Tourism and economic growth: a meta-analysis of panel data studies

1 TOURISM AND ECONOMIC GROWTH: A me a-analysis o panel da a s udies Me cedes Cas o-Nuñoª, José A. Molina-Toucedo b and Ma ía del P. Pablo-Rome o c a. Depa men o Economic Analysis and Poli ical Economy, A da. Ramon y Cajal, 1 Uni e si y o Se ille, E-41018 Se ille, Spain Tel.: (+34) 954 55 44 77 Fax: (+34) 954 55 76 29 [email p o ec ed]s b. Depa men o Economic Analysis and Poli ical Economy, A da. Ramon y Cajal, 1 Uni e si y o Se ille, E-41018 Se ille, Spain Tel.: (+34) 954 55 75 29 Fax: (+34) 954 55 76 29 [email protected] c. Depa men o Economic Analysis and Poli ical Economy, A da. Ramon y Cajal, 1 Uni e si y o Se ille, E-41018 Se ille, Spain Tel.: (+34) 954 55 76 11 Fax: (+34) 954 55 76 29 mpablo[email p o ec ed] Abs ac Al hough o decades i has been acknowledged ha ou ism likely con ibu es o economic g ow h, heo e ical models ha conside a causal ela ionship be ween bo h a e a ecen phenomenon. F om a sample o 11 s udies based on panel da a echniques published h ough o 2011, and o a o al o 87 he e ogeneous es ima ions, a me a- analysis is pe o med by applying models o bo h ixed and andom e ec s, wi h he main objec i e being o calcula e a summa y measu e o he e ec s o ou ism on economic g ow h. While he esul s ob ained poin o a posi i e elas ici y be ween economic g ow h and ou ism, he magni ude o he e ec was ound o a y acco ding o he me hodological p ocedu e employed in he o iginal s udies o empi ical es ima ions. KEYWORDS: Panel da a, Tou ism, Economic G ow h, Me a-analysis, Elas ici ies. JEL Codes: C33, C83, L83, O40, O50. 2 1. INTRODUCTION Al hough o decades i has been ecognized ha ou ism could con ibu e in some way o economic g ow h, heo e ical models ha conside a causal ela ionship be ween ou ism and economic g ow h a e a ecen phenomenon (Kim e al., 2006). Lanza & Piglia u (2000) we e he i s o in es iga e his ela ion om an empi ical poin o iew, while Balague & Can a ella-Jo da (2002) we e he i s o analyse he ou ism-led g ow h hypo hesis (TLG) – i.e. he hypo hesis acco ding o which ou ism gene a es economic g ow h – om an econome ic pe spec i e. F om he i s a emp , an inc easing numbe o a icles wi h he same objec i e – al hough o di e en coun ies, using di e en me hodologies and ob aining di e en esul s –ha e been published. Mos o he s udies a e based on ime se ies and e e o a single economy. Among hem, and wi hou p o iding an exhaus i e lis , se e al wa an men ion such as ha o D i sakis (2004), Du ba y (2004), Ongan & Demi oz (2005), Gunduz & Ha emi-J (2005), Oh (2005), Kim e al. (2006), Ka i cioglu (2007, 2009, 2010), Lee & Chien (2008), B ida & Risso (2009). B ida e al. (2010), Chen & Chiou-Wei (2009) Jin (2011), Lean & Tang (2010), and A slan u k e al. (2011). O hese, mos suppo he TLG hypo hesis. Among hese s udies he e is a g oup which uses panel da a analysis o in es iga e he TLG hypo hesis. Al hough he numbe o hese s udies is smalle han ha o ime-se ies s udies, a la ge numbe o coun ies a e included. This pe mi s an unde s anding o be gained o he ela ionships ha occu ac oss a g oup o coun ies (Lee & Chang, 2008) and an e alua ion o be made o he b oade o global impac o ou ism (Sequei a & Nunes, 2008). Fu he o his, he ela ion be ween G oss Domes ic P oduc (GDP) and ou ism in hese s udies is usually no isola ed because o he a iables ha a e essen ial o g ow h a e also conside ed. 3 The aim o he p esen wo k is o e i y whe he ou ism con ibu es o economic g ow h and o de e mine he magni ude o his con ibu ion by calcula ing global measu es based on published scien i ic e idence a ailable h ough un il 2011. To his end, me a-analysis echniques ha e been used, which allow he quan i a i e syn hesis o nume ous es ima ions ob ained in p e ious s udies, as well as o de e mine how ce ain me hodological app oaches in luence alues ob ained in hese es ima es. We will only conside hose s udies based on panel da a o co obo a e he TLG hypo hesis as hey p o ide global es ima ions