key: cord-0916148-sh5nmyxc authors: Lee, Chien-Chiang; Chien, Mei-Se title: Structural breaks, tourism development, and economic growth: Evidence from Taiwan date: 2008-04-04 journal: Math Comput Simul DOI: 10.1016/j.matcom.2007.03.004 sha: 5bcea04970de79f73abf85267d6e02f60df15a6c doc_id: 916148 cord_uid: sh5nmyxc Our paper differs from previous studies by examining the issue of whether regime changes have broken down the stability of the long-run relationships between tourism development and real GDP in Taiwan for the 1959–2003 period. We empirically investigate the co-movements and the causal relationships among real GDP, tourism development, and the real exchange rate in a multivariate model. We use two different tourism variables—international tourism receipts and number of international tourist arrivals. To employ the unit root tests and the cointegration tests allowing for a structural break, the empirical evidence clearly shows that the causality between tourism and economic growth is bi-directional. Lastly, the international and cross-strait political change, economic shocks, and the relaxing of some tourism control and policies would break down the stability of the relationships between tourism development and economic growth. Overall, we do find the structural breakpoints, and they look to match clearly with the corresponding critical economic, political, or tourist incidents. The Asia Pacific region has become a hastily growing tourism destination and has even exceeded the Americas to become the world's second largest tourist-receiving region since 2001. As part of the emerging market in tourism, Taiwan has an abundance and diversity of natural and cultural resources, offering great potential for the development of tourism. The main purpose of this paper is to investigate whether regime changes have broken down the stability of the long-run relationship between tourism development and real GDP in Taiwan. We empirically examine the co-movements and the causal relationships among real GDP, tourism development, and the real exchange rate in a multivariate model. To employ the unit root tests and the cointegration tests allowing for a structural break, the empirical evidence clearly shows that the causality between tourism and economic growth is bi-directional. Moreover, international and cross-strait political changes, economic shocks, and relaxing of some tourism control and policies would break down the stability of the relationship between tourism development and real GDP. Taiwan has always been a great travel destination for business travelers and tourists from neighboring countries. Over the period 1960-2000, total tourist arrivals to Taiwan increased from 23,636 to 2,624,037. International tourism revenue of Taiwan increased from US$ 1.477 million in 1960 to US$ 3.738 billion in 2000 (see Table 1 ). Further analysis of overseas visitors by country shows that came from countries within the neighboring Asian region, which provided nearly 77% of all visitors in 2004. The top five of Taiwan's tourist-source residences in 2004 were Japan, Hong Kong, and Macau region, United States, South Korea, and Singapore. To date there are very few empirical studies that have investigated inbound tourism development in Taiwan. For instance, Huang and Min [23] established a model for Taiwan's inbound demand to predict the volume of visitor arrivals after the 921 earthquake in 1999. Wang [64] provides empirical evidence using the grey theory and fuzzy time series to predict tourism demand. Furthermore, Min [42] employs a seasonal autoregressive integrated moving average (SARIMA) model for measuring the impact of the Severe Acute Respiratory Syndrome (SARS) on Taiwan's inbound demand over the period from March to July 2003. Kim et al. [28] examine the causal relationships between tourism development and economic growth in Taiwan. Thus far, these empirical studies for Taiwan have not undertaken structural breaks in the relationships between tourism development and economic growth. Over the past several decades, the island's economic and tourist developments have covered a somewhat volatile time of economic incidents, international political shocks, and the adjustment of tourism regimes in Taiwan. Will these changes cause structural breaks in the relationship between tourism development and economic growth in Taiwan? In the 1960s and 1970s, tourism in Taiwan grew very fast. Over the late 1970s to the early 1990s, Taiwan faced a turbulent international political relationship and world economic crises, which eroded the growth of Taiwan's tourism development. After the United States Government changed its diplomatic recognition from Taipei to Beijing on 1 January 1979, this policy prompted Taiwan to face a difficult relationship with mainland China and the international community. Only around 30 countries maintain formal diplomatic relations with Taiwan at the moment. Substantial relations between Taiwan and those countries without diplomatic relations were confronted with many challenges, including foreigners applying for visas and the rights of international airlines. This severe international political situation resulted in a prolonged slowdown of Taiwan's tourism growth. The grim situation has also been aggravated by two oil price shocks, occurred respectively in 1973 and in 1979, and the change of Taiwan's exchange rate regime in 1987. Being in the presence of the circumstances, Taiwan's government relaxed some tourism controls and policies to reinvigorate the tourism industry in the stage. First, in the 1980s the government established 6 national parks and 12 national scenic areas. Furthermore, there was the lifting of travel restrictions-for example, the government's institution of a 5-day visa-free entry program for 15 countries in 1994. Over the last decade the growth rate of tourist arrivals to Taiwan has risen again, which is attributed to many factors. As well as the above tourism polices, the rapid economic growth of the neighboring Asian counties has also fueled overseas arrivals to Taiwan. 