Introduction
Temporary Event-Based Development in Small Cities
Hosting mega-events is generally expected to have substantial benefits to cities, such as profit from event ticket sales and tourism, city branding, lower unemployment rates, greater industry output (Atkinson et al., 2008; Gratton and Preuss, 2008; Vanolo, 2008, 2015; Zhang and Zhao, 2009; Bottero et al., 2012; Minnaert, 2012; Chappelet and Lee, 2016). In general, however, changes related to mega-events are radical, temporary and top-down. In contrast to gradual bottom-up development, these event-led changes have an immediate effect, but they do not guarantee any kind of sustainability. Moreover, these temporary changes often raise operational and management concerns related to the sustainability of infrastructure after the event has taken place (Lei and Spaans, 2009; Müller, 2015). Therefore, besides difficulties in measuring the direct and immediate impact leading up to and during mega-events, the efforts necessary to measure the indirect and long-lasting impacts, those that influence a city’s economic sustainability while also justify the motto of “sustainable Games,” are also quite significant (Miyoshi and Sasaki, 2016; Garcia, 2017; Lauermann, 2019). In contrast to the Summer Olympics, which were mostly hosted in metropolitan cities such as London, Barcelona and Paris, the Winter Olympics are often hosted in small cities with appropriate geographical conditions to accommodate winter sports (Essex and Chalkley, 2007). Moreover, hosting the Olympics is considered an important opportunity for small, declining cities to revitalize and boost their industry and economy. For example, Nagano, the 1998 winter Olympic host city, was a small city with a population of 0.3 million, with appropriate geographical characteristics for winter sports. Turin, usually referred to as the “Detroit of Italy,” has been suffering from a continuous decline since the automotive industry, the city’s most important industry, left. When Turin was selected to host the 2006 Winter Olympics, the city government strategically planned Olympic venues and a new railway station in close relation to a previous urban re-vitalization project in the “Lingotto” area, which had been abandoned since the Fiat factory left the district. However, the impact of the Olympics was not long-lasting in either city. Nagano is often referred to as a failed case. Although Turin is seen as a successful case, the Olympic impact was not enough to fully revitalize the city and to make it endure the subsequent economic crisis in 2008. The Athletes’ residences in Turin became either an abandoned building where homeless and drunk people slept or was redeveloped as a new residential area.
Structural Changes in Olympic Host Cities: Turin, Nagano and Gangneung
Among winter Olympic host cities, Turin, Nagano, and Gangneung were further studied in terms of spatial changes during Olympic infrastructure preparation. The case of Turin was selected because it has been often reported as a successful urban revitalization case based on Olympic. While, Nagano was selected as the case has been criticized as a failure case due to excessive spending for post management of Olympic infrastructure. Lastly, Gangneung was focused study site for this research because the city had Olympic ahead and appropriate to investigate the changes before and after Olympic, and advantageous to collect the field data as needed.
Hosting the Olympics brings substantial changes to both the physical settings and the spatial structure of a city (Fig. 1). These changes can be largely categorized into three: accessibility enhancement, venue construction, and new residence preparation (Kim, 2017). In terms of accessibility, new transport connections to major cities and within cities, such as railways and subways, were constructed. In Turin, for example, an old tram line connecting the old city center to commercial and cultural amenities was extended into the newer city center called Lingotto, where Olympic venues were located. These areas were then further connected through existing subway lines to the outskirts of the city. The purpose of this plan was to revitalize the declining old industrial area, making it function as a new city center (Fig. 1). For the Nagano Olympics, the government constructed a new Shinkansen express railway to connect to Tokyo (Fig. 1). A new railway station was built in the old city center where the city hall, a historical temple, museums, and the city plaza are located. More recently, the Gangneung city government, in preparation for the 2018 Winter Olympics, built a new KTX express railway connecting the city to Seoul and the Incheon Airport, which reduced the travel time from Gangneung to Seoul to approximately one hour, drawing tourists into the old downtown where the new KTX station is located (Fig. 1). Finding a location for the new railway station was a controversial issue. The first plan proposed the station to be constructed in the outskirts of the city in order to save on the budget for construction. However, due to opposition from downtown merchants, the government finally decided to build the station in the city center. For that, part of the rails was constructed underground, at a higher cost.
In addition, new Olympic venues and residences are extensively developed in host cities based on different local conditions. In Turin, the historical and cultural amenities in the old center and the Olympic district in the new center were located within a 4 km radius. Olympic venues and public spaces were concentrated in two or three spots and intensely planned in abandoned industrial areas. In the case of Nagano, the Olympic venues, residences, and cultural amenities are loosely distributed within a 7 to 8 km radius, based on the low-density development in the city. In Gangneung, unlike Turin and Nagano where the Olympic venues spread over the city, the venues are planned in one place, a large Olympic park.
