0% found this document useful (0 votes)
38 views19 pages

9141-Article Text-26325-1-10-20170429 PDF

Uploaded by

Aleksandra
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
38 views19 pages

9141-Article Text-26325-1-10-20170429 PDF

Uploaded by

Aleksandra
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

European Scientific Journal April 2017 edition Vol.13, No.

10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

Measuring Tourism Efficiency of European Countries


by Using Data Envelopment Analysis

Halenur Soysal-Kurt, MSc.


Department of Management Information Systems
Osmaniye Korkut Ata University, Turkey

doi: 10.19044/esj.2017.v13n10p31 URL:http://dx.doi.org/10.19044/esj.2017.v13n10p31

Abstract
This study aims to measure relative efficiency of 29 European
countries with the data of the year 2013 using input-oriented and constant
returns to scale Data Envelopment Analysis and to offer improvement
suggestions for the countries found inefficient based on their measured
relative efficiency scores. Three input and three output variables are used to
assess relative performances of the countries. In this study, tourism expenses,
number of employees and number of beds are used as input variables;
tourism receipts, tourist arrivals and number of nights spent are used as
output variables. As the result of the analysis, 16 countries are found
relatively efficient and 13 countries are found relatively inefficient. This
study is one of the few publications within the scope of European countries
based on data envelopment analysis. Unlike most researches evaluating the
efficiency of tourism establishments at the micro level, this paper is thought
to contribute to the related literature as it evaluates relative efficiency of the
countries at the macro level for tourism industry. Considering the variables
used in the analysis, it is expected to give ideas to relatively inefficient
European countries on efficiency improvement.

Keywords: Data Envelopment Analysis (DEA), Tourism, Efficiency,


European countries

Introduction
It is known that tourism sector is one of the largest and fastest
growing industries in the world. Thanks to generation of employment, export
revenues, investments and infrastructure developments, tourism sector makes
serious contribution to socio-economic process directly and indirectly.
According to the report of United Nations World Tourism Organization
(UNWTO, 2016), on a global basis, tourism revenues provided by
destinations was US$ 2 billion in 1950; in 2015 it amounted to US$ 1260

31
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

billion. According to the previous year it increased by 4.4% in real terms. In


addition to tourism revenues, total value of tourism exports provided by
international transportation services reached US$ 1.5 trillion in 2015.
International tourist arrivals were 25 million in 1950; in 2015 they reached
1186 million. International tourist arrivals increased by 4.6% according to
the previous year. In 2015, tourism industry contributed 10% to world gross
domestic product (GDP) regarding direct, indirect and induced effects. In
terms of employment, it contributed employment one out of every eleven.
Efficiency is the ratio of output to input in general manner and deals
with the operational performance of a firm (at micro level) or country (at
macro level) (Joppe & Li, 2016). The processes which produce more outputs
per input have bigger efficiency. If the largest possible output per input is
provided, optimum efficiency will be achieved. Without using new
technologies or making various changes, it is impossible to increase the
efficiency (Sherman & Zhu, 2006).
Efficiency assessment can be divided into two categories as
parametric and non-parametric. In parametric methods, production function
is preset and changes that may randomly affect production are considered
(factor analysis, regression analysis, stochastic frontier approach etc.). In
non-parametric methods, analyses are done without presetting production
function through linear programming (data envelopment analysis, back error
propagation, artificial neural network etc.) (Yi & Liang, 2015; Oukil et al.,
2016).
A tourist destination is a geographical area where tourists enjoy many
different experiments and can be divided into several levels as country,
region, city, holiday resort. It is possible to analyze the tourist destination
like a commercial business. In this case, it aims to produce maximum output
managing its inputs effectively (Barros et al., 2011). Physical and human
resources of the tourist destination are the inputs of virtual tourist production
process. From this point, outputs are affected by factors like tourist arrivals,
bed-nights, added-value, employment, customer satisfaction, scientific
information, capital resources, infrastructure, etc. (Cracolici et al., 2008;
Porter, 1990).
A tourist region may be inefficient for the following reasons
(Cracolici et al., 2008):
• The capacity may not be used efficiently because of public
regulations (limitations on hotel construction and transportation
infrastructure, etc.).
• Global effects may influence outcomes of strategic and operational
marketing actions (growing interest in the natural environment, restructuring
of the economies around the world, growing complex technology-human
resource interface, etc.).

32
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

• Physiography, culture and social impacts may limit operations on the


tourist destinations (environmental limits, relief of residents, etc.).
The objective of this study is to measure and compare tourism
efficiencies of 29 European countries relatively. Because of this purpose,
relative tourism efficiencies are measured with variables accepted in the
literature by using the non-parametric method Data Envelopment Analysis
(DEA). DEA method is preferred in this paper because of some reasons. One
of these, the calculation process is not relatively difficult. The method allows
identification and modification of factors affecting efficiency. It evaluates
the efficiency of a DMU regarding other DMUs with common factors. It
provides optimal weights for each factor and DMU. It permits that poor
decision making units (in this paper, countries) are easily detected. Input-
oriented CCR model used in this paper focuses on what should be the
optimum amount of input corresponding to a certain amount of output. In the
analysis, input-oriented CCR model is preferred from the DEA approaches
because it may be difficult to make changes in the amount of the outputs of
the countries.
The paper consists of five parts. Following the first and introduction
part, in the second part literature review made in the field of DEA on
tourism, in the third part DEA methodology, in the fourth part efficiency
analysis of European countries and its findings, in the last part conclusions
are located in the paper.

