This study captures the relationship between three constructs, i.e. the economy, the infrastructure, and tourism to explain the direction and magnitude of the relationship between these constructs. To do so, an apriori multidimensional causal model was hypothesised and tested using Structural Equations Modelling methodology. The economy was hypothesised to have a positive significant relationship with the infrastructure construct and a negative relationship with the tourism construct, while the infrastructure construct was hypothesised to have a positive significant relationship with the tourism construct. The sample comprised of 189 countries, which was extracted from the World Bank (1995) data set. Results supported the structural relationship between economy and infrastructure and infrastructure and tourism. Contrary to the hypothesised relationship, the direct relationship between the economy and tourism was found to be insignificant. The significance of this study is in terms of verifying causal relationships between constructs at the aggregate level using countries as the unit of analysis.
The importance of tourism development and its contribution to the growth of countries has been well documented through prior research. "From the most optimistic perspective, tourism may be seen to foster economic growth, improve human living standards, promote intercultural understanding, and nurture world peace" (BeIk and Costa, 1995). These contributions have been even more pronounced to the growth of developing countries encouraged by international organisations such as UNESCO, The United Nations, the International Monetary Fund, The World Tourism Organisation, and the World Bank (Belk and Costa, 1995). The contributions of tourism to a given region or country is well researched in terms of the economic and social perspective vis-à-vis effects on GDP, multiplier effects, culture and so forth. The direction of research initiatives has been to study direct and indirect effects of tourism on the economy and society in order to assess its present and/or future contributions.
Approaches to modelling in the tourism industry have revolved around time-series forecast models and regression models with the primary intent to predict and/or explain tourism-related issues. These models developed in tourism studies are predominantly single equation models used to explain demand at the aggregate or cross-country level of analysis (Sinclair, 1998). This study is a step taken towards verifying the theoretical underpinnings (structure) pertaining to the variables that affect tourism. Although it is apparent that the relationship between variables may be self-explanatory based on past research, significant contributions have not been made to develop models in tourism research to confirm the theoretically hypothesised structural relationships between the constructs/variables at the aggregate level. Moreover, a majority of the research efforts have focussed on the effects of tourism on the economy and society; and not vice versa, which would help explain the effects of the state of the economy on tourism.
With the aforementioned points as the precursor, the overall objective of this paper is to confirm the relationship between the variables that have been used in prior research to explain the impact on tourism. By identifying the underlying structural relationship between variables used in tourism based studies, the intent is to develop a causal model in order to explicate the relationship between constructs. To accomplish this, Structural Equation Modelling was used to confirm a hypothesised model derived from theoretical underpinnings. The model will help integrate the issues dealt with by other researchers in isolated studies. Since this approach is at the aggregate level with countries taken as the unit of analysis, the outcome of this study will help understand and verify the relationship between constructs and/or variables; that is, whether the theory established through prior research really holds good at the aggregate level. The important question that the model will help address is if relationships between the constructs and variables exist at the aggregate level that span countries from both the developing and the developed world?
IDENTIFYING AND DEFINING VARIABLES IN TOURISM ECONOMICS
Tourism encompasses travellers away from home and the business and people that serve them by expediting or otherwise making the travel easier or more entertaining (Lundberg et al, 1995). The impact of tourism on the economy of a given region or country is studied through tourism economics, which measures the amount of travel and its economic consequences, direct, indirect and induced (Lundberg et al, 1995). Tourism has also been studied in terms of its influence on psychological, sociological, ecological, and political structure of a region/country. The essence of tourism lies not only on how it impacts the economy of the region being studied but also on how the economy and infrastructure of a given country/region affects the tourism potential of that country. In order to understand this concept, it is important to study the variables that have a direct/indirect impact on tourism inflow.
It is a well-established fact that countries with the highest per capita gross domestic product (GDP) have the highest propensity to travel (Lundberg et al, 1995). However, the relationship between the state of the economy as well as infrastructure of the destination country would impact the number of tourism arrivals to that country. Therefore, it would be interesting to study the relationship between the economy, the infrastructure, and tourism for a sample of countries to test if a relationship actually exists between the constructs. In other words, the question that needs to be addressed is: how does tourism revenues and/or tourism arrivals vary based on the infrastructure of a given country? Do countries that are more developed as far as infrastructure is concerned have a higher potential for tourism as compared to those that are not as developed while assuming that the effects of tourist attraction is constant? Would this indicate that there is a significant relationship between the infrastructure of a given country and its tourism potential, and most importantly, if this relationship can be generalizable across countries? Similarly, the impact of the economy on the infrastructure as well as tourism arrivals to a country needs to be confirmed across both developing and developed nations.
