Impacts of Exported Turkish Soap Operas and Visa-free ...haticeozerballi.weebly.com/uploads/1/4/2/7/... · Impacts of Exported Turkish Soap Operas and Visa-free Entry on Inbound Tourism

Impacts of Exported Turkish Soap Operas and

Visa-free Entry on Inbound Tourism to Turkey

January 24, 2013

Abstract

We examine the main determinants of the recent boost in the number of tourist inflows

to Turkey, focusing on the indirect marketing effect of the Turkish soap operas exported

abroad and recent changes in the Turkish government’s foreign policies. Applying a tradi-

tional tourist demand gravity model, we explore that the recent increase in the popularity

of the Turkish soap operas in the Middle East and Eastern Europe has boosted the number

of inbound tourists to Turkey from those countries, and as the number of hours of Turkish

soap operas aired in a particular country increases, the tourist flows from that country to

Turkey increase as well. We also consider the Turkish government’s recent bilateral agree-

ments with other countries to waive the visa requirements for ordinary foreign visitors, and

indicate that the termination of visa requirements has increased the tourist flows from those

countries to Turkey.

JEL Codes: F41, L83.

Keywords: Dynamic Panel, Film Tourism, Gravity Model, Turkish Soap Operas,

Tourist Inflows.

1

1 Introduction

The visual media, i.e. TV shows and films, are the most important vehicles for attracting

people’s attention. Most of the world’s population have used visual media to entertain them-

selves and spend their leisure time. A wide number of studies focus on the importance of the

visual media on marketing purposes. Some of those studies focused on the role of the greater

extent of visual media for influencing international tourism demand (Butler, 1990; Urry, 1994;

Edensor, 2001; Kim and Richardson, 2003; Laing and Crouch, 2009; Croy, 2010; Kim, 2012;

Portegies, 2010). Among the visual media, TV shows and films have undeniable power for

influencing audiences. Riley et al., 1998; Beeton, 2005); Macionis and Sparks, 2009; and Mor-

due, 2009 have extensively studied, the media –TV shows and movies– on tourists’ motivation

for selecting destinations; the literature has called this“film/movie induced tourism”.

On August 30, 2008, eighty-five million Arab viewers were stuck to their TV sets for the

finale of the Arabian-dubbed Turkish soap opera named “Noor.”1 Later, in 2009, following

Noor ’s unbeatable success, another Turkish soap opera succeeded in attracting TV watchers

across the Middle East. Sixty-eight million Arabian TV viewers were sitting on their couch to

watch the finale of Sanawat al Dayaa (Buccanti, 2010)2. The Arabian watchers’ loyalty to the

Turkish soap operas is not confined to just sitting down and watching the soap operas: the

sales of t-shirts and posters of Noor even surpassed those featuring Arab leaders like Saddam

Hussein or Yasser Arafat (Khaleej Times, 2008). Giving another example of Turkish soap

operas’ successes abroad, in 2010, Slovakia’s leading private broadcaster operated by Central

European Media Enterprises (CEME) started airing, A Thousand and One Nights, which has

been breaking viewer rating records. The Slovakians are so obsessed with the characters of A

Thousand and One Nights that a Slovakian rock band named Kissuck featured a pair of models

of the main characters of this series on their music clips to attract people’s attention. These

examples truly show the success of the Turkish soap operas in foreign markets, and this success

is observed in record-breaking TV ratings as well as in how they influence foreign audiences’

preferences.

The Turkish soap opera industry, due to the competition among domestic TV channels in

1Noor is a Turkish melodrama originally broadcasted in Turkey as Gumus. The series became a pop-culturephenomenon when it was aired across the Arab world as Noor by the Middle Eastern Broadcast Company(MBC).

2Sanawat al Dayaa (The Lost Years) is another Turkish soap opera originally broadcasted in Turkey as Ihla-murlar Altinda. It was aired to Arab TV watchers by MBC.

2

Turkey, has developed substantially in recent years. Turkish entrepreneurs have looked for the

possibility of exporting Turkish soap operas since 2001. A big jump in soap opera exports has

occurred since 2005. Now, Turkish soap operas are considered a growing export item under

Turkish export industry. According to the Turkish Statistical Institute, the value of Turkish

soap operas exported has reached to 100 million USD (United States dollars) in 2011, even

though this is a small amount relative to the total value of Turkey’s export, which is around 135

billion USD in 2011. This paper, however, does not study the impact of Turkish soap operas on

Turkish exports; rather, it emphasizes the role of the exported soap operas in affecting Turkey’s

international tourism demand. According to the latest World Tourism Organization (WTO)

statistics, Turkish inbound tourism growth has outstripped the European Union and global

averages, and the inflows have more than doubled since 2002. In this paper, we aim firstly to

relate the boost in the number of tourist inflows to the Turkish soap operas exported abroad.