Literature Review
There are many studies in the literature that measure efficiency in the
tourism industry through DEA. A large part of these studies is carried out at
the micro level, assessing the efficiency of the hotels and restaurants. Banker
and Morey’s study (1986) is one of the early studies which implement DEA
in the tourism industry at the micro level. They measure technical and scale
efficiencies of 60 restaurants in fast food chain using six input and three
output variables. Inputs are expenditures for supplies and materials,
expenditures related to labor, advertising expenditures, the age of the store,
whether the store is located in an urban or rural area, and whether it has a
drive-in window. Outputs are breakfast sales, lunch sales and dinner sales. It
is understood from the application that the target output can be obtained by
saving the inputs. Johns et al. (1997), firstly use DEA in the hotel sector,
compare 15 hotels with each other through DEA using four input and three
output variables in 12-month period. The input variables are number of room
nights available, total labor hours, total beverage and food costs, and total
utilities cost. The output variables are number of room nights sold, total
covers served and total beverage revenue. In the conclusion of the study,
DEA is determined to be usable for diagnosing and identifying outstanding

33
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

behavior regarding gross profit and measured productivity. Anderson et al.


(2000) apply DEA method to 48 hotels in United States using the data of the
year 1994. Input variables are the full-time equivalent employees, the
number of rooms, total gaming related expenses, total food and beverage
expenses, and other expenses. As output variable, total revenue generated
from the chosen inputs except employees for the year ending 1994 is used.
Using these variables, overall, allocative, technical, pure technical and scale
efficiencies of the hotels are measured. As the result of the study, they
understand that efficient firms allocate more resources to food and beverage
operations, and the less efficient firms spend more on hotel operations and
other expenses, and have employees and rooms overmuch. Hwang and
Chang (2003) use DEA and Malmquist index to measure managerial
performance and efficiency changes between the years 1994-1998 for 45
hotels with the data of the year 1998. Inputs are number of full-time
employees, guest rooms, total area of meal department and operating
expenses. Outputs are room revenue, food and beverages revenue, and other
revenues. In conclusion, they find there are serious differences on efficiency
changes derived from customer resources and management styles. Barros
and Alves (2004) try to analyze the efficiency of 42 hotels in Portuguese
public-owned hotel chain for the period of 1999-2001 by using output-
oriented DEA. They rank the hotels regarding their total factor productivity
changes. Number of full-time workers, cost of labor, book value of property,
operating costs and external costs are used as inputs. As outputs, number of
guests and nights spent in the hotel are used. Consequently, they find that
while most hotels provide efficient technical change, they do not provide
technological change. Barros and Mascarenhas (2005) measure technical and
allocative efficiency through DEA with the data of the year 2001 for 43
hotels in Portuguese state-owned chain. Inputs are labor, capital and the
number of rooms. Outputs are sales, the number of guests and nights spent.
Input prices are price of labor, price of physical capital and price of rooms.
In the conclusion of the study, they recognize DEA has several managerial
advantages. One of the advantages of using DEA is identifying strategically
important pousadas. Another advantage is discovering overall competence
and capability of a pousada. Oukil et al. (2016) make a two-stage DEA
application for 58 hotels in Sultanate of Oman. Inputs are number of beds
and salary of employees. Outputs are annual revenue, number of guests,
number of nights and occupancy rate. In the result of the analysis, they find a
major part of the hotels in Oman technically inefficient. They determine that
location of the most hotels found efficient is at the capital of Muscat, and
variables affecting the hotel efficiency mostly are star rating and cultural
attractions.

34
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

One of the studies evaluating tourism efficiency at the macro level is


the study of Peypoch (2007). Peypoch measures tourism efficiency for
French tourism industry with the data set of the years 2000-2003 by
Luenberger productivity indicator. The analysis is performed on a
nationality-by-nationality basis. Seven countries and one continent are
participated in the analysis. He uses the number of tourist bed-nights in
hotels and campsites as input variables, and uses tourism receipts based on
tourist nationality as output variables. He concludes that compared to the
other non-parametric methods like Malmquist productivity index,
Luenberger productivity indicator are more useful than the other because of
not needing input or output orientation. Cracolici et al. (2008) measure 103
Italian regions’ efficiencies with the data set of the year 2001 by constant
returns to scale output-oriented DEA and Cobb-Dougles stochastic frontier
production function. Inputs are the regional state-owned cultural patrimony
and heritage, tourist school graduates and the labor units employed in the
tourism sector. Output is bed-nights. According to DEA analysis, only seven
regions are found efficient. According to the results of stochastic frontier
approach, labor units are found the most effective variable. Compared two
methods with each other, the results show a low consistency. Bi et al. (2011)
use a two-stage DEA model to measure the efficiency of China’s 31 regions.
In the first stage, manpower, materials and financial resources are used as
inputs. The second stage is the transformation process of tourist facilities and
services to economic and social benefits. Inputs used in the first stage are
fixed assets and persons engaged in the tourism industry. Outputs are number
of star-rated hotels, number of travel agencies and number of other tourism
corporations. Inputs used in the second stage are number of excellent tourism
cities, number of star-rated hotels, number of travel agencies and number of
other tourism corporations. Outputs are the total number of visitors, foreign
exchange earnings, tourist income, tax delivery, profits and rate of job
growth. In the results of the study, they find that compared to developed
countries Chinese regional tourism industry develops at a low level and there
are serious efficiency differences between 31 regions. Barros et al. (2011)
evaluate the efficiency of 22 French regions for the years 2003-2007 using a
two-stage DEA method in their study. In the first stage, they calculate the
efficiency scores; in the second stage, they use a bootstrapped truncated
regression model. Inputs used to measure the efficiency are accommodation
capacity and arrivals; output is nights slept. In the regression model tourism
attractors are used. These are monuments, museums, theme parks, beaches,
ski resorts and Natural Parks. According to the study’s findings, primary
factors that affect the efficiency are sea, sun and strategy based on beaches
endowment. They believe that developing theme parks, monuments, ski
resorts and Natural Parks can enhance tourism attraction and improve the