Past research has delved into the factors that affect the influx of tourism in developed and developing countries. For instance in 1993, the World Tourism Organisation (WTO) estimated that Africa gets about 4% of the tourist arrivals and 2% of the worlds total tourism receipts. South Asia was estimated to get 1 % of the world's receipts and 1 % of the arrivals (Lundberg etal, 1995). Similarly, the Middle East was estimated to get 2% of the worlds arrivals and 2% of the receipts. On the contrary, Hawaii received more visitors and generated more tourism income as compared to regions like South America, South Asia, and the Middle East. Similarly, as pointed out by Hassan (2000), "a country such as Singapore with 3 million people attracts 6.4 million tourists and generates $7.6 billion (excluding transport) in tourism receipts." Countries like Hawaii and Singapore have higher tourism arrivals than other countries, which can be attributable to the economic and infrastructural factors that favour the growth of tourism. This can be confirmed testing the direction and magnitude of relationship between the economy, the infrastructure, and international tourism for countries that are both developing and developed in order to generalise that the effects hold good across countries.
Factors that affect tourism inflow to a region/country have been attributed to various economic variables. For instance, Martin and Witt (1987) have shown that consumer price index is a reasonable proxy for the cost of tourism. Other researchers have used exchange rates to show the impact of tourism arrivals in a given country based on the buying power of the currency of the country of origin (Uysal and Crompton, 1984). Witt and Witt (1995) state that exchange rates are also sometimes used separately to represent tourist living costs. It is essential to considéra unification rate along with exchange rate to understand the effects on tourism costs. According to Lundberg et al. (1995), change in currency exchange rates can have considerable effects on tourism, and therefore, the economic stability of countries. The authors' state that, from the perspective of tourism economics, inflow of foreign currencies to pay for tourism services can be critical to the economic well being of a country. Currency exchange rates are influenced by other economic factors like real income, inflation rates, and interest rates. The effect of currency exchange rate on tourism has been documented in terms of tourist arrivals to a country. Travel to a foreign country increases ordecreases based on the prevalent exchange rate between the country of origin and destination country. Also, the economic conditions within a country affect the purchasing power of individuals within that country as well as inbound tourists to that country. The cost of living comparisons between the destination country and the country of origin can be analysed by comparing the purchasing power parity conversion factor between the two countries. This is also done by comparing the conversion factorin terms of the purchasing power of the local currency with respect to the U.S. dollar. An increase in the factor indicates that the prevalent cost of living in the destination country can affect the international traveller to that destination because of the difference in the purchasing power of the currency used by the traveller.
The sources of funds available to develop the infrastructure of the destination country are an important variable that needs to be analysed. Foreign Direct Investments (FDI) that are channelled into the destination country can be used to develop tourism infrastructure. For instance, according to the World Bank, funds are mobilised into developing countries to help develop their infrastructure to cater to the tourism demand as well as to tap tourism potential through the International Finance Corporation (IFC) that has been involved in tourism-related investments and project financing. According to Price (1998), "in 1997, the IFC's portfolio in tourism involved US $ 447 million committed in 86 investments." IFC invests in commercial hotels, resorts and real estate development. Therefore, foreign direct investment is an indicator of how funds inflow can be used to develop infrastructure to cater to tourism demand and tap the potential.
The IFC has also contributed to the development of small and medium sized enterprises, with the main objective of stimulating greater involvement of the private sector's smaller businesses in terms of environmental protection initiatives. The goal of this initiative is to promote eco-tourism by preserving natural resources in a given region of countries. Target countries are considered to be those countries that have few development prospects aside from tourism, where the industry has the potential to become an important component of the economy and contribute to the dispersal of economic activities throughout a country. Other organisations that fund the growth of tourism are Multilateral Investment Guarantee Agency (MIGA). According to the study conducted by MIGA, FDI is an important feature of tourism development financing. In terms of the model, it is not certain what the outcome of the relationship between FDI and tourism economy would be, as there exists a financial gap for tourism projects in many developing countries. Given the approach of IFC and other such organisations that finance infrastructural development of countries, FDI is considered as an important element that may have a direct positive impact on the economy, and in turn may have an indirect positive effect on tourism potential. The essence of FDI is that it helps maintain economic growth in most countries. Witt & Witt (1995), state that the economy is an indicator of business tourism. According to the authors, the level of economic activity in a destination is likely to influence the demand for tourism, especially the business segment. And therefore, an increase in FDI may suggest that the direct effects could be increased economic activities and the indirect effects, an increase in business tourism potential.
