74 MODELLING HIGH SPEED RAIL IMPACT ON TOURISTS’ DESTINATION CHOICES – Contributions by Francesca Pagliara
Introduction
The increase of accessibility to a given tourist destination, thanks to new interventions in the transportation system, represents an important factor for tourism development. The principle is that suitable means of transport can change the image of dead centres of tourist interest by transforming them into active places for visitors.
With the advent of new technologies, the transportation system, has experienced a change during the centuries. Specifically, rail, thanks to the High Speed Rail (HSR) deployment, is one of sectors mainly affected by their introduction. Several are the projects already realized, others are under construction, and more are in the pipeline all over the globe.
Europe is among the most visited continents in the world, and it is normal to consider that HSR projects are likely to affect tourist’s choices of a given tourist destination. HSR systems were not conceived at the beginning as a real alternative transport mode for tourists, their impact on tourism market came only later. In general, a decrease in travel time, brought by HSR, is perceived by tourists as a good motivation for choosing a destination connected with this service.
Several are the approaches able to model the impact of HSR on tourists’ behaviour. In this chapter the main ones are reported.
Discrete choice analysis : disaggregate approach
Discrete Choice Analysis (DCA) is generally used for measuring and predicting individuals’ preferences and alternatives ’choices, providing quantitative measures of the impact of different attributes related to tourism destinations, products, or services, also considering tourists’ Willingness To Pay (WTP) for different services.
DCA has found applications in different fields of research including transportation, marketing, retailing, health, and environmental economics, since it can well explain and predict preferences and individuals’ choice behaviour. Examples of DCA models in the tourism context include, for example, estimating how attractive tourists find competing destinations with respect to their attributes, on what premises they choose a transport mode for their trip or with whom they decide to travel. The fundamental concept underlying the methods is the Random Utility Model (RUM) theory, according to which a tourist chooses an alternative destination maximizing his/her utility. The utility is made of several attributes related to the tourists as well as to the destination, such as the transport accessibility. The alternative mode HSR systems can be considered as one of these attributes.
In Delaplace et al. (2014), two case studies were compared, i.e two capitals, the one of Rome in Italy (Valeri et al., 2012) and that of Paris in France. Through DCA, the main outcome of this work was that several factors influenced tourists’ choices of these destinations, like the presence of architectural sites, the quality of promotion of the destination itself, the presence of events, etc. Moreover, HSR system affected the choice of Paris and Rome differently. The two cities belong to two different countries in which the history of HSR service is very different. Indeed, in France, TGV (Train à Grande Vitesse – High Speed Train) is considered a real transport mode alternative, while in Italy it is a relatively new system which still needs a campaign of promotion to be well accepted among the tourists. The main motivation of this could also be also justified in the high travel costs.
In Pagliara (2015) the same approach was followed for the case study of the city of Naples, in the south of Italy. From the model estimation it resulted that several were the factors influencing the choice of this city as a tourist destination, but the presence of HSR connecting Naples with other Italian cities, played its role.
In Pagliara et al. (2015a) and Pagliara et al. (2015b) a comparison between two key tourist destinations in Europe, namely Paris and Madrid, was made to identify the factors influencing tourists’ choice of these capitals. Based on two surveys, it was found that the presence of architectural sites, the quality of promotion of the destination itself, and cultural and social events had an impact on tourists’ choices. However, the availability of the HSR systems affected the choice of Paris and Madrid as tourist destinations in a different way. For Paris, TGV was considered a real transport mode alternative among tourists. On the other hand, Madrid was chosen by tourists irrespective of the presence of an efficient HSR network. Data collected from the two surveys were used for a further quantitative analysis aimed at the identification of the factors influencing holidaymakers to revisit Paris and Madrid and visit other tourist places accessible by HSR from these capitals. The analysis demonstrated that HSR was chosen to visit places close to Madrid, such as Toledo as well as places close to Paris.
Delaplace et al. (2016) identified the extent and the nature of the relationship between HSR and destination choice in the case of theme parks. Theme parks are tourist specific places that can be considered as “stay tourism”, (in contrast with “circuit tourism”) that is, a kind of tourism with a unique motivation and in a given area. It was expected that the link between a given HSR station and a theme park is stronger when the HSR station was closer to tourist facilities and even stronger when it was conceived for the tourist structure itself. However, the question was also to investigate if tourists coming to these theme parks by HSR were also visiting other nearby places or places connected by HSR. Two theme parks, Disneyland Paris and Futuroscope Park that are both served by an HSR station, Marne-la- Vallée-Chessy and Futuroscope TGV respectively, were considered as case studies. The main outcome was that only few tourists visited Futuroscope by HSR and they would have come without HSR. Moreover, the tourists visiting other places near Futuroscope did not choose HSR. The link between Futuroscope and more generally tourism in the region and HSR was not very significant. Nevertheless, Futuroscope TGV station could also be useful for the Futuroscope technopole. Concerning Disney, the link was more significant. Tourists coming by HSR were numerous and would not have come without HSR. Nevertheless, the diffusion process was not linked to HSR. The probability of visiting other places was linked to RER (Réseau Express Régional – Regional Express Network) and to the proximity of Paris which allowed visiting different places. The surveys highlighted that the accessibility to HSR had not always had an impact on the theme park destination choice.