Long-Lasting Olympic Impact in a Spatial and Temporal Context
As seen in the cases of Winter Olympic host cities, despite its temporariness, mega event-led development influences the spatial structure of cities and the spillover effects are noticeable. Lauermann (2016b) argues that, although temporary projects such as Olympic bidding and hosting are focused on the short-term, the impact of these events are long-lasting. Though these events are often considered to be fixed and isolated, and to take place in a linear manner directed towards existing development agendas, they actually play a significant role in redefining and adjusting existing agendas in a non-linear manner. Lauermann (2016b) points out that Olympic bidding plans recursively adjust long-term development plans rather than being discarded after they fail to win hosting rights. Although Lauermann’s study focused on the impact of unrealized bidding plan, it implies how the preparation and hosting the Olympic event greatly influences city planning from a long-term perspective. Therefore, it is important to measure the long-lasting, spillover effects of mega-events than their short-term, direct impacts, since the former can impact a city’s economic sustainability in the long run.
Spatially, the influences go far beyond a few Olympic venue spots as it extends further into other parts of city, such as new transport modes, residential districts, public outdoor spaces, and existing cultural and commercial amenities. In order to quantitatively measure spillover effects of the Olympic, this study conducted longitudinal map survey for tourists’ visited places in six major attractions including Olympic park. Particularly, the study aims to see whether the visitors’ flows are expanded with new Olympic-related attractions after Olympic, as well as, to local traditional attractions during Olympic. For data analysis, this study utilized spatial network analysis. Beyond individual attraction, the network analysis can reveal how they are related and whether the new Olympic attractions and traditional ones are well connected. The study site is Gangneung, South Korea facing winter Olympic, therefore comprehensive to collect empirical data around Olympic period.
Mega-Event and Strategical Urban Development
Numerous researches (Table 1) have pointed out the contemporary tendency of the strategic integration of mega event planning with large urban developments such as infrastructure provision and regeneration in which the event has been regarded as a catalyst or justification for development projects (Bramwell, 1997; Burbank et al., 2002; Lei and Spaans, 2009; Müller, 2015). For instance, Salt Lake City and Atlanta considered Olympic preparation and planning as a part of a long-term development agenda (Burbank et al., 2002). In addition, Sheffield had taken an integrated approach connecting the game, event programming, and tourism for city revitalization (Branwell, 1997). Additionally, London strategically planned the Olympic Stadium to regenerate the derelict East end area (Stewart and Rayner, 2016).
However, researchers criticized that the current model of urban planning associated with mega-events is not sustainable. Gaffney (2013) maintained that a long-term agenda can be damaged by forcing a hectic planning cycle into changing city infrastructure, and it does consider the local stakeholders’ opinion. Stewart and Rayner (2016) pointed out the inconvenient truth that the commercial feasibility and post-management of the Olympic stadium was well known and predicated problems from the beginning based on their interview study.
Not many studies have suggested a specific solution. Müller (2015) suggested a decentralized model that opposed tying the event to large-scale development. Instead, it focused on spreading demand and distributing facilities spatially rather than constructing a permanent complex for peak demand. Some cities including Nagano, Turin, and Gangneung as introduced, have taken a decentralized and integrated approach, however, the impact of those planning attempts have not been evaluated.
While events occur temporarily at a special spot, tourism does not exist temporarily and exists before, during, and after the event (Lauermann, 2016a; Lauermann, 2016b). Since the organizations taking ownership of mega-events are not the same as those who are influenced by the long-term impact of the events, they are often not interested in creating long-lasting vitalization (Jago et al., 2010). Accordingly, the need of long-term research and monitoring as well as independent urban planning evaluation has been outlined (Bramwell, 1997; Gaffney, 2013).
Attraction Compatibility Theory and Multiple Attractions Visit
Wang and Jin (2019) maintained that a mega-event plays an important role in developing attractions; however, most event attraction marketing may be short-lived without proper planning and management. They emphasized sharing the economy, collaboration across stakeholders, and complementary distribution for exchange of mutual values. For strategies, they suggested the reduction of pressure to core spots, redistribution of tourists, and connecting existing local attractions (Wang and Jin, 2019).
