TER manifest in the CCEC?

Q2. What are the differences in the subsystems of TER within the CCEC before and after the pandemic?

Q3. What is the formation mechanism of TER differences from an asymmetric perspective?

2. Study Area and Data

2.1. Study Area

Located in southwestern China (28°10′–32°03′ N, 103°49′–110°11′ E), the CCEC (see Figure 1) is a vital economic growth pole in the upper Yangtze River region. With a total area of 185,000 km², it encompasses Chongqing Municipality and 15 cities in Sichuan Province: Chengdu, Zigong, Luzhou, Deyang, Mianyang (excluding Pingwu and Beichuan counties), Suining, Neijiang, Leshan, Nanchong, Meishan, Yibin, Guangan, Dazhou (excluding Wanyuan City), Ya’an (excluding Tianquan and Baoxing counties), and Ziyang—16 administrative units in total. The region is characterized by a subtropical monsoon climate, with an average annual precipitation of 1000–1200 mm and an average annual temperature of 16–18 °C. Its diverse topography—including the Sichuan Basin, parallel ridge-valley systems, hilly and mountainous terrain, and the river network formed by the Yangtze River and its tributaries—provides unique resource advantages for developing comprehensive tourism. In 2023, the CCEC received over 500 million tourist visits, generating total tourism revenue exceeding RMB 500 billion, making tourism a pillar industry and a key driver of regional development. As of 2023, the area boasts six UNESCO World Heritage sites (e.g., Jiuzhaigou Valley and Leshan Giant Buddha), multiple 5A-level scenic spots (e.g., Mount Emei and Qingchengshan–Dujiangyan), and dozens of 4A-level attractions, forming the largest and most diverse cluster of world-class natural and cultural heritage sites and urban tourism destinations in western China. Given these distinctive features, the CCEC was selected as the study area to investigate the evolutionary characteristics of TER and its impact on the complex ecological environment within this extensive region, representing high strategic and representative significance. Based on TER theory and data including tourist numbers, total tourism revenue, and ecotourism development from 2013 to 2023, the analysis of TER in the CCEC was divided into two phases: 2013–2019 (pre-COVID-19, a period of rapid tourism growth and enhanced ecological conservation policies) and 2020–2023 (post-COVID-19, a phase of responding to major external shocks and testing resilience). Furthermore, set-theoretic methods and Boolean algebra were employed to analyze the configurational effects of factors influencing TER.

2.2. Data Sources and Processing

The study period from 2013 to 2023 was selected for quantitative analysis across 16 regions within the CCEC. Data were primarily collected from the Sichuan Statistical Yearbook, Chongqing Statistical Yearbook, Sichuan Culture and Tourism Yearbook, and Chongqing Tourism Statistical Bulletins https://www.cnki.net/ (accessed on 4 June 2025). To ensure data completeness, interpolation methods were employed to supplement missing values for individual years. It should be noted that certain data for the year 2021, which coincided with the COVID-19 outbreak, were estimated based on available yearbooks, bulletins, and local and national datasets. Furthermore, although the planned boundary of the CCEC includes only parts of Kaizhou, Yunyang, Mianyang, Dazhou, and Ya’an, the entire administrative area of each city was included in the study for consistency and ease of processing, using city-wide data as an approximation representation.

3. Methodology

3.1. Indicator System Construction

The tourism ecosystem is regarded as the primary research object in tourism ecology, and it is widely acknowledged in academia that this system integrates tourism, ecological, social, and economic subsystems, exhibiting structural complexity and evolutionary dynamics as well as developmental vulnerability. Therefore, the construction of a Tourism Ecological Resilience (TER) indicator system must adhere to the principle of ecological priority, highlight the capacity for infrastructure development in tourism, reflect the ability to ensure both high-quality growth of the tourism industry and the equitable sharing of ecological conservation outcomes, and fully demonstrate the innovation capacity for transformational renewal of the tourism ecosystem. Drawing on the conceptual foundations of resilience theory and taking into full consideration the complex evolutionary nature and shock-sensitive characteristics of the tourism ecosystem, TER is defined in this study as: the ability of the tourism ecosystem to defend against and resist internal and external disturbances—such as disordered tourism development, ecological environmental damage, and exceptional external events—its capacity to recover from damage and maintain normal functional operation, and its regenerative capability to transform into an improved state through innovation, optimization, and transitional upgrading. With due consideration to data availability and in reference to previous research [32,33,34], the TER indicator system is constructed in accordance with the comprehensive, interrelated, and systemic nature of the tourism ecosystem, comprising three dimensions: resistance, recovery, and innovation (see

Table 1. TER Indicator System for the CCEC.

Criterion Layer	Indicator Layer	Weight
Resistance	GDP per capita (CNY)	0.0240 (+)
	Tourism revenue as % of GDP	0.0259 (+)
	Urbanization rate (%)	0.0217 (+)
	Number of star-rated hotels	0.1408 (+)
	Domestic tourist arrivals (10,000 person-times)	0.1229 (+)
	Tourism traffic pressure (%)	0.0041 (−)
Recovery	Per capita tourism expenditure (CNY)	0.0071 (+)
	Road area per capita (m2)	0.0149 (+)
	Green space per capita (m2)	0.0180 (+)
	Green coverage rate in built-up areas (%)	0.0136 (+)
	Number of medical beds	0.1256 (+)
Innovation	Fiscal expenditure as % of GDP	0.0334 (+)
	Environmental investment as % of GDP	0.0407 (+)
	Internal R&D expenditure (10,000 CNY)	0.2091 (+)
	Number of college students enrolled	0.1824 (+)
	Sewage treatment rate (%)	0.0250 (+)

).