A number of studies have investigated how the visual media influences audiences in terms of

tourism marketing purposes, both theoretically and empirically (Beeton, 2005; Kim et al., 2007;

Kim et al., 2009; Beeton, 2010; Kim, 2012). However, all these studies have investigated the

effect of certain TV shows on single market destinations. For instance, Kim et al. (2007) ex-

plored the effects of the Korean TV drama series named Winter Sonata on potential and actual

Japanese tourist flows to South Korea. In this study, they conducted a survey on Japanese

tourists and explored the effect of the Korean drama series broadcasted via Japanese TV chan-

nels on their choice of holiday destinations. In a different study, Kim et al. (2009) related tourist

inflows into Korea to the increase in Korean TV shows and soap operas, focusing on multiple

markets where Korean soap operas were exported. This study is limited in that it provides only

comparative statistics between the exported TV shows and tourist inflows, rather than building

a model to measure the impact of soap operas on inbound tourists.

Our paper has also been motivated by the recent changes in the Turkish government’s foreign

policies. Since 2006, the Turkish government has eliminated visa requirements for ordinary

foreign visitors from many countries from Central and Northern Africa, Central and East Asia,

the Middle East and Latin America. Even though visa-free entry is an important tool to increase

international tourism demand, there has been limited academic research into the effects of these

arrangements. Choong-Ki et al. (2010) has studied the effect of visa elimination in a paper

investigating the number of Korean tourists visiting Japan. They showed that visa elimination

favors the number of international tourist flows. Considering that the recent changes in Turkey’s

3

visa policies have been effective for multiple countries, in this paper, we also investigate the

effect of visa-free entry on the tourist inflows to Turkey. We employ a gravity model to explore

the determinants of the tourist inflows from 81 different countries between 1995 and 2010.

International tourism is considered a form of international trade. Thus modelling bilateral

tourist flows would be similar to creating a gravity model, as in bilateral trade models. Apart

from conventional determinants of the gravity model, we employ the bilateral trade volumes

between the source countries and Turkey to capture the comparative advantage of transportation

costs between countries. Bilateral trade volume is widely used to measure the “extent” of

economic integration between countries; this variable also captures the ease of transportation

costs between two centers, thereby measuring if there is a comparative advantage between these

two destinations. More importantly, we test for the effect of the Turkish soap operas exported

to multiple markets on the tourist inflows to Turkey. Using the number of hours of soap operas

exported to different countries taken from Calinos Incorporation and the web–sites of national

broadcasting companies in the Middle East, Slovakia, Hungary and Russia for the years between

2001 and 2010, we are able to discuss the isolated effect of soap operas on inbound tourists.

Again, new to the literature, we control for the recent government policy changes in international

relations, i.e. the Turkish government’s bilateral agreements, mostly with the countries in the

Middle East, Central and Eastern Europe, East Asia, Central and North Africa, and Latin

America to waive the visa requirements for ordinary foreign citizens.

The remainder of this paper is organized as follows: Section 2 discusses the data and descrip-

tive statistics for the variables used in the paper. Section 3 presents the conventional gravity

model and its empirical results. Section 4 outlines the results for the dynamic panel model.

Section 5 concludes the paper.

2 Data and Descriptive Statistics

The data have been collected from different number of sources. In obtaining our dataset, we

employ 81 source countries for a period between 1995 and 2010. The tourist volumes between

Turkey and the source countries have been gathered from the Ministry of Culture and Tourism of

the Republic of Turkey (MCTRT). The bilateral trade volume data have been obtained from the

International Monetary Fund’s (IMF) Direction of Trade Statistics Database (DOTS). Popula-

tion and Gross Domestic Product (GDP) per capita (in USD) have been obtained from IMF’s

4

International Financial Statistics (IFS) database. The relative Consumer Price Index (CPI)

variable, i.e. the CPI of Turkey adjusted to USD has also been collected from the IFS database.

The accommodation variable created by collecting the number of hotel rooms in Turkey has

been obtained from MCTRT. The binary variables –having a colonial relationship (COLONY),

sharing the same border (BORDER), sharing the same language (COMMON LANGUAGE), practis-

ing the same religion (RELIGION) and physical distance between the source country and Turkey

(DISTANCE)– have been obtained from Centre d’Etudes Prospectives et d’Informations Inter-

nationales (CEPII), an independent French institute for research into international economics.

