The impact of e-commerce on the tourist purchase decision : an empirical analysis

DEPARTAMENTO DE ESTADÍSTICA, ESTRUCTURA ECONÓMICA Y O.E.I. Plaza de la Victoria, 2 28802 Alcalá de Henares (Madrid) Teléfono: 91 885 42 01 http://www.uah.es/centros_departamentos/departamentos Alcamentos Departamento de Estadística, Estructura y O.E.I. 0801 THE IMPACT OF E-COMMERCE ON THE TOURIST PURCHASE DECISION: AN EMPIRICAL ANALYSIS Juan Muro, Cristina Suárez y María del Mar Zamora Universidad de Alcalá y Alcamétrica 1 DEPARTAMENTO DE ESTADÍSTICA, ESTRUCTURA ECONÓMICA Y O.E.I. Plaza de la Victoria, 2 28802 Alcalá de Henares (Madrid) Teléfonos: 91 885 42 01 http://www.uah.es/centros_departamentos/departamentos Alcamentos Nº: 0801 THE IMPACT OF E-COMMERCE ON THE TOURIST PURCHASE DECISION: AN EMPIRICAL ANALYSIS Juan Muro, Cristina Suárez y María del Mar Zamora Universidad de Alcalá y Alcamétrica 2 THE IMPACT OF E-COMMERCE ON THE TOURIST PURCHASE DECISION: AN EMPIRICAL ANALYSIS Juan Muro Cristina Suárez* María del Mar Zamora1 Universidad de Alcalá and Alcamétrica January, 2007 Abstract The development in the tourist industry linked with the rapid growth in e-commerce has put in evidence the existence of a new customer. We empirically investigate the microeconomic determinants of the internet purchased tourist goods. We adopt a reduced form demand for online goods model, extended to incorporate possible selectivity biases stemming from interactions between unobserved individual heterogeneity associated with specific internet use choice. The model is estimated using a very rich dataset from EGATUR (Encuesta de Gasto Turístico), the Spanish Foreign Tourist Expenditure Survey. The sample allows us to explore the influence of price and income related variables as well as personal characteristics on internet purchased goods. Price and income results are consistent with theory. Unobserved individual heterogeneity linked with the use of the internet is significantly correlated to unobserved individual heterogeneity related to online purchases. Keywords: e-Commerce, Tourism, Binary choice model with selectivity JEL classification: C25, L83 *Corresponding autor. Facultad de Ciencias Económicas y Empresariales Universidad de Alcalá Plaza de la Victoria, 2 28802 Alcalá de Henares, Madrid (SPAIN) e-mail: cri[email protected] 1 We would like to thank participants at ATMC held in Valencia (2007) for the highly constructive comments received. Errors remain our sole responsibility. 3 THE IMPACT OF E-COMMERCE ON THE TOURIST PURCHASE DECISION: AN EMPIRICAL ANALYSIS Juan Muro, Cristina Suárez and María del Mar Zamora 1. Introduction The Internet phenomenon is changing people’s habits in developed countries and many socio-economic studies are currently being developed related to this event, in particular with the on-line retail sales of goods and services (for example OECD (1999)). In general, the Internet could be deemed as an information system and also as an electronic market place, so, these characteristics allow the Internet to be considered as an intermediary between buyers and sellers to exchange information about prices and product offerings. One of the Internet markets that has been developing towards higher levels of sales is the online travel-tourist market which has increased by as much as 34% from 2004 to 2005 (Marcussen (2006)). This market has special characteristics for both buyers and sellers. One of the main features of the tourist product is its intangible nature when purchased; it is merely a piece of information stored in a reservations system, subsequently buyers do not require sizable investments in organizational transformations. These electronic market systems reduce the search costs that buyers must pay to obtain information about the prices and product offerings available in the market, some authors, as Combes and Patel (1997), described the customer environment for Internet-based travel services as allocation where consumers could be compared with ease and, also, they can inquire about various aspects of a travel destination without having to speak to a travel agent or they can quickly and simply find the lowest fare anywhere. In