Economic Modelling 33 (2013) 65–70

Contents lists available at SciVerse ScienceDirect

Economic Modelling
journal homepage: www.elsevier.com/locate/ecmod

Terrorism and tourism: A conjunction and ramification in Pakistan

Syed Ali Raza ⁎, Syed Tehseen Jawaid 1
IQRA University, Karachi-75300, Pakistan

a r t i c l e i n f o a b s t r a c t

Article history: This study investigates the impact of terrorism activities on tourism in Pakistan by using the annual time
Accepted 19 March 2013 series data from the period of 1980 to 2010. Johansen and Jeuuselius and ARDL bound testing cointegration
Available online xxxx approach confirms the valid long run relationship between terrorism and tourism. Results indicate the signif-
icant negative impact of terrorism on tourism in the long run as well as in the short run. Results of rolling
JEL classification:
window estimation method indicate that terrorism having negative coefficients throughout the sample peri-
L83
C32
od. Results of dynamic ordinary least square (DOLS) suggest that the results will remain the same in the fu-
ture up to lead 2. Results of Granger causality test, Toda and Yamamoto Modified Wald causality test and
Keywords: variance decomposition test confirm the unidirectional causal relationship between terrorism and tourism,
Tourism causality runs from terrorism to tourism. Hye and Khan (2012) confirm tourism led growth hypothesis for
Time series analysis Pakistan. It is suggested that the government should play a significant role to reduce terrorism activities
Terrorism that should be decreased in the country to boost tourism activities that lead to increase in income from tour-
Cointegration
ism in Pakistan.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction
Many studies have been conducted on tourism activities. Some
of them show the relationship between tourism activities with eco-
nomic growth and found significant positive relationship between
tourism and economic growth. 2On the other hand, some studies are
done on the relationship between terrorism and tourism activities
and conclude the negative and significant effect of terrorism on tour-
ism activities. 3
In Pakistan, during last three decades, the average tourism activi-
ties have significantly increased. In the 1980's, the average tourism
receipt was 41 million, in the 1990's it was 45 million and in the
last decade the average tourism receipt was 196 million. In contrast,
terrorist activities have also increased in last three decades. In the
1980's, the average terrorist activities was 27, in the 1990's it was
163 and in the last decade it was 251.
Fig. 1.1 shows the trend in terrorism and tourist activities. Both activ-
ities show a mix trend but in the last few years, the terrorist activities
have drastically increased in the country. The question is that, are terror-
ist activities correlated with tourism in Pakistan? This study examines
this question by using long time series data from 1981 to 2010.
⁎ Corresponding author. Tel.: +92 333 344 8467.
E-mail addresses: syed_aliraza@hotmail.com (S.A. Raza), stjawaid@hotmail.com
(S.T. Jawaid).
1
Tel.: +92 345 309 4838.
2
Gunduz and Hatemi (2005), Brida et al. (2010) and Hye and Khan (2012).
3
Seddighi et al. (2001), Chen and Noriega (2004), Floyd et al. (2003) and Kingsbury
and Brunn (2004).
The rest of the paper is organized as follows: following the introduc-
tion Section 2 reviews some selected studies, Section 3 discusses empir-
ical strategy, Section 4 shows estimations and results, Section 5 shows
results of rolling window estimation, Section 6 discusses the results of
dynamic ordinary least square, Section 7 shows the results of causal re-
lationship between tourism and terrorism and the final section con-
cludes the study and provides some policy implications.
2. Review of literature
Most of the empirical research suggests that terrorism has signifi-
cant negative effect on tourism activities. Some selected studies are
reviewed below.
Cook and McCleary (1983) and D'Amore and Anuza (1986) argue
that previous international experience of tourists also influences their
reaction to terrorism. In contrast, Sonmez and Graefe (1998) suggest
the indirect impact of past international experience on future behavior.
The nature of previous travel also has impact on future travel behavior
(Mazursky, 1989).
Hartz (1989) argues that tourists modify their traveling behavior
to risky destination because of the risk of terrorism. Cost of experi-
ence increased due to increase in tourist perceived risk caused by
the risk from the terrorism at destination resulting in the substitution
of that destination with one perceived as safe (Gu and Martin, 1992).
Enders and Sandler (1991), Enders et al. (1992) and Mansfeld (1996).
Mansfeld (1996) claims that the country's level of involvement in secu-
rity situation correlates with its visitor's number. Conversely, Mansfeld
(1996), Enders et al. (1992) and Sonmez (1998) argue that tourists