The Turkish government has initiated bilateral visa agreements to waive the visa requirements

for ordinary visitors. To quantify this policy change, we create a binary time series variable,

which takes 1 if the ordinary citizens of the source country (i) has no visa requirements for

entering Turkey starting at time t and 0 otherwise. We obtain this information from the Resmi

Gazete, the official gazette of the Republic of Turkey.

Turkish soap operas have been exported by Turkish entrepreneurs since 2001. First, soap

operas were sold to Central Asian countries, including Kazakhstan and Uzbekistan, where coun-

tries share a common ethnic background with Turkey. Later, export of the Turkish soap operas

has been expanded to a wide range of countries including all Middle Eastern and North African

countries, Central Asian and a number of Eastern European countries: Bulgaria, Greece, Roma-

nia, Slovenia, Slovakia, Hungary, Ukraine, Moldova, Serbia, Macedonia, and Russia. To capture

the effect of the Turkish soap operas on inbound tourists, we created a variable, SOAP OPERA,

assigning the number of hours of Turkish soap operas exported to country (i). These data have

been gathered from many sources. We obtained most of these data from the Calinos Group of

companies, which exports nearly 80% of the Turkish soap operas on its own. We obtained the

rest of the data from different TV channels, including from MBC, a Saudi Arabia-based TV

broadcasting company, and other TV channels in Slovakia, Russia, Lebanon and Hungary.

Examining the dataset, one might raise the issue of the overlap of the countries that have

eliminated visa requirements with Turkey and import Turkish soap operas. However, the issue

is not clear–cut. For example, for some countries, for one or two years of observations (say 2007

and 2008), there was both a visa–waiving policy and soap operas were exported to that country,

but for the other years, overlapping does not exist. We are able to say that there are 22 countries

with a visa termination policy with Turkey and that imported soap operas (occurring at least

for one year of observation). The number of the countries with a visa termination policy but

5

that did not import Turkish soap operas is 34 (at least one year’s observation), the number of

the countries without visa termination but which imported soap operas is 14 (at least one year’s

observation) and countries without visa termination and which did not import soap operas is

33 (at least one observation).3

Table 1 contains descriptive statistics for the variables used in this paper. We provide the

mean, the standard deviation (SD), the number of observations, minima and maxima for each

variables to check whether the variables used in the tables have been collected appropriately.

We obtain the pair–wise correlations of the variables between a range of –0.45 and 0.52 in our

sample. For the sake of brevity, we did not provide the pair–wise correlation table. Additionally,

we show the annual percentage change in tourist inflows to Turkey in Figure 1. It contains two

series: one is the annual percentage change in tourist inflows from the world; the other indicates

the annual percentage change in tourist flows from the countries Turkish soap operas have been

exported to. We intend to compare the growth of tourist inflows from the countries that

imported Turkish soap operas with the overall growth of tourist flows in Turkey. Particularly

after 2005, the white bars are higher than the black bars, indicating that the annual percentage

change of tourist inflows from the countries that imported Turkish soap operas are higher than

the total annual percentage change of tourist inflows to Turkey. In the paper, we claim that

exported soap operas might be one of the reasons for this difference.

3 The Static Panel Model

Tourism is widely considered to be a form of international trade. Thus, modelling bilateral

tourist flows would be similar to creating a gravity model for bilateral trade (Tinbergen, 1962

and Poyhonen, 1963) or bilateral financial asset flows (Lane and Milesi-Ferretti, 2008). In the

basic forms of the gravity model, the amount of flows (trade) between the source and destina-

tion countries is assumed to increase with their size (population, GDP per capita and market

capitalization are the variables used in the economic sense) and to decrease with the distance be-

tween the economic centers (distance and cost of transportation). Adding to the gravity model,

some studies (Witt and Witt, 1995; Eliat and Einav, 2004; Lane and Milesi-Ferretti, 2008; Balli

et al., 2011; Okowa and van Wincoop, 2012) have been successful in including some similarity

3Adding up all subsets gives a result higher than 81 (total number of the countries) since the export dataand visa termination for some countries takes both 0 and 1, and counted in different groups. Say, for the stateof Qatar, the bilateral visa termination was signed in 2010, the binary variable takes 0 before 2010 and takes 1after 2010, and thus Qatar is double counted.