Tourism is a composite product that consists of multiple products such as accommodation, transport, food and beverage production and service establishments, entertainment, and other services (Sinclair, 1998). The components that form an essential part of infrastructure to develop tourism are travel, lodging and environment related. Good infrastructure for tourism would mean that destination countries need to have for instance, good paved roads, railway network properly laid out for easy access to destinations within a country, well laid out communication network, and entertainment facilities. Good communication channels within the country helps build links with origin countries as well as help increase the tourists' awareness of what destinations to visit. Television and telephone, and personal computers are good indicators of how well developed are the communication channels in a destination country that can have a direct impact on the factors that influence infrastructure. Other variables that capture the travel infrastructure of a country are the number of vehicles, as this would indicate the propensity to travel as well as infrastructure provided for road travel.
Forecasting models developed in tourism studies have used tourism demand in terms of the number of visits (arrivals) from a country of origin to a destination country (Sheldon, 1990). Market share or demand for a particular destination is expressed in terms of tourist visits or tourist expenditures/receipts (Witt & Witt, 1995; Sinclair, 1998). Tourist expenditures on products and services related to tourism capture real demand for these products and services. Travel services provided by the destination countries have a direct impact on tourist arrivals. The above indicators were considered for the tourism construct study, i.e. tourism arrivals, tourism receipts and travel services. The variables/indicators used for the study are provided in Table 1 and their description, in Appendix 1.
EXOGENOUS AND ENDOGENOUS CONSTRUCTS
As listed in Table 1, the variables used for the study are expected to load on three constructs, i.e. economy, infrastructure, and tourism. Constructs are unobservable, and exist only if it is determined that the variables that are theoretically considered to be part of a given construct explain a common variance. The model domain needs to be defined at the outset, in that the endogenous constructs and exogenous constructs need to be specified a priori. As indicated in Table 1, the exogenous constructs are the economy and the infrastructure. By definition, exogenous constructs are those that are influenced by factors outside the model, whereas endogenous constructs are those that are influenced by factors specified within the model. The same applies to the variables, that is, those variables that are influenced by variables outside the model are exogenous and those that are influenced by the variables specified within the model are endogenous. In the present study, since the objective is to derive a model that explains tourism by studying economic and infrastructure related factors that are considered to be causal factors, the economy was considered as the exogenous construct and the infrastructure and tourism were considered as an endogenous constructs. Infrastructure was considered as endogenous as it is dependent on the economic policies of a given country. And since the effect was measured in terms of impact on tourism, this was taken as the endogenous construct. In a nutshell, the constructs of the tourism model have been identified as economy, infrastructure, and tourism.
DEVELOPING THE HYPOTHESES
The economy of a region/country affects tourism arrivals and revenues; therefore some economic variables would have a positive impact while others would have a negative impact on international tourism arrivals to a given region/country. As pointed out in the previous section, variables such as consumer price index and purchasing power parity influence international tourism arrivals to a given region. However, this effect may be negative, that is, if the cost of visit goes up for international travellers to visit a given region, this could affect the inflow of tourists to that region. Therefore, it is essential to verify if the direction of relationship between the economy construct and the tourism construct is influenced by variables that make up the economy construct. In this study, the direct effects of two of the four variables, i.e. consumer price index and purchasing power parity conversion factor used in the economy construct may have a negative impact (direct effects) on tourism arrivals to a given region/country. The other two variables, i.e. foreign direct investment and industry value added may not have a direct effect on tourism arrivals, rather these variables effect infrastructure directly, which in turn may effect tourism arrivals.
The "purchasing power parity conversion factor" was used as a variable to represent the economy construct. This variable indicates how many units of a given currency will need to be spent to consume a given product in a domestic market as one dollar would in the United States. For instance, higher the inflation rate in the destination country, the lesser the value of the local currency as compared to the dollar in terms of purchasing power. There are other economic factors that affect the value of the local currency, which would then determine buying power of the international traveller in the destination country. Therefore, purchasing power parity needs to be assessed along with consumer price index, which influences the cost of living in order to analyse the effect of these variables on international tourism arrivals and international tourism revenues. Higher the cost of living in the destination country, the more it will impact the international tourism arrivals to that country. Similarly, higher the differential in the purchasing power of the currency between the destination country and the country of origin, the more it will impact tourism arrivals to that country. Therefore, there would be a negative relationship between international tourism and the economy constructs if these two variables alone are considered as part of construct.
