Microsoft Word - 17.docx PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X DETERMINATION OF NUMBER OF ARRIVING TOURISTS AND NIGHT SPENT IN ACCOMMODATION RELATIONS WITH ECONOMIC GROWTH: THE CASE OF TURKEY Ress. Asst. Seda KARAGOZ ZEREN Trakya University karagozseda@hotmail.com ABSTRACT The tourism-based growth hypothesis (TLGH), which indicates that tourism is the determinant of economic growth and provides economic growth, suggests a positive relationship between tourism expenditures and economic growth. Within the context of the tourism-based growth hypothesis, it is known that several of the factors affecting tourism expenditures are the number of tourists coming to the country and the length of their stay. This study is aimed at determining the relationship with this context in 2000-2015 years with 81 provinces of data from arrivals tourist numbers and night spent in accommodation in Turkey with variables gross domestic product per capita Depending on this purpose, the horizontal and cross-sectional dependencies of the variables are first analyzed with the CD proposed by Pesaran (2004) and the BA-LM tests proposed by Pesaran, Ullah and Yamagata (2008). According to the test results, the null hypothesis, which suggests that there are no horizontal-section dependencies, has been rejected. Then, the CIPS Panel Unit Root test, which is sensitive to horizontal-section dependency, was performed and the stationarity of the variabilities was determined. In addition, the cointegration test, which is sensitive to horizontal-section dependency, was applied and a cointegration relationship was found between the number of arriving tourists and the length of stay and economic growth. The slope heterogeneity test results, which are sensitive to horizontal-slice dependence applied to variables, show that slope heterogeneity is present in the variables. Dynamic CCEMGE (Dynamic Common Correlated Effects Mean Group Estimator) model was used to test the TLGH hypothesis because our variables have horizontal-section dependencies on one side and slope heterogeneity on the other. Dynamic CCEMG model results indicate that the results of TLGH hypothesis applies to the provinces of Turkey. Moreover, the relationship between the number of tourists and the length of stay and economic growth varies according to the results of the study. PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Key Words: Number of Incoming Tourists, Accommodation Time, Economic Growth, Tourism-Led Growth Hypothesis, Dynamic CCEMGE Model 1. INTRODUCTION PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X The advanced hypothesis that international tourism constitutes the main source of economic growth is the tourism-based growth hypothesis (Brida, Sanchez Carrera, & Risso, 2008: 1). This economic link is in the ratio of Tourism Capital Import to Growth (TKIG) (Cortes- Jimenez, Pulina, Prunera, & Artis , 2009:4). The growth of imports of investment goods by increasing tourism revenues. The economic growth of countries is possible by developing international tourism as non-traditional exports (Chang, Khamkaew, & McAleer, 09.08.2018). Tourism is an invisible export pen (Bahar , 2006: 140). In the tourism-based growth hypothesis, tourism is considered to provide economic growth in the long run (Balaguer & Cantavella- Jorda, 2002: 878). In other words, tourism expenditures made by foreign tourists in another country are an export effect as they are in exports of goods, in terms of foreign exchange income provided to that country (Çağlayan, Güriş, & Öskönbayev, 2012: 