which, consis en wi h Lee & Chang (2008), e e o la ge samples o coun ies. Despi e he ex ao dina y g ow h o he s udies examining he ela ionship be ween ou ism and economic g ow h using panel da a, in such a sho ime, no quan i a i e sys ema ic e iew ha in eg a es all o he in o ma ion has ye been made. The use o me a-analysis was in oduced by Glass, (1976). In con as o he adi ional na a i e e iew, he basic pu pose o me a-analysis is o p o ide he same me hodological igo o a li e a u e e iew ha is equi ed o expe imen al esea ch (Rosen hal, 1995). In he case o economic g ow h and de elopmen s udies, his echnique has been used o in eg a e indings on he e ec s o iscal policies (Nijkamp & Poo , 2004; Phillips & Goss, 1995), he in luence o income inequali y condi ions o poli ical s uc u es (De Dominicis, Flo ax, & De G oo , 2008; Doucouliagos & Ulubasglu, 2008), he con ibu ion o social capi al o economic g ow h (Wes lund & Adam, 2010) and popula ion g ow h (Headey & Hodge, 2009), o he e ec i eness o de elopmen aid (Doucouliagos & Paldan, 2008). In he ield o ou ism, gene al applica ions o his echnique can be ound in ou ism esea ch (Dann, Nash & Pea ce, 1988), ou ism o ecas ing (Calan one, Di Benede o & Bojanic, 1987), and mo e speci ically on ou is and economic impac s udies (Wagne , 2002; Wagne & Wobe , 2003). 4 Me a-analyses o pa icula impo ance in his ield conce n hose pe o med on ou ism income mul iplie s (Baaijens, Nijkamp & Van Mon o , 1998), egional ou ism mul iplie s (Baaijens & Nijkamp, 2000), and ou ism demand (C ouch, 1995; Lim, 1999). Mo e ecen ly, epo s ha e been published conce ning speci ic b anches o ou ism such as ha by Ca lsen & Boksbe ge (2011) on wine ou ism, Weed (2009) on spo s ou ism and Sa iisik, Tu kay & Ako a (2011) on yach ing ou ism. This wo k is di ided in o six sec ions which desc ibe he me a-analysis app oach aken wi h espec o ou ism-economic g ow h. 2. METHODOLOGY. Following Glass e al. (1981) and Lipsey & Wilson (2001), he me a-analysis me hod consis s o deducing a summa y e ec based on he combina ion o di e en es ima ions (e ec sizes) om a selec ed sample o s udies by means o di e en s a is ical echniques: he ixed-e ec s model (FEM) and he andom-e ec s model (REM). Unde a FEM (Bo ens ein e al. 2009) he selec ed s udies a e combined on he p emise ha he e is no he e ogenei y among hem and he only de e minan s o he weigh o each s udy in he me a-analysis would be i s sample size and i s own a iance o wi hin-s udy a iance (in e se a iance weigh ed me hod: Bi ge, 1932 and Coch an, 1937). Assuming a sample o "m" es ima es o e ec sizes, (i = 1, 2… m), ep esen ing a measu e o an analyzed e ec called T i , a summa y e ec   may be o mula ed as (Bo ens ein e al., 2009):   ∑  ∑ [1], whe e w i is he s a is ical weigh o he i- h es ima ion:    1   [2] , and   he a iance o he i- h es ima ion, so ha : ∑   1 . I is possible ha he a iabili y among s udies is highe han ha expec ed by pu e andomness, which would be de ec able in he i s ins ance by es ing he 5 hypo hesis o homogenei y. The mos widely used es was o iginally de eloped by Coch an (1954); i calcula es he pa ame e Q, acco ding o:   ∑        [3]. Because o he low powe o his es , highligh ed by Takkouche e al. (1999) i is ecommended ha a subg oup analysis o s udies o ha addi ional p ocedu es o quan i y he possible he e ogenei y, such as he pa ame e I 2 (Higgins e al., 2003), which indica es he p opo ion o he a ia ion be ween s udies (be ween-s udies a iance) in he o al a ia ion due o he e ogenei y:       [4], whe e   (Tau- Squa ed) is he be ween-s udies a iance and   he wi hin-s udy a iance (due o andomness). I he e ogenei y is de ec ed, a REM should be app op ia ed (Bo ens ein e al., 2009), which conside s ha he es ima ed e ec s o he included s udies a e only a andom sample o all hose possible, and he ue e ec sizes o hem would be dis ibu ed abou a mean e ec (wi h wo possible sou ces o a ia ion: ha exis wi hin he s udies o andom e o and he a ia ion be ween s udies o ue