addition 4 to their ability to reduce search costs, if we take into account the fact that the tourist is normally not able to try the product until the moment agreed, we can draw the conclusion that the tourist product fits perfectly with the new technologies. The use of the Internet with a tourism purpose is twofold: it can act as a promotional tool or it can be used focusing on its capacity to do e-trade. At first, the promotional application was the reason why the tourist became so interested in the Internet. The second reason was ecommerce, which should be understood as the reservations and/or shopping for tourist products. Given that the relatively new phenomenon of e-commerce has important repercussions for tourist travel decisions, the aim of this paper is to analyse the microeconomic determinants of the tourists’ purchase choices for foreign tourism arriving in Spain. The empirical literature on this subject is very scarce. It is limited to some mainly descriptive papers, for Spain consult, for example, IBIT (2001), or for others on very specific questions not related to the subject of tourism consult, for example, Goolsbee (2000) and Alm and Melnik (2005). The purpose of the article focuses on the influence that e-commerce is having on the tourist sector by assuming that they compare the stochastic utility of several alternatives and select the one that maximizes their utility. Also, this paper demonstrates how access to the Internet with a tourism purpose is important in order to buy online tourist travel products or services. To do so, we use a probit model with sample selection in which the probability of the tourist e-commerce choice is conditional to the access to the Internet with a tourism purpose. 5 The model has been estimated with Spanish data on foreign tourism. We utilize the 2004 wave of a very rich database coming from EGATUR (Encuesta de Gasto Turístico) the Spanish Foreign Tourist Expenditure Survey. The survey is a questionnaire answered by more than a sixty thousand foreign tourists visiting Spain and it requests information on tourists’ socioeconomic characteristics, attributes of the trip and other relevant variables including the e-commerce choice. The Egatur sample does not have problems of selection bias because the data include all types of tourists arriving in Spain and not only those that use the Internet. The goal in this article is to analyze a new phenomenon, the impact of e-commerce on the tourist purchase decision for foreign tourists visiting Spain. The next section reviews a conceptual framework that explains the tourists’ choice of commerce mode in section 2. This is followed by a description of the database and we also present, in section 3, the empirical results and analyse the main determinants of e-commerce choice. Finally, in the last section, we sum up with our main conclusions of the impact of e-commerce on the tourist purchase decision experience in Spain. 2. The Model Online commerce presents Internet users with another method for purchasing goods. Almost all goods traded online can also be purchased in traditional commerce. In this respect, the Internet presents simply another venue for purchasing the same goods, and hence Internetpurchased goods can be considered as perfect substitutes to some goods purchased in traditional commerce. 6 We can therefore structure the consumer decision to purchase goods online in the following way. First, we assume that the utility function of the representative tourist is U=U(q1, …, qk, z1, …, zn, d1, …, dr), where q=(q1, …, qk) represents the vector of goods that can be purchased preferably in traditional commerce, for example restaurant meals; z =(z1, …, zn) denotes consumer goods that can be purchased in online commerce and in traditional commerce, where they are perfect substitutes, for example, hotel beds; and finally, d = (d1, …, dr) represents a good that can be purchased preferably in online commerce, for example, low-cost airlines. The consumer balance will be reduced to: Max U= U(q1, …, qk, z1, …, zn, d1, …, dr) Subject to: pq q + pz z + pd d = Y where pq, pz and pd are the vectors of prices, and Y represents the income