See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/288690645

The socio-economic determinant of tourism demand in Turkey: A panel data

approach

Article  in  International Research Journal of Finance and Economics · November 2010

CITATIONS READS

18 1,042

2 authors:

Simal Gormus İsmet Göçer

Centennial College Aydın Adnan Menderes University
7 PUBLICATIONS   42 CITATIONS    69 PUBLICATIONS   215 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by İsmet Göçer on 03 February 2016.

The user has requested enhancement of the downloaded file.

International Research Journal of Finance and Economics
ISSN 1450-2887 Issue 55 (2010)
© EuroJournals Publishing, Inc. 2010
http://www.eurojournals.com/finance.htm

The Socio-Economic Determinant of Tourism Demand in

Turkey: A Panel Data Approach

Şakir Görmüş
Sakarya University, Faculty of Economics and Administrative Sciences, Turkey
E-mail: [email protected]
Tel: +90-506-6128525

İsmet Göçer
Adnan Menderes University, Faculty of Economics and Administrative Sciences, Turkey
E-mail: ı[email protected]

Abstract

During the last several years a large number of economist and institution have
expressed increasing concerns regarding to importance of tourism income to sustain current
account deficit and create jobs. Given the high contribution of tourism sector in Turkish
economy, this study attempts to investigate the socio-economic determinant of international
tourism demand in Turkey. After the several diagnostic tests, we concluded that the two-
way random effect model is best for our estimation. To accomplish this objective the two-
ways random effect models were carried out for annual time series from 2000 to 2006 for
32 countries. The first contribution of this paper is to extent the previous studies related to
Turkish tourism demand using additional explanatory variables and including more
countries in panel data set. The second contribution of this paper is to analyze the effect of
the official visits of Turkish President and Prime Minister (to sending countries) on Turkish
tourism demand. To our best knowledge, this will be the first study to investigate the effect
of the official visits of President and Prime Minister on tourism demand.The results from
the two-ways random effect models show that real income of sending countries, trade value
between sending countries and Turkey and accommodation capacity are positively related
with tourism demand. Distance between sending countries and Turkey is negatively related
with tourism demand. However, contrary to expectations, relative prices and reel exchange
rate have positive relationship with tourism demand. If it is thought that Turkey offers
cheap holiday packages due to its low cost of living this result may not be surprise.
Competitive prices between Turkey and Greece, Spain, Egypt are negatively related with
tourism demand as expected. 2001 Turkish financial crisis has positive and 2003 Gulf War
II has negative effect on tourism demand. However, even if it has right sign, official visits
of Turkish President and Prime Minister were not significant in explaining tourism
demand.

Keywords: Tourism demand, panel data estimation

JEL Classification Codes: L83, C23
International Research Journal of Finance and Economics - Issue 55 (2010) 89

1. Introduction
In the 21th century, tourism becoming an important and the fastest growing sector for many countries
after the telecommunication and information sectors (Crouch-Ritchie, 1999). Growing tourism sector
has huge potential to generate income, investment, employment and foreign exchange. According to
the World Tourism Travel Council (WTTC, 2010), the contribution of travel & tourism sector to GDP,
total employment and total investment are 9.8%, 8.6% and 9.8%, respectively. Because of the
increasing importance of international tourism sector in terms of its contribution to the national
economy, during the last several years there have been made vast theoretical and empirical studies
about the determinant of tourism demand, especially for countries, which has great dependence on
tourism sector.
To understand the importance of the tourism sector for the Turkish economy over the past
several years, it should be mentioned that the number of tourism arrival (foreign tourism receipts)
increased from 7.9 million (US$ 5.9 billion) to 18.1 million (US$ 24.1 billion) peoples between years
1996–2005 and the ratio of foreign tourism receipts to export (trade deficit) is 25% (56%) in year
20051. At the same time, according to WTTC, the Turkish tourism sector provide employment to 1.7
million peoples in 2008 and expected to increase 2.5 million peoples in 2010.
There are many studies in the literature, which investigate the effect of selected variables2 on
international tourism demand. Studies have attempted to model tourism demand using three different
models3 (Regressions model, Co integration model and Almost Ideal Demand System (AIDS) model).
However, in this field, there are limited studies related to Turkey4.
The contribution of this paper is two-fold. The first contribution of this paper is to expand the
database using new variables and including more countries. Previous studies related to Turkish tourism
demand used several variables (real income, relative prices, exchange rates, population,
accommodation capacity and some country specific dummies) and maximum 18 countries included in
panel data. This paper use additional explanatory variables5 as literature suggest and including 32
countries6. The second contribution of this paper is to analyze effect of the official visits of Turkish
President and Prime Minister on Turkish tourism demand. To our best knowledge, this is the first study
to investigate the effect of the official visits of President and Prime Minister on tourism demand.
Therefore, the main objective of this study is to investigate the effect of selected variables on
international tourism demand in Turkey over the 2000 – 2006 periods by utilizing the several two-way
random effect models.
The rest of the paper is organized as follows. Section 2 outlines the importance of the tourism
sector in Turkish economy. Section 3 provides a brief literature review about variables and data
sources. Section 4 explains the specification of the tourism demand function to estimate. Section 5
presents the results from the two-way random effect model estimations. The final section concludes.