In the tourism field, the term, cumulative attraction, has been used to indicate the ability of the attraction that can have an influence on an adjacent attraction (Lue et al., 1993). The cumulative attraction is closely associated with the compatibility concept in which if different attractions affect each other positively, they are considered compatible (Nelson, 1958). Hunt and Crompton (2008) emphasized the co-operative strategies and complementarity among attractions (Hunt and Crompton 2008). Several studies empirically proved the attraction compatibility theory based on multi-destination travel pattern analysis, and subsequently revealed influential factors such as spatial and program proximity (Wall, 1978; Lue et al., 1993; Oppermann, 1995; Fyall et al., 2001; Hunt and Crompton, 2008; Weidenfeld et al., 2010; Chen, 2019).
This study examined how the compatibility among existing local attractions and new facilities related to the Olympics has changed before, during, and after the Olympics. In addition, this study investigated the influential factors affecting the attraction compatibility such as travel type, transportation mode, and duration of stay.
Network Analysis Application to Tourism
Network analysts argue that causation or mechanism does not come from the individuals or attributes; however, they are derived from the structure or context where the individuals are embedded in (Marin and Berry, 2011). Unlike conventional analysis focusing on the individuals and their attributes, the network analysis focuses on their relationships, for example, the variables are explained not by their attributes, but by their relationships.
The network is composed of nodes and connecting edges. The results of network analysis are interpreted based on the centrality of nodes and the strength of the connecting edges (Granovetter, 1973; Czernek-Marszałek, 2018). The degree of centrality measures the engagement of the node in the network, calculating the number of nodes that a certain node is connected to or the total sum of the weight of the ties that are placed into the node. In addition, the strength of edges measures the number of ties or total sum of tie weights between two different nodes (Freeman, 1978; Bonacich, 1987; Borgatti et al., 2006).
Beyond individual attractions, this study aims to reveal how they are related and whether the new Olympic-related attractions are well connected to existing ones based on the visitor movement. As one of research methodologies that is being increasingly used, social network analysis has frequently been applied to spatial network research (Lee et al., 2013; Leung et al., 2016; Liu et al., 2017; Kang et al., 2018; Lee and Kim, 2018; Erlström et al., 2022). More recently, this has been applied to tourism research to reveal visitors’ movement pattern analysis (Liu et al. 2017; Lee and Kim 2018).
Attraction Network Changes and Influential Factors
Attraction network change can be a good indicator of the spatial spillover effect of mega-event development. It calculates how the core value of each attraction (e.g., the Olympic park) changes over time. In addition, it calculates how the connecting strength among attractions changes over time. Thus, attraction network is effective in showing how Olympic locations’ impact spills over and extends further into other areas. Few studies have examined attraction network among multiple destinations, which could also lead to new insights on the synergetic effect that mega-event infrastructure development can have in relation to other existing attractions (Lee et al. 2013; Lee and Kim 2018).
Extending from the investigation of network characteristics, this study took a more advanced approach by exploring the mechanism behind the attraction network formation. As mentioned in Liu et al. (2013), “the formation of attraction networks can be more explicable than it appears” (p. 134). There has been research on the factors that influence attraction visitation, such as seasonality (Hui and Yuen, 2002; Song et al., 2011), event host (Leung et a l., 2016; Rocha and Fink, 2017), travel settings (Woodside and Sherrell, 1977; Decrop, 2010), and tourists’ characteristics and background (Hill, 2000; Butler, 2001; Anwar and Sohail, 2004; Lau and McKercher, 2004; Lepp and Gibson, 2008; Mutinda and Mayaka, 2012; Caldeira and Kastenholz, 2015, 2017; Garcia-Palomares et al., 2015; Karl et al., 2015). For example, empirical evidence has shown seasonal patterns of tourism demand, and emphasizes the importance of comprehensive understanding and accurate predictions of seasonality (Song et al. 2011). Most visited attractions are found to become more diverse, for instance, the traditional core tourism area extends with newly developed Olympic related attractions (Leung et al., 2016). However, the research on factors affecting attraction network is still limited. Few recent studies started to investigate the influential factors affecting attraction networks, such as the various types of proximity in terms of physical locations, program types, evaluation status, and staying period (Jurowski et al., 2016; Liu et al., 2017; Kang et al., 2018; Abdelhalim, 2022). For example, location and tenure proximity were found to have positive relationships with attraction network, which means the attractions are closer each other or they are developed in similar period, stronger co-visitation connections are identified (Liu et al., 2017). In addition, the attraction visitation patterns were significantly differentiated depending on the length of stay (Kang et al., 2018).
