PROPOSAL SKRIPSI
DIAJUKAN UNTUK MEMENUHI TUGAS UJIAN AKHIR SEMESTER
PADA MATA KULIAH METODOLOGI PENELTIIAN
Oleh:
PENDAHULUAN
Dalam kehidupan modern saat ini pariwisata tidak lagi hanya dinikmati oleh
orang-orang tertentu, namun pariwisata saat ini menjadi salah satu pilihan bagi
kehidupan manusia. Pariwisata telah berkembang dan menjadi salah satu sektor
yang berkontribusi pada perekonomian. Ditandai dengan perkembangan jumlah
wisatawan dan kontribusi sektor pariwisata terhadap PDB. Berdasarkan laporan
Kementrian Pariwisata (Kemenpar), jumlah wisatawan mancanegara ke Indonesia
pada tahun 2018 mencapai 15,784,233 juta jiwa dan pada tahun 2019 mencapai
16,137,565 juta jiwa.
Pariwisata Indonesia yang telah dikenal di dunia dan memiliki daya saing
di pariwisata Internasional dapat menjadi salah satu sektor yang dapat memberikan
sumbangan pendapatan yang signifikan terhadap pertumbuhan ekonomi jika
dikelola dan dikembangkan dengan baik. Pertumbuhan ekonomi adalah proses
kenaikan output perkapita diproduksi dengan Produk Domestik Regional Bruto
(PDRB) perkapita (Boediono, 1985). Terdapat beberapa faktor yang mempengaruhi
pertumbuhan ekonomi antara lain : tenaga kerja, tingkat teknologi, sumber daya
alam, dan kondisi sosial budaya.
Tabel 1.1
Berdasarkan Tabel 1.1 dapat dilihat bahwa PDRB atas dasar harga konstan
tertinggi di Indonesia diduduki oleh DKI Jakarta dengan jumlah 1.736.195 miliar.
Kemudian diikuti oleh Jawa Timur dengan jumlah 1.563.759 miliar rupiah, serta
Jawa Barat dengan jumlah 1.419.689 miliar rupiah. Jawa Timur sebagai provinsi
terbesar di pulau Jawa PDRB nya menempati peringkat ke 2 secara nasional. PDRB
Jawa Timur meningkat setiap tahunnya pada periode 2015-2019 mengalami
peningkatan mencapai 1.650.143 miliar rupiah pada tahun 2019.
Sesuai dengan yang telah diuraikan pada latar belakang, maka tujuan dari
penelitian ini adalah menguji pengaruh variabel jumlah pengunjung wisata, jumlah
restoran, dan jumlah kamar hotel terhadap pertumbuhan ekonomi sektor pariwisata
di Jawa Timur.
Sistematika penulisan penelitian ini terdiri dari lima tahapan yang saling
berkaitan untuk mencapai tujuan penelitian, yaitu :
BAB 1 : PENDAHULUAN
Dalam bab ini berisi latar belakang, rumusan masalah, tujuan penelitian, dan
sistematika penulisan dalam penelitian.
Dalam bab ini berisi landasan teori dari permasalahan yang dibahas,
penelitian-penelitian terdahulu, dan kerangka berfikir teoritis.
BAB 4 : PEMBAHASAN
Dalam bab ini berisi gambaran umum penelitian, deskripsi hasil uji empiris,
analisis model, serta pembahasan hasil penelitian
Dalam bab ini berisi seluruh penjelasan yang ditarik dari hasil penelitian
dalam bentuk simpulan, dan disertai saran-saran yang diperlukan.
BAB 2
TINJAUAN PUSTAKA
Jumlah
Kamar Hotel
Wisatawan
Jumlah
Restoran
3.2 Rancangan Model
Penelitian ini menggunakan model data panel dinamis yang mengacu pada
beberapa jurnal seperti A Dynamic Panel Data Study of the German Demand for
Tourism in South Tyrol yang ditulis oleh Brida dan Risso (2009), jurnal Bilateral
Trade and Tourism Demand yang diteliti oleh Hanafiah et al (2010) dan Current
Issues in Tourism yang ditulis Habibi (2017). Berdasarkan acuan jurnal tersebut
maka penelitian ini melakukan modifikasi model dengan data yang dimiliki. Maka
model dalam penelitian ini adalah sebagai berikut :
𝑙𝑛𝑃𝐷𝑅𝐵𝑖𝑡 = 𝛽0 + 𝛽1 𝑙𝑛𝐽𝑢𝑚𝑙𝑎ℎ𝑃𝑒𝑛𝑔𝑢𝑛𝑗𝑢𝑛𝑔𝑊𝑖𝑠𝑎𝑡𝑎𝑖𝑡 + 𝛽2 𝐽𝑢𝑚𝑙𝑎ℎ𝑅𝑒𝑠𝑡𝑜𝑟𝑎𝑛𝑖𝑡
+ 𝛽3 𝑙𝑛𝐽𝑢𝑚𝑙𝑎ℎ𝐾𝑎𝑚𝑎𝑟𝑖𝑡 + 𝜀𝑖𝑡
Keterangan :
lnPDRB : Logaritma natural dari variabel Produk Domestik
Regional Bruto sektor pariwisata
lnJumlahPengunjungWisata : Logartima natural dari variabel jumlah pengunjung
wisata
JumlahRestoran : Variabel Jumlah Restoran
lnJumlahKamar : Logartima natural dari variabel jumlah kamar
𝛽1 𝛽2 𝛽3 : Koefisien/Intersep regresi
𝛽0 : Konstansta
𝜀 : Error
3.2.1 Definisi Variabel
PDRB adalah nilai total seluruh output akhir yang dihasilkan oleh daerah
baik penduduk di dalam daerah tersebut maupun penduduk di luar daerah tersebut.