1
Turkish Ministry of Culture, www.kultur.gov.tr.
2
Real GDP, relative prices, the cost of transportation, the exchange rate, accommodation capacity, competitive prices,
trade value and country specific dummies, etc.
3
For a detailed survey of literature, see Syriopoulos and Sinclair (1993), Crouch (1994), Witt and Witt (1995), Kulendran
(1996), Lim (1997), Kim and Song (1998), Kulendran and Witt (2001), Song, Witt and Jensen (2003), Kumar (2004), Li,
Song and Witt (2005), Proença and Soukiazis (2005), Ouerfelli (2008), Song and Li (2008).
4
Uysal and Crompton (1984), Var, Mohammed and Icoz (1990), Akis (1998), Icoz, Var and Kozak (1998), Akal (2004),
Halıcıoğlu (2008), Aslan, Kula and Kaplan (2009).
5
Trade value between host and sending countries, distance between host and sending countries, marketing budget,
competitive (substitute) prices between Turkey and Greece, Egypt, Spain, Gulf War and European Union member’s
dummies.
6
Australia, Austria, Azerbaijan, Belgium, Bulgaria, Canada, China, Denmark, Finland, France, Germany, Georgia, Greece,
Iran, Ireland, Israel, Italy, Japan, Kazakhstan, The Netherlands, New Zealand, Norway, Poland, Portugal, Romania,
Russia, South Korea, Spain, Sweden, Switzerland, UK and USA.
90 International Research Journal of Finance and Economics - Issue 55 (2010)

2. The Importance of Tourism Sector in Turkish Economy

Tourism is an important sector for Turkish economy and its growth and benefits to economy has been
impressive in recent years. To see the importance and share of tourism sector in Turkish economy, this
section provides some important statistics of international tourism in Turkey. Table 1 shows the share
of the tourism sector in the Turkish economy.

Table 1: Some International Tourism Statistics of Turkey

Years Number of Tourist Tourism Share of Tourism Ratio of Tourism Ratio of Tourism
Arrivals Receipts Receipts in GDP Receipts to Export Receipts to Trade
(Thousand) (Million) (%) (%) Deficit (%)
1996 8 614 5 650 3,2 17.6 55
1997 9 689 7 008 4,2 21.8 47
1998 9 752 7177 2.9 23.3 51
1999 7 464 5 193 2.1 17.9 53
2000 10 412 7 636 2.9 24.8 35
2001 11 569 8 090 5.1 23.3 241
2002 13 247 8 481 5.2 20.9 133
2003 14 030 9 677 4.3 25.2 98
2004 17 517 12 125 4.1 23.2 70
2005 21 124 13 929 3.8 23.2 55
2006 19 819 12 553 3.2 18.1 41
2007 23 341 13 990 2.8 16.2 40
2008 26 379 16 761 3.0 16.0 41
Sources: www.tursab.org.tr, www.kultur.gov.tr and www.evds.tcmb.gov.tr

International tourist arrivals (receipts) to Turkey increased from 8.6 million ($5.6 billion) in
1996 to 26.3 million ($16.7 billion) in 2008. Since 1996, visitor arrivals and tourism earning have
grown except 1999 and 2006. Decline of 1999 and 2006 can be explained both internal and external
factors such as 1998 Russian financial crisis, 1999 Turkish earthquake, increase in terrorism, bird-flu
and murder of priest Andrea Santoro in 2006.
Share of Tourism receipts in GDP changes from 2.8 % (2007) to 5.2% (2002). In addition, ratio
of tourism receipts to export changes from 16 % (2008) to 25.2 % (2003). Finally, ratio of tourism
receipts to trade deficit changes from 35 % (2000) to 241 % (2001). Given that, tourism receipts are a
main source of trade deficit finance and have a significant contribution to GDP. The importance of
tourism receipts for Turkish economy can be seen more clearly after 2001 financial crisis.

3. Literature Review and Data

There are too many both survey7 and applied8 papers in the literature to analyze the effect of selected
socio-economics variables on tourism demand. Those studies have been used a broad variety of
econometric techniques (OLS, VAR, ARDL, AIDS, etc.), data set (time series or panel data) and
variables (GDP, relative prices, the cost of transportation, the exchange rate, accommodation capacity,
competitive prices, trade value and country specific dummies, etc.) to explain tourism demand.
Tourism Demand: The tourism demand literature shows that there is several measurements for
international tourism demand such as: the number of the tourist arrivals, the number of nights spent by
tourist or the receipts from tourism. According to Song and Li (2008), the number of tourist arrivals is
still the most popular measurement in tourism demand studies. The main reason for this choice has

7
Crouch (1994), Witt and Witt (1995), Lim (1997), Li, Song and Witt (2005) and Song and Li (2008).
8
Uysal and Crompton (1984), Var, Mohammed and Icoz (1990), Syriopoulos and Sinclair (1993), Kulendran (1996), Kim
and Song (1998), Icoz, Var and Kozak (1998), Kumar (2004), Halıcıoğlu (2004), Proença and Soukiazis (2005), Ouerfelli
(2008), Aslan, Kula and Kaplan (2009)
International Research Journal of Finance and Economics - Issue 55 (2010) 91

