City University Research Journal

Volume 06 Number 02 July 2016 PP 311-320

MODELING FOR TOURISM THROUGH ECONOMIC

INDICATORS OF PAKISTAN
Muhammad Kamran*, Amjad Alib**, Alamgir*** and Dost Muhammad Khan****

ABSTRACT
The purpose of this study is to find an appropriate model for tourism data of Pakistan.
Log tourism has been used as a dependent variable while Financial Exchange Rate,
Gross Domestic Product, Employment, Visitor Exports and Investment are taken as
independent variables. Data has been collected from secondary source Tourism
Department of Pakistan and Economic survey of Pakistan 2014, for the period of 2004-
2014. For final model selection, the model selection techniques; all possible
regression, forward selection procedure and backward elimination procedure have
been used. The final selected model can be used for future prediction of tourism.
According to our findings two independent variables; employment and GDP
significantly contribute to tourism. This model can be used for short run predication,
because our data consist of very few years, which is the main hurdle that this model
cannot be applied for long run prediction.

Keywords: Tourism, all possible regression, forward selection and backward

selection.

INTRODUCTION
Since the beginning of recorded history, humans are driven to travel for the purpose of
business (trade), exploration, food, water and safety. In the intervening time, the
purpose of trading has increased to entertainment and delight. Traveling also increased
due to advanced technologies, which facilitate the people to move easily from one place
to another. In the past, travelers used to walk or ride on tamed animals. With the
advancement in different field specifically, the invention of the wheel, planes and sea
ships provided new mode of traveling. There is always a gradual increase with
advancement, like chain of roads increased, governments stabilized and interest in
traveling increased for sightseeing, education and religious purposes.
Tourism not only provides the inner pleasure to the tourists but also gives strength to the
nation. The impact of tourism has many positive effects on the economy like, increased
employment ratio, better currency exchange rate, diverse education, cultural exchange
and improving hoteling etc. According to UNWTO World Tourism in 2009 the number
of world tourist is 880 million, whereas, it is estimated that it to be1.6 billion by 2020. In
the last few decades, the tourism industry in Pakistan was badly affected due to unstable
condition of Pakistan (Getz, 2009; Wall & Mathieson, 2006). Tourism and economic
growth has long history. There are too many consequences of economic growth on
tourism, and there are many developed techniques in order to estimate the economic
* Department of Management Science, City University of Sc. & IT, Peshawar
**Department of Statistics, Islamia College Peshawar, KPK, Pakistan
***Department of Statistics, University of Peshawar, KPK, Pakistan
****Department of Statistics, Abdul Wali Khan University Mardan, KPK, Pakistan.
C 2016 CURJ, CUSIT 311
Muhammad Kamran et al.

impact on tourism (Andersson, Armbrecht, & Lundberg, 2008).

In Asia, tourism is considered to be more important factor for economic growth.
Through tourism, increased employment, per capita income, foreign exchange rate,
trade in commodities is obvious. This foreseen high economic growth attracts the tourist
for business tours and exchange programs. Tourism is an important sector for
authorities, consultants and tourism developers for measuring the economic
developments and its impact on tourism.
Generally, tourism is an assortment of industry, services and activities that distributes
travel experience. According to the World Tourism Organization (WTO) presently
tourism is the largest industry in the world, with annual revenues of over $3 trillion and it
also provides more than 6 million jobs only in US. This industry contributes more than
10% in the world GDP, also this is the source for the income of generation, job
developments and so many other advantages can be obtained for a state. According to
the world tourism organisation, global incomes from tourism in 2002 were high up to $
474 billion which is likely to increase 1.5 trillion by 2011 (WTTC, 2010).
Unfortunately, due to political instability, war in terror, floods, bomb attacks or suicide
attacks, tourism is neglected in Pakistan and the most beautiful valleys are affected in
the war of terrorism. Recently created Pakistan has recovered all the deficiencies. The
elevation of tourism in the country stared in the 1960 in the shape of a small department
of tourism which was created in 1960 as a cell in the Ministry of Railway. And afterward
the Department of Tourism was transferred from one Minister to another. Now the
Tourism in Pakistan department has gained the status of tourism Department under the
Ministry of Tourism (Khan et al. 2011).