The other two variables, i.e. higher levels of foreign direct investment and industry value added (manufacturing) might have a positive or negative impact on the economy of the destination country. The destination country benefits from higher levels of investments that are available from foreign sources. Therefore, the relationship should be positive but only to the extent that the levels of investment is not too high that might increase the risk ratings of the industries of the destination countries. This is so because the higher a country relies on foreign direct investments, the more it indicates that the country relies on external sources to fund its growth, which affects economic variables such as interest rates and cost of living. The same applies to "industry value added," in that the destination country benefits from increased manufacturing industry value added from industries such as mining, electricity, and construction. Therefore, a positive relation between the variable and the economic construct is expected from higher value additions from the manufacturing sector.
However, if the proportion of manufacturing sector value added is more than service value added for a given country, it may indicate that the country's service sector is not well developed, which could suggest that the service sector of the country is not a major contributor to the country's GDP as compared to a fully developed country. Typically, developed countries rely more on the service industry for value addition, which might indicate that a negative economic impact might if the value addition through the manufacturing sector is high. However, in this study the industry value added is compared among the countries of the sample in absolute terms, and therefore a higher value may indicate higher contribution to the well being of the economy and vice versa. It should be noted that variables such as FDI and Industry Value Added have a direct positive effect on the infrastructure of a given country, which in turn affect tourism arrival and/or tourism revenues. Therefore, the direct effects of these variables on tourism arrivals were not considered to be significant, which leads to the hypothesis that the direct effect between the economy and the tourism constructs would be negative as a result of the direct effects of the other two variables, i.e. Consumer Price Index and Purchasing Power Parity Conversion Factor on the tourism arrivals and/or revenues.
The direct effects of the economy of a given country on the infrastructure of that country need to be considered. In this study, the economy construct is hypothesised to have a direct impact on the infrastructure construct, which will in turn affect tourism. The relationship between the economy construct and the infrastructure construct in this study is hypothesised to be positive. This is so because the relationship between the economy and the infrastructure of a country is direct. As a result, the better the economy of a given country, the better the likelihood that the infrastructure of that country is good. This relationship may be more apparent when time series comparisons are made between the two constructs. Likewise, the relationship between the infrastructure and the tourism constructs will be positive. That is, better the infrastructure of a given country, the better the possibility that the country can tap its tourism potential. It should be noted that a linear relationship between variables is assumed in this study. Good infrastructure in destination countries will have a positive impact on international tourism arrivals and revenues as compared to countries with a poor infrastructure. In this study, the infrastructure construct comprises of variables such as paved roads; number of computers; television, and radio per 1000 population; number of vehicles per 1000 population; and transportation services, which will positively affect international tourism arrivals and revenues. Based on the logic presented in this section, the model hypotheses were developed, which is explained in the following section.
The Hypothesised Measurement Model
The hypotheses for the measurement model were common for all indicators. In other words, since all indicators were hypothesised to have a positive relationship with their latent construct, a common hypothesis was developed for the relationships between the indicators and the constructs, i.e. economy, infrastructure, health, and tourism:
Hypothesis (H1): There will exist a significant positive or negative relationship between the indicators and the latent construct for each of the indicators identified in the study.
The Hypothesised Structural Model
Based on the literature outlined in the previous section, the following structural relationships are hypothesised between the constructs - economy, infrastructure and tourism:
Hypothesis (H2): There will be a significant positive relationship between the economy and the infrastructure constructs.
Hypothesis (H3): There will be significant positive relationship between the infrastructure and tourism constructs.
Hypothesis (H4): There will be a significant negative relationship between the economy and tourism constructs.
DATA DESCRIPTION, ANALYSIS, AND FINDINGS
Data used for the study was taken from World Bank Development Indicators, 1998. The data series was chosen for the year 1995, as there were lesser missing values in the data set as compared to the ones provided for 1994 and 1996. World Bank reports data on the economy, infrastructure, society, and tourism related issues. The data is reported country wise, year wise and indicator wise. Data for the indicators outlined in the previous section was drawn from the data set for 212 countries. All the indicators were continuous variables or ratio type data. Since all the countries did not have data reported for the indicators, some of the countries were deleted to arrive at the final sample size of 189 countries. The data set had missing values in it, which needs to be considered as a limitation to the study. As a result, pair-wise deletion was used to arrive at the correlation matrix. The standard deviation for the selected variables reflected large values indicating that the data was nonnormally distributed. This was because the countries ranged from fully developed to developing countries and hence the values ranged considerably across countries. In order to address this problem, the data was transformed by taking the natural log of the indicator values. The mean and standard deviation for the transformed data indicated that most variables were normally distributed, with the exception of one variable with standard deviation of above three. Bivariate correlation between selected variables revealed that the indicator "Gross Domestic Product Per Capita" was highly correlated with almost all other indicators. This variable was deleted in order to reduce the effects of multicollinearity. Correlation of above 0.80 was considered as high. The correlation matrix is provided in Table 2.