35
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

efficiencies. Assaf (2012) measures and compares the efficiencies of leader


tour operators and hotel firms in 12 Asian Pacific countries for the years
2007-2009. He uses a combination of stochastic frontier in Bayes framework
and DEA as the method. Inputs for the hotels are number of rooms, number
of full time equivalent employees and other operational costs. For the tour
operators, inputs are the same as the hotels; in addition to those the number
of rooms by the number of outlets is used. Total revenues are used as output
for both hotels and tour operators. In the results, Australia has the highest
efficiency in hotel and tour operator industry. The author believes that the
factors affecting that finding are outstanding air freight and tourism
infrastructure, and serious tourism appeal. Hadad et al. (2012) examine
tourism efficiency of 34 developed and 71 developing countries in two parts
using DEA based on super-efficiency with the data of the year 2008. Inputs
are number of employees, number of rooms, natural resources and cultural
resources. Outputs are number of tourists and expenditure per tourist. As the
result of the study, in developed countries Luxembourg, France, Ireland and
Singapore; in developing countries Ukraine, Bosnia, Brunei, Bahrain, Puerto
Rico, Hungary and Croatia are found efficient. It indicates that globalization
and accessibility play an important role in the efficiency of the tourism
industry. Assaf and Barros (2013) measure and compare performances of
hotel industries of countries worldwide with the stochastic frontier semi-
parametric approach. 26 from 37 countries are analyzed for the first time in
the literature. Using the data set of the years 2006-2008, they analyze 519
hotels with three input and three output variables. Inputs are number of
outlets, number of full-time equivalent employees and other operational
costs. Outputs are operational revenue, annual occupancy rate and market
share. According to the findings, they find the most efficient countries in
hotel industry as Spain, France, Germany, the USA, United Kingdom (UK),
United Arab Emirates (UAE) and Turkey. Kosmaczewska (2014) evaluates
tourism interest in terms of efficiency for the years 2007-2009. She analyzes
27 European Union (EU) countries through CCR, BCC and NIRS (non-
increasing returns to scale) models. As input, the arrivals in tourist
accommodation establishments are used; as outputs, collective tourist
accommodation establishments and GDP per inhabitant are used in the
analyses. Consequently, she determines that the efficiency of richer and
poorer countries resemble each other. In Atan and Arslantürk (2015)’s study,
efficiency analyses of 91 worldwide countries are performed for the years
2006-2010 through CCR and BCC models. Inputs are number of arrivals,
total activity, tourism expenses, tourism expenses for passenger
transportation needs and tourism expenses for travel needs. Outputs are
tourism revenues, tourism revenues for passenger transportation needs and
tourism revenues for travel needs. As the result of the study, according to

36
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

CCR model 65 countries, BCC model 67 countries are found efficient. The
findings of both models resemble each other.
In recent years, a considerable amount of study has been done on
efficiency and productivity analysis in the tourism literature. Both its
economic contributions and increasing competition in the global tourism
market have promoted the number of studies on tourism efficiency.
However, the majority of the tourism efficiency studies is bounded by travel
agents, hospitality industry and food & beverage facilities (micro-level)
(Hadad et al., 2012). When it is considered that efficiency measurement is
usually carried out for commercial businesses, countries in this paper are
considered as commercial businesses aiming to produce maximum output
managing their inputs effectively. The efficiency of the tourism industry at
the macro level has been discussed in few studies. Unlike most studies
evaluating the efficiency of tourism establishments, the study is thought to
contribute to the related literature as it evaluates relative efficiencies of the
countries at the macro level for tourism industry. In addition, it is one of the
few publications within the scope of European countries based on data
envelopment analysis.

Data Envelopment Analysis


DEA is firstly developed by Charnes, Cooper and Rhodes (1978)
(CCR) and is a non-parametric method that is used to assess the relative
efficiency of decision making units (DMUs) in the situations where multiple
inputs and outputs are available (Despotis & Simirlis, 2002). DEA
determines the efficiency of DMUs with common inputs and outputs through
the ratio of weighted outputs to weighted inputs of each DMU. DMUs are
compared with each other by constituting linear programming model (LP).
The efficient frontier consisting of the best observations is specified so that
the efficiencies of all DMUs are calculated by the distance from this frontier.
DMUs on the efficient frontier are relatively efficient; those outside the
frontier are relatively inefficient. Reference sets consisting of the efficient
DMUs are determined for the inefficient DMUs, thus suggestions providing
to be efficient for the inefficient DMUs can be given.
DEA is firstly used to improve planning and controlling of the
activities of public institutions by Charnes et al. (1978). Afterwards, it is
used to measure the relative efficiency in many areas and institutions as
hospitals, schools, factories, government business enterprises, service
industry, etc. One of the areas DEA used in is tourism industry that is
discussed in this paper.
Figure 1 shows the classification of DEA approaches by returns to
scale and orientation. DEA models are divided into two categories according
to scale as constant returns (CRS) and variable returns (VRS). CRS assumes