been easy availability of tourist arrivals data. In this study, tourism demand is measured in terms of
number of tourist arrivals from an origin country to destination country (Turkey).
Real Income (GDP): In terms of the major factors influencing the demand for tourism, income
and prices are the most commonly used variables (Lim, 1997). Tourism is a normal good. As peoples’
income increase, they are more willing to travel abroad. Therefore, it is expected that an increase in
incomes may cause an increase in demand for tourism. Because of easy availability of data most
studies have used real (per capital) personal income or real gross domestic product (GDP) as measure
for income in sending countries. In this study, the real gross domestic products (GDP) of sending
countries are used as a proxy for national income.
Relative Prices and Competitive Prices: Before making any decision the most of the tourists
compare the cost of living at the tourist destination country relative to the origin country and possibly
to alternative destination countries. An increase in the price level of the destination country relative to
sending country discourages tourists to travel to this place or reallocate their demand to other relatively
cheaper alternative tourism destinations. Therefore, two types of relative prices have to be considered
in the demand function of international tourism: the first one is relative price between the destination
and the sending country; the second one is relative prices between the destination and different
competing countries (the substitution price effect). The selection of alternative destinations was limited
to Greece, Spain and Egypt. They are considered as alternative destinations due to their geographic and
cultural similarities to Turkey.
The relative price variable, which is normally used in the demand for tourism function, is the
ratio of the consumer price indexes between the destination and the sending countries adjusted by the
exchange rates. The competitive price variable is used in the demand for tourism function is the ratio of
the consumer price indexes between the destination and the alternative destinations countries adjusted
by the exchange rates. Both prices should be inversely related to tourism demand, as higher prices in
the destination country relative to sending countries, as well as in alternative destinations, are likely to
influence the visitor’s decision whether or not to travel and/or where to travel. In this study, same
definitions for relative prices and competitive prices are used.
Trade Value and Distance: In last decade, increasing number of studies9 have used gravity
model framework to explain tourism demand. These models include trade value and distance variables
between sending and destination countries in tourism demand function. Studies try to attempt that there
is a positive link between trade and tourism demand. They argued that trade can influence tourism
demand in two ways. First, bilateral trade can make home-country product more preferable. Second, it
can reduce transportation cost between sending and destination countries. In this study, bilateral trade
between the destination and the sending country are used.
Even if transportation cost is a main determinant of tourism demand, there has been less
attention on transportation cost in the literature due to measurement problem and data unavailability.
Durbarry (2000 and 2008) and Khandaroo and Seetanah (2007) used gravity model framework to
explain tourism demand (Archibald and LaCorbiniere and Moore, 2008). In addition to well-known
explanatory variables, the distances between origin and destination countries have been employed in
gravity model as a proxy for transportation cost. It is expected that as the distance between sending and
destination countries (transportation cost) increases, tourism demand will decrease. In this study,
distance between major cities of the destination and the sending country are used. Inclusion of trade
value and distance variables make this model the gravity-based tourism demand model.
Accommodation Capacity and Marketing Expenditure: Supply conditions of the destination
country are important to attract more tourists. Even if there are several supply factors, accommodation
capacity and marketing expenditure are the most used in the literature due to easy availability of the
data. In this study, the advertisement expenses of Turkish Minister of Culture and the number of beds

9
Kulendar and Wilson (2000), Archibald and LaCorbiniere and Moore (2008), Phakdisoth and Kim (2008) and Leitao
(2010).
92 International Research Journal of Finance and Economics - Issue 55 (2010)

available to tourists who visit Turkey measure marketing expenditure and accommodation capacity,
respectively. An increase in the supply factors are expected to increase tourism demand.
Real Exchange Rate (REER): The real exchange rate changes may have a significant effect
on decision of international tourists. If the price of foreign currency decreases (travel become cheaper)
then, more peoples are willing to travel and tourism demand increases. In generally, because of easy
comparison tourists use the real exchange rates as a proxy for destination prices. A depreciation in the
real exchange rate means that it is now cheaper to visit the destination; as a result, it is likely to be
positively associated with tourism demand.
Population: It is obvious that the tourism demand is related to the population of the sending
counties (the amount of potential customer). It is expected that as the population of sending countries
(the amount of potential customer) increase tourism demand will increase too.
Dummies: We can expect that February 2001 Turkish economic crisis, EU membership of
sending countries, Gulf War II in March 2003, the terrorist attacks on the USA in September 11 2001
and official visit of Turkish President and Prime Minister to sending countries had some effect of
Turkish tourism demand. These events are incorporated in our model using dummy variables.
A dummy variable was used to capture the effects of the EU member countries. The dummy
variable takes the value of one if sending countries are EU member and zero otherwise. In addition,
economic crises can be a major deterrent for tourists. Hence, a dummy variable is used to account for
the February 2001 economic crisis. The dummy variable takes the value of one in year 2001 and zero
otherwise. It is expected that both dummy variables have positive effect on Turkish tourism demand.
There has been huge depreciation on Turkish currency after the February 2001 economic crisis.
Therefore, Turkey becomes relatively cheaper and tourism demand expects to increase.
A dummy variable was included in order to evaluate the effect of the Gulf war in March 2003.
The dummy variable takes the value of one in year 2003 and zero otherwise. September 11 2001 could
also deter potential tourists around the world and Turkey and a dummy variable for September 11 is
accounted for. The dummy variable takes the value of one in year 2002 and zero otherwise. It is
expected that both dummy variables have negative effect on Turkish tourism demand.
Finally, a dummy variable was included in the model to capture the effect of official visits of
Turkish President and Prime Minister to sending countries. It is assumed that when Turkish President
or Prime Minister visits a country there will be more news on TV and newspaper about Turkey, hence
more peoples willing to visit Turkey. We assumed that most of the tourists prefer to summer for
vacation and decisions on holidays are generally planned. Therefore, if Turkish President or Prime
Minister visits to sending countries in first three months of the year, the dummy variable takes the
value of one in same year and zero otherwise. if Turkish President or Prime Minister visit to sending
countries in the rest of the year, the dummy variable takes the value of one in the next year and zero
otherwise.
The choice of explanatory variables based on a review of literature, special features of Turkey
and data availability. Table 1 shows selected variables, expected signs, data sources and references.
Data for all the variables are obtained from various statistical sources.
International Research Journal of Finance and Economics - Issue 55 (2010) 93
Table 2. Selected Variables, Expected Signs, Data Sources and References

