Anatolia: An International Journal of Tourism and Hospitality Research
Anatolia: An International Journal of Tourism and Hospitality Research
To cite this article: Chew Ging Lee (2012) Tourism, trade, and income: evidence from Singapore,
Anatolia: An International Journal of Tourism and Hospitality Research, 23:3, 348-358, DOI:
10.1080/13032917.2012.701596
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Anatolia – An International Journal of Tourism and Hospitality Research
Vol. 23, No. 3, November 2012, 348–358
Nottingham University Business School, The University of Nottingham, Malaysia Campus, Jalan
Broga, 43500 Semenyih, Selangor Darul Ehsan, Malaysia
(Received 16 January 2012; final version received 7 June 2012)
This paper examines the short-run and long-run dynamic interactions between exports,
imports, international tourism, and economic growth for Singapore using annual data
over 1980– 2007. Since the sample size is relatively small, the long-run relationship of
these variables is investigated with the bounds test. Granger causality test is then used
Downloaded by [University of Otago] at 08:14 08 January 2015
to examine the dynamic interactions of these variables. The results suggest that the
dynamic interactions of these four variables are complex. The results support growth-
led tourism, tourism-led imports, and export-led tourism hypotheses in the short run.
The results also show that imports have positive effects on economic growth in the long
run. It is also found that tourism has indirect effects on economic growth in the long run
through import activities. There is a bidirectional causal relationship between exports
and income, which, in turn, affects the tourism industry.
1. Introduction
Recently, researchers have studied the relationship between international trade and inter-
national tourism. However, the number of studies is limited. Only four papers have been
found that studied this relationship with the Granger causality test. The first paper is that of
Kulendran and Wilson (2000), which investigates the relationship between international
trade and international travel between Australia and its four travel and trading partners:
Japan, New Zealand, the UK, and the US. They have approximated international trade with
exports, imports, and total trade. International travel is approximated by total travel,
business travel, and holiday travel. Using the Engle and Granger (1987) approach, they find
evidence of cointegration between some of the proxies for international trade and inter-
national travel for Australia’s main travel and trading partners. Applying the standard
Granger causality test in a VAR, they find the directions of causality between these variables
are mixed depending on the proxies being used and Australia’s main travel and trading
partner being studied. For instance, total trade Granger causes total travel if Australia’s main
travel and trading partner is the US, whereas, for Japan, total travel Granger causes total
trade.
Subsequently, Shan and Wilson (2001) investigate the relationship between China’s
tourist arrivals and total trade with respect to four of its major tourist origin countries:
Australia, Japan, the UK, and the US. Using the Granger non-causality approach
*Email: [email protected]
developed by Toda and Yamamoto (1995) and control variables, such as income in the
origin country, the living cost, and exchange rate in China, they find a bidirectional
causality between tourist arrivals and total trade for all China’s main tourist origin
countries. Khan, Toh, and Chua (2005) have duplicated the study of Kulendran and Wilson
(2000) with the data of Singapore and its major trading partners, such as ASEAN,
Australia, Japan, the UK, and the US. They find evidence of cointegration between some
proxies of international trade and international travel for some of Singapore’s main trading
partners. However, in this study, the results of the Granger causality test shows less
evidence to support the causality between these two variables in comparison to that of
Kulendran and Wilson (2000). Katircioglu (2009) examines the interaction of international
trade, international tourism, and economic growth with respect to Cyprus. The methodology
and approach used by this author is similar to this paper.
This paper extends the earlier studies by including one additional variable, the income of
the destination country into the empirical model. The inclusion of this additional variable is
Downloaded by [University of Otago] at 08:14 08 January 2015
important because the empirical models of the earlier papers have been mis-specified in
terms of omitting this important variable. This paper tests the nature of the causal
relationship between international tourism, international trade, and economic growth using
data from Singapore. There are six main reasons for choosing Singapore as a case study:
First, tourism is one of the fastest growing industries in Singapore, growing at annual
average of 15% over the last three decades (Khan et al., 2005). Second, tourism contributes
to approximately 7% of Singapore’s gross domestic product (GDP) and provides jobs to
approximately 14% of its labour force (Khan et al., 2005). Third, Singapore is one of the top
20 tourist destinations in the world (Khan & Abeysinghe, 2002). Fourth, as a major entrepot
country, Singapore’s total trade is more than 300% of its GDP (Singapore Department of
Statistics, 2001). Fifth, the Singaporean government plays an active role in the development
of tourism in Singapore. It aims to attract 17 million international visitor arrivals and earn
S$30 billion per year by 2015. Lastly, in 2010, Singapore earned tourism receipts of S$18.9
billion and received international arrivals of 11.6 million (Singapore Tourism Board, 2011).