Dalam artian umum PDRB didefinisikan sebagai nilai barang dan jasa yang
diproduksi di suatu wilayah dalam periode waktu tertentu. Variabel PDRB dalam
penelitian ini adalah PDRB atas dasar harga konstan pada 38 Kabupaten/Kota di
Provinsi Jawa Timur periode 2014-2018.
c) Jumlah Restoran
3.2.2 Hipotesis
4.1 Teori/Konsep
Dalam penelitian ini terdapat keterbatasan dalam teori dan konsep yaitu
penelitian ini hanya menjelaskan konsep mengenai pengertian dari pariwisata dan
pertumbuhan ekonomi.Secara teori tidak dijelaskan mengenai pengaruh sektor
pariwisata secara langsung namun secara empiris terdapat pengaruh besar, sektor
pariwisata terhadap pertumbuhan ekonomi.
4.2 Model
Metode GMM yang digunakan dalam penelitian ini hanya melihat
hubungan searah antara pariwisata dan PDRB sektor pariwisata. Namun pada
dasarnya apabila ketersediaan data terpenuhi maka dapat digunakan metode Panel
Granger Causality untuk melihat hubungan dua arah antara pariwisata dan PDRB
sektor pariwisata.
4.3 Data
Dalam penelitian ini hanya menggunakan variabel Produk Domestik
Regional Bruto, jumlah pengunjung wisatawan, jumlah kamar hotel, jumlah
restoran. Namun pada dasarnya terdapat banyak variabel yang dapat digunakan
sebagai indikator untuk mengetahui hubungan dan pengaruh antara pariwisata dan
PDRB sektor pariwisata meliputi investasi, penerimaan pariwisata, pengeluaran
pariwisata, dan travel cost.
Abumound, I., Ibrahim, A., & Hijawi, L. (2019). Estimating the Economic Impact
Baltagi, B. H., Bratberg, E., & Holmås, T. H. (2005). A panel data study of
physicians’ labor supply: The case of Norway. Health Economics, 14(10),
1035 1045. https://doi.org/10.1002/hec.991
Brida, J. G., & Risso, W. A. (2009). A Dynamic Panel Data Study of the German
Demand for Tourism in South Tyrol. Tourism and Hospitality Research,
9(4), 305–313. https://doi.org/10.1057/thr.2009.15
Chang, C. L., & Mcaleer, M. (2012). Aggregation, heterogeneous autoregression
and volatility of daily international tourist arrivals and exchange rates.
Japanese Economic Review, 63(3), 397–419.
Firdaus, M. (2011). Aplikasi Ekonometrika untuk Data Panel dan Time Series.
Bogor : IPB Press
Gujarati, D.N., & Porter, D.C. (2009). Basic Econometrics (5𝑡ℎ 𝐸𝑑). New York :
McGraw-Hill Irwin
Habibi, F. (2017). The determinants of inbound tourism to Malaysia: a panel data
analysis. Current Issues in Tourism, 20(9), 909–930.
https://doi.org/10.1080/13683500.2016.1145630
Hanafiah, M.H.M., Harun, M.F.M., & Jamaluddin, M.R. (2010). Bilateral Trade
https://jdih.kemenkeu.go.id/fullText/2009/10TAHUN2009UU.HTM.
Witt, S.F & Witt, C.A. (1995). Forecasting tourism demand: A review of empirical
© 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313
www.palgrave-journals.com/thr/
Brida and Risso
total number of beds in the hospitality structures Table 2: Origin of the tourist in the South Tyrol in
is 21, revealing a small-scale business structure. the last two years (in percentage)
The occupancy rate of beds increased from
Country 2005/06 2006/07
25 per cent in 1970 to 40 per cent in 2007.
The output of the tourism industry in South Germany 47.80 46.90
Tyrol increased from less than 5 per cent of Italy 36.90 37.00
the region’s GDP in 1970 to 12 per cent in Austria 2.60 2.70
2007. As a consequence, the South Tyrolean Switzerland 3.60 3.70
economy is strongly dependent on tourism. Benelux 3.20 3.50
Income from tourism industry has exhibited Other countries 5.80 6.20
constant growth during the last 40 years. The Source: ASTAT, 2008.
increase in overnight stays in the same period
was from 11 million to more than 27 million.
In the last 15 years, the rate of growth of 47.1 per cent of total trips. The second place
the number of overnight stays was around is taken by the airplane, increasing the share in
1 per cent but in the last 30 years there was a the period 2002–2007.
decrease in the average number of overnight Table 2 shows the percentage of German
stays from 9 to 5 days. This can be explained tourists in South Tyrol in the last two touristic
by an almost linear increase of tourist arrivals years. It represents 47.8 and 46.9 per cent
to the region from less than 500 000 in 1970 (2005/06 and 2006/07 respectively) of the total
to more than 5 million in 2007. In average per tourists, and 75.87 and 74.44 per cent of the
year, around 30 000 persons are employed in total international tourists measured as over-
the tourism sector, accounting for more than nights stays.
12 per cent of total employments in South Table 3 shows that, most of the German
Tyrol, where women are 58 per cent of the tourists arrive from Bayern. Germans from
total employees in the sector. Bayern are the main tourists in the Dolomites,
Most of the tourists arriving in South Tyrol Aurina and the other regions, both in winter
comes from Germany. During the last years, and summer seasons. The second place is taken
German tourists represented around 50 per by Nordrhein-Westfalen.
cent of the total tourists and more than 74 per In this study, a dynamic demand model
cent of the total international tourists measured proposed by Arellano and Bond of the German
as overnights stays. Tourism from Germany has demand for tourism in South Tyrol is intro-
been constantly increasing in the last 20 years. duced. The model is used to identify the main
Muñoz (2007) asserted that like most European determinants of the demand and their relative
holidaymakers, Germans prefer to use their contribution to the volume of German in
own cars for travelling. Table 1 shows that in tourism to South Tyrol. Using a dynamic
2007, the percentage share of travels by car was model, we study the impacts of the demand
306 © 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313
Dynamic panel data study of the German demand for tourism
Table 3: German tourists of 2004/05 according to destination, season and German region
for tourism in the previous period on actual panel data model to estimate short-run and
demand. We show that the past consumption long-run elasticities for tourists visiting the
behaviour has a positive and relevant effect on province of South Tyrol. Panel data analysis
current demand, reflecting loyalty of the has rarely been used in previous empirical
German tourists. The results of this paper may research. Panel data permit the use of a
give valuable help for South Tyrolean policy- complete database in order to explain the influ-
makers in the decision-making process. ence of several variables in decisions made by
The rest of the paper is organised as follows. tourists. The possibility of using a large number
The next section provides a brief literature of observations provides more degrees of
review and discussion of the methodologies freedom in the estimation process. This
used by previous studies. The subsequent approach also reduces the problem of multicol-
section describes the data and the econometric linearity and provides more degrees of freedom
methods used for estimation. The penultimate in the model estimation. Panel data also allow
section presents the empirical results and their a better representation of adjustment dynamics.