Variables Expected Data Sources10 References

Signs
Tourist Arrivals BTS, www.kultur.gov.tr [1], [3], [4], [5], [6], [8], [9], [13]
GDP + World Bank’s WDI [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11],
[12], [13], [14]
Trade Value + IMF Direction of Trade [6], [8], [12]
Statistics
Distance - CEPII, www.cepii.fr [5], [6], [8]
Accommodation Capacity + BTS, www.kultur.gov.tr [2], [4], [7], [11]
Relative Prices - Calculated IMF data [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12],
[13], [14]
Population + World Bank’s WDI [4], [6]
Marketing Expenditure + BTS, www.kultur.gov.tr [13], [15]
Real Exchange Rate - World Bank’s WDI [1], [3], [5], [6], [14], [15]
Competitive Prices - Calculated IMF data [3], [5], [10], [11], [12], [13]
Official Visit Dummy + Official Newspaper of
Government
2001 Crisis Dummy +
2003 Gulf War Dummy - [10]
EU Dummy +
September 11 - [7], [10]
2001Dummy
[1] Katafona and Gounder (2004), [2] Proença and Soukiazis (2005), [3] Walsh (1996), [4] Zhang and Jensen (2007), [5]
Archibald, LaCorbiniere and Moore (2008), [6] Leitao (2010), [7] Aslan, Kula and Kaplan (2009), [8] Phakdisoth and Kim
(2007), [9] Akış (1998), [10] Song, Witt and Li (2003), [11] Ouerfelli (2008), [12] Habibi and Rahim (2009), [13] Uysal
and Crompton (1984), [14] Lee, Var and Blaine (1996), [15] Chummi (2001)

4. Methodology
In this study, the data set consist of annual panel data for 32 countries for a period of 7 years (2000–
2006). The use of annual data is preferable in the literature in order to avoid the seasonality problems,
which are dominant in tourism sector and panel data including more countries, and years, which has
greater acceptability in terms of generalized results. According to Hsiao (2003) and Balgati (2001), the
utilization of panel data set enables us to have more degrees of freedom, efficiency, controls to omitted
variable bias and reduces the multicolinearity problem, hence improving the accuracy of estimation.
In terms of the estimation techniques, the log-linear two-way random effect model will be
employed. There are two major advantages in using the log-linear two-way random effect models.
First, estimated coefficients can be interpreted as elasticities. Second, Breush Pagan LM tests results
show that two-way random effect model is best fitting in this data set.
Gravity base models to the analysis of Turkish international tourism demand can be written in
log-linear form as:
ln TAi,t = α i + β1 lnYi,t + β 2 ln TVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 D01i,t + ε i,t (1)
ln TAi,t = α i + β1 lnYi,t + β 2 ln TVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 ln Popi,t + ε i,t (2)
ln TAi,t = α i + β1 lnYi,t + β 2 ln TVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 ln MEi,t + ε i,t (3)
ln TAi,t = α i + β1 lnYi,t + β 2 ln TVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 ln CPGREi,t + ε i,t (4)
ln TAi,t = α i + β1 ln Yi,t + β 2 ln TVi,t + β3 ln Disi,t + β 4 ln ACi,t + β5 ln RPi,t + β 6 ln CPEGPi,t + ε i,t (5)
lnTAi,t = α i + β1 lnYi,t + β 2 lnTVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 ln CPSPAi,t + ε i,t (6)
ln TAi,t = α i + β1 lnYi,t + β 2 ln TVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 ln REERi,t + ε i,t (7)

10
BTS: Bulletin of Tourism Statistics, WDI: World Development Indicators, CEPII: French Research Center in
International Economics.
94 International Research Journal of Finance and Economics - Issue 55 (2010)

ln TAi,t = α i + β1 lnYi,t + β 2 ln TVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 D02i,t + ε i,t (8)

ln TAi,t = α i + β1 lnYi,t + β 2 ln TVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 D03i,t + ε i,t (9)
ln TAi,t = α i + β1 lnYi,t + β 2 ln TVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 D04i,t + ε i,t (10)
lnTAi,t = α i + β1 lnYi,t + β 2 lnTVi,t + β 3 ln Disi,t + β 4 ln ACi,t + β 5 ln RPi,t + β 6 D06i,t + ε i,t (11)
Where, TAi,t is tourist arrivals (or demand) from country i to Turkey at time t;
Yi,t is real income of sending country at time t;
TVi,t is trade value between sending country and Turkey at time t;
Disi,t is distance between sending country and Turkey;
ACi,t is accommodation capacity of Turkey at time t;
RPi,t is relative price between sending country and Turkey at time t;
Popi,t is population of sending country at time t;
MEi,t is marketing expenditure of Turkish Minister of Culture at time t;
CPGREi,t is competitive price between Turkey and Greece at time t;
CPEGPi,t is competitive price between Turkey and Egypt at time t;
CPSPAi,t is competitive price between Turkey and Spain at time t;
REERi,t is reel effective Exchange rate at time t;
D01i,t is dummy variable to capture official visit of Turkish President and Prime
Minister to sending country;
D02i,t is dummy variable to capture February 2001 economic crisis;
D03i,t is dummy variable to capture March 2003 Gulf War;
D04i,t is dummy variable to capture EU member countries;
D05i,t is dummy variable to capture 2001 September 11 events;
εi,t is the error term.