By recognizing these precedent factors considered important, we chose the relevant touristic destination-related factors that can explain the attraction network formation for this study context, and are meaningful for providing attraction management and planning. Considering two types of destination management potentials for maximizing the synergetic effect, we chose spatial characteristic factors (location adjacency and contents similarity) and temporal type factors (mega-event and seasonality). Based on the factor types, the relevant analysis was introduced: 1. network density comparison analysis for proving significant differences among different periods (event host and seasonality) and 2. regression analysis for proving two relational concepts as predictors of the attraction network (location proximity and contents proximity).
Research Methods
Study Location
For this study, Gangneung, a 2018 winter Olympic host city, was selected. Gangneung is a city located on the east coast of South Korea, famous for its natural landscape, historical sites and adjacent fishing villages. There are popular sandy beaches and lakes, such as Kyeong-po beach, Anmok beach, and Kyeong-po lake, as well as sites of historical significance, such as the traditional Korean style house of a great scholarly family. In 2009, the International Olympic Committee (IOC) selected Pyeongchang as the host city for the 2018 Winter Olympics. While the snowing events took place in Pyeongchang, all indoor ice events took place in Gangneung, located 80 km (1.5 hour driving) from Pyeongchang.
In more detail, the specific area of this study is a total of 19 regions. They are classified into three types: nature- based tourist attractions such as Gyeongpo Beach, cultural and historical resource-based tourist attractions such as O-jukhun, and local amenities such as markets or parks. The site selection was based on the information in the ‘tourist attractions’ category of the Gangneung City Hall website. Furthermore, through the research team discussion, major regional amenities that were considered to have an impact on Olympics were added.
Data Collection and Respondents’ Profile
A city map survey with marks indicating the location of nineteen local attractions was developed and the participants were asked to check all the places they had visited and specify hours they had spent. Four student researchers participated in pre-training sessions and conducted 48 surveys during one and half year between September 2017 and January 2019. The participants were recruited from six different types of tourist attractions (O-Juk-heon complex, Kyong-po beach, Olympic Park, and Walwha, a commercial street), the express bus terminal, and the express railway station during six different seasons (Table 2). The survey dates are evenly distributed considering week, weekend and the four seasons.
In addition, the participants were asked to report their transportation method and staying period. Regarding personal factors, the participants were asked to answer questions about their gender, age, residence and frequency of visit. Using a convenient sidewalk sampling method, a total of 527 visitors participated in the survey. An average of 70 to 80 visitors participated each season, with the exception of the Olympic period, when 29% of the visitors were recruited (Table 2). Survey participants were provided informed consent and research project details and the survey materials were reviewed by Institutional Review Board at Gangneung-Wonju National University.
The respondents’ profile indicates that the participants’ gender was reasonably balanced: 56% male and 44% female. In terms of age, the majority of the respondents were between 20 and 30 years old (55%), the second largest group was between 40 and 50 (33%), and respondents of either less than 20 or more than 70 were around 11%. The residence profile showed that about half the participants were tourists from other Korean cities (49%). Local residents (35%) were the second largest group. Visitors from foreign countries were the third largest group, making up 17% of those surveyed. Most foreign participants (98%) were recruited during the Olympics. The most common form of transportation can be divided into three major groups: on foot, bicycle and public bus (29%), own car or taxi (46%) and mixed-use (25%).
In addition to participants’ background as described above, the visitation frequencies and duration of stay for each of the 19 regions of this study were collected in a subjective form so that they could be used in network analysis.
Network Analysis
Different types of network analysis were conducted. In the first phase, a descriptive analysis of the six sets of networks (i.e. eigenvector centrality) was conducted. The data were divided into six data sets based on six periods of data collection. For each season, a random seed function was applied to set same sample number of observations. The raw data matrix based on a 2-mode was transformed into a 1-mode data matrix. The eigenvector centrality of each node and the weights of each edge were estimated.
In the second phase, in order to investigate the temporal impact of the Olympic event and seasonality, the seven different networks were compared using Network density comparison analysis considering single and dual impacts of both event hosting and seasons. Utilizing UCINET, a bootstrap paired sampled t-test was conducted to test whether there is a significant difference between the two paired network data. All statistical significances were tested at a 95% significance level. The number of permutations for bootstrapping was set up to 10,000.
The last phase of the analysis included a regression analysis- a QAP (Quadratic Assignment Procedure). It is a useful approach for testing hypotheses regarding the relationships among multiple networks. In a study by Liu et al. (2017), QAP regression was reviewed and it was inferred that it could effectively avoid auto-correlation problems. According to their review, network analysis using OLS will result in biased estimators because the nodes in the unit of analysis respond with reference to one another, however, the QAP explicitly accepted the auto-correlated errors and can circumvent the problem. Like a traditional regression analysis, QAP calculates the influence significance of independent variables on the dependent variables; it also generates a pseudo-R2 that is analogous to the R2 in an ordinary least squares (OLS) regression. For this study, two hypothesized relationships between two destination characteristic factors and attraction network are tested as follows.