6

variables, such as a sharing the same border dummy, a having a colonial relationship dummy, a

common language dummy and common currency or common religion dummies. Following the

earlier literature, we use the following equation to model inbound tourist numbers:

TOURISTij,t = β0 + β1GDPi,t + β2TRADEij,t + β3POPULATIONi,t + β4ACCOMMODATIONj,t

+β5CPIj,t + β6VISAij,t + β7SOAP OPERAij,t−1 +Xij,t ∗ β + εij,t. (1)

The dependent variable, TOURISTij,t, is the natural logarithm of the tourist inflows from

country (i) to Turkey (j ) at time t . TRADEij,t is the trade value (exports + imports) between

Turkey (j ) and country (i) in USD. GDPi,t and POPULATIONi,t are the real GDP per capita and

population levels of source country (i) in logarithmic terms, respectively. ACCOMMODATIONj,t is

the accommodation (total number of the rooms in hotels in logarithms) available in Turkey at

time t, CPIj,t is the relative CPI of Turkey (j ) adjusted to USD. VISAij,t is a binary variable that

takes 1 if ordinary passport holders from country (i) are able to enter to Turkey without any

visa requirements and takes 0 otherwise. SOAP OPERAij,t−1 is simply the total number of the

hours of Turkish soap operas exported to country (i), at time t−1. We consider that Turkish

soap operas would not influence foreign audiences’ preferences instantaneously but we allowed

a one–year period for foreign visitors from country (i) to adjust their preferences in deciding

on foreign destinations to visit. Xij,t contains the bilateral variables that are widely used to

model bilateral trade, investment and immigration volumes such as the sharing the same border

dummy, the sharing the same language dummy and the practicing the same religion dummy.

We present the variables that are statistically significant in models. Apart from the variables

described above, we also control for different events including the so–called “one–minute” event

with a binary variable created for 2009 and afterwards. When the Turkish prime minister had

an argument with the President of Israel in Davos, Switzerland, this increased the popularity

of Turkey in the Middle East after 2009. We also employed another dummy for year 2007 and

afterwards: a dummy for the year of second selection of the governing party in Turkey, mostly

considered the beginning of the Turkish government facing towards the East instead of the

West.

After finding that all variables are stationary via the panel unit root test of Im et al. (2003),

7

we record the regression results for Equation (1) in Table 2. We performed four different re-

gressions to test the baseline variables and to capture the effects of the visa-waiving agreements

and exported Turkish soap operas, separately and jointly. We used Jarque–Bera statistics for

testing the normality of the residuals and we fail to reject the normality of residuals, which is re-

ported in Table 2.4 In addition, we test for autocorrelation and heteroskedasticity in the models

and find evidence for both of them in our models; therefore, we present the heteroskedasticity

and autocorrelation corrected standard errors in the table. For all regressions, to overcome the

possible endogenity problem of a bilateral trade variable (consistent with previous studies), we

use the colonial relationship dummy as an instrumental variable.

The first column contains the regression analysis for the baseline variables. Consistent with

Witt and Witt (1995) and Lim (1997), sharing the same border and the same language have

highly significant and positive coefficients, although our estimate is on the lower side compared

to that of Witt and Witt (1995) and Lim (1997) for explaining the number of inbound tourists

to Turkey. The results suggest that a common language or geographical proximity may facilitate

an easier trip to Turkey. Apart from those variables, following Balli et al. (2011), the practising

the same religion variable also has a positive and significant effect (0.94 with a SD of 0.19),

confirming that cultural similarities (based on religion) play a significant role in attracting

more tourists to Turkey, particularly from Islamic countries. Istanbul, formerly the capital of the

Islam-dominated Ottoman Turks, contains the oldest and some of the most fascinating mosques,

and other respectable Islamic symbols and places. This has attracted many tourists from the

Islamic countries to visit those places, which would be the reason why the religion dummy is

highly significant. GDP and POPULATION variables also have solid and intuitive coefficients (0.31

and 0.21, SDs of 0.07 and 0.09, respectively) which suggest that the richer the source country

is, more tourists you expect to have from that country; the same applies for the population

variable (Chadee and Mieczkowski, 1987; Witt and Witt, 1995; Naudee and Saayman, 2005).

As far as relative prices in the destination are concerned, it is common in previous studies (Eilat

and Einav, 2004; Naudee and Saayman, 2005) to use the CPI of a destination country adjusted

by the USD as a proxy for relative tourism prices. This measure of relative prices captures

changes in the real exchange rate over time as well as cross–sectional variations in the cost of

travel. The coefficient of relative CPI (–0.31 with a SD of 0.06) shows that differences in the cost

of living matter to an extent and that tourists are sensitive to the price level in Turkey. Following

4The null hypothesis for the Jarque–Bera test is that the residuals are normally distributed.

