Does tourism effectively stimulate Malaysia's economic growth?Chor Foon Tang a, *, Eu Chye Tan baCentre for Policy Research and International Studies, Universiti Sains Malaysia, 11800 USM, Penang, MalaysiabDepartment of Economics, Faculty of Economics and Administration, University of Malaya, 50603 Kuala Lumpur, Malaysiahi ghli ghts We examine the validity of the tourism-led growth hypothesis in Malaysia. We test the hypothesis within the Solow growth model. Wend that tourism Granger-causes economic growth in the short- and long-run. Tourism is catalyst for Malaysia's economic growth. Policies to promote inbound tourism could effectively stimulate economic growth.arti cle i nfoArticle history:Received 30 September 2013Accepted 28 June 2014Available online 18 July 2014Keywords:CausalityCointegrationEconomic growthMalaysiaTourismabstractThis study attempts to further verify the validity of the tourism-led growth hypothesis in Malaysia usinga multivariate model derived from the Solow growth theory. It employs annual data from 1975 to 2011.Wend that economic growth, tourism and other determinants are cointegrated. Specically, tourismhasapositiveimpactonMalaysia'seconomicgrowthbothintheshort-runandinthelong-run. TheGrangercausalitytestindicatesthat tourismGranger-causeseconomicgrowth. All thisprovidestheempirical support for the tourism-led growth hypothesis in Malaysia. In light of this, any policy initiativethat promotes tourism could contribute to Malaysia's economic growth. 2014 Elsevier Ltd. All rights reserved.1. IntroductionTraditionally, economic prosperity has been linked to growth inthe agricultural and manufacturing sectors as well as the inux offoreign capital. Whilst, the role of tourism in economic growth hasoftenbeendownplayedandregardedasanon-growthorientedsector, henceattractinglittleattentionof botheconomists andpolicymakers(Papatheodorou, 1999). Today, tourismhasbecomeone of the rapidly growing services sectors of the world. This hasprompted the Malaysian government to set tourism as a key sectorfor invigoratingMalaysia's long-termeconomic growth. Specif-ically, the 10th Malaysia Plan (2011e2015) has identied thetourism sector as one of the National Key Economic Areas (NKEAs)for transforming Malaysia into a high income nation by 2020.Nevertheless, asglobalisationgathersmomentum, viewsalsoexist that tourismmayinactual fact not signicantlystimulatelong-termeconomic growthas manyinformal agents bringinillegal workers to Malaysia using tourism as a channel. It may behardtodifferentiatebetweengenuine tourists andthosewhoactually arrive in search for jobs (Kassim, 1997). In 1995, only 600thousandforeignworkersinMalaysiawereillegal. Thenumbersubsequently increased to 2.1 million as observed during theimplementationof theIllegal Immigrant ComprehensiveSettle-ment Programme (Augustin& Lee, 2012).In view ofthese coun-terfactual data, doubts have arisen regarding the appropriateness ofemphasisingontourismas akeysector for drivinglong-termeconomicgrowthinorder toattainthehighincomestatusby2020. Asnotall touristarrivalsinvolvegenuinetourists, higherratesofarrivalsdonotnecessarilymeanhigherratesoftourismearnings. In fact, UNWTO (2012) noted that Malaysia's ranking intermsoftourismearningswasmuchlowerthantherankingbytouristarrivals. Inviewofthesereservations, thereisanurgentneed for a more accurate empirical assessment of the actual impactof tourism on Malaysia's economic growth.From our reading of earlier studies on the relationship betweentourism and economic growth, we nd that this topic has attractedalot ofinterest amongst researchers,particularly afterthestudyofBalaguerandCantavella-Jord a(2002). Recently, Brida, Cortes-*Corresponding author. Tel.: 60 4 653 2044; fax: 60 4 658 4820.E-mail addresses: tcfoon@usm.my, tcfoon@gmail.com (C.F. Tang).Contents lists available at ScienceDirectTourism Managementj ournal homepage: www. el sevi er. com/ l ocat e/ t ourmanhttp://dx.doi.org/10.1016/j.tourman.2014.06.0200261-5177/ 2014 Elsevier Ltd. All rights reserved.Tourism Management 46 (2015) 158e163Jimenez and Pulina (2014), Castro-Nu~ no, Molina-Toucedo, andPablo-Romero (2013), and Pablo-Romero and Molina (2013) havepublishedthreecomprehensivesurveysontherelationshipbe-tween tourism and economic growth. To conserve space, we wouldonly review some selected studies and those related to Malaysia.Forexample, NarayanandPrasad(2003), Dritsakis(2004), Brida,Carrera, and Risso (2008), Katircio glu (2009, 2011), Tang andAbosedra (2014a, 2014b), Hye andKhan(2013), Al-mulali, Fer-eidouni, Lee, andMohammed(2014), BouzahzahandEl Menyari(2013), and Jalil, Mahmood, and Idrees (2013) discovered thattourism affects economic growth. Therefore, they concluded that thetourism-led growth hypothesis is valid. However, Oh (2005), PayneandMervar (2010), andLee (2012) arguedthat it is economicgrowth that affects tourism