1 The fast economic growth rate of Taiwan has also brought more business visitors, which exceeded tourist visitors to become the first group from 1998 to 2000. Another important reason is that the policy for mainlanders visiting Taiwan experienced a gradual door opening from 1988 as there was an initial reconciliation between Taiwan and mainland China. In recent years, Taiwan's tourism development is still facing many uncertainties and challenges, including economic disturbances of the Asian financial crisis in 1997, and natural damages of Taiwan's 921 earthquake in 1999, and human diseases of the SARS epidemic in 2003. Some advantages may vitalize Taiwan's tourism industry in the future. First, the opening up of Taiwan tourism to mainlanders was permitted starting from 2002, which will effectively boost tourism development. After many traditional industries moved into mainland China, tourism development has recently become a major policy in Taiwan for increasing employment and economic growth. For example, the Doubling Tourism Arrivals Policy is proposed by the government of Taiwan. This plan sets to double the number of tourists visiting Taiwan in 2008. The purpose and contribution of this paper are as follows. In the beginning, we tested Taiwanese data for the last 50 years, which may be effected greatly by economic shocks, political incidents, or tourism policy alterations. This may yield structural changes from the data of GDP or tourism development variables, which therefore influence the result of the stationary test. Secondly, we use Johansen's [25] multivariate cointegrated method to test the cointegration relation between GDP and tourism development. Based on the multivariate cointegration model, we probe into the causality between GDP and tourism development with the weak exogeneity test. Finally, it is possible that structural breaks have occurred which might affect the result of the cointegration test. Using two kinds of Hansen [21] and Gregory and Hansen [18] structural change tests, which to our knowledge has not been previously applied in this area, we examine the cointegration relationship altogether and commented whether there is instability or not between GDP and tourism development. The paper is organized as follows. Section 2 reviews recent literature in the area of economic effects of tourism development and describes the general framework. The empirical results are presented in Section 3. Concluding remarks close the article in Section 4. In the analysis of tourism, economists emphasize the economic effects of tourism on the economy. The speedy growth of tourism causes an increase of household incomes and government revenues through multiplier effects, improvements in the balance of payments, and growth in the number of tourism-promoted government policies. As such, the development of tourism has usually been considered a positive contribution to economic growth (e.g. [27, 34, 35, 51] ). Given the aforementioned reasons, a steady stream of empirical tourism economics literature has been done in recent years, especially focusing on the issue of tourism demand. Tourism demand has been extensively analyzed in empirical literature. 2 Lim [36] reviews a large number of empirical studies on modeling international tourism demand and integrates those results according to the important explanatory variables used (income, transportation costs, and tourism prices), the proportion of significant findings, and the effect sizes of these major explanatory variables. Some studies demonstrate significant results in a direction opposite to the overall conclusion. In a review of 70 articles applying meta-analysis, such outcomes occur in 2.8%, 6.2%, and 8.3% of all cases for the models used for the income, transportation cost, and tourism price variables, respectively. Applying time series analysis is the most frequent econometric approach in the tourism demand literature. According to the review of Narayan [45] , it is essentially divided into two categories. One is for pre-1995 and the other is post-1995. It was found that pre-1995 and some post-1995 research studies may have the so-called "spurious regression" problem for ignoring the unit root test and cointegration. Most post-1995 literature has applied modern econometric techniques including the unit root test, cointegration, or the error correction model. 3 Lim and McAleer [38] predict tourism demand by past behavior of tourist variables, which is a simple and straightforward method, rather than relating them to other variables in a model. 