1. Introduction
Along with the rapid development of tourism, tourism links between destinations are getting closer, gradually forming a tourism economic network. The tourism network helps destinations function in a complex and competitive world, which acts as a system to organize and integrate destinations for the participating tourism companies and to improve the performance and quality of destinations [
1,
2,
3]. Moreover, tourism economic connections could promote the coordinated development of the regional tourism economy and provide a basis for scientifically formulating regional tourism coordinated development policies [
4]. Studies have focused on the characteristics, impact, and influencing factors of attraction and destination networks [
5,
6,
7]. However, few studies have evaluated and discussed the evolution of the spatial network structure of the tourism economy in tourism destinations. Therefore, examining the tourism economic network structure in tourism destinations is significant to tourism scholars.
The Tibetan Plateau is extremely rich in tourism resources, such as a variety of landscapes and special ethnic flavors, which attract a large number of visitors every year [
8]. Since 2000, the number of tourists on the Tibetan Plateau has been on the rise. In 2019, the total tourism revenue of Qinghai and Tibet was equivalent to 22.94% and 38.29% of GDP, respectively, and tourism has become a pillar industry in the economy of the Tibetan Plateau region. The development of the tourism economy plays an increasingly important role in promoting regional economic development. However, cities still lack a range and in-depth level of tourism economic interactions. The study of urban tourism networks on the Tibetan Plateau is of great significance.
1.1. Network Analysis in Tourism
In recent years, social network analysis in tourism research has witnessed an increasing number of studies. The research streams mainly focus on interorganizational relations, virtual networks, and the networks of attractions and destinations [
6].
From the perspective of interorganizational relations, research applying network analysis has focused on stakeholders’ collaborations and network dynamics. Tran et al. [
9] analyzed the pattern of the network between tour operators and travel agencies, as well as between tour operators. Kim and Scott [
10] attempted to trace the evolution of the organization of tourism stakeholders.
From the virtual networks perspective, the research focuses on the influence of particular users or stakeholders in social media. Scott et al. [
11] took the island of Elba as an example to compare the communication network of tourist organizations and the virtual network formed by the websites of these organizations.
Tourism destinations are composed of different types and levels of tourism suppliers and tourism organizations [
5]. From a tourist demand perspective, research has been conducted on the network of attractions [
4,
12,
13], the formation mechanism of network relationships [
14], and the structure and retrieval paths of destination image [
15]. Wang et al. [
13] studied the structure of tourism economic networks in national parks in central China and modified the gravity model using the total revenue of the national parks. Liu et al. [
14] used social network analysis to explore the mechanisms of tourist attraction network informed by tourist flows. From the supply perspective, the existing literature generally concentrates on the characteristics of the network [
16], the influence of network structure, and the influencing factors of the network [
7]. Wang et al. [
17] explored the spatial network characteristics of the tourism economy of 15 cities in Xinjiang by combining the modified gravity model with the comprehensive tourism quality assessment index. Based on the modified gravity model and QAP analysis, Wang and Xia [
18] visualized the structural morphological characteristics of the spatial network of China’s provincial tourism economy and analyzed the influencing factors. Yang et al. [
19] used the gravity model and social network analysis to analyze the spatial structure of the tourism economy in the Beijing–Tianjin–Hebei–Xiong region in 2013 and 2016. In conclusion, research on social networks has shifted from focusing on the analysis and description of network structures to exploring the formation, development, and effects of relational networks.