rather than the other way round. Hence,these studies instead support the growth-led tourism hypothesis.As far as Malaysia is concerned, several studies have been con-ductedtoanalyse the role of tourismineconomic growthassummarised in Table 1. Generally, the table shows that the causalrelationshipbetweentourismandeconomicgrowthinMalaysiaremains a controversial subject. For example, Nanthakumar,Ibrahim, andHarun(2008), Kadir, Nayan, andAbdullah(2010)andTang(2011a)foundthattourismexpansionislesslikelytopromote economic growth. Whilst,Lau,Oh, and Hu (2009), LeanandTang(2010), OthmanandSalleh(2010), Kadir andKarim(2012), Othman, Salleh, andSarmidi (2012), Cheam, Mahmood,Abdullah, and Ong (2013), and Tang (2013) discovered thattourismexpansioncouldplayanimportant roleinstimulatingMalaysia's economic growth. All these studies however haveweaknesses that this study aims to address. They mainly involvedtheuseof bi-variatemodelsand/orad-hocmodel specicationswhich are not based upon any theoretical model.1Although esti-mation via ad-hoc model specication is relatively easy and simple,itishard tointerprettheresultsifthey arenotbaseduponanyeconomic theory. Apart fromthese, noneof thesestudies hasconsideredthepossibleimpactof structural breaksinunitroottesting.Accordingto Perron (1989),standardunit root tests mayhave lowpower whendata series containstructural break(s).Therefore, theresultsofthesepastMalaysianstudiesarehighlyquestionable and may be inaccurate.Motivated by the aforementioned shortcomings, the goal of thispaper is tore-investigatetheimpact of tourismexpansiononMalaysia'seconomicgrowthinamultivariateframework. Unlikethe earlier studies, we contribute to the literature by analysing theroleof tourisminMalaysia'seconomicgrowthbasedupontheneoclassical growthmodel. Variouseconometricapproachesareemployed in this study. First, apart from the standard augmentedDickey-Fuller (ADF) unit root test, we also employ the endogenousbreakunit root test developedbyZivotandAndrews(1992) todeterminetheorder of integrationof eachseries. Second, thesystem-wide cointegration technique proposed by Johansen (1988)and Johansen and Juselius (1990) is used to determine the presenceoflong-run equilibriumrelationshipsamongst economic growth,tourism and other determinants. Gonzalo (1994) revealed that thistechnique performs better than the other cointegration techniqueseven when the disturbance term is non-spherically distributed andthe lag structure is mis-specied. In addition, Tang (2011b) notedthat the Johansen-Juselius cointegration test is not sensitive to thechoice of the dependent variable because it treats all variables asendogenous. Lastly, thecausalrelationshipbetweentourismandeconomic growth in Malaysia will be ascertainedby the Grangercausality test.The rest of this paper is organised as follows. The next sectionwill explain the theoretical model and the data used in this study.Section3willdiscusstheeconometricproceduresfollowed. Theempiricalndings will then be presented in Section 4 followed byconcluding remarks in Section 5.2. Theoretical model and dataInthis study, we use the following Solow's (1956) growthframework derived from the CobbeDouglas production function tojustify the role of tourism in economic growth:ln

YtLt

b0q ln Z a1 aln st a1 alnn g dt(1)where ln denotes the natural logarithm, Yt is the output, Lt is totallabour, st is savings, n is the population growth rate, d is the rate ofdepreciationof capital stock, gis thegrowthrateof technicalprogress and Z is a vector of factors that affect the level oftech-nology and efciency in the economy.In this study,Z representstourismexpansionandaninstitutional factor suchas politicalstability. Therefore, the long-run growth model used in this studycan be written as:lnGNPt b0q1 lnTOURt q2 lnPSt b1 lnGNStb2 lnn g dt t(2)where t is the disturbance term. GNPt is per capita real gross na-tional product (GNP), TOURt per capita real tourism receipts, PSt ispolity2 which is an overall measure of political stability, GNSt is percapitareal grossnational savings(GNS), nisthegrowthrateofpopulation, g is the growth rate of technical progress andd is theTable 1Summary of published studies on the tourism-growth nexus in Malaysia.Authors SampleperiodMain methodologies MainndingsCointegratedDirectionofcausalityNanthakumaret al. (2008)1980e2007 Johansen test; GrangercausalityNo EG/TOURLau et al. (2009) 1972e2004 Johansen test; GrangercausalityYes TOUR/EGKadir et al.(2010)1994:Q1e2004:Q4Johansen test; GrangercausalityYes EG/TOURLean and Tang(2010)1989:M1e2009:M2TYDL Grangercausalitye TOUR4EGOthman andSalleh (2010)1976e2005 Johansen test; GrangercausalityYes TOUR/EGTang (2011a) 1995:M1e2009:M2ECM-based t-test;Granger causalityYes EG/TOURKadir and Karim(2012)1998e2005 Pedroni test; PanelGranger causalityYes TOUR/EGOthman et al.(2012)1970e2010 ARDL; GrangercausalityYes TOUR4EGCheam et al.