4 However, this univariate time series forecasts cannot detect the hypothesis of tourism-led economic growth. Only a few empirical researchers have investigated Notes: "Tourism ⇒ growth" denotes the causality running from tourism to economic growth. "Growth ⇔ tourism" denotes the causality running from economic growth to tourism development. tourism-led economic growth. Tourism-led growth occurs when tourism manifests a motivating influence across the entire economy in the form of spillovers and other externalities [41, 51] . Balaguer and Cantavella-Jorda [6] examine the role of tourism's long-run economic development in Spain. The hypothesis of tourism-led economic growth was confirmed by applying cointegration and causality tests. Durbarry [12] also supports the hypothesis of tourism-led economic growth by using the cointegration and causality tests to study the case of Mauritius. Some research studies did not arrive at the same conclusion. Table 2 presents previous empirical results between tourism and economic growth, but we still cannot see an explicit result. Nevertheless, different empirical evidence shows different policy implications, which cannot only contribute to distinguishing the innate characters of the tourism industry, but can also be used as the basis for how a government can resolve the symbiosis policies of tourism businesses and economic development [62] . In many countries much hope for economic development is laid on tourism due to its capacity in generating employment. The development of tourism as such mostly requires investment, principally in hospitality, transportation, basic health, and recreation in order to motivate the accommodation and the stay of visitors. Finally, governments have to be active in laying out legislation and ordaining the development of tourism, so that the environment might be preserved and the future of tourism activities might be insured over time [65] . Eugenio-Martín et al. [16] investigate the relationship between tourism and economic growth for Latin American countries from 1985 through 1998, being based on a panel data approach and the Arellano-Bond [3] estimator for dynamic panels. The empirical results show that tourism development can contribute to the economic growth of mediumor low-income countries, while such a role is unclear for developed countries. In South Korea, the tourism-led economic growth hypothesis did not hold according to the research of Oh [51] , which studies the causal relations between tourism growth and economic expansion. This research obtains the one-way causal relationship of economic-driven tourism growth. Kim et al. [28] examines the causal relationship between tourism development and economic growth in Taiwan. Although Taiwan and South Korea have similar experiences of economic development, their empirical results are different from Oh's [51] . They find a reciprocal relationship between tourism development and economic growth in Taiwan. The factor of owing to structural breaks is a common problem in macroeconomic series as they are usually affected by exogenous shocks or regime changes in economic events. Therefore, this relationship is likely to be subject to variation as a result of changes in the structure of the economy like changes in tourism development policy or economic development regimes, reforms in tourism regulation, or institutional developments. Using different unit root tests, Narayan [48, 49] investigates whether military coups in 1987 and 2000 have had a transitory or a permanent effect on tourist expenditures in Fiji. Narayan [48] uses the Vogelsang [63] test, but Narayan [49] uses two different unit root tests-the Zivot and Andrews [67] and the Lumsdaine and Papell [40] tests. However, these two studies yield the same conclusions, which coup of 1987 had only a transitory effect on tourist expenditures in Fiji. Unfortunately, most of these above studies in the literature did not test structural breaks for exogenous shocks or regime change in critical tourism or economic events. If the estimation periods cover the somewhat volatile time of tourism development and economic innovation, then it is important to check the cointegration relationship for structural breaks. Narayan [48, 49] has studied the structural break of tourist expenditure but has not examined the stability of the relationship between tourism development and economic growth. The instability of the economic system may in fact be reflected to the parameters of the estimated models that, when used for inference or forecasting, can induce misleading results. For these reasons, this paper examines the stability of the relationship between tourism development and GDP. According to the total of 70 studies under review by Lim [36] , 65 and 48 studies include the income and tourism price variables, respectively. Following Balaguer and Cantavella-Jorda [6] and Lee and Chang [33] , the model includes real GDP, a tourism development variable, and real exchange rate, which can be written as All the variables are expressed in natural logarithms so that elasticities can also be interpreted. Here, LY represents real GDP. LTOUR represents tourism development. LRQ is real exchange rate (a proxy variable of external competitivity, Dritsakis [10] ). Term u is the error term. All the data used are annual observations of the variables, and the estimation period is 1959-2003. Naturally, compared with monthly or quarterly data, 5 annual data can even react to seasonal adjustment phenomena. The nominal GDP series is transformed into real gross product in 2001 prices, using the consumer price index (CPI). The start of the sample period is 1959 which is determined by the availability of data for CPI. Terms LY and LRQ used in this study are taken from the AREMOS economic-statistic data banks, created jointly by the Ministry of Education in Taiwan. Furthermore, we use two different types of tourism development variables-LTOUR1 [6, 33] which is international tourism earnings in real terms and LTOUR2 [16, 33] which is the number of tourists. We collect all tourism variables from the Annual Report on Tourism by Taiwan. We apply the unit root tests of the augmented Dickey-Fuller (ADF, Dickey and Fuller [9] ), Phillips-Perron (PP; Phillips and Perron [53] ), DF-GLS [13] , and MZ GLS α [50] . The lag order of models is selected based on the modified Akaike information criteria (MAIC). In order to take into account this possible shift in regime in the unit root test, Zivot and Andrews ( [67] ; hereafter ZA) and Perron [52] develop a new category of tests that allows an endogenous structural break. Two tests are allowed for a unit root against the alternative of a trend stationarity process with a structural break. We adopt the Akaike information criteria (AIC) to select the model and the lag lengths for the ZA [67] and Perron [52] tests. Fig. 1 plots the actual values of these series, indicating that the entire series exhibit trends. Consequently, all unit root test regressions are run with a constant and trend terms. We find the results of unit root tests without structural change which indicate that different tests may yield contradictory results. 6 Except for the PP test in LTOUR2, all other statistics are not able to significantly reject the null of the unit root. Furthermore, the DF-GLS and MZ GLS α indicate 5 Hakkio and Rush [19] pointed out that when using monthly or quarterly data in an empirical analysis, increasing the number of observations does not add any robustness to the results. 6 We do not report the details of all of the unit root tests we use here in order to conserve space, but all results are available upon request. that all difference variables do not reject the null hypothesis of I(2), apart from LRQ that are integrated with order one. However, the two tests by the ZA and Perron approaches point out unanimously that all variables are I(1). Also noteworthy is that the ZA test results indicate that the breakpoint occurs in 1992 for real GDP, in 1965 and 1966 for tourism development variables, and in 1987 for the real exchange rate. We find critical economic and tourism incidents for Taiwan that can match with the structural breaks of these series. First, what caused the breakpoint of real GDP in 1992? Due to the collapse of Taiwan's bubble economy in 1990, the asset markets went bust, the stock price index sank from around 12,000 to under 3000 in half a year, and the real estate market became sluggish if not outright collapsed. Second, why did tourism development variables have a structural break in 1965 and 1966? With the end of the U.S. aid program in Taiwan at 1965, the Military Assistance Advisory Group (MAAG), with members of about 10,000 who had stayed in Taiwan, was withdrawn and this event struck the tourism development of Taiwan. Fortunately, Japan relaxed its ban on overseas travel in 1964, and the movement of Japanese tourists to Taiwan picked up the pace. Japan has displaced the U.S. to become the dominant country generating travel to Taiwan from 1967 until the present. There was a breakpoint of the real exchange rate in 1987, in which the government released foreign exchange controls and substituted the floating foreign exchange rate system for a managed foreign exchange rate system. Under this new system, the foreign exchange rate was determined by market forces [66] . We follow an alternative route consisting of three steps: (i) test of a cointegration relationship using the full systems model; (ii) reduction of the model from the systems formulation to single equations using weak exogeneity tests to examine long-run causality; (iii) test for cointegration between the variables allowing for structural change or parameter instability [55] . As a consequence of the presence of non-stationary series, Eq. (1) is estimated using the Johansen and Juselius [26] procedure for cointegration. Both λ max and trace tests suggest the presence of one cointegrating vector for each type of tourism development variables, indicating the existence of a long-run relationship between tourism, real income, and the real exchange rate in Taiwan. 7 The normalized cointegrating coefficients are shown as the following The above indicates that tourism and the real exchange rate positively affects Taiwan's economic growth over time. The signs of the variables conform to the theory in the literature. This shows that tourism development contributes to economic growth, as in Durbarry [12] . At the same time, a rising real effective exchange rate can be harmful to tourism development [8] . However, the long-run tourism-elasticities are estimated to be 3.50-4.67. The long-run real exchange rate-elasticities are estimated to be 0.43-0.81. According to these estimations, income has a positive relationship with tourism and a positive link with the real exchange rate. Furthermore, the likelihood ratio test on the statistical significance of the parameters rejects the null hypothesis in all cases for the tourism variables, but does not reject the null