107). Tourism is the whole of the events that occur as a result of the accommodation and the accommodation of the goods and services produced by the tourism enterprises, so that the individuals are permanently residing and traveling outside the places where they work and not aiming to settle in the area and gain profit (Barutçugil, 1998). According to the data of 2015, foreign tourist traffic increased by 4.4% to 1 billion 184 million. In 2016, the tourism sector is expected to grow by 4% (UNWTO (World Tourism Organization), 09.08.2018). In this context, the figures show that the tourism sector has become a constantly developing and growing sector in the world economy. In this study, tourism-led growth hypothesis, 2000-2015 year the number of tourists coming to Turkey and their stay provinces with data, economic growth has been tested using variables. Panel data studies performed in the tourism sector in Turkey is to test the existence of co- integration between provinces generally related variables. Aslan (2008) study made by the Turkey's long-term economic development in the tourism role of the 1992-2007 period of growth hypothesis based on tourism by examination to examined Johansen be confirmed by co- integration and Granger causality tests. According to the study variables of tourism in Turkey has determined that the impact on economic growth (Arslan, 2008: 1). Kaya and Canlı (2013) is the work they have done investigated the determinants of international tourism demand for Turkey. 1990-2018 / 2010 for the period of 24 selected OECD countries towards Turkey have made an analysis of international tourism demand. Within findings they obtained that determine the demand for international tourism to Turkey from OECD countries have determined that the PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X variable income. In this context, the increase in the income levels of OECD countries leads to an increase in the demand for tourism to Turkey (Kaya & Canlı, 2013: 43). 2. DATA AND METHOD The number of tourists and accommodation for the duration provinces of Turkey between the years 2000-2015 in the study was attempted to test whether there is a relationship on economic growth. Working tourism enterprises located in 81 provinces in the number of tourists coming to the certificate of accommodation facilities with municipal certificate accommodations for their stay and Turkey were examined with per capita GDP falling in each province separately. Tourism data and falling GDP per capita in the provinces Turkey Statistical Institute (TÜİK- http://www.tuik.gov.tr) was obtained from the database. Descriptive statistics of the data used in Table 1 are presented below. Table 1. Descriptive Statistics Variables Observation Mean Standard Deviation Min. Max. Lg (GSYH Per Capita) 1296 10.039,25 7.493,607 455 43.645 Ttg (Number of Tourists Coming to Accommodation Facilities with Tourism Operation Certificate) 1296 336.052,51 1.216.467,581 148 14.657.471 Btg (Number of Tourists Coming to Municipal Licensed 1296 206.679,52 435.180,035 658 3.642.438 PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Accommodation Facilities) Tg (Number of nights spent in Accommodation Facilities with Tourism Operation Certificate) 1296 1.060.511,29 5.508.824,294 366 70.527.186 Bg (Number of nights spent in Municipal Certified Accommodation Facilities) 1296 445.042,98 1.294.190,872 697 11.920.172 Source: TÜİK, https://biruni.tuik.gov.tr/, 01.06.2018. In order to determine the relationship between the number of arriving tourists and the length of stay and economic