3.2. Research Method

Comprehensive Level Index Measurement. The calculation procedure was conducted as follows: (1) All indicator data were standardized; (2) Dimensionless processing was applied to the indicator data; (3) The entropy weight method was used to determine the weights of each indicator; (4) A weighted summation approach was employed to calculate the comprehensive scores of TER development levels for each province in different years [35].

Fuzzy-Set Qualitative Comparative Analysis (fsQCA). To investigate the asymmetric causal mechanisms influencing TER in the Chengdu-Chongqing Economic Circle (CCEC), the fsQCA method was adopted. This approach integrates fuzzy logic with qualitative comparative analysis (QCA), combining quantitative empirical testing with qualitative inductive reasoning [29,36]. Grounded in set theory and Boolean algebra, QCA enables the construction of complex causal relationships with small sample sizes and facilitates intuitive interpretation through set-theoretic principles [29,37]. The method offers three key advantages: (1) minimal sample size requirements, allowing the revelation of complex causal patterns with limited cases; (2) equifinality, meaning multiple configurations of conditions can lead to the same outcome; and (3) causal asymmetry, where the occurrence and non-occurrence of an outcome require distinct explanatory paths [38]. Using fuzzy set theory and Boolean algebra, fsQCA identifies how antecedent conditions combine and what outcomes emerge with or without these conditions.

Calibration and computation of the fuzzy truth table represent the core steps of fsQCA. The calibration process involves transforming the values of antecedent and outcome variables into fuzzy sets by determining their membership scores. Three qualitative anchors were set for all variables to describe their degree of membership in a set: 1 indicates full membership, 0.5 represents the crossover point, and 0 denotes non-membership [29,39]. Mathematically, precise membership scores of 1 or 0 are unattainable; thus, full membership and non-membership thresholds were defined as 0.95 and 0.05, respectively. The maximum value of each variable was set as the full membership threshold, the minimum as the non-membership threshold, and the median as the crossover point. The truth table lists configurations of antecedent conditions, encompassing all possible combinations predicting the outcome. Configurations were refined based on frequency and consistency criteria [40]. For predicting TER levels, a frequency threshold of 1 and a consistency threshold of 0.8 were applied as widely accepted standards [34]. The consistency and coverage metrics provided by fsQCA accurately reflect the explanatory power of causal configurations for specific outcomes, analogous to correlation and determination coefficients in symmetric methods.

Overall Data Analysis Procedure. The data analysis in this study was conducted following four steps: First, the comprehensive level index of TER was calculated using the entropy weight method. Second, the fsQCA method was employed to investigate the asymmetric causal mechanisms influencing TER level. Third, predictive validity analysis was performed using additional samples to verify the model’s predictive effectiveness. Statistical data from 2022 were used to compute the average values of TER levels and antecedent variables for each city in the CCEC, with in-depth analysis conducted using fsQCA 3.0 software.

4. Results

4.1. Temporal Evolutionary Characteristics of TER

Based on the constructed TER indicator system, the weights of each indicator were determined using Python software 3.7 (see

Table 2. TER Levels in the CCEC, 2013–2023.

City	2013	2014	2015	2016	2017	2018	2019	2020	2021	2022	2023	MV
Chongqing	0.479	0.499	0.507	0.576	0.665	0.684	0.697	0.676	0.740	0.684	0.796	0.637
Chengdu	0.367	0.393	0.396	0.440	0.410	0.452	0.550	0.550	0.507	0.568	0.659	0.481
Zigong	0.067	0.069	0.067	0.064	0.075	0.086	0.071	0.069	0.053	0.067	0.089	0.071
Luzhou	0.072	0.079	0.084	0.081	0.097	0.096	0.092	0.097	0.086	0.123	0.113	0.093
Deyang	0.077	0.076	0.075	0.081	0.095	0.105	0.100	0.103	0.085	0.101	0.112	0.092
Mianyang	0.142	0.126	0.169	0.130	0.140	0.146	0.163	0.171	0.135	0.182	0.191	0.154
Suining	0.068	0.081	0.079	0.083	0.105	0.088	0.084	0.091	0.066	0.079	0.070	0.081
Neijiang	0.056	0.061	0.058	0.055	0.060	0.069	0.067	0.082	0.055	0.079	0.095	0.067
Leshan	0.085	0.094	0.094	0.091	0.097	0.102	0.106	0.094	0.066	0.081	0.118	0.093
Nanchong	0.070	0.092	0.092	0.087	0.103	0.103	0.097	0.121	0.086	0.108	0.121	0.098
Meishan	0.062	0.072	0.073	0.065	0.079	0.075	0.067	0.078	0.055	0.065	0.071	0.069
Yibin	0.069	0.067	0.076	0.076	0.092	0.094	0.092	0.104	0.075	0.114	0.105	0.088
Guang’an	0.047	0.075	0.075	0.070	0.101	0.095	0.067	0.063	0.043	0.067	0.059	0.069
Dazhou	0.039	0.049	0.053	0.050	0.048	0.058	0.060	0.071	0.052	0.073	0.074	0.057
Ya’an	0.107	0.127	0.137	0.078	0.100	0.098	0.089	0.110	0.093	0.125	0.100	0.106
Ziyang	0.078	0.091	0.097	0.054	0.063	0.066	0.051	0.063	0.043	0.081	0.052	0.067

Note: MV stands for Mean Value.