hypothesis on the real exchange rate. In other words, comparing the effects on Taiwan's real GDP variables (LY) between the tourism variables (LTOUR1 and LTOUR2) and the real exchange rate (LRQ), the former is more significantly effective. In order to examine the long-run causal relationship, we test for weak exogeneity among the cointegrating relationship [26, 15] . Hall and Milne [20] and Arestis et al. [4] interpret weak exogeneity in a cointegrated system as a notion of long-run causality. Weak exogeneity is rejected for LY, TOUR1, and TOUR2 at 5%, which indicates that bi-directional causal linkages between GDP and tourism development is identified. These empirical findings have important implications for Taiwan. The bi-directional causality between tourism and GDP in the long-run, which indicates the level of economic activity and tourism development, mutually influences each other in that a high level of economic growth leads to a high level of tourism development and vice versa. This suggests that tourism and GDP are endogenous and therefore any single equation forecast of one or the other could be misleading [59] . This is different from the empirical results of previous related literature in studying the causal relationship between tourism and economic development, such as Balaguer and Cantavella-Jorda [6] , Durbarry [12] , Eugenio-Martín et al. [16] , and Oh [51] -all of who obtain the one-way causal relationship of economic-driven tourism growth. 8 It is now extensively accepted that the relationship between economic and financial time series are mainly nonlinear. 9 Non-linear dependence may exist if tourism, real income, and the real exchange rate are generated by non-linear processes-this hypothesis is theoretically plausible and empirically substantiated. Several studies investigate the tests of non-linear causality relations, such as Baek and Brock [5] and Hiemstra and Jones [22] . Baek and Brock [5] show that the linear Granger causality tests have low power against the non-linear alternatives. However, any test based on the assumption of linearity fails to find any non-linear dependence. Savit [56] indicates that the difference between the linear and non-linear adjustments to any deviation from the equilibrium lies in whether or not the size of adjustment is proportional to the deviation. A proportional adjustment means a linear relationship, but this kind of adjustment cannot cause the randomness observed in financial and commodity markets. The estimation periods for this study cover the somewhat volatile time of the tourism development and economic innovation in Taiwan. Consequently, it is important to check the cointegration relationship for structural breaks. 10 In this context, Hansen [21] proposes three tests (L c , MeanF, and SupF) for parameter instability based on the full modified statistics. When the constant term is included in the specification, the data do not support the parameter stability. These tests show signs of instability. 11 Furthermore, Gregory and Hansen ([26] , henceforth GH) provide an alternative approach with tests that are based on the notion of regime change and are a generalization of the usual residual-based cointegration test. GH consider three alternative models-a level shift (model C), a level shift with trend (model C/T), and a regime shift that allows the slope vector to shift as well (model C/S). GH have developed versions of the cointegration ADF tests of Engle and Granger [14] , as well as the Z t and Z α tests of Phillips-Quliaris [54] , whereby all of them are modified according to the alternative considered. Taking into account that the date of the change is unknown, they compute the values of ADF * t = inf λ ∈ J ADF t , Z * t = inf λ ∈ J Z t , and Z * α = inf λ ∈ J Z α . GH test provides clear evidence of not finding cointegration even when we allow for a structural break in the relationship between LY, LTOUR1, and LRQ. However, the relationship between LY, LTOUR2, and LRQ reveals a structural break in ADF * and Z * at the 10% level. According to the ADF statistic criterion, the structural break years estimated on the basis of the three models are mainly in 1980 and 1983. The structural break years of 1980 and 1983 were caused by the following: first, the oil price shock of 1979, which preceded economic recessions and decreased tourism activities. Second, another shock was the change in international political relationships. As with the descriptions of the last section, in 1979 the United States changed its diplomatic recognition from Taiwan, and this policy resulting in Taiwan facing difficult international and cross-strait political relationships. This severe international political situation prompted a prolonged slowdown in Taiwan's tourism growth. Finally, the adjustments of tourism policies also created some effects. In 1980, the Statute for the Development of Tourism was amended to encourage the construction of international tourist hotels. Moreover, the first management offices of the national scenic area administration were set up in 1982. The structural break years are different between tests of Hansen [21] and GH [18] . The former is from 1975 to 1997, which were caused by many factors, including the changes of international and cross-strait political relationships, world economic crises, and relaxing tourism policies. The latter is 1980 and 1983 for the oil price shock of 