5. Empirical Results and Analysis

Panel data techniques are used to estimate the demand function of tourism in Turkey. In panel data,
choosing the right econometric technique is crucial to get consistent and efficient estimates. A number
of diagnostic tests were implemented in order to evaluate the best estimation method and to reach the
most acceptable estimation results. If all of the country specific effects are constant and equal across
the countries and there are no the time specific effects, then ordinary least squares (OLS) estimation
method is consistent. However, panel data may includes country (individual) effects, time effects or
both of them and these effects can be either fixed effects or random effects. A fixed effect model is not
applicable in our panel data set because of short time dimension (7 years) and long cross-section
dimension (32 countries).
First, it has to be tested that which model (OLS or Random Effects) would be appropriate for
our panel data set estimation. Hypothesis test for Random Effects model against OLS model can be
written as:
H 0 : σ µ2 = σ λ2 = 0
H1 : σ µ2 ≠ 0 or σ λ2 ≠ 0 or both of them
The null hypothesis assumes that there are no random country specific effects or random time
specific effects. If random country specific effects (σ µ2 ≠ 0) or random time specific effects (σ λ2 ≠ 0) are
present, and then we can conclude that random effects model is better than OLS model.
This hypothesis can be tested using the Breusch-Pagan LM test (Balgati, 2001 and Erlat, 2006)
as:
2 2

LM =
 N (
NT  ∑i =1 ∑t =1 uˆ it
T 2
)− 1

 +
 T (
NT  ∑t =1 ∑i =1 uˆ it
N 2
)− 1

 ≈ χ 2 (1) (12)
2(T − 1)  ∑ N ∑T uˆ it2  2( N − 1)  ∑ N ∑T uˆ it2 
 i =1 t =1   i =1 t =1 
International Research Journal of Finance and Economics - Issue 55 (2010) 95

If LM exceeds the critical value of the Chi-square with one degree of freedom, we reject the
null hypothesis in favor of random effects. The LM test consist of two parts which first part tests
random country specific effects, second part tests random time specific effects and finally both part
together tests combine country specific effects and random time specific effects.

Table 3: Test Results for Random Effects versus OLS model

Time Effect Individual Effect Time and Individual Effect

Breusch-Pagan LM test 1,6 552,5 554,05
Model 1
P-value 0,2 0,001 0,001
Breusch-Pagan LM test 1,6 552,45 554,05
Model 2
P-value 0,2 0,001 0,001
Breusch-Pagan LM test 1,78 558,76 560,54
Model 3
P-value 0,18 0,001 0,001
Breusch-Pagan LM test 2,2 561,98 564,18
Model 4
P-value 0,14 0,001 0,001
Breusch-Pagan LM test 1,47 556,15 557,62
Model 5
P-value 0,22 0,001 0,001
Breusch-Pagan LM test 1,47 556,15 557,62
Model 6
P-value 0,22 0,001 0,001
Breusch-Pagan LM test 2,14 561,59 563,73
Model 7
P-value 0,14 0,001 0,001
Breusch-Pagan LM test 1,36 555,12 556,49
Model 8
P-value 0,24 0,001 0,001
Breusch-Pagan LM test 1,99 560,49 562,48
Model 9
P-value 0,15 0,001 0,001
Breusch-Pagan LM test 1,45 551,05 552,51
Model 10
P-value 0,22 0,001 0,001
Breusch-Pagan LM test 1,55 554,05 551,51
Model 11
P-value 0,23 0,001 0,001
Individual effect: (σ µ2 ≠ 0) , Time effect: (σ λ2 ≠ 0) and Time and Individual effect (two-way random effect): (σ µ2 ≠ 0) and (σ λ2 ≠ 0) .
P-values are in bold number shows that null the hypothesis is rejected in favor of random effect models

Results for Breusch-Pagan LM tests are presented in table 3. LM value for all individual
random effect models and individual and time random effects (two-way random effects) models are
greater than the critical Chi-squares. Therefore, two-way random effect models will be preferred to
OLS model in our estimation.
Second, it has to be tested that error terms are correlated across cross-section units (countries).
Zeller (1962) proposed that if error terms were correlated across cross-section units, then Seemingly
Unrelated Regressions (SUR) model would be appropriate for our panel data set estimation (Greene,
2003). Hypothesis test for Random Effect model against SUR model can be written as:
H 0 :Cov(ε k ,i ,t , ε k , j ,t ) = 0
H 1 :Cov(ε k ,i ,t , ε k , j ,t ) ≠ 0 at least for one i ≠ j
The null hypothesis assumes that there is no correlation between error terms of cross-section
units. If correlations across cross-sections are present, then we can conclude that SUR model is better
than Random Effect model. This hypothesis can be tested using the Breusch-Pagan (1980) LM tests
(Kónya, 2006) as:
N i −1
λ = T ∑ ∑
i= 2 j =1
rij2 (13)

Where rij is correlation coefficient between countries i and j at time t. N is number of countries.