THE MEASUREMENT MODEL
Statistical analysis was conducted by using Structural Equation Modelling methodology through the LISREL 8.3 software. The first step was to check whether the measurement model, i.e. the loadings of each of the indicators on the factors were significant. The measurement model was run constraining the path of the first variable loading on the factor and freeing the other paths that loaded on the same factor. The first model was based on the "Economy" and "Health" factors. Since health factor had only two indicators, the health and economy factors were correlated in order to run the measurement model. Results indicated that the loadings were significant for the economy factor (refer to Table 2). However, this was not the case for the health factor. The squared multiple coefficients for the health factor were not significant at p=.05, suggesting that the indicators used to measure health related issues did not have any commonality in terms of shared variance. The correlation between health and economy factors was significant. Similarly, the measurement model was tested for infrastructure and tourism variables. For infrastructure, all loadings were significant as detailed in Table 2. This was observed in the case of the tourism factor, which showed significant loadings for all the three indicators, i.e. tourism arrivals, tourism revenue, and tourism services. The measurement models indicated that the infrastructure construct was valid except for the indicator "Personal Computers per 1000 people." Hence, this indicator was dropped from the construct. The health factor had two indicators, i.e. immunisation and number of hospital beds per 1000 people. Both these indicators had poor loadings and hence, this factor was not considered as a significant construct by itself.
A second test on the measurement model was conducted, this time omitting the health construct from the prior model. All indicators loaded significantly on the economy factor. The values of the loadings are listed in Table 3. Infrastructure factor was tested again by analysing a measurement model without the indicator 'Personal Computers." The loadings were significant for all factors. The final measurement model for the economy factor, infrastructure factor, and tourism factor included the indicators as shown in Table 3. Hypothesis H(1) was supported as the direction and magnitude of all loading were significant.
THE STRUCTURAL MODEL
The structural model was based on the hypothesis that the economy factor has a direct relationship with both infrastructure and tourism. As pointed out earlier, it was expected that the gamma coefficient would be negative for "Economy-Tourism" path and positive for the "Economy-Infrastructure" path. This is justified by the fact that, a better economy would be able to provide a better infrastructure for tourism than a weaker economy. Similarly, as stated earlier, the gamma coefficient for economy and tourism was expected to load negatively. Again, this was primarily because of the inverse relationship between the variables used as part of the economy construct and the variables that were included as part of the tourism construct.
For instance, the relationship between economic variables such as "CPI" and "Purchasing Power Parity Conversion Factor" have an inverse relationship with tourism related variables such as "International Tourism Arrivals" and "International Tourism Revenues." Higher the CPI for a given year for a given destination country, the more it will affect the international tourism arrivals to that country and the tourism revenues generated for that year, as CPI affects cost of products and services, which in turn affects the cost of travel for the international travellers. The same is true for the variable "Purchasing Power Parity Conversion Factor." Finally, the beta coefficient between infrastructure and tourism was expected to load positively. This is based on the fact that the infrastructure has a positive effect on tourism of a country. The better the infrastructure, the more satisfaction it creates for domestic as well as international travellers; the effect perhaps being more pronounced in the case of the latter.
Since the health factor and economy were correlated, a model with health factor as the endogenous variable and economy as the exogenous variable was tested in order to verify the structural relationship between these constructs. The results indicate that health factor was not significantly related with economy (Gamma = .14). Also, the Gamma coefficient between health and tourism was negative and non-significant (Gamma = -.03). Since the measurement loadings for the health factor were not significant, therefore this model does not fit the data well. Furthermore, the fit indices for the model were not good (see Table 4a for model fit indices). Subsequently, the structural model without the health factor was tested that included economy and infrastructure related variables. The results indicated that the gamma coefficient for economy and infrastructure was positive (Gamma = .79) and significant. Similarly, the gamma coefficient for the economy and tourism relationship was negative (Gamma = -0.1) but not significant. This is in tandem with what was hypothesised during model specification. Although this may be the case, it is essential to include more indicators that influence infrastructure and tourism to verify the findings of this study. The Chi-square difference test between this model and the model without health factor is significant indicating that the model without the health factor is more relevant in terms of fit. In other words, the model as given in Figure 1 without the health factor fits the data better than the one with health factor included. In the case of the relationship between infrastructure and tourism, the beta coefficient (Beta = 0.92) was both positive and significant, clearly supporting the hypothesis that good infrastructure is a prerequisite for tapping international tourism potential of a given country. The overall model supports the hypotheses H(2), H(3), and H(4). Although, the negative yet non-significant relationship between economy and tourism is justifiable, it needs to be explored and researched further to ensure that the reasons for the negative relationship are attributable to the variables that make up the economy construct.