1.2. Network Analysis in the Tourism Economic Network
The tourism economic network is a point-line-surface interwoven tourism spatial structure based on the regional space, tourism activities, and other elements, such as tourism passenger flow and tourism goods flowing among tourism destinations [
20], which is the core category of the tourism destination network. Tourism economic linkages are a comprehensive reflection of the flow of tourism factors and to a certain extent reflect the spatial structure of regional urban tourism. Cities with development advantages can exemplify the development of tourism economies in other cities, thus creating a radiating effect of tourism [
21]. The spatial network structure of the tourism economy can facilitate the spatial integration between tourism destinations and promote the development of the tourism industry [
22]. In terms of the research perspective, the main focus is on economically developed urban zones and urban agglomerations [
4,
23]. To date, there are two general approaches to the study of urban economic linkages: (1) the combination of gravity model and social network analysis [
24] and (2) the use of various factor flows data to reflect urban economic linkages; Mou et al. [
25] found that online travel diaries can accurately reflect the spatial characteristics of tourism flow. Matsumoto and Domae [
26] assessed the competitive relationship between Eurasian cities based on data on air passenger and cargo traffic. Monzon et al. [
27] discussed that new high-speed rail lines significantly increase the level of accessibility of Spaniards to most destinations. However, the gravity model has the advantages of easy access to data, comprehensiveness and wide applicability [
28], and is therefore widely used. In addition, the depth and breadth of research on urban networks from a tourism perspective is progressively advancing, but mostly on intercity hierarchical ranking while lacking in the exploration of the mechanisms driving urban tourism economic networks. Social networks enable a deeper understanding of urban tourism networks from a dynamic, evolutionary perspective. In addition, there are few studies on the network space of the tourism economy in underdeveloped areas [
13]. However, research on Tibetan Plateau cities with great potential for tourism development has not been conducted.
1.3. Tourism Development in the Tibetan Plateau
The Tibetan Plateau is one of the key areas for the construction of ecological civilization in China and an important cultural preservation site of Chinese national characteristics. The Chinese government attaches importance to the exploration of the Tibetan Plateau for tourism development. Since 2010, the Communist Party of China (CPC) Central Committee’s Tibet Work Symposium has attached importance to the development strategy of making Tibet a world tourism destination [
29]. Tourism development is also important to poverty alleviation among farmers and herders in the Tibetan Plateau region. Tourism is a way to alleviate poverty through income and employment generation, especially for rural areas in developing countries [
30]. By integrating tourism more closely into the local economy and thus promoting other businesses, linkages can be enhanced [
31]. Tibet’s tourism industry has helped 32,000 poor people out of poverty, including rural tourism income of RMB 1.2 billion. Tourism is a green way of poverty alleviation for the Tibetan Plateau, in line with regional economic positioning and green development goals, but attention should be paid to scientific planning and rational development [
32].
Highland tourism has received extensive attention from scholars, and research on the Tibetan Plateau has mainly focused on tourists and tourism resources. In the tourist context, studies focus on tourists’ travel constraints and revisit intention [
33], and spatiotemporal behavior [
34]. From the resource and economic perspective, tourism resource spatial patterns [
35] and the impact of climate change on tourism [
36] have been discussed. In a study evaluating the potential for ecotourism development on the Tibetan Plateau, Zhang et al. [
37] found that although the region within the Tibetan Plateau is extremely rich in tourism resources, it is limited by infrastructure and other constraints, and there is thus a divide between tourists and destinations. However, the tourism economic network in cities on the Tibetan Plateau has not been explored. Therefore, this study takes the main cities on the Tibetan Plateau as the case study, and we first built a modified gravity model to identify the tourism economic association from a network perspective. Second, we explore the dynamic evolution of tourism economic connections and cooperation between cities, and analyze the influencing factors by quadratic assignment procedure (QAP) methods. Finally, this study aims to provide a theoretical and practical scientific reference for enhancing tourism economic connections and development.
This paper has been divided into five parts. Following the introduction and literature review, the second part introduces the study area and a rigorous description of the methods employed. The third part presents the empirical results of the study, in which we describe the tourism economic network structure of the Tibetan Plateau. The fourth part discusses the findings, while the last part concludes the analysis and presents practical implications.
3. Results
3.1. Strength of the Tourism Economic Connection
Following the modified gravity model, this paper calculated the strength of tourism economic connections among the 18 major cities on the Tibetan Plateau during 2015 and 2019 (
Table 2).
Figure 2 shows that each node indicates one city, and the connecting line presents the interaction between two cities. The width of the connecting line indicates the degree of interaction.
In 2015, the top 3 cities in terms of tourism economic connections were Xining, Haidong, and Hainan, which together accounted for 79.31% of the total tourism economy. The bottom 3 cities, Changdu, Naqu, and Ali, had weak tourism economic ties with other cities. In 2019, although the number of economic connections to cities generally improved, there was little change in city rankings. In recent years, the development of tourism routes, such as the Sichuan–Tibet route and the Qinghai–Tibet route, has led to the development of tourism in cities along the route and promoted overall Tibetan Plateau tourism development.
