Tourism is one of the increasingly important industries in many countries in
the world. Its contribution to the national income, export and employment opportunities
are important to some countries especially in countries with established tourism
destinations. At the same time, particularly in the last two decades, many countries
mainly from the developing countries has seriously developed their tourism industry.
Within the last 15 years, between 1995 and 2010, for example, international
tourist arrivals had increased from 540.6 to 935 million. During this period
international tourist arrivals grew at an average rate of 3.72% per annum. At
the same time the total tourism receipts had increased from USD 410.7 billion
to USD 919 billion or with an annual average growth of 5.52% (http://unwto.org/en).
This clearly shows the development of the global tourism industry brings a positive
impact on the economic growth and national income.
The effect of the positive contribution of this industry, many countries in
the world is taking initiatives to further enhance and stimulate the development
of their tourism industry. However, the share of international tourist arrivals
and receipts are still concentrated in the traditional markets namely the European
and American markets. However, within the last two decades the share of international
tourist arrivals and receipts began to change in which new tourism destinations,
especially the Asia and the Pacific market which recorded an increasing share,
especially the Chinese market (Salleh et al., 2011a).
Rapid increase in international tourist arrivals to new markets, particularly
to Asia and the Pacific mainly due to various factors such as tourists interest
to explore new destinations which offering new tourism products, especially
their natural environment that have not polluted and local heritages (Salleh
et al., 2007). Within this region there are abundant of relatively
untouched natural environment such as islands that are gazetted as marine parks,
mangrove areas and caves that can be promoted as attractive ecotourism destinations
(Yacob et al., 2009; Sathya
and Sekar, 2012; Biswas et al., 2011). In
addition, cheaper travel costs also encouraged more people travelling to the
area. The governments in those countries were also taking serious initiatives
to promote tourism in their respective countries.
Various strategies and incentives were given to tour operators, particularly
private operators to spur the development of their tourism industry. In Malaysia,
for example, basic facilities such as good and extensive infrastructure facilities
required by the tourists have been developed. In fact, the government tried
hard to encourage local and foreign investors to participate in the tourism
industry by providing them various attractive investment incentives (Othman
and Salleh, 2006).
Development of an integrated tourism industry requires a large sum of investment
funds. In many developing countries such investment funds are rather limited.
Many poor countries such as Indonesia, foreign direct investments are highly
needed to finance its economic development projects (Hadiwibowo,
2010). According to UNCTAD the inflow of foreign direct investment (PLA)
could boost the tourism industry, particularly upgrading the facilities and
basic services such as hotels, restaurants and recreation centers and supporting
physical infrastructure and services that are still inadequate. PLA from the
developed countries in tourism industry in the developing countries is expected
to spur the development of this industry in developing countries. However, the
inflow of PLA into the developing countries tourism sector only accounts for
about 10% of the overall PLA at the global level (UNCTAD,
2007). This means that the tourism sector is quite lagging behind in terms
of getting PLA as compared to other economic sectors such as services, manufacturing,
agriculture and others. Despite this, the tourism sector is still growing well
and even managed to surpass the growth of other sectors. Therefore, countries
that would like to develop their tourism industry, especially in the NSM should
design appropriate policies and strategies to attract the inflow of PLA that
could accelerate the development of this industry (UNCTAD,
2007). On the other hand a good development in tourism can be an attractive
factor to draw the inflow of PLA.
Economic developments as manifested by the performance of the GDP are also
expected to influence the development of tourism industry. Economic development
in the form of supplying of basic infrastructural facilities directly or indirectly
also influenced the development of the tourism industry since these facilities
and services are needed by tourists. At the same time the development of tourism
may also help the economic development of any country (Othman
and Salleh, 2010a).
Realizing the existence of interconnections between tourism development, economic growth and foreign direct investments as described above, an interesting issue here is to find empirical evidence about the possibility and the pattern of relationship between these variables. For that purpose, the study causal analysis for the ARR, GDP and the PLA is carried out. A total of 18 major international tourism destinations/countries are used as the case study.
For this purpose, cointegration analysis is utilized. In this study the number of tourist arrivals is a proxy for ARR, the gross domestic product is for GDP and the inflow foreign direct investment is for PLA. If cointegration relationship exists between the variables studied their direction of relationship can be identified.
In general, this study attempts to analyze the causal relationship between the ARR, GDP and the PLA in several major international tourist destinations. In particular the research objectives are as follows:
||Identifying the possibilities of cointegration relations between
the ARR, GDP and PLA
||Determining the direction of relationship between the ARR, GDP and PLA
Past studies: There are many studies done on causality analysis to identify the pattern of relationship between economic variables. Among the common variables studied is economic growth with other variables such as foreign direct investment (PLA), technology transfer and trade/export. Lately, tourism variable was also considered in such study. Causality analysis is also employed in various other areas such as in macroeconomic, energy, calorie demand and medicine. Table 1 represents the summary of some previous studies that discuss the relationships among the selected variables including tourism.