The subsector of tourism industry, Business Travel, Meetings, Incentives, Exhibitions, and
Conventions (BTMICE) sector, attracted a total of 3.1 million business travellers,
contributing to almost 27% of all visitors to Singapore, and generated approximately S$5.4
billion in tourism receipts in 2010. The Singaporean government has also actively been
involved in attracting business travellers, through the Singapore Exhibition and Convention
Bureau (SECB), a subsidiary of the Singapore Tourism Board. SECB aims to establish
Singapore as a dynamic business events destination by continuously developing existing
conventions and exhibitions, and attracting new business events. In 2010, 108 new events,
such as MedTech 2010, RoboCup 2010, and Blueprint, were held in Singapore.
This study contributes to the existing literature in the following ways: First, this paper
analyses the interactions of international trade, international tourism, and economic growth
simultaneously. It tackles the possible mis-specification of a model by going beyond the
simple bivariate causal relationship. Second, total exports and total imports of Singapore
have been used as the two proxies for trade. Total exports and total imports are allowed to
interact with Singapore’s total international tourist arrivals and GDP simultaneously. It is
believed that this is a better approach, instead of using total trade, the sum of total exports
and imports, because Singapore, as a city state, is resource lacking. It imports raw materials,
intermediate, and final goods. Some of these imports will be used in the production of
goods for export and domestic consumption. Some will be re-exported. Therefore, the
understanding of import and export interactions is important in this study. Instead of using
the exports and imports of Singapore to each of its main trading partners, the total exports
350 C.G. Lee
and total imports of Singapore are used in this study because we believe that only total
exports and imports can have an effect on Singapore’s income, based on the export-led
growth and import-led growth hypotheses. Third, an appropriate statistical procedure for
estimating cointegration when the sample size is small has been utilized. The bounds test
developed by Pesaran, Shin, and Smith (2001) is used to overcome many of the
shortcomings of other popular estimation methods.
The rest of this paper is presented as follows. Section 2 reviews the related literature.
Section 3 describes the data sources, presents the econometric methodology, and analyses
the empirical findings. The final section concludes this paper with implications for policy
purposes.
2. Literature review
Many studies have examined for the presence of tourism-led growth hypothesis, but the
Downloaded by [University of Otago] at 08:14 08 January 2015
results of the direction of causality are mixed. Oh (2005) is unable to find a cointegrating
relationship between these two variables and shows that income Granger causes tourism
for Korea in the short run. Balaguer and Cantavella-Jordá (2002) confirm tourism-led
growth hypothesis with Spanish quarterly data. Dritsakis (2004) finds a cointegrating
relationship and bidirectional causality between tourism and income for Greece. Using
annual data from 1970 to 2007, Belloumi (2010) finds cointegration between tourism and
economic growth in Tunisia. The results of Granger causality test also suggest that tourism
has a positive effect on economic growth. Brida, Lanzilotta, Lionetti, and Risso (2010)
find a cointegrating relationship between real GDP per capita of Uruguay, real expenditure
of Argentinian tourists to Uruguay, and the relative price between Uruguay and Argentina.
They also find a unidirectional causality from real expenditure of Argentinian tourists to
Uruguayan real GDP. Gunduz and Hatemi-J (2005) find a tourism-led growth hypothesis
in the case of Turkey with leveraged bootstrap causality tests. Using heterogeneous panel
cointegration, Lee and Chang (2008) find unidirectional causality from tourism
development to economic growth in OECD countries and bidirectional causality between
these two variables in non-OECD countries. Tang (2011) performs full sample and rolling
window Granger causality tests to study the short-run and long-run dynamic interactions
between tourist arrivals, real output, and real effective exchange rate in Malaysia. The
obtained results support tourism-led growth hypothesis for both types of Granger causality
tests, but such relationship is unstable after 2005.