At the provincial level, Qinghai Province had the highest tourism economic corporations among cities. This is because Qinghai Province has rich tourism resources and convenient transportation facilities. The 13th Five-Year Plan for Tourism Development in Qinghai Province proposes to systematically plan the province as a large scenic area and enhance the overall concept. The tourism economic connection in Tibet is the lowest, which is related to the low level of economic development and traffic accessibility.
3.2. Tourism Economic Network Structural Characteristics
3.2.1. Structural Characteristics and Evolutionary Trend of the Overall Network
This study identified the tourism economic network structure of Tibetan Plateau cities and established a relation matrix.
Figure 3 plots the evolutionary trends of the network density and association relations of the tourism economic network from 2015 to 2019, which shows that the network density exhibited an upwards trend overall. We used ArcGIS 10.5 to visualize the structure of the spatial association network in 2015 and 2019, as shown in
Figure 4. The network density and the number of association relations in 2019 increased by 32.80% and 32.89%, respectively, compared with the data in 2015. In summary, these results showed a marked improvement in the regional tourism economic connections between cities.
3.2.2. Evolutionary Trend of the Centrality
This study used the data from 2015 and 2019 to calculate the degree, betweenness, and closeness of the tourism economic network of the Tibetan Plateau. The spatial distributions of degree, betweenness, and closeness centrality are shown in
Figure 5.
The degree centrality index showed that 7 cities were above the mean value (49.67) in 2015; the highest value was Xining, followed by Haixi, Lhasa, Aba, Haidong, Ganzi, and Hainan, suggesting that the 7 cities had more economic tourism connections between the surrounding cities. In contrast, Naqu and Ali ranked in the last two, suggesting that these cities had a low connection with other cities. The overall degree centrality value showed a distribution characteristic of “high in the east and low in the west”, which indicates that the tourism connection to Tibet is still in the initial development stage and that the connection with other regions is weak. In 2019, in addition to the 7 cities mentioned above, the centrality of Rikaze was larger than the average (66.01). Overall, although the degree centrality of tourism connections on the Tibetan Plateau still exhibited the characteristics of the east being stronger than the west, the mean value of the degree centrality in 2019 increased by 32.89% compared to 2015, indicating that the imbalance tends to moderate with the enhancement of the radiation effect.
In terms of the closeness centrality index, 11 cities were above the mean value (39.54) in 2015, and the top 3 cities were Xining, Haixi, and Lhasa, suggesting that these cities had a high level of control over other cities in the tourism economic network. In 2019, the average closeness degree increased to 11.08, while the top three cities remained the same. This shows that the three cities had a stronger ability to gain tourism economic factors than other cities.
Figure 5 shows that the closeness of Tibetan cities is generally low, and they are marginalized in tourism economic connections from 2015 to 2019.
The betweenness centrality index showed that 5 cities were above the mean value (2.45) in 2015, followed by Xining, Haixi, Lhasa, Haidong, and Aba, suggesting that the control power of these cities was stronger than that of other cities on the Tibetan Plateau and acted as a bridge in the spatial network of the tourism economy. The regional network of tourism links was overly dependent on these cities, with a low intermediate centrality of other municipalities, weak controllability of other cities, and an imbalance in the development of the entire tourism network. Naqu, Ali, and Shannan ranked in the last 3, with a betweenness centrality value of 0, indicating that 3 cities had weak controllability for other cities. In 2019, the mean value of betweenness centrality decreased to 2.12, and the value of Rikaze was significantly higher, while the value of Ali and Naqu remained at 0.
Figure 5 shows that Xining, Haixi, and Lhasa still ranked in the top three, suggesting that these cities regulated the tourism economic connections in the Tibetan Plateau.
3.2.3. Structural Holes
In the tourism economic connection network, the cities with a high level of structural holes can connect the cities that are not directly connected, representing the degree of homogeneous competition.
Table 3 shows that in 2015, Lhasa, Xining, and Haixi had a large effective size and high efficiency, lower constraints, and a high level of structural holes in the tourism economic network. These cities can promote communication between different groups in the overall network, such as tourists and economic flow. However, Naqu, Shannan, Haibei, Guoluo, and Gannan have a relatively low effective size and efficiency and a higher constraint, which indicates that these cities lack competitive advantages.