). The degree of closeness between each evaluated object and the optimal solution was calculated, with this proximity defined as the TER level for each region. From a temporal dynamic perspective, the comprehensive development levels of cities in the CCEC exhibited fluctuating trends from 2013 to 2023, with significant disparities observed overall. Except for Chongqing and Chengdu, the comprehensive scores for the remaining cities were generally low. Specifically, the average comprehensive scores of Chongqing and Chengdu were significantly higher than those of other regions, at 0.634 and 0.491, respectively, indicating stronger comprehensive development capabilities. Among the remaining cities, Mianyang, Deyang, Luzhou, and Yibin had relatively higher average scores, though these remained within the range of 0.090–0.155, suggesting substantial room for improvement in overall development levels. Chongqing’s comprehensive score demonstrated a steady upward trend, rising from 0.479 in 2013 to 0.796 in 2023, reflecting strong sustainable development capabilities and resilience. This performance was attributed to Chongqing’s advantageous geographical location, substantial policy support, continuous industrial structure optimization, and in-depth development of cultural tourism resources, collectively enhancing its comprehensive competitiveness. Chengdu’s comprehensive score also showed a fluctuating upward trend, increasing from 0.367 in 2013 to 0.659 in 2023—a growth of approximately 79.5%—indicating robust developmental resilience. Chengdu maintained relatively stable development amid changing macroeconomic conditions, leveraging its strong economic foundation, technological innovation capabilities, openness to the global economy, and cultural tourism appeal. Among other cities, Mianyang and Deyang had average comprehensive scores of 0.155 and 0.092, respectively. Although these ranked higher relative to other non-core cities, they still lagged significantly behind the dual cores of Chongqing and Chengdu. Luzhou and Yibin exhibited fluctuating upward trends, with scores recovering notably during 2020–2022, indicating potential for resilience. In contrast, regions such as Ziyang, Dazhou, and Guangan consistently had low comprehensive scores, with averages not exceeding 0.070, reflecting weaker developmental resilience and inadequate accumulation and enhancement mechanisms for comprehensive development capabilities.

Overall, significant internal disparities were observed within the CCEC, with the dual-core driving effect being prominent. Most peripheral cities remained at low developmental stages, necessitating enhanced regional coordination and policy support to promote overall coordinated development.

4.2. Spatial Differentiation Pattern of TER

4.2.1. Spatial Differentiation Pattern of TER Before the Pandemic

Based on the spatial distribution maps generated using ArcGIS 10.7 (see Figure 2), TER of the CCEC in 2013, 2016, and 2019 was characterized by distinct spatial heterogeneity. Overall, a stable “dual-core” structure was maintained in the TER pattern throughout the study period, with imbalanced regional development and significant inter-regional disparities. Chongqing and Chengdu consistently exhibited high TER levels, serving as regional tourism centers with prominent advantages in tourism scale and resource endowment. A clear polarization effect was observed between the two cores, with intermediate regions such as Ziyang, Meishan, and Neijiang continuously demonstrating low TER levels. Chongqing exhibited a strong siphon effect, attracting resources from surrounding areas and resulting in lower resilience levels in regions such as Dazhou and Guang’an. Specifically, although Mianyang was maintained at a relatively high level, a considerable gap remained compared to the dual cores. Between 2010 and 2019, Dazhou and Nanchong were elevated from low to relatively low TER levels, indicating enhanced resilience. Yibin, Zigong, and Suining fluctuated between medium and relatively low levels. Deyang and Leshan declined from relatively high to medium levels, Luzhou decreased from medium to relatively low, and Guang’an declined from relatively low to low, showing a downward trend in resilience. In contrast, Ziyang, Meishan, and Neijiang remained consistently at low levels with no significant changes.

4.2.2. Spatial Differentiation Pattern of TER After the Pandemic

Based on the spatial distribution maps generated using ArcGIS 10.7 (see Figure 3), the TER of the CCEC from 2020 to 2023 was characterized by significant spatial heterogeneity. Overall, a “dual-core” structure continued to be observed in the TER pattern during this period, accompanied by imbalanced regional development and pronounced inter-regional disparities. Chongqing and Chengdu were consistently classified at high TER levels, demonstrating strong systemic stability and recovery capacity. A pronounced polarization effect between the two cores remained evident, with intermediate regions such as Ziyang, Meishan, and Neijiang persistently exhibiting low TER levels. Chongqing was characterized by a strong resource agglomeration and siphon effect, which suppressed surrounding areas and resulted in lower resilience levels in regions such as Dazhou and Guang’an. Specifically, Mianyang was maintained at a relatively high level, though a significant gap compared to the dual cores persisted. Between 2020 and 2023, influenced by the COVID-19 pandemic, TER fluctuations were observed in most regions: Nanchong and Dazhou were elevated from low to relatively low levels, indicating a certain degree of recovery; Yibin and Zigong fluctuated repeatedly between medium and relatively low levels; Deyang and Leshan declined from relatively high to medium levels, Luzhou decreased from medium to relatively low, and Guang’an remained consistently at a low level, reflecting a general weakening of resilience to varying degrees. In contrast, Ziyang, Meishan, and Neijiang remained stable at low levels with no significant changes.

4.3. Evaluation and Analysis of Subsystems of TER

To visually illustrate the spatial heterogeneity of TER subsystems before and after the COVID-19 pandemic in 2020, their spatial distributions were mapped using ArcGIS 10.7.