Table 4: Test Results for Random Effects versus SUR model

Model Model Model Model Model Model Model Model Model Model Model
1 2 3 4 5 6 7 8 9 10 11
LM test 0,61 0,47 0,45 0,44 0,44 0,44 0,46 0,5 0,42 0,52 0,48
P-value 0,43 0,48 0,49 0,5 0,5 0,5 0,49 0,47 0,51 0,46 0,45
P-value for all is lowers than 0.01. Therefore, null the hypothesis is accepted in favor of random effect models.
96 International Research Journal of Finance and Economics - Issue 55 (2010)

Results for Breusch-Pagan LM test are presented in table 4. LM values for all are lower than
the critical Chi-squares and P-value for all are lower than 0.01. Therefore, null the hypothesis is
accepted in favor of random effect models. Two-way random effect models will be preferred to SUR
model in our estimation too.
Above tests results showed that, the two-way random effect model is appropriate empirical
technique to estimate tourism demand in Turkey. The two-way random effect model estimation
requires two conditions. First, number of variables in the model (including constant) cannot be more
than time dimension of data (7 years). Second, number of variables in the model (including constant)
cannot be more than cross-section (number of countries).
Therefore, maximum six explanatory variables may include in the tourism demand model. To
see the effect of broad variety of explanatory variables on tourism demand, eleven different tourism
demand function were estimated. The table 5 gives the results from the two-way random effect
estimation models by using the panel Estimated Generalized Least Squares (EGLS). Real GDP,
relative prices, accommodation capacity, trade value and distance variables are used all 11 models. The
adjusted R-squared (R2) values, as measures of goodness of fit, are between 0.36–0.62.

Table 5: Two-Way Random Effect Models

Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11
C 6,50 5,79 9,14 -24,30 -9,63 -23,32 6,79 6,20 5,52 5,92 6,10
(2,80) (1,73) (2,15) (-1,42) (-1,45) (-1,44) (2,69) (2,80) (1,67) (1,88) (2,72)
lnY 0,19 0,19 0,19 0,20 0,20 0,19 0,21 0,17 0,19 0,20 0,19
(2,90)* (2,99)* (3,03)* (3,17)* (3,25)* (3,17)* (3,31)* (2,63)* (3,16)* (3,08)* (2,96)*
lnTV 0,23 0,20 0,22 0,20 0,19 0,20 0,17 0,28 0,20 0,21 0,22
(4,08)* (3,10)* (3,43)* (3,04)* (2,92)* (3,06)* (2,41)** (4,63)* (3,10)* (3,24)* (3,68)*
lnDis -0,81 -0,87 -0,82 -0,83 -0,84 -0,83 -0,85 -0,79 -0,83 -0,84 -0,82
(-4,80)* (-5,18)* (-5,11)* (-5,24)* (-5,25)* (-5,24)* (-5,88)* (-4,74)* (-5,22)* (-4,55)* (-4,86)*
lnAC 0,70 0,79 0,84 3,19 2,03 3,11 0,69 0,69 0,80 0,77 0,74
(3,97)* (3,00)* (3,50)* (2,33)** (3,77)* (2,40)** (3,54)* (4,10)* (3,08)* (3,26)* (4,30)*
lnRP 0,0009 0,0009 0,0009 0,0009 0,001 0,0009 0,0009 0,0009 0,0009 0,0009 0,0009
(3,08)* (3,41)* (3,34)* (3,45)* (3,43)* (3,45)* (3,32)* (3,06)* (3,43)* (3,36)* (3,07)*
D01 0,018 - - - - - - - - - -
(0,47)
lnPop - 0,05 - - - - - - - - -
(0,91)
lnME - - -0,23 - - - - - - - -
(-1,09)
lnCPGRE - - - -1,43 - - - - - - -
(-1,79)***
lnCPEGP - - - - -0,79 - - - - - -
(-2,56)**
lnCPSPA - - - - - -1,37 - - - - -
(-1,84)***
lnREER - - - - - - 0,00004 - - - -
(1,73)***
D02 - - - - - - 0,09 - - -
(2,07)**
D03 - - - - - - - - -0,15 - -
(-
1,85)***
D04 - - - - - - - - - -0,05 -
(-0,17)
D05 - - - - - - - - - - -0,02
(-0,55)
Adjusted R2 0,62 0,38 0,43 0,36 0,47 0,37 0,53 0,63 0,39 0,42 0,62
F Statistic 61,48 22,71 27,99 21,01 32,97 21,68 41,48 63,30 23,76 26,64 61,50
Observations 224 224 224 224 224 224 224 224 224 224 224
The values in parentheses are t-ratios (heteroskedasticity corrected). (*), (**), (***), Indicates that the estimated coefficient
is statistically significant at the 1%, 5% and 10% significance level, respectively
International Research Journal of Finance and Economics - Issue 55 (2010) 97