In 2019, the level of structural holes in the Tibetan Plateau urban network generally decreased, indicating that the network structure is becoming more complex. The advantages in Lhasa, Xining, and Haixi were still obvious. The effective size and efficiency of Rikaze, Ganzi, and Yushu significantly improved, indicating that their ties with other cities have been strengthened and can conveniently obtain more information from other cities. Ali and Naqu still had the lowest level of structural holes. They should strengthen cooperation with the surrounding cities in an advantageous position, such as Lhasa and Rikaze.
3.2.4. Core–Periphery Model
Through the analysis of the core–periphery structure of urban tourism economic ties on the Tibetan Plateau (
Table 4 and
Figure 6), it can be seen that the overall tourism economic network on the Tibetan Plateau presented a significant core–periphery distribution pattern.
In 2015, the overall network structure developed unevenly, showing a spatial pattern of the “western part is higher than the eastern part”. Except for the isolated core Lhasa, the southeast and northeast cities, such as Haixi, Xining, Aba, and Ganzi, were at the network’s core. In contrast, remote cities, such as Ali, Rikaze, and Naqu, were marginalized. These core cities had more spatial associations with other regions and contributed to the whole tourism association of the Tibetan Plateau.
In 2019, the number of cities located in the core grew to 10, of which Haibei, Yushu, and Rikaze became new core areas and Huangnan became marginal areas. The layout of the core area shows the distribution of two areas: the Qinghai–Sichuan area and the Lhasa–Rikaze area. On the whole, the degree of tourism economic connection is gradually rising, and the cities in the marginal areas gradually join the core structure. Lhasa, Xining, Ganzi, and Aba have always been at the core of the tourism economic network. Ali and Naqu have weak tourism economic connections with other cities and states on the Tibetan Plateau. On the one hand, they have fewer tourism attraction resources. On the other hand, they are far away from the tourist source market, and the high altitude also limits the physical condition of regional tourists. As a result, it has formed the periphery of the tourism economic network with relatively weak attractiveness.
3.3. Factors Influencing the Tourism Economic Network
The previous analysis addressed the centrality of cities in the urban tourism economic network of the Tibetan Plateau as well as the core–edge structure. Next, the factors that influence the spatial association network were discussed. First, the dependent variable is the spatial association, i.e., the spatial association matrix obtained above for 2019. Second, the following seven relationship variables were selected to examine the factors influencing spatial association: (i) the A-level attractions—expressed by the sum value matrix of the number of A-level attractions between cities; (ii) the difference in elevation—expressed by the absolute value matrix of the difference in the elevation between cities; (iii) the star-rated hotels—expressed by the sum value matrix of the number of star-rated hotels between cities; (iv) the percentage of total tourism revenue in GDP—expressed by the sum value matrix of the percentage of total tourism revenue in GDP between cities; (v) road density—expressed by the sum value matrix of the road density between cities; (vi) geographical adjacency—if two cities are geographically adjacent, then the matrix element is assigned with 1, otherwise, it is 0; and (vii) the administrative connection—if two cities belong to the same province, then the matrix element is assigned with 1, otherwise, it is 0.
The direction of spatial association was not considered in this study. We used the absolute values of the differences across cities to construct a difference matrix for indicators i to vii to make various indicators comparable. The range method was used to normalize the relation matrix.
3.3.1. The QAP Correlation Analysis
The QAP correlation analysis compares the similarity of two matrix values based on matrix data arrangement. UCINET software was used to randomly arrange the matrix data 5000 times to obtain the correlation coefficient between the spatial correlation matrix of the Tibetan Plateau tourism economic connection network and its influencing factors. The results are shown in
Table 5.
Table 5 shows that the correlation coefficient between the star-rated hotels and the spatial association is significantly positive at the 1% level. The A-level attractions and road density have a significant positive correlation with spatial relations at the 5% level. The difference in elevation, the percentage of total tourism revenue in GDP, geographical adjacency, and administrative connection do not pass the significance test, indicating that these four variables have no significant impact on the spatial association.