The first proposition is that location adjacency is positively related to a tourist attraction network. Spatial planning and zoning have been considered important in tourism or leisure planning literatures to guide tourist flows and demands (Inskeep, 1991; Gunn, 1997; Liu et al., 2017; Liu et al., 2020). In general, the attractions that are located close can be considered as a group by tourists. The second proposition is that program similarity is positively related to the attraction network. There is little research that explored the influence of similar themes on the attraction network (Jurowski et al., 2016; Liu et al. 2017).
To examine the predictability of destination factors including adjacency and program similarity, location and type proximity were set as independent variables. The attraction networks for six periods were set as dependent variables. It is composed of a valued 19 × 19 matrix generated based on visitations to multiple attractions. To apply OAP network analysis to the data sets, the cell values in the data sets needed to be transformed into binary data. For location adjacency and program similarity, if attraction A and B are located in same designated region or categorized as the same type, the values of a cell (x, y) in location and type matrix are 1; otherwise, the cell is given the value 0. For the attraction network, the mean of all cell values was selected as a cut-off and the cell values were reassigned with either the value 0 or 1 (Shih, 2006).
Results and Discussion
Changes in Centrality
Table 3 and 4 show eigenvector centralities calculated with two types of attraction network; one is based on the visitations to each attraction (Table 3), and the other is based on the visitors’ duration of stay at various attractions (Table 4 and Fig. 2).
During the pre-Olympic period, the results found that most attractions have a centrality value of more than 0.5, except for the Olympic Park (0.35–0.39). Kyungpo beach (1.0), Anmok beach (0.95–0.99), O-juk-hun (0.94–0.96), downtown (0.87–1.00), and Kyungpo lake (0.90–0.93) received centrality values more than 0.9. Subsequently, Gangmun beach (0.85–0.89), Jungang market (0.85–0.87), GWNU campus (0.78–0.86), Chodang village (0.75–0.83), Museums (0.70–0.80), Seonkyo house (0.76) and Songjung beach (0.71–0.74) displayed centrality values more than 0.7. Finally, Walwha Street (0.57–0.64), Gasiyeon (0.56–0.58), Namhangjin port (0.56), Namdae River (0.53–0.54) and the Olympic Park (0.35–0.39) reported centrality values less than 0.7.
Centrality results based on staying time show similar pattern, however, they tended to present lower centralities and bigger differences among the attractions. The attractions receiving more than 0.9 centrality values were less than the visitation network and includes Kyungpo beach (0.91–1.00), Anmok beach (0.92–0.93). Downtown (0.84–1.00) and Jungang market ( 0 .70–0 .73) received centrality values more than 0 .7. Gangmun beach (0.59–0.78), Kyungpo lake (0.52–0.75), O-jukhun (0.57–0.66) reported values less than 0.7 and more than 0.5. Lastly, Namhangjin port (0.36–0.37), Seonkyo
House (0.32–0.37), Dongbu market (0.33–0.35), Gasiyeon (0.21–0.39), Museums (0.31), Huhnansulheon park (0.21–0.35), Walwha Street (0.26–0.31), Namdae River (0.22–0.33), and Olympic park (0.11–0.23) received values less than 0.5.
During the Olympic period, the results were completely opposite. The calculated centrality of all the attractions decreased 0.3–0.4 on average, with the exception of the Olympic Park and downtown. The Olympic Park, initially in the lowest group, received highest value of 1 and the centrality of the downtown area increased (0.07) compared to the pre-Olympic period. Centrality results based on duration of stay showed that the centrality of most attraction range from 0.1–0.3 in average, except for the beach areas, such as Kyungpo beach (0.66), Anmok beach (0.58) and Gang-mun beach (0.54), downtown (0.86) and the local market (0.46).
For the post-Olympic period, the centrality of attractions in general tended to increase again and were restored to Pre-Olympic levels. More specifically, in the pre-Fall and post-Spring season comparison, there was an increase in centrality for about half of the attractions, including downtown and Dong-bu markets. However, there was a decrease for the remaining others, including O-jukhun and museums. In the first Winter after the Olympics, there was a tendency to decrease. When comparing pre-Olympic Winter data with post-Olympic Winter data, all centrality of attractions decreased, with the exception of the Olympic Park and Walwha Street, which showed an increase compared to the pre-Olympic period.