The results in Table 5 show that real GDP of sending countries, trade value and accommodation
capacity (distance) have a positive (negative) and highly significant effect in all 11 models as expected.
An increase in these variables will increase (decrease) international tourism demand in Turkey. The
estimated values of the income elasticity suggest that the economic conditions of tourists who visit
Turkey are important factor in determining tourism demand in Turkey. Income elasticities are less than
one (between 0.17–0.21), indicated that tourism travel is not a luxury good. In addition, the coefficients
of trade value and distance indicate that trade is an important vehicle to expand tourism and the gravity
base tourism demand function is approrapriate.
However, contrary to expectations, the estimated coefficient of relative prices and real
exchange rate do not have the correct signs in any of the models, even though it is significant. Positive
relationship between relative prices and tourism demand could be explained in two ways. First, there
can be a positive relationship between relative prices and tourism demand for “high-budget” tourists.
Second, relative prices may not be reflecting true cost of vacation (because of damping, promotion
etc.). A tourism price index would be more accurate which is not easily available (Katafono and
Gounder, 2004). Also, Patsouratis, Frangouli and Anastasopoulos (2005) argues that Turkey offers
cheap holiday packages due to its low cost of living and has under value exchange rate policy thus one
can expect positive or no relationship between relative prices (real exchange rate) and tourism demand.
The estimated coefficients of marketing expenditure and population variables are insignificant.
The actual impact of marketing expenditure can be distributed over time and depend on marketing
expenditure of alternative destinations countries (Walsh, 1996).
The sign of the coefficients of competitive prices have correct sign (negative) and statistically
significant for all three alternative destination countries (Greece, Spain and Egypt). An increase in
Turkish price levels relative to those of alternative destinations will encourage the tourist to travel the
alternative destinations. Since the substitute price elasticities (coefficients) are more than one for
Greece and Spain, it can be said that tourism in Turkey is very sensitive to the price changes in these
alternative destinations. However, the substitute price elasticities are less than one for Egypt, it can be
said that tourism in Turkey is not very sensitive to the price changes in this alternative destinations.
In the models, there are five dummies. The February 2001 economic crisis (D02) and the Gulf
war in March 2003 (D03) dummy variables are statistically significant and expected signs. At the time
of the February 2001 economic crisis, There has been huge depreciation (150%) on Turkish currency
thus Turkey become relatively cheaper and tourism demand expects to increase. At the time of the Gulf
War in March 2003, there was war between coalition power and south neighbor (Iraq) of Turkey.
Therefore, tourists hesitated to travel Turkey. The EU membership of sending countries (D04) and the
September 11 2001 (D05) dummy variables are statistically insignificant.
Finally, the dummy variable was introduced in the model to capture the effect of official visits
of Turkish President and Prime Minister to sending countries on tourism demand. Visit (D01) dummy
variable is statistically insignificant even if it has right sign (positive).

6. Conclusion and Policy Implication

Because of the increasing importance of international tourism sector in terms of its contribution to the
national economy, during the last several years there have been made vast theoretical and empirical
studies about the determinant of tourism demand, especially for countries, which has great dependence
on tourism sector. Therefore, the main objective of this paper was to identify the determinants of
Turkish tourism demand and estimate it utilizing the best econometric techniques. After the several
diagnostic tests, the two-way random effect models were used to measure the effect of selected socio-
economic variables on tourist arrivals from 32 sending countries to Turkey between years 2000 and
2006, and it was estimated by using the EGLS estimation method for panel data set.
The results from the two-ways random effect models suggest that the most important factors in
determining the level of tourist arrivals into Turkey are real GDP of sending countries, trade value,
distance, accommodation capacity and competitive prices level, which is followed by the February
98 International Research Journal of Finance and Economics - Issue 55 (2010)

2001 economic crisis (D02) and the Gulf war in March 2003 (D03) dummy variables. Contrary to
demand theory and expectations, relative prices and real exchange rate are significant and positively
related to the demand for tourism in Turkey. Even if it has right sign, official visits of Turkish
President and Prime Minister were not significant in explaining tourism demand.
Based on estimation results, we can make following suggestions to policy makers and tourism
supplier. First, when tourists decide to travel they are more care about competitive price instead of
relative prices or real exchange rates. Therefore, policy makers and suppliers must closely monitor the
price changes in alternative destination (Greece, Spain, and Egypt). Second, government should
provide investment in tourism infrastructure and private tourism sector increases accommodation
capacity. Finally, the coefficients of trade value and distance variables indicate that trade are an
important vehicle to expand tourism. Hence, government’s trade increase policies may increase
tourism demand too. In addition, comfortable and cheaper transportation have to be provided to
tourists.

References
[1] Akal, M., 2004. “Forecasting Turkey’s tourism revenues by ARMAX model”, Tourism
Management, Vol. 25, No. 5, pp. 565–580.
[2] Akis, S., 1998. “A Compact econometric model of tourism demand for Turkey”, Tourism
Management, Vol 19, No. 1, pp. 99–102.
[3] Archibald X., LaCorbiniere J. and Moore W., 2008. “Analysis of Tourism Competitiveness in
the Caribbean: A Gravity Model Approach”, 29th Annual Review Seminar, Research
Department of Central Bank of Barbados, July.
[4] Aslan, A., Kula, F. and Kaplan, M., 2009. “International Tourism Demand for Turkey: A
Dynamic Panel Data approach”, Research Journal of International Studies, Vol. 29, pp. 65–73.
[5] Baltagi, H. B., 2001. “Econometric Analysis of Panel Data”, John Wiley & Sons ltd. West
Sussex, England.
[6] Breusch, T. S. and Pagan, A. R., 1980. “The Lagrange Multiplier Test and Its Applications to
Model Selection in Econometric.”, Review of Economics Studies, Vol. 47, pp.239–253.
[7] Chummi, M., 2001. “Tourism Demand Model: Determinant of Thailand’s International Tourist
Receipts”, Unpublished Master Thesis, Chulalongkorn University.
[8] Crouch, G., 1994. “The Study of International Tourism Demand: a Survey of Practice” Journal
of Travel Research, Vol. 32, No. 4, pp. 41–55.
[9] Crouch, G. and Ritchie, J.R.B., 1999. “Tourism, Competitiveness, and Societal Prosperity”,
Journal of Business Research, Vol. 44, pp.137–152.
[10] Erlat, H., 2006. “Panel Data: A Selective Survey”, First Revision, Discussion Paper Series No:
97–04, August 1997, Middle East Technical University, Ankara.
[11] Greene, W.H., 2003. “Econometric Analysis”, 5th Ed., Upper Saddle River, N.J.: Prentice –
Hall.
[12] Habibi, F., and Rahim, K. A., 2009. “A Bound Test Approach to Cointegration of Tourism
Demand”, American Journal of Applied Sciences, Vol. 6(11), pp. 1925–1932.
[13] Halicioglu, F., 2004. “An ARDL Model of Aggregate Tourism Demand for Turkey”, Global
Business and Economics Review, pp. 614–624.
[14] Hsiao, C., 2003. “Analysis of Panel Data”, 2nd Edition, Cambridge University Press,
Cambridge, United Kingdom.
[15] Icoz, O., Var, T. and Kozak, M., 1998. “Tourism demand in Turkey”, Annals of Tourism
Research, Vol. 25, No. 1, pp. 236–240.
[16] Katafono, R. and Gounder, A., 2004. “Modelling Tourism Demand in Fiji”, Working Paper,
Reserve Bank of Fiji, December.
[17] Kim, S. and Song, H., 1998. “Analysis of Tourism Demand in South Korea: A Cointegration
and Error Correction Approach”, Tourism Economics, Vol. 3, No. 1, pp. 25–41.
 International Research Journal of Finance and Economics - Issue 55 (2010) 99