growth, the model presented in equation (1) is used: 𝑳𝒈𝒊𝒕 = 𝝀𝒊𝒅𝒕 + 𝜶𝟏𝒊𝑻𝒕𝒈𝒊𝒕 + 𝜶𝟐𝒊𝑩𝒕𝒈𝒊𝒕 + 𝜶𝟑𝒊𝑻𝒈𝒊𝒕 + 𝜶𝟒𝒊𝑩𝒈𝒊𝒕 + 𝒖𝒊𝒕 (1) 𝒖𝒊𝒕 = 𝜽𝒊𝒇𝒕 + 𝜺𝒊𝒕, i=1,2,…,N and t=1,2,…,T In the equation, Lg-per capita GDP, Ttg-Number of Tourists Coming to Accommodation Facilities with Tourism Operation Certificate, Btg-Number of Tourists Coming to Municipal Licensed Accommodation Facilities, Tg-Number of nights spent in Accommodation Facilities with Tourism Operation Certificate, Bg-Number of nights spent in Municipal Certified Accommodation Facilities, 𝑑6 and 𝑓6 are observed and unobserved joint effects, and 𝜀96 is the error term. In the applied analysis, it was investigated whether or not there is horizontal section dependency among variables. Whether horizontal section dependency is an important problem in panel data analysis. In traditional first-generation panel data models, it is assumed that there are no horizontal section dependencies between the error terms and the slopes are homogeneous. Not investigating the horizontal section dependency can lead to many problems. One of these problems is that in the case of horizontal section dependency between error terms, PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X it is shown that reaching many dimensions in the results of traditional first generation unit root tests and using fixed and changing effect models to reach inconsistent and misleading results (Atılgan, Ertuğrul, & Basar, 2017: 425). The CD test proposed by Pesaran (2004) and the LM (Bias adjusted LM test) proposed by Pesaran, Ullah and Yamagata (2008) were used to analyze whether there is horizontal section dependency. Because of the horizontal section dependency between the variables, the CIPS unit rootstocks recommended by Pesaran (2007) were used. Once the series were found to be stationary in the first order, the cointegration relation was examined using the Gauss 10 program. After determining the existence of cointegration between the variables, the slope heterogeneity test proposed by Pesaran and Yamagata (2008) was applied. The slope is the assumption of slope homogeneity in traditional first generation panel data models that are due to the application of the heterogeneity test. If there is no slope homogeneity, the estimation results in the constructed model can be misleading. The Dynamic Common Correlation Effects Mean Group Estimator-Dinamic CCEGM model developed by Chudik and Pesaran (2015), which takes into account both horizontal and vertical gradient heterogeneity, is used. The dynamic CCEGM model is set up as shown in Equation (2) below. 𝒚𝒊𝒕 = 𝜶𝟎𝒊𝒚𝒊𝒕 − 𝟏 + 𝜶𝟏𝒊 + 𝜷𝒊𝒙𝒊𝒕 + ∑ 𝜽𝒊𝒚@𝒊𝒕A𝒋𝒏𝒋D𝟏 + ∑ 𝜽𝒊𝒙@𝒊𝒕A𝒋 + 𝒏 𝒋D𝟏 𝝋𝒊𝒇𝒕 + 𝝐𝒊𝒕 (2) In the equation, the 𝑦96 dependent variable is the fixed effects of the time-invariant heterogeneity among the 𝛼I9 groups, the 𝑥96 descriptive variables vector, �̅�96AL 𝑎𝑛𝑑 𝑦P96AL delayed horizontal section averages, 𝛽9 the specific slopes of the observed variables, 𝑓6 common factors that can not be observed with the heterogeneous factor𝜑9, and the 𝜖96 error term. In the applied analysis, CCEGM and AMG (Augmented Mean Group) models were predicted in addition to the dynamic CCEGM model. 