Aggregate approaches
Parametric approach: Generalized Estimating Equation (GEE)
Pagliara et al. (2017) attempted, through the case study of the High Speed/High Capacity Rail project in Italy, to evaluate the expected impacts on the tourism market, considering the Italian visitors and analysing their diﬀerent behaviour. The speciﬁcation of a panel model, simultaneously considering the eﬀects of HSR on the number of visitors and the number of nights spent at destination in all the Italian cities served by a HSR line, was proposed.
Specifically, an empirical analysis was carried out with the aid of a database containing information both on tourism and transport for the Italian municipalities, during a predefined time-period.
In this study the dependent variables took only non-negative integer values, the statistical treatment diﬀered from that of the normally distributed one, which could assume any real value, positive or negative, integer or fractional. Count data can be modelled by using diﬀerent methods, the most popular is the Poisson distribution, which is applied to a wide range of transportation count data contexts. Panel model analysis provides a general, ﬂexible approach in these contexts, since it allows modelling a wide variety of correlation patterns.
To consider these possible unknown correlations, Generalized Estimating Equations (GEEs) were considered. The main conclusion of this analysis was that HSR had a positive impact on local tourism, but a denser HSR network would have increased signiﬁcantly the potential of HSR obtaining not only a local but also a global impact.
Pagliara et al. (2021) used the same GEEs model specification but applied to the case study of China. In this work, two GEEs models were estimated one for the Chinese tourists and another one for the foreign tourists. Variables such as Resorts, Gross Domestic Product and Passengers had a positive impact on the number of domestic tourists. It was clear that Chinese tourists were strongly influenced, in the choice of destinations, both by the presence of international airports and by the presence of HSR connections. The second GEE model estimated considered as dependent variable the number of foreign tourists. Comparing the two models, it was possible to note that foreign tourists preferred as a tourist destination one in which there were many hotels, which were not so fundamental for domestic tourists. The presence of HSR stations was more significant in the second model, implying that the presence of a HSR network had a higher impact on the choices of foreign tourists compared to those of the domestic ones. The presence of international airports seemed to have a higher impact on Chinese tourists than on the foreign ones.
Non-parametric approach: Classification and Regression Tree (CART)
The application of the non-parametric Classification and Regression Tree (CART) model does not require an a priori probabilistic knowledge of the phenomenon under study and the fulfilment of strict hypotheses, neither on the type of relationship, nor on the form of distribution of the dependent variable. These aspects represent the main advantages over parametric techniques.
Each node of the tree indicates the predicted value, the number of experimental units contained in the node and its descriptive percentage. Very few are the contributions in the literature applying the CART methodology in the context of tourism.
Another advantage is that the CART analysis can effectively handle collinearity problems. When a serious correlation between independent variables exists, the variability of the estimated coefficients will be inflated. It follows that an interpretation of the relationship between independent and dependent variable is difficult to define. On the other hand, regression tree methods are also not sensitive to outliers since the splitting is based on the samples proportion within the split ranges and not on the absolute values. From the applied perspective, the regression tree methods are very intuitive and easy to explain. Moreover, they have the advantage of giving each variable the chance of appearing in different contexts with different covariates, and thus better reflecting its potential impact on the dependent variable. However, unlike a linear regression model, a variable in the CART algorithm can be considered highly important even if it never appears as a node split.
Two case studies applied this methodology with the objective of evaluating the impact of HSR on tourist’ choices, one was the analysis of the effect of HSR in the case study of Italy (Pagliara et al., 2020) and the other one in the case study of China (Pagliara et al. 2021).
Geographically Weighted Poisson Regression (GWPR)
Previous research studies have considered various empirical and methodological aspects of modelling the eﬀect of HSR on tourists’ behaviour. However, they have been unable to consider the presence of both the spatial autocorrelation and the unobserved heterogeneity. Speciﬁcally, the relationship between the dependent variable and any independent variable is assumed to be stationary over space. Due to this aspect, the use of an overall ﬁxed relationship between the dependent variable and the explanatory variables, for the case study under analysis, can aﬀect the adjustments of the estimates of the individual destination, generating instability in the models’ coeﬃcients.