dispe sion). Applying he a iance weigh ed me hod, unde he REM, exp ession [2] is ans o med and we ha e, o each i- h es ima ion, adjus ed weigh s (    acco ding o [5]:         [5], whe e   (Tau-Squa ed) is he be ween-s udies a iance and w i he s a is ical weigh o an i- h es ima ion unde a FEM. Wi h ega ds o he summa y e ec   (i.e. a mean e ec ob ained om "a dis ibu ion o e ec sizes") we can calcula e:   ∑  ∑  [6]. The possibili y o ob aining a biased summa y e ec mus be assessed, which is de i ed om he p esence o publica ion biases as a esul o he ac ha many comple ed s udies a e no ac ually published because hey do no achie e signi ican e ec s, because hey a e un a o able o because hey ha e nega i e ou comes (S e ne e al., 2000; Tho n on & Lee, 2000). 6 Analy ically, he publica ion biases can be de ec ed by he s a is ical me hods o Begg (Begg & Mazumba, 1994), and Egge (Egge e al., 1997), and g aphically by he namely unnel plo s diag ams. Howe e , he limi a ions o hese me hods (Tho n oln & Lee, 2000; Macaskill e al., 2001), equi e applica ion o Du al and Tweedie's T im and Fill echnique (Du al & Tweedie, 2000), which allows he numbe o missing s udies o be de e mined and added o he analysis, ollowing which he combined e ec is ecompu ed. Finally, o assess he obus ness o s abili y o he calcula ed summa y e ec , sensi i i y analysis is pe o med. 3. PANEL DATA STUDIES AND ESTIMATIONS. De ails o 13 s udies published h ough o 2011 which use panel da a o analyze he ela ionship be ween ou ism and economic g ow h a e gi en in Table 1. These s udies we e iden i ied by li e a u e sea ch echniques using Scopus, ScienceDi ec , Google Schoola and he main jou nals in ou ism esea ch 1 , using e ms as: ou ism, economic g ow h, ou ism led g ow h hypo hesis and ela ed e ms. Pape s om o he s udies iden i ied we e also used, which include no only a icles in scien i ic jou nals lis ed in Jou nal Ci a ion Repo s (JCR) o o he da abases, bu also wo king pape s (Wpape ) published on he In e ne ha ha e eached a ce ain scien i ic ecogni ion on accoun o hei quali y o numbe o ci a ions. All he s udies included in he analysis a e shown in Table 1 wi h an iden i ica ion code, and he numbe o es ima ions in each s udy. Each o hese es ima ions di e s depending on he es ima ion model, whe he o no addi ional a iables we e used o explain economic g ow h, he ype o a iable used o measu e 1 Ryan (2005) shows he anking and a ing o academics and jou nals in ou ism esea ch. 7 ou ism, whe he he sample was classi ied in o subsamples, and he inclusion o no o ins umen al a iables o dummy a iables o econome ic es ima ion. Table 1: Panel da a s udies showing he ela ionship be ween ou ism and economic g ow h. Au ho Yea o s udy Code Classi ica ion o s udy Sample Analysed pe iod No. o es ima ions Eugenio- Ma in e al. 2004 Eug Wpape La in Ame ican coun ies 1985-1998 4 Sequei a & Campos 2005 Seca Wpape 72 coun ies 1980-1999 6 Sequei a & Nunes 2008 Sequ JCR. Q3 Small, poo and no mally de eloped coun ies 1980-2002 16 Fayissa e al. 2008 Fayi JCR. Q3 Sub-Saha an coun ies 1995-2004 4 Lee & Chang 2008 Lee JCR. Q1 OCDE, Asia, La in Ame ican and sub- Saha an coun ies 1990-2002 20 Co és- Jiménez 2008 Co JCR. Q3 Coas al egions o I aly and Spain 1990-2000 12 P oenca & Soukiazis 2008 Sou JCR. Q4 Po ugal egions NUT II and NUT III 1993-2001 6 Fayissa e al. 2009 Fay Wpape La in Ame ican coun ies 1995-2004 4 Adamau & Cle ides 2010 Adam open jou nal 162 coun ies 1980-2005 10 Na ayan e al. 2010 Na a JCR. Q3 4 islands 1988-2004 2 Holzne 2011 Holz JCR. Q1 99 coun ies 1970-2007 4 See enah 2011 See JCR. Q1 Paci ic Islands and de eloped coun ies 1995-2007 6 D i sakis 2011 D i JCR. Q3 Medi e anean coun ies 1980-2007 2 Sou ce: Own elabo a ion. Two di e en empi ical models a e gene ally es ima ed: dynamic and non- dynamic. The i s is de ined econome ically as ollows [7], in gene al e ms: i ii i i i uXTyy ε λ β φ α + + + + + = −1 [7], whe e y is he loga i hm o eal pe capi a GDP, T is a measu e o ou ism de elopmen