interpretation. Some concluding remarks are Dynamic panel methodologies, based on both
including in the final section. cross-section and time series data, have the
advantage that they take account of all the
LITERATURE REVIEW information relating to the data set under
Tourism demand has been studied from consideration. In spite of its advantages, the
different points of view, using both quantita- panel data approach has rarely been applied to
tive and qualitative methods. As pointed in the tourism demand analysis. Song and Li (2008)
very recent review (Song and Li, 2008), the in their recent review of the published studies
majority of the published studies used quanti- on tourism demand modelling and forecasting
tative methods, in particular time-series and show only four exceptions in the post-2000
econometric models. This study employs a literature. (1) Ledesma-Rodrýguez et al (2001)
© 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313 307
Brida and Risso
used the panel data method to estimate short- each of the 17 Spanish destinations for the
run and long-run elasticities for tourists visiting period 1991–2003 to examine German demand
the island of Tenerife. (2) The paper by Naude for tourism in Spain. This paper is very close
and Saayman (2005) uses both cross-section to our study, both for the methodology and
data and panel data for the period 1996–2000 the results. It also coincides. As for South Tyrol,
to identify the determinants of tourism arrivals most of the international tourism arriving in
in 43 African countries, taking into account Spain comes from Germany. Garín-Muñoz
tourists’ country of origin. (3) Roget and (2006) is similar to the previous paper. In this
Gonzalez (2006) employ panel data models to case the application of the GMM-DIFF
examine the demand for rural tourism in dynamic model uses the panel structure of the
Galicia (Spain). Finally, Sakai et al (2000) used data set for the 15 most important countries of
the panel data approach to analyse the effects origin of tourists over the period 1992–2002
of demographic change on Japanese people’s and other models are also shown for compar-
travel propensity. We extended the review of ison. The results of the study suggest that
Song and Li to the last two years finding six tourism demand to Canary Islands must be
more papers modelling tourism demand by considered as a luxury good and is highly
using panel data models. For instance, the dependent on the evolution of relative prices
paper by Kuo et al (2008) investigate the and cost of travel between origin and destina-
impacts of infectious diseases including Avian tion country. Finally, also the study by Garín-
Flu and severe acute respiratory syndrome Muñoz and Montero-Martín (2007) uses a
(hereafter SARS) on international tourist dynamic model estimated with a panel data set
arrivals in Asian countries using panel data to identify and measure the impact of the main
procedures. Eilat and Einav (2004) used a determinants of the inbound international
generalisation of the multinomial logit model tourism flows in the Balearic Islands.
and panel data set to estimate the demand of Less attention has also been paid to the
international tourism. Eugenio-Martin et al dynamics of tourism demand systems. Most of
(2008) introduce a cross-sectional time series the models are static regressions, not permitting
model which deals with autocorrelation and to consider time as a fundamental variable of
heteroskedasticity in the error term that is the process. We specify dynamic econometric
applied to a worldwide panel data set of desti- models for modelling short-term impacts as
nation countries for tourists from Australia, well as long-term dynamics. The observed
France, Germany, Japan, Spain, the United significance of the lagged dependent variable
Kingdom and the United States to examine in such models indicates that time is likely to
whether economic development of a particular be a necessary variable of the model specifica-
destination is relevant for tourism demand. tions. (Fujii and Mak, 1981; Garin-Muñoz,
Khadaroo and Seetanah (2007) use panel data 2006) Although the importance of including
to investigate the importance of transport on explicit dynamic adjustments in demand anal-
the demand for tourism of Mauritius, showing ysis has generally been recognised, specific
that international tourists are particularly sensi- research in tourism demand using dynamic
tive to the island’s transport infrastructure. systems is not abundant, and only a few authors
Maloney and Montes Rojas (2005) use dynamic have addressed these issues in empirical studies.
panel data analysis of tourist flows from eight Some papers have tried to find alternative
origin countries to 15 Caribbean destinations. tourism demand specifications including
In the paper by Garín-Muñoz (2007), the previous consumption in the model (Witt and
dynamic model GMM-IFF proposed by Martin, 1987; Ledesma-Rodrýguez et al, 2001;
Arellano and Bond is applied to a panel data Garin-Muñoz, 2006). These models permit to
set consisting of inbound German tourism in handle the dynamic structure of preferences.
308 © 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313
Dynamic panel data study of the German demand for tourism
According to Garin-Muñoz (2006), there are major advantage when compared with the
two main reasons for including previous utilisation of time series. Tourism demand is a
consumption as an explanatory variable of the function of several variables that are to be
model: first, there is less uncertainty associated selected, together with the particular form of
with a previous country where you travel than the function. Once the variables and the func-
an unvisited country; second, people with tional form of the demand are selected, we
knowledge about the visited destination talk have to choose the estimation procedure to
about their holidays reducing the uncertainty determine the parameters of the model.