4.3.1. Evaluation of Subsystems of TER Before the Pandemic

From 2013 to 2019, significant differences and dynamic changes were observed among the subsystems of TER in the CCEC, as illustrated in Figure 4. In terms of resilience, Chongqing was consistently ranked the highest, increasing steadily from 0.807 in 2013 to 0.966 in 2019, maintaining an exceptionally high level throughout. Chengdu showed a moderate increase with fluctuations, with an average value of 0.473, indicating a notable gap compared to Chongqing. Most other cities exhibited a “U-shaped” adjustment trend around 2016, while areas such as Ziyang experienced a continuous decline in resilience. Regarding recovery capability, Chongqing remained consistently at a very high level (0.928 in 2013, 0.975 in 2019), whereas Chengdu demonstrated some volatility (0.665 in 2013, 0.620 in 2019). Cities such as Nanchong and Mianyang were maintained at low to medium levels, while Guang’an and Ziyang showed a significant decline by 2019, indicating systemic challenges in recovery capacity across the region. Innovation capability displayed pronounced polarization. Chongqing (average 0.880) and Chengdu (average 0.821) jointly formed high-level dual cores, both peaking in 2016. In contrast, innovation capabilities in other cities were generally weak: Mianyang experienced a gradual decline, Ya’an saw a sharp decrease, and most regions had average values below 0.14, reflecting a severe lack of regional innovation momentum. Overall, the TER of the CCEC was characterized by a spatial pattern described as “prominent dual cores, a collapsed central region, and vulnerable peripheries,” with significant developmental imbalances among the subsystems.

4.3.2. Evaluation of Subsystems of TER in the Post-Pandemic Era

From 2020 to 2023, significant differences and dynamic characteristics were observed among the subsystems of TER in the CCEC during the impact and recovery phases of the pandemic (see Figure 5). In terms of resilience, a pattern characterized by “one superior and multiple weak, with partial fluctuations” was identified. Chongqing consistently led with an average value of 0.915, demonstrating strong buffering stability. Although Chengdu ranked second (average 0.467), significant fluctuations were observed. Most other cities generally exhibited low resilience, with insufficient stability in areas such as Ziyang and Meishan. Regarding recovery capability, Chongqing was maintained at an extremely high and stable level (average 0.961), while Chengdu steadily improved to 0.669. However, the average recovery capacity of other cities, such as Nanchong and Mianyang, mostly remained below 0.14, indicating overall weak recovery resilience. A declining trend was particularly noted in Guang’an and Ziyang, reflecting uneven regional recovery momentum in the post-pandemic phase. Innovation capability demonstrated a pattern of “dual-core dominance with significant gaps.” Chongqing and Chengdu formed two regional innovation poles with average values of 0.905 and 0.859, respectively, continuously playing a leading role. In contrast, third-tier cities such as Mianyang and Ya’an had average values ranging only from 0.09 to 0.26, with most showing declining or fluctuating trends. Overall regional innovation momentum remained insufficient, constraining the coordinated enhancement of post-pandemic resilience. In summary, during the pandemic, the TER of the CCEC was characterized by a prominent core, vulnerable peripheries, and uneven recovery. Chongqing stably led the region, Chengdu experienced a slow rebound, while the vast majority of cities demonstrated weaknesses across all three subsystems.

4.4. Asymmetric Analysis of Influencing Factors of TER

4.4.1. Data Calibration and Truth Table Construction

Prior to conducting fsQCA analysis, data calibration is essential as the method relies on Boolean algebra and requires all data to fall within the [0, 1] interval—a condition not met by raw data. Adopting the continuous calibration approach proposed by Ragin (2009), this study defines continuous fuzzy sets within the [0, 1] range and transforms variable data into fuzzy-set membership scores via linear scaling, indicating the degree to which a case belongs to a set [29]. Calibrated data are presented in

Table 3. Examples of Data Calibration.

	GOV	TRAN	INDUS	TECH	SER	CONS	RESI
1	0.17	0.95	0.05	0.51	0.86	0.11	0.95
2	0.92	0.7	0.05	0.53	0.95	0.8	0.91
3	0.12	0.06	0.08	0.22	0.23	0.41	0.05
4	0.24	0.51	0.74	0.31	0.13	0.65	0.23
5	0.5	0.11	0.05	0.5	0.61	0.23	0.13
6	0.39	0.43	0.86	0.56	0.56	0.71	0.53
7	0.43	0.08	0.37	0.15	0.21	0.42	0.07
8	0.47	0.12	0.82	0.16	0.63	0.29	0.07
9	0.95	0.12	0.54	0.13	0.54	0.58	0.08
10	0.17	0.41	0.51	0.14	0.55	0.69	0.16
11	0.17	0.05	0.19	0.45	0.2	0.08	0.05
12	0.26	0.16	0.93	0.36	0.05	0.45	0.18
13	0.05	0.05	0.29	0.06	0.09	0.41	0.05
14	0.11	0.42	0.11	0.07	0.6	0.95	0.06
15	0.95	0.05	0.94	0.2	0.52	0.67	0.24
16	0.82	0.06	0.95	0.05	0.21	0.05	0.08

.

The truth table, which lists all logically possible combinations of causal conditions and the outcome [29], serves as the foundation for fsQCA’s configurational analysis. After calibration, a truth table is constructed to facilitate Boolean minimization using the Quine–McCluskey algorithm. Theoretically, the truth table contains 2ⁿ rows, where n is the number of antecedent conditions, with each row representing a unique configuration. To refine the truth table and identify meaningful configurations, consistency and frequency thresholds are applied. Frequency threshold refers to the minimum number of cases with membership scores above 0.5 in a configuration that merits investigation [29]. For small-sample studies, this threshold is typically set to 1, while larger samples may require a higher minimum. It is recommended that frequency filtering retains at least 80% of the cases. The resulting truth table is shown in

Table 4. Examples of Truth Table Construction.