exp essed in loga i hmic e ms, X 8 ep esen s a ec o o o he explana o y a iables, α is a pe iod-speci ic in e cep e m o cap u e changes common o all coun ies, u is an unobse ed coun y-speci ic and ime-in a ian e ec , ε is he e o e m and he subsc ip s i and ep esen coun y and ime pe iod, espec i ely. Non-dynamic models a e speci ied simila ly, bu wi hou he e m 1−i y φ . They can be de ined in gene al as ollows in [8]: i ii i i uXTy ε λ β α + + + + = [8]. The pa ame e β, which e lec s he es ima ed impac o ou ism on he GDP, eaches a di e en in e p e a ion: in non-dynamic models, i e lec s he elas ici y o p oduc i i y wi h espec o ou ism (because he a iables a e exp essed in na u al loga i hms); while in dynamic models; i e lec s only pa o he e ec o ou ism on p oduc i i y (which is p oduced in he same pe iod). The e ec s o ou ism expand in ime, which is o say ha ou ism has an e ec on p oduc i i y a ious pe iods he ea e depending on he alue o φ in [7]. I espec i e o whe he he models a e dynamic o no , he s udies also di e in e ms o hose ha use addi ional a iables such as educa ion, physical capi al, e c., o explain he g ow h o eal pe capi a GDP, compa ed wi h hose ha ela e only o he ou ism g ow h a iable (A o B espec i ely) . O he impo an di e ences can be summa ized as ollows: 1. Whe he he empo al e ec is included by i ue o he coe icien α in he es ima ion ( ime dummies used); 2. The p oxy o ou ism expansion, which is used o de ine T 2 (indica o s o ou ism a i als s. indica o s o ou ism eceip s); 3. Those es ima ions ha using ins umen al a iables in es ima ing he unc ion o no ; 4. Depending on he wide sample o coun ies o a speci ic se o coun ies ha make up he panel da a. 2 Soukiazis & P oenca (2008) use as a p oxy o he ou ism a iable he accommoda ion capaci y o he ou ism sec o . This p oxy is no based on ou ism a i als and ou ism eceip s, and is he e o e no classi ied in ou me a-analysis ollowing his c i e ion. 9 Gi en hese di e ences, he me a-analysis conside s di e en g oups o simila es ima ions, as shown in Table 2. The e a e wo se s o es ima ions ( ype 1 scena ios) – dynamic and non-dynamic – because as s a ed abo e, he es ima ed β coe icien s a e no di ec ly compa able be ween he wo model ypes. Wi hin each ype 1 scena io, 3 clus e s can in u n be made: hose ha con empla e he whole sample o each scena io (o e all), and es ima ions ha include only ype A o ype B es ima ions ( ype 2 scena ios). Fu he mo e, wi hin each ype 2 scena io, 9 clus e s can be o med: hose ha con empla e he whole sample o es ima ions o he scena io (o e all) and ype 3 scena ios. These combina ions gi e ise o a o al o 42 scena ios. 4. META-ANALYSIS RESULTS. Thi een s udies we e iden i ied in ou sea ch, bu he es ima ions o Sequei a & Campos (2005) and Fayissa e al. (2009), we e excluded om he me a-analysis because he da a p o ided we e insu icien . The s udy hus encompassed he empi ical esul s om 11 p e ious s udies (Table 1) ha ga e ise o a o al o 87 es ima ions ( he sample o ou me a-analysis amewo k) epo ed in he o m o elas ici ies , which exp ess he impac o he ou ism sec o on economic g ow h. Table 2 summa izes he main esul s o he me a-analysis pe o med on ha sample. 16 o e all sample o dynamic and non-dynamic scena ios, as well o hei espec i e A and B subg oups. I was ound ha he inclusion o empo a y a iables and he use o ins umen al a iables ended o dec ease he andom poin es ima ions alue in all o he ype 2 scena ios, and ha he andom poin es ima ions alue ob ained o hese scena ios we e highe when a el income was used o measu e ou ism han when he numbe o a i als was used. Fu he o his, he andom poin es ima ion ended o be g ea e when he es ima ions ha conside only la ge samples o coun ies (gene al coun ies) we e used. In such cases, he es ima ed alue is unbiased. I es ima ions e e only o speci ic coun ies, i.e. samples e e only o coun ies wi h a ce ain p o ile, he elas ici y ends o be lowe . Howe e , i mus be aken in o accoun ha g oups o coun ies in his scena io we e e y di e se, anging om Asian, La in Ame ican, Medi e anean and sub-Saha an coun ies, o island g oups, economically poo coun ies, small coun ies. This sugges s ha a mo e de ailed s udy o elas ici ies is equi ed based on he cha ac e is ics o he coun ies in hose samples. 