Objectives of the study:

The main objectives of the present study is to investigate the best model for tourism in
Pakistan selection through model selection criterion, as well as to find the best sub set of
predictors and identified that the final model can be used for prediction or not.

LITERATURE REVIEWS
Several researchers have discussed the importance of tourism industry and its
contribution in economic growth all over the world, impact on cultural, social,
environmental growth (Deery & Jago, 2010; Dogan, 1989; Gösslingt & Hall, 2008;
Lankford & Howard, 1994; Pizam, 1978; Turner & Ash, 1975). According to Kulendran
and Wilson (2002) are studied the empirical analysis of China and Australia, they
observed form their study a strong relationship between international tourism and
economic growth. Further they concluded that through international and domestic
tourist not only caused increase economic growth but this also increased the opportunity
of employment and trade etc.
Similarly, the excellence in economic growth attracts people for business tours and they
openly invest money in the developed industry as a result labor force and employment
also might be increased (Bahmani- Oskooee & Alse, 1993; Chow, 1987; Marin, 1992).
According to UN World Tourism Barometer (2009) total 924 million visitors were seen
worldwide by the end of 2008 which was 2% high compared to 2007. This results show
that the number of visitor's increases year by year but now in the present conditions
312 C 2016 CURJ, CUSIT
 Modeling For Tourism Through Economic Indicators...

people liked travelling more and this ratio is expected very high in the recent years.
Nissan et al (2010) discussed those determinants which affecting the tourism sector.
Also found the feedback between income and tourism. The finding of the study suggest
that tourism not only create money for financial firms but also motivates local firms and
creates new jobs opportunities that increase the labor force. Variables that have
important effects on tourism sector like entrepreneurship and prices of different
commodities. Results suggest that tourism has a positive impact on economic growth,
entrepreneurship, price and income.
Lord & Brain (2009) in their study focused to evaluate those opportunity which promote
the high tourism culture with respect to visitor perspective. The recommendation in that
study suggests that high quality, distinctiveness and infrastructure are the major benefits
to maximizing the tourism industry.
Sequeria & Nunes (2010) focused on the tourism specialized countries. They described
in their study the economic growth of these countries are high than those countries who
are not focused on tourism. In the proposed study researcher used panel data for
studying the relationship among economic growth and tourism. Further they conclude
that tourism has positive effect on economic growth but tourism contribution is least in
small and poor countries because they have insufficient opportunities to develop this
industry.
Po & Huang (2008) they used cross sectional data for 88 countries over the year 1995 to
2005. They used nonlinear regression between tourism sector and economic growth.
Analysis indicates that data for these countries should be distributed into different
groups according to their geographical condition. Further they conclude that tourism
has positive significant effect on economic growth.

Figure 1: Theoretical Framework

HYPOTHESES
H1: Financial exchange rate has significant impact on log tourism
H2: Gross Domestic Product has significant impact on log tourism
H3: Employment has significant impact on log tourism
H4: Investment has significant impact on log tourism
H5: Visitor export has significant impact on log tourism
C 2016 CURJ, CUSIT 313
Muhammad Kamran et al.

MATERIALS AND METHODS

In the present study, annual data has been used from 2004 to 2014 which has been
collected from the secondary source like, Tourism department of Pakistan and
Economic survey of Pakistan-2014. Log tourism (LnTUR) used as response / dependent
variable and five independent variables financial exchange rate (FER), gross domestic
product (GDP), employment (EMP), visitor export (VEXP) and investment (INV) have
been included.
Augmented Dickey Fuller (ADF) test has been used for testing the stationarity of the
variables; all the original variables are stationary at level. The test for stationarity is
important for long term relation and for prediction purpose. Furthermore all possible
regression, forward selection procedure and backward elimination criteria were used
for model selection on the basis of R-Square, Adjusted R- Square and Mean Square
Error (MSE). Data has been analyzed in Statistical Software for Social Sciences (SPSS).