level. In this setting each tourist is assumed to have to choose between tourist goods that can be purchased in online commerce and in traditional commerce. Due to the cross-sectional nature of our database we assume a myopic behaviour. For any given tourist, defined by means of individual observed characteristics, his/her utility is derived from a number of observed goods attributes and travel features and a set of unobservable ones. 7 The probability that a tourist i will choose to buy online equals the probability associated with a positive difference in the comparisons between the utility derived from buying online and the utility related to traditional commerce. The difference between the online commerce and the traditional commerce can be represented as an unobserved latent variable Yi*. So Yi* = Xi’ β + ui, [1] such that one observes only the binary outcome, Yi = 1 if Yi*> 0 and Yi = 0 if Yi* ≤ 0. However, one only observes Yi for observation i if the tourist has decided to obtain access to the Internet (Ci =1), where Ci* follows Ci* = Zi’ γ + ε i, [2] where Ci = 1 if Ci*> 0 and Ci = 0 if Ci* ≤ 0. 8 Xi and Zi are variable vectors of individual characteristics that can be common or not in both specifications [1] and [2]. ui and ε i are the error terms for equations [1] and [2], respectively, distributed as bivariate normal with mean zero, unit variance, and ρ = Corr(ui, ε i). After controlling by observables our model allows for correlation between unobservables in equations [1] and [2]. As is well known, when ρ ≠ 0, standard probit techniques applied to equation [1] yield biased results, and the probit model with sample selection provides consistent, asymptotically efficient estimates for all the parameters in such models. 3. Empirical analysis In this section we present the empirical results of the analysis proposed in the last section that can be summarised with the following equations: the selection equation which is related to internet access with a tourism purpose (equation [2]) and the main equation which is related to online commerce and is only observed if internet access exists (equation [1]). These equations are estimated simultaneously according to maximum likelihood and the method is adapted from the article by Van de Ven and Van Pragg (1981), in which both equations have binary dependent variables. All specifications incorporate a group of common variables included in Xi and Zi which are related to characteristics that can influence tourist purchase choices and the possibility to undertake certain activities and they are common in both decisions (Internet access versus no Internet access, and online commerce versus traditional commerce). These variables are related 15 It may be assumed that the cost of searching for new alternatives is generally too high and the expected gains associated with new alternatives too uncertain. If we analyse tourists with ten or more visits to Spain, we can observe that they prefer to use e-commerce with a marginal effect of 8.1% on the probability of buying online. Also evidenced in Table 2 is the fact that the use of a low cost airline to come to Spain increases the probability of e-commerce. This is one of the most important characteristics of this type of company, which prefers direct access to a consumer only through call centres and the Internet and also it is important to remark that the type of service most demanded in the online market is air travel with 56% of the demand. The shorter the length of the stay the greater the probability of buying online, with positive means marginal effects ranging from 8.8% (between 1 and 3 days) to 5.3% (between 4 and 7 days). 4. Conclusions The impact of the Internet in activities related to tourism has seen a significant growth in the last years. Given that online commerce has important repercussions for tourist decisions, the aim of this paper is to analyse the microeconomic determinants of the tourists’ e-commerce choices for foreign tourism arriving to Spain in 2004. 