Model: (TRAN, INDUS, TECH, SER, CONS)
Conditional Variables					Number	RESI	Raw Consist.	PRI Consist.
TRAN	INDUS	TECH	SER	CONS
0.00	0.00	0.00	0.00	0.00	4.00	0.00	0.31	0.00
0.00	1.00	0.00	1.00	1.00	3.00	0.00	0.51	0.00
0.00	1.00	0.00	0.00	0.00	2.00	0.00	0.39	0.00
0.00	1.00	0.00	1.00	0.00	1.00	0.00	0.48	0.00
1.00	0.00	1.00	1.00	0.00	1.00	1.00	0.94	0.84
1.00	1.00	0.00	0.00	1.00	1.00	0.00	0.71	0.00
0.00	0.00	0.00	1.00	1.00	1.00	0.00	0.45	0.10
1.00	0.00	1.00	1.00	1.00	1.00	1.00	0.94	0.81

.

4.4.2. Analysis of Necessary Conditions

In this study, fsQCA 3.0 software was used to examine whether individual condition variables (including negated sets) constitute necessary conditions for TER. Necessity analysis refers to the extent to which the outcome set is a subset of the condition set, reflecting the proportion of the intersection of the two variables in the fuzzy set of the outcome—i.e., the absence of a condition can directly prevent the outcome from occurring [39]. The formula for necessity analysis is as follows:

$$\begin{array}{r}\mathrm{Consistency}(X_i\le Y_i)=\sum \{\min (X_i,Y_i)\}/\sum (X_i)\end{array}$$

$$\begin{array}{r}\mathrm{Coverage}(X_i\le Y_i)=\sum \{\min \left(X_i,Y_i\right)\}/\sum \left(Y_i\right)\end{array}$$

where X_i represents the membership score of the i case in the condition combination, and Y_i represents the membership score of the i case in the outcome variable. The operator min denotes taking the minimum value between the condition combination and the outcome variable. The results of the fsQCA necessity analysis (see

Table 5. Necessity Analysis of High-Level and Low-Level TER.

Conditional Variables		High Level TER		Low Level TER
		Consistency	Coverage	Consistency	Coverage
Government Support	GOV	0.760	0.435	0.456	0.826
	~GOV	0.695	0.288	0.688	0.901
Transportation Development	TRAN	0.854	0.766	0.228	0.647
	~TRAN	0.607	0.199	0.918	0.952
Tourism Industry	INDUS	0.521	0.267	0.557	0.905
	~INDUS	0.815	0.367	0.549	0.784
Technological Innovation	TECH	0.763	0.666	0.280	0.775
	~TECH	0.742	0.246	0.879	0.922
Tourism Service	SER	0.917	0.507	0.426	0.746
	~SER	0.542	0.230	0.719	0.965
Tourist Consumption	CONS	0.745	0.381	0.533	0.864
	~CONS	0.734	0.332	0.618	0.885

Note: “~” denotes the meaning of “not”, indicating the absence of that condition.

) indicate that, among the antecedent conditions for high-level TER, tourism service level (SER, consistency = 0.917) exceeded the threshold of 0.9, suggesting that it may approximate a necessary condition for high-level resilience. Transportation development level (TRAN, consistency = 0.854) also exhibited relatively high consistency. For low-level TER, underdeveloped transportation (~TRAN, consistency = 0.918) and low technological innovation level (~TECH, consistency = 0.879) were particularly prominent, with the consistency of underdeveloped transportation exceeding 0.9, indicating that it is highly likely to be a necessary condition for low resilience. Nevertheless, the consistency values of most other variables remained below 0.9, suggesting that the formation of TER still depends on the combined effects of multiple factors. According to existing research, when dealing with such complex socio-ecological outcomes, configuration analysis is required to clarify sufficient combinations of multiple antecedent conditions [29]. Therefore, it is necessary to further employ the fsQCA method to systematically examine the sufficient condition configurations leading to high or low levels of TER.

4.4.3. fsQCA Results

Based on the configuration analysis results of the causal models for high-level TER using fsQCA (see Figure 6), three configurations composed of six antecedent conditions were identified, all demonstrating strong explanatory power for high-level TER. The consistency threshold for the data (>0.75) was exceeded, with all configurations exhibiting consistency values above 0.84, reaching a maximum of 0.959, indicating that these condition combinations can be considered sufficient paths for the outcome [33]. In terms of coverage, the raw coverage of each configuration ranged between 0.391 and 0.453, while the unique coverage ranged between 0.102 and 0.125, suggesting that each path independently and significantly explains a substantial portion of the cases. The overall solution consistency was 0.888, reflecting strong explanatory power of the model.

Configuration 1 (~GOV*TRAN*~INDUS*TECH*SER*~CONS) indicates that, even with low government support, underdeveloped tourism industry, and low tourist consumption levels, high levels of transportation development, technological innovation, and tourism service quality can still drive TER to a high level. Configuration 2 (~GOV*~TRAN*INDUS*TECH*SER*CONS) suggests that, despite insufficient government support and underdeveloped transportation, a strong tourism industry foundation, high technological innovation capability, superior tourism services, and high tourist consumption capacity can still form a path to high resilience. Configuration 3 (GOV*TRAN*~INDUS*TECH*SER*CONS) demonstrates that, with strong government support, well-developed transportation, high technological innovation, superior service quality, and high consumption capacity, high ecological resilience can be achieved even if the tourism industry level is not high (see

Table 6. Complex Configurations of Sufficient Conditions for Predicting High-Level TER.