Analysis and Discussion:

According to the objectives of present study researchers use various statistical tools to
carry out their analysis. In the present study we used three different statistical tools, to
identify best model out of all possible regression models and also to find the best sub set
of predictors. For this purpose, we used all possible regression, backward elimination
and forward selection procedures.

All Possible Regression:

All possible regression means to identify the best sub-set of predictor's regression model
out of all possible regression models. In the present study used five explanatory
variables which established 25-1=31 models. For best model selection criterion were
used such as coefficient of determination (R-Square), Adjusted R-Square and Mean
square Error (MSE). In Table-1 have been used five single variable regression models in
which just only one regression model with (X3) independent variable have been
selected, having R-square is 76.2% indicates that 76.2% of the variability explained in
dependent variable by its linear relationship with the independent variable (X3) and
remaining is unexplained due to other factors. Also in this model Adjusted R-square
value is maximum 73.5% and the mean square error is too minimum out of all single
variable models.
Now for the two independent variables we have ten possible regression models in which
(X2,X3) have been selected having R-square value is 84.4%, indicates that 84.4% of the
variability in dependent variable explained by its linear relationship with the
independent variable (X2,X3) and remaining due to other factors, this also indicates
that (X2) is an important variables. The value of Adj R-square 80.5% is also increased as
compared to single variable model. Further if we observed the value of mean square
error remain same as single variable model. Now for the three independent variable we
have also ten models in which (X2,X3,X4) have been selected, having high R-square
88.7%, indicates that 88.7% of the variability in dependent variable explained by its
linear relationship with the independent variable (X2,X3,X4) and remaining is
unexplained due to other factors. Also the value of Adj R-Square value 81.2% increased
as compared to two variable model, indicate that (X4) has significant effect on
314 C 2016 CURJ, CUSIT
 Modeling For Tourism Through Economic Indicators...

dependent variable when appear with (X2,X3), further the value of MSE has to
decreased as compared to previous model, so three variable model is most suitable as
compared to previous models.
In four independent variables we have five regression models in which (X1,X2,X3,X4)
have been selected, having high R-square 89.1%, indicates that 89.1% of the variability
explained in dependent variable due to its linear relationship with the independent
variable (X1,X2,X3,X4) and remaining due to other factors. The value of Adj R-square
81.9% were decreased as compared to three variable model by adding fourth variable
(X1), but the MSE remain the same like the previous model.
In full model in which all independent variables are included (X1,X2,X3,X4,X5) its R-
square values is not much different than four variable model. But if we observed that
there is no major difference between three and four variable selected model, but in this
case the most appropriate model is two variable models because the R-square value is
higher than 80% and the second advantage is it deal easily as compare to larger models.
At the end from the whole analysis conclude that among these 31 models only the two
variables model (X2, X3) is important model and more appropriate and used for
predication and other necessary calculation. In simple word we can say that the most
appropriate model is log-linear regression

So, this model shows that gross domestic product (X2) and employment (X3) have more
contribution in tourism; more simply tourism is highly effect through these two
variables. Now we will check the model selection and the best sub set of independent
variables by another criterion known as backward and forward selection criterion. The
analysis for these two criterions follows as:

Table-1: Model Selection and Selection of Significant Variables by Coefficient of

Determination, Adj R-square and Mean square error Criteria.