37
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

that there is no substantial relationship between scale and efficiency of the


DMU. If inputs change in a proportion, outputs change in that proportion. In
VRS, there are increasing, decreasing and constant returns to scale for
production process. According to the orientation, DEA differs depending on
input-oriented, output-oriented and nonoriented. In input-oriented models, it
is aimed to minimize the amount of inputs to produce predetermined outputs.
In output-oriented models, it is aimed to produce maximum output using
predetermined inputs. CCR (Charnes, Cooper & Rhodes, 1978) is the first
DEA model that calculates total efficiency based on constant returns to scale.
In BCC (Banker, Charnes & Cooper, 1984) model, there is no obligation to
be constant returns to scale. BCC model investigates local returns to scale
under the assumption of VRS (Charnes et al., 1994). Each DMU must
provide both technical and scale efficiency to be CCR-efficient, while it is
sufficient to provide only technical efficiency to be BCC-efficient (Bowlin,
1998). Because input-oriented constant returns to scale CCR model is used in
this study, only this model is discussed in the following section.
Input-
CCR-input
oriented
Constant
Nonoriented
returns to Nonoriented
scale (CRS) CRS

Output-
CCR-output
oriented
DEA
Approaches
Input-
BCC-input
oriented
Variable
returns to Nonoriented Additive
scale(VRS)
Output-
BCC-output
oriented
Figure 1. Classification by returns to scale and orientation (Ali, 1994)

Input-oriented CCR model


Input-oriented CCR model focuses on what should be the optimum
amount of input corresponding to a certain amount of output. In CCR model,
the efficiencies of DMUs are provided by the ratio of virtual outputs to
virtual inputs.
Assume that n is the number of DMUs, s is the number of outputs
and m is the number of inputs; CCR model for DMUo is as follows (Charnes
et al., 1978):

38
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

∑𝑠𝑟=1 𝑢𝑟 𝑦𝑟𝑜
𝑚𝑎𝑥ℎ𝑜 =
∑𝑚𝑖=1 𝑣𝑖 𝑥𝑖𝑜
subject to:
∑𝑠𝑟=1 𝑢𝑟 𝑦𝑟𝑗
≤1 𝑗 = 1, … , 𝑛
∑𝑚 𝑖=1 𝑣𝑖 𝑥𝑖𝑗
𝑣𝑖 , 𝑢𝑟 ≥ 0 ; 𝑖 = 1, … , 𝑚 ; 𝑟 = 1, … , 𝑠
Because the model above is fractional programming form, for
facilitating the solution it is transformed into linear programming form. The
results of both models are the same. CCR model in LP form for DMUo is as
follows (Cooper et al., 2006):
𝑠

max 𝜃 = � 𝑢𝑟 𝑦𝑟𝑜
𝑟=1
subject to:
𝑚

� 𝑣𝑖 𝑥𝑖𝑜 = 1
𝑖=1
𝑠 𝑚

� 𝑢𝑟 𝑦𝑟𝑗 − � 𝑣𝑖 𝑥𝑖𝑗 ≤ 0 𝑗 = 1, … , 𝑛
𝑟=1 𝑖=1
𝑣𝑖 , 𝑢𝑟 ≥ 0 ; 𝑖 = 1, … , 𝑚 ; 𝑟 = 1, … , 𝑠
In the model above, by generating maximization problems for all
DMUs, efficiency scores of each DMU are obtained. To assess the
efficiencies correctly, the number of DMUs must be greater than the sum of
the number of inputs and outputs three times (Raab & Lichty, 2002). vi and ur
respectively represent the weights of each input and the weights of each
output (the relative importance degrees). The equality constraint represents
the sum of the virtual inputs of DMUo. Inequality constraint states that the
sum of the weighted outputs cannot be greater than the sum of the weighted
inputs for each DMU. The inequalities as many as the number of DMUs are
added on the LP model. The objective function represents the virtual outputs
of DMUo. If the optimum value of the objective function (θ*) is equal to 1,
DMUo will be efficient. If the value θ* is smaller than 1, DMUo will be
inefficient relatively.
The dual form of CCR model for DMUo is as follows (Banker et al.,
2004):
𝑚 𝑠

𝑚𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝜃 − 𝜀 �� 𝑠𝑖− + � 𝑠𝑟+ �


𝑖=1 𝑟=1
subject to:

39
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

0 = 𝜃𝑥𝑖𝑜 − � 𝑥𝑖𝑗 𝜆𝑗 − 𝑠𝑖− ,


𝑗=1
𝑛

𝑦𝑟𝑜 = � 𝑦𝑟𝑗 𝜆𝑗 − 𝑠𝑟+ ,


𝑗=1
0 ≤ 𝜆𝑗 , 𝑠𝑖− , 𝑠𝑟+ ∀𝑖, 𝑗, 𝑟.
The dual model with adding slack variables contains information for
inefficient DMUs about what should be done to become efficient.
θ is a value between zero and one. It determines how much the input
xio should be reduced to θxo radially to get DMUo to the efficient frontier
(Kulshreshtha & Parikh, 2002; Cooper et al., 2006).
The values λj are the density values of the elements in the reference
sets that provide DMUo to be efficient.
The value 𝑠𝑖− (input excesses) is the slack input value belonging to i.
input of DMUo. The value 𝑠𝑟+ (output shortfalls) is the slack output value
belonging to r. output of DMUo.
According to dual model, if the weighted inputs of relatively
inefficient DMUo are converted to (𝜃𝑥𝑖𝑜 − 𝑠𝑖− ) and the weighted outputs of
relatively inefficient DMUo are converted to (𝑦𝑟𝑜 + 𝑠𝑟+ ), DMUo will be
efficient.
To determine the possible input excesses and output shortfalls, two-
stage LP model is solved (Cooper et al., 2007):

Stage I
The dual form of the model is solved and then the value θ* is found.
The value θ* is the same as the efficiency value calculated in the primal
linear model. The value θ* obtained will be used in Stage II.

Stage II
Using the value θ*, the following model is solved:
max
− +
𝑤 = 𝑒𝑠 − + 𝑒𝑠 +
𝜆,𝑠 ,𝑠
subject to:
𝑠 − = 𝜃 ∗ 𝑥𝑜 − 𝑋𝜆
𝑠 + = 𝑌𝜆 − 𝑦𝑜
0 ≤ 𝜆𝑗 , 𝑠𝑖− , 𝑠𝑟+
e = (1,…, 1) a vector whose elements equal to 1.
𝑒𝑠 − = ∑𝑚
𝑖=1 𝑠𝑖

; 𝑒𝑠 + = ∑𝑠𝑟=1 𝑠𝑟+
The purpose of Stage II is to find a solution which makes the sum of
the input excesses and output shortfalls maximum maintaining θ = θ*. In

40
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

order to be CCR-efficient with optimal solution values (𝜃 ∗ , 𝜆∗ , 𝑠 −∗ , 𝑠 +∗ ),


DMUo must satisfy following two criteria (Cooper et al., 2007):
I. θ* = 1
II. All slacks (𝑠𝑖−∗ , 𝑠𝑟+∗ ) are equal to zero.
But in some cases, it can be seen that only the first constraint is
satisfied. In this situation, DMUo is characterized as “weak efficient”.

Analyses and Findings


In this study, DEA method is applied to 29 European countries with
high income and upper-middle income using variables compiled from the
studies in the literature (Hwang & Chang, 2003; Oukil et al., 2016; Hadad et
al., 2012; Assaf, 2012). The application is performed using constant returns
to scale (CRS) and input-oriented (CCR) DEA through MDEAP 2 software
developed by Michel Deslierres. In the analysis, three input and three output
variables thought to affect the tourism efficiency and used commonly in the
literature are used. In order for the analysis to produce more reasonable
results, countries with similar income levels are preferred for the analysis.
The grouping of countries by income levels is based on the order of The
World Bank. Due to limited data availability, only European countries are
participated in the analysis.

Inputs
Number of employees: Employed persons by full-time/part-time
activity in the tourism sector.
Tourism expenses: The money spent for tourism activities.
Number of beds: Number of beds in tourist accommodation
establishments.

Outputs
Tourist arrivals: The number of the tourist coming from other
countries. (overnight visitors).
Tourism receipts: The money spent by international tourists.
Number of nights spent: 1 night or over nights spent in tourist
accommodation establishments.
Descriptive statistics of variables used to measure the efficiency of
European countries are given in Table 1. Data set of the variables is accessed
from Eurostat, The World Bank and World Economic Forum. As it is seen
from the standard deviation values in Table 1, data is spread over a wide
range.

41
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

Table 1. Descriptive statistics of variables for European countries


Variables Minimum Maximum Mean Std. Deviation
Number of employees
175,9 39.531,4 8.299,121 10.545,1351
(thousand)
Tourism expenses
384 91.366 12.507,97 19.841,258
(million $)
Number of beds
38,40 5.049,73 1.074,2069 1.484,45145
(thousand)
Arrivals (thousand) 944 84.726 16.156,24 19.910,034
Tourism receipts
1.190 66.049 17.196,34 19.790,611
(million $)
Number of nights spent
3.079,21 1.312.992,49 218.993,5648 353.920,05784
(thousand)

In Table 2, European countries are numbered from 1 to 29 to identify


reference countries which will be discussed in later section. The findings of
input-oriented (CCR) and constant returns to scale (CRS) DEA model are
given in Table 2.
According to the findings, 16 countries are found efficient; 13
countries are found inefficient. Efficient countries whose efficiency scores
are equal to 1 are Cyprus, Croatia, Estonia, Finland, France, Greece,
Hungary, Ireland, Latvia, Luxembourg, Malta, Poland, Portugal, Spain,
Switzerland and Turkey. Inefficient countries whose efficiency scores are
found less than 1 are Austria, Belgium, Bulgaria, Czech Republic, Denmark,
Germany, Italy, Lithuania, Netherlands, Romania, Slovak Republic, Slovenia
and United Kingdom. Regarding variables used, the countries with the
lowest efficiency score are Italy with 0,455 efficiency score, Netherlands
with 0,616, Belgium with 0,651 and United Kingdom with 0,66. It should be
considered that the efficiency scores calculated in DEA are relative
measures.
Table 2. Efficiency scores of European countries for the year 2013
No Countries Efficiency scores
1 Austria 0,813
2 Belgium 0,651
3 Bulgaria 0,792
4 Cyprus 1,000
5 Croatia 1,000
6 Czech Republic 0,783
7 Denmark 0,848
8 Estonia 1,000
9 Finland 1,000
10 France 1,000
11 Germany 0,760
12 Greece 1,000
13 Hungary 1,000
14 Italy 0,455
15 Ireland 1,000
16 Latvia 1,000
17 Lithuania 0,879