6. CONCLUSIONS Theo e ical models ha conside a causal ela ionship be ween economic g ow h and ou ism a e a mo e ecen phenomenon. Since 2002, an inc easing numbe o a icles ha ha e in es iga ed his ela ion om an econome ic pe spec i e ha e been published. A conside able p opo ion o hese s udies a e based on panel da a o analyze e ec s o ou ism on economic g ow h ac oss a la ge numbe o coun ies. Acco ding o he me a-analysis p esen ed in his pape , om 87 es ima ions ob ained using panel da a echniques we can conclude ha ou ism posi i ely a ec s economic g ow h. Howe e , 17 he du a ion o his posi i e e ec depends on me hodological ea u es o es ima ions made in he o iginal s udies. Thus, he me a-analysis applied o es ima ions based on dynamic unc ions shows ha elas ici y ( he p oduc i i y wi h espec o ou ism) in he sho e m is small, yielding a andom poin e-adjus ed es ima ion o 0.002. Howe e , he ini ial e ec is p olonged in ime, so ha in he long e m he a e age alue o he elas ici y is aised o 0.179, o signi ican and s able es ima ions. The me a-analysis applied o es ima ions based on non-dynamic unc ions showed ha he elas ici ies had an a e age alue o 0.266 o he o e all sample. The alue o hese elas ici ies is none heless a ec ed by a ange o ea u es used in he es ima ions ca ied ou . We ound ha as he model becomes mo e speci ic, he alue o elas ici y, i espec i e o he case, ends o dec ease. Thus, he inclusion o explana o y a iables o economic g ow h, in addi ion o ha o ou ism, ends o educe he alue o he elas ici y, especially wi h espec o non-dynamic models. Also, when empo a y a iables and ins umen al a iables a e conside ed, he alue o he elas ici y ends o diminish. Fu he mo e, he a iable used o measu e ou ism also a ec s he elas ici y. I ou ism is measu ed in e ms o a el income, hen elas ici y ends o be highe han i he ou ism is measu ed in e ms o ou is a i als. Finally, i should be no ed ha he a e age elas ici y calcula ed in ou me a- analysis was highe only when he s udies included we e based on a la ge sample o ype-speci ic coun ies. Based on he me a-analysis p esen ed, we we e unable o deduce a clea pa e n o how elas ici y a ies ac oss de ined g oups o coun ies, because he es ima ions ob ained using panel da a o hese de ined g oups we e e y he e ogeneous and di icul o compa e in his ega d. Thus, i may be in e es ing o pe o m a me a-analysis on which he a e age elas ici y is de ined o g oups o 18 coun ies and ob ained using es ima ions om ime se ies and ela ed o indi idual coun ies. REFERENCES A slan u k, Y., Balcila , M., & Ozdemi , Z. A. (2011). Time- a ying linkages be ween ou ism eceip s and economic g ow h in a small open economy. Economic Modelling, 28, 664– 671. Baaijens, S.R., Nijkamp, P., & Van Mon o , K. (1998). Explana o y me a-analysis o he compa ison and ans e o egional ou is income mul iplie s. Regional S udies, 32, 839-849. Baaijens, S.R., & Nijkamp, P. (2000). Me a-analy ic me hods o compa a i e and explo a o y policy esea ch: An applica ion o he assessmen o egional ou is mul iplie s. Jou nal o Policy Modeling, 22, 821–858. Balague , J., & Can a ella-Jo dà, M. (2002). Tou ism as a long- un economic g ow h ac o : he Spanish case. Applied Economics, 34, 877-884. Begg, C.B., & Mazumba , M. (1994). Ope a ing cha ac e is ics o a ank co ela ion es o publica ion bias. Biome ics, 50, 1088-1101. Bi ge, R.T. (1932). The calcula ion o e o s by he me hod o leas squa es. Physical Re iew, 40, 207–227. Bo ens ein, M., Hedges, L., Higgins, J.P.T., & Ro hs ein, H.R. (2009). In oduc ion o me a- analysis. Chiches e , UK: Wiley. B au, R., Lanza, A., & Piglia u, F. (2007). How as a e small ou ism coun ies g owing? E idence om he da a o 1980-2003. Tou ism Economics, 13(4), 603-613. B ida, J.G., Lanzilo a, B., Lione i, S., & Risso, W.A. (2010). The ou ism-led g ow h hypo hesis o U uguay. Tou ism Economics, 16 (3), 765–771. B ida, J.G. & Risso, W. A. (2009). Tou ism as a ac o o long- un economic g ow h: An empi ical analysis o Chile. Eu opean Jou nal o Tou ism Resea ch, 2(2), 178-185. 