8

Lane and Milesi–Ferretti (2008), who explored the relationships between bilateral investment

flows and trade volumes between source and destination countries, we employ bilateral trade

volume as a solid determinant of the inbound tourist numbers, given that higher trade volumes

between two countries would create a comparative advantage in transaction costs. Apart from

Smith and Toms (1967) and Eilat and Einav (2004), the tourism literature has not explored

the direct relationship between trade volume and tourist inflows extensively. We show that the

coefficient of bilateral trade volume (1.20 with a SD of 0.26) is highly significant and positive

for all regressions, meaning that international tourists are more inclined to visit Turkey, as their

nation has more economic ties with Turkey. Our estimate for the bilateral trade variable is

more powerful compared to that of Eilat and Einav (2004), where they find mixed results on

the effect of bilateral trade on tourist flows.

Given that all baseline variables of column (1) have meaningful and significant coefficients

for explaining the determinants of tourist flows to Turkey, in the second column, we test if the

Turkish government’s recent visa-waiving agreements have boosted tourist inflows to Turkey.

The coefficient of the visa dummy is significant and positive (0.64 with a SD of 0.17), suggesting

an increase in the tourist flows from the countries where visa requirements have recently been

waived for regular citizens. The third column tests for the possible effect of exported Turkish

soap operas. We estimated a significant and positive coefficient (0.06 with a SD of 0.02),

meaning that exports of Turkish soap operas (in number of hours) have a positive marketing

effect for Turkish tourism.5 The adjusted R2 also increases from 0.69 to 0.75 when we add

the soap opera variable to the list of explanatory variables for tourists inflows to Turkey. The

last column shows the results for testing for the joint effect of the visa-waving policy and the

exported soap operas. It shows that both the coefficients of exported Turkish soap operas and

visa-waving policy dummy are jointly significant and effective in explaining the inbound tourist

numbers to Turkey. The result of the F -test for soap opera and visa-waving policy variables

strongly rejects the null hypothesis of these variables jointly being insignificant.

5Thanks to the anonymous referee, we control for the ethnic relationship of Turkey with its neighbouringcountries. For example, some Turkish citizens have relatives living on the neighbour countries. This mightaffect the estimations. Accordingly, we control for the border effect (Syria, Iraq and, to some extent, Bulgariaand Greece), by dropping these countries and rerunning the regressions. However, the results have not changedsubstantially.

9

4 Dynamic Panel Data Model

There are two important issues in the panel data model. First, time-invariant country char-

acteristics (fixed effects), such as geography and demographics, may be correlated with the

explanatory variables. The fixed effects contained in the error term of Equation (1) consist of

the unobserved country-specific effects. Second, in the previous section, we omitted persistency

issues in international tourism demand. For instance, if you are a tourist, you may consider

visiting the same country next year, since you like the place a lot and/or you do not want to run

the risk of ruining your holiday by visiting an unknown place. This is called the “persistence

and reputation effect” (Naudee and Saayman, 2005).

Incorporating these dynamics into the previous model, the new model would be as follows:

TOURISTij,t = β0 + β1TOURISTij,t−1 + β2GDPi,t + β3TRADEij,t + β4POPULATIONi,t

+β5ACCOMMODATIONj,t + β6CPIj,t + β7VISAij,t + β8SOAP OPERAij,t−1 +Xij,t ∗ β + εij,t. (2)

Due to the inclusion of the TOURISTij,t−1 variable, a problem of endogeneity arises and

regular Ordinary Least Square (OLS) estimations provide biased results (Naudee and Saay-

man, 2005; Khadarooa and Seetanah, 2008). To overcome this problem, instrumental variables

(IVs) need to be used to overcome the endogeneity problem in dynamic models. The possi-

bility of having strong instruments for the TOURISTij,t−1 variable is low. In the case of having

weak instruments for the TOURISTij,t−1 variable, IV estimators are likely to be biased towards

the OLS. Accordingly, to cope with this problem, the difference Generalized Method of Mo-

ments (GMM) of Arellano-Bond (1991) estimator is widely used in the literature. The lagged

levels of the endogenous variable are included as instruments. This makes the endogenous

variables pre-determined and, therefore, they are not correlated with the error term.

In order to remedy the problem mentioned above, we use first–differences (first step) to

transform equation (2) into:

∆TOURISTij,t = β0 + β1∆TOURISTij,t−1 + β2∆GDPi,t + β3∆TRADEij,t + β4∆POPULATIONi,t

10

+β5∆ACCOMMODATIONj,t + β6∆CPIj,t + β7∆VISAij,t + β8∆SOAP OPERAij,t−1 + ∆εij,t. (3)

The left-hand side is the log difference in tourism flows from the source country (i) to

Turkey (j ) at period t.