for potential visitors. If people are satisfied with
the destination there is a larger probability of Variables
coming back and tell other people about their In this paper, the tourism demand is measured
experiences. Hence, the coefficient of the in terms of the number of overnights stays of
lagged dependent variable may be interpreted international visitors in hotels and apartments
as a measure of habit formation and inde- in each one of the 116 destinations of South
pendent preferences. In the present study we Tyrol. We use this variable to represent
consider past consumption as an explanatory German tourism demand. This is motivated by
variable and some traditional demand variables. the fact that more than 60 per cent of the total
It means we consider income of the German number of international tourists corresponds to
tourist and price-type factors. From a theoretical Germans. This is true for all the period covered
point of view, the demand for tourism is a in this study. The function of several variables
function of the quantity of German tourism defining the tourism demand gives causal
demanded in the past, the relative cost of living relationships between the tourism demand
between the Italy region and Germany, the price (dependent) variable and its influencing factors
of travel and the income of the tourists. (explanatory or independent variables). The
literature suggests several possible independent
ECONOMETRIC METHODOLOGY variables to model tourism demand. The most
AND DATA SET commonly used are income, relative prices,
The objective of this paper is to introduce a exchange rates and transportation costs but
model to investigate the German demand for variables of time trend, marketing expenditure
tourism in South Tyrol. We have in mind two and special events are also included (Song and
main properties that our model must verify: it Li, 2008). The use of these variables, however,
must be a dynamic model and employ a panel depends on the availability of data and its quan-
data set. The model utilised in this study is the titative measurement, the non-existence of
dynamic data panel model proposed by collinearity between explanatory factors and
Arellano and Bond (1991) and we apply it to the form of the model. This paper chose
a panel data set collected from 116 different income, relative prices and transportation costs
tourism destinations of South Tyrol. The avail- as explanatory or independent variables to
able data consist of 2436 observations: the investigate how adequately these variables can
annual overnights stays of international tourists explain the German demand for tourism in
in 116 different tourism destinations of South South Tyrol.
Tyrol (that coincides with 116 communes of
the province) from 1987 to 2007. Using a Data
panel allows us to use annual data and avoids Data for our study were selected based on data
the problems of seasonality. The availability of availability, reliability of data sources, ability to
panel data will allow us to measure the effects represent and theoretical considerations. The
of variables with little changes within cities and data utilised in this study are annual time series
more variability across cities. This represents a data for the period 1987–2007. Data for the
© 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313 309
Brida and Risso
number of overnights stays of international which t and i are the time and destination-
visitors in hotels and apartments in each one specific effects, respectively. The error
of the 116 destinations of South Tyrol are component i,t is assumed to be serially uncor-
taken from the Provincial Statistics Institute related with zero mean and independently
(ASTAT) of Bolzano. Income is represented distributed across destinations, but heteroske-
by the German per capita GDP at constant dasticity across time and destinations is allowed
price of 2000 and the relative prices were esti- for. Moreover, i,t is assumed to be uncor-
mated using the ratio between the consumer related with the initial condition ln Qi,t, for
index price of Italy and Germany. Both series t = 2, 3, …, T, and with the individual effects i
were obtained from the International Mone- with for any t.
tary Fund (IMF). Several studies sustain that When a model for panel data includes lag
German tourists prefer to travel to the North dependent explanatory variables, the simple
of Italy by car and this is the motivation to estimation procedures are asymptotically valid
proxy transportation costs by the average of only when there is a large number of observa-
Dated Brent, West Texas Intermediate and the tions in the time dimension (T); in our case
Dubai Fateh price of crude oil. This time series T = 21 is not enough. The current available
was also obtained from the IMF. response to this problem (Holtz-Eakin, 1988;
Arellano and Bond, 1991; Hsiao, 2003) is to
Model specification first difference the model and use lags of the
The tourism demand function is expressed in dependent variable as instruments for the lagged
the form: dependent variable. In the present study we
used the generalised method of moments
Qi ,t = f (Qi ,t -1 , PTt , PCOt ,GDPt )
(GMM-DIFF) framework proposed by Arellano
(1)
and Bond (1991). As the model is in differ-
ences, the problems stemming from possible
where Qi,t is the number of overnight stays
non-stationarity are avoided and this method
destination i during the year t; PTt is the rela-
will generate consistent and efficient estimates
tive cost of living of German tourist in Italy;
of the parameters of interest and standard
PCOt is the price of crude oil and GDPt is
errors. The dynamic model to be estimated
the real per capita gross domestic product
will therefore be
in Germany. Equation (1) is a theoretical rela-
tionship among the variables under considera- lnQi ,t = b1lnQi ,t −1 + b 2 lnPTt + b3 lnPCOt
tion. In practice, we need to specify the + b4 lnGDPt + e i ,t (3)
functional form of the model. In the present
study the tourism demand model has adopted
where i = 1, …, 116; t = 1987, …, 2007; and all
the double-logarithmic form. Then the model
the variables are in first differences.
to be estimated would be equation (2);
Because of the double-logarithmic form of
however, the economic time series generally
the model, the parameters may be interpreted
present unit roots and then the model to
as elasticities. The parameter 1 indicates to
be estimated is in differences as shown in
what degree current tourism purchase is deter-
equation (3):
mined by the value of previous consumption.
As it is a dynamic model, the estimated coef-
ln Qi ,t = a + b1lnQi ,t -1 + b 2 lnPTt + b3 lnPCOt
ficients are the short-run elasticities. Long-run
+ b4 lnGDPt + lt + mi + e i ,t (2) elasticities can be obtained by dividing each of
the coefficients by (1 − 1).
In equation (2), i,t = t + i + i,t is the fixed As Garín-Muñoz and Montero-Martín
effects decomposition of the error term, in (2007) assert, one of the advantages of using a
310 © 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313
Dynamic panel data study of the German demand for tourism
dynamic model is that both short-run and as in the linear model the coefficients are
long-run elasticities may be obtained. A further significant and the signs are the expected.
advantage relates to the fact that, by differ- The coefficients are the short-run demand
encing data, we avoid the problem of non- elasticities; to obtain the long-run elasticities
stationarity and this method will give us we divide each of the estimated coefficients by
confidence in the reported coefficient and (1 − 1). The long-run elasticity values are
standard errors. income elasticity = 1.92, cost of living = − 4.36
and price travel elasticity = − 0.19.