Causal Model of High-Level TER	Raw Coverage	Unique Coverage	Consistency
Model: RESI = f (GOV, TRAN, INDUS, TECH, SER, CONS)
M1: ~GOV*TRAN*~INDUS*TECH*SER*~CONS	0.430	0.104	0.959
M2: ~GOV*~TRAN*INDUS*TECH*SER*CONS	0.391	0.102	0.843
M3:GOV*TRAN*~INDUS*TECH*SER*CONS	0.453	0.125	0.935
Solution coverage: 0.659
Solution consistency: 0.888

Note: In the logical calculus of Boolean algebra, “Negation” is represented by “~”; “And” is represented by “*”.

).

Through further comparison of the parsimonious solutions, it was identified that technological innovation (TECH) and tourism service level (SER) were consistently present across all three configurations, indicating their role as core conditions, while transportation condition (TRAN) was also found to play a critical role in most configurations. This suggests that the formation of high-level TER predominantly relies on a dual-core driving of “technology-service,” which often functions in combination with factors such as transportation or consumption. Unlike traditional linear analyses, this study revealed that the same variable may exhibit divergent effects across different configurations. For instance, the absence of government support (GOV) in certain paths was still observed to achieve high resilience, reflecting the highly context-dependent nature of antecedent variable effects and corroborating Ordanini’s assertion that “configurations are more critical than single variables” [29]. Therefore, enhancing TER requires a focus on the synergistic combination of multiple conditions rather than isolated factors. To intuitively illustrate the asymmetric relationships between configurations of antecedent conditions and their outcomes, a fuzzy XY plot is drawn. The values adjacent to the horizontal and vertical axes represent the levels of coverage and consistency, respectively (see Figure 7).

Based on the configuration analysis results of causal models for low-level TER (~RESI) using fsQCA (see

Table 7. Complex Configurations of Sufficient Conditions for Predicting Low-Level TER.

Causal Model of Low-Level TER	Raw Coverage	Unique Coverage	Consistency
Model: ~RESI = f (GOV, TRAN, INDUS, TECH, SER, CONS)
M1: ~GOV*~TRAN*~TECH*~SER*~CONS	0.410	0.113	1.000
M2: ~TRAN*INDUS*~TECH*~SER*~CONS	0.373	0.050	1.000
M3: ~GOV*~TRAN*~TECH*SER*CONS	0.260	0.044	1.000
M4: ~TRAN*INDUS*~TECH*SER*CONS	0.267	0.025	1.000
M5: ~GOV*~TRAN*INDUS*SER*CONS	0.191	0.002	0.963
M6:~GOV*TRAN*INDUS*~TECH*~SER*CONS	0.188	0.013	1.000
M7: ~GOV*~TRAN*INDUS*~TECH*~CONS	0.283	0.013	1.000
Solution frequency: 0.646; Solution consistency: 0.989

Note: In the logical calculus of Boolean algebra, “Negation” is represented by “~”; “And” is represented by “*”.

), multiple condition combinations were found to exhibit high consistency (all ≥ 0.963, mostly 1.000) and significant coverage for low-level TER, indicating that these configurations represent sufficient paths leading to low resilience. The overall solution consistency was 0.989, with satisfactory coverage, demonstrating strong explanatory power of the model. The analysis identified multiple condition combinations that could lead to low TER. These combinations are not simply the opposites of high-resilience configurations, reflecting the multifaceted causal complexity of outcome formation. Typical paths include: Configuration 1 (~GOV~TRAN~TECH~SER~CONS) indicates that the absence of government support, underdeveloped transportation, low technological innovation, inferior tourism services, and weak tourist consumption capacity collectively result in low ecological resilience. Configuration 2 (~TRANINDUS~TECH~SER~CONS) suggests that even with a certain tourism industry foundation, poor transportation conditions, insufficient technological innovation, low service levels, and weak consumption capacity still lead to inadequate resilience. In other configurations, combinations such as the absence of government support (~GOV), underdeveloped transportation (~TRAN), and low technological innovation levels (~TECH) frequently recur.

Through comparison with parsimonious solutions, it was further identified that the absence of government support (~GOV), underdeveloped transportation (~TRAN), and low technological innovation (~TECH) are core conditions stably present across multiple configurations, while other variables play auxiliary roles in different contexts. This suggests that the lack of resilience in tourism ecosystems primarily stems from the collective weakness of “systemic supporting elements” such as policy, infrastructure, and innovation. These configurations can be categorized as the “systemic support deficiency” type. This finding aligns with the “multiple conjunctural causation” perspective in complexity theory, which asserts that the same outcome can be triggered by various condition combinations, and the role of variables must be understood within specific configurations [29]. Therefore, enhancing TER requires systematically strengthening foundational support systems such as government support, transportation conditions, and technological innovation, while avoiding allowing single weaknesses to become sources of systemic vulnerability.