3.3.2. The QAP Regression Analysis
The QAP regression analysis is used to explore the regression relationship between multiple independent variable matrices and a dependent variable matrix. This study removed four variables: the difference in elevation, percentage of total tourism revenue in GDP, geographical adjacency, and administrative connection. We selected the remaining three driving factors with significant correlation coefficients as the independent variables for QAP regression analysis. UCINET software was used to randomly permute the data 20,000 times, and the results are shown in
Table 6. The regression coefficients of the relation matrix variables and their significance test results imply that A-level attractions and star-rated hotels significantly affect the formation of the spatial association network.
The coefficient of the A-level attractions matrix is significantly positive at the 5% level, indicating that the A-level attractions play an important role in promoting the formation of the spatial association. The reason is that tourist attraction is the significant component of tourism products and the center of the tourism industry. Lhasa and Xining have far more tourism resources than other cities on the Tibetan Plateau, so it is easier to have tourism economic associations with other cities.
The coefficient of star-rated hotels is significantly positive at the 1% level, indicating that star-rated hotels have a positive effect on the formation of the spatial association. The reason is that the scale and grade of star-rated hotels have become an important yardstick to measure local economic indicators, and star-rated hotels in Western China have an obvious driving effect on the tourism economy [
58]. However, the accommodation conditions in the high-altitude areas of Tibet and Qinghai restrict people’s travel choices [
33]. Ordinary travel agencies will also prioritize the safety and comfort of accommodations when arranging travel, resulting in more tourism links between cities with a large number of hotels.
In addition, road density has no significant impact on spatial association.
4. Discussion
The purpose of this paper was to explore the tourism economic network of the Tibetan Plateau. This study not only measured the strength of tourism economic connections on the Tibetan Plateau, but also analyzed the dynamic evolution and driving factors of the spatial association network.
First, the tourism market and traffic distance have become important factors affecting tourism economic connection strength between cities, which is consistent with the studies of Yu et al. [
59]. The number of economic connections to cities generally improved from 2015 to 2019, but there was little change in city rankings. Xining had the highest strength of tourism economic connections, while Ali had the lowest. On a provincial scale, cities in Qinghai Province generally had higher tourism economic connection strength, more bordering provinces around them, and a series of government policies issued to promote overall tourism development. Tibet had a lower strength of economic connection. This may be due to the local tourism industry being in a development stage. In terms of the structural characteristics of the overall network, the density value and association relations of the network increased during the study period, indicating that regional tourism economic connections increased significantly, with cities becoming increasingly connected.
Second, the results of degree, closeness, betweenness centrality, and structural holes analyses showed that Xining, Haixi, and Lhasa presented the characteristics of strong centrality with a structural hole advantage. Ali, Naqu, and Guoluo presented the characteristics of low centrality. Although the overall development of the network tended to be balanced during the study period, it still showed a greater value of centrality in the western part of the Tibetan Plateau than in the eastern part. At the provincial level, the centrality of cities in Qinghai Province was stronger than that in other regions.
Third, the urban tourism economic network of the Tibetan Plateau shows an obvious core–edge structure, with uneven development of the overall network structure in 2015, showing the spatial pattern that determines that the “western part is higher than the eastern part”. The year 2019 shows the layout of the core area, showing the distribution of two areas: the Qinghai–Sichuan area and the Lhasa–Rikaze area. In general, the scope of the core area gradually expands, and the cities in the periphery gradually integrate into the high-density core structure. Core cities also promote the development of surrounding cities, which corroborates the finding from the study conducted by Gan et al. [
4].
Finally, based on the QAP regression results, we can see that A-level attractions and star-rated hotels can significantly affect the tourism economy connection. These findings provide a theoretical basis on which policy-makers can promote tourism economic integrations and connections. The Tibetan Plateau’s resource endowments dictate that its development needs to rely on tourism, which plays an important role in driving local herders out of poverty and increasing their income, as well as in the sustainable development of ecotourism. Through a method of quantifying different tourism development objectives, Zhang et al. [
29] found that there was an urgent need for Tibet to develop tourism-related infrastructure, which is consistent with our results. Therefore, the results obtained through our QAP analysis are consistent with the actual needs of tourism development on the Tibetan Plateau. However, the relatively backward economic conditions and fragile ecological environment dictate that the scale of tourism must be strictly controlled in the process of tourism development on the Tibetan Plateau according to the existing ecological carrying capacity. The government should pay attention to the rational development of scenic spots and build a reasonable number of star-rated hotels.