EMPIRICAL RESULTS FOR Notice the significance of the coefficient of
SOUTH TYROL the lagged dependent variable. This significance
This section presents the estimation of equation could be explained by the habit of persistence
(3) using the GMM-DIFF estimator of Arel- and/or the word-of-mouth effects.
lano and Bond and the linear regression to The estimated coefficient for the income vari-
compare. We have used STATA v.9.0 econo- able has the expected sign. The long-run income
metric software to estimate the model. The elasticity (1.92) means that tourism in South
hypothesis of no second-order autocorrelation Tyrol is a luxury for the German tourist.
in the errors is tested as the methodology The short-run ( − 1.05) and long-run
assumes. We conducted a test of autocorrela- ( − 4.36) price elasticities indicate that the
tion and the Sargan test of over-identifying tourism in the region is elastic, being sensitive
restrictions as derived by Arellano and Bond to a variation of prices. This may be the reflec-
(1991). Failure to reject the null hypothesis in tion of numerous alternatives in the region
both tests gives support to the model. (Austria, Switzerland and Southern Germany).
The empirical results in Table 4 show that Consequently, care must be taken by the
the model performs satisfactorily. Note that industry to maintain or improve price compet-
itiveness. In contrast to the cost of living, the
Table 4: Estimation results for the linear and dynamic
short-run ( − 0.04) and long-run ( − 0.19) cost
model (1987–2007) of travel elasticities suggest that tourism is
inelastic to variations in the cost of travel.
Variable GMM-DIFF estimator of
Arellano Bond CONCLUSIONS
ln Qi,t − 1 0.7588
This article expands significantly upon previous
− 34.74 studies of Croes and Vanegas (2005), Garín-
ln GDPt 0.4639 Muñoz (2006, 2007) and Garín-Muñoz and
− 2.19 Montero-Martín (2007). Their contribution to
ln PTt − 1.0529 the literature is the application of a dynamic
( − 4.83) econometric modelling of international tourism
ln PCOt − 0.0458 demand to estimate a dynamic panel data model
( − 3.54) for the German demand of tourism in South-
Tyrol. The dynamic methodologies used in
Autocorrelation 2 0.14, P=0.886 this study are based on both cross-section and
Sargan (df) 115.88, P=0.357
time series data, having the advantage of taking
Wald test 28300000
into account all the information relating to the
No. observations 2204 data set under consideration. The model used
Dependent variable: Logarithm of the number of overnights in this paper provides short- and long-run elas-
stays (ln Qi,t ). t-statistics in parenthesis. The estimates are ticities. The study shows that the long-run
obtained by using the instruments ln Qi,t lagged up to seven income elasticity of 1.92 indicates that tourism
periods. in South Tyrol is a luxury and the short-run
© 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313 311
Brida and Risso
( − 1.05) and long-run ( − 4.36) price elasticities case of the South Tyrolean region’. We thank
indicate that the tourism in the region is very the helpful comments of anonymous referees.
sensitive to variation of prices. The latter may A preliminary version of this paper was
reflect the numerous alternatives in the region. presented at the conference ‘Advances in
In particular, Austria, Switzerland and Southern Tourism Economics’ (ATE2009), Research
Germany offer similar and substitute products. Centre in Tourism, Innovation and Services
South Tyrol should give importance to main- (CITIS) of Universidade Lusiada de Lisboa,
tain or improve price competitiveness. Promo- 23–24 April 2009.
tion of other kinds of tourism in alternative to
mountain tourism to diversify the supply would REFERENCES
also be very advisable. From the other hand, Arellano, M. and Bond, S. (1991) Some tests of spec-
the elasticities of the cost of travel are negative ification for panel data: Monte Carlo evidence
and smaller than one in absolute value, both and an application to employment equations.
in the short-run and long-run ( − 0.04 Review of Economic Studies 58: 277–297.
and − 0.19, respectively). Then the cost of ASTAT. (2007) Tourism in South Tyrol: Year
travel has a negative impact on the German 2005/06. Bolzano, Italy: ASTAT.
ASTAT. (2008) Tourism in South Tyrol: Year
demand for tourism in South Tyrol. Given that
2006/07. Bolzano, Italy: ASTAT.
the majority of the German tourists travel by Croes, R.R. and Vanegas, M. (2005) An econometric
car, the South Tyrolean authorities can improve study of tourist arrivals in Aruba and its implica-
the results of the industry by giving incentives tions. Tourism Management 26(2): 879–890.
to travel by low cost airlines to the airport of Dritsakis, N. (2004) Cointegration analysis of
Bolzano or by subsidising part of the costs of German and British tourism demand for
travelling by car. The elasticities estimated Greece. Tourism Management 25: 111–119.
in this study are in line with the results of Eilat, Y. and Einav, L. (2004) Determinants of
previous studies (see Dritsakis, 2004; Croes and international tourism: A three dimensional
Vanegas, 2005; Garín-Muñoz, 2006, 2007 and panel data analysis. Applied Economics 36(12):
Garín-Muñoz and Montero-Martín, 2007). 1315–1327.
This study should be of significant assistance Eugenio-Martin, J.L., Martín-Morales, N. and
Sinclair, M.T. (2008) The role of economic
to clarify the application and benefits of model-
development in tourism demand. Tourism
ling in analysing tourism demand in small Economics 14(4): 673–690.
regions. Consequently, the study should help Fujii, E. and Mak, J. (1981) Forecasting tourism
in developing economic policies for the demand: Some methodological issues. Annals
promotion of tourism. The tools used in this of Regional Science 15(2): 359–380.
paper can be applied to estimate the German Garín-Muñoz, T. (2006) Inbound international
demand of tourism in other regions of Italy. tourism to Canary Islands: A dynamic panel data
The results of these studies could be used to model. Tourism Management 27(2): 281–291.
compare with South Tyrol. The research can Garín-Muñoz, T. (2007) German demand for
also be generalised by including the prices of tourism in Spain. Tourism Management 28(1):
alternative destinations as a new variable of the 12–22.
model to test if there are substitute markets. Garín-Muñoz, T. and Montero-Martín, L. (2007)
Tourism in the Balearic Islands: A dynamic
model for international demand using panel
ACKNOWLEDGEMENTS data. Tourism Management 28(2): 1224–1235.