However, the centrality results based on duration hours showed different results. In a pre-Fall and post-Spring comparison, the centrality of most attractions increased, particularly the Olympic Park and Walwha Street showed a great increase. Similarly, in a pre- and post-Winter comparison, two-thirds of the attractions showed increased centrality values and some increases were noticeable, such as for the Olympic Park and Walwha Street. In the first Summer after the Olympics, the centrality of most attractions, including the beach area, increased and showed the highest values. In general, the results based on the duration hours showed that the centrality of most attractions tended to increase, barring a few exceptions, and the range of increase was larger than the centrality results based on visitations.
Connecting Edges’ Strength Changes
The attraction network has a total of 171 connecting edges between attraction nodes. The individual edge strength was measured based on the counts of co-visitation occurrences between the two nodes, and standardized to a value between 0–100. Both results based on visitations and duration of stay was graphically represented in the maps (Fig.s 3 and 4). The different widths of the yellow lines were based on the continuous values of strengths. Fig. 5 depicts the numerical results for the strengths of the connecting edge for three representative periods (before, during, and after the Olympics).
In general, the results correspond with that of degree centralities. During the pre-Olympic period, the results of the attraction network displayed complex patterns of connections centered on several attraction points that are traditionally popular (Fig. 3). Similarly, Fig. 5 presents the level of distribution of strengths, indicating a wider distribution than the Olympic period, and the differences between the weakest and strongest groups were noticeable (Fig. 5).
As the Winter Olympics were taking place mainly around the Olympic Park, the visitation network results displayed very strong connections among the Olympic park, the Northern coast, and the Southern city’s downtown. A few attractions such as Gang-mun beach and Kyeong-po lake in the Northern area, serve as minor connections to a major one. However, except the major connections, all other connecting strengths were presented with noticeably weaker strengths when compared to the pre-Olympic period (Fig. 3, 4, and 5).
During the first spring after the Olympics, the edge strengths were restored to levels similar to those of the pre-Olympic period. Compared to pre-Olympic period, the most noticeable change was the reduced gap between the strongest and weakest group, indicating that the edge strengths tend to distribute equally. Another change was that the strengths of connections to the Olympic Park became stronger, indicating that the Olympic legacy was included as one of the touristic attractions. The results of the following summer exhibited patterns similar to those in spring. The results of the following winter indicated that the connections among the Northern and Eastern coasts, the Western historical sites, and the Southern downtown area remained strong and connections to the downtown became more prominent with new powerful attraction points. However, the results should be interpreted based on the winter season impact.
Region and Type Proximity Correlation Analysis
In order to investigate the factors affecting attraction networks, hypothesized relationships were modeled and QAP analysis was conducted. The results of the regression include standardized coefficients of the independent variables and the R2 of the QAP regression models, as shown in Table 5. Both variables of location adjacency and program similarity were not significant factors in determining the attraction network. As a result of six hypothesized models’ testing, in most models, p-values were more than . 05, except in two seasons including the pre-Olympic Fall season (.042*) and the post-Olympic Summer season data (.046*). Table 5 reports that region proximity has no significant effect on the tourist attraction network (p > .05), indicating that there is no tendency that travelers move around the attractions adjacent to each other. Type proximity was found to have a positive effect on the tourist attraction network during pre-fall and post-summer, indicating that travelers tend to move around the attractions with similar themes, for instances, they may focus on touring the natural sandy beaches during the summer or they move around a few popular spots before Olympic. However, the adjusted R squares of .0069 and .005 indicate that only a small percentage of variance in the tourist attraction network can be explained by independent variables; only 0.5%–0.7% of the variance in the tourist attraction network can be explained by the program similarity.
Network Density Comparison
Similar to the classical paired sample t-test, network density comparison is effective when making statistical comparisons of the densities of two networks based on the same nodes. “Network density” represents the portion of actual connections from all the connections that can potentially exist between two nodes.
In this study, in order to investigate the influence of the Olympics as well as seasonality on the attraction network, the seven pairs of attraction networks were statistically compared based on the data collection periods (Table 6). To test the impact of the Olympics, two networks in the same season category were paired depending on whether they occurred before or after the Olympics. Fall and spring were considered as same season category. Similarly, to test the impact of the seasons, two networks from the period either before or after the Olympics, and from different seasons were paired.