(2013)1974e2010 Johansen test; GrangercausalityYes TOUR4EGTang (2013) 1974e2009 ARDL; GrangercausalityYes TOUR4EGTang and Tan(2013)1995:M1e2009:M2Combinedcointegration test;Recursive GrangercausalityYes TOUR/EGNotes: TOUR TourismandEG GDPor output growth. /represents uni-directional causality running from the variable on the left to the right, whereas 4represents bi-directional causality.1Studies of Lau et al. (2009), Lean and Tang (2010), Othman and Salleh (2010),Tang(2011a), KadirandKarim(2012) andTangandTan(2013) usedbi-variatemodels whichmay besubject to theomission of relevantvariables biasproblemas noted in Ltkepohl (1982).C.F. Tang, E.C. Tan / Tourism Management 46 (2015) 158e163 159depreciation rate of the capital stock. According to Mankiw, Romer,and Weil (1992), the growth rate of gd is approximately 5 per centand constant over time.Thisstudyutilisesannualtimeseriesdatafrom1975to2011extracted from variousdatabases. Annualdataon per capitarealGNP, per capitareal GNS, per capitareal tourismreceipts andpopulation growth rate are extracted fromthe InternationalFinancial Statistics(IFS), theWorldDevelopment Indicators(WDI)and the CEIC databases.However,the annual series on polity2 iscollected from Polity IV Project (Marshall, Gurr, & Jaggers, 2013).This dataset has been extensively used by earlier studies to measurepolitical stability (e.g. Aisen & Veiga, 2006, 2013; Zureiqat, 2005).The score of polity2ranges fromstronglyautocratic (10) tostrongly democratic (10). Therefore, high positive scores of polity2indicate politically stability. All variables are transformed intonatural logarithm to induce stationarity in the variance-covariancematrix and that therst differences of the variables can be inter-preted as growth terms.3. Econometric methods3.1. Zivot-Andrews unit root testIn order to examine the order of integration of the data series,weemployboththestandardADFunitroottestandtheZivot-Andrewsunitroottestwithonestructural break. Asthetestingprocedure for the ADF unit root test is well explained, the presentsectionmerelyfocusesontheZivot-Andrewsprocedure. Inthisstudy, we use two versions ofthe endogenousbreak model pro-posedbyZivot andAndrews(1992) todeterminetheorderofintegration. Model Aallows for a breakinthe intercept only,whereas Model C allows for a break in both the intercept and theslope of the trend function. The testing models can be written as:Model A : Dwt k0a1wt1b1t k1DUtXpj1cjDwtjx1t(3)Model C : Dwt k0a1wt1b1t k1DUt b2DTtXpj1cjDwtjx2t(4)whereDistherstdifferenceoperator, xitarewhitenoiseandnormally distributed disturbance terms. Dwtjare the laggeddependent variablesaddedintotheequationstoremoveserialcorrelationandtoensurethatthedisturbancestermsarewhitenoise. p is the optimal lag length determined by the Akaike's Infor-mation Criterion (AIC).DUt is the dummy variable for astructuralbreak in intercept occurring at time TB, whereas DTt is the dummyvariable for a trend shift, where DUt 1 and DTt tTB if t > TB and0otherwise. Thepotential breakpoint(TB)isselectedwheretheabsolute term of the ADF t-statistic for wt1 is maximised.3.2. System-wide JohanseneJuselius cointegration testTo perform the system-wide JohanseneJuselius cointegration test,we estimate the following vector error-correction model (VECM):DWt FD PWt1Xk1i1GiDWtiyt(5)whereWtisavectorof vevariables, namely, lnGNPt, lnTOURt,lnPSt,lnGNStand ln(ngd)t. D is a vector of deterministic com-ponents(i.e. theconstantsanddeterministictrends). Then, F, PandG are the 5 5 matrices of unknown parameters to be esti-matedandytisthedisturbanceterm. ThematrixPcontainsin-formation about the long-run relationship among the Wt variablesinthevector. Toexaminethepresenceofcointegratingrelation-ships, weapplytheLikelihoodRatio(LR)tracetestLR(ltrace)assuggested by Johansen (1988) and Johansen and Juselius (1990) inwhichLRltrace TPnir1ln1 li, wherelistheeigenvalueand T is the number of observations. If the LR(ltrace) test statistic isgreater than the critical value, this indicates that the null hypoth-esis of no cointegration can be rejected. Therefore, one canconclude that the variables are cointegrated and a meaningful long-run relationship can be derived.3.3. Granger causality analysisIn this section, we discuss the testing procedure for the Grangercausality test. In the context of the present study, we may either usethe rst difference vector autoregression (VAR) model or the vectorerror-correctionmodel (VECM)toexaminethedirectionofcau-sality between Malaysia's economic growth and tourism. Accordingto Granger (1988), if a set of variables are cointegrated, one shoulduse the VECM because it takes into account the short-run and long-runelements. However, intheabsenceofcointegration, onecanonly detect the short-run causal relationship using therst differ-ence VAR model. Assuming that the variables are cointegrated, thedirection