Table 3 reports the first-step GMM estimator of Equation (3). We used Jarque–Bera statis-

tics for testing for the normality of the residuals and we fail to reject the normality of the resid-

uals. In addition, we test for the autocorrelation and heteroskedasticity of the models and find

evidence for heteroskedasticity. We have presented the heteroskedasticity–corrected standard

errors. In all three regressions, the first– and second–order correlation Arellano-Bond (AB) tests

have p–values greater than 10%, which means that there is not enough evidence to support that

there is autocorrelation. This finding validates the use of suitably lagged endogenous variables

as instruments. Additionally, the p–values of the Sargan test of over–identifying restrictions

fails to reject the null hypothesis that the instruments are exogenous in any specification.

The main findings are not different from the first model in general. On the top of the previous

results, the lagged coefficient of TOURIST is highly significant and around 47% for all models,

which validates the persistence and reputation effects. The coefficients of POPULATION and GDP

are again significant (0.99 and 0.62, with SDs of 0.08 and 0.04, respectively), as expected: the

greater (higher) the population (GDP per capita) of the origin country, the greater the number

of tourists flows to Turkey. CPI is negative and significant (–0.46 with a SD of 0.04), confirming

Table 2 about the cost motivation of the tourists. TRADE is highly significant as well, confirming

the importance of the economic ties between the origin country and Turkey on inbound tourists

to Turkey. In the dynamic model, time–invariant variables, i.e. practising the same religion,

sharing the same border, distance, sharing the same language have been deleted automatically.

Focusing on the control variables, SOAP OPERA is highly significant (0.12 with a SD of 0.02

and 0.14 with a SD of 0.04 in the last column), confirming once again–more strongly–that the

Turkish soap operas indeed influence foreign audiences of these soap operas and attract them to

Turkey. The Turkish government’s visa-waiving policy variable is also highly significant (0.31

with a SD of 0.06) in modelling the tourist inflows to Turkey. Once again, this finding supports

the view that the recent visa-waiving agreements of the Turkish government have boosted the

tourist flows from those countries.

11

5 Concluding Remarks

In this paper, we examine the determinants of the recent increase in inbound tourist numbers to

Turkey, focusing on both the impact of exported Turkish soap operas and visa-free entry polices.

We show that previously determined factors, i.e. relative CPI, bilateral trade, population,

geographic and cultural factors, are effective in explaining the international tourism demand

of Turkey. Since 2001, there has been a boost in the volume of Turkish soap operas exported

abroad, particularly to Middle Eastern, Eastern European and North African countries. We

show that soap operas exported to these countries have influenced foreign audiences’ preferences,

leading to a sharp increase in the number of tourists inbound from those countries. Apart from

the soap operas, the government’s visa–waiving polices with other nations have also helped

ordinary foreign visitors to visit Turkey more easily, contributing to the recent increase in

number of visitors from these countries.

12

References

Arellano, M., and Bond, S. (1991). Some tests of specification for panel data: Monte Carloevidence and an application to employment equations. Review of Economic Studies, 58,277–297.

Balli, F., Jean Louis, R., and Osman, M. (2011). The patterns of cross-border portfolio invest-ments in the GCC Region: Do institutional quality and the number of expatriates play arole? Journal of Economics and Finance, 35(4), 434–455.

Beeton, S. (2005). Film-induced tourism. Clevedon: Channel View Publications.

Beeton, S. (2010). The advance of film tourism. Tourism and Hospitality Planning & Develop-ment, 7(1), 1–6.

Buccianti A. (2010) Dubbed Turkish soap operas conquering the Arab world: social liberationor cultural alienation? Retrieved from Arab Media and Society on 05 May 2012.(http://www.arabmediasociety.com/articles/downloads/20100330130359)

Butler, R. (1990). The influence of the media in shaping international tourist patterns. TourismRecreation Research, 15(2), 46–53.

Chadee, D. and Mieczkowski, Z. (1987). An empirical analysis of the effects of the exchangerate on Canadian tourism, Journal of Travel Research, 26(1), 13–17.

Choong-Ki L., Hak-Jun S., and Lawrence J. B. (2010). The impact of visa-free entry on out-bound tourism: A case study of South Korean travelers visiting Japan. Tourism Ge-ographies: An International Journal of Tourism Space, Place and Environment, 12(2),302–323.

Croy, G. (2010). Planning for film tourism: active destination image management. Tourismand Hospitality Planning & Development, 7(1), 21–30.

Edensor, T. (2001). Performing tourism, staging tourism: (re)producing tourist space and prac-tice. Tourist Studies, 1(1), 59–81.

Eilat, Y., and Einav, L. (2004). Determinants of international tourism: a three-dimensionalpanel data analysis. Applied Economics, 36, 1315–1327.

Im, K. S., Pesaran M., and Yongcheol S. (2003). Testing for unit roots in heterogeneous panels,Journal of Econometrics, 115, 53–74.