Backward Selection Criterion:

This selection criterion, starts from the full model and then one by one variable is
removed from the model. The process of eliminating the variable from the model will be
continuing until the best model and the sub set of best predictors are achieved. In this
procedure we start from the full model removed all those variables which are
insignificant, the process will be continued until we get the significant variables.
Now, observed model summary Table-2 in the model (1) the value of R-square is
maximum, but we dropped one by one independent variables the value of R-square will

C 2016 CURJ, CUSIT 315

Muhammad Kamran et al.

be decreased. If one can compared the value of R-square of model (1) and model (4)
there is no major difference among these two models clearly shows the values which are
0.892 and 0.844 respectively. In the last model (4) we included two variables which are
(X2, X3) and the R-square value also greater than 80%.Another side, if we compared
Adj R-square values, for the first model is too minimum among all four models. This
indicates that in the full model some insignificant variables are present by which the Adj
R-square is smaller. But in the last model the value of Adj R-square is high compared to
the first model and the standard error also minimum. So again according to this criterion
two variable models is best one.

Table-2: Model Summary

Table-3: ANOVA

Table-4: Coefficients

In Table-3 clearly observed the residual sum of squares, which is least in the suggested
model, is 0.012, and the value of F-statistic is 21.642, shows that the overall model is
highly significant. This criterion also suggests the two variable model when we
removed variables from the model the performance of the model also increased. In the
(Table -4), only discussed the finalized model coefficients and the significance of
variables. In first stage we added all explanatory variables and then one by one
316 C 2016 CURJ, CUSIT
 Modeling For Tourism Through Economic Indicators...

insignificant variables will be eliminated until the significant variables were left. At the
end two variables are left in the model, which is our final model, X2 has negative
significant and X3 has positive significant impact on dependent variable.

Forward Selection Criterion

This criterion is the reciprocal of backward selection criterion, in this criterion, we
taking start from constant with no predictors at the first stage, and the second stage we
added one highly significant variable in the model, in this way we obtained one best
variable in the model. In the second stage we obtained the second significant variable.
The process will continue until all insignificant variables are left. In Table-5 R-square,
with X3 alone (step 1), 76.2% of the variance was accounted for both X3 and X2 (step
2), 84.4% of the variance was accounted for.

Table-5: Model Summary

Table-6: ANOVA

Table-7: Coefficients

This Table-7 now gives two F-test one for each step of the procedure. Both steps had
overall significant results. Table-8 gives beta coefficients so that we can construct the
regression equation. Betas shows change, in dependent variable, depending on which
predictors are included in the model. These are the weights that we want, , for an
C 2016 CURJ, CUSIT 317
Muhammad Kamran et al.

an equation that includes just X3 and X2 (the two best predictors). The equation would
be;

All the three different selection procedure gives the same results. So we can conclude
from the whole analysis the two variable model “employment” and “gross domestic
product” is the best one. These two variables can be used to predict the tourism.

Results for Research Hypothesis:

Alternative hypothesis Accepted Rejected
H1 × 

H2  ×

H3  ×

H4 × 

H5 × 

According to selected model, our first alternative is rejected, because financial

exchange rate has no significant effect on tourism. The second alternative hypothesis is
accepted GDP has negative significant effect on tourism. Third alternative hypothesis
also accepted employment has positive significant effect on tourism and the fourth and
fifth hypothesis has been rejected.

CONCLUSION
In present study, we used log tourism as dependent variable financial exchange rate,
gross domestic product, employment, visitor exports and investment as independent
variables. Ten year data has been used, collected from the secondary source,
Three different statistical tools were applied for data analysis, all possible regression,
backward elimination and forward selection procedures, to achieve the objectives and
research hypothesis. From three different results we obtained the same results. In all
possible regression, total number of possible regression equations are 32, in which one
is constant model with no predictors, five are one variables models, ten are with two
variables models, ten are with three variables models, five are four variables models and
the last model is full variable model mean with five predictors.
This method suggest us two variable model is more efficient out of all models, which has
maximum R-square, Adj R-squared and minimum MSE, named of these two variables
are employment (X3) and gross domestic product (X2). Backward elimination and
forward selection procedure also suggest discussed model. According to selected
model, our first alternative is rejected, because financial exchange rate has no
significant effect on tourism. The second alternative hypothesis is accepted GDP has
negative significant effect on tourism. Third alternative hypothesis also accepted
employment has positive significant effect on tourism and the fourth and fifth
hypothesis also rejected.
318 C 2016 CURJ, CUSIT
 Modeling For Tourism Through Economic Indicators...