The model fit for the model without health factor was good without variable × 7 (no. of computers per 1000 people). This variable that loads on the infrastructure factor has high correlation with telephone lines per 1000 people. Hence, by deleting this variable, it was expected that a better model fit would result. The fit indices also showed significant improvement across the models shown in Table 5 (a). The fit indices for the final model suggest that although the chi-square goodness of fit measure was significant, the overall model (final) was better than the other two models in terms of fit criteria. Typically, the chi-square goodness of fit test should be non-significant, which will indicate that there is no significant difference between the theoretical model and the hypothesised model. Since the sample size affects the results obtained using the chi-square goodness of fit criteria, the chi-square divided by degrees of freedom index is used to check the overall fitness of the model. Results indicate (as given in Table 5a) that the final model had a value of between two and four, which is considered to be better than value above four. Moreover, other fit indices for the final model met the criteria for good fit, which Model 1 and Model 2 failed to meet. For instance, the final model met the criteria laid out by researchers for the GFI, CFI, NFI, RMSEA, Standardised RMSR indices (GFI>0.9, CF1>0.9, NFI>0.9, RMSEA~ 0.08, Std. RMR~0). Moreover, the Chi-square difference test between Model 2 and 3 was significant as indicated in Table 5(b). Hence, Model 3 was accepted as the final model.
DISCRIMINANT AND CONVERGENT VALIDITY
Once the measurement model was confirmed, discriminant validity was confirmed by testing a three-factor correlated model. The purpose was to check the correlation between the factors in order to identify if they were constructs that were distinctly different from each other or not. All elements of the "phi matrix" were fixed except the off diagonal elements below the diagonal, which were freed. Results indicate that the correlation between the economy, the infrastructure, and the tourism constructs were significant. Moreover, the fit indices for the three-factor correlated model provided in Table 5(b) indicate that the overall model is good. Convergent validity was established as all loadings were significant. The better each variable loads on to its construct, the better the convergent validity.
LIMITATIONS, IMPLICATIONS, AND CONCLUSIONS
The role of the economy and infrastructure is significant in driving revenue in the tourism industry. The model proposed through the present research is a step taken in that direction to show a casual relationship between economy, infrastructure, and tourism. Relationships between these constructs were tested, which indicated that there exists a positive significant relationship between the economy and infrastructure constructs, and the infrastructure and tourism constructs; whereas a negative insignificant relationship exists between the economy and tourism constructs. The relationship between economy and infrastructure needs to be further researched especially with respect to the economy construct. Countries need to focus on infrastructure related issues to promote tourism as indicated by the model. It is also apparent that the economy needs to be in good shape for tourism to be a value driver for countries. Although this has been established through prior research to some extent, this study confirms the relationship at the global level based on the unit of analysis used for the study.
What emerges from this study is that although the economy and infrastructure constructs have a significant direct and indirect effects on tourism, the variables that make up the constructs play an important role in defining and capturing the overall variance accounted for by the model. The study confirms the multidimensional simultaneous effects of variables on each other while explaining a reasonable level of variance in the constructs they loaded on individually. This further helped explain the structural relationships between the constructs. The significant contribution of this study for policy makers is that it provides a perspective of how resource allocation decisions can affect the economy, infrastructure, and the tourism industry not only in aggregate terms but also in disaggregate terms. Importance should be given to the role indicators play in sustaining the growth of the tourism industry. Policy makers especially in developing countries need to pay particular attention to the role of the manufacturing sector and service sector in terms of their contribution to the growth of the economy. Although cross sectional data was used for the study, the relationship between variables was confirmed across countries, which further establishes the significance of the study.