42
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

18 Luxembourg 1,000
19 Malta 1,000
20 Netherlands 0,616
21 Poland 1,000
22 Portugal 1,000
23 Romania 0,962
24 Slovak Republic 0,828
25 Slovenia 0,975
26 Spain 1,000
27 Switzerland 1,000
28 Turkey 1,000
29 United Kingdom 0,660

To interpret the findings in more detail, let us take Belgium, which is


inefficient, and Switzerland, which is efficient. One reason for Belgium’s
inefficiency relative to Switzerland may be the fact that the number of nights
spent per bed in Switzerland is about 503,7, while in Belgium it is about
258,8. Furthermore, tourist arrivals per bed in Switzerland is 36,4, while in
Belgium 20,5. These mean that Switzerland uses its bed capacity more than
Belgium. In addition, the number of tourist arrivals per employee in Belgium
is lower. Because the ratio is about 2,01 in Switzerland, it is about 1,68 in
Belgium. Another example is United Kingdom, which is inefficient. The
number of nights spent per bed is about 220,9 in United Kingdom. Similarly,
the ratio of tourist arrivals per bed is 7,8 in United Kingdom. Both imply that
United Kingdom cannot use its bed capacity as well as Switzerland.
Table 3. Reference sets and λ density values of European countries according to CCR model
Country Reference countries and λ density values
no No λ No λ No λ No λ No λ
1 4 3,10869 5 0,0114 10 0,0167 19 9,98518
2 9 0,14296 15 0,41309 18 0,34945 27 0,39034
3 5 0,08693 19 2,04323 28 0,07176
4 4 1,000
5 5 1,000
6 9 0,11111 21 0,06912 26 0,14683
7 4 2,32512 9 0,41505 15 0,21843
8 8 1,000
9 9 1,000
10 10 1,000
11 9 9,06177 27 0,91642
12 12 1,000
13 13 1,000
14 5 0,04979 13 0,00083 19 12,2053 26 0,45896
15 15 1,000
16 16 1,000
17 8 0,35955 13 0,00482 21 0,05458 28 0,00172
18 18 1,000
19 19 1,000
20 4 3,31776 9 1,51475 10 0,00668
21 21 1,000
22 22 1,000

43
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

23 21 0,10707 26 0,06358
24 8 1,50457 13 0,12338 21 0,0227 28 0,00626
25 4 0,23432 16 0,17319 22 0,07169 26 0,00998 28 0,00605
26 26 1,000
27 27 1,000
28 28 1,000
29 4 2,54309 9 4,6719 22 0,61776 26 0,2705

Reference sets determined to give suggestions which will provide to


be efficient for the inefficient countries and λ density values are given in
Table 3. According to the reference sets given in the table, the most
referenced countries are Finland (6 times), Cyprus (5 times), Spain (5 times),
Turkey (4 times) and Poland (4 times). Thanks to the countries in the
reference set of an inefficient country and λ density values, input quantities
to achieve to be efficient are calculated. For relatively inefficient countries,
actual input values, target input values calculated according to CCR model
and potential improvement rates for the input variables are respectively given
in Table 4 and Table 5.
Table 4. Actual and target input values of European countries according to input-oriented CCR model
Actual Input Values Target Input Values
Tourism
Efficiency Number of Number of Number of Tourism Number of
No Countries expenses
scores employees beds employees expenses beds
(million
(thousand) (thousand) (thousand) (million $) (thousand)
$)
1 Austria 0,813 4.104,7 10.277 981,07 3.339,08 8.360,1 798,08
2 Belgium 0,651 4.530,3 21.876 372,87 2.952,9 10.913,5 243,04
3 Bulgaria 0,792 2.934,9 1.527 302,43 2.323,28 1.208,78 239,4
4 Cyprus 1,000 365,1 1.222 87,14 365,1 1.222 87,14
5 Croatia 1,000 1.524,0 903 867,11 1.524,0 903 867,11
Czech
6 0,783 4.937,1 4.595 740,67 3.865,56 3.597,71 579,92
Republic
7 Denmark 0,848 2.687,6 10.132 417,59 2.279,45 6.392,18 354,17
8 Estonia 1,000 621,3 995 55,48 621,3 995 55,48
9 Finland 1,000 2.456,7 5.292 254,11 2.456,7 5.292 254,11
10 France 1,000 25.763,6 42.907 5.049,73 25.763,6 42.907 5.049,73
11 Germany 0,760 39.531,4 91.366 3.326,82 26.349,94 62.666,23 2.528,58
12 Greece 1,000 3.513,2 2.437 1.207,39 3.513,2 2.437 1.207,39
13 Hungary 1,000 3.892,8 1.908 422,04 3.892,8 1.908 422,04
14 Italy 0,455 22.190,5 26.972 4728,18 10.092,12 12.266,73 2.150,35
15 Ireland 1,000 1.881,2 6.201 211,03 1.881,2 6.201 211,03
16 Latvia 1,000 893,9 715 38,4 893,9 715 38,4
17 Lithuania 0,879 1.292,8 966 69,29 1.135,9 848,76 60,88
18 Luxembourg 1,000 238,7 3.804 66,75 238,7 3.804 66,75
19 Malta 1,000 175,9 384 43,36 175,9 384 43,36
20 Netherlands 0,616 8.285,2 20.056 1.404,85 5.104,72 12.357,01 707,76
21 Poland 1,000 15.568,0 8.675 679,45 15.568,0 8.675 679,45
22 Portugal 1,000 4.429,4 4.143 491,1 4.429,4 4.143 491,1
23 Romania 0,962 8.549,1 2.056 303,24 2.766,66 1.978,09 291,75
Slovak
24 0,828 2.329,2 2.367 190,31 1.928,25 1.959,55 157,55
Republic
25 Slovenia 0,975 905,9 923 105,56 883,51 900,19 102,95