0264-9993/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.econmod.2013.03.008
66 S.A. Raza, S.T. Jawaid / Economic Modelling 33 (2013) 65–70
Source: Authors' Construction
Fig. 1.1. Trend of tourism and terrorism in Pakistan. Source: Authors' construction.
(Inter-Regional) generalized argument to the whole region. This is
called “Generalization Effects”.
Enders and Sandlers (1991) investigate the causal relationship be-
tween transnational terrorism and tourism in Spain. Monthly time se-
ries data have been used from 1970 to 1988. Vector auto regression
method has been applied. Results indicate that unidirectional causal-
ity run from terrorism to tourism exists.
Drakos and Kutan (2001) discuss the regional effect of terrorism
on tourism in three Mediterranean countries namely Greece, Israel and
Turkey. Seemingly unrelated regression model is used by employing
monthly time series data from January 1996 to December 2000. Results
indicate that tourism industry in Turkey and Israel is more sensitive to
terrorism than Greece.
Seddighi et al. (2001) affirm that political instability may be caused
by terrorist attacks resulting to turn down in arrivals in some tourist des-
tination. Chen and Noriegi (2004), Floyd et al. (2003) and Kingsbury and
Brunn (2004) show that demand to cancel travel or holiday plan has in-
creased after 9/11 terrorist attacks.
Alsarayreh et al. (2010) investigate the effect of terrorism on in-
ternational tourism activities. Primary data has been collected from
42 different countries through a questioner. Descriptive statistics in-
dicate that most of the respondents believe that terrorism has re-
duced international tourism activities.
Henderson et al. (2010) examine the effect of terrorism and tour-
ism specifically to the hotel industry of Singapore. Four interviews
have been conducted by security measures of Singapore's superior
standard hotels and some are done from national security industry in-
stitute and academic experts. It is concluded that terrorist activities
have unfavorable impact for destination where it happens. Hotels
are attractive marks for attacks. The risks are considered and manage-
ment is trying secure properties against damages.
Many time series and cross-sectional studies have been conducted
on tourism. These previous researches have been divided into two
groups. The first group of studies tests the relationship of tourism
with economic growth. Researches in the second group identify the
causal relationship between tourism and economic growth.
International tourism is considered as a source of foreign ex-
change earnings and if these earnings are used to import capital
goods, it leads to enhance economic growth (Mckinnon, 1964).
Ghali (1976) examines the role of tourism on economic growth. It is
concluded that in the absence of tourism, personal income which has
been recorded was 17 percent low. Balaguer and Cantavella-Jorda
(2002) examine that tourism led growth hypothesis in Spain by
employing quarterly time series data. Result indicates the long run mul-
tiplier effect growth. Skerritt and Huybers (2005) investigate the effect
of international tourism on economic growth of 37 developing coun-
tries. It is concluded that tourism has significantly contributed to boost
economic growth in developing countries.
Gunduz and Hatemi (2005) examine tourism led growth hypoth-
esis in Turkey. Bootstrap causality test has been used. Results indicate
that unidirectional causality exists, runs from tourism to economic
growth exists. On the other hand, Dritsakis (2004) concludes that bi-
directional causality exist between tourism and economic growth in
Greece. Cortes-Jemenez and Pulina (2006) have found that support-
ive evidence of tourism led growth hypothesis for Spain by employing
multivariate Granger causality test. Khalil et al. (2007) investigate the
relationship between tourism and economic growth in Pakistan.
Annual time series data have been used from 1960 to 2005. Results
indicate that significant positive relationship exists between tourism
and economic growth in the short run. Results also confirm bidirec-
tional causality between tourism and economic growth.
Katircioglu (2009) re-examine tourism led growth hypothesis for
Turkey by using Johansen and bound testing approach to cointegration.
No causal relationship has been found between tourism and economic
growth. Kreishan (2010) investigates the causal relationship between
tourism earnings and economic growth. Result shows unidirectional cau-
sality run from tourism earnings to economic growth. Hye and Khan
(2012) examine that tourism led growth hypothesis in Pakistan by
using annual time series data from 1971 to 2008. Johansen and Juselius
and autoregressive distributive lag model for cointegration are used.
Results show that long run relationship exists between income from
tourism and economic growth in Pakistan.
Overall, it is hypothesized that tourism has negative relationship
with terrorism activities, while, most of the studies suggest that tour-
ism has positive effect on economic growth.
3. Empirical framework
In this study, the 31-year annual time series data of Pakistan has
been used from 1980 to 2010. Data of tourism is gathered from sever-
al issues of economic survey of Pakistan, while, data of terrorism ac-
tivities is gathered from the Global terrorism database. 4All variables
are used in logarithm form.
3.1. Unit root test
Augmented Dickey Fuller (ADF) 5and Phillip Perron (PP) 6unit root
test are used to examine the stationary properties for long run rela-
tionship of time series variables. Augmented Dickey Fuller (ADF)
test is based on equation given below:
X
k
ΔY t ¼ α 0 þ α 1 Y t−1 þ dj ΔY t−j þ εt
j¼1
Where εt is pure white noise error term, Δ is first difference oper-
ator, Yt is a time series, α0 is the constant and k is the optimum
numbers of lags of the dependent variable. Augmented Dickey Fuller
(ADF) test determines whether the estimates of coefficients are equal
to zero. ADF test provides cumulative distribution of ADF statistics.
The variable is said to be stationary, if the value of the coefficient δ
is less than the critical values from fuller table. Phillip and Perron
(PP) unit root test equation is given below:

ΔY t ¼ α þ ρ Y t−1 þ εt
The Phillip and Perron unit root test is also based on t-statistics
that is associated with estimated coefficients of ρ⁎.
4
It is maintained by the National Consortium for the Study of Terrorism and Re-
sponses to Terrorism (START), based at the University of Maryland. Web Link: http://
www.start.umd.edu/gtd/.
5
See, Dicky and Fuller (1979).
6
See, Phillips and Perron (1988).
 S.A. Raza, S.T. Jawaid / Economic Modelling 33 (2013) 65–70 67

3.2. Cointegration method Table 4.1

Stationary test results.
Source: Authors' estimation.
Johansen and Jeuuselius J.J (1990) and Auto Regressive Distributed
Lag (ARDL) cointegration methods are used to examine the long run Variables Augmented Dickey-Fuller (ADF) Phillips–Perron (PP)
relationship between terrorism and tourism. The J. J. cointegration I(0) I(1) I(0) I(1)
test is based on λtrace and λmax statistics. First “trace test” cointegration
C C&T C C&T C C&T C C&T
rank ‘r’ is as follow:
TOR 0.11 −1.24 −4.79 −4.80 −0.08 −1.40 −4.87 −4.86
X
n   TER −1.56 −1.99 −5.22 −5.12 −1.57 −1.99 −5.24 −5.13
λtrace ¼ −T In 1−λj Note: The critical values for ADF and PP tests with constant (c) and with constant
j¼rþ1
& trend (C&T) 1%, 5% and 10% level of significance are − 3.711, − 2.981, − 2.629
and − 4.394, − 3.612, − 3.243 respectively.
Second, λmax maximum number of cointegrating vectors against
r + 1 is presented in the following way:
The error correction model shows the speed of adjustment needed
 