Future research should focus on developing a more complex construct for tourism, infrastructure and the economy. Some questions that emerge from the finding reported above need to be explored in future studies. For instance, would cross sectional data actually capture the relationship between the economy and the infrastructure for a group of countries included in the sample? In other words, even if the economy of a given country is good for a given year, it may not indicate that the infrastructure of the country would also be good. In fact, it may be hard to determine what variable actually precedes the other variable in terms of identifying cause and effect relationships. Although a strong economy would certainly help in the process of improving the infrastructure of a country, it cannot be stated with certainty if good infrastructure is a prerequisite for strong economic growth. For instance, developing countries may show strong economic growth over a given year, yet these countries may not have good infrastructure to sustain the economic growth. Therefore, the cross sectional analysis between the two variables may not be appropriate to test the causal relationship between these variables.
Indicators like "Industry Value Added" loaded positively on the economy construct, which supports the hypothesis. Higher the value addition by the manufacturing industries, the higher will be the overall economic growth. Although, most developed and some developing nations may be high on the "Services Value Added" factor, the findings of this study indicate that the manufacturing industries' contribution to the aggregate value added for a given country is significant. The gamma coefficient for the economy and tourism indicates that there is a negative non-significant relationship between economy and tourism (Gamma = -0.1). The reason for the direction of the relationship is attributable to the direct effects of the variables, i.e. "Consumer Price Index" and "Purchasing Power Parity Conversion Factor" that were included as part of the economy construct, which were determined to have an inverse relationship with the tourism variables. However, the insignificance of this structural path suggests that these variables by themselves were not sufficient to determine the negative relationship between the economy and tourism constructs. Future research could delve into a more comprehensive construct for the economy construct so that the significance of the magnitude and direction between the economy and tourism structural paths could be determined.
Tourism construct can be improved by including variables such as number of hotels and restaurants in countries that may help strengthen the model. Also, including domestic tourism receipts will help improve the tourism construct further. For instance, the effects of infrastructure and the economy on domestic tourism can be studied, which will help explain the contribution made by domestic tourism to the overall growth of the tourism industry for countries across the world. This will be of particular interest to policymakers in those developing countries that have not exploited their tourism potential in context of both domestic and international tourism. Also, more economic variables can be used in the study to verify the complexity of the relationship between variables that make up the economy construct. This will help address the issue pertaining to this study in terms of the negative relationship between the economy and the infrastructure. The presence of more variables will also factor out any spurious relationships that may exist between variables and constructs. Moreover, it should be noted that this study captures the effects between constructs and variables by using cross-sectional data. For instance, the effects of economic variables such as CPI on tourism inflow will be immediate, and therefore cross sectional data serves the purpose of capturing the relationships among variables and constructs. Similarly, the infrastructure variables such as paved roads and transportation affects tourism on the immediate or short run. However, future research could focus on the lag effects of some of the independent variables pertaining to the economy and the infrastructure constructs on the dependent variable pertaining to the tourism construct. This will be particularly helpful in testing the effects of the economy on the infrastructure as the economy of a given country may impact the infrastructure over a period of time rather than at a particular point in time.
While including more variables it must be kept in mind that parsimonious models have better fit than models that are highly complex in terms of number of variables used forthe study. In this connection, theoretical underpinnings that describe the role of the variable should be used as the screening test to include and/or exclude variables from the model. The same applies forthe infrastructure variables, which were well represented by the variables in the present study. However, other variables for infrastructure need to be considered that may help improve the overall model.
The limitations of this paper were related to working on a data set with missing values, which may have affected the results. Primary research needs to be conducted or other secondary data sets need to be used to plug in these gaps. Future studies could also test the effects of tourism on the economy and the infrastructure. In the present study, tourism was used as a dependent or endogenous construct, while economy and infrastructure were considered as independent or exogenous constructs. Since tourism helps build the economy as well as the infrastructure, the causal relationship while considering tourism and infrastructure as the independent variables may help explain the role of the tourism industry in sustaining economic growth. Furthermore, future research can also distinguish between developing and developed countries in terms of the contribution made by the tourism industry. The sample can be divided into two sub-samples (developing and developed countries) and the model can be tested out separately for these samples in order to verify the theory. This was not possible in the present study because of limited sample size. Although structural models fall into the category of linear models, they help explain the simultaneous causal effects between variables, which help explain the relationship between multidimensional constructs.
Belk, R.W. and JA Costa (1995) "International Tourism: An assessment and overview," Journal of Macromarketing, 15:33-48.
Financial Times (2000a) November.
Financial Times (2000b) October 6.
Financial Times (2000c), October 31.