1979 and the change of diplomatic recognition between Taiwan and the United States also in 1979. The structural break timing of the two tests is different, but both tests show the same phenomenon that international political change, economic shocks, and relaxing of some tourism controls and policies causing the structural breaks. The GDP-tourism development specification, enveloping the changing economic incidents, does raise some important questions on the long-run relationships between these series. These tests suggest that structural change in the cointegration vector is important and needs to be taken care of in the specification of GDP-tourism for Taiwan. This finding is reassuring since the endogenous estimation procedure produces structural breaks that correspond to recognizable tourism and economic events. It implies that, within the context of tourism development, households and the government may respond differently when the economy is in a different regime. Hence, the specification of GDP-tourism, enveloping the changing of tourism and economic events, does raise some important questions on the long-run relationships between GDP, tourism, and the real exchange rate. Taiwan's 921 earthquake in 1999 and the SARS epidemic in 2003 did not bring about an unstable relationship between tourism development and GDP in the long run as either the results of the Hansen [21] test or the GH [18] test show. Reviewing the investigation of past empirical literature [23, 42] , the above disasters indeed struck Taiwan's international tourism industry in the short run, but in the long run these disasters did not generate structural breaks as our empirical results present. After the 921 earthquake in 1999, the Tourism Bureau of Taiwan implemented a series of measures to help the island's tourism industry recover. The Tourism Bureau assisted in reconstructing public facilities under relief loans for tourist industries. More international promotions were carried out to attract foreign tourists [23] . After facing the detrimental effects of SARS, Taiwan's government launched a US$ 75 million promotional campaign to recover from the devastating impact of SARS on the tourism industry [42] . In the long run, the prompt and effective efforts of Taiwan's government have helped the tourism industry to recover and have concurrently improved Taiwan's awareness around the globe. These efforts avoided sapping Taiwan's tourist development in the long run. 12 All studies in the previous literature have not examined parameter constancy across policy changes, even though the countries they investigated have been under substantial structural changes due to a number of altering tourism and economic policies. This paper has studied the issue of whether regime changes have broken down the stability of the relationship between tourism development and economic growth in Taiwan from 1959 to 2003. We empirically reexamine the long-run co-movement and causal relationship between real GDP and tourism development in a multivariate model. By following the procedure of Pradhan and Subramanian [55] , we apply a three-step testing procedure to inquire into the stability of the relationship between tourism development and economic growth in Taiwan. The empirical evidence was based on an elaborate methodology, by first identifying the full systems model. Then, it was reduced to the single equation framework. Finally, testing was done for the structural break with unknown timing. This paper demonstrates the following empirical results. First, it indicates that all variables have the phenomena of a breakpoint. The breakpoint in 1992 was for real GDP, because of Taiwan's collapsing bubble economy in 1990. With the end of the U.S. aid program in Taiwan in 1965 plus Japan relaxing its ban on overseas travel in 1964, there was a breakpoint of tourism development variables in 1965 and 1966. The breakpoint of the real exchange rate in 1987 was due to the government of Taiwan releasing foreign exchange controls and altering its foreign exchange rate system. Next, tourism development looks to act as an engine of economic growth for Taiwan. Continuous economic growth will also generate a continuous rise on tourism development. However, the bi-directional causality between tourism and GDP in the long-run, which indicates the level of economic activity and tourism development, mutually influences each other in that a high level of economic growth leads to a high level of tourism development and vice versa. Third, the Hansen [21] test shows that the relationship between tourism development and GDP may indeed be unstable. At the same time, the GH [18] test does suggest that a structural change in the cointegration vector is important and needs to be taken care of in the specification of tourism development and GDP. We find that the breakpoints in various GDP-tourism match compatibly with critical economic, political, or tourist incidents of Taiwan. Finally, our empirical results also show that the development of Taiwan's tourist industry has coped with the economic development and political relationship between mainland China and Taiwan. Moreover, international and cross-strait political change, economic shocks, and relaxing of some tourism control and policies would break down the stability of the relationship between tourism development and economic growth. And a final observation worth noting, past