19 Calan one, R.J., Di Benede o, C.A., & Bojanic, D. (1987). A comp ehensi e e iew o he ou ism o ecas ing li e a u e. Jou nal o T a el Resea ch, 26, 28-39. Ca lsen, J., Boksbe ge , P. (2011). The blending o wine and ou ism expe iences: A me a- analysis. In M.J. G oss (Ed.), CAUTHE 2011 Na ional con e ence ou ism: c ea ing a b illian blend (pp. 983-989). Adelaide, S.A.: Uni e si y o Sou h Aus alia. Chen, C.-F., & Chiou-Wei, S. Z. (2009). Tou ism expansion, ou ism unce ain y and economic g ow h: New e idence om Taiwan and Ko ea. Tou ism Managemen , 30, 812–818. Coch an, W. G. (1937). P oblems a ising in he analysis o a se ies o simila expe imen s. Jou nal o he Royal S a is ical Socie y, Supplemen a y, 4, 102-118. Coch an, W. G. (1954). The combina ion o es ima es om di e en expe imen s. Biome ics, 10, 101-129. Co és-Jiménez, I. (2008). Which ype o ou ism ma e s o he egional economic g ow h ? The cases o Spain and I aly. In e na ional Jou nal o Tou ism Resea ch, 10, 12-139. C ouch, G. I. (1995). A me a-analysis o ou ism demand. Annals o Tou ism Resea ch, 22, 103-118. Dann, G., Nash, D., & Pea ce, P. (1988). Me hodology in ou ism esea ch. Annals o Tou ism Resea ch, 15, 1-28. De Dominicis, L., Flo ax, R.J., & De G oo , H.L. (2008). A me a-analysis on he ela ionship be ween income inequali y and economic g ow h. Sco ish Jou nal o Poli ical Economy, 55, 654-682. Doucouliagos, H., & Ulubaşoğlu, M.A. (2008). Democ acy and economic g ow h: A me a- analysis. Ame ican Jou nal o Poli ical Science, 52, 61–83. Doucouliagos, H., Paldam, M. (2008). Aid e ec i eness on g ow h: A me a s udy. Eu opean Jou nal o Poli ical Economy, 24, 1-24. D i sakis, N. (2004). Tou ism as a long- un economic g ow h ac o : an empi ical in es iga ion o G eece using causali y analysis. Tou ism Economics, 10(3), 305-316. Du ba y, R. (2004). Tou ism and economic g ow h: The case o Mau i ius. Tou ism Economics, 10(4), 389-401. 20 Du al, S., & Tweedie, R. (2000). T im and ill: a simple unnel plo based me hod o es ing and adjus ing o publica ion bias in me a-analysis. Biome ics, 56, 455-463. Egge , M., Smi h, G.D., Schneide , M., & Zinde , Ch. (1997). Bias in me a-analysis de ec ed by a simple, g aphical es . B i ish Medical Jou nal, 31, 629-634. Egge , M., Smi h, G.D. & Al man, D.G. (Eds.). (2001). Sys ema ic e iews in heal h ca e: Me a-analysis in con ex (2ª ed.). London: BMJ Books. Eugenio-Ma ín, J. L., Mo ales, N. M., & Sca pa, R. (2004). Tou ism and economic g ow h in la in ame ican coun ies: a panel da a app oach. Fondazione Eni En ico Ma ei Wo king Pape Se ies, No a di La o o 26 Fayissa B., Nsiah C., & Tadasse B. (2008). Impac o ou ism on economic g ow h and de elopmen in A ica. Tou ism Economics, 14(4), 807–818. Fayissa B., Nsiah C., & Tadasse B. (2009). Tou ism and economic g ow h in La in Ame ican coun ies (LAC): Fu he empi ical e idence. Wo king Pape s 200902, Middle Tennessee S a e Uni e si y, Depa men o Economics and Finance Glass, G. V. (1976). P ima y, seconda y, and me a-analysis o esea ch. Educa ional Resea che , 5, 3-8. Glass, G.V., McGaw, B., & Smi h, M.L. (1981). Me a-analysis in social esea ch. Be e ly Hills, CA: SAGE Publica ions. Goco ali, U. (2010). Con ibu ion o ou ism o economic g ow h in Tu key. An In e na ional Jou nal and Tou ism and Hospi ali y Resea ch 21(1), 139-153. Gunduz, L., & Ha emi-J, A. (2005). Is he ou ism-led g ow h hypo hesis alid o Tu key?. Applied Economics Le e s, 12, 499-504. Headey, D., & Hodge, A. (2009). The e ec o popula ion g ow h on economic g ow h: a me a- eg ession analysis o he mac oeconomic li e a u e. Popula ion and De elopmen Re iew, 35, 221–248. Higgins, J.P.T., Thompson, S.G., Deeks, J.J., & Al man, D.G. (2003). Measu ing inconsis ency in me a-analyses. B i ish Medical Jou nal, 327, 557-560. 