4.1. Theoretical Implications
This study has several theoretical implications. First, in terms of the research method, this study uses the social network approach to analyze the network structure characteristics of Tibetan Plateau cities from the perspective of the network, which enriches the related research on tourism spatial networks.
Second, the modified gravity model is used to calculate the strength of tourism economic connections between cities. Tourism revenue and the number of tourists are chosen to represent tourism quality, the ratio of geospatial distance to the difference of GDP of two cities is taken as the distance between two cities, and the proportion of tourism trips to the total number of tourism trips of two cities is used to correct the empirical constants. Thus, a tourism economic spatial connection matrix is established.
Third, most of the previous studies focus on urban clusters in economically developed regions. However, this paper makes up for the lack of research in this area by studying cities on the Tibetan Plateau, which possess great potential but face challenges in development. It contributes to the sustainable development of the tourism economy.
Finally, this study identifies the evolution of different cities’ positions in the tourism economic network from 2015 to 2019 and explores the driving factors of the evolution of the network structure.
4.2. Practical Implications
First, in the Tibetan Plateau city tourism economic network, Lhasa, Xining, Haixi, Aba, and other cities located in the core position should enhance the role of a bridge, with the advantages of tourism economic growth pole radiating to surrounding cities. Marginal areas, such as Ali and Nagqu in Tibet and Guoluo in Qinghai, should promote tourism economic development policies and formulate scientific tourism development plans. Moreover, based on the tourism lines of the Sichuan–Tibet South Line, Sichuan–Tibet North Line, and Qinghai–Tibet Line, various characteristic cultures are spatially combined to improve the exchange network, and advance the formation and development of regional tourism integration on the Tibetan Plateau.
Second, Tibetan Plateau cities should benefit from the plateau characteristics of tourism resources to improve the development and construction of scenic spots and service levels to attract tourism investment. In particular, cities located in low-value areas, such as Ali and Nagqu, should fully explore the ecological and cultural resources, such as the Guge dynasty, sacred mountains, sacred lakes, Qiangtang nature reserve, and other tourism resources with world-class value, and turn the unique and high-quality ecological and cultural resource advantages into economic advantages.
Third, under the premise of protecting the environment, the construction of star-rated hotels and other service facilities in moderation promotes the strengthening of tourism economic ties. The tourism reception capacity and service quality are improved, providing tourists with a more convenient and comfortable tourism experience.
5. Conclusions
Based on the data for the Tibetan Plateau cities from 2015 to 2019, the modified gravity model was used to determine the spatial association. In addition, the social network analysis method was adopted to study the structural characteristics and evolutionary trends of the overall network and centrality. Finally, the QAP correlation and regression analysis methods were employed to explore the driving factors. The study lead to the following conclusions:
The strength of tourism economic connections between cities on the Tibetan Plateau witnessed an improvement from 2015 to 2019, but it was unbalanced in spatial distribution. Tourism economic strength was generally higher in Qinghai Province and lower in Tibet.
The overall network structure from 2015 to 2019 showed an overall upwards trend in the density and association relations. The tourism and economic connections between cities on the Tibetan Plateau increased significantly, and the tourism and economic ties grew closer. However, network connectivity was significantly higher on the eastern Tibetan Plateau.
The spatial association network presented a significant core–edge distribution pattern. Xining, Haixi, and Lhasa were all higher than other cities in the network, which was the bridge with the advantage of structural holes. Meanwhile, the centrality of Rikaze significantly improved during the study period. However, Naqu, Ali, and Guoluo were always at the edge of the network.
The QAP regression analysis showed that A-level attractions and star-rated hotels could significantly promote the formation of spatial association.
Several limitations to this study need to be acknowledged. As a result of the difficulty in obtaining data on the tourism economy of some cities, the current study only examined the data from 2015 to 2019; thus, it is difficult to study the evolution of the tourism economy of cities on the Tibetan Plateau over a longer period of time. In addition, in further research, we can try to explore the economic links of tourism between cities by studying different tourism flows. Moreover, this study analyzed the influencing factors on the structure of the tourism economic network. However, empirical studies are unable to determine the mechanisms of influencing factors. Future research should therefore concentrate on methods to explore the impact mechanism of these factors.