3. ESTIMATION RESULTS PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X In the study, firstly, it is examined whether there is horizontal section dependency between variables. For this purpose, the CD test proposed by Pesaran (2004) and the Bias adjusted LM test proposed by Pesaran, Ullah and Yamagata (2008) were used. The horizontal section dependency test results are presented in Table 2 and Table 3. Table 2. Horizontal Cross Section Dependency Test Results- CD Test p Value CD Test Lg 0.000 226.71 Ttg 0.000 143.88 Btg 0.000 64.96 Tg 0.000 144.29 Bg 0.000 77.06 The values show a level of significance of 5 percent. The null hypothesis is that there is no horizontal section dependency. Table 3. Horizontal Cross Section Dependency Test Results- LM / Bias Adjusted LM Test Statistic p Value LM 5318 0.000 LM adj. 32.75 0.000 The values show a level of significance of 5 percent. The null hypothesis is that there is no horizontal section dependency. When Table 2 and Table 3 were examined, the null hypothesis that there was no horizontal section dependency in the CD and LM tests was rejected. Horizontal section dependency exists between variables. The results of the CIPS tests proposed by Peseran (2007) are shown in Table 4 for the unit root examination after the horizontal cross section dependence. Table 4. CIPS Unit Root Test Results PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Level First Difference Fixed Fixed + Trend Fixed Fixed + Trend Lg -2.005 -2.022 -3.402 -3.502 Ttg -2.343 -2.685 -3.735 -3.795 Btg -2.517 -2.616 -3.864 -3.819 Tg -2.317 -2.779 -3.702 -3.739 Bg -2.647 -2.657 -3.799 -3.931 The values show a level of significance of 5 percent. The basic hypothesis for the CIPS test is that the unit is root-containing. When Table 4 is examined, the variables become stable after the first difference is taken. Analysis of the cointegration relation between the variables after the series were determined to be stationary was carried out by Westerlund Durbin-Hausman test in the Gauss 10 program and the critical value at the 5% significance level was found to be 1.645> 0.05. The results of the cointegration analysis showed that the number of arriving tourists and the duration of accommodation were cointegration with GDP. Following this analysis, the slope heterogeneity test proposed by Pesaran and Yamagata (2008) was applied. Slope heterogeneity test results are shown in Table 5. Table 5. Slope Heterogeneity Test Results Value Swamy 𝑆U 911.083 ∆W 25.323 ∆W𝒂𝒅𝒋 31.481 ∆W 23.063 ∆W𝒂𝒅𝒋 1.736 The values show a level of significance of 5 percent. The basic hypothesis is slope homogeneity. The results of the slope hetoregion tests show that the slope, which is the main hypothesis, is rejected as being homogeneous. In other words, it is revealed that the slope in the model is heterogeneous and the estimators considering the heterogeneity of slope should be used. For PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X the coefficient estimates, Dynamic CCEGME estimator which considers these problems have been used since both the horizontal section dependency and the slope heterogeneity exist in the model. The dynamic CCEGME estimator results are presented in Table 6, Table 7, Table 8 and Table 9. Table 6. Dynamic CCEMG Estimator Results- Lg dependent variable Ttg independent variable Dependent Variable (Lg) Coefficients Lg (-1) 0.397 Ttg 0.003 Lg- Horizontal section mean (-1) -0.042 Ttg- Horizontal section mean (-1) 0.000 C -124.624 The values show a level of significance of 5 percent. Table 7. Dynamic CCEMG Estimator Results- Lg dependent variable Btg independent variable Dependent Variable (Lg) Coefficients Lg (-1) 0.355 Btg 0.001 Lg- Horizontal section mean (-1) -0.388 Btg- Horizontal section mean (-1) 0.002 C -29.217 The values show a level of significance of 5 percent. Table 8. Dynamic CCEMG Estimator Results- Lg dependent variable Tg independent variable Dependent Variable (Lg) Coefficients PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Lg (-1) 0.385 Tg 0.002 Lg- Horizontal section mean (-1) -0.420 Tg- Horizontal section mean (-1) 0.000 C -96.392 The values show a level of significance of 5 percent. Table 9. Dynamic CCEMG Estimator Results- Lg dependent variable Bg independent variable Dependent Variable (Lg) Coefficients Lg (-1) 0.326 Bg 0.002 Lg- Horizontal section mean (-1) -0.369 Bg- Horizontal section mean (-1) 0.000 C -52.297 The values show a level of significance of 5 percent. The coefficients of the independent variables Ttg, Btg, Tg and Bg were found to be positive and statistically significant in the expectation direction. The fact that this coefficient is positive and meaningful indicates that there is a relationship between the number of tourists and the length of stay and GDP. The dynamic CCEGM Model gives the coefficients for each province. The coefficients obtained for the 81 provinces examined are presented in Table 10. Table 10. Provincial Coefficients Obtained from the Dynamic CCEGM Model Province Coefficients Ttg Btg Tg Bg Erzurum 0.017 -0.002 -0.003 -0.001 Erzincan -0.015 -0.017 0.018 0.018 Bayburt 0.320 0.006 -0.270 0.008 Agri -0.004 0.010 0.032 -0.009 Kars 0.011 0.011 -0.011 -0.004 PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Igdir -0.007 0.008 0.019 -0.001 Ardahan 0.073 0.117 -0.042 -0.007 Malatya 0.011 0.000 -0.004 -0.002 Elazığ 0.001 0.011 0.014 -0.006 Bingöl 0.108 0.005 -0.038 -0.016 Tunceli 0.044 0.014 -0.016 -0.017 Van -0.023 -0.003 0.001 0.001 Muş -0.002 0.123 -0.010 -0.047 Bitlis -0.086 -0.002 0.078 -0.018 Hakkari 0.001 0.014 -0.004 0.006 Gaziantep 0.035 -0.008 0.002 0.926 Adıyaman 0.002 -0.022 -0.021 0.664 Kilis 0.228 -0.025 -0.135 0.042 Şanlıurfa 0.008 -0.006 -0.007 0.004 Diyarbakir -0.016 -0.001 0.009 0.001 Mardin -0.023 0.031 0.021 -0.022 Batman 0.001 0.113 -0.008 -0.094 Şırnak 0.021 0.018 -0.007 -0.013 Siirt -0.971 -0.112 1.267 0.086 İstanbul 0.003 -0.001 -0.002 0.001 Tekirdag -0.058 -0.007 0.022 -0.001 Edirne -0.019 0.008 0.016 -0.001 Kirklareli -0.034 0.011 0.012 -0.002 Balıkesir -0.005 0.001 0.003 -0.001 Çanakkale -0.005 -0.002 0.001 0.002 İzmir 0.001 0.001 -0.001 -0.001 Aydin 0.000 -0.001 0.000 0.000 Denizli 0.004 -0.002 -0.005 0.000 Mugla 0.001 -0.001 0.000 0.000 Manisa -0.049 -0.005 0.009 0.002 Afyonkarahisar 0.002 0.003 -0.002 -0.001 Kütahya 0.000 -0.001 0.004 0.000 Usak -0.023 0.029 0.008 -0.061 PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Bursa -0.006 -0.002 0.003 -0.003 Eskisehir -0.041 0.011 0.035 -0.021 Bilecik -0.152 0.298 0.123 -0.103 Kocaeli 0.052 0.064 -0.023 -0.038 Sakarya 0.018 -0.003 -0.008 -0.061 Düzce -0.119 -0.009 0.067 0.007 Bolu 0.020 0.004 -0.011 0.003 Yalova 0.005 -0.042 0.011 0.038 Ankara 0.033 0.004 -0.017 -0.005 Konya -0.003 -0.009 -0.006 0.007 Karaman -0.001 0.207 0.032 -0.204 Antalya 0.000 0.000 0.000 0.001 Isparta 0.514 -0.028 -0.048 0.009 Burdur 0.075 0.016 -0.097 -0.017 Adana 0.001 0.000 0.002 0.001 Mersin 0.000 0.000 0.000 0.000 Hatay 0.003 -0.005 -0.001 0.005 Kahramanmaras 0.003 0.001 0.000 0.000 Osmaniye -0.063 0.081 0.072 -0.067 Kirikkale 0.111 0.023 -0.151 0.028 Aksaray 0.014 -0.018 -0.004 0.016 Nigde 0.020 -0.047 -0.008 0.048 Nevşehir -0.004 -0.001 0.002 0.002 Kirsehir -0.019 0.026 0.002 -0.026 Kayseri -0.019 -0.015 0.018 0.001 Sivas 0.009 0.001 0.000 0.001 Yozgat 0.016 0.012 -0.009 -0.013 Zonguldak 0.082 0.052 -0.034 -0.028 Karabük 0.027 -0.024 -0.018 0.024 Bartin -0.031 0.003 0.013 -0.004 Kastamonu -0.020 0.007 -0.005 -0.007 Çankırı 0.049 0.013 -0.039 -0.014 Sinop -0.079 0.011 0.054 -0.007 PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Samsun 0.007 