16 The econometric analysis employed to obtain these results used a probit model with sample selection. This model is necessary for the aim of controlling e-commerce effects for the likelihood of Internet use. We came to this conclusion showing the statistical significance of the correlation coefficient. Using a probit model with sample selection we have estimated the probability of e-commerce choice, for the tourists that use Internet with a tourism purpose, as opposite to buying tourist products using traditional commerce. Our results allow us to define characteristics influencing e-commerce. In general, the results show that tourist users of online shopping meet the following requirements. Younger people are more likely to buy via internet, tourists who come to Spain looking to relax or for beach and sun (leisure tourists) prefer to purchase the tourist product online, tourists without package holidays and who travel by low cost companies have more probability of buying online than tourists planning the travel with package holidays or travel by air, in a full service airline, or by road. Furthermore, geographical characteristics show that tourist coming from United Kingdom and going to the beach in Community of Valencia have the greatest probability of using Internet for looking up information and also of buying online. 5. References Alm, J. and M. I. Melnik (2005) Sales Taxes and the Decision to Purchase Online, Public Finance Review, 33(2), 184-212. Combes, G.C. and J.J. Patel (1997) Creating lifelong customer relation-ships: why the race for customer acquisition on the Internet is so strategically important, Iword, 2(4), Hambrecht & Quist. 17 Goolsbee, A. (2000) In a World without Borders: The Impact of Taxes on Internet Commerce, Quarterly Journal of Economics, 115(2), 561-76. IBIT (2001) Study on Electronic commerce in the value chain of the tourism sector. Marcussen, C.H. (2006) Trends in European Internet distribution of travel and tourism services, http://www.crt.dk/uk/staff/chm/trends.htm. OECD (1999) The economic and social impact of electronic commerce. Preliminary findings and research agenda. Van de Ven, W. P. M. M. and B. M. S. Van Pragg (1981) The Demand for Deductibles in Private Health Insurance: A Probit Model with Sample Selection, Journal of Econometrics, 17, 229-252. 18 Figure 1: Percentage of Tourist with Internet Access and Online Shopping 39.96% 24.50% 60.04% 15.46% 0% 20% 40% 60% 80% 100% Internet Access Online Shopping Yes No Source: EGATUR and own elaboration. 19 Table 1: Percentage of e-commerce by tourist’ characteristics, trip attributes and other control variables Tourists’ characteristics Total tourists Internet access Trip attributes Total tourists Internet access Age Size of travel group <= 24 years 36.19% 23.91% Alone 39.53% 38.32% 24 < age <= 44 29.69% 23.59% Couple 29.64% 30.90% 44 < age <=64 21.60% 24.75% More than two 30.83% 30.78% >= 65 years 12.52% 27.76% Tourist main destination Purpose of the trip Rest of Spain 13.43% 13.18% Work and Business 22.74% 30.66% Andalusia 12.06% 16.25% Sun and beach (or relax) 32.39% 30.17% Balearic Island 16.44% 16.70% Other motives 44.87% 39.17% Canary Island 8.16% 7.76% Age & Purpose of the trip Catalonia 15.88% 14.64% <= 24 years & Sun and beach (or relax) 32.56% 22.59% Community of Valencia 22.46% 18.31% 24 < age <= 44 & Sun and beach 31.35% 23.18% Madrid 11.57% 13.16% 44 < age <=64 & Sun and beach 23.16% 25.51% Length of stay >= 65 years & Sun and beach 12.93% 28.71% 1 < days < 3 28.94% 35.46% 4 < days < 7 37.96% 32.71% <= 24 years & Work and Business 38.19% 26.15% >= 8 days 33.10% 31.83% 24 < age <= 44 & Work and Business 23.68% 23.91% Type of accomodation 44 < age <=64 & Work and Business 20.32% 25.21% Other type of accomodation 33.50% 33.27% >= 65 years & Work and Business 17.81% 24.72% Free accomodation 41.62% 40.43% Tourism resort 24.88% 26.30% <= 24 years & Other motives 37.61% 23.57% Type of travel 24 < age <= 44 & Other motives 33.53% 25.56% Full Service Airline 26.01% 31.09% 44 < age <=64 & Other motives 18.78% 24.19% Low Cost Company 67.26% 51.64% >= 65 years & Other motives 10.07% 26.68% Road 6.73% 17.28% Level of education Other control variables Basic education 20.62% 27.16% Seasonality Medium education 35.00% 35.99% First Quarter 24.20% 26.16% University education 44.38% 36.85% Second