Bootstrap approach results in which 5000 sub-samples are generated are reported in the output. The difference between mean density ranges from 4.40 to 20.94. The standard error of the difference in the classical method ranges from 0.58 to 0.66; the standard error by bootstrap estimate ranges from 1.88 to 2.14. In the bootstrap paired sample t-test, all the networks compared showed significant level of differences. That is, the observed differences would very rarely arise by chance in random samples drawn from these networks.
More specifically, the results revealed with great confidence that the density of attraction ties in the Summer and Spring is greater than the density of attraction ties in the Winter. This applied to both pre- and post-Olympic periods. In addition, the density of attraction ties during the post-Olympic period is greater than the density of attraction ties during the pre-Olympic period. This applied to the Winter comparison (pre: 9.0392, post: 13.7251) as well as to the pre-Olympic Fall (14.1637) and the post-Olympic Spring (18.9573) season comparison. The biggest density difference was reported when there is dual impact of Olympic and season. For example, the density of attraction ties in the Summer of post-Olympic period (30.0556) is greater than the density of attraction ties in the case of the pre-Olympic Winter period (9.1152). Therefore, both Olympic and seasonality were significant factors affecting attraction network density.
Conclusion
Research Implications
Although mega-events such as the Olympics are generally expected to have both short- and long-term impacts on hosting cities, very few studies have quantitatively proved those effects based on demand-oriented evidence and through long-term data collection (Gaffney, 2013; Garcia, 2017). This study focused on tourists’ visiting pattern changes after Olympic particularly the changes of the relations among attractions based on attraction compatibility theory. The study utilizes a network analysis and the results can be used for providing co-operative strategy and complementarity among attractions (Hunt and Crompton, 2008; Lue et al., 1993; Nelson, 1958) in Olympic host cities. This is especially the case for smaller cities where, although the government and local community eagerly anticipate a return of investment and durable benefits of Olympic hosting, there have been disappointing cases from a long-term perspective, such as Nagano and Turin. Therefore, there is an urgent need for base-line data, which empirically proves and quantitatively measures the temporal and spatial extent of Olympic impact.
This study examined changes in visitors’ movement pattern among major touristic attractions during a year and half period, including pre- and post-Olympic seasons in Gangneung, a 2018 Winter Olympics host city. At first glance, the results reveal that spillover effects of the Olympics have not been significant in either spatial or temporal terms from a visitors’ demand perspective. The core values of most attractions, with the exception of the Olympic Park and the downtown area, significantly decreased during the Olympics, which means that a large portion of visitors did not visit other local attractions during the event. The spatial extent of impact mainly concentrates on two spots, the KTX railway station and the Olympic Park. Similarly, in terms of connecting strengths, when compared to the pre-Olympic period, for which the connecting strengths were quite evenly distributed and presented a web-like pattern, those for the Olympic event were concentrated on a few relations that show as a simpler triangle-like pattern. The other connections became weaker. These results provide empirical evidence against the belief that during the mega-event visitors would stop at various local attractions besides the event venues.
However, the longitudinal observations revealed that there is a noticeable durable impact of hosting the Olympics. To some extent, Olympic hosting and the KTX railway construction seem to have changed tourist visitation patterns. More specifically, the study found that, although the popularity of most attractions remained the same or even tended to decrease after the Olympics in terms of visitation frequencies, the results regarding centrality considering duration of stay revealed that the lowest centrality group moved closer to the highest centrality group. The results indicate that the tourist visitation pattern was re-organized around a new downtown core. In addition, the gap between the core and minor group decreased, the popular attractions diversified, and minor local amenities group underwent an upward leveling. When comparing seasons, the core values of most attractions increased at least in one pair of seasons in comparison with the pre- and post-Olympic seasons. The range of increase was the largest in the local amenities group, including downtown, Walwha Street, and local markets located in downtown. This implicates that the new KTX express railway construction contributed to an increase in visitors’ attention to the downtown area, which may lead, to some extent, to an area-wide vitalization.
Although the results warn that the expectation of spillover and durable effect can be hasty, the following results imply that the redevelopment project induced by a mega event has brought the changes and reorganized the travel pattern in small city. Based on the empirical data, this study supports the governments’ efforts to integrate mega-event planning with urban redevelopment, as well as, connecting the event with tourism for city revitalization (Bramwell, 1997; Burbank et al., 2002; Lei and Spaans, 2009; Müller, 2015).
The results can also be interpreted in relation to the reduced travel time based on new transportation connections. There was a great increase in daytrips to Gangneung after express railway connection. This shows that travel patterns changed from needing to schedule multiple days of relaxation in a coastal area to short-term travel dropping off at trending places downtown and to short-stay visits to the coast. On the other hand, the increased centrality values implicate that the Olympics may contribute to a diversification and expansion of attraction cores from more traditionally popular coastal areas to local hotspots and cultural or historical local sites. Geographically, attraction network expands from the North and Eastern coastal area into the Southern downtown.