44
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

26 Spain 1,000 17.139,0 16.414 3.437,36 17.139,0 16.414 3.437,36


27 Switzerland 1,000 4.460,7 16.053 246,49 4.460,7 16.053 246,49
28 Turkey 1,000 25.519,8 4.817 1.051,16 25.519,8 4.817 1.051,16
29 UK 0,660 29.952,5 52.748 4.001,02 19.778,31 34.830,69 2.641,96

For example, considering the inefficient country Denmark with 0,848


efficiency score, its reference countries are found as Cyprus with λ4=2,32512
density value, Finland with λ9=0,41505, Ireland with λ15=0,21843. The
calculation of Denmark’s target values of input variables to become efficient
is as follows:
XDenmark = (X1, X2, X3) = {(365,1; 1.222; 87,14) x 2,32512 + (2.456,7;
5.292 ; 254,11) x 0,41505 + (1.881,2; 6.201; 211,03) x 0,21843
= {(2.279,45; 6.392,2; 354,17)}
Potential improvement rates for the inputs in Table 5 are calculated
𝑇𝑎𝑟𝑔𝑒𝑡 𝑣𝑎𝑙𝑢𝑒−𝐴𝑐𝑡𝑢𝑎𝑙 𝑣𝑎𝑙𝑢𝑒
according to the formula � ∗ 100�.
𝐴𝑐𝑡𝑢𝑎𝑙 𝑣𝑎𝑙𝑢𝑒
Table 5. Potential improvement rates of European countries according to input-oriented
CCR model
Potential Improvement Rates (%)
No Countries Number of
Tourism expenses Number of beds
employees
1 Austria -18,65 % -18,65 % -18,65 %
2 Belgium -34,82 % -50,11 % -34,82 %
3 Bulgaria -20,84 % -20,84 % -20,84 %
4 Cyprus 0% 0% 0%
5 Croatia 0% 0% 0%
6 Czech Republic -21,7 % -21,7 % -21,7 %
7 Denmark -15,19 % -36,91 % -15,19 %
8 Estonia 0% 0% 0%
9 Finland 0% 0% 0%
10 France 0% 0% 0%
11 Germany -33,34 % -31,41 % -23,99 %
12 Greece 0% 0% 0%
13 Hungary 0% 0% 0%
14 Italy -54,52 % -54,52 % -54,52 %
15 Ireland 0% 0% 0%
16 Latvia 0% 0% 0%
17 Lithuania -12,14 % -12,14 % -12,14 %
18 Luxembourg 0% 0% 0%
19 Malta 0% 0% 0%
20 Netherlands -38,39 % -38,39 % -49,62 %
21 Poland 0% 0% 0%
22 Portugal 0% 0% 0%
23 Romania -67,64 % -3,79 % -3,79 %
24 Slovak Republic -17,21 % -17,21 % -17,21 %
25 Slovenia -2,47 % -2,47 % -2,47 %
26 Spain 0% 0% 0%
27 Switzerland 0% 0% 0%
28 Turkey 0% 0% 0%
29 United Kingdom -33,97 % -33,97 % -33,97 %

45
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

For example, according to Table 5, some improvements are required


to make Denmark efficient. To make Denmark efficient, number of
employees (thousand) should be reduced from 2.687,6 to 2.279,45; tourism
expenses (million $) should be reduced from 10.132 to 6.392,2; number of
beds (thousand) should be reduced from 417,59 to 354,17. Thus, it can be
applied -15,19 % improvement to number of employees, -36,91 %
improvement to tourism expenses and -15,19 % improvement to number of
beds. Because the analyses are for input-oriented CCR model, there will be
no change in outputs.