Our research was supported by the Compe- Holtz-Eakin, D. (1988) Testing for individual
tence centre in Tourism Management and effects in autoregressive models. Journal of
Tourism Economics (TOMTE) of the Free Econometrics 39: 297–308.
University of Bolzano, project: ‘Tourism, Hsiao, C. (2003) Analysis of Panel Data, 2nd edn.,
growth, development and sustainability. The Cambridge: Cambridge University Press.
312 © 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313
Dynamic panel data study of the German demand for tourism
Khadaroo, J. and Seetanah, B. (2007) Transport Naude, W.A. and Saayman, A. (2005) Determinants
infrastructure and tourism development. of tourist arrivals in Africa: A panel data regres-
Annals of Tourism Research 34(4): 1021–1032. sion analysis. Tourism Economics 11: 365–391.
Kuo, H.I., Chen, C.C., Tseng, W.C., Ju, L.F. and Roget, F.M. and Gonzalez, X.A.R. (2006) Rural
Huang, B.W. (2008) Assessing impacts of tourism demand in Galicia, Spain. Tourism
SARS and Avian Flu on international tourism Economics 12: 21–31.
demand to Asia. Tourism Management 29(5): Sakai, M., Brown, J. and Mak, J. (2000) Population
917–928. aging and Japanese international travel in the
Ledesma-Rodriguez, F.J., Navarro-Ibanez, M. and 21st century. Journal of Travel Research 38:
ez-Rodrýguez, J.V. (2001) Panel data and 212–220.
tourism: A case study of Tenerife. Tourism Song, H. and Li, G. (2008) Tourism demand
Economics 7: 75–88. modeling and forecasting – A review of recent
Maloney, W.F. and Montes Rojas, G.V. (2005) research. Tourism Management 29(2): 203–220.
How elastic are sea, sand and sun? Dynamic Witt, S. and Martin, C. (1987) Econometric models
panel estimates of the demand for tourism. for forecasting international tourism demand.
Applied Economics Letters 12(5): 277–280. Journal of Travel Research 25(3): 23–30.
© 2009 Palgrave Macmillan 1467-3584 Tourism and Hospitality Research Vol. 9, 4, 305–313 313
World Applied Sciences Journal 10 (Special Issue of Tourism & Hospitality): 110-114, 2010
ISSN 1818-4952
© IDOSI Publications, 2010
Mohd Hafiz Mohd Hanafiah, Mohd Fauzi Mohd Harun and Mohd Raziff Jamaluddin
Abstract: Tourism industry can be claimed as one of the important sectors in the modern Malaysian economic
development. It has been identified as second largest foreign exchange earning sector and had strengthen up
the performance of Malaysia Gross Domestic Product and trade balance. Numbers of studies have been done
in order to attempt the theory that there is a link between trade and tourism and most of them showed that it
is positively correlated. This paper specifies tourism demand models in Malaysia and estimates demand
equations using tourist inflow data for the period 1997-2008. Researchers identified that bilateral trade,
population, income, tourism price and geographical distance between Malaysia and the Asian countries are
the main determinants of tourism. It clearly shows that tourism demands relates significantly with the trades.
With more proactive action taken by the policymaker in understanding the dynamic of this industry, Malaysia
will able to increase the quality of tourism industry in order to be the main center for tourism Asia region and
the world beyond 21st century.
Corresponding Author: Mohd Hafiz Mohd Hanafiah, Universiti Teknologi Mara, Malaysia.
Tel: +603 5543 5699, Fax: +603 5543 5698, E-mail: hafizhanafiah@salam.uitm.edu.my.
110
World Appl. Sci. J., 10 (Special Issue of Tourism & Hospitality): 110-114, 2010
developed and estimated in tourism literature to identify TRADE = The bilateral trade;
the relationship between tourist arrivals in a particular POP = The total population in tourist generation
country and the factors that influence arrivals. In countries;
international tourism demand modeling, most of the DIS = The geographical distance.
previous studies have used the demand function
approach to identify the quantitative [5]. Some studies Empirical literature reviews on tourism demand
have used the gravity model approach. However, the recommend that tourist demand is measured as in tourist
gravity model approach lacks a firm theoretical foundation arrivals and departures, with tourist arrivals as the
and the final form of the model closely resembles the dependent variable. On the other hand, empirical models
demand function [6]. of tourism demand use consumer’s income, the price of
The gravity-type models are applied in physics and services, exchange rate and distance as explanatory
have achieved increasing recognition in the analysis of variables [2, 5, 10, 11].
economic phenomena related to the flow of goods and/or Most of the empirical studies used an autoregressive
services [8]. This type of model has yet to be applied, model in demand estimation to evaluate the determinants
with a proper theoretical base to tourism demand. of international tourism demand. The econometric model
Although some studies may claim to have estimated a was formulated to evaluate the international tourism
variant of the gravity model, omitted variables from the demand by residents of Argentina, Brazil, Colombia and
model and incorrect estimation technique bias the results’. Venezuela to Aruba [12].
There are sets of potential determinants that can Additionally, from a research on Caribbean tourism
influence the decision to travel classified into the demand, the empirical results identified income as the
following categories: socioeconomic factors, such as, most significant variable in explaining tourism demand
income level, relative prices between the origin and the [7]. Referring to other researchers who concluded that
destination places, demography, urbanization and length forecasting Portugal tourism demand is important in
of the leisure time; technical factors related to easier order to analyze the country’s economy position and the
communications and transport facilities; psychological econometric methodology showed that co-integration
and cultural factors reflecting personal preferences and is non-existent in Portuguese tourism demand [12]. The
the style of life of the potential travelers; and random authors concluded that communication for transportation
factors related to unexpected events, like political infrastructures, destination risk, bilateral trade and
instability, weather conditions, natural disasters, epidemic geographical distance are the main determinants of
diseases, etc [7]. tourism.