The spatial heterogeneity within local models, such as Geographically Weighted (GW) models, provides a better platform allowing exploring the diﬀerent spatial relationships between HSR and tourism. Considering that global regression provides estimates with a low degree of accuracy in some areas, it might be more useful to specify local regression models, since they are able to incorporate spatial relations among variables for a given study area.
The Geographically Weighted Regression (GWR) approach was proposed in the geography literature to allow relationships in a regression model to vary over space. In contrast to the traditional linear regression models, where the regression coeﬃcients are constant over space, the regression coeﬃcients in this case are estimated locally at spatially referenced data points with GWR. GWR was proposed originally as an extension to the ordinary regression model to predict a continuous variable with a Gaussian (normal) error. However, when predicting a non-continuous variable, the dependent variables are count data with discrete and non-negative integer values, it is more appropriate to use diﬀerent regression models. In the literature, the Generalized Linear Models (GLMs) are considered the most suitable ones to determine the relationship between count data and the dependent variables. GLMs aim at extending ordinary regression models to encompass non-normal response distributions. Then, Geographically Weighted Generalised Linear model (GWGL) was developed by integrating the GLM and the GWR ones and extending the concept of the Geographical Weighted Regression (GWR) models in the context of the Generalized Linear Models (GLM). Given that the dependent variables are count data with discrete and non-negative integer values, GWR models were performed by using the Poisson distribution error. The Geographically Weighted Poisson Regression (GWPR) approach proposed with the objective of capturing the heterogeneity of the independent variables, with respect to each tourist destination.
The spatial heterogeneity within local models, such as Geographically Weighted (GW) models, provides a better platform allowing exploring the diﬀerent spatial relationships between HSR and tourism. In Pagliara and Mauriello (2020), a spatio-temporal analysis was proposed to evaluate the variables aﬀecting tourists ‘choices and speciﬁcally the impact of HSR on both the Italian and foreign tourists. The GWPR modelling approach has been proposed, which considers the problem of the temporal and spatial autocorrelation in a diﬀerent way with respect to the Generalized Estimating Equations method. Indeed, the results of this study supported the use of the GWPR as a promising tool for tourism planning, especially because it made possible to model non-stationary spatially counting data. The diﬀerent behaviours of Italian and foreign tourists was reported, considering the impact of the presence of a HSR service on them.
Other approaches
Tourism Spatial Interactions (TSI) and Coefficients of Variation (CV)
City’s size appears to be an important determinant of the impacts of HSR on tourism. A proof of this assertion needs to use a region with cities of different sizes as an analysis case. Case studies have been carried out in France, Spain, and China; however, none of them have examined the impact of an HSR network that connects cites of all different sizes in an area on the tourism spatial structure. An HSR network that connects multilevel cities in one area with several lines and nodes fosters cooperation and competition among them and makes the area a destination network. Such HSR and destination networks will be found in more countries in the future with the development of HSR projects. For this reason, understanding the impacts of HSR networks on a destination network with multilevel cities is very important both theoretically and practically.
Yin et al. (2019) contributed to the existing literature on HSR and tourism management by filling this gap and developing an analysis of multilevel cities in one area with an HSR network. Specifically, it investigated HSR’s influence on the tourism spatial structure of the Capital area of China through the specification of a spatial interaction model. The Capital area of China was chosen because of the different sizes of the cities, and because some of them are served by HSR and others are not. Specifically, this study attempted to understand the impact that HSR could have on cities of different sizes. The spatial interaction model was chosen for two main reasons. First, distance (the actual distance or travel time) was an important independent variable in the model. HSR would change the travel time in a region, and the model could examine the HSR network’s influence. Secondly, a city’s size was also an independent variable in the model specified and could be measured by population. The study’s objective was to determine whether an HSR network exerted different influences on cities of different sizes in one region. The differences in spatial interaction between two scenarios were considered, i.e., one with an existing HSR network and the other with a planned HSR network.
A Tourism spatial interaction (TSI) model was specified together with a Coefficient of Variation (CV), which was used to compare the changes in tourism at the regional level in the two scenarios.
Fixed-effect econometric model
The effects of HSR on cultural tourism are a subject of interest though the scarcity of bibliography shows that sufficient attention has not been paid in this regard up to now. Although some analytic studies using the econometric models have already emerged dealing with the influence of HSR on tourism, some of them even with a meritorious high level of disaggregation, the specific impact on the cultural tourism, and, more specifically, on museums and monuments, remain unexplored up to now. Hence, the higher returns and greater positive effects on society by the tourist industry, and in an outstanding way by cultural tourism, make it an appropriate field of research that should receive special attention.