In recent years, the relationship between international trade and economic growth has
also become a central topic in empirical research, for instance, the works of Awokuse
(2006, 2008), Siliverstoves and Herzer (2006), Abu-Qarn and Abu-Bader (2004) and
others. Awokuse (2006), using the method developed by Toda and Yamamoto (1995) and
directed acyclic graph, reveals that the causal relationship between Japanese exports and
GDP growth is bidirectional. Siliverstovs and Herzer (2006) also find evidence to support
export-led growth hypothesis for Chile. Abu-Qarn and Abu-Bader (2004) investigate for
the presence of export-led growth hypothesis with the data of nine Middle East and North
African countries. When total exports are used, only one country supports export-led
growth hypothesis. However, when manufactured exports are used, export-led growth is
supported by the countries with relatively high share of manufactured exports in the total
merchandise exports. Recognizing that export expansion and import expansion may lead
to economic growth, Awokuse (2008) examines the interaction of exports, imports, and
GDP growth for three countries: Argentina, Colombia, and Peru with a set of control
variables. He is unable to find evidence to support export-led growth hypothesis both in the
Anatolia – An International Journal of Tourism and Hospitality Research 351
short run and long run for Argentina and Peru. Export-led growth hypothesis is supported
only in the long run for Colombia. However, there is evidence to support import-led
growth for all these countries. There is a short-run causality from imports to economic
growth for Argentina and Peru, and there is a short-run causality and a long-run causality
from imports to economic growth for Colombia. Based on the works in tourism-led growth
and trade-led growth hypotheses, it should be recognized that there are not just the
interactions between international trade and international tourism, there are also the
interactions between international trade and income, and between international tourism
and income. According to Reizman, Summers, and Whiteman (1996), omitting the
important variables can result in spurious conclusions regarding the export-led growth
hypothesis. Following the suggestion of Reizman et al. (1996), this study will address the
interactions of international tourism, international trade, and economic growth into a
single empirical framework.
Downloaded by [University of Otago] at 08:14 08 January 2015
ADF KPSS
Variable
Level First difference Level First difference
EX 0.1490[0] 2 3.8954[0]** 0.1819[3]** 0.0842[2]
IM 2 0.4230[0] 2 4.2578[0]** 0.1569[3]** 0.0902[3]
GDP 2 0.4094[0] 2 3.9472[0]** 0.1407[3]* 0.0944[1]
TOU 2 3.8551[5]** 2 1.9124[5] 0.0884[2] 0.0726[1]
*,** and *** Statistically significant at the 10%, 5%, and 1% levels, respectively.
t-Statistic and LM-statistic are reported for ADF and KPSS tests, respectively. The brackets beside t-statistic
indicate the number of lagged first differences of ADF selected based on the Schwarz information criterion. The
brackets beside LM-statistic indicate the choice of bandwidth parameter in the Bartlett-kernel-based sum-of-
covariances estimator selected based on Newey –West data-based automatic bandwidth parameter methods.
Downloaded by [University of Otago] at 08:14 08 January 2015
suggest that TOU is stationary at level but non-stationary at first difference. This type of
property is impossible to occur. The results of KPSS test suggest that TOU is stationary at
level and first difference. Therefore, it is concluded that TOU is stationary.
To investigate for a cointegrating relationship between these variables, the bounds test
within the autoregressive distributed lag (ARDL) framework and based on the F-statistic
proposed by Pesaran et al. (2001) is used. The bounds test instead of other popular
approaches, such as Engle and Granger (1987), Johansen (1988), and Johansen and
Juselius (1990) approaches, is used because the bounds test can be applied irrespective of
whether the variables are purely I(1), purely I(0), or mutually cointegrated and has better
small sample properties. Pesaran and Shin (1999) have shown that the ARDL-based
estimators of the long-run coefficients are super-consistent. For this study, it is believed
that the bounds test is more suitable because the sample has only 28 observations and the
variables are either I(1) or I(0).
The bounds test investigates whether a cointegrating relationship exists in the
following unrestricted error correction models:
X
q X
q X
q
DGDPt ¼ a0 þ aGi DGDPt2i þ aTi DTOUt2i þ aIi DIMt2i
i¼1 i¼0 i¼0
X
q
þ aEi DEXt2i þ a1 GDPt21 þ a2 TOUt21 þ a3 IMt21 þ a4 EXt21 þ 11t ð1Þ
i¼0
X
q X
q X
q
DTOUt ¼ b0 þ bTi DTOUt2i þ bGi DGDPt2i þ bIi DIMt2i
i¼1 i¼0 i¼0
X
q
þ bEi DEXt2i þ b1 GDPt21 þ b2 TOUt21 þ b3 IMt21 þ b4 EXt21 þ 12t ð2Þ
i¼0
X
q X
q X
q X
q
DIMt ¼ c0 þ cIi DIMt2i þ cGi DGDPt2i þ cTi DTOUt2i þ cEi DEXt2i
i¼1 i¼0 i¼0 i¼0
X
q X
q X
q
DEXt ¼ d0 þ dEi DEXt2i þ d Gi DGDPt2i þ d Ti DTOUt2i
i¼1 i¼0 i¼0
X
q
þ d Ii DIMt2i þ d1 GDPt21 þ d2 TOUt21 þ d3 IMt21 þ d4 EXt21 þ 14t ð4Þ
i¼0
5.61) at the 1% significance level. The null of no cointegration is rejected if the computed
F-statistic obtained from each null hypothesis exceeds the respective upper critical value.
If the computed F-statistic is lower than the respective lower critical value, the null
hypothesis is accepted. If the computed F-statistic falls inside the lower and upper critical
values, no conclusive decision can be made unless the order of integration of the variables
under consideration is known. The null hypothesis of no cointegration of each equation is
stated in the second column of Table 2. The null of no cointegration is rejected at the 1%
significance level for Equation (1). The null of no cointegration is accepted even at the
10% significance level for Equations (2) and (4). For Equation (3), the bounds test shows
inconclusive result at the 10% significance level and accepts the null of no cointegration at
the 5% level. Therefore, it is concluded that there is no cointegration for Equation (3).
Since there is evidence of cointegration when GDP is used as the dependent variable,
the Granger causality test applied on the case where GDP is the dependent variable will be
augmented with an error correction term based on the suggestion of Engle and Granger
(1987). To obtain the cointegrating relationship, the following ARDL model is estimated:
ð1 2 f1 L 2 · · · 2 fp L p ÞGDPt ¼ b0 þ ð1 2 b1 L 2 · · · 2 bq L q ÞTOUt
þ ð1 2 a1 L 2 · · · 2 ar L r ÞIMt ð5Þ
þ ð1 2 u1 L 2 · · · 2 us L s ÞEXt þ nt
The values of p, q, r, and s in Equation (5) are selected based on SIC. However, because of
a relatively small sample size and annual data used in this study, the maximum possible
values of p, q, r, and s are set at 1. SIC selects p ¼ 1, q ¼ 0, r ¼ 0, and s ¼ 1. Equation (5)
is reparameterized to obtain the long-run coefficients and the corresponding asymptotic
Equation H0 F-Value q
(1) a1 ¼ a2 ¼ a3 ¼ a4 ¼ 0 31.1500*** 2
(2) b1 ¼ b2 ¼ b3 ¼ b4 ¼ 0 0.4735 1
(3) c1 ¼ c2 ¼ c3 ¼ c4 ¼ 0 3.0144 2
(4) d1 ¼ d2 ¼ d3 ¼ d4 ¼ 0 1.0514 2
*,** and *** Statistically significant at the 10%, 5%, and 1% levels, respectively.
354 C.G. Lee
where h^t is the error correction term and the asymptotic standard errors are reported in
parentheses. IM is statistically significant at the 5% significance level. The estimated
coefficient of IM is positive implying that IM has positive effects on GDP in the long run.
TOU is statistically insignificant even at the 10% level. The statistically insignificant
coefficient of TOU is positive. EX is marginally significant at the 10% level because the
p-value of the null hypothesis of the coefficient of EX equal to zero is only 9.6%. But, the
sign of the estimated coefficient of EX is negative indicating that EX has negative impacts
on GDP. Instead of providing support for export-led growth, the result suggests that an
increase in export activities lead to a fall in the income of Singapore. Since EX is only
Downloaded by [University of Otago] at 08:14 08 January 2015
marginally significant and its coefficient has a wrong sign, it is suggested that EX has no
effect on GDP in the long run.
To implement the Granger causality test, the following models are specified:
X
q X
q X
q
DGDPt ¼ e0 þ eGi DGDPt2i þ eTi DTOUt2i þ eIi DIMt2i
i¼1 i¼1 i¼1
X
q
þ eEi DEXt21 þ e1 ECTt21 þ m1t ð7Þ
i¼1
X
q X
q X
q
DTOUt ¼ f 0 þ f Ti DTOUt2i þ f Gi DGDPt2i þ f Ii DIMt2i
i¼1 i¼1 i¼1
X
q
þ f Ei DEXt2i þ m2t ð8Þ
i¼1
X
q X
q X
q
DIMt ¼ g0 þ gIi DIMt2i þ gGi DGDPt2i þ gTi DTOUt2i
i¼1 i¼1 i¼1
ð9Þ
X
q
þ gEi DEXt2i þ m3t
i¼1
X
q X
q X
q
DEXt ¼ h0 þ hEi DEXt2i þ hGi DGDPt2i þ hTi DTOUt2i
i¼1 i¼1 i¼1
X
q
þ hEi DIMt2i þ m4t ð10Þ
i¼1
where ECTt21 is the error correction term. h^t21 of Equation (6) is used to replace ECTt21
of Equation (7) because ECTt21 is not observable. The value of q in each equation is
selected with the same approach as that applied to Equations (1) – (4). The results of short-
run and long-run Granger causality are presented in Table 3. The long-run Granger
causality is implemented only on Equation (7). To investigate for the presence of long-run
Granger causality, the t-test is used to check whether the coefficient of the error correction
Anatolia – An International Journal of Tourism and Hospitality Research 355
term is different from zero. Further, the sign of the coefficient of the error correction term
Downloaded by [University of Otago] at 08:14 08 January 2015
must be negative. The rejection of this null suggests that long-run Granger causality is
present. The short-run Granger causality can be investigated with the f-test. If the lagged
first differences of an explanatory variable are jointly insignificant, there will be no short-
run Granger causality from that explanatory variable to the dependent variable.
In Equation (7), only the lagged first differences of EX are jointly significant. These
results suggest that there is only short-run Granger causality running from EX to GDP. The
coefficient of error correction term is statistically significant with the correct sign. Granger
(1988) points out that there will be causality among the variables at least in one direction,
if there is a cointegrating relationship among them. However, based on the results of
Equation (6), it is concluded that there is a long-run Granger causality running from IM to
GDP because only IM is statistically significant. In Equation (8), the lagged first
differences of GDP are jointly significant at the 1% level. This result suggests that there is
a short-run Granger causality running from GDP to TOU. The lagged first differences of
EX are also jointly statistically significant at the 10% level. This result suggests EX
Granger causes TOU in the short-run. In Equation (9), the lagged first differences of GDP
and EX are jointly significant at the 1% level. This implies that there is a short-run Granger
causality running from GDP to IM and from EX to IM. The lagged differences of TOU are
TOU GDP
EX IM
also jointly significant only at the 10% level. This suggests that there is a short-run
Granger causality running from TOU to IM. In Equation (10), only the lagged first
differences of GDP are statistically significant. This indicates that GDP Granger causes
EX in the short-run. Figure 1 summarizes the results from the f- and t-tests.
supported growth-led tourism, tourism-led imports, and export-led tourism in the short run.
The results also show that imports have positive effects on economic growth in the long run.
The findings indicate that tourism has indirect effects on economic growth in the
long run. International tourism Granger causes imports in the short run, but imports
Granger causes income in the long run. This justifies the intense involvement of the
Singaporean government at promoting and increasing its tourism demand that make tourism
an indirect determinant of overall economic growth in the long run. The expansion of
Singapore’s economy, in terms of higher GDP, will have direct and indirect effects on the
tourism industry. GDP Granger causes tourism directly because there are few natural attrac-
tions in Singapore, and significant resources are required for the upgrading and maintenance
of existing attractions and infrastructure through economic growth (Lee, 2008). GDP also has
indirect impacts on tourism through export activities. This can be supported by the suggestion
of Kulendran and Wilson (2000) that business travellers arrive in a country for the purpose of
buying a product from that country. Finally, there is a bidirectional causal relationship
between exports and income. This suggests that the expansion of income has effects on the
export activities and vice versa. The interactions of GDP and exports of Singapore have
impacts on its tourism industry.
The totality of the results suggests that promoting international tourism without
promoting exports and imports may not lead to the potential growth-enhancing contribution
of the tourism industry. For Singapore, export promotion, as a strategy of economic growth
should be the focus of the government. Export promotion would be an effective tool, which
can lead to economic growth both directly and indirectly. The findings suggest that there is a
need for policy makers to focus on promoting trade openness to capitalize the effects of
international tourism on economic growth. In summary, this study suggests that exclusion
of imports and exports, and focus on just the role of international tourism as the engine of
growth may be misleading. Current evidence does not provide a direct effect of
international tourism on economic growth. Overall, this study indicates that international
tourism, exports, imports, and economic growth are interacting and reinforcing each other
either directly or indirectly.
References
Abu-Qarn, A., & Abu-Bader, S. (2004). The validity of the ELG hypothesis in MENA region:
Cointegration and error correction model analysis. Applied Economics, 36(15), 1685– 1695.
Awokuse, T.O. (2006). Export-led growth and the Japanese economy: Evidence from VAR and
directed acyclic graphs. Applied Economics, 38(5), 593– 602.
Anatolia – An International Journal of Tourism and Hospitality Research 357
Awokuse, T.O. (2008). Trade openness and economic growth: Is growth export-led or import-led?
Applied Economics, 40(2), 161– 173.
Balaguer, J., & Cantavella-Jordá, M. (2002). Tourism as a long-run economic growth factor: The
Spanish case. Applied Economics, 34(7), 877– 884.
Belloumi, M. (2010). The relationship between tourism receipts, real effective exchange rate and
economic growth in Tunisia. International Journal of Tourism Research, 12(5), 550– 560.
Brida, J.G., Lanzilotta, B., Lionetti, S., & Risso, W.A. (2010). Research note: The tourism-led
growth hypothesis for Uruguay. Tourism Economics, 16(3), 765– 771.
Dickey, D.A., & Fuller, W.A. (1979). Distributions of the estimators for autoregressive time series
with a unit root. Journal of the American Statistical Association, 74, 427– 431.
Dickey, D.A., & Fuller, W.A. (1981). Likelihood ratio statistics for autoregressive time series with a
unit root. Econometrica, 49(4), 1057– 1072.
Dritsakis, N. (2004). Tourism as a long-run economic growth factor: An empirical investigation for
Greece using causality analysis. Tourism Economics, 10(3), 305– 316.
Engle, R.F., & Granger, C.W.J. (1987). Cointegration and error correction presentation: Estimation
and testing. Econometrica, 55(2), 251– 276.
Downloaded by [University of Otago] at 08:14 08 January 2015
Shan, J., & Wilson, K. (2001). Causality between trade and tourism: Empirical evidence from China.
Applied Economics Letters, 8(4), 279– 283.
Shareef, R., & McAleer, M. (2007). Modelling the uncertainty in monthly international tourist
arrivals to the Maldives. Tourism Management, 28(1), 23 – 45.
Sheldon, P.J. (1993). Forecasting tourism: Expenditures versus arrivals. Journal of Travel Research,
32(1), 13 –20.
Siliverstovs, B., & Herzer, D. (2006). Export-led growth hypothesis: evidence for Chile”. Applied
Economics Letters, 13(5), 319– 324.
Singapore Department of Statistics (2001). Yearbook of statistics. Singapore: Singapore Department
of Statistics.
Singapore Tourism Board (2011). Singapore Tourism Board Annual Report 2010/2011. Singapore:
Singapore Tourism Board.
Tang, C.F. (2011). Tourism, real output and real effective exchange rate in Malaysia: A view from
rolling sub-samples. MPRA Paper 29379. Munich, Germany: University Library of Munich.
Toda, H.Y., & Yamamoto, T. (1995). Statistical inferences in vector autoregression with possibly
integrated processes. Journal of Econometrics, 66(1– 2), 225– 250.
Downloaded by [University of Otago] at 08:14 08 January 2015