[18] Kónya, L., 2006. “Exports and Growth: Granger Causality Analysis on OECD Countries with a
Panel Data Approach”, Economic Modeling, Vol. 23, pp. 978–992.
[19] Kulendran, N., 1996. “Modeling quarterly tourist flows to Australia using cointegration
analysis”, Tourism Economics, Vol. 2, No. 2, pp. 203–222.
[20] Kulendran, N. and Wilson, K., 2000. “Modeling Business Travel”, Tourism Economics, Vol. 6,
pp. 47–59.
[21] Kulendran, N. and Witt, S. F., 2001. “Cointegration Versus Least Squares Regression”, Annals
of Tourism Research, Vol. 28, No. 2, pp. 291–311.
[22] Kumar, N. P., 2004. “Fiji’s Tourism Demand: the ARDL Approach to Cointegration”, Tourism
Economics, Vol. 10, No. 2, pp. 193–210.
[23] Lee, C., Var, T. and Blaine, T., 1996. “Determinant of Inbound Tourism Expenditures”, Anal of
Tourism Research, Vol. 23, No. 3, pp. 527–542.
[24] Leitao N. C., 2010. “Does Trade Help to Explain Tourism Demand? The Case of Portugal”,
Theoretical and Applied Economics, Vol. XVII, No. 3, pp. 63–74.
[25] Li, G., Song, H. and Witt, S. F., 2005. “Recent Developments in Econometric Modeling and
Forecasting”, Journal of Travel Research, Vol. 44, No. 1, pp. 82–89.
[26] Lim C., 1997. “Review of International Tourism Demand Model”, Annals of Tourism
Research, Vol. 24, pp.835–849.
[27] Ouerfelli, C., 2008. “Co-integration Analysis of Quarterly European Tourism Demand in
Tunisia”, Tourism Management, Vol. 29, pp.127–137.
[28] Patsouratis, V., Frangouli, Z. and Anastasopoulos, G., 2005. “Competition in Tourism among
the Mediterranean Countries”, Applied Economics, 37, pp. 1865-1870.
[29] Phakdisoth, L. and Kim, D., 2007. “The Determinants of Inbound Tourism in Laos”, ASEAN
Economic Bulletin, Vol. 24 (2), pp. 225- 237.
[30] Proenca, S. A. and Soukiazis, E., 2005. “Demand for Tourism in Portugal: A Panel Data
Approach”, Document to de Trabalho Rabalho / Discussion Paper (February), No: 29.
[31] Song, H., Witt, S.F. and Li, G., 2003. “Modeling and Forecasting the Demand for Thai
Tourism”, Tourism Economics, Vol. 9, pp. 363–387.
[32] Song, H. and Li, G., 2008. “Tourism Demand Modeling and Forecasting: A Review of Recent
Research”, Tourism Management, Vol. 29, No. 2, pp. 203–220.
[33] Song, H., Witt, S. F. and Jensen, T. C., 2003. “Tourism Forecasting: Accuracy of Alternative
Econometric Models”, International Journal of Forecasting, Vol. 19, No. 1, pp. 123–141.
[34] Syriopoulos, T. and Sinclair, M.T., 1993. “An Econometric Study of Tourism Demand: The
AIDS Model of US and European Tourism in Mediterranean Countries”, Applied Economics,
Vol. 25, pp. 1541–1552.
[35] Uysal, M. and Crompton, J.L., 1985. “Deriving a Relative Price Index for Inclusion in
International Tourism Demand Estimation Models”, Journal of Travel Research, Vol. 24, No. 1,
pp.32–34.
[36] Walsh M., 1996. “Demand Analysis in Irish Tourism”, Journal of the Statistical and Social
Inquiry Society of Ireland, Vol. XXVII, Part IV.
[37] Var, T., Mohammad, G. and Icoz, O., 1990. “Factors Affecting International Tourism Demand
for Turkey”, Annals of Tourism Research, Vol. 17, No. 3, pp. 606–610.
[38] Witt, S. and Witt, C., 1995. “Forecasting Tourism Demand: A Review of Empirical Research”,
International Journal of Forecasting, Vol. 11, No. 3, pp. 447–475.
[39] Zhang J. and Jensen C., 2007. “Comparative Advantage Explaining Tourism Flows”, Annals of
Tourism Research, Vol. 34, No.1, pp. 223–243.

View publication stats