21 Holzne , M. (2011). Tou ism and economic de elopmen : The beach disease?. Tou ism Managemen , 32, 922-933. Jin, J.C. (2011). The E ec s o ou ism on economic g ow h in Hong Kong. Co nell Hospi ali y Qua e ly, XX(X), 1–8. Ka i cioglu, S.T. (2007). Tou ism, ade and g ow h: he case o Cyp us. Applied Economics, 41, 2741-2750. Ka i cioglu, S.T. (2009). Re ising he ou ism-led-g ow h hypo hesis o Tu key using he bounds es and Johansen app oach o coin eg a ion. Tou ism Managemen , 30, 17-20. Ka i cioglu, S. (2010). Tes ing he ou ism-led g ow h hypo hesis o Singapo e – an empi ical in es iga ion om bounds es o coin eg a ion and G ange causali y es s. Tou ism Economics, 16(4), 1095-1101. Kim, H. J., Chen, M. H., & Jang, S. (2006). Tou ism expansion and economic de elopmen : he case o Taiwan. Tou ism Managemen , 27(5), 925–933. Lanza A., & Piglia u F. (2000). Tou ism and economic g ow h: does coun y’s size ma e ?. Ri is a In e nazionale di Scienze Economiche e Comme ciali, 47, 77-85. Lean, H.H., & Tang, C.F. (2010). Is he ou ism-led g ow h hypo hesis s able o Malaysia? A no e. In e na ional Jou nal o Tou ism Resea ch, 12(4), 375-378. Lee, C.-C., & Chang, C-P. (2008). Tou ism de elopmen and economic g ow h: A close look o panels. Tou ism Managemen , 29, 80-192. Lee, C-C., & Chien, M-S. (2008). S uc u al b eaks, ou ism de elopmen , and g ow h: E idence om Taiwan. Ma hema ics and Compu e s in Simula ion, 77, 358-368. Lim, C. (1999): A Me a-Analy ic Re iew o In e na ional Tou ism. Jou nal o T a el Resea ch, 37, 273-284 Lipsey, M.W., & Wilson, D.B. (2001). P ac ical me a-analysis. Thousand Oaks: Sage. Lucas, R. E. (1988). On he mechanics o economic de elopmen . Jou nal o Mone a y Economics, 22, 3-42 Macaskill, P., Wal e , S.D., & I ing, L. (2001). A compa ison o me hods o de ec publica ion bias in me a-analysis. S a is ics in medicine, 20, 641-654. 22 Malik, S., Chaudh y, I.S., Sheikh, M.R., & Fa ooqi F.S. (2010). Tou ism, economic g ow h and cu en accoun de ici in Pakis an: E idence om co-in eg a ion and causal analysis. Eu opean Jou nal o Economics, Finance and Adminis a i e Sciences, 22, 1450-2275. Na ayan, P.K., Na ayan, S., P asad, A., & P asad, B.C. (2010). Tou ism, and economic g ow h: a panel da a analysis o Paci ic Island coun ies. Tou ism Economics, 16(1), 169-183. Nijkamp, P., & Poo , J. (2004): Me a-analysis o he e ec o iscal policies on long- un g ow h. Eu opean Jou nal o Poli ical Economy, 20, 91-124. Oh, C-O. (2005). The con ibu ion o ou ism de elopmen o economic g ow h in he Ko ean economy. Tou ism Managemen , 26, 39-44. Ongan, S., & Demi öz, D. M. (2005). The con ibu ion o ou ism o he long- un Tu kish economic g ow h. Ekonomicky casopis (Jou nal o Economics) 53 (9), 880-894. Palma-Pé ez, S., & Delgado-Rod íguez, M. (2006). P ac ical conside a ions on de ec ion o publica ion bias. Gace a Sani a ia, 20, 10-16. Phillips, J.M., & Goss, E.P. (1995). The e ec o s a e and local axes on economic de elopmen : a me a-analysis, Sou he n Economic Jou nal, 62, 320-333. P oenca, S., & Soukiazis, E. (2008). Tou ism as an al e na i e sou ce o egional g ow h in Po ugal. Cen o de es udios de la Unión Eu opea 34. Ryan C. (2005). The anking and a ing o academics and jou nals in ou ism esea ch. Tou ism Managemen , 26, 657–662 Rosen hal, R. (1995). W i ing me a-analy ic e iews. Psychological Bulle in, 118, 183-192. Sa iisik, M., Tu kay, O., & Ako a, O. (2011). How o manage yach ou ism in Tu key: A swo analysis and ela ed s a egies. P ocedia Social and Beha io al Sciences, 24, 1014-1025 Schube , F.S., B ida, J.G., & Risso, W.A. (2010). The impac s o in e na ional ou ism demand on economic g ow h o small economies dependen o ou ism. Tou ism Managemen , 32, 377-385. Sengup a, J. K., & Espana, J. R. (1994). Expo s and economic g ow h in Asian NICs: an econome ic analysis o Ko ea. Applied Economics, 26, 41–51. 