of causality between economic growth and tourism canbe tested by estimating the following equations.DlnGNPt a0u1DUM98 Xkj1hjDlnGNPtjXkj04jDlnTOURtjXkj0gjDlnPStjXkj0fjDlnGNStjXkj0wjDlnn g dtjp1t1m1t(6)DlnTOURt a1u1DUM98 Xkj14jDlnTOURtjXkj0hjDlnGNPtjXkj0gjDlnPStjXkj0fjDlnGNStjXkj0wjDlnn g dtjp2t1m2t(7)whereD is therst difference operator, and t1 is the one periodlagged error-correction term derived from the long-run equation.mit's are the disturbance terms assumed to be normally distributedandwhitenoise. Totestthat DlnTOURtdoesnotGranger-causeDlnGNPtintheshort-run, weestimateequation(6)andtestthenull hypothesis H0:fj 0usingtheLikelihoodRatio(LR) test.Rejectionof thenull hypothesisindicatesthattourismGranger-causeseconomicgrowthintheshort-run. Similarly, totestthateconomicgrowthdoesnot Granger-causetourism, weestimateequation (7) and test the null hypothesis H0:hj0 using the LR test.For the long-run causality, one can test the signicance of the oneC.F. Tang, E.C. Tan / Tourism Management 46 (2015) 158e163 160period lagged error-correction term. From equation (6),p1s0 im-plies that there is causality fromtourism to economic growth in thelong-run, whereas p2 0inequation(7)meansthateconomicgrowth does not Granger-cause tourism in the long-run.4. Empirical resultsPrior to testing for the presence of cointegration, it is necessary todeterminetheorderof integrationof eachseries. Inthepresentstudy, we performthe ADF unit root test as well as the Zivot-Andrewsunit root test with structural breaks to check the orders of integrationof the data series. Table 2 reports the results of both unit root tests. Atthe 5 per cent signicance level, the results of the ADF test indicatethat all the variables are non-stationary at levels, but they are sta-tionary afterrst differenced. Likewise, the Zivot-Andrews unit roottestnds no additional evidence against the results of the ADF unitroottest. Therefore, bothunitroottestsconsistentlysuggestthatlnGNPt, lnTOURt, lnPSt, lnGNStand ln(ngd)tare integrated of orderone orI(1). This is in line with thendings of Nelson andPlosser(1982) that most of the macroeconomic series are I(1).Since both unit root tests consistently suggest that all variablesareI(1), wecanproceedtotest for thepresenceof along-runequilibrium relationship between Malaysia's economic growth andits determinants using the system-wide JohanseneJuselius cointe-grationapproach. Inordertoperformthiscointegrationtest, weshouldcorrectlydeterminethelaglengthandthedeterministiccomponents(i.e. constant andtrend) intheVARmodel. Thisisbecause the statistical inferences of the JohanseneJuselius cointe-gration test are very sensitive to these two issues (e.g. Hall, 1991;Johansen, 1995). Toaddresstheseissues, weemploythesystem-wide Akaike's Information Criterion (AIC) to select an optimal laglengthandthePantulaprincipletodeterminethedeterministiccomponentsintheVARmodel. Thecalculatedsystem-wideAICstatistics suggest using a VAR model with 3 lags, while the Pantulaprinciple indicates that Model 3 (i.e. intercept in the cointegratingequationandalineartrendinthedata)isthemostappropriatemodel for our dataset. Apart from these, it is also important to pointout here that the standard JohanseneJuselius statistics for cointe-gration are biased toward rejecting the null hypothesis of no coin-tegration (Cheung &Lai, 1993). On the basis of Monte Carlosimulation, Podivinsky (1998) added that the size distortion prob-lemwith the JohanseneJuselius cointegration test is obvious,especially when the sample size is less than 100 observations. Giventhe sample size of this study is small (i.e. T 37), we circumvent thesize distortionproblemby applying the formula suggestedbyReinselandAhn(1992)toadjustthestandardJohanseneJuseliusstatistics. Theresultsofthesystem-wideJohanseneJuseliuscoin-tegration test are presented in Panel A of Table 3. Importantly, wend that both the adjusted and the unadjusted LR(ltrace) statisticsconsistently reject the null hypothesis of no cointegration at the 1per cent signicance level. Therefore, we are able to conclude thatlnGNPt, lnTOURt, lnPSt, lnGNStandln(ngd)tarecointegrated.Thus, thereisameaningful long-runrelationshipbetweeneco-nomic growth and its determinants in Malaysia.Having found that the variables are cointegrated, we estimatethelong-runrelationshipbetweenMalaysia's economicgrowthand its determinants using the Ordinary Least Squares (OLS) esti-matorbecauseithassuperiorperformance insmallsamples.2Inaddition, Stock (1987) revealed that when the variables are found toTable 2The results of unit root tests.Variables ADF Zivot-Andrews unit root test with one breakTest statisticsModel A Model CBreakyearTeststatisticsBreakyearTeststatisticsLevel:lnGNPt3.060 (3) 1998 2.933 (1) 1998 2.749 (1)lnTOURt3.035 (1) 1990 4.112 (1) 1994 4.407 (1)lnPSt1.152 (0) 2005 2.001 (0) 2005 3.881 (0)lnGNSt3.068 (0) 1998 3.070 (0) 1998 