Khaleej Times (2008). Noor ’s lights up beacons of change. Retrieved on 28 July 2008 fromhttp://www.arabwashingtonian.org/english/article.php?articleID=853.

Khadarooa, J. and Seetanah B. (2008). The role of transport infrastructure in internationaltourism development: a gravity model approach. Tourism Management, 29, 831–840.

Kim, H. and Richardson, S. L. (2003). Motion picture impacts on destination images. Annalsof Tourism Research, 30(1), 216–237.

Kim, S. S., Argusa, J., Lee, H., and Chon, K. (2007). Effects of Korean television dramas onthe flow of Japanese tourists. Tourism Management, 28(5), 1340–1353.

Kim, S., Long, P., and Robinson, M. (2009). Small screen, big tourism: the role of popular Ko-rean television dramas in South Korean tourism. Tourism Geographies: An InternationalJournal of Tourism Space, Place and Environment, 11(3), 308–333.

13

Kim, S. (2012) Audience involvement and film tourism experiences: emotional places, emo-tional experiences, Tourism Management, 33(2), 387–396.

Laing, J. H. and Crouch, G. I. (2009). Exploring the role of the media in shaping motivationsbehind frontier travel experiences. Tourism Analysis, 14(2), 187–198.

Lane, P.R. and Milesi-Ferretti, G. M. (2008). International investment patterns. The Reviewof Economics and Statistics, 90(3), 538–549.

Lim, C. (1997). Review of international tourism demand models. Annals of Tourism Research,24(4), 835–849.

Macionis, N. and Sparks, B. (2009). Film-induced tourism: an incidental experience. TourismReview International, 13(2), 93–101.

Mordue, T. (2009). Television, tourism, and rural life. Journal of Travel Research, 47(3), 332–345.

Naudee, W. A. and Saayman, A. (2005). Determinants of tourist arrivals in Africa: a paneldata regression analysis. Tourism Economics, 11(3), 365–391.

Okowa, Y. and van Wincoop, E. (2012) Gravity in international finance. Working paper 072010.Hong Kong: Institute for Monetary Research.

Portegies, A. (2010). Places on my mind: exploring contextuality in film in between the globaland the local. Tourism and Hospitality Planning & Development, 7(1), 47–58.

Poyhonen, P. (1963). A tentative model for the volume of trade between countries. WeltwirtschaftlichesArchiv, 90(1), 93–99.

Riley, R., Baker, D., and Van Doren, C. (1998). Movie-induced tourism. Annals of TourismResearch, 25(4), 919–935.

Smith, A.B. and Toms, J.N. (1967). Factors affecting demand for international travel to andfrom Australia, Occasional paper no.11. Canberra: (Bureau of Transport Economics).

Tinbergen, J. (1962). Shaping the world economy: Suggestions for an international economicpolicy. New York: The Twentieth Century Fund.

Urry, J. (1994). Cultural change and contemporary tourism. Leisure Studies, 13, 233–238.

Witt, S. and Witt, C. (1995). Forecasting tourism demand: a review of empirical research. In-ternational Journal of Forecasting, 11, 447–475.

14

Figure 1: The annual percentage change in the number of tourist inflows to Turkey

-30%

-20%

-10%

0%

10%

20%

30%

40%

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

TOTAL

SOAP IMPORTINGCOUNTRIES

Source: Ministry of Culture and Tourism of Republic of Turkey and authors' own calculations. Total refers to the annual percentage change in tourist flows from all countries in our analysis, whereas Soap importing countries refers to the annual percentage change in tourist flows from countries which imported Turkish soap operas.

15

Table 1: Descriptive Statistics

Mean SD Min Max Obs

ACCOMMODATIONj 242751 21019 202483 278255 1296BORDERij 0.06 0.28 0 1 1296COLONYij 0.04 0.31 0 1 1296COMMON LANGUAGEij 0.07 0.26 0 1 1296CPIj 0.49 2.12 12.12 10.12 1264DISTANCEij 4027 3774 502 17234 1264GDPi 14494 17328 321 118672 1296POPULATIONi 63652 185804 267.47 1318194 1296RELIGIONij 0.21 0.46 0 1 1296SOAP OPERAij 39.33 6.03 0 213.11 1296TOURISTij 200189 487358 108 4488350 1220TRADEij 1615 3430 12 4488350 1257VISAij 0.51 0.48 0 1 1296