Conclusion
In many studies in the literature, tourism efficiency analyses have
been done in subcategories of tourism industry like hospitality industry,
travel agencies, and restaurants. In some of the studies, economic efficiency
assessment has been also considered. This study evaluates tourism efficiency
at the macro level (based on countries) using basic characteristics (number of
employees, tourism expenses, number of beds, tourism receipts, tourist
arrivals, number of nights spent) that affect the efficiency of tourism
industry. For this purpose, tourism efficiency of 29 European countries with
high income and upper-middle income is tried to be analyzed. In the
analysis, CCR-DEA method is used with the data set of the year 2013.
According to the findings of the analysis, 16 countries are identified
efficient; 13 countries are identified inefficient. The countries identified
relatively efficient are Cyprus, Croatia, Estonia, Finland, France, Greece,
Hungary, Ireland, Latvia, Luxembourg, Malta, Poland, Portugal, Spain,
Switzerland and Turkey. Relatively inefficient countries are Austria,
Belgium, Bulgaria, Czech Republic, Denmark, Germany, Italy, Lithuania,
Netherlands, Romania, Slovak Republic, Slovenia and United Kingdom. The
countries with the lowest efficiency scores are found as Italy, Netherlands,
Belgium and United Kingdom. It should be considered that the efficiency
scores calculated in DEA are relative measures and there are other
controllable and uncontrollable factors (globalization, physical capitals,
cultural and natural resources, accessibility, environmental sustainability,
safety, etc.) affecting the efficiency. It is necessary to use various indicators
for more detailed evaluation of the efficient use of resources. From this
point, these findings do not give the countries definite results about their
tourism efficiency, give only a preliminary idea on the balance of
inputs/outputs according to the given variables. This paper may be
suggestive in terms of resource allocation and long-term decisions of local
governments and major tourism companies.
Due to the limitation of data availability, the number of countries and
variables used in the analysis are limited. For a more comprehensive

46
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

analysis, variables affecting tourism efficiency (natural and cultural


resources, etc.) can be enhanced. In this study, when the relative efficiencies
of European countries are analyzed through DEA method, only input-
oriented CCR model is used. By revising the number of countries and using
various DEA models, it is possible to obtain different findings.

References:
1. Ali, A. I. (1994). Computational aspects of DEA. In A. Charnes, W.
W. Cooper, A. Y. Lewin and L. M. Seiford (Eds.), Data envelopment
analysis: Theory, methodology, and application (1st ed., pp. 63-88).
New York: Springer Science+Business Media, LLC.
2. Anderson, R. I., Fok, R. & Scott, J. (2000). Hotel industry efficiency:
An advanced linear programming examination. American Business
Review, 18(1), 40-48.
3. Assaf, A. G. (2012). Benchmarking the Asia Pacific tourism industry:
A Bayesian combination of DEA and stochastic frontier. Tourism
Management, 33(5), 1122-1127.
4. Assaf, A. G. & Barros, C. P. (2013). A global benchmarking of the
hotel industry. Tourism Economics, 19(4), 811-821.
5. Atan, M. & Arslantürk, Y. (2015). Dünya ülkelerin turizm
potansiyelinin etkinliği. Gazi Journal of Economics and Business,
1(1), 59-76.
6. Banker, R. D. & Morey, R. C. (1986). Efficiency analysis for
exogenously fixed inputs and outputs. Operations Research, 34(4),
513-521.
7. Banker, R. D., Charnes, A. & Cooper, W. W. (1984). Some models
for estimating technical and scale efficiencies in data envelopment
analysis. Management Science, 30(9), 1078-1092.
8. Banker, R. D., Cooper, W. W., Seiford, L. M., Thrall, R. M. & Zhu,
J. (2004). Returns to scale in different DEA models. European
Journal of Operational Research, 154(2), 345-362.
9. Barros, C. P. & Alves, F. P. (2004). Productivity in tourism industry.
International Advances in Economic Research, 10(3), 215-225.
10. Barros, C. P. & Mascarenhas, M. J. (2005). Technical and allocative
efficiency in a chain of small hotels. Hospitality Management, 24(3),
415-436.
11. Barros, C. P., Botti, L., Peypoch, N., Robinot, E., Solonandrasana, B.
& Assaf, A. G. (2011). Performance of French destinations: Tourism
attraction perspectives. Tourism Management, 32(1), 141-146.
12. Bi, G., Luo, Y. & Liang, L. (2011). Efficiency evaluation of tourism
industry with data envelopment analysis (DEA): A case study in
China. Journal of China Tourism Research, 7(1), 104-116.

47
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

13. Bowlin, W. F. (1998). Measuring performance: An introduction to


data envelopment analysis (DEA). The Journal of Cost Analysis,
15(2), 3-27.
14. Charnes, A., Cooper, W. W. & Rhodes, E. (1978). Measuring the
efficiency of decision making units. European Journal of
Operational Research, 2(6), 429-444.
15. Charnes, A., Cooper, W. W., Lewin, A. Y. & Seiford, L. M. (1994).
Data envelopment analysis: Theory, methodology, and application.
(1st ed.) New York: Springer Science+Business Media, LLC.
16. Cooper, W. W., Seiford, L. M. & Tone, K. (2006). Introduction to
data envelopment analysis and its uses: With DEA-Solver software
and references. New York: Springer Science & Business Media, Inc.
17. Cooper, W. W., Seiford, L. M. & Tone, K. (2007). Data envelopment
analysis: A comprehensive text with models, applications, references
and DEA-Solver software (2nd Ed.). New York: Springer Science &
Business Media, LLC.
18. Cracolici, M. F., Nijkamp, P. & Rietveld, P. (2008). Assessment of
tourism competitiveness by analysing destination efficiency. Tourism
Economics, 14(2), 325-342.
19. Despotis, D. K. & Simirlis, Y. G. (2002). Data envelopment analysis
with imprecise data. European Journal of Operational Research,
140(1), 24-36.
20. Hadad, S., Hadad, Y., Malul, M. & Rosenboim, M. (2012). The
economic efficiency of the tourism industry: A global comparison.
Tourism Economics, 18(5), 931-940.
21. Hwang, S.-N. & Chang, T.-Y. (2003). Using