23 Sequei a, T. N., & Campos, C. (2005). In e na ional ou ism and economic g ow h: A panel da a app oach. Wo king Pape s 141, Fondazione Eni En ico Ma ei. Sequei a, T. N., & Nunes, P.M. (2008). Does ou ism in luence economic g ow h? A dynamic panel da a app oach. Applied Economics, 40, 2431-2441. See enah, B. (2011). Assessing he dynamic economic impac o ou ism o island S e ne, J.A.C., Ga aghan, D., & Egge , M. (2000). Publica ion and ela ed bias in me a- analysis: powe o s a is ical es s and p e alence in he li e a u e. Jou nal o Clinical Epidemiology, 53, 1119-1129. S e ne, J.A.C., Egge , M., & Smi h, G.D. (2001). In es iga ing and dealing wi h publica ion and o he biases in me a-analysis. B i ish Medical Jou nal, 323,101-105. Takkouche, B., Cada so-Suá ez, C., & Spiegelman, D. (1999). E alua ion o old and new es s o he e ogenei y in epidemiologic me a-analysis. Ame ican Jou nal o Epidemiology, 150, 206-215. Tho n on, A,. & Lee, P. (2000): Publica ion bias in me a-analysis: i s causes and consequences. Jou nal o Clinical Epidemiology, 53, 207-216. Wagne , D. M. (2002). Compa ing Eu opean ci ies ou ism gues su eys: an in elligen me a- analy ical app oach. In K.W. Wobe (Ed.), Ci y ou ism. (pp. 150-162). Wien: Sp inge - Ve lag Wien. Wagne , D. M., & Wobe , K. W. (2003). He e ogeneous ma ke esea ch in o ma ion in ou ism—implemen ing me a-analy ical ha moniza ion p ocedu es in a isi o su ey da abase. In A.J. F ew, M. Hi z & P. O’Conno (eds), In o ma ion and Communica ion Technologies in Tou ism 2003: P oceedings o he In e na ional Con e ence in Helsinki, Finland. (pp.76-85). Wien: Spinge -Ve lag. Weed, M. (2009). P og ess in spo s ou ism esea ch? A me a- e iew and explo a ion o u u es. Tou ism Managemen , 30, 615–628. Wes lund, H., & Adam, F. (2010). Social Capi al and Economic Pe o mance: A Me a-analysis o 65 S udies. Eu opean Planning S udies, 18, 893-919. 24 Wo ld Tou ism O ganiza ion (WTO) (2011). h p://www.wo ld- ou ism.o g/ egional/a ica/ p og amme/speci ic_p og amme.h m Wol , F.M. (1986). Me a-analysis: Quan i a i e me hods o esea ch syn hesis. Be e ly Hills, CA: Sage. ANNEX I Table 3. Long-Te m dynamic mul iplie o ou ism on economic g ow h CODE Model ype β φ Dynamic mul iplie (DM) Adam 1 A 0.002** (0.00056) -0.1** (0.0077) 0.02ª Adam 3 A 0.00018** (4.50E-05) -0.01** (0.0073) 0.018ª Adam 4 A 0.00012* (0.00006) -0.09** (0.017) 0.001ª Adam 6 A 0.0041** (0.00096) -0.101** (0.0076) 0.041ª Adam 7 A 0.0039** (0.0012) -0.101** (0.018) 0.038ª Adam 8 A 0.00048** (0.00012) -0.1** (0.0074) 0.004ª Holz 1 A 0.011** (2.00) 0.941*** (35.98) 0.186 Holz 2 A 0.018*** (2.84) 0.95*** (35.49) 0.360 Holz 3 A 0.008** (2.08) 0.97*** (52.93) 0.267 Co 1 A 0.001** (n.a) 0.895*** (n.a) 0.010 Co 2 A 0.006* (n.a) 0.884 *** (n.a) 0.052 Co 3 A 0.001** (n.a) 0.919*** (n.a) 0.012 Co 4 A 0.006*** (n.a) 0.907*** (n.a) 0.065 Co 6 A -0.015 (n.a) 0.891*** (n.a) -0.138 Co 8 A -0.017** (n.a) 0.942*** (n.a) -0.293 Co 9 A 0.001* (n.a) 0.831*** (n.a) 0.006 Co 10 A 0.007*** (n.a) 0.830*** (n.a) 0.041 Co 11 A 0.001** (n.a) 0.869*** (n.a) 0.008 Co 12 A 0.006*** (n.a) 0.857*** (n.a) 0.042 Eug 1 A 0.00036* (1.68) 0.777* (19.30) 0.007 Eug 2 A -0.0002* (2.54) 0.765* (12.64) -0.001 Eug 3 A 0.00063* (1.92) 0.738* (10.16) 0.002 Eug 4 A 0.00062* (2.63) 0.597* (4.14) 0.002 Fayi 3 A 0.0249*** (0.0081) 0.568*** (0.073) 0.058 See 1 A 0.12* (1,95) 0.24** (215) 0.158 See 2 A 0.06* (1.95) 0.23*** (2.52) 0.078 See 3 A 0.064* (1.96) 0.34*** (2.43) 0.097 See 4 A 0.14* (2.04) 0.17* (2.17) 0.169 See 5 A 0.033* (1.87) 0.25** (2.15) 0.044 See 6 A 0.08* (1.89) 0.37** (2.19) 0.127 Sequ 2 A 0.041** 0.927*** 0.562 25 (2.42) (17.47) Sequ 3 A 0.026* (1.92) 0.931*** (18.37) 0.379 Sequ 4 A 0.025* (1.85) 0.891*** (20.77) 0.229 Sequ 5 A 0.048** (3.77) 0.943*** (24.73) 0.842 Sequ 6 A 0.041** (2.69) 0.924*** (23.14) 0.539 Sequ 7 A 0.095*** (4.44) 0.87*** (7.96) 0.731 A e age alue o DM - - - 0.179 No e: Signi icance a ***1%, **5%, *10%, espec i ely. The es ima ed unc ion is yy β θ + = ∆ −1 . So yy β θ + + = −1 )1( , and θ β =MD n.a. No a ailable. Sou ce: Own elabo a ion.