The finalized model can be used for short run predication, because our data consist of
very few years, which are the main cause this model cannot be used for long run
prediction.

REFERENCES
Andersson, T. D., Armbrecht, J., & Lundberg, E. (2008). Impact of Mega-Events on the
Economy. Asian Bus Manage, 7(2), 163-179.

Bahmani-Oskooee, M., & Alse, J. (1993). Export growth and economic growth: An
application of Cointegration and error correction modeling. Journal of
Developing Areas, 27, 535–542.

Chow, P. C. Y. (1987). `Causality between export growth and industrial development:

Empirical evidence from the NICs. Journal of Development Economics, 26,
55–63.

Deery, M., & Jago, L. (2010). Social impacts of events and the role of anti-social
behavior. International Journal of Event and Festival Management, 1(1), 8
-28.

Dogan, H. Z. (1989). Forms of adjustment: Socio-cultural impacts of tourism. Annals of

Tourism Research, 16(2), 216-236.

Getz, D. (2009). Policy for sustainable and responsible festivals and events:
institutionalization of a new paradigm. Journal of Policy Research in Tourism,
Leisure and Events, 1(1), 61 - 78.

Gössling, S., & Hall, M. C. (2008). Swedish Tourism and Climate Change Mitigation:
An Emerging Conflict? Scandinavian Journal of Hospitality and Tourism,
8(2), 141 - 158.

Khan, J. A., Syed, S., Zaman, K., & Ahmad, F. (2011).Students' Perception toward
Social, Economic and Environmental Impacts of Tourism in Chiral–Pakistan.
Journal of International Academic Research, 11(1).

Kulendran, N., & Wilson, K. (2000). Is there a relationship between international trade
and international travel? Applied Economics, 32, 1001–1009.

Lankford, S. V., & Howard, D. R. (1994).Developing a tourism impact attitude scale.

Annals of Tourism Research, 21(1), 121-139.

Lord Cultural Resources & Brain Trust. (2009). "Ontario cultural and heritage tourism
product research paper.”

Marin, D. (1992). Is the export-led hypothesis valid for industrialized countries?

Review of Economics and Statistics, 74, 678–688.

Nissan, E., Galindo, M., & Mendez, M. (2010)."Relationship between tourism and
economic growth." The Services Industries Journal, 31(10), 1567–1572.
C 2016 CURJ, CUSIT 319
Muhammad Kamran et al.

Pizam, A. (1978). Tourism's Impacts: The Social Costs to the Destination Community as
Perceived by Its Residents. Journal of Travel Research, 16(4), 8-12.

Po, W. & Huang, B. (2008)."Tourism development and economic growth – a nonlinear

approach." Physica A, 387 (2008), 5535–5542.

Sequeria, T. & Nunes, P. (2010). "Does tourism influence economic growth? A dynamic
panel data approach." Applied Economics, 40(18), 2431–2441.

Turner, L., & Ash, J. (1975).The golden hordes: international tourism and the pleasure
periphery. New York: St. Martin's Press.

UNWTO (2009). UNWTO World Tourism Barometer, Jan 2009 Edition. Retrieved
from:<http://www.unwto.org/facts/eng/pdf/barometer/UNWTO_Baro09_2_
enexcerpt.pd [accessed 24 February 2010].

Wall, G., & Mathieson, A. (2006).Tourism: changes, impacts, and opportunities (2.
Ed.). Harlow, Eng. New York: Pearson Prentice Hall.

World Tourism Organization (2002). News from the World Tourism Organization.
Retrieved July 22, 2002 from http://www.world-tourism.

WTTC (2010). World Travel and Tourism Council. Travel and Tourism Economic
Impact: Pakistan. Retrieved from:
<http://www.wttc.org/bin/pdf/original_pdf_file/pakistan.pdf> [accessed 11
March 2010].

320 C 2016 CURJ, CUSIT