Quarter 24.57% 24.63% Country of residence Third Quarter 23.90% 23.67% France 6.84% 13.40% Fourth Quarter 27.33% 25.54% Germany 14.29% 14.58% Number of visits United Kingdom 24.09% 20.90% Number of visits >=10 47.58% 55.55% Italy 17.98% 17.44% Number of visits < 10 52.42% 44.45% Netherlands 21.69% 18.30% Rest of the World 15.11% 15.38% Level of income High 30.07% 31.01% Médium 32.81% 31.99% Low 37.12% 36.99% Organization of the trip without package tour 70.19% 66.49% with package tour 29.81% 33.51% 20 Table 2: Estimation of the probit model with sample selection E-commerce Coef. Std. Err. Internet access Coef. Std. Err. Age & Purpose of the tripe Age <= 24 years & Sun and beach (or relax) 0.1994 (0.048) *** <= 24 years 1,1751 (0.029) 24 < age <= 44 & Sun and beach 0.1427 (0.033) *** 24 < age <= 44 0.9609 (0.024) *** 44 < age <=64 & Sun and beach 0.0984 (0.035) *** 44 < age <=64 0.5836 (0.024) *** <= 24 years & Work and Business relations -0.2038 (0.125) Level of education 24 < age <= 44 & Work and Business relat. -0.5062 (0.045) *** Basic education -0.3718 (0.020) *** 44 < age <=64 & Work and Business relat. -0.4441 (0.073) *** Medium-High education -0.2868 (0.013) *** Level of education Country of residence Basic education -0.2641 (0.042) *** France -0.8092 (0.030) *** Medium education -0.0646 (0.026) ** Germany -0.1312 (0.027) *** Country of residence United Kingdom 0.0945 (0.026) *** France -0.2326 (0.063) *** Italy -0.2623 (0.034) *** Germany 0.0186 (0.042) Rest of the world -0.2643 (0.027) *** United Kingdom 0.2602 (0.040) *** Level of income Italy -0.1976 (0.052) *** High 0.2142 (0.052) *** Rest of the world -0.2270 (0.042) *** Medium 0.1591 (0.050) *** Level of income Organization with package tour -0.6599 (0.019) *** High -0.1021 (0.099) Purspose of the trip Medium -0.0959 (0.097) Work and Business relations -0.5606 (0.024) *** Organization with package tour -0.7892 (0.039) *** Sun and beach 0.0761 (0.018) *** Type of travel Size of travel group <=3 0.0998 (0.015) *** Full Service Airline 1.0015 (0.040) *** Tourist main destination Low Cost Company 1.7268 (0.046) *** Rest of Spain 0.1120 (0.025) *** Size of travel group Andalusia -0.4450 (0.025) *** Alone 0.0671 (0.032) * Canary Island 0.1409 (0.018) *** Couple 0.0185 (0.025) Catalonia 0.0959 (0.022) *** Tourist main destination Community of Valencia 0.1966 (0.024) *** Rest of Spain -0.2219 (0.042) *** Madrid -0.1903 (0.030) *** Andalusia -0.2107 (0.051) *** Length of stay Canary Island -0.6237 (0.036) *** 1 < days < 3 -0.1862 (0.019) *** Catalonia -0.0710 (0.037) * 4 < days < 7 0.0610 (0.013) *** Community of Valencia 0.1940 (0.042) *** Type of accomodation Madrid -0.3672 (0.048) *** Free accomodation -0.1250 (0.017) *** Length of stay Seasonality 1 < days < 3 0.1806 (0.034) *** Second Quarter 0.0696 (0.017) *** 4 < days < 7 0.1411 (0.021) *** Third Quarter 0.0129 (0.016) Type of accomodation Fourth Quarter 0.1613 (0.017) *** Free accomodation -0.0124 (0.041) Number of visits >=10 -0.1709 (0.013) *** Tourism resort -0.1227 (0.039) *** Constant -0.6922 (0.069) *** Seasonality Second Quarter -0.0833 (0.030) *** rho 0.2411 (0.077) *** Third Quarter -0.1356 (0.029) *** Log pseudolikelihood -477860.49 Fourth Quarter -0.0557 (0.030) * Number of obs. 60011 Number of visits>=10 0.1661 (0.026) *** Censored obs. 36031 Constant -0.4411 (0.131) *** Uncensored obs. 23980 aIndividual reference: more than 64 years old, other motives of travel, University education, Netherlands, low level of income, without package tour, travel by road, size of travel group over two, Balearic Island, length of stay over 8 days, other type of accommodation, first quarter, less than ten visits. ***Level of significance 1%, **level of significance, 5%, *level of significance 10%. 