0.001 -0.004 -0.001 Tokat 0.029 -0.011 -0.021 0.016 Çorum 0.021 0.019 -0.014 -0.003 Amasya 0.165 -0.015 -0.098 0.012 Trabzon 0.002 0.002 -0.001 -0.002 Ordu 0.001 0.001 0.002 -0.005 Giresun -0.030 0.009 0.016 0.007 Rize 0.016 0.013 -0.014 -0.012 Artvin 0.008 0.000 -0.008 0.004 Gümüshane -0.133 0.031 0.077 -0.034 RESULT In this study, there is a positive relationship between the number of arriving tourists and the length of stay and economic growth. Number of tourists and duration of stay were examined separately for tourism certified enterprises and municipal certified enterprises. Also applied to the variables under study CCEGM Dynamic modeling results in terms of the number of incoming tourists and their stay in Turkey's 81 provinces have been shown to have different effects on economic growth. REFERENCES Arslan, A. (2008). Türkiye’de Ekonomik Büyüme ve Turizm İlişkisi Üzerine Ekonometrik Analiz. Erciyes Üniversitesi Sosyal Bilimler Enstitüsü Dergisi, 1(24), 1-12. PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Atılgan, E., Ertuğrul, H., & Basar, D. (2017). Sağlık Harcamaları ve Ekonomik Büyüme İlişkisi: OECD Ülkeleri için Bir Uygulama. Internatıoanl Economics Research and Financial Markets Congress Proceedings (s. 423-430). Edirne: IERFM. Bahar , O. (2006). Turizm Sektörünün Türkiye’nin Ekonomik Büyümesi Üzerindeki Etkisi: VAR Analizi Yaklaşımı. Yönetim ve Ekonomi, 13(2), 137-150. Balaguer, J., & Cantavella-Jorda, M. (2002). Tourism as a long-run economic growth factor: the Spanish case. Applied Economics, 877-884. Barutçugil, İ. (1998). Turizm İşletmeciliği. İstanbul: Beta Basın Yayın Dağıtım A.Ş. Brida, J., Sanchez Carrera, E., & Risso, W. (2008). Tourism’s Impact on Long-Run Mexican Economic Growth. Economics Bulletin, 3(21), 1-8. Chang, C.-L., Khamkaew, T., & McAleer, M. (09.08.2018). Estimation of a Panel Threshold Model of Tourism Specialization and Economic Development. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1583242. adresinden alınmıştır Chudik, A., & Pesaran, M. (2015). Common correlated effects estimation of heterogeneous dynamic panel data models with weakly exogenous regressors. Journal of Econometrics, 188(2), 393-420. Cortes-Jimenez, İ., Pulina, M., Prunera, C., & Artis , M. (2009). Tourism and Exports as a means of Growth. Research Institute of Applied Economics, 1-28. Çağlayan, E., Güriş, B., & Öskönbayev, Z. (2012). Turizme Dayalı Büyüme Hipotezinin Kuzey Kıbrıs Türk Cumhuriyeti için Geçerliliğinin Analizi. Trakya Üniversitesi Sosyal Bilimler Dergisi, 14(2), 105-122. Kaya, A., & Canlı, B. (2013). Türkiye’ye Yönelik Uluslararası Turizm Talebinin Belirleyenleri: Panel Veri Yaklaşımı. Anadolu Üniversitesi Sosyal Bilimler Dergisi, 43-54. Pesaran, M. (2004). General diagnostic tests for cross section dependence in panels. ESRC Working Paper. Pesaran, M. (2007). A simple panel unit root test in the presence of cross-section dependence. Journal of Applied Econometrics, 22(2), 265-312. Pesaran, M., & Yamagata, T. (2008). Testing slope homogeneity in large panels. Journal of Econometrics, 142(1), 50-93. PRIZREN SOCIAL SCIENCE JOURNAL / Volume 2, Issue 2; May - August 2018 / ISSN: 2616-387X Pesaran, M., Ullah, A., & Yamagata, T. (2008). A bias-adjusted LM test of error crosssection independence. The Econometrics Journal, 11(1), 105-127. TÜİK.(01.06.2018). https://biruni.tuik.gov.tr/bolgeselistatistik/degiskenlerUzerindenSorgula.do?durum=a cKapa&menuNo=273&altMenuGoster=1&secilenDegiskenListesi=. UNWTO (World Tourism Organization). (09.08.2018). Annual Report 2017. www.unwto.org. adresinden alınmıştır