Notes: The dependent variable, TOURISTij , is the number of tourist flows from source country (i) to Turkey (j ).COMMON LANGUAGEij is a binary variable that takes 1, if more than 80% of the population of both the source anddestination countries speak same language and 0 otherwise. Similarly, RELIGIONij is another binary variable thattakes 1 if more than 80% of the population of both the source and destination countries practise same religion and0 otherwise. ACCOMMODATIONj is the accommodation (total number of the rooms in hotels) available in Turkey.BORDERij is another binary variable that takes 1, if source country (i) and Turkey (j) shares same border and 0otherwise. DISTANCEij is the physical distance between the capital cities of source country (i) and Turkey (j) (inkilometres). GDPi is the real GDP per capita in source country (i), in USD. CPIj is the CPI of the destinationcountry adjusted by the exchange rate in USD. POPULATIONi is the total population of source country (i) inthousands. TRADEij is the total trade value (exports + imports) between source country (i) and Turkey (j ) inmillion USD. VISAij is again another binary variable that takes 1 if there are no visa requirements for the ordinarycitizens of country (i), entering Turkey, and 0 otherwise. SOAP OPERAij shows the number of hours of Turkishsoap operas exported to country (i).

16

Table 2: Static Panel Data Estimation

Dependent Variable: The logarithm of the number of tourist flows from source countries to Turkey

(1) (2) (3) (4)

COMMON LANGUAGEij,t 1.82‡(0.46) 1.59‡(0.31) 2.19‡(0.21) 1.65‡(0.22)

DISTANCEij,t –0.94‡(0.19) –0.88‡(0.20) –0.87‡(0.20) 0.89‡(0.12)

BORDERij,t 1.44‡(0.36) 1.17‡(0.37) 1.43‡(0.40) 1.17‡(0.21)

RELIGIONij,t 0.94‡(0.19) 0.78‡(0.24) 0.91‡(0.19) 0.76‡(0.10)

GDPi,t 0.31‡(0.07) 0.24†(0.12) 0.32‡(0.14) 0.21‡(0.10)

TRADEij,t 1.20‡(0.26) 1.16‡(0.29) 3.92‡(0.35) 1.17‡(0.11)

POPULATIONi,t 0.21†(0.09) 0.23‡(0.06) 0.24‡(0.06) 0.22‡(0.05)

ACCOMMODATIONj,t 0.47‡(0.12) 0.46‡(0.10) 0.48‡(0.13) 0.65‡(0.11)

CPIj,t –0.31‡(0.06) –0.20†(0.08) –0.31‡(0.10) –0.22‡(0.11)

VISAij,t – 0.64†(0.17) – 0.64†(0.07)

SOAP OPERAij,t−1 – – 0.06‡(0.02) 0.09‡(0.02)

SAMPLE 1132 1132 1132 1132Jarque–Berra p–value 0.41 0.44 0.45 0.40ADJUSTED R−SQUARE 0.69 0.71 0.75 0.78

Notes: ∗, † and ‡ indicate that the relevant coefficient is significant at the 10%, 5% and 1% level, respectively.See Table 1 for the variable definitions. heteroskedasticity and autocorrelation–corrected standard errors arereported in parentheses.

17

Table 3: Dynamic Panel Data Estimation

Dependent Variable: The log difference of the number of tourist flows from source countries to Turkey

(1) (2) (3) (4)

∆TOURISTij,t−1 0.47‡(0.03) 0.46‡(0.03) 0.47‡(0.03) 0.46‡(0.04)

∆GDPi,t 0.62‡(0.04) 0.57‡(0.04) 0.60‡(0.04) 0.57‡(0.04)

∆TRADEij,t 0.11‡(0.02) 0.08‡(0.01) 0.41‡(0.05) 0.13‡(0.04)

∆POPULATIONi,t 0.99‡(0.08) 1.12‡(0.10) 0.68†(0.10) 1.07†(0.10)

∆ACCOMMODATIONj,t 0.04(0.03) 0.06(0.14) 0.12†(0.06) 0.12†(0.06)

∆CPIj,t –0.46‡(0.04) –0.31‡(0.05) –0.42†(0.14) –0.32‡(0.04)

∆VISAij,t – 0.31‡(0.04) – 0.31‡(0.06)

∆SOAP OPERAij,t−1 – – 0.12‡(0.02) 0.14‡(0.04)

SAMPLE 1010 1010 1010 1010Jarque–Berra p–value 0.34 0.38 0.44 0.36AB(1) test p–value 0.21 0.26 0.20 0.41AB(2) test p–value 0.26 0.23 0.31 0.35Sargan statistic p–value 0.81 0.85 0.89 0.88

Notes: ∗, † and ‡ indicate that the relevant coefficient is significant at the 10%, 5% and 1% levels, respectively.See Table 1 for the variable definitions. Heteroskedasticity–corrected standard errors are reported in parentheses.

18