The findings of this study provide important implications for destination management, aiming at generating a synergetic effect based on the relationship among attractions. First, the study results suggest the influence of attractions and the potential of co-marketing with the connected attractions. The analyzed attraction network informs the critical players (i.e. Olympic park, coastal area, KTX station) and supportive assistants in network formation, based on which destination management organizations and tourism department in local government or tourism development corporation can develop effective management strategies.
The study also suggests the need of integrated event planning with an urban redevelopment agenda for host cities. Rather than the sole, immediate impact of event, the impact of urban development facilitated by event hosting can be more critical such as transportation upgrade and downtown renovation. Taken together, in order to increase the sustainability of Olympic-based urban development, it is desirable to establish a balanced development in conjunction with urban development of various scales, such as regional urban master plans or urban regeneration plans. It is also necessary to select major and sub-areas of urban development to promote mid- to long term development rather than developing tourist destinations based on past event areas. Furthermore, as observed in the Turin and Nagano cases that are suffering from underused Olympic venues, a separate plan for the post-management of the Olympic stadium and secure financial resources for maintenance are another crucial issues that need to be addressed from a policy point of view.
Network density comparison results confirmed the centrality results and, more specifically, identified and differentiated the impact of the Olympics and those of different seasons. This result confirms previous studies proving the seasonality and event host impacts on tourist movement pattern (Hui and Yuen, 2002; Leung et al., 2016; Song et al., 2011). In correlation results, over six periods, there were no evidences that the attraction network is influenced by location adjacency. This result is incongruent with the previous findings that proved the positive effect of region proximity on tourist attraction networks or the influence of significant distance among attractions on a multi-attraction trip (Hwang and Fesenmaier, 2003; Liu et al., 2013). Program similarity was also identified to have no significant impact on the attraction network. This result is congruent with previous research findings that visitors may move around attractions with some variety (Jurowski et al., 2016; Liu et al., 2017).
Finally, the study suggests that complementing a scarce program can be an effective strategy when planning attraction upgrades in preparation of an event, for instance, improving the facilities and access to local cultural and amenity programs to Gangneung where natural sandy beach attractions were traditionally competent as a summer vacation city. This can reduce the seasonal gap in tourist inflow, thus expanding summer-oriented contents to a four-season basis. Event hosting can be an important opportunity to make up for a lack of contents when considering the results that the impact of seasonal gap was even larger than the impact of the event hosts.
The quantitative results of this study showed the extent of the spatiotemporal impact of hosting the Olympics on a smaller city. During the event, the influence of the Olympics in a spatial context was complete opposite to the general expectation. However, in seasons after the Olympics, this impact was perceptible in the intensification of attractions’ core values and in the diversification of the core attractions. For instance, the visitors’ focus expanded from natural and historical attractions to local amenities. This result is consistent with an earlier study, in which it is argued that a sustainable process of city image transformation induced by culture-led regeneration, such as what is enabled through Olympic hosting, needs long-term based evidences rather than short-term evidences during the event (Garcia, 2017). This also provides evidence that mega-event impact sometimes lags behind the city government and local community’s schedule, which explains the need for a long-term evaluation.
Limitations and Future Research
One of the limitations of this study is its demand-oriented approach to Olympic impact. The indirect, long-lasting impact of mega-events can, of course, also be studied based on other aspects, such as changes in real estate value, local business profits, and amenity enhancement. (Müller 2014; Miyoshi and Sasaki 2016). In addition, this study has not considered changes in size in the local economy nor the scale of changes in other benefits. Particularly, diversified attractions may be profitable to local businesses, however, we cannot say with certainty whether the overall “profit pie” has grown or whether it has just been more fairly distributed. A second limitation is related to the fact that the study did not differentiate between the impact of the event and that of redevelopment projects motivated by the Olympics. Although this study emphasizes the synergetic impact when integrating event hosting with urban redevelopment, understanding the single impact from event programming would be beneficial for cities planning to host international events and invest money, however have no plan for either transportation upgrade or further redevelopment. In addition, this study has a limitation in that it has not been able to draw more meaningful implications by using information such as survey respondents’ gender, age, frequency of visit, and means of transportation. Finally, further studies on the Olympic impact on attraction network based on a longer timeline would likely provide more comprehensive conclusions.
PDF Links
PubReader
ePub Link
Full text via DOI
Download Citation