The work by Campa et al. (2019) provides a deeper insight into this interesting subject for the first time by implementing an econometric model and taking into consideration the number of tourists of 64 museums and monuments in 25 Spanish municipalities, using a validated methodology which could be suitable in other countries with a HSR network and cultural attractiveness. Although the previous studies focused on general tourism in Spain, little or no influence of HSR was detected, the results in this paper showed signs that this mode of transportation could play a positive role in the reinforcement of cultural tourism. This contribution also highlighted the different roles that HSR played in the Spanish cultural tourist markets of museums and monuments, with a significant increase in the number of tourists in some regions, while these outcomes were not significant in others. A doubly controversial effect of distance to the HSR stations was also observed.
While little or no effect was detected in museums located in the same municipality, museums located in a different municipality received an appreciable significant increase in the number of tourists. This was interpreted as an indication that HSR increased the action radius of tourists surely due to the gain in available time at a destination and reinforced hidden potentialities of further museums. Moreover,
the central position of the HSR stations was also detected as a significant beneficial factor and suggested that the location of the HSR with respect to the city was also an important factor in addition to the mere connection to the HSR network.
Conclusions
Over these 10 years, we have tried to contribute to answering the following questions:
Does High Speed Rail have an impact on tourists’ travel choices?
Transport accessibility is fundamental for tourists’ destination choice, HSR systems have proven to play their role
What are the modelling approaches able to analyse this impact?
Several are the methodologies which have been tested and each of them has played a significant role with respect to the available data set, aggregate vs disaggregate information.
What remains?
Other methodologies deserve to be tested. An example is the application of Geographical and Temporal Weighted Regression (GTWR) type models in this context to consider also the local effects from the temporal point of view.
Written by Francesca Pagliara, Department of Civil, Architectural and Environmental Engineering, University of Naples Federico II, Naples, Italy
Read Francesca’s letter to future generations of tourism researchers
References
Campa, J. L., Pagliara, F., Lopez-Lambas, M.E., Arce, R., & Guirao, B. (2019). Impact of High-Speed Rail on Cultural Tourism Development: the Experience of the Spanish Museums and Monuments, Sustainability, 11(20), 5845.
Delaplace, M., Pagliara, F., Perrin, J., & Mermet, S. (2014). Can High Speed Rail foster the choice of destination for tourism purpose? Procedia Social and Behavioral Sciences, 111, 166-175.
Delaplace, M., Pagliara, F., & La Pietra, A. (2016). Does high-speed rail affect destination choice for tourism purpose? Disneyland Paris and Futuroscope case studies, Belgeo, 3, 1-23.
Pagliara, F. (2015). High-Speed Rail Systems and Tourists’ Destination Choice: The Case Study of Naples, Journal of Traffic and Transportation Engineering.
Pagliara, F., Delaplace, M., & Vassallo, J. M. (2015a). High Speed Rail systems and tourist’destination choice: the case studies of Paris and Madrid, International Journal of Sustainable Development & Planning, 10(3), 399–410.
Pagliara, F., La Pietra, A., Gomez, J., & Vassallo, J. M. (2015b). High Speed Rail and the tourism market: Evidence from the Madrid case study, Transport Policy, 37, 187-194.
Pagliara, F., Mauriello, F., & Garofalo, A. (2017). Exploring the interdependences between High Speed Rail systems and tourism: Some evidence from Italy, Transportation Research Part A: Policy and Practice, 106, 300-308.
Pagliara, F., & Mauriello, F. (2020). Modelling the impact of High Speed Rail on tourists with Geographically Weighted Poisson Regression, Transportation Research Part A: Policy and Practice, 132, 780-790. ISSN: 0965-8564.
Pagliara, F., Mauriello, F., & Russo, L. (2020). A Regression Tree Approach for Investigating the Impact of High Speed Rail on Tourists’ Choices, Sustainability, 12, 910.
Pagliara, F., Mauriello, F., & Yin, P. (2021). Analyzing the Impact of High-Speed Rail on Tourism with Parametric and Non-Parametric Methods: The Case Study, Sustainability, 13, 3416.
Valeri, E., Pagliara, F., & Marcucci, E. (2012, July). A destination choice model for tourism purpose. Paper presented at the ASRDLF 2012 conference special session on High Speed Rail, Tourism and Territories, Belfort, France.
Yin, P., Pagliara, F., & Wilson, A. (2019). How Does High-Speed Rail Affect Tourism? A Case Study of the Capital Region of China, Sustainability, 11(2), 472.