Furthermore, the above variables are correlated with
immigration and most of the immigrants prefer home- Econometric Model: A common model used in tourism
country products [6]. For that reason, this phenomenon demand studies is a single equation model with demand
can reduce the trading transaction costs. In particular, explained by the income of tourists in their countries of
tourism can be thought as a form of temporary origin, the cost of tourism in the destination and
international migration. It clearly shows that inbound alternative destinations and the substitute price variable.
tourist could promote the tourism of their host country As to explanatory variables, the definitions of income
with their family and friends [9]. Not only that, used are usually personal disposable income or gross
international trade theory is simply international domestic product of the country of origin. In this study,
localization theory. Therefore, most of the economists income is measured by industrial output value, due to
linked tourism through an input-output matrix [6]. difficulties in obtaining figures for monthly GNP and
From the literature, it indicates that the demand for population, or for per capita income and even disposable
travel export is equal to: income for the originating countries. Therefore, the
researchers employed real income, cost of living,
TA: f (GDP, RP, TRADE, POP, DIS) exchange rate, arrival from origin country, inflation rate
and dummy variable (economic crisis) as the independent
Where variables.
TA = The number of tourist arrivals; The dependent variable used is the number of
GDP = The income in tourist generation countries; arrivals at destination (i) from origin (j) in a given year
RP = The relative price; (TA). Data set from The World Development Indicators
111
World Appl. Sci. J., 10 (Special Issue of Tourism & Hospitality): 110-114, 2010
and World Statistical Yearbook and The Tourism log(RPim) = log [CPIm / CPIi ]
Compendium and Yearbook of Tourism Statistics have
been used for the explanatory variables and dependent Where CPIm is the consumer price index in
variable respectively. Malaysia and CPIi is the consumer price index of origin
The researchers converted the GDP per capita in country. The ratio will give the relative price of tourism
tourist countries into logarithm modes in order to find out and influences the decision for the tourist to decide
whether tourism demand will be influenced by income in whether to travel to Malaysia or not. Meanwhile,
the country of origin of the tourists or not. B Based on the logarithm between the geographical distance and the
empirical literature, the researchers expected for a positive partner country were compute in order to find out whether
sign. Income in the origin country is the most frequently the cost of transports will moderate the tourist arrivals or
used explanatory variable. Most researchers have relied not. A negative sign is expected for both variables [10,
on national income and GDP as measures for income in 11].
the origin. This paper considers that income is highly
significant as a determinant of demand. Other than that, Model Specification: This paper also emphasizes on
this paper is also to find out whether a trade partner is an international trade either it can brings positive impact on
important vehicle to expand tourism or not. Malaysia tourism demand or not. From the recent
A study on the determinants of inbound tourism in literature review, it shows that tourism demand was a
Laos found a positive sign when they used this proxy dynamic phenomenon. Therefore, this study uses a panel
[13]. data set between Malaysia and the Asian countries
(China, Japan, Thailand, Indonesia and United Arab
Log TRADE = Xt + Mt Emirates) for the period of 1997-2008. The proposed panel
GDPm + GDPi data applies Newton’s gravitational law and the gravity
models are based on the gravity law of spatial interaction.
Where The gravity models utilized the gravitational force
Xt = The annual exports of Malaysia to the concept as an analogy in order to explain the volume of
country of origin of each tourist at time t; trade, capital flows and migration of people among the
origin at time t; countries in the world. These models have been accepted
Mt = The annual import of Malaysia from each in tourism studies [14]. Several researchers modified the
tourist’s country of origin at time; Tinbergen Gravity Model for the purpose of estimating
GDPi = The Gross Domestic Product per capita in tourism demand. A slight adjustment has been made by
tourist countries; converting the variables in order to suit the tourism
related variables [10, 15].
This research has been conducted in order to Nearly 80 of 100 empirical studies on tourism
discover whether population changes will affect the demand or arrivals used single equation of linear and/or
number of immigration from a particular country or not log linear equation [2]. As a result, estimated coefficients
and according to the literature on forecasting tourism can be interpreted as elasticities. Therefore, the variables
demand, we expected a positive sign. The result shows such as log(GDPm), log(RPim), log(TRADEim), log(DIS im)
that the level of tourism depends not only on the and log(POPim) are treated as endogenous variables in
population of the origin country, but also on the this study. The log-linear model based on the modified
immigrants from the destination country. Therefore, Gravity Model is as follows:
population of the origin country is also part of
explanatory variable [7]. log(TAim) = á + $1 log(GDPm) + $2 log(RPim) + $3
Nevertheless, most studies do not consider this log(TRADEim) + $4 log(DISim) + $5 log(POPim)
variable because population tends to be highly + ui
correlated with income and at the same time will lead
towards multicollinearity and autocorrelation problems. Where
The researchers then manipulate Macroeconomic á = Constant
variables (tourism price) and distance (transportation TAim = The number of tourist arrivals in Malaysia
cost) to analyze their influence on the cost of tourist GDPim = The gross domestic product
service such as: RPim = The relative price
112
World Appl. Sci. J., 10 (Special Issue of Tourism & Hospitality): 110-114, 2010
TRADEim = The bilateral trade Table 1: The Impact of Trade on Malaysian Tourism Demand
POPim = The population Variables Coefficient
DISim = The distance LogGDP 0.582**
ui = The error LogRP -0.626**
LogTRADE 0.118**
Findings: From the analysis, it is found that the LogDIS -0.572**
explanatory power is quite high (Adjusted R2=0.738). LogPOP 0.620**
All the explanatory variables are significant (LogGDP, Adj. R2 0.738
LogTRADE, LogDIS and LogPOP at 95%) confidence N 84
interval, with the exception of LogRP. Gross Domestic Durbin Watson 0.674692
Product is statically significant, with an expected nR2 = P2 61.992
positive sign. ** - Statistically significant at 5% confidence intervals
The variables LogTRADE and LogRP are statistically
significant with positive relationship with tourism The critical value is then opted from Chi-square table
demand. This outcome really shows that trade partners and since the nR2 value of 61.992 is lower than the 5%
and relative price are crucial vehicle in order to inflate critical value, therefore we can accept that no
tourism demand. The coefficient of LogDIS (Distance) heteroskedasticity problem exists.