21 Table 3: Marginal effects and pseudo-elasticities of the probit model with sample selection Marginal Effects Pseudo elasticity Use Internet for booking Direct effects Indirect effects Total Direct effects Indirect effects Total Age & Purpose of the tripe <= 24 years & Sun and beach (or relax) 0.0749 0.075 22.7% 22.7% 24 < age <= 44 & Sun and beach 0.0531 0.053 16.1% 16.1% 44 < age <=64 & Sun and beach 0.0363 0.036 11.0% 11.0% <= 24 years & Work and Business relations -0.0701 -0.070 -21.3% -21.3% 24<age<= 44 & Work and Business relations -0.1578 -0.158 -47.9% -47.9% 44<age<=64 & Work and Business relations -0.1415 -0.142 -42.9% -42.9% Level of education Basic education -0.0892 0.0123 -0.077 -27.1% 10.5% -16.6% Medium education -0.0230 0.0187 -0.023 -7.0% 16.0% 9.0% Country of residence France -0.0793 0.0492 -0.030 -24.1% 42.1% 18.0% Germany 0.0067 0.0107 0.017 2.0% 9.1% 11.2% United Kingdom 0.0986 0.0002 0.099 29.9% 0.2% 30.1% Italy -0.0681 0.0088 -0.059 -20.7% 7.5% -13.1% Rest of the world -0.0775 0.0070 -0.071 -23.5% 6.0% -17.6% Level of income High -0.0361 -0.0199 -0.056 -10.9% -17.1% -28.0% Medium -0.0339 -0.0158 -0.050 -10.3% -13.6% -23.8% Organization with package tour -0.2203 -0.0181 -0.238 -66.8% -15.5% -82.3% Type of travel Full Service Airline 0.3828 0.383 116.2% 116.2% Low Cost Company 0.5712 0.571 173.3% 173.3% Size of travel group Alone 0.0246 0.025 7.5% 7.5% Couple 0.0067 0.007 2.0% 2.0% Tourist main destination Rest of Spain -0.0759 -0.0194 -0.095 -23.0% -16.6% -39.6% Andalusia -0.0723 0.0218 -0.050 -21.9% 18.7% -3.3% Canary Island -0.1861 -0.0479 -0.234 -56.5% -41.0% -97.5% Catalonia -0.0253 -0.0102 -0.035 -7.7% -8.7% -16.4% Community of Valencia 0.0728 -0.0083 0.065 22.1% -7.1% 15.0% Madrid -0.1201 -0.0087 -0.129 -36.4% -7.4% -43.8% Length of stay 1 < days < 3 0.0676 0.0200 0.088 20.5% 17.2% 37.7% 4 < days < 7 0.0525 0.0004 0.053 15.9% 0.3% 16.2% Type of accomodation Free accomodation -0.0045 -0.004 -1.4% -1.4% Tourism resort -0.0431 -0.043 -13.1% -13.1% Seasonality Second Quarter -0.0296 -0.0089 -0.038 -9.0% -7.6% -16.6% Third Quarter -0.0475 -0.0076 -0.055 -14.4% -6.5% -20.9% Fourth Quarter -0.0199 -0.0141 -0.034 -6.0% -12.1% -18.1% Number of visits>=10 0.0621 0.0185 0.081 18.8% 15.9% 34.7% 22 Appendix Tourists’ characteristics: Age: The socio-demographic characteristics have been defined including information related to the age of the tourist. We have established four categories: Under 24, between 24 and 44, between 45 and 64 and over 64. Level of Education: The educational level has been established in three different categories: Basic, Secondary and University Education. Country of residence: We have considered six different origins: France, Germany, United Kingdom, Italy, Netherlands and the rest of the world. Level of income: This variable considers different income levels which are placed into the following categories: High income level, Medium income level, and Low income level. Purpose of the trip: These variables identify tourists whose principal motives of the Spain visit is Work and Business relations, Sun and Beach and other motives. Organization of the trip: This variable recognizes if the tourists have visited Spain with a package tour or not. Trip attributes: Size of travel group: With this variable we identify if the tourist travels Alone, as a Couple or in a Group of more than two persons. Tourist main destination: In order to collect the main tourism destinations in Spain we have defined seven dummy variables: Andalusia, Canary Islands, Balearic Island, Catalonia, Community of Valencia, Madrid and other destinations, respectively. Length of stay: In order to identify the tourist’s fidelity, we have considered the categories: More than once in a year, Once in a year and Less than once a year to refer the number of times that the tourists visit Spain in one year. Type of accommodation: We use three different categories: Tourism resort, Free accommodation and other type of accommodation. Type of travel: We use three different categories: Full Service Airline, Low Cost Company and Road. Other control variables: Seasonality: these variables identify the quarter during which the trip is made. Number of visits: In order to identify the tourist’s fidelity, we have considered the categories: More than ten and Less than ten to refer the number of times that the tourists visit Spain in one year. 23 Relación de títulos publicados en la colección ALCAMENTOS. Nº Autor/es Título 0801 Juan Muro Cristina Suárez Maria del Mar Zamora The impact of e-commerce on the tourist purchase decision: An empirical micro analysis