Hassan, S.S. (2000) "Determinants of Market Competitiveness in an Environmentally Sustainable Tourism Industry," Journal of Travel Research, 38(3): 239-245.
Lundberg, D., M. Krishnamoorthy and M. Stavenga (1995) Tourism Economics, John Wiley and Sons.
Price, A. (1998) "The World Bank Group & Tourism," Travel and Tourism Analyst, 75-90.
Sheldon, RJ. (1990)"AReviewofTourism Expenditure Research," in C.R Cooper(ed.), Progressin Tourism, Recreation, and Hospitality Management, London: Belhaven.
Sinclair, MT. (1998) "Tourism and Economic Development: ASurvey," The Journal of Development Studies, 34(5): 1-51.
Uysal, M. and J.L. Crompton (1984) "Determinants of Demand for International Tourist Flow in Turkey," Tourism Management, 5(4): 288-97.
Witt, S. and C. Witt (1995) "Forecasting Tourism Demand," International Journal of Forecasting, 447-475.
Prakash Chathoth is currently at the Department of Hospitality and Tourism Management, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, USA.
Appendix 1 : Defining the Variables
The indicators as defined by World Bank used for the analysis were the following :
* Consumer price index (1987 =100)'. Consumer price index reflects changes in the cost to the average consumer of acquiring afixed basket of goods and services.
* Foreign direct investment, net inflows (% of GDP): Foreign direct investment is net inflows of investment to acquire a lasting management interest (10 percent or more of voting stock) in an enterprise operating in an economy other than that of the investor. It is the sum of equity capital, reinvestment of earnings, other long-term capital, and short-term capital as shown in the balance of payments.
* Hospital beds (per 1,000 people): Hospital beds include inpatient beds available in public, private, general, and specialised hospitals and rehabilitation centres. In most cases beds for both acute and chronic care are included.
* Immunisation, DPT (% of children under 12 months): Child immunisation measures the rate of vaccination coverage of children under one year of age. A child is considered adequately immunised against DPT (diphtheria, pertussis or whooping cough, and tetanus) after receiving two or three doses of vaccine, depending on the immunisation scheme.
* Industry, value added (% of GDP): Industry corresponds to ISIC divisions 10-45 and includes manufacturing. It comprises value added in mining, manufacturing (also reported as a separate subgroup), construction, electricity, water, and gas. Value added is the net output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or depletion and degradation of natural resources.
* International tourism, number of arrivals: International inbound tourists are the number of visitors who travel to a country other than that where they have their usual residence for a period not exceeding 12 months and whose main purpose in visiting is otherthan an activity remunerated from within the country visited.
* International tourism receipts (current US$): International tourism receipts are expenditures by international inbound visitors, including payments to national carriers for international transport. These receipts should include any other prepayment made for goods or services received in the destination country.
* PPP conversion factor (LCU per international $): Purchasing power parity conversion factor is the number of units of a country's currency required to buy the same amounts of goods and services in the domestic market as $1 would buy in the United States.
* Radios (per 1,000 people): Radios are the estimated number of radio receivers in use for broadcasts to the general public, per 1,000 people.
* Roads, paved (%): Paved roads are roads that have been sealed with asphalt or similar road-building materials.
* Telephone mainlines (per 1,000 people): Telephone mainlines are telephone lines connecting a customer's equipment to the public switched telephone network. Data are presented per 1,000 people for the entire country.
* Television sets (per 1,000 people): Television sets are the estimated number of television sets in use, per 1,000 people.
* Travel services (% of total service imports): Travel covers goods and services acquired from an economy by travellers for their own use during visits of less than one year in that economy for either business or personal purposes.
* Vehicles (per 1,000 people): Motor vehicles per 1,000 people include cars, buses, and freight vehicles but do not include two-wheelers.
Table 1: Variables selected for the Tourism model
1. ×8 - Purchasing Power Parity
2. ×4 - Industry Value Added
3. ×1 - Consumer Price Index
4. ×2 - Foreign Direct Investment
1. ×3 - Immunisation
2. ×15 - No. of hospital beds per 1000 people
1. ×9 - Roads paved (KM)
2. ×10 - Television per 1000 people
3. ×11 - Telephone lines per 1000 people
4. ×14 - No. of Vehicles per 1000 people
1. ×6 - International Tourist Arrivals (number)
2. ×5 - International Tourism Revenue
3. ×13 - Travel Services
7 The description of the variables was extracted from the World Bank variable definitions.
8 Note: Not all of the variables were included in the final model.…