Major tourism destinations: The main international tourists destinations based on tourist arrivals in 2007, according to the World Tourism Organization (WTO) were led by France (80.9 million), Spain (58.7 million), United States (56.0 million), China (54.7 million), Italy (43.7 million), United Kingdom (30.9 juta), Germany (24.4 million), Ukraine (23.1 million), Turkey (22.2 million), Mexico (21.4 million), Malaysia (21.0 million), Austria (20.8 million), Russia (20.6 million), Canada ( 17.9 million), Hong Kong (17.2 million), Greece (16.2 million), Poland (15.0 million), Thailand (14.5 million), Macau (12.9 million) and Portugal (12.3 million).
Judging from above the data, it appears that tourist arrivals are not only focused on American and European destinations but has spread to other countries, especially Asian and Pacific markets. South Africa and the Middle East destinations also received attention but still are insignificant as compared to other established destinations.
This situation clearly shows that there is a good demand for new tourism destinations
and this is a good indication for the development of tourism industry in those
|| Researchers of past studies and empirical findings
|Source: Paraphrased from past studies
This condition is also supported by the awareness among the governments in
the developing countries about the potential of this industry in generating
national income and foreign exchange earnings. Various policies, strategies
and promotional travel incentives are being implemented in an effort to attract
tourists and boost the tourist service providers to be actively involved in
various sub-sectors in the industry.
Realizing limited development fund available, governments in developing countries
through various efforts, especially attractive tax reduction incentives try
to encourage the inflow of PLA from abroad particularly from developed countries
to finance their development projects including tourism projects. As indicated
by UNCTAD (2007) PLA from developed countries had helped
tourism industry infrastructure such as hotels, restaurants and recreational
facilities in developing countries. However, whether the PLA has directly influence
the economic and tourism development of the tourism industry or the economic
and tourism industry development that attracts the inflow of PLA is less empirically
verifiable. The pattern of relationships between economic growth (GNP), foreign
direct investment (PLA) and tourism development (ARR) are still unclear and
need to be proved empirically. However, based on the real data, over the years
these variables in general seemed to have similar upward trends (Salleh
and Othman, 2011; Salleh et al., 2011b).
Utilizing the ARDL approach this study attempts to identify whether there is
any relationship between them and further to determine the pattern of their
Research methodology: In determining the presence of relationship between the variables discussed and pattern of relationships of these variables, this study is using the ARDL approach and causal analysis. This method involves three steps or tests need to be implemented. The first step in applying cointegration methods is to ensure that data is purely stationary. For this purpose, the test stationarity or unit root test would be done first.
Once the data is confirmed stationary, the second step is to test for cointegration for the three variables GDP, ARR and PLA for all countries selected. Cointegration test with the ARDL approach would use in this study.
Finally, Granger causal the test is carried out to test whether there is a one-way or two-way relationship between the three variables of the study.
Unit root test: In economic analysis, unit root test is done to determine
the stationarity of time series data used. A time series data is stationary
when the mean and variance are constant over time while the autocovariance series
do not depend on time. A number of stationarity tests are often adopted in economic
studies including Dickey and Fuller (1979) and Phillips
and Perron (1988). However, this study only uses Augmented Dickey Fuller
test that is the test using a single parameter autoregression to approach the
structural error in regression tests, as shown by the following equation:
With variable ΔYt shows the unit root test using the logarithm of Y for all variables of the model (ARR, GDP and PLA) at time t. While; the variables ΔYt - 1 is the first differential lag where Δ is the symbol of differentiation. μ is the estimation error and α, β and δ are parameters to be estimated. The hypothesis tested is:
||HO: β = 0 indicates the data is stationary
||H1: β <0 indicates the data is not stationary
Cointegration: Stationary data as discussed above is important to prevent the occurrence of spurious regression that would lead the results from that regression is meaningless in economic analysis. To overcome this problem a number of studies in economics have used differentiated data (Cochrane-Orcutt method). However, an analysis using data that have been differentiated only show the short-term relationship and unable to estimate the long run relationship which is very important for policy formulations.
The above shortcomings are resolved by utilizing cointegration method as proposed
by Engle and Granger (1987), Johansen
(1988) and Johansen and Juselius (1990). Cointegration
methods allow non-stationary data to be used in economic studies and the resulting
regression is not spurious if the data is stationary after it is differentiated.