empirical literature have showed that natural and human disasters, like Taiwan's 921 earthquake in 1999 and the SARS epidemic in 2003, indeed struck Taiwan's international tourism industry in the short run. Nevertheless, in the long run these disasters did not generate a structural break as our empirical results show. Furthermore, economic variables may exhibit non-linear dependencies, and in future research we suggest that researchers should consider the non-linear factor in the dynamic GDP-tourism relation. Linear and nonlinear Granger causality: evidence from the U.K. Stock Index Futures Market A compact econometric model of tourism demand for Turkey, Tourism Manage Some test of specification for panel data Monte Carlo evidence and an application to employment equations Financial development and economic growth: the role of stock markets A general test for nonlinear Granger causality: bivariate model, Working Paper Tourism as a long-run economic growth factor: the Spanish case Forecasting tourism revisited, Tourism Manage Spectral analysis of overseas tourists' expenditures in the United Kingdom Likelihood ratio statistics for autoregressive time series with a unit root Cointegration analysis of German and British tourism demand for Greece, Tourism Manage An econometric investigation of Greek tourism: a note Tourism and economic growth: the case of Mauritius Efficient tests for an autoregressive unit root Cointegration and error correction: representation, estimation, and testing Tourism and economic growth in Latin American countries: a panel data approach Tourism and economic growth: an empirical study Residual-based tests for co-integration in models with regime shifts Cointegration: how short is the long run The relevance of p-star analysis to UK monetary policy Tests for parameter instability in regressions with I(1) processes Testing for linear and nonlinear Granger causality in the stock price-volume relation Earthquake devastation and recovery in tourism: the Taiwan case, Tourism Manage Tourism demand in Turkey Statistical analysis of cointegration vectors Testing structural hypotheses in a multivariate cointegration analysis of the PPP and the UIP for UK The multiplier effect: Singapore's hospitality industry Tourism expansion and economic development: the case of Taiwan, Tourism Manage Forecasting international quarterly tourist flows using error correction and time series models Cointegration versus least squares regression The implications of tourism specialization in the long run: an econometric analysis for 13 OECD economies Determinants of inbound tourist expenditures Tourism development and economic growth: a closer look at panels, Tourism Manage Importance of secondary impact of foreign tourism receipts on the South Korean economy Review of international tourism demand models A meta-analytic review of international tourism demand Monthly seasonal variations: Asian tourism to Australia Forecasting tourist arrivals A cointegration analysis of annual tourism demand by Malaysia for Australia Multiple trend breaks and the unit root hypothesis Is the export-led growth hypothesis valid for industrialized countries? SARS devastation on tourism: the Taiwan case The price-volume relationship in the crude oil futures market-some results based on linear and nonlinear causality testing A tourism demand model for Fiji Tourism demand modeling; some issues regarding unit roots, co-integration and diagnostic tests Determinants of tourist expenditure in Fiji: a cointegration approach Fiji's tourism exports: an ARDL approach to cointegration Did Rabuka's military coups have a permanent effect or a transitory effect on tourist expenditure in Fiji: evidence from Vogelsang's structural break test, Tourism Manage The structure of tourist expenditure in Fiji: evidence from unit root structural break tests Lag length selection and the construction of unit-root tests with good size and power The contribution of tourism development to economic growth in the Korean economy, Tourism Manage Further evidence on breaking trend functions in macroeconomic variables Testing for unit roots in time series regression Asymptotic properties of residual based tests for cointegration On the stability of demand for money in a developing economy: some empirical issues When random is not random: an introduction to chaos in market prices The demand for tourism in North East England with special reference to Northumbria: an empirical analysis, Tourism Manage Tourism Demand Modelling and Forecasting: Modern Econometric Approaches Forecasting international tourist flows to Macau, Tourism Manage Univariate modelling using periodic and non-periodic analysis: inbound tourism to Japan Evaluation of demand: US tourists to Aruba Growth, development and tourism in a small economy: evidence from Aruba Wald-type test for detecting breaks in the trend function of a dynamic time series Predicting the tourism demand using fuzzy time series and hybrid grey theory, Tourism Manage Annual reports, progress and priorities 2004/05, The World Travel and Tourism Council published The evolution of commercial banking and financial markets in Taiwan Further evidence on the great crash, the oil-price shock, and the unit-root hypothesis The authors thank Prof. Michael McAleer, the editor, and an anonymous referee for their valuable suggestions and comments. Errors and omissions, if any, are our own.