In summary, no single antecedent condition was found to be necessary for either high or low levels of tourism ecological resilience (TER). However, multiple distinct combinations of antecedent conditions were identified as sufficient to lead to high or low resilience, reflecting the characteristics of multiple conjunctural causation and asymmetry in its formation pathways. High-level TER demonstrated a “technology-service dual-core driving” pattern, primarily manifested through three sufficient paths:

Path 1: ~GOVTRAN~INDUSTECHSER*~CONS

Path 2: ~GOV~TRANINDUSTECHSER*CONS

Path 3: GOVTRAN~INDUSTECHSER*CONS

Among these, high technological innovation (TECH) and high tourism service level (SER) were consistently identified as core conditions across all paths. In contrast, low-level TER exhibited a “systemic support deficiency” pattern. Multiple paths repeatedly featured combinations such as lack of government support (~GOV), underdeveloped transportation (~TRAN), and low technological innovation (~TECH), indicating that the collective deficiency of systemic elements—public policy, infrastructure, and innovation capability—constitutes the core cause of weak resilience. A comparative analysis revealed that the mechanisms underlying high and low resilience are not simply symmetric: High resilience is driven by the dual-core of “technology-service” combined with varying auxiliary conditions, while low resilience is commonly attributed to deficiencies in multiple systemic supporting elements. This outcome underscores the holistic and configurational nature of TER construction. It suggests that policy practices must emphasize the coordinated development of multiple elements and systematically strengthen weak links to prevent the combined absence of key conditions from constraining the overall resilience of the system. The asymmetry among the seven models, as illustrated in Figure 8, is empirically validated.

4.4.4. Prediction Validity

A well-fitted model does not guarantee strong predictive capability; therefore, predictive validity was employed to evaluate the predictive power of the hypothesized asymmetric causal models across different datasets. The raw data were divided into two subsamples. Subsample 1 was subjected to fsQCA asymmetric causal modeling [41]. Subsequently, Subsample 2 was used to test the causal models derived from Subsample 1.

As shown in

Table 8. Validity Analysis of the QCA Results.

Configuration	Raw Coverage	Unique Coverage	Consistency
High-Level TER
M1: ~GOV*TRAN*~INDUS*TECH*SER*~CONS	0.502	0.502	0.983
solution coverage: 0.502; solution consistency: 0.983
Low-Level TER
M2: ~GOV*~TRAN*~TECH*SER*CONS	0.345	0.077	1.000
M3: ~TRAN*INDUS*~TECH*SER*CONS	0.317	0.029	1.000
M4: ~GOV*~TRAN*~INDUS*~TECH*~SER*~CONS	0.312	0.084	1.000
M5:GOV*~TRAN*INDUS*~TECH*~SER*~CONS	0.357	0.112	1.000
M6: ~GOV*TRAN*INDUS*~TECH*~SER*CONS	0.264	0.038	1.000
solution coverage: 0.694; solution consistency: 1

Note: In the logical calculus of Boolean algebra, “Negation” is represented by “~”; “And” is represented by “*”.

, when the configurations of antecedent conditions for high/low tourism ecological resilience (TER) in Subsample 1 were analyzed using the fsQCA model, the results were consistent with those of the full sample, with both coverage and consistency performing excellently. Specifically, Figure 9 reveals a sufficient yet non-necessary asymmetric relationship between X (M1–M6, representing a configuration model of antecedent conditions) and Y (TER), as illustrated. Subsequently, Subsample 2 was utilized to validate the effectiveness of the M2 model constructed based on Subsample 1. This model similarly demonstrated a comparable asymmetric relationship, with excellent performance in both coverage (0.954) and consistency (0.944). In summary, the asymmetric causal models were found to exhibit significant predictive validity.

5. Discussion

5.1. Spatiotemporal Differences in TER

Temporally, the regional TER during the study period was characterized by a fluctuating upward trend overall, yet with significant internal disparities. Chongqing and Chengdu consistently maintained leading positions, while most other cities remained at relatively low levels, reflecting a typical “dual-core driven, peripheral lag” development pattern. This finding is highly consistent with the “core–periphery” structure observed in regional economic development and tourism industry agglomeration [42,43], indicating that TER is significantly influenced by economic foundations, policy support, and resource endowments.

Spatially, the TER of the CCEC exhibited stable and pronounced dual-core polarization. High-resilience areas were consistently concentrated in Chongqing and Chengdu, demonstrating strong siphon effects and regional leadership. Intermediate transitional zones (e.g., Ziyang, Meishan, Neijiang) and peripheral cities persistently showed low resilience levels, failing to form an effective resilience transmission network. Between 2013 and 2019, although resilience levels in cities such as Nanchong and Dazhou slightly improved, they generally remained at low to medium levels. Meanwhile, resilience decline was observed in Deyang, Leshan, and Guang’an, indicating limited systemic regulatory capacity in some regions when facing internal and external disturbances. This spatial differentiation pattern aligns with typical findings in tourism destination resilience research, suggesting that the resilience of tourism systems depends not only on resource endowments but also on regional collaborative capacity and institutional responsiveness.

The pandemic further highlighted the unbalanced responses of subsystem resilience. During 2020–2023, resilience was characterized by a “one superior, multiple weak” pattern: Chongqing remained at a high level (0.915), Chengdu at a medium level (0.467), and other cities generally at low levels (below 0.15), indicating that core cities possessed stronger buffering capacity under sudden public crises. In terms of recovery capability, Chongqing maintained extremely high stability (0.961), while Chengdu improved to 0.669. However, recovery levels in other cities such as Nanchong and Mianyang mostly remained below 0.14, reflecting overall weak regional recovery resilience. This is consistent with findings that post-crisis tourism recovery exhibits significant spatial heterogeneity [25]. For innovation capability, a “dual-core dominance with significant gaps” was observed: Chongqing and Chengdu achieved average values of 0.905 and 0.859, respectively, while third-tier cities such as Mianyang and Ya’an averaged only 0.09–0.26, with most showing declining or fluctuating trends. This indicates that insufficient innovation momentum severely constrained the coordinated enhancement of regional resilience, echoing the view that “lack of innovation is a major source of vulnerability in tourism systems” [44].