However, important pre-conditions to be met in applying the cointegration methodology
is the data must be stationary at the same level (order). This means that if
data is not stationary at the same level, the cointegration analysis cannot
Pesaran and Shin (1995), Pesaran
et al. (1996), Pesaran and Shin (1999) and
Pesaran et al. (2001) has improved this deficiency
by introducing new methods in cointegration approach that is the Autoregressive
Distributed Lag (ARDL bound test). With the ARDL bound test approach, the cointegration
tests can be performed using data with different levels of stationarity but
limited to data that stationary at level I(0) and the first difference I(1)
only (Salleh et al., 2007).
This study utilizes ARDL together with the computer package Microfit 4.0. The
long run model for estimating ARR, GDP and PLA is given in Eq.
2a-c. ARDL bound test model and Error Correction Model
(ECM) are as in Eq. 3 and 4, respectively:
Δ is the symbol of differentiation, εt is the error/residual (white noise) and α and β are parameters that to be estimated. ARR, GDP and PLA refer to the development of tourist industry, economic growth and foreign direct investment from 18 countries selected, respectively.
The test of bound test is based on the F test or Wald-statistic. F test is to test the two following hypotheses as below:
||H0: α2 = α3 = 0, no cointegration
between the variables
||Ha: α3 ≠ α2 ≠ 0, there is cointegration
between the variables
Equation for cointegration tests can also be written as follows:
Since the F-test does not have a standard distribution (non-standard distribution),
the two critical values are given for the Upper Critical Bound (UCB) and Lower
Critical Bound (LCB) (Pesaran et al., 2001).
LCB value assumes all variables are I (0), meaning there is no cointegration
between variables, while the UCB assume all variables are I (1) which means
that there is cointegration between variables. If the calculated F value is
greater than the UCB, hypothesis H0 is rejected. Therefore, there is cointegration
between the variables. The opposite is true if the calculated F value is smaller
than the LCB. This means that the hypothesis H0 failed to be rejected.
If the F calculated is between the LCB and UCB, this indicates that the results
could not be ascertained.
Engle Granger causal analysis: Engle and Granger
(1987) and Granger (1988) stated if there are two
variables of time series data that are cointegrated, then at least there is
one relationship between variables in either long run or in the short run. Error
Correction Model (ECM) can be used in analyzing the causal relationships between
variables as shown in Eq. 4, 6 and 7:
The hypothesis to be tested are as follows:
||H0: no Granger causal relationship between variables
||H1: there is Granger causal relationship between variables
Granger causal relationship is important to test whether:
||Tourism development (ARR) is the determinant to the inflow
of PLA (ARR-led PLA) or otherwise, PLA as the cause to tourism development
||Tourism development (ARR) is the determinant to GDP growth (ARR-led GDP)
or vice versa as GDP growth is the determinant to tourism development (GDP-led
||GDP is the determinant o the inflow of PLA (PLA-led GDP) or vice versa
as PLA is the determinant to GDP growth (PLA-led GDP)
RESULTS AND DISCUSSIONS
Unit root test: ADF test results are shown in Table 2 for 18 selected major countries. Results indicate that H0 is rejected at the 5% level of significance after the first difference for most of the variables in all countries. This means that all variables have unit root and is stationary at the first difference and denoted as I (1).
PLA variable from Austria, Canada, China, Germany, Greece, Hong Kong, Italy,
Spain, United Kingdom and Singapore as well as GDP of China, the United Kingdom
and the Netherlands also found to be stationary at level and denoted as I (0)
when utilizing the model with intercept and trend.
|| Results of ADF for Unit Root Test at Level and First Difference
|Note: ln GDP is the logarithm of the rate of growth of Gross
Domestic Product (GDP), ln PLA is the logarithm of PLA for Foreign Direct
Investment (PLA) and ln ARR is the logarithm of ARR for the development
of tourism industry economic growth (ARR). Numbers in parentheses is the
length of lag that are being used in ADF test (as determined from the set
of SIC to a maximum of 7) for the rejection of serial correlation in the
residuals. ADF unit root test is estimated with intercept and with intercept
and trends. *Significant at 5% level of significance (95% confidence level)
|| Results of cointegration test: Bound test approach
|Nota: Value of UCB at 5% = 5.473, Value of LCB at 5% = 4.267,
*Significant level at 5%
As a conclusion, ARR, GDP and the PLA variables in the study are stationary
at I (0) an I (1) as shown in Table 2.
Using Bound test the results of cointegration test have shown the existence of cointegration between variables for all countries as indicated by the value of F statistic which is exceeding the critical value of UCB. The details are given in Table 3.
Granger causal analysis: Results of the Granger causal analysis will
indicate the direction relationship between variables studied.
|| Pattern of relation ships between variables
There are three forms of relationships between variables that is of bidirectional
or two way relations, one-way and no relationship at all. Roughly this relationship
is illustrated in Fig. 1.