3.362 (0)ln(n gd)t 1.866 (3) 1988 2.424 (3) 1994 4.037 (3)First difference: Critical valuesaDlnGNPt4.301 (0)***Model A: Model C:DlnTOURt6.123 (3)***1 per cent 5.340 1 per cent 5.570DlnPSt5.793 (0)***5 per cent 4.800 5 per cent 5.080DlnGNSt5.271 (1)***10 per cent 4.580 10 per cent 4.820Dln(ngd)t 3.367 (2)**Notes: *** and** denoterejectionat the1and5per cent signicancelevels,respectively. (.) indicates the optimal lag length for ADF and Zivot-Andrews tests asdetermined by Akaike's Information Criterion (AIC).aCritical values for the Zivot-Andrews unit root test with break are obtained fromZivot and Andrews (1992).Table 3The results of the cointegration test.Panel A: JohanseneJuselius cointegration testHypotheses Likelihood ratio (LR) testsH0HAUnadjusted-LR(ltrace) Adjusted-LR(ltrace)r 0 r1 164.327*** 97.708***r1 r2 98.788*** 58.739***r2 r3 59.803*** 35.559***r3 r4 24.303*** 14.451r4 r5 1.365 0.812Panel B: OLS estimator e Long run equationVariables Coefcients Std. errors t-statisticlnTOURt0.1403 0.0179 7.8145***lnPSt0.0718 0.0259 2.7652***lnGNSt0.4244 0.0302 14.0166***ln(ngd)t0.4778 0.1344 3.5527***Constant 4.0002 0.3688 10.8478***Panel C: Vector Error-Correction Model (VECM)Dependent variable: D ln GNPtVariables Coefcients Std. errors t-statisticsDlnTOURt0.0903 0.0238 3.7997***DlnPSt0.0324 0.0349 0.9310DlnGNSt0.4316 0.0883 4.8867***Dln(ngd)t0.9214 0.1704 5.4074***t10.5806 0.1116 5.2008***DUM98 0.0763 0.0134 5.6899***Constant 0.0020 0.0047 0.4330Panel D: Diagnostic testsc2NORMAL1.664 (0.4351)c2SERIAL[2]: 1.067 (0.5867)c2ARCH[1]: 0.971 (0.6155)c2RESET[2]: 2.234 (0.3272)CUSUM Stable at the 5 per cent signicance levelCUSUMSQ Stable at the 5 per cent signicance levelNotes: Theasterisks***denotesstatisticalsignicanceat the1percent level. [.]representstheorderofthediagnostictest, while(.)representsthep-value. Thecritical values obtained from MacKinnon, Haug, and Michelis (1999) are used to testfor cointegration.2For the case of small samples, the results of Monte Carlo simulation byAbeysingheandTan(1999) revealedthat theOLSestimator is relativelymorerobust thantheother velong-runestimators suggestedbyJohansen(1988),Bardsen(1989), PhillipsandHansen(1990), EngleandYoo(1991)andStockandWatson (1993).C.F. Tang, E.C. Tan / Tourism Management 46 (2015) 158e163 161be cointegrated, OLS estimation for long-run coefcient(s) is super-consistent andhighlyefcient. Therefore, theOLSestimator isappropriate for our study with the limited sample size. Panel B ofTable 3 shows the estimated long-run coefcients, whereas Panel Cof Table 3 shows the estimated short-run coefcients of the vectorerror-correctionmodel (VECM).3Fromtheresults, we ndthattourism, political stability, savings and ln(ngd)t are statisticallysignicant at the 1 per cent level in the long-run. In the short run,only political stability is insignicant while the rest of the variablesare statistically signicant at the 1 per cent level. Specically, theresultsindicatethattourismhasapositiveimpactoneconomicgrowth in both the short-run and the long-run. A 10 per cent in-creaseinpercapitareal tourismreceipts, onaverage, wouldin-crease per capita real GNP by 0.9 per cent and 1.4 per cent in theshort-run and in the long-run respectively. In summary, these re-sultsindicatethat tourismcouldstimulateMalaysia'seconomicgrowth, thus supporting the tourism-led growth hypothesis.Next, we further attempt to conrm the validity of the tourism-led growth hypothesis in Malaysia using the Granger causality test.Since the system-wide JohanseneJuselius cointegration resultsshows the presence of cointegration, the Granger causality test isconducted in the VECM framework to capture both short-run andlong-runcausalities. Table4reports theresults of theGrangercausality test. For the short-run, wend that the Likelihood Ratio(LR) test statistic rejects the null hypothesis that tourism does notGranger-cause economic growth and also the null hypothesis thateconomicgrowthdoesnotGranger-causetourism. Theseresultsthus indicate that tourismand Malaysia's economic growthGranger-cause each other (i.e. bi-directional causality) in the short-run. Turning to the long-run causality, wend that the one periodlagged error-correction term(t1) is statistically signicant at the 1per cent level when economic growth is the dependent variable.However, when tourismis the dependent variable, we nd that t1becomes insignicant at even 10 per cent level. Hence, based uponthe signicance of the t1 term, we are able to conclude that thereisauni-directional causalityrunningfromtourismtoeconomicgrowth in Malaysia in the long run. In light of all thesendings,the tourism-led growth hypothesis is well supported by theMalaysian experience. Tourismcould be an effective catalyst for thegrowth of the Malaysian economy. This is contrary to the ndings ofNanthakumar et al. (2008) and Kadir et al. (2010), but consistentwith those of Lau et al. (2009), Lean and Tang (2010), Othman et al.(2012) and Tang (2013) though it must be borne in mind that theseearlier studies are methodologically weak.5. Concluding remarksUsingamultivariatemodel derivedfromtheSolowgrowththeory, we attempt to analyse the impact of tourism on economicgrowth of Malaysia over the period of 1975e2011. The econometrictechniques involved in this study include tests for unit root, coin-tegration and Granger causality.As wefoundthat all thevariables areI(1), weappliedtheJohanseneJuselius cointegration test to determine the presence ofcointegration. Theresults of thetest suggestedthat Malaysia'seconomic growth, tourism, political stability, savings and popula-tion growth are cointegrated. Therefore, we estimatedthe short-run and long-run relationships between economic growth and itsdeterminantsinMalaysia. Remarkably, our ndingssuggestthattourism is positively related to Malaysia's economic growth in theshort-run and in the long-run. Similarly, the results of the GrangercausalitytestintheVECMframeworkalsoshowedthattourismGranger-causes economic growth both in the short-run and long-run. In conclusion, our empiricalndings support the tourism-ledgrowthhypothesis. TourismcanbereliedupontodriveMalay-sia's economic growth and development and hence, policy-makersshould pay greater attention towards promoting inbound tourism.This would also contribute to the structural diversication of theMalaysian economy and thus its resilience.SuchndingsonMalaysiahavegeneral applicabilitytoothercountriesaswell particularlythoseseekingfor newsourcesofgrowthanddevelopment. Like Malaysia, theycouldcount ontourismforeconomicgrowthif theyareabletoattractgenuinetourist arrivals rather than would-be migrant workers who pose astourists. Theymust strive toextendverywarmhospitality totourists, have years designated as Visit-the-Country Years, establisha very high-level tourismdevelopment and promotion governmentagencyandpromotemultilingualismamongst theservicepro-viders. Apart from these, they could advertise themselves as highlyinterestingtouristdestinationswithinternational televisionsta-tions. They could also try to cater to tourists from different regionsor countries by specically creating tourist spots that suit the variedtaste of differentnationalities.In a way,this couldmake touristsfeel at home while holidaying abroad.AcknowledgementThe earlier draft of this paper was presented at the InternationalConference on Business, Accounting, Finance and Economics (BAFE,2013): Paving the Way for Value Creation toward Sustainability, atUniversiti Tunku Abdul Rahman, Perak, Malaysia on October 4th,2013.We would liketo thank theanonymousreviewersand theeditorof TourismManagement, ChrisRyanfortheirconstructivecommentsandsuggestionsgivenontheearlierdraft. Any errorsthat remain in the paper are of our own responsibility.ReferencesAbeysinghe, T., &Tan, K. B. (1999). Small sampleestimationof acointegratingvector: an empirical evaluation of six estimation techniques. Applied Economics,6(10), 645e648.Aisen, A., & Veiga, F. J. (2006). Does political instability lead to higher ination? apanel data analysis. Journal of Money, Credit, and Banking, 38(5), 1379e1389.Aisen, A., &Veiga, F. J. (2013). Howdoes political instabilityaffect economicgrowth? European Journal of Political Economy, 29, 151e167.Al-mulali, U., Fereidouni, H. G., Lee, J. M. Y., & Mohammed, A. H. (2014). Estimatingthe tourism-led growth hypothesis: a case study of the Middle East countries.Anatolia: AnInternational Journal of TourismandHospitalityResearch, 25(2),290e298.Augustin, S., & Lee, K. K. (2012). Two-thirds of foreign workers in Malaysia are illegal.NewStraitsTimeshttp://news.asiaone.com/News/AsiaOne%2BNews/Malaysia/Story/A1Story20120412-339195.html Accessed 13.02.13.Table 4The results of the Granger causality test.Null hypotheses Short-run causality Long-run causalityLR statistics t-statisticsDlnTOURt KDlnGNPt 29.353*** 5.208***DlnGNPt KDlnTOURt 20.501*** 1.674Notes: *** denotes statistically signicant at the 1 per cent level. Kdenotesdoesnot Granger-cause.3Numerous diagnostic tests have been applied to verify the reliability of statis-tical inferences derived from the ECM. In line with the assumptions of the classicallinear regression model (CLRM), theJarque-Bera normality test conrms that theresiduals are normally distributed and the Breusch-Godfrey LM test indicates thatthemodelisfree fromserial correlation problem. Inaddition, theautoregressiveconditional heteroskedasticity (ARCH) LM test suggests that the model is free fromheteroskedasticity, whereas the Ramsey RESET test conrms no general specica-tion errors or functionality problem. Finally, the CUSUM and CUSUMSQ test revealthattheestimatedparametersarealsostableoverthesampleperiod. Therefore,our estimation model is correctly specied and reliable.C.F. Tang, E.C. Tan / Tourism Management 46 (2015) 158e163 162Balaguer, J., & Cantavella-Jord a, M. (2002). Tourism as a long-run economic growthfactor: the Spanish case. Applied Economics, 34(7), 877e884.Bardsen, G. (1989). Estimation of long run coefcients in error correction models.Oxford Bulletin of Economics & Statistics, 51(3), 345e350.Bouzahzah, M., & El Menyari, Y. (2013). International tourism and economic growth:the case of Morocco and Tunisia. Journal of North African Studies, 18(4), 592e607.Brida, J. G., Carrera, E. J. S., &Risso, W. A. (2008). Tourism'simpactonlong-runMexican economic growth. Economics Bulletin, 3(21), 1e8.Brida, J. G., Cortes-Jimenez, I., &Pulina, M. (2014). Hasthetourism-ledgrowthhypothesis been validated? A literature review. Current Issues in Tourism. http://dx.doi.org/10.1080/13683500.2013.868414.Castro-Nu~ no, M., Molina-Toucedo, J. A., & Pablo-Romero, M. P. (2013). Tourism and GDP:a meta-analysis of panel data studies. Journal of Travel Research, 52(6), 745e758.Cheam, C. L., Mahmood, R., Abdullah, H., &Ong, S. C. (2013). Economicgrowth,tourism and selected macroeconomic variables: a triangular causal relationshipin Malaysia. Margin: The Journal of Applied Economic Research, 7(2), 185e206.Cheung, Y. W., & Lai, K. S. (1993). Finite-sample sizes of Johansen's likelihood ratiotests for cointegration. Oxford Bulletin of Economics and Statistics, 55(3), 313e328.Dritsakis, N. (2004). Tourismas along-runeconomic growthfactor: anempiricalinvestigation for Greece using causality analysis. Tourism Economics, 10(3), 305e316.Engle, R. F., & Yoo, B. S. (1991). Cointegrated economic time series: an overview withnewresult. InR. F. Engle, &C. W. J. Granger(Eds.), Longruneconomicrela-tionship: Readings in cointegration. New York: Oxford University Press.Gonzalo, J. (1994). Fivealternativemethodsof estimatinglong-runequilibriumrelationships. Journal of Econometrics, 60(1e2), 203e233.Granger, C. W. J. (1988). Some recent development in a concept of causality. Journalof Econometrics, 39(1e2), 199e211.Hall, S. G. (1991). The effect of varying length VAR models on the maximum like-lihood estimates of cointegrating vectors. ScottishJournalofPoliticalEconomy,38(4), 317e323.Hye, Q. M. A., & Khan, R. E. A. (2013). Tourism-led growth hypothesis: a case study ofPakistan. Asia Pacic Journal of Tourism Research, 18(4), 303e313.Jalil, A., Mahmood, T., & Idrees, M. (2013). Tourism-growth nexus in Pakistan: evi-dence from ARDL bounds tests. Economic Modelling, 35, 185e191.Johansen, S. (1988). Statistical analysis of cointegration vectors. Journal of EconomicDynamics and Control, 12(2e3), 231e254.Johansen, S. (1995). Likelihood-based inference in cointegrated vector auto-regressivemodels. Oxford: Oxford University Press.Johansen, S., & Juselius, K. (1990). Maximum likelihood estimation and inference oncointegration withe applications to the demand for money. Oxford Bulletin ofEconomics and Statistics, 52(2), 169e210.Kadir, N., &Karim, M. Z. A. (2012). TourismandeconomicgrowthinMalaysia:evidence from tourist arrivals from ASEAN-5 countries. Ekonomska Istrazivanja-Economic Research, 25(4), 1089e1100.Kadir, N., Nayan, S., &Abdullah, M. S. (2010). Thecausal relationshipbetweentourismandeconomicgrowthinMalaysia: evidencefrommultivariatecau-sality tests. Tourism & Management Studies, 6, 16e24.Kassim, A. (1997). Illegal alien labour in Malaysia: its inux, utilization, and rami-cations. Indonesia and the Malay World, 25(71), 50e81.Katircio glu, S. T. (2009). Revisitingthetourism-led-growthhypothesisforTurkeyusing the bounds test and Johansen approach for cointegration. Tourism Man-agement, 30(1), 17e20.Katircio glu, S. T. (2011). TourismandgrowthinSingapore: newextensionfrombounds test tolevel relationships andconditional Granger causality tests.Singapore Economic Review, 56(3), 441e453.Lau, P. H., Oh, S. L., &Hu, S. S. (2009). TouristarrivalsandeconomicgrowthinSarawak. Sarawak Development Journal, 9, 36e45.Lean, H. H., &Tang, C. F. (2010). Isthetourism-ledgrowthhypothesisstableforMalaysia? A note. International Journal of Tourism Research, 12(4), 375e378.Lee, C. G. (2012). Tourism, trade, and income: evidence from Singapore. Anatolia: AnInternational Journal of Tourism and Hospitality Research, 23(3), 348e358.Ltkepohl, H. (1982). Non-causality due to omitted variables. Journal of Economet-rics, 19(2e3), 367e378.MacKinnon, J. G., Haug, A. A., & Michelis, L. (1999). Numerical distribution functionsof likelihood ratio tests for cointegration. Journal of Applied Econometrics, 14(5),563e577.Mankiw, N. G., Romer, D., &Weil, D. N. (1992). A contributionto theempiricsofeconomic growth. Quarterly Journal of Economics, 107(2), 407e437.Marshall, M. G., Gurr, T. R., &Jaggers, K. (2013). PolityIVProject: Political regimecharacteristics and transitions, 1800e2012. Center for Global Policy, GeorgeMason University.Nanthakumar, L., Ibrahim, Y., &Harun, M. (2008). Tourismdevelopment policy, strategicalliances andimpactof consumerpriceindexontourist arrivals: thecaseof Malaysia.TOURISMOS: An International Multidisciplinary Journal of Tourism, 3(1), 83e98.Narayan, P. K., & Prasad, B. C. (2003). Does tourism granger causes economic growthin Fiji? Empirical Economics Letters, 2(5), 199e208.Nelson, C. R., & Plosser, C. R. (1982). Trends and random walks in macroeconomictimeseries:someevidenceandimplications. Journal of MonetaryEconomics,10(2), 139e162.Oh, C. O. (2005). The contribution of tourism development to economic growth inthe Korea economy. Tourism Management, 26(1), 39e44.Othman, R., &Salleh, N. H. M. (2010). Hubunganpembangunanindustri pelan-congandanpertumbuhanekonomi di beberapanegarautamaASEAN. Inter-national Journal of Management Studies, 17(1), 171e188.Othman, R., Salleh, N. H. M., &Sarmidi, T. (2012). Analysisofcausalrelationshipbetweentourismdevelopment, economicgrowthandforeigndirect invest-ment: an ARDL approach. Journal of Applied Sciences, 12(12), 1245e1254.Pablo-Romero, M. P., & Molina, J. A. (2013). Tourism and economic growth: a reviewof empirical literature. Tourism Management Perspectives, 8, 28e41.Papatheodorou, A. (1999). ThedemandforinternationaltourismintheMediter-ranean region. Applied Economics, 31(5), 619e630.Payne, J. E., &Mervar, A. (2010). ThetourismgrowthnexusinCroatia. TourismEconomics, 16(4), 1089e1094.Perron, P. (1989). The great crash, the oil price shock, and the unit root hypothesis.Econometrica, 57(6), 1361e1401.Phillips, P. C. B., & Hansen, B. E. (1990). Statistical inference in instrumental variableregression with I(1) processes. Review of Economic Studies, 57(1), 99e125.Podivinsky, J. M. (1998). Testing misspecied cointegration relationships. EconomicsLetters, 60(1), 1e9.Reinsel, G. C., & Ahn, S. K. (1992). Vector autoregressive models with unit roots andreduced rank structure: estimation, likelihood ratio test and forecasting. Journalof Time Series Analysis, 13(4), 353e375.Solow, R. (1956). A contribution to the theory of economic growth. Quarterly Journalof Economics, 70(1), 65e94.Stock, J. H. (1987). Asymptotic properties of least square estimators of cointegratingvectors. Econometrica, 55(5), 277e302.Stock, J. H., & Watson, M. W. (1993). A simple estimator of cointegrating vectors inhigher order integrated systems. Econometrica, 61(4), 783e820.Tang, C. F. (2011a). Is the tourism-led growth hypothesis valid for Malaysia? A viewfrom disaggregated tourism markets. International Journal of Tourism Research,13(1), 97e101.Tang, C. F. (2011b). MultivariateGrangercausalityandthedynamicrelationshipbetweenhealthcarespending, income, andrelativepriceof healthcareinMalaysia. Hitotsubashi Journal of Economics, 52(2), 199e214.Tang, C. F. (2013). TemporalGrangercausality andthedynamicsrelationshipbe-tween real tourism receipts,real income, real exchange ratesin Malaysia. In-ternational Journal of Tourism Research, 15(3), 272e284.Tang, C. F., &Abosedra, S. (2014a). Small sampleevidenceonthetourism-ledgrowth hypothesis in Lebanon. Current Issues in Tourism, 17(3), 234e246.Tang, C. F., & Abosedra, S. (2014b). The impacts of tourism, energy consumption andpolitical instability on economic growth in the MENA countries. Energy Policy,68, 458e464.Tang, C. F., & Tan, E. C. (2013). How stable is the tourism-led growth hypothesis inMalaysia?A viewfromdisaggregatedtourismmarkets. TourismManagement,37, 52e57.UNWTO. (2012). Tourism highlights: 2012 edition. Madrid: World Tourism Organization.Zivot, E., & Andrews, D. W. K. (1992). Further evidence of the great crash, the oil-priceshockandtheunit-root hypothesis. Journal of Business andEconomicStatistics, 10(3), 251e270.Zureiqat, H. M. (2005). Political instability and economic performance: a panel dataanalysis. Award Winning Economics Papers, 1, 1e33.Chor Foon Tang is a Senior Lecturer at the Centre for PolicyResearch and International Studies, Universiti SainsMalaysia. He earned his Ph.D. in Economics from UniversityofMalaya, Malaysia. Hisresearchinterestsprimarilyfocuson DevelopmentEconomics, AppliedMacroeconomics, andTourismEconomics. He was awarded the Bright SparksScholarship in 2010 and won the University of Malaya Excel-lence Award in 2012. He is the reviewer and editorial boardmember for a wide range of international and local scholarlyjournals. Moreover, he is also one of the top economists inMalaysia as well as in Asia ranked by the Research Papers inEconomics (RePEc).EuChyeTanis aProfessor at theDepartment of Eco-nomics, Facultyof Economics andAdministration, Uni-versityof Malaya, KualaLumpur, Malaysia. Heobtainedhis Ph.D. in Economics from University of Warwick, UnitedKingdom. His research interest encompasses DevelopmentEconomics, Monetary Economics and InternationalEconomics.C.F. Tang, E.C. Tan / Tourism Management 46 (2015) 158e163 163