In summary, the TER of the CCEC demonstrated steady temporal improvement but persistent internal divergence, while spatially it exhibited a clear gradient distribution and dual-core agglomeration. The pandemic further tested and exacerbated imbalances among subsystems. These findings not only deepen the understanding of the spatiotemporal evolution of tourism ecosystem resilience but also provide empirical evidence for regional resilience policy formulation.

5.2. Mechanisms Underlying Differences in TER

Unlike previous studies based on symmetric thinking and linear assumptions, this research systematically reveals the multiple pathways and complexity of TER formation mechanisms from a configurational perspective and asymmetric logic. The results demonstrate that the driving mechanisms of high and low resilience are not symmetric; their antecedent combinations are not simply inverses and cannot be explained by the positive or negative directions of the same variables, thereby transcending the limitations of traditional symmetric causal frameworks. Variables such as government support and technological innovation exhibit high context dependence. For example, government support can be either absent or present in high-resilience paths, whereas its absence serves as a core condition in multiple configurations for low resilience. Technological innovation stably appears as a core element in high-resilience configurations, deepening the understanding of its role in complex tourism ecosystems and echoing the view that technological empowerment requires synergy with institutions and facilities [45]. In contrast, technological innovation is generally lacking in low-resilience configurations. Whether high resilience can ultimately be achieved depends on its combination with conditions such as transportation and services [46]. Transportation conditions are mostly present in high-resilience configurations and absent in low-resilience ones, but their effects also require synergy with high-level services and technology. The fsQCA method adopted in this study aims to identify sufficient condition combinations, emphasizing the interconnectedness and concurrent effects of variables rather than their independent net effects, thereby revealing the heterogeneous roles of variables across different configurations and contributing to the understanding of multiple conjunctural causation mechanisms in TER formation [46]. Thus, this study does not negate the value of traditional methods but provides an important complement and extension from a configurational perspective, collectively promoting a more comprehensive understanding of the complex causes of TER.

6. Conclusions

6.1. Conclusions

Based on data from 2013 to 2023, a comprehensive assessment and mechanistic analysis of TER in the CCEC was conducted, leading to the following conclusions: and

(1)

Temporal Evolution: TER demonstrated an overall upward trend, though significant internal regional disparities were observed. Chongqing and Chengdu consistently maintained leading positions, with average values of 0.634 and 0.491, respectively, while most other cities exhibited averages below 0.155, reflecting a development pattern characterized by “prominent dual cores and lagging peripheries.” Chongqing showed steady growth, reaching 0.796 in 2023, while Chengdu experienced fluctuations with an upward trend, achieving 0.659 in 2023. Most cities exhibited low resilience levels, indicating a need for enhanced regional coordination.

(2)

Spatial Pattern: A distinct “dual-core” structure was identified, with Chongqing and Chengdu consistently classified as high-resilience areas, demonstrating strong polarization and radiation effects. Cities such as Mianyang, Deyang, and Nanchong were categorized at medium levels, while Ziyang, Meishan, Neijiang, and Guang’an persistently exhibited low resilience, highlighting pronounced regional development imbalances. The spatial structure remained generally stable before and after the pandemic, though fluctuations were observed in subsystems of some cities.

(3)

Subsystem Comparison Pre- and Post-Pandemic: Resilience: A pattern of “one superior, multiple weak” was observed, with Chongqing averaging 0.915, Chengdu at 0.467, and most other cities below 0.15. Recovery: Chongqing maintained extremely high stability (average 0.961), while Chengdu steadily improved to 0.669. However, recovery capabilities in other cities were generally insufficient, mostly below 0.14. Innovation: A “dual-core dominance” pattern was evident, with Chongqing and Chengdu averaging 0.905 and 0.859, respectively, while other cities mostly averaged below 0.12, indicating severe inadequacy in innovation momentum.

(4)

Influencing Mechanisms: fsQCA analysis revealed that TER is influenced by configurational effects of multiple factors, with no single necessary condition identified. High-resilience pathways relied on the dual-core driving of “technology-service” (technological innovation and tourism services), combined with auxiliary conditions such as transportation, consumption, or government support. In contrast, low resilience stemmed from multiple systemic deficiencies, including lack of government support, underdeveloped transportation, and insufficient technological innovation. The causal mechanisms exhibited significant asymmetry, indicating that pathways to high and low resilience are not simply opposites.

In summary, the TER of the CCEC is characterized by a distinct core–periphery structure and multiple path dependencies. Future policies should focus on systematically addressing weaknesses and promoting synergy among technology, services, infrastructure, and policy to enhance overall regional resilience.

6.2. Limitations and Suggestions

This study systematically analyzed the TER level and its influencing mechanisms in the CCEC, but three limitations remain to be addressed in future research: First, TER is a relatively new concept. Although a comprehensive evaluation index system was constructed by integrating systemic characteristics and resilience connotations, the comprehensiveness and applicability of the index system can still be improved, constrained by the scope of knowledge and data availability. Second, the CCEC was used as the case region in this study. Future research could expand to horizontal comparisons or longitudinal tracking of multiple urban agglomerations or conduct more micro-level analyses of tourist behaviors or tourism flow patterns in specific cities or regions to enrich the scale and dimensions of the research. Finally, the analysis of the dynamic evolution mechanisms of TER under sudden public crises (e.g., the COVID-19 pandemic) remains insufficient. The interaction between short-term shocks and long-term resilience construction was not fully revealed. Future studies could introduce evolutionary resilience theory to strengthen the dynamic analysis of resilience formation, adaptation, and transformation processes. Simultaneously, we recommend using long-term time series to conduct an asymmetrical comparative analysis of TER before and after the pandemic.