Bilateral relations between the two variables are when both variables are the
causal to one another. From Fig. 1, if there is bidirectional
relation between tourism development and economic growth, this means tourism
development is the determinant to economic growth and at the same time economic
growth is also determinants to tourism development.
|| Result of ARDL approach for Granger causality analysis
|***,**,*Significant at 1%, 5% and 10% level of significance,
One-way relationship shows one variable is the determinant to other variable
but not vice versa. In the context of tourism development and economic growth,
a one-way relationship means that tourism development in the cause of economic
growth but economic growth is not the cause of tourism development. While no
relationship indicates that each variable has no relationship with any other
According to Granger causal analysis, research findings as presented in Table 4 can be summarized in Table 5. Overall, the relationship pattern between variables is quite clear. The most important relationship is between ARR and GDP, followed by ARR and PLA relation, While the relationship between PLA and GDP is least important.
In the case of bidirectional relationship, the ARR and GDP variables are the
|| Summary of relationships pattern between ARR, GDP and PLA
Seven countries have two-way relationship namely the United Kingdom, Malaysia,
Singapore, Austria, Canada, Netherlands and Turkey. This means that the country's
tourism development and economic growth have complementary roles. The development
of tourism helps in promoting economic growth and at the same time the general
economic growth also supports the development of tourism industry. The second
important two-way relationship is for variables PLA and ARR which involve four
countries, namely France, Mexico, China and Hong Kong. While the least important
for two-way relationship is for the PLA and GDP variables that only involve
two countries, Austria and Mexico.
For one-way relationship, it is not involve many countries. In France, Germany and Italy it is found tourism development lead to economic growth in those countries. This is possibly due to tourism industry in these countries are already established and developed thus, tourism contribution to GDP is significant and help their economic growth. While in China, GDP is the determinant to ARR. This situation may be due to the country's tourism industry is still growing and requires rapid economic development to stimulate tourism industry development. Other variables that have a one-way relationship is PLA is the determinant to ARR which involve three countries, that is Austria, Canada and Malaysia and GNP as determinant to PLA also involves three countries, Germany, the USA and China.
Absence of any relationship between variables is quite clear for GDP and PLA involving 12 countries, including Malaysia. Other countries are Canada, France, Greece, Hong Kong, Portugal, Spain, Thailand, Turkey, United Kingdom, Netherlands and Singapore. This situation may be due to the value of PLA inflow into those country is relatively small as compared to their GDP. This is followed by the absence of any relationship between ARR and PLA variables that involve 10 countries, namely Germany, Greece, Italy, Portugal, Spain, Thailand, Turkey, USA, Netherlands and Singapore. This might indicate that the inflow of PLA into the country is possibly less invested in the tourism industry and at the same time the development in tourism industry is not attractive enough encourage the inflow of PLA to these countries. While the absence of relationship between ARR and GDP is only involved eight countries, China, Greece, Hong Kong, Mexico, Portugal, Spain, Thailand and USA.
This empirical study aims to examine long-run and causal relationship between
tourism development (ARR) which is proxied by tourist arrivals, foreign direct
investment (PLA) and economic growth (GDP) in selected major tourist destinations
represented by Austria, Canada, China, France, Germany, Greece, Hong Kong, Italy,
Malaysia, Mexico, Portugal, Spain, Thailand, Turkey, United Kingdom, United
States, Netherlands and Singapore.
For this purpose, cointegration analysis and causal test are done. The first step is to test unit root. The test results clearly show the data have unit root and stationary after first difference. There are also data that is stationary at level. Overall, the data set is stationary at level and the first difference and denoted as I (0) and I (1).
Since the data are stationary at different stages (at level and first difference),
the method to be utilized with such data is the ARDL cointegration approach.
In addition this approach is fit to be applied to small data samples (Salleh
et al., 2007).
Research findings prove the existence of long-term relationships between the variables studied. Due to the presence of long-term relationship, it is necessary to do causality test. Findings from the causality tests have shown there are several different types of relationships exist between the three variables.
Overall it is found that variables that have important relationships are between ARR and GDP and between ARR and PLA. While the relationship between the PLA and the GDP is not so important. The least importance of the relationship between GDP and PLA may be due to relatively the amount of PLA those countries studied is not large enough to stimulate the economic growth of the countries concerned. At the same time the economic development of those countries was not attractive enough to motivate the inflow of PLA. Similarly, the relationship between PLA and ARR is also not important. Such insignificant relationship is may be because of the PLA data available does not demonstrate a specific amount of PLA invested in the tourism industry. If there is a better and detailed data for PLA is available, the findings might be different. It can be concluded that there is a fairly strong relationship between variables ARR of GDP as compared to the relationship between variables ARR and the PLA and the relationship between the PLA and the GDP variables.