is negative, as expected. This result proves the Next, the variance inflation factor (VIF) has been
justification of gravitational model and explains that high conducted in order to find out any Multicollinearity
traveling cost really reduce the tourism demand. Finally, problem with the variables. To test whether the variables
the variable population (LogPOP) finds a positive sign. have Multicollinearity problem, the variance inflation
A series of t-test have been applied in each
factor (VIF) had been conducted. If the VIF result is
independent variable against the dependent variable.
less than 4, it indicates that there is no serious
From the above gravity model equation, the variables
Multicollinearity problems occurred. If the VIF result is
log(GDPm), log(RPim), log(TRADEim), log(DISim) and
more than 4, it indicates that the research model has
log(POPim) share the same significant level of 95 %.
serious Multicollinearity problems and the models need
Taken as a whole, all independent variables are
to be re-checked.
significant towards explaining the dependent variables
and it shows that there are significant relationships
By entering VIFPF scalar formula, we will get:
between these variables. In other words, there are
significant positive relationships between tourism
VIF = 1 / (1 - 0.6579)
demands with income, trade and population. On the other
hand, there are significant negative relationship between VIF = 2.93 < 4
tourism demand and distance and also tourism price.
F-test has been used in order to verify the As the VIF value is less than 4 (rule of thumb),
relationships between the independent variables there is supporting evidence that TAim is in absence of
(price, income, distance, population and economic crisis) Multicollinearity.
as a group and the dependent variable (tourism demand)
[10]. As indicated in Figure 1, the F-value (18.0327) is CONCLUSION
more critical than the F-distribution value. Therefore,
one can reject the null hypothesis and concludes with This paper investigates the relationship between
95% confidence interval that the independent variables bilateral trade and tourism demand in Malaysia. Any
as a whole explained a significant changes in tourism analysis of international tourism demand and the
demand. In a simple word, income, distance, population, sensitivity of economic parameters in particular must be
tourism price and bilateral trade as a group have given viewed with caution. There may be some unknown bias
significant effect either positive or negative towards in the derived demand parameters due to the use of a
tourism demand. small sample and that was the reason why the researchers
White’s test was applied to eliminate used a modified gravity model. Thus, in order to
heteroskedasticity problem by treating the data as understand the dynamic nature of tourism, the researcher
degrees of freedom equal to the number of slope constructed a log linear econometric model and the
coefficient excluding the constant in the regression [11]. research findings confirmed the above assumption.
113
World Appl. Sci. J., 10 (Special Issue of Tourism & Hospitality): 110-114, 2010
The logged tourism demand showed an expected positive 3. Kulendran, N. and K. Wilson, 2000. Is there a
sign while other explanatory variables such as relationship between international trade and
geographical distance and tourism price are statistically international travel? Applied Economics.
significant. Moreover, the results were aligned with key 4 World Travel and Tourism Council (WTTC), 2006.
economic theories. These results prove that an augment Malaysia: Travel and Tourism Climbing to New
technique is needed in order to understand more about Heights. World Travel and Tourism Council. London.
the demand for tourism determinants. Furthermore, this 5. Turner, L.W. and S.F. Witt, 2001. Factors influencing
study generated new information on the affects and demand for international tourism: Tourism demand
sensitivity of economic parameters and also their analysis using structural equation modelling.
influence over demand for international tourism. As the Tourism Economics, 7(1): 21-38.
quality of data for international travel and tourism 6. Witt, S. and C. Witt, 1995. Forecasting Tourism
increases and more data become available, the Demand: A Review of Empirical Research.
methodology and the database developed in this study International J. Forecasting, 11(3): 447-475.
can be extended to obtain more vigorous parameter 7. Carey, K., 1991. Estimation of Caribbean Tourism
estimates. Basically, recommendations for further study Demand: Issuers and Methodology. Atlantic
were only intended in order to give guidelines for Economic J., 19(3): 32-40.
improvement on this research paper and for better 8. Wang, S. Yu., 2008. The impact of crisis events and
understanding on the discussed topic. macroeconomic activity on Taiwan’s international
As a conclusion, this report meets its objectives by inbound tourism demand. Tourism Management,
conducting an exploratory research and identifying pp: 1-8.
key economic indicators that have significant effects 9. Dwyer, L., I. Burnley, P. Murphyand and P. Forsyth,
towards tourism industry development and performance 1992. Tourism - Immigration Interrelations. Report
measures for future applications. Further, this research Prepared for Bureau of Immigration Research.
forms an understanding of the multitude of economic 10. Hanafiah, M.H. and M.F. Harun, 2010. Application of
factors and issues that affect the nation’s Gravity Model in Estimating Tourism Demand in
competitiveness and the effective operation of the Malaysia. Proceedings of 2010 International
government’s strategies in the global tourism arena. Conference on Business, Economics and Tourism
Additionally, to the investors and businessmen exploring Management, Singapore, 12(2): 121-129.
good potential investment opportunities in Malaysia, 11. Hanafiah, M.H. et al., 2008. Estimating Tourism
this report offers a glimpse into various significant Demand in Malaysia. Proceedings of 7th Asia Pacific
tourism related economics that local companies can take Forum for Graduate Students’ Research Tourism,
advantage on current and future world economic Malaysia, 1(3): 141-146.
condition. Lastly, this paper has amply satisfied it aims 12. Daniel, A. and F. Ramos, 2002. Modelling Inbound
and at the same time, a similar study should be done using International Tourism Demand Portugal. International
a bigger sample, more variables and current data in order J. Tourism Res., 4(1): 193-209.
to strengthen this study. 13. Phakdisoth, L. and D. Kim, 2007. The Determinants
of Inbound Tourism in Laos. ASEAN Economic
REFERENCES Bulletin, 24(2): 225- 237.
14. Tinbergen, J., 1962. Shaping the World Economy.
1. World Travel and Tourism Council (WTTC), 2006. New York: Twentieth Century Fund.
Malaysia: Travel and Tourism Climbing to New 15. Rodrigue, J.P., 2004. The geography of transport
Heights. World Travel and Tourism Council. London. systems. London: Routledge.
2. Lim, C., 1977. Review of International Tourism
Demand Models. Annals of Tourism Res.,
24(4): 835-849.
114