λ max ðr; r þ 1Þ ¼ −T In 1−λj to restore the long run equilibrium following a short run shock. The n
is the coefficient of error correction term in the model that indicates
the speed of adjustment.
The Auto Regressive Distributed Lag method of cointegration de-
veloped by Pesaran and Pesaran (1997), Pesaran and Shin (1999) 4. Estimations and results
Pesaran et al. (2000, 2001) has been used with the help of
unrestricted vector error correction model to investigate the long To check the stationary properties we use Augmented Dickey Full-
run relationship between terrorism and tourism. The ARDL approach er (ADF) and Phillip Perron (PP) unit root tests. Table 4.1 represents
has several advantages upon other cointegraion methods. ARDL ap- the results of stationary tests. First, these tests are applied on the
proach may apply irrespective of whether underlying variables are level of variables then on their first difference.
purely I(0), I(1) or mutually co-integrated. 7ARDL approach has esti- Results of Table 4.1 show that all variables are stationary and inte-
mated better small sample properties. 8In the ARDL procedure the es- grated at first difference. This implies that the series of variables may
timation of results is even possible if the explanatory variables are exhibit a valid long run relationship.
endogenous. 9The ARDL model is developed for estimations as follow: Table 4.2 represents the calculated and critical values of Trace sta-
X
p X
p tistics and maximum eigenvalue statistics of Johansen and Jeuuselius
ΔTORt ¼ ψ0 þ ψ1 ΔTORt−1 þ ψ2 ΔTERt−1 þ γ1 TORt−1 (1990) cointegration method. Results indicate the rejection of null
i¼1 i¼1 hypothesis of no cointegration in model at significance level of 5 per-
þ γ2 TERt−1 þ μ t cent in Pakistan in favor of alternative hypothesis that is the existence
of one or more cointegrating vectors. The findings confirm the exis-
Where ψ0 is constant, μt is white noise error term, TOR is tourism tence of long run relationship between terrorism and tourism in
receipts and TER is terrorism activities. The error correction dynamics Pakistan.
is denoted by a summation sign while the second part of the equation Autoregressive distributed lag method for cointegration is also
corresponds to long run relationship. Schwarz Bayesian criteria (SBC) used to estimate the long run relationship between terrorism and
has been used to identify the optimum lag of model and each series. tourism. The first step is to determine the optimal lag length of the
In the ARDL model we first estimate the F-statistics value by using model. The order of optimal lag length is decided by using the
the appropriate ARDL models. Secondly, the Wald (F-statistics) test Schwarz Bayesian Criterion. 10Table 4.3 shows the results of ARDL
is used to investigate the long run relationship among the series. cointegration method.
The null hypothesis of the above cointegration equation is that The ARDL results suggest the rejection of null hypothesis of no
(H0 = γ1 = γ2 = 0). The null hypothesis of no cointegration is cointegration in model because the value of the F-statistics is greater
rejected if the calculated F-test statistics exceeds the upper critical than the upper bound critical value at 5% level of significance in favor
bound (UCB) value. The results are said to be inconclusive if the of alternative hypothesis that the valid long run relationship exists
F-test statistics falls between the upper and lower critical bound. between terrorism and tourism in Pakistan.
Lastly, the null hypothesis of no cointegration is accepted if the Now we estimate the lag length order of all variables through
F-statistics is below the lower critical bound. If long run relationship unrestricted vector auto regression method. The decision criterion is
between terrorism and tourism is found then we estimate the long based on minimum value of Schwarz Bayesian criterion.
run coefficients. The following model will be used to estimate the Table 4.4 represents the results of lag length order of all variables. Re-
long run coefficients: sults of Schwarz Bayesian criterion indicate that terrorism and tourism
both should be included in the model at 1st lag. After having the valid
X
p X
p
TORt ¼ ζ 0 þ ζ 1 TORt1 þ ζ 2 TERt1 þ μ t evidence of long run relationship between terrorism and tourism, now
i¼1 i¼1 we applied the ARDL method to estimate the long run and short run co-
efficients. The model for the long run coefficients is as follow:
If we find evidence of long run relationship between terrorism and
tourism then we estimate the short- run coefficients by employing X
p X
p X
p
TORt ¼ ζ 0 þ ζ 1 TORt−1 þ ζ 2 TERt þ ζ 3 TERt−1 þ μ t
the following model: i¼1 i¼1 i¼1

X
p X
p
ΔTORt ¼ φ0 þ φ1 ΔTORt−1 þ φ2 ΔTERt−1 þ nEC −1 þ μ t Table 4.5 shows the results of long run ARDL estimations. Results
i¼1 i¼1 indicate the significant negative impact of terrorism on tourism in

10
The SBC is slightly superior to the AIC (Pesaran and Shin, 1999). Besides, SBC is par-
7
Pesaran and Shin (1999). simonious as it uses minimum acceptable lag while selecting the lag length and avoid
8
Haug (2002). unnecessary loss of degrees of freedom. Therefore, SBC criterion has been used, as a cri-
9
Pesaran and Shin (1999) and Pesaran et al. (2001). terion for the optimal lag selection, in cointegration estimation.
68 S.A. Raza, S.T. Jawaid / Economic Modelling 33 (2013) 65–70

Table 4.2 Table 4.5

J. J. cointegration test results of tourism model. Long run results using ARDL approach.
Source: Authors' estimation. Source: Authors' estimation.

Null hypothesis Trace 5% critical Max. eigenvalue 5% critical Variables Coeff. t-stats Prob.
no. of CS(s) statistics values statistics values
C 0.607 2.243 0.036
None * 31.430 25.872 23.789 19.387 TOR(-1)) 1.060 19.419 0.000
At most 1 7.641 12.518 7.641 12.518 TER −0.383 −2.200 0.039
At most 2 4.655 12.321 4.107 11.225 TER(-1)) −0.072 −1.164 0.259
Adj. R2 0.953
D.W. stats 2.423
F-stats (Prob.) 215.149 (0.000)

Table 4.3
Lag length selection & bound testing for cointegration.
Source: Authors' estimation. 6. Dynamic ordinary least square
Lags order AIC HQ SBC F-test statistics
The robustness of the relationship between dependent variable
0 5.669 5.697 5.765 9.321⁎
and explanatory variables in the long run is tested through dynamic
1 1.919⁎ 2.005⁎ 2.207⁎ 7.564⁎
2 2.15 2.294 2.631 5.213⁎
ordinary least square (DOLS) technique developed by Stock and
Watson (1993). This method involves estimating the dependent
⁎ 5% level of significance.
variable on explanatory variable in levels, leads and lags. This method
resolves the issues of small sample bias, endogeneity and serial corre-
lation problems by adding the leads and lags of explanatory variable
Pakistan. The coefficient of terrorism is showing that the 1% increase
(Stock and Watson, 1993). The equation of DOLS model is given
in terrorism activities causes the decreases in the tourism receipt by
below:
0.38%. It is concluded that terrorism activities are worsening the tour-
ism growth in Pakistan. The following model is used to check the p X
X k
short run relationship among the considered variables with the dif- Y t ¼ φ0 þ φ1 X t þ θji ΔX i;t−j þ εt
ferent lag length. j¼−p i¼1

X
p X
p X
p Where Yt is the dependent variable, Xt is the vector of explanatory
ΔTORt ¼ φ0 þ φ1 ΔTORt−1 þ φ2 ΔTERt þ φ3 ΔTERt−1 þ nEC t−1 þ μ t
variables and Δ is the lag operator. Table 6.1 represents the results
i¼1 i¼1 i¼1
of dynamic ordinary least square of terrorism and tourism model. It
is confirmed from the results that the coefficients of terrorism re-
Table 4.6 represents the short run relationship between terrorism
main with the same sign and significance after taking the lead in
activities and tourism receipt. Results indicate that the lagged error
the model. Consequently it can be concluded that the relationship
correction term for the estimated tourism receipt equation is both
of terrorism and tourism will remain the same in the future up to
negative and statistically significant. This confirms a valid short run
lead 2.
relationship between terrorism and tourism in Pakistan. The coeffi-
cient of error term − 0.58 suggests that about 58% of disequilibrium
7. Causality analysis
is corrected in the current year. Results indicate the negative and sig-
nificant short run relationship between terrorism activities and tour-
In this section three different causality analyses have been performed
ism receipt.
to check the causal relationship between terrorism and tourism. At first,
by using Granger causality11analysis, secondly by using the Toda and
5. Rolling window method
Yamamoto12procedure and lastly by using the variance decomposition
method.13
The stability of coefficients of the model in the sample size is eval-
uated by using the rolling window estimation method. Fig. 5.1 repre-
7.1. Granger causality analysis
sents the results of coefficients of terrorism.
Two standard deviation bands show the upper and lower
The direction of causality between dependent and independent
bounds. Results indicate the considered variable terrorism having
variables is analyzed by Granger (1969) causality test. We determine
negative coefficients throughout the sample period. The result of
the causality analysis of our tourism model on lag one. Jones (1989)
Fig. 5.1 shows that the coefficient of terrorism increases from
favors the ad hoc selection method for lag length in Granger causality
1990 to 1996 then it remains steady from 1997 to 2003. Further
test over some of the other statistical methods to determine optimal
drastic increase in coefficient of terrorism has been seen from
lag. The equation of Granger causality model is given below:
2005 to 2010.
X
t X
t
Y¼ α i X t−i þ βi Y t−i þ μ
i¼1 i¼1
Table 4.4 X
t X
t

Lags defined through VAR of variables.

X¼ λi X t−i þ δi Y t−i þ v
i¼1 i¼1
Source: Authors' estimation.

Lag 0 1 2 Selected lags

11
SBC SBC SBC SBC See, Granger (1969).
12
See, Toda and Yamamoto (1995)
TOR 2.324 −0.388⁎ −0.124 1 13
The Variance Decomposition method is estimated through VAR framework, it
TER 3.880 2.549⁎ 2.767 1
shows the proportion contribution in one variable caused by the shocks in other vari-
⁎ Indicate minimum AIC and SBC values. ables, Pesaran and Shin (1998).
 S.A. Raza, S.T. Jawaid / Economic Modelling 33 (2013) 65–70 69

Table 4.6 Table 7.1.1

Short run results using ARDL approach. Results of Granger causality test.
Source: Authors' estimation. Source: Authors' estimations.

Variables Coeff. t-stats Prob. Variables F-statistic Probability

C −0.018 −0.311 0.760 TOR does not Granger cause TER 1.644 0.212
D(TOR(-1)) 1.399 3.502 0.002 TER does not Granger cause TOR 3.528 0.072
D(TER) −0.014 −2.303 0.034
Note: The lag length is 1.
D(TER(-1)) −0.286 −1.306 0.209
ECM(-1) −0.577 −2.391 0.029
Adj. R2 0.848
D.W. stats 2.093 when SUR models are used in the estimation. So, the model can be
F-stats (Prob.) 60.549 (0.000) specified as follows:

X
kþd X
kþd
Y t ¼ α1 þ γ 1i Y t−i þ γ 2i X t−i þ εyt
i¼1 t−i
X
kþd X
kþd
Xt ¼ α2 þ δ1i Y t−i þ δ2i X t−i þ εxt
i¼1 t−i

Where k is the optimal lag order, d is the maximum order of inte-

gration of the series in the system, and εyt and εxt are error terms that
are assumed to be white noise. Usual Wald tests are then applied to
the first k coefficient matrices using the standard χ 2 – statistics. The
main hypothesis can be drawn as follows:
(a) Xt “Granger causes” Yt if γ2i ≠ 0
(b) Yt “Granger causes” Xt if δ2i ≠ 0
The results of Toda and Yamamoto (1995) based causality test are
reported in Table 7.2.1. Results indicate that unidirectional causal re-
lationship is found between terrorism and tourism, and causality runs
from terrorism to tourism.
Fig. 5.1. Coefficient of TER and its two S.E. bands based on rolling OLS (Dependent
variable: TOR).
7.3. Variance decomposition analysis

Generalized forecast error variance decomposition method under

It is assumed that μ and v are uncorrelated. There are two variables
and dealt with bilateral causality. The above equation states that Y is
related to its lag values and X is related to its lag values.
The results of Granger causality test are reported in Table 7.1.1.
The results show that unidirectional causal relationship found in
runs from terrorism to tourism.
7.2. Toda and Yamamoto modified Wald test causality analysis
The direction of causality between dependent and independent
variables is analyzed by using the causality test based on Toda
and Yamamoto (1995) procedure. This test uses a modified Wald
(MWALD) test which can be applied irrespective of whether the un-
derlying variables are purely I(0), I(1) or mutually co-integrated.
Toda and Yamamoto (1995) augmented Granger causality test uses
the Seemingly Unrelated Regression (SUR) technique through esti-
mating a two-equation system. The Wald test improves efficiency
vector autoregressive (VAR) system has been used to analyze the
strength of the causal relationship of terrorism and tourism. The
variance decomposition method provides the magnitude of the
predicted error variance for a series accounted for by innovations
from each of the independent variable over different time period.
Wong (2010), Hye (2012) and Shahbaz et al. (2012) have used this
approach to find causal relationship among considered variables.
Table 7.3.1 represents the results of variance decomposition analysis.
Results of Table 7.3.1 show that in the first round the change in tour-
ism explains 98.84% by its own innovations and 1.16% by the innovation
of terrorism, in the second period 96.57% explained by its own innova-
tion and 3.43% by terrorism. In period five the shocks in tourism explain
93.75% by own innovation and 6.25% by terrorism innovation. In the
tenth period the shocks in terrorism explain 66.11% by own shocks,
while, 33.89% is explained by innovations of terrorism. The shocks in
terrorism explain 0.00%, 10.32% and 1.72% by innovation of tourism in
periods 1, 5 and 10 respectively. These findings suggest that unidirec-
tional causal relationship is found between terrorism and tourism, and
causality runs from terrorism to tourism.

Table 6.1
Table 7.2.1
Results of dynamic ordinary least square DOLS.
Results of Toda and Yamamoto causality test.
Source: Authors' estimation.
Source: Authors' estimations.
Variables Coeff. t-stats Prob.
Dependent variable Modified Wald – Statistics
C −0.596 −2.972 0.007
TOR TER
TOR(-1) 0.880 18.101 0.000
TER(1) −0.349 −2.083 0.048 TOR – 6.676
TER(2) −0.232 −3.226 0.004 0.018
Adj. R2 0.952 TER 1.853 –
D.W. stats 2.215 0.189
F-stats (Prob.) 250.611 (0.000)
Note: The lag length for TOR and TER is 1. As per Schwarz Bayesian criteria.
70 S.A. Raza, S.T. Jawaid / Economic Modelling 33 (2013) 65–70

Table 7.3.1 Dicky, D.A., Fuller, W.A., 1979. Distribution of the estimators for autoregressive time se-
Results of variance decomposition approach. ries with a unit root. Journal of the American Statistical Association 74, 427–431.
Source: Authors' estimation. Drakos, K., Kutan, A.M., 2001. Regional Effects of Terrorism on Tourism: Evidence from
Three Mediterranean Countries. Working Paper B 26. Center for European Integra-
Period TOR TER tion Studies, Rheinische.
Dritsakis, N., 2004. Tourism as a long-run economic growth factor: an empirical inves-
Variance decomposition of TOR tigation for Greece using causality analysis. Tourism Economics 10 (3), 305–316.
1 98.838 1.162 Enders, W., Sandler, T., 1991. Causality between transnational terrorism and tourism:
2 96.574 3.426 the case of Spain. Terrorism 14 (1), 49–58.
3 97.045 2.955 Enders, W., Sandler, T., Parise, G.F., 1992. An econometric analysis of the impact of ter-
4 96.360 3.640 rorism on tourism. Kyklos 45, 531–554.
5 93.751 6.249 Floyd, M.F., Gibson, H., Thapa, B., Pennington-Gray, L., 2003. The effect of risk percep-
6 89.311 10.689 tions on intentions to travel in the aftermath of September 11. Journal of Travel
7 83.642 16.358 and Tourism Marketing 15 (2/3), 19–38.
Ghali, M.A., 1976. Tourism and economic growth: an empirical study. Economic Develop-
8 77.507 22.493
ment and Cultural Change 24 (3), 527–538.
9 71.537 28.463
Gu, Z., Martin, T.L., 1992. Terrorism, seasonality, and international air tourist arrivals in Cen-
10 66.111 33.889
tral Florida: an empirical analysis. Journal of Travel and Tourism Marketing 1, 3–15.
Gunduz, I., Hatemi, J.A., 2005. Is the tourism-led growth hypothesis valid for Turkey.
Variance decomposition of TER Applied Economics Letters 12 (8), 499–504.
1 0.000 100.000 Hartz, C., 1989. Business Executives as International Terrorist Targets. In: Buckwalter,
2 13.213 86.787 J.R. (Ed.), International Terrorism: The Decade Ahead. The University of Illinois,
3 12.398 87.602 Office of International Criminal Justice, Chicago, pp. 21–28.
4 11.443 88.557 Haug, A., 2002. Temporal aggregation and the power of co-integration tests: a Monte
5 10.322 89.678 Carlo study. Oxford Bulletin of Economics and Statistics 64, 399–412.
6 9.005 90.995 Henderson, J.C., Shufen, C., Huifen, L., Xiang, L.L., 2010. Tourism and terrorism: a hotel
7 7.467 92.533 industry perspective. Journal of Tourism, Hospitality and Culinary Arts 2 (1).
8 5.693 94.307 Hye, Q.A.M., 2012. Long term effect of trade openness on economic growth in case of
Pakistan. Quality and Quantity 46 (4), 1137–1149.
9 3.716 96.284
Hye, Q.M.A., Khan, R.E.A., 2012. Tourism-led growth hypothesis: a case study of
10 1.722 98.278
Pakistan. Asia Pacific Journal of Tourism Research. http://dx.doi.org/10.1080/
10941665.2012.658412.
Katircioglu, S.T., 2009. Revisiting the tourism-led growth hypothesis for Turkey using
8. Conclusion and recommendations bonds test and Johansen approach for cointegration. Tourism Management 30
(1), 17–20.
Khalil, S., Mehmood, K.K., Waliullah, 2007. Role of tourism in economic growth: empirical
This study investigates the impact of terrorism activities on tour- evidence from Pakistan economy. The Pakistan Development Review 46 (4), 985–995.
ism in Pakistan by using the annual time series data from the period Kingsbury, P.T., Brunn, S.D., 2004. Freud, tourism and terror: traversing the fantasies of
of 1980 to 2010. Johansen and Jeuuselius and ARDL bound testing post September 11 travel magazines. Journal of Tourism and Travel Marketing 15
(2–3), 39–61.
cointegration approach confirms the valid long run relationship be- Kreishan, F.M.M., 2010. Tourism and economic growth: the case of Jordan. European
tween terrorism and tourism. Results indicate the significant negative Journal of Social Sciences 15 (2), 229–234.
impact of terrorism on tourism in the long run as well as in the short Mansfeld, Y., 1996. Wars, Tourism and the “Middle East” Factor. In: Pizam, A.,
Mansfeld, Y. (Eds.), Tourism, Crime and International Security Issues. Wiley,
run. Results of rolling window estimation method indicate that ter- New York, pp. 265–278.
rorism is having negative coefficients throughout the sample period. Mazursky, D., 1989. Past experience and future tourism decision. Annals of Tourism
Results of dynamic ordinary least square (DOLS) suggest that the Research 16, 333–344.
Mckinnon, R., 1964. Foreign exchange constraints in economic development and effi-
results will remain the same in the future up to lead 2. Results of cient aid allocation. The Economic Journal 74 (294), 388–409.
Granger causality test, Toda and Yamamoto Modified Wald causality Pesaran, M.H., Pesaran, B., 1997. Working with Microfit 4.0: Interactive Econometric
test and variance decomposition test confirm the unidirectional caus- Analysis. Oxford University Press, Oxford.
Pesaran, M.H., Shin, Y., 1998. Generalized impulse response analysis in linear multivar-
al relationship between terrorism and tourism, causality runs from iate models. Economic Letters 58 (1), 17–29.
terrorism to tourism. Hye (2012) confirms that tourism led growth Pesaran, M.H., Shin, Y., 1999. An autoregressive distributed lag modeling approach to
hypothesis for Pakistan. It is suggested that the government should co-integration analysis. In: Strom, S. (Ed.), Econometrics and Economic Theory in
the 20th Century: The Ragnar Frisch Centennial Symposium. Cambridge University
play a significant role to reduce terrorism activities and should be de-
Press, Cambridge.
creased in the country to boost tourism activities that lead to increase Pesaran, M.H., Shin, Y., Smith, R.J., 2000. Structural analysis of vector error correction
income from tourism in Pakistan. models with exogenous I(1) variables. Journal of Econometrics 97, 293–343.
Pesaran, M.H., Shin, Y., Smith, R.J., 2001. Bounds testing approaches to the analysis of
level relationships. Journal of Applied Economics 16, 289–326.
References Phillips, P.B., Perron, P., 1988. Testing for a unit root in time series regression.
Biometrika 75, 335–346.
Alsarayreh, M.N., Jawabreh, O.A.A., Helalat, S.S., 2010. The influence of terrorism on the Seddighi, M.W., Nuttall, A.L., Theocharous, H.R., 2001. Does cultural background of
international tourism activities. European Journal of Social Sciences 13 (1). tourists influence the destination choice? An empirical study with special refer-
Balaguer, J., Cantavella-Jorda, M., 2002. Tourism as a long-run economic growth factor: ence to political instability. Tourism Management 22 (2), 181–191.
the Spanish case. Applied Economics 34 (7), 877–884. Shahbaz, M., Islam, F., Aamir, N., 2012. Is devaluation contractionary? Empirical evi-
Brida, J.G., Barquet, A., Risso, W.A., 2010. Causality between economic growth and tour- dence for Pakistan. Economic Change and Restructuring 45.
ism expansion: empirical evidence from Trentino-Alto Adige. TOURISMOS: An In- Skerritt, D., Huybers, T., 2005. The effect of international tourism on economic develop-
ternational Multidisciplinary Journal of Tourism 5 (2), 87–98. ment: an empirical analysis. Asia Pacific Journal of Tourism Research 10 (1), 23–43.
Chen, R.J.C., Noriega, P., 2004. The impacts of terrorism: perceptions of faculty and stu- Sonmez, S., 1998. Tourism, terrorism, and political instability. Annals of Tourism
dents on safety and security in tourism. Journal of Tourism and Travel Marketing Research 25 (2), 416–456.
15 (2–3), 81–97. Sonmez, S., Graefe, S., 1998. Influence of terrorism risk on foreign tourism decisions.
Cook, R.L., McCleary, K.W., 1983. Redefining vacation distances in consumer minds. Annals of Tourism Research 25 (1), 112–144.
Journal of Travel Research 22 (2), 31–34. Stock, J., Watson, M.W., 1993. A simple estimator of cointegrating vector in higher
Cortes-Jemenez, I., Pulina, M., 2006. A further step into the ELGH and TLGH for Spain order integrated systems. Econometrica 6 (4), 783–820.
and Italy Sustainable Development Series. Fondazione Eni Enrico Mattei (FEEM) Toda, H.Y., Yamamoto, T., 1995. Statistical Inference in Vector Autoregressions with
(Working Paper No. 2006/118). Possibly Integrated Processes. Journal of Econometrics 66, 225–250.
D'Amore, L.J., Anuza, T.E., 1986. International terrorism: implications and challenge for
global tourism. Business Quarterly 20–29 (November).