趙磊+方成
[摘 要]旅游業(yè)會通過溢出效應對經(jīng)濟增長產(chǎn)生非線性影響,但尚缺乏相關經(jīng)驗證據(jù)。文章基于1999-2013年省級面板數(shù)據(jù),以旅游業(yè)發(fā)展水平作為轉換變量,采用面板平滑轉換回歸模型(PSTR),對旅游業(yè)與經(jīng)濟增長之間的非線性關系進行了實證檢驗。結果表明:旅游業(yè)對經(jīng)濟增長具有正向促進效應,旅游導向型經(jīng)濟增長假說在中國真實有效;旅游業(yè)與經(jīng)濟增長之間的關系存在非線性的旅游業(yè)門檻效應,旅游業(yè)發(fā)展水平與旅游業(yè)經(jīng)濟影響效應顯著負相關。隨著旅游業(yè)發(fā)展水平的提高,旅游業(yè)經(jīng)濟影響效應處于高機制,當旅游業(yè)發(fā)展跨越門檻值之后,旅游業(yè)經(jīng)濟影響效應處于低機制,旅游業(yè)發(fā)展會弱化其對經(jīng)濟增長正向影響的邊際效應。因此,可通過優(yōu)化旅游產(chǎn)業(yè)結構和強化經(jīng)濟增長其他決定因素對旅游業(yè)的外部性效應,來保持旅游業(yè)對經(jīng)濟增長的穩(wěn)定持續(xù)貢獻。
[關鍵詞]旅游業(yè);經(jīng)濟增長;非線性;面板平滑轉換回歸
[中圖分類號]F59
[文獻標識碼]A
[文章編號]1002-5006(2017)04-0020-13
Doi: 10.3969/j.issn.1002-5006.2017.04.008
引言
旅游業(yè)發(fā)展與經(jīng)濟增長之間的關系歷來是業(yè)界和學界關注的熱門話題。2014年,旅游業(yè)對全球經(jīng)濟的綜合貢獻達7.58萬億美元,約占全球GDP的9.8%,創(chuàng)造就業(yè)機會2.77億個,占全球就業(yè)人數(shù)的9.4%[1]。同期,中國旅游業(yè)對GDP的綜合貢獻則達6.61萬億元,占GDP的10.39%。旅游業(yè)直接和間接就業(yè)人數(shù)為7873萬,占全國就業(yè)總人數(shù)的10.19%。對比來看,中國旅游發(fā)展對經(jīng)濟增長的貢獻程度略高于全球平均水平。旅游業(yè)因其可以為目的地創(chuàng)造外匯、帶動就業(yè)、增加稅收和平衡收支,逐漸成為一國或地區(qū)促進經(jīng)濟增長的戰(zhàn)略工具[2-4]。進而,全球范圍內,圍繞旅游發(fā)展與經(jīng)濟增長關系的應然性研究層出不窮[5],并為特定的國家或地區(qū)旨在發(fā)展經(jīng)濟過程中,在制定與旅游業(yè)發(fā)展相關政策時提供了理論認知和判斷依據(jù)。
1998年中央經(jīng)濟工作會議提出將旅游業(yè)作為國民經(jīng)濟新的增長點,成為中國旅游業(yè)發(fā)展模式由“計劃事業(yè)型”向“市場產(chǎn)業(yè)型”轉變的標志。自此,中國旅游業(yè)發(fā)展進入快速發(fā)展時期。通過觀察中國1999—2014年旅游業(yè)專業(yè)化(旅游總收入GDP占比)變化趨勢可以發(fā)現(xiàn),盡管對旅游發(fā)展變化趨勢的線性擬合呈現(xiàn)單調增高,但卻無法掩蓋旅游發(fā)展對經(jīng)濟增長綜合貢獻階段變化的非一致性。毫無疑問,中國旅游業(yè)絕對規(guī)模正經(jīng)歷高速擴張期,通過發(fā)展旅游業(yè)帶動經(jīng)濟增長,是否能成為經(jīng)濟轉型期產(chǎn)業(yè)結構優(yōu)化的有益選擇還需深入研究。圖1所隱含的一個重要信息是,由于時變環(huán)境的存在,旅游業(yè)發(fā)展同時具有波動性,從而導致其對經(jīng)濟增長的產(chǎn)業(yè)貢獻也并非持續(xù)穩(wěn)定。這種實踐現(xiàn)象,實際上在研究旅游發(fā)展與經(jīng)濟增長之間關系時需要謹慎對待[6-7]。顯然,這直接關系到旅游發(fā)展對經(jīng)濟增長影響效應的時變非線性。
目前,對旅游業(yè)發(fā)展與經(jīng)濟增長關系的研究主要涉及3個方向:其一,測算旅游經(jīng)濟貢獻[8];其二,應用增長模型框架[9];其三,檢驗兩者因果關系[10]。然而,如果將時變環(huán)境因素納入旅游業(yè)影響經(jīng)濟增長的研究框架,細察來看,既有文獻無論是在方法論,還是在分析框架上,均存在與實踐現(xiàn)象相悖的學理弊端。首先,旅游經(jīng)濟影響的評估模型主要反映的是旅游業(yè)對經(jīng)濟增長的靜態(tài)貢獻,并且忽視了旅游業(yè)發(fā)展與經(jīng)濟增長關系的動態(tài)特征,受到Lean和Tang以及Tang和Tan的質疑[11-12]。其次,盡管現(xiàn)有研究遵循將旅游業(yè)納入經(jīng)典增長模型基礎上的實證分析思路,但具體研究范式仍拘泥于線性框架,因而無法反映出旅游業(yè)與經(jīng)濟增長之間關系的非線性特征,尤其很難識別出旅游業(yè)影響經(jīng)濟增長的門檻效應。
盡管,主流觀點認為旅游業(yè)發(fā)展對經(jīng)濟增長具有積極貢獻[13-14],但也有與此相左的論點[15-16]。該種爭論始于20世紀90年代,焦點在于理論解析和實證檢驗兩方面。假設理論差異在于研究視角的不同,那么實證檢驗結論的迥異則會受到截面異質性的干擾[17]。因此,為了提高旅游業(yè)發(fā)展與經(jīng)濟增長之間非線性關系的估計效率,本文引入由González等所發(fā)展的面板平滑轉換回歸(panel smooth transition regression, PSTR)模型對旅游業(yè)經(jīng)濟影響非線性效應進行實證檢驗[18]。PSTR模型是以捕捉面板數(shù)據(jù)的截面異質性為主要研究目的的非線性回歸模型,可以有效刻畫面板數(shù)據(jù)的截面異質性,因而更符合社會經(jīng)濟的現(xiàn)實情境。
本文對旅游經(jīng)濟研究文獻的補充和推進主要體現(xiàn)在如下諸端:第一,在研究視角上,已有對旅游業(yè)發(fā)展與經(jīng)濟增長的經(jīng)驗研究主要停留在線性模型基礎上,然而,無論是對兩者關系的實踐發(fā)現(xiàn),抑或理論判斷,旅游業(yè)發(fā)展與經(jīng)濟增長之間的非線性關系更加貼近經(jīng)濟現(xiàn)實。鑒于此,從旅游業(yè)經(jīng)濟影響非線性效應這一視角切入,檢驗與評估旅游業(yè)發(fā)展影響經(jīng)濟增長的關系與效應,對如何制定旅游產(chǎn)業(yè)政策以提升經(jīng)濟貢獻具有重要的現(xiàn)實意義。第二,在研究方法上,PSTR模型進一步放松了非線性面板門檻回歸(panel threshold regression, PTR)模型的嚴格約束條件,在有效刻畫截面異質性特征的同時,允許估計參數(shù)隨轉換變量進行平滑變化,相比傳統(tǒng)的面板固定效應或隨機效應模型估計更具效率。第三,在研究內容上,豐富了旅游業(yè)影響經(jīng)濟增長方面的研究文獻,尤其是拓展了旅游導向型經(jīng)濟增長(tourism-led growth, TLG)假說的研究體系,基于中國省際面板數(shù)據(jù),藉以探索旅游業(yè)發(fā)展影響經(jīng)濟增長的內在機理,從而為“TLG在中國是否有效”的學術論證提供一種經(jīng)驗解釋。
1 文獻評述與理論探索
歷史上,經(jīng)濟繁榮主要依賴于農(nóng)業(yè)和制造業(yè)部門增長,然而,旅游業(yè)在經(jīng)濟活動中常被低估,并且被認為是非增長導向部門(non-growth oriented sector),很少受到經(jīng)濟學者和政策制定者的青睞[19]。然而,當前旅游業(yè)已成為全球快速增長的服務業(yè) 部門之一,其發(fā)展速度已經(jīng)超過了全球整體經(jīng)濟增速[20]。旅游業(yè)發(fā)展可通過溢出效應和外部性對經(jīng)濟活動產(chǎn)生積極影響,進而促進地區(qū)經(jīng)濟增長[21]。
關于旅游業(yè)影響經(jīng)濟增長的研究文獻,早期主要集中探討旅游業(yè)收入的經(jīng)濟貢獻,最初思想主要來源于McKinnon的“旅游創(chuàng)匯說”[22]。隨后,Gray通過測算發(fā)現(xiàn),美國對世界其他地區(qū)的人均旅游需求收入彈性為5.13,加拿大則為6.6,從而初步證實了國際旅游收入對經(jīng)濟增長的貢獻潛力[23]。再者,旅游需求只有轉換為旅游支出,并借助消費的乘數(shù)效應,才能綜合提升旅游業(yè)發(fā)展對經(jīng)濟增長的拉動作用[24]。于是,如何測度旅游消費支出變動所產(chǎn)生的經(jīng)濟效應成為旅游經(jīng)濟學研究的一個重要分支,代表性研究方法包括投入產(chǎn)出分析[25]、一般均衡模型[26]和旅游衛(wèi)星賬戶[27]。
為了尋求旅游業(yè)經(jīng)濟貢獻的理論支撐,探析旅游業(yè)發(fā)展影響經(jīng)濟增長的溢出途徑便成為應然性的研究關照。第一,旅游業(yè)對經(jīng)濟增長的直接貢獻表現(xiàn)在提供外匯收入、創(chuàng)造就業(yè)崗位和增加稅收收入[28-30]。第二,地區(qū)間旅游業(yè)投資競爭效應提升了旅游相關企業(yè)生產(chǎn)率,進而規(guī)模經(jīng)濟擴大,生產(chǎn)成本降低,有益于經(jīng)濟增長[31]。第三,旅游業(yè)通過對其關聯(lián)產(chǎn)業(yè)的間接誘導效應帶動經(jīng)濟增長[32]。第四,旅游業(yè)也是技術知識擴散、研發(fā)投入和人力資本積累的重要因素[33-34]。除此以外,在Feder的經(jīng)濟模型中,出口導向型增長(export-led growth)假說為旅游業(yè)促進經(jīng)濟增長開辟了另一種認識視角[35]。既有出口導向型增長文獻主要關注可貿(mào)易品與經(jīng)濟發(fā)展之間的關系,并未考慮到非貿(mào)易品。但是,隨著非貿(mào)易品與經(jīng)濟增長之間關系的理論模型逐漸成為近期研究熱點[36],基于這一新的認識視角,旅游業(yè)作為一種非貿(mào)易品的出口部門[37],某種程度上就引申出了旅游業(yè)是否會相應地促進經(jīng)濟增長的問題。沿著國際旅游業(yè)與貿(mào)易之間關系的研究脈絡,Gray和Keintz最早對兩者關系進行了探索[38-39],包括最近的大部分研究[40-41],均支持旅游業(yè)和貿(mào)易之間存在協(xié)整關系。
旅游業(yè)與經(jīng)濟增長之間的關系貌似硬幣的正反面。旅游業(yè)作為經(jīng)濟增長的工具同樣會受到質疑[42-43]。Sánchez-Rivero等指出,一國旅游業(yè)不會自動引發(fā)經(jīng)濟增長,除非有鼓勵這一過程的條件[44]。例如,需要強化旅游業(yè)部門人力資本投資[24]、增加公共安全支出[45]和實施環(huán)境保護政策[46]等。
對旅游業(yè)正面影響經(jīng)濟增長的經(jīng)典批判當屬Copeland的“去工業(yè)化”學說[47]和Chao等的“荷蘭病”效應[48],兩種觀點的理論進路相似。Copeland認為旅游業(yè)擴張增加了非貿(mào)易品消費,貿(mào)易條件得到改善,但資源配置從可貿(mào)易部門(資本密集型)到非貿(mào)易部門(勞動密集型)的資本縮減過程,導致實際匯率升值,進而削弱了可貿(mào)易部門的外部競爭力,最終出現(xiàn)去工業(yè)化(de-industrialization)現(xiàn)象[47]。資本和勞動力要素從傳統(tǒng)貿(mào)易部門流向非貿(mào)易部門,實際匯率升值,就會產(chǎn)生一種經(jīng)濟“病”,即“荷蘭病”,旅游業(yè)對經(jīng)濟增長的短期積極效應會引起長期經(jīng)濟體萎靡[49]。Chao等認為旅游業(yè)擴張分別通過資源效應和消費效應引發(fā)“荷蘭病”,一方面,旅游業(yè)作為繁榮部門(booming sector)需要來自其他部門的資源要素保持生產(chǎn);另一方面,旅游業(yè)擴張改善了貿(mào)易條件,外匯收入的增加刺激了對非貿(mào)易品的消費需求,抬升了非貿(mào)易品相對價格,進而又引起非貿(mào)易部門對資本和勞動需求的擴大,使得傳統(tǒng)貿(mào)易部門凋敝[48]。最終,實際匯率升值和國內商品價格上升,競爭力受到削弱,經(jīng)濟縮水。Capó等就研究發(fā)現(xiàn)在西班牙著名的旅游島嶼,即巴里阿里群島和加納利群島(the Balearic and Canary Islands)存在“荷蘭病”效應[50]。
倘若旅游業(yè)擴展會致使地區(qū)經(jīng)濟發(fā)生“荷蘭病”,則會對社會福利產(chǎn)生負面影響。Chao 等揭示出旅游業(yè)在短期和中期可能會增加居民整體福利,這是由于旅游業(yè)抬高了非貿(mào)易品價格[48],但從長遠來看,福利卻在下降,因為這是一個長期資本消耗過程。對此,Holzner以1970—2007年世界134個國家為研究樣本,對旅游依賴型國家(tourism-dependent countries)是否存在“荷蘭病”效應進行了實證檢驗[51]。結果發(fā)現(xiàn),上述國家并不存在發(fā)生荷蘭病效應的危險。相反,旅游依賴型國家不僅未出現(xiàn)實際匯率升值和去工業(yè)化情況,卻經(jīng)歷了高于平均樣本國家的經(jīng)濟增速。
旅游業(yè)發(fā)展影響經(jīng)濟增長的理論爭端必然會掀起相應的實證檢驗。肇自Ghali對夏威夷旅游業(yè)產(chǎn)出彈性的估計[52],從實證角度對旅游業(yè)與經(jīng)濟增長之間關系的切實研究要始于Lanza和Pigliaru的探索[9],尤其是以Balaguer和Cantavella-Jordà所提出的旅游導向型增長(tourism-led growth, TLG)假說為標志[10],諸多實證文獻開始關涉此話題,并分別利用時間序列或面板數(shù)據(jù)計量經(jīng)濟模型對單一國家或多個國家TLG假說真實性進行實證檢驗。旅游業(yè)與經(jīng)濟增長之間主要存在4種實證關系[53]:支持旅游導向型增長假說[54-62];支持經(jīng)濟驅動型旅游業(yè)增長(economic-driven tourism growth, EDTG)假說[63-65];旅游業(yè)與經(jīng)濟增長之間存在雙向因果關系[66-70];旅游業(yè)與經(jīng)濟增長之間不存在因果關系[71-73]。
綜上可知,一個基本的事實是,主流觀點支持旅游導向型增長假說,并得到了Pablo-Romero和Molina的述評佐證,其在對關于旅游業(yè)與經(jīng)濟增長關系的87篇國外文獻進行綜述后發(fā)現(xiàn),支持TLG假說的文獻比率為63%,僅有4篇文獻并未證實兩者之間存在關系[5]。進一步深入到TLG假說文獻內部,可以發(fā)現(xiàn),Lanza等的實證貢獻推進了TLG假說研究內容的深入[74]。Lanza等是第一篇采用面板數(shù)據(jù)模型對TLG假說進行研究的實證文獻,在對1977—1992年13個世界經(jīng)濟合作與發(fā)展組織(OECD)國家旅游業(yè)和經(jīng)濟增長關系進行實證檢驗后發(fā)現(xiàn),旅游專業(yè)化程度在長期并沒有不利于經(jīng)濟增長,主要是因為以旅游業(yè)為基礎的經(jīng)濟體(tourism-based economy)相對較低的生產(chǎn)率增速可以通過逐步提高旅游專業(yè)化得以彌補[74]。自此,后續(xù)文獻開始轉向旅游專業(yè)化如何影響旅游業(yè)和經(jīng)濟增長之間的關系方面,其中,以Gunduz和Hatemi-J為代表,指出旅游業(yè)占一國經(jīng)濟總量的比重是旅游業(yè)影響經(jīng)濟增長的重要決定因素,旅游專業(yè)化程度越高,旅游業(yè)對經(jīng)濟增長的影響力越大[56],這一觀點隨后得到Kaplan和?elik、Sequeira和Nunes、Adamou和Chloride以及Brida等的研究支持[75-78]。
既然旅游專業(yè)化會影響旅游業(yè)和經(jīng)濟增長之間的關系,這就意味著,隨著旅游專業(yè)化程度的變化,旅游業(yè)和經(jīng)濟增長之間的關系則會呈現(xiàn)出非線性特征。Brau等首先實證捕捉到此種門檻效應,其通過將143個國家1980—2003年平均人口小于100萬且旅游平均專業(yè)化水平高于10%的國家定義為“小國”,運用虛擬變量回歸發(fā)現(xiàn),人口規(guī)模小的國家只有在旅游專業(yè)化程度很高的情況下才會支持TLG假說[79]。Sequeira和Nunes則通過動態(tài)面板估計發(fā)現(xiàn),當同時將經(jīng)濟體規(guī)模和旅游專業(yè)化作為分組變量時,一國經(jīng)濟是否增長并非是由經(jīng)濟體規(guī)模決定,而是受到其旅游專業(yè)化的影響[76]。Po和Huang進一步運用較為先進的面板門檻回歸(panel threshold regression, PTR)方法,將入境旅游專業(yè)化作為門檻變量,通過對88個國家1995—2005年面板數(shù)據(jù)研究發(fā)現(xiàn),入境旅游專業(yè)化存在兩個門檻值,只有入境旅游專業(yè)化低于4.05%或高于4.73%時,入境旅游才對經(jīng)濟增長具有顯著正向關系[80]。Chang等的研究邏輯與Po和Huang相同,同樣支持入境旅游專業(yè)化對經(jīng)濟增長影響的門檻效應[81]。緊接著,Pan等拓展了Po和Huang以及Chang等的研究方法,首次引入面板平滑轉換回歸模型對15個OECD國家1995—2010年TLG假說進行重新檢驗,并以匯率收益率和通貨膨脹率為轉換變量,實證發(fā)現(xiàn)入境旅游與經(jīng)濟增長之間存在非線性關系[82]。
正如Pablo-Romero和Molina所言,主流文獻目前主要停留在旅游專業(yè)化是旅游業(yè)影響經(jīng)濟增長的決定因素這一研究共識,毫不諱言,眾多學者無論是理論探索,還是實證檢驗,都為此做出了重要貢獻,但對這種影響機制是如何發(fā)生,影響效應到底如何變化卻知之甚少[5]。在拓展對旅游業(yè)和經(jīng)濟增長因果關系的認識視域中,隨著復雜精妙的計量統(tǒng)計技術的應用,對非線性行為認知的逐漸加深有益于動態(tài)刻畫TLG假說的實踐內涵[83]。盡管已有文獻證實TLG假說在中國真實有效[84],但卻鮮有文獻對中國旅游業(yè)與經(jīng)濟增長之間的非線性關系進行有說服力的實證研究。
本文是對Po和Huang以及Chang等研究內容的推進和深化,主要體現(xiàn)在平滑式面板非線性計量方法的應用方面。盡管Hansen面板門檻回歸模型摒棄了對門檻變量進行主觀分組的傳統(tǒng)非線性檢驗手段,轉向對門檻變量異質性信息進行內生分組,來考察不同門檻區(qū)間內估計參數(shù)的跳躍轉換[85],但這種在門檻值前后發(fā)生的估計參數(shù)突變的假定,顯然并不符合宏觀經(jīng)濟變量之間因果關系漸進連續(xù)的客觀事實。鑒于此,與Pan等研究緊密相關,本文選擇使用目前較為前沿的非線性計量經(jīng)濟模型,即PSTR模型來實證檢驗旅游業(yè)影響經(jīng)濟增長的非線性關系。一方面,PSTR模型允許截面異質性,同時還可以捕捉既有文獻所一直忽視的旅游業(yè)對經(jīng)濟增長影響關系的時變性;另一方面,PSTR模型進一步放松了PTR模型的限制條件,但又與Pan等研究不同,本文重點識別旅游業(yè)處于不同發(fā)展階段時,旅游業(yè)與經(jīng)濟增長之間的關系變化形態(tài),所以引入一個連續(xù)的以旅游業(yè)(國內旅游和入境旅游)發(fā)展水平作為轉換變量的一般轉換函數(shù)來替代PTR模型中特殊的離散轉換函數(shù),從而允許模型中旅游業(yè)經(jīng)濟影響效應隨轉換變量的變化而連續(xù)地平滑轉換,這一點顯然與不斷變化的宏觀經(jīng)濟現(xiàn)實更為契合。
2 方法、模型與變量
2.1 研究方法
由González等提出[18],經(jīng)由Fouquau等完善的面板平滑轉換回歸(PSTR)模型是經(jīng)典的檢驗變量之間非線性關系的前沿計量技術[86],通過放松Hansen所開發(fā)的PTR模型的約束條件擴展而來,與傳統(tǒng)的面板數(shù)據(jù)固定和隨機效應模型相比,PSTR模型不僅可以有效刻畫模型參數(shù)的截面異質性,可以有效克服內生性所導致的參數(shù)估計量偏誤問題,尤其是允許模型參數(shù)隨轉換變量做緩慢平滑的非線性轉換。
式中,c是一個m維的轉換發(fā)生的位置參數(shù)向量,[γ]是平滑參數(shù),決定轉換函數(shù)的轉換速度,[γ>0]。
可見,在PSTR模型中,變量估計系數(shù)由線性部分[β0]和非線性部分[β1?g(?)]共同構成。顯然,模型存在兩種機制,當[g(?)=0]時,模型存在低機制(low regime);當[g(?)=1]時,模型存在高機制(high regime)。同時,隨著轉換函數(shù)值[0,1]之間平滑移動時,模型估計系數(shù)會以c為中心在[β0~β0+β1]之間單調轉換。
2.2 模型設定
基于經(jīng)典的旅游業(yè)經(jīng)濟增長模型[88],并根據(jù)前述理論分析,為了深入揭示在不同旅游專業(yè)化階段,中國TLG假說有效的復雜機制,本文通過構建旅游業(yè)影響經(jīng)濟增長的PSTR模型來對兩者之間的非線性關系進行實證檢驗,計量模型設定如下:
經(jīng)濟增長水平采用人均實際GDP對數(shù)衡量。旅游業(yè)發(fā)展水平采用旅游專業(yè)化衡量,即省份旅游總收入占GDP比值[17]。相關控制變量:短期內投資水平提高有助于經(jīng)濟增長[89],采用投資規(guī)模占GDP比值來衡量投資份額([invest]);經(jīng)濟增長來自人力資本的積累[90],人力資本([lnhuman])是經(jīng)濟增長的重要決定因素,采用人均勞動受教育年限衡量;政府支出可以反映“看得見的手”對經(jīng)濟增長的干預程度[91],采用政府支出占GDP比值衡量政府規(guī)模(govern);出口貿(mào)易可以通過促進技術進步推動經(jīng)濟增長[90],采用進出口貿(mào)易總額占GDP比值衡量貿(mào)易開放(open);產(chǎn)業(yè)結構變遷與經(jīng)濟增長密切相關[92],采用第三產(chǎn)業(yè)就業(yè)人員比重衡量產(chǎn)業(yè)結構(indstu)。
2.3 數(shù)據(jù)來源
考慮到數(shù)據(jù)可得性與一致性,本文使用1999—2013年中國大陸30個省、市、自治區(qū)(西藏剔除)省級面板數(shù)據(jù)。旅游產(chǎn)業(yè)數(shù)據(jù)來源于《中國旅游年鑒(2000—2014)》,其他原始數(shù)據(jù)分別來源于《中國統(tǒng)計年鑒(2000—2014)》、中經(jīng)網(wǎng)統(tǒng)計數(shù)據(jù)庫和CEIC中國經(jīng)濟數(shù)據(jù)庫。
3 實證結果分析
應用PSTR模型實證檢驗旅游業(yè)與經(jīng)濟增長之間的非線性關系,需要遵循3個步驟:(1)檢驗模型非線性;(2)確定平滑參數(shù)[γ]和位置參數(shù)c;(3)模型穩(wěn)健性檢驗。
3.1 模型非線性檢驗
在建立PSTR模型之前,首先對方程(3)進行非線性檢驗,以考察是否存在非線性機制轉換效應,即對原假設[H0:γ=0]進行檢驗,由于模型包含未識別參數(shù)[γ]和c,故而無法對模型進行傳統(tǒng)的非線性檢驗。為了檢驗截面異質性,González等建議遵循Luukkonen等的做法[93],考慮設置同質性零假設[H0:γ=0],并在[γ=0]處用轉換函數(shù)一階泰勒展開式替代,從而構造出輔助回歸方程:
式中,T為時間長度,N為截面?zhèn)€數(shù),k為外生變量個數(shù),[SSR0]和[SSR1]分別為接受和拒絕原假設的殘差平方和。經(jīng)檢驗,[LMF]統(tǒng)計量為6.35,并在1%水平上顯著拒絕[H*0],所以接受模型存在非線性的假設。
3.2 模型參數(shù)估計
通過檢驗發(fā)現(xiàn)異質性存在,則應考慮PSTR模型參數(shù)估計,要比線性模型能更好地克服參數(shù)異質性問題,從而得到穩(wěn)定可靠的估計結果。PSTR模型的參數(shù)估計主要采用非線性最小二乘法(nonlinear least squares, NLS)得到估計值[94]。其中,轉換函數(shù)的斜率系數(shù)[γ]和位置參數(shù)c可采用網(wǎng)格搜索法(grid search arithmetic)或模擬退火法(simulated annealing arithmetic)得到。鑒于網(wǎng)格搜索法受限于搜索精度,本文首先采用模擬退火法獲得平滑參數(shù)[γ]和位置參數(shù)c的初始值,然后采用NLS方法對方程(3)進行估計。
以旅游專業(yè)化為門檻變量的PSTR模型估計結果顯示,PSTR模型發(fā)生非線性轉換的位置參數(shù)c為0.073,表明以旅游專業(yè)化衡量的旅游業(yè)發(fā)展門檻值為0.073,模型存在兩個機制。其中,旅游專業(yè)化低于門檻值([TRi,t≤0.073])時,轉換函數(shù)[g(TRi,t;γ,c)]取值趨于0,并且共有166個觀測值,占全部觀測值比重為36.9%;旅游專業(yè)化高于門檻值([TRi,t>0.073])時,轉換函數(shù)[g(TRi,t;γ,c)]取值趨于1,并且共有284個觀測值,占全部觀測值比重為63.1%。模型在旅游業(yè)經(jīng)濟影響效應機制之間平滑的斜率系數(shù)[γ]為27.23,表明模型在低高機制之間轉換速度相對較快,并呈現(xiàn)平滑漸進變化趨勢(圖2)。簡言之,當旅游業(yè)處于不同發(fā)展階段時,旅游業(yè)與經(jīng)濟增長之間的關系出現(xiàn)了平滑轉換。
PSTR模型同時報告出,[TR]估計系數(shù)[β0]為1.585,在1%水平上顯著,而[TR×g(?)]估計系數(shù)[β1]為-0.531,在5%水平上顯著,表明旅游業(yè)經(jīng)濟影響效應具有動態(tài)性和非線性。當轉換函數(shù)[g(TRi,t;γ,c)=0]時,旅游業(yè)經(jīng)濟影響效應為1.585([β0]),模型處于高機制;當轉換函數(shù)[g(TRi,t;γ,c)=1]時,旅游業(yè)經(jīng)濟影響效應為1.054([β0+β1]),模型處于低機制,旅游業(yè)經(jīng)濟影響效應在低與高機制之間以旅游專業(yè)化門檻值0.073為中心,隨著自身狀態(tài)變量的變動,旅游業(yè)經(jīng)濟影響效應在[1.054,1.585]之間平滑轉換。結合圖3決定,旅游專業(yè)化與旅游業(yè)經(jīng)濟影響效應顯著負相關,即盡管旅游業(yè)對經(jīng)濟增長依然具有正向促進作用,但隨著旅游業(yè)專業(yè)化程度不斷增高,旅游業(yè)發(fā)展對經(jīng)濟增長影響的邊際效應遞減。具體而言,當旅游業(yè)發(fā)展水平較低時,旅游業(yè)經(jīng)濟影響效應處于高影響狀態(tài),當旅游業(yè)發(fā)展跨越門檻值0.073時,旅游業(yè)經(jīng)濟影響效應開始逐漸從高影響狀態(tài)向低影響狀態(tài)轉換,并最終持續(xù)處于低影響狀態(tài)。
這一研究結論與Adamou和Clerides對1980—2005年全球162個國家旅游業(yè)與經(jīng)濟增長之間非線性關系的實證發(fā)現(xiàn)基本一致,在旅游專業(yè)化初期階段,旅游業(yè)會較大幅度推動經(jīng)濟增長,但推動效果會逐漸減弱,即當旅游專業(yè)化達到一定程度之后,其會弱化旅游業(yè)對經(jīng)濟增長的貢獻率[77]。究其原因,第一,單純的旅游產(chǎn)業(yè)刺激政策很難在長期保持旅游業(yè)對經(jīng)濟增長的持續(xù)貢獻,這是因為旅游產(chǎn)品吸引力具有時間衰減規(guī)律,如果目的地旅游產(chǎn)品創(chuàng)新能力缺乏或者旅游產(chǎn)業(yè)結構調整滯后,都可能會引起旅游客源消費市場的“心理倦怠(psychological tiredness)”效應[5];第二,旅游業(yè)盲目快速擴張,最直接的后果是旅游投資扭曲引致資源配置效率降低,初級觀光旅游產(chǎn)品“產(chǎn)能過?!敝率孤糜尾块T生產(chǎn)要素邊際生產(chǎn)率下降,旅游業(yè)產(chǎn)出規(guī)模報酬遞減最終導致旅游業(yè)對經(jīng)濟增長的貢獻下降。第三,鑒于某些外部影響因素,諸如公共投資、人力資本、經(jīng)濟資本和產(chǎn)業(yè)結構可能會影響旅游業(yè)和經(jīng)濟增長之間的關系,如果旅游業(yè)在發(fā)展過程中脫離于上述相關宏觀經(jīng)濟變量的外部約束,同樣會減弱旅游業(yè)對經(jīng)濟增長的影響效應。
在控制變量對經(jīng)濟增長的影響方面,投資率、人力資本和產(chǎn)業(yè)結構均對經(jīng)濟增長存在顯著正向效應,這與既有理論預期相符。需要指出的是,貿(mào)易開放并未對經(jīng)濟增長產(chǎn)生顯著促進效應,這與陸銘和陳釗的經(jīng)驗結果相一致[95],表明貿(mào)易開放是否成為促進中國經(jīng)濟增長的持續(xù)動力有待商榷,可能的原因是由于國際金融危機、人民幣匯率升值和出口產(chǎn)品創(chuàng)新附加值低等原因所共同導致的出口貿(mào)易受挫有關。政府規(guī)模對經(jīng)濟增長存在顯著正向效應,這與王小魯?shù)鹊难芯拷Y論相反[96],但與張杰等在幾乎相同樣本期內(1999—2012)所得出的實證結論一致[97]。對比既有研究結論表明,在不同發(fā)展階段,維持中國經(jīng)濟增長的動力正在發(fā)生顯著變化,猶如經(jīng)濟開放度和政府干預。
在普通面板模型線性估計中,旅游專業(yè)化估計系數(shù)為1.107,由于遺漏了旅游業(yè)與經(jīng)濟增長之間的非線性關系,導致旅游業(yè)經(jīng)濟影響效應傾向于低估旅游業(yè)對經(jīng)濟增長的線性影響效應,同時高估旅游業(yè)對經(jīng)濟增長的整體影響效應,處于旅游業(yè)經(jīng)濟影響效應低機制與高機制之間。因而,相比于普通面板模型線性估計,PSTR模型更好地刻畫了旅游業(yè)對經(jīng)濟增長的動態(tài)影響。
PTR模型主要采用的是網(wǎng)格搜索法,通過迭代,直到殘差平方和最小時的最優(yōu)估計所對應的門檻值則為初始值,q為0.067。表1中PTR模型估計結果顯示,旅游業(yè)經(jīng)濟增長影響效應存在基于旅游專業(yè)化的正向非單調性“門檻效應”,即旅游業(yè)經(jīng)濟增長影響效應存在非線性。當旅游專業(yè)化低于門檻值0.067時,旅游業(yè)經(jīng)濟影響效應為2.022;當旅游專業(yè)化高于門檻值0.067時,旅游業(yè)經(jīng)濟影響效應為1.227。由此發(fā)現(xiàn),當旅游業(yè)處于不同發(fā)展水平時,旅游業(yè)對經(jīng)濟增長的影響效應不同,表現(xiàn)出顯著門檻特征。此外,通過構建上述門檻值與虛擬變量的乘積項,當[TRi,t≤q]([TRi,t>q])時,虛擬變量定義為[D1]([D2]),并對乘積項進行普通面板模型估計,發(fā)現(xiàn)乘積項的估計系數(shù)和顯著性均與PTR模型估計結果基本一致。因此,無論是PTR模型估計,還是虛擬變量乘積項估計,均表明PSTR模型對旅游業(yè)經(jīng)濟影響效應的非線性估計結果具有穩(wěn)健性。
3.3 模型穩(wěn)健性分析
本文采用旅游業(yè)發(fā)展水平另一代理變量,旅游人次比作為度量指標[36],進行PSTR模型穩(wěn)健性檢驗,估計結果列于表2。首先,以旅游人次比作為門檻變量的PSTR模型估計結果顯示,位置參數(shù)c為7.305,表明當旅游人次比低于門檻值([TPi,t≤7.305]),且[g(TPi,t;γ,c)=0]時,旅游業(yè)經(jīng)濟影響效應為0.026,模型處于高機制;當旅游人次比高于門檻值([TPi,t>7.305]),且[g(TPi,t;γ,c)=1]時,旅游業(yè)經(jīng)濟影響效應為0.008,模型處于低機制,旅游業(yè)經(jīng)濟影響效應在低與高機制之間以人次比門檻值7.305為中心,隨著自身狀態(tài)變量的變動,在[0.008,0.026]之間平滑轉換。平滑參數(shù)[γ]為1.287,結合圖4,表明模型在位置參數(shù)前后機制轉換速度較慢,旅游業(yè)經(jīng)濟影響效應在低與高機制之間轉換速率為1.287。當采用旅游人次比度量旅游業(yè)發(fā)展水平時,隨著旅游業(yè)處于不同發(fā)展階段,尤其是當旅游人次比跨越門檻值之后,由圖5所示,旅游業(yè)經(jīng)濟影響效應開始由高影響狀態(tài)向低影響狀態(tài)轉換,旅游業(yè)對經(jīng)濟增長影響的邊際效應開始降低。綜上分析,旅游業(yè)影響經(jīng)濟增長的非線性PSTR模型估計結果具有穩(wěn)健性,在旅游業(yè)不同發(fā)展階段,旅游業(yè)經(jīng)濟影響效應并非線性,而是在低與高機制之間平滑轉換。
表2中PTR模型結果顯示,同樣采用最優(yōu)網(wǎng)格搜索法確定門檻值,當旅游人次比低于門檻值7.637時,旅游人次比估計系數(shù)為0.027,且在1%水平上統(tǒng)計顯著;當旅游人次比高于門檻值7.637時,旅游人次比估計系數(shù)為0.004,但統(tǒng)計不顯著。鑒于此,循上邏輯,首先定義基于PTR模型門檻值的虛擬變量,再通過對旅游人次比與虛擬變量的乘積項進行固定效應模型估計,結果發(fā)現(xiàn)當旅游人次比低于門檻值7.637時,旅游人次比估計系數(shù)為0.026,當旅游人次比高于門檻值7.637時,旅游人次比估計系數(shù)為0.006,并且分別在1%和10%水平上統(tǒng)計顯著,進而佐證了以旅游人次比為門檻變量,PSTR模型對旅游業(yè)和經(jīng)濟增長之間非線性關系進行檢驗的穩(wěn)健性。在普通面板模型線性估計中,旅游人次比估計系數(shù)為0.008,并且統(tǒng)計顯著,同樣由于未考慮旅游業(yè)與經(jīng)濟增長之間的非線性關系,導致旅游業(yè)經(jīng)濟影響效應極大低估了旅游業(yè)對經(jīng)濟增長的線性影響效應,同時與旅游業(yè)對經(jīng)濟增長的整體影響效應相近,相比來看,PSTR模型估計更能客觀地反映出旅游業(yè)對經(jīng)濟增長影響的非線性變化。此外,控制變量估計系數(shù)符號和顯著性也基本穩(wěn)健。
4 結論與啟示
追尋著Pablo-Romero和Molina對旅游業(yè)和經(jīng)濟增長之間關系的述評方向[5],隨著對旅游業(yè)影響經(jīng)濟增長理論認識的深化,以及實證計量技術的進步,如何往前推進TLG假說的研究視域,尤其是深入、客觀和精確地揭橥旅游業(yè)對經(jīng)濟增長影響效應的非線性變化成為當前國際旅游經(jīng)濟學需要破解和厘清的核心問題。盡管Pan等首次將非線性經(jīng)典PSTR模型應用到TLG假說實證檢驗[82],但其與本文研究內容相比,仍存在兩點局限:一是,研究對象主要是世界經(jīng)濟合作與發(fā)展組織(OECD)國家,卻忽視了世界上最大的發(fā)展中國家;二是,主要考察的是政策變量(匯率收益率和通貨膨脹率)作為轉換變量時,入境旅游與經(jīng)濟增長之間的非線性關系,但并未識別旅游業(yè)處于不同發(fā)展階段時,由于其狀態(tài)變量的變化所導致的旅游業(yè)與經(jīng)濟增長之間的非線性關系。
本文基于中國1999—2013年省級面板數(shù)據(jù),采用非線性面板平滑轉換回歸模型,對旅游業(yè)與經(jīng)濟增長之間的非線性關系進行實證檢驗,有效克服了普通面板數(shù)據(jù)模型因遺漏了非線性因素而導致的無法捕捉旅游業(yè)經(jīng)濟影響效應的動態(tài)機制轉換,從而使得估計結論更符合經(jīng)濟現(xiàn)實情境。旅游業(yè)發(fā)展水平對旅游業(yè)經(jīng)濟影響效應具有正向非線性特征影響,即在旅游業(yè)不同發(fā)展階段,旅游業(yè)對經(jīng)濟增長均具有顯著正向促進效應,但旅游業(yè)發(fā)展水平則與旅游業(yè)經(jīng)濟影響效應顯著負相關,隨著旅游業(yè)發(fā)展水平的提高,旅游業(yè)經(jīng)濟影響效應處于高機制,當旅游業(yè)發(fā)展跨越門檻值之后,旅游業(yè)經(jīng)濟影響效應處于低機制,旅游業(yè)規(guī)模的擴張反而會弱化旅游業(yè)對經(jīng)濟增長的邊際貢獻,同時以位置參數(shù)為中心,旅游業(yè)經(jīng)濟影響效應在高低機制之間平滑轉換,當以旅游專業(yè)化度量旅游業(yè)發(fā)展水平時,平滑轉換速率要高于以旅游人次比度量旅游業(yè)發(fā)展水平時。
本文研究結論為地區(qū)實施旅游業(yè)促進經(jīng)濟增長的經(jīng)濟政策提供了理論基礎。首先,TLG假說在中國客觀有效,表明旅游業(yè)發(fā)展對經(jīng)濟增長的綜合貢獻能力值得信任,鼓勵旅游業(yè)發(fā)展可以作為促進區(qū)域經(jīng)濟增長的有效工具。其次,在保持旅游業(yè)規(guī)模擴張的同時,還需注重內在發(fā)展質量,尤其是提升旅游業(yè)發(fā)展效率,核心要義是優(yōu)化旅游產(chǎn)業(yè)結構。質言之,現(xiàn)代旅游產(chǎn)品愈發(fā)具有技術知識密集型特點,突破大眾觀光旅游的粗放型發(fā)展模式窠臼,通過優(yōu)化資源要素配置和提升產(chǎn)品創(chuàng)新能力來共同引領旅游業(yè)內涵式集約化發(fā)展,才能穩(wěn)健地發(fā)揮旅游業(yè)對經(jīng)濟增長的持續(xù)貢獻。另外,鑒于當前中國旅游業(yè)發(fā)展仍主要依賴資源要素驅動,在旅游業(yè)發(fā)展未跨越門檻值階段,由資源比較優(yōu)勢所帶來的“要素紅利”和規(guī)模經(jīng)濟使得旅游業(yè)經(jīng)濟影響效應處于高機制,然而,隨著旅游資源消耗殆盡,旅游業(yè)影響經(jīng)濟增長的邊際效應則會隨著時間推移而逐漸減弱。因此,為了推動旅游業(yè)經(jīng)濟影響效應由低機制再向高機制轉換,一方面,可以通過鼓勵與旅游業(yè)相關的經(jīng)濟活動來培育旅游業(yè)與其關聯(lián)產(chǎn)業(yè)之間的產(chǎn)業(yè)融合,以拓寬旅游業(yè)對經(jīng)濟增長的影響渠道和傳導路徑;另一方面,在充分重視經(jīng)濟增長的決定因素時,還需調整和強化相關宏觀經(jīng)濟變量對旅游業(yè)發(fā)展的溢出效應,進而為旅游業(yè)影響經(jīng)濟增長提供有利外部性條件。
參考文獻(References)
[1] World Travel and Tourism Council. Travel and tourism economic impact [EB/OL]. www.wttc.org/research/economic-research, 2015-04-01. [世界旅行及旅游理事會. 旅行及旅游的經(jīng)濟影響[EB/OL]. www.wttc.org/research/economic-research, 2015-04-01.]
[2] Archer B. The impact of international tourism on the economy of Bermuda, 1994 [J]. Journal of Travel Research, 1995, 34(2): 27-30.
[3] Archer B, Fletcher J. The economic impact of tourism in the Seychelles [J]. Annals of Tourism Research, 1996, 23(1): 32-47.
[4] Sahli M, Nowak J J. Does inbound tourism benefit developing countries? A trade theoretic approach [J]. Journal of Travel Research, 2007, 45(4): 426-434.
[5] Pablo-Romero M P, Molina J A. Tourism and economic growth: A review of empirical literature [J]. Tourism Management Perspectives, 2013, 8(1): 28-41.
[6] Arslanturk Y, Balcilar M, Ozdemir Z A. Time-varying linkages between tourism receipts and economic growth in a small open economy [J]. Economic Modelling, 2011, 28(1): 664-671.
[7] Balcilar M, Eyden R V, Gupta I L R. Time-varying linkages between tourism receipts and economic growth in South Africa [J]. Applied Economics, 2014, 46(36): 4381-4398.
[8] Dwyer L, Forsyth P, Spurr R. Tourisms contribution to a state economy: A multi-regional general equilibrium analysis [J]. Tourism Economics, 2003, 9(4): 431-448.
[9] Lanza A, Pigliaru F. Tourism and Sustainable Economic Development [M]. New York: Springer, 2000: 57-69.
[10] Balaguer J, Cantavella-Jordá M. Tourism as a long-run economic growth factor: The Spanish case [J]. Applied Economics, 2002, 34(7): 877-884.
[11] Lean H H, Tang C F. Is the tourism-led growth hypothesis stable for Malaysia? A note [J]. International Journal of Tourism Research, 2010, 12(4): 375-378.
[12] Tang C F, Tan E C. How stable is the tourism-led growth hypothesis in Malaysia? Evidence from disaggregated tourism markets [J]. Tourism Management, 2013, 21(8): 1579-1591.
[13] Khan H, Phang S Y, Toh R S. The multiplier effect: Singapores hospitality industry [J]. Cornell Hospitality Quarterly, 1995, 36(1): 64-69.
[14] Lee C, Kwon K. Importance of secondary impact of foreign tourism receipts on the South Korean economy [J]. Journal of Travel Research, 1995, 34(2): 50-54.
[15] Forsyth P, Dwyer L, Clarke H. Problems in use of economic instruments to reduce adverse environmental impacts of tourism [J]. Tourism Economics, 1995: 265-282.
[16] Palmer T, Riera A. Tourism and environmental taxes. With special reference to the “Balearic ecotax” [J]. Tourism Management, 2003, 24(6): 665-674.
[17] Lee C C, Chang C P. Tourism development and economic growth: A closer look at panels [J]. Tourism Management, 2008, 29(1): 180-192.
[18] González A, Terasvirta T, Dijk D V. Panel Smooth Transition Regression Models [R]. Stockholm: Stockholm School of Economics, 2005: 1-15.
[19] Papatheodorou A. The demand for international tourism in the Mediterranean region [J]. Applied Economics, 1999, 31(5): 619-630.
[20] Pérez-Rodríguez J V, Ledesma-Rodríguez F, Santana-Gallego M. Testing dependence between GDP and tourisms growth rates [J]. Tourism Management, 2015, 48(3): 268-282.
[21] Marin D. Is the export-led hypothesis valid for industrialized countries? [J]. Review of Economics and Statistics, 1992, 74(4): 678-688.
[22] Mckinnon R. Foreign exchange constraints in economic development and efficient aid allocation [J]. Economic Journal, 1966, 76(301): 388-409.
[23] Gray H P. The demand for international travel by the United States and Canada [J]. International Economic Review, 1966, 7(1): 83-92.
[24] Sinclair M. Tourism and economic development: A survey [J]. Journal of Development Studies, 1998, 34(5): 1-51.
[25] Frechtling D C, Horváth E. Estimating the multiplier effects of tourism expenditures on a local economy through a regional input-output model [J]. Journal of Travel Research, 1999, 37(4): 324-332.
[26] Dwyer L, Forsyth P, Spurr R. Evaluating tourism's economic effects: New and old approaches [J]. Tourism Management, 2004, 25(3): 307-317.
[27] Ivanov S, Webster C. Measuring the impact of tourism on economic growth [J]. Tourism Economics, 2006, 13(3): 379-388.
[28] Belisle F J, Hoy D R. The perceived impact of tourism by residents a case study in Santa Marta, Colombia [J]. Annals of Tourism Research, 1980, 7(1): 83-101.
[29] Davis D, Allen J, Consenza R M. Segmenting local residents by their attitudes, interests, and opinions toward tourism [J]. Journal of Travel Research, 1988, 27(2): 2-8.
[30] Uysal M, Gitelson R. Assessment of economic impacts: Festivals and special events [J]. Festival Management and Event Tourism, 1994, 2(1): 3-10.
[31] Krueger A. Trade policy as an input to development [J]. American Economic Review, 1980, 70(2): 188-292.
[32] Slee B, Farr H, Snowdon P. The economic impact of alternative types of rural tourism [J]. Journal of Agricultural Economics, 1997, 48(3): 179-192.
[33] Braun P. Regional tourism networks: The nexus between ICT diffusion and change in Australia [J]. Information Technology and Tourism, 2003, 6(4): 231-243.
[34] Nowak J J, Sahli M, Cortés-Jiménez I. Tourism, capital good imports and economic growth: Theory and evidence for Spain [J]. Tourism Economics, 2007, 13(4): 515-536.
[35] Feder G. On exports and economic growth [J]. Journal of Development Economics, 1983, 12(1): 59-73.
[36] Kim H J, Chen M H, Jang S C. Tourism expansion and economic development: The case of Taiwan [J]. Tourism Management, 2006, 27(5): 925-933.
[37] Theobald W F. Global Tourism, (the 2nd Edition) [M]. Butterworth & Heinemann, Oxford. 2001: 25.
[38] Gray H P. International Arrivals: International Trade [M]. Lexington, MA: D. C. Heath. 1970.
[39] Keintz R M. A study of the demand for international travel to and from the United States [J]. The American Economist, 1971, 15(1): 137-138.
[40] Kadir N, Jusoff K. The cointegration and causality tests for tourism and trade in Malaysia [J]. International Journal of Economics and Finance, 2010, 2(1): 138-143.
[41] Lionetti S, Gonzalez O. On the relationship between tourism and growth in Latin America [J]. Tourism and Hospitality Research, 2012, 12(1): 15-24.
[42] Brohman J. New directions in tourism for third world development [J]. Annals of Tourism Research, 1996, 23(1): 48-70.
[43] Diagne A K. Tourism development and its impacts in the Senegalese petite c?te: A geographical case study in centre-periphery relations [J]. Tourism Geographies, 2004, 6(4): 472-492.
[44] Sánchez-Rivero M, Pulido-Fernández J I, Cárdenas-García P J. Quantitative Methods in Tourism Economics [M]. Berlin: Physica-Verlag HD, 2013: 235-251.
[45] Dunn H S, Dunn L L. Tourism and popular perceptions: Mapping Jamaican attitudes [J]. Social and Economic Studies, 2002, 51(1): 25-45.
[46] Gursoy D, Rutherford D G. Host attitudes toward tourism: An improved structural model [J]. Annals of Tourism Research, 2004, 31(3): 495-516.
[47] Copeland B R. Tourism, welfare and de-Industrialization in a small open economy [J]. Economica, 1991, 58(232): 515-29.
[48] Chao C C, Hazari B R, Laffargue J P. Tourism, Dutch diseasr and welfare in an open dynamic economy [J]. Japanese Economic Review, 2006, 57(4): 501-515.
[49] Corden W M, Neary J P. Booming sector and deindustrialization in a small open economy [J]. Economic Journal, 1982, 92(368): 825-848.
[50] Capó J, Font A R, Nadal J R. Dutch disease in tourism economies: Evidence from the Balearics and the Canary Islands [J]. Journal of Sustainable Tourism, 2007, 15(6): 615-627.
[51] Holzner M. Tourism and economic development: The beach disease? [J]. Tourism Management, 2011, 32(4): 923-933.
[52] Ghali M A. Tourism and economic growth: An empirical study [J]. Economic Development & Cultural Change, 1976, 24(3): 527-38.
[53] Chatziantoniou I, Filis G, Eeckels B. Oil prices, tourism income and economic growth: A structural VAR approach for European Mediterranean countries [J]. Tourism Management, 2013, 36(3): 331-341.
[54] Blake A, Sinclair M T. Tourism crisis management: US response to September 11 [J]. Annals of Tourism Research, 2003, 30(4): 813-832.
[55] Dritsakis N. Tourism as a long-run economic growth factor: An empirical investigation for Greece using causality analysis [J]. Tourism Economics, 2004, 10(3): 305-316.
[56] Gunduz L, Hatemi-J A. Is the tourism-led growth hypothesis valid for Turkey? [J]. Applied Economics Letters, 2005, 12(8): 499-504.
[57] Katircioglu S T. Revisiting the tourism-led-growth hypothesis for Turkey using the bounds test and Johansen approach for cointegration [J]. Tourism Management, 2009a, 30(1): 17-20.
[58] Akinboade O A, Braimoh L A. International tourism and economic development in South Africa: A granger causality test [J]. International Journal of Tourism Research, 2010, 12(2): 149-163.
[59] Fayissa B, Nsiah C, Tadesse B. Tourism and economic growth in Latin American countries-further empirical evidence [J]. Tourism Economics, 2011, 17(6): 1365-1373.
[60] Tang C F, Abosedra S. Small sample evidence on the tourism-led growth hypothesis in Lebanon [J]. Current Issues in Tourism, 2012, 17(3): 234-246.
[61] Tang C F, Abosedra S. The impacts of tourism, energy consumption and political instability on economic growth in the MENA countries [J]. Energy Policy, 2014, 68(5): 458-464.
[62] Hye Q M A, Khan R E A. Tourism-led growth hypothesis: A case study of Pakistan [J]. Asia Pacific Journal of Tourism Research, 2013, 18(4): 303-313.
[63] Narayan P K. Fijis Tourism demand: The ARDL approach to Cointegration [J]. Tourism Economics, 2004, 10(2): 193-206.
[64] Oh C O. The contribution of tourism development to economic growth in the Korean economy [J]. Tourism Management, 2005, 26(1): 39-44.
[65] Payne J E, Mervar A. The tourism-growth nexus in Croatia [J]. Tourism Economics, 2010, 16(4): 1089-1094.
[66] Durbarry R. Tourism and economic growth: The case of Mauritius [J]. Tourism Economics, 2004, 10(4): 389-401.
[67] Chen C F, Chiou-Wei S C. Tourism expansion, tourism uncertainty and economic growth: New evidence from Taiwan and Korea [J]. Tourism Management, 2009, 30(6): 812-818.
[68] Seetanah B. Assessing the dynamic economic impact of tourism for island economies [J]. Annals of Tourism Research, 2011, 38(1): 291-308.
[69] Apergis N, Payne J E. Tourism and growth in the Caribbean-evidence from a panel error correction model [J]. Tourism Economics, 2012, 18(2): 449-456.
[70] Ridderstaat J, Croes R, Nijkamp P. Tourism development, quality of life and exogenous shocks: A systemic framework [J]. International Journal of Society Systems Science, 2013, 5(4): 321-336.
[71] Eugenio-Martin J L, Morales M N, Scarpa R. Tourism and Economic Growth in Latin American Countries: A Panel Data Approach [R]. Canary Islands: Universidad of Las Palmas de Gran Canaria, 2004: 1-20.
[72] Tang C H, Jang S C. The tourism-economy causality in the United States: A sub-industry level examination [J]. Tourism Management, 2009, 30(4): 553-558.
[73] Katircioglu S T. Testing the Tourism-led Growth Hypothesis: The case of Malta [J]. Acta Oeconomica, 2009b, 59(3): 331-343.
[74] Lanza A, Temple P, Urga G. The implications of tourism specialisation in the long run: an econometric analysis for 13 OECD economies [J]. Tourism Management, 2003, 24(3): 315-321.
[75] Kaplan M, C?elik T. The Impact of Tourism on Economic Performance: The case of Turkey [J]. International Journal of Applied Economics and Finance, 2010, 4(4): 254-259.
[76] Sequeira T N, Nunes P M. Does tourism influence economic growth? A dynamic panel data approach [J]. Applied Economics, 2008, 40(18): 2431-2441.
[77] Adamou A, Clerides S. Prospects and limits of tourism-led growth: The international evidence [J]. Review of Economic Analysis, 2010, 2(3): 287-303.
[78] Brida J G, Punzo L F, Risso W A. Tourism as a factor of growth: The case of Brazil [J]. Tourism Economics, 2011, 17(6): 1375-1386.
[79] Brau R, Lanza A, Pigliaru F. How fast are small tourism countries growing? Evidence from the data for 1980-2003 [J]. Tourism Economics, 2007, 13(4): 603-613.
[80] Po W C, Huang B N. Tourism development and economic growth: A nonlinear approach [J]. Physica A Statistical Mechanics and Its Applications, 2008, 387(22): 5535-5542.
[81] Chang C L, Khamkaew T, McAleer M. IV estimation of a panel threshold model of tourism specialization and economic development [J].Tourism Economics, 2012, 18(1): 5-41.
[82] Pan S C, Liu S Y, Wu P C. Re-testing the tourism-led growth hypothesis using panel smooth transition regression models [J]. Tourism Economics, 2014, 20(1): 39-50.
[83] Sugihara G, May R, Ye H. Detecting causality in complex ecosystems [J]. Science, 2012, 338(6106): 496-500.
[84] Zhao Lei. Tourism development and economic growth: Empirical evidence from China [J]. Tourism Tribune, 2015, 30(4): 33-49. [趙磊. 旅游發(fā)展與經(jīng)濟增長——來自中國的經(jīng)驗證據(jù)[J]. 旅游學刊, 2015, 30(4): 33-49.]
[85] Hansen B E. Threshold effects in non-dynamic panels: Estimation, testing, and inference [J]. Journal of Econometrics, 1997, 93(2): 345-368.
[86] Fouquau J, Hurlin C, Rabaud I. The Feldstein-Horioka puzzle: A panel smooth transition regression approach [J]. Economic Modelling, 2007, 25(2): 284-299.
[87] Granger C W. Ter?svirta T. Modelling Non-linear Economic Relationships [M]. Oxford: Oxford University Press, 1993: 15-26.
[88] Ekanayake E M, Long A E. Tourism development and economic growth in developing countries [J]. International Journal of Business & Finance Research, 2012, 6(1): 61-63.
[89] Mankiw N G, Romer D, Weil D N. A contribution to the empirics of economic growth [J]. Quarterly Journal of Economics, 1992, 107(2): 409-437.
[90] Lucas R E. On the mechanics of economic development [J]. Journal of Monetary Economics, 1988, 22(1): 3-42.
[91] Barro R J. Inequality and growth in a panel of countries [J]. Journal of Economic Growth, 2000, 5(1): 5-32.
[92] Chenery H B. Interactions between industrialization and exports [J]. American Economic Review, 1980, 70(2): 281-287.
[93] Luukkonen R, Saikkonen P, Ter?svirta T. Testing linearity against smooth transition autoregression [J]. Biometrika, 1988, 75(3): 491-499.
[94] Thanh S D. Threshold effects of inflation on growth in the ASEAN-5 countries: A panel smooth transition regression approach [J]. Journal of Economics, Finance and Administrative Science, 2015, 20(38): 41-48.
[95] Lu Ming, Chen Zhao. Fragmented growth: Why economic opening may worsen domestic market segmentation? [J]. Economic Research Journal, 2009, (3): 42-52. [陸銘, 陳釗. 分割市場的經(jīng)濟增長——為什么經(jīng)濟開放可能加劇地方保護?[J]. 經(jīng)濟研究, 2009, (3): 42-52.]
[96] Wang Xiaolu, Fan Gang, Liu Peng. Transformation of growth pattern and growth sustainability in China [J]. Economic Research Journal, 2009, (1): 44-47. [王小魯, 樊綱, 劉鵬. 中國經(jīng)濟增長方式轉換和增長可持續(xù)性[J]. 經(jīng)濟研究, 2009, (1): 44-47.]
[97] Zhang Jie, Gao Debu, Xia Yinlei. Do patents economic growth in China: An explanation based on government patent subsidy policy [J]. China Industry Economics, 2016, (1): 83-98. [張杰, 高德步, 夏胤磊. 專利能否促進中國經(jīng)濟增長——基于中國專利資助政策視角的一個解釋[J]. 中國工業(yè)經(jīng)濟, 2016, (1): 83-98.]
Abstract: Tourism has become a strategic tool by which to promote economic growth for destinations, due to its importance in creating foreign exchange, offering job opportunities, increasing tax revenue, and balancing payments. The tourism-led growth (TLG) hypothesis has been previously validated by scholars; however, the existing papers focus on the linear relation between tourism and economic growth, which clearly does not conform to the law of tourism economy. In fact, a nonlinear relationship exists between tourism and economic growth through the spillover effect, although there is still no empirical evidence for this. Therefore, this paper applies the panel smooth transition regression (PSTR) model to examine the nonlinear relationship between tourism and economic growth, using tourism as a transition variable for 30 provinces in China during the period of 1999 to 2013. We also check the robustness by using panel threshold regression (PTR). The empirical results indicate that tourism has a significant positive effect on economic growth, thus the tourism-led growth hypothesis is valid in China. In addition, the relationship between tourism and economic growth is nonlinear, and varies inversely with tourism industry specialization. With the development of tourism, the effect of tourism on economic growth is in the high regime; however, the relationship between tourism and economic growth shows a decreasing marginal tendency as the degree of specialization grows, when above the threshold level. This paper enriches the field of research regarding tourism economics, especially the TLG hypothesis. Based on Chinas provincial panel data, this paper explores the inherent mechanism of the effects of tourism on economic growth, then provides empirical evidence for determining whether or not TLG is valid in China.
The conclusions of this paper lay the theoretical foundation for the effects of regional tourism development on economic growth. First, the TLG hypothesis is proven to be valid in China, which signifies that tourism contributes to economic growth. Therefore, tourism is an effectively strategic tool by which to enhance economic growth. Second, while maintaining the expansion of the scale of the tourism industry, attention should be paid to the quality of tourism, especially efficiency. Meanwhile, given the fact that the tourism development modes in China depend heavily on resources, when the level of tourism development is lower than its threshold values, the comparative advantages of tourism resources and economies of scale lead to the effects of tourism on economic growth in the high regime. However, with the depletion of tourism resources, the marginal effects of tourism on economic growth will gradually weaken over time. Therefore, in order to transfer the effects of tourism on economic growth from the low regime to the high regime, several suggestions should be offered. On the one hand, it is necessary to encourage the industrial integration of industries related to the tourism industry, in order to broaden the channels and transmission paths of the effects of tourism development on economic growth; on the other hand, when the determinants of economic growth are fully paid attention to, it is necessary to adjust and strengthen the spillover effects of relative macro-economic variables on tourism development, thus providing the favorable external conditions for the tourism industry to promote economic growth.
Keywords: tourism; economic growth; nonlinearity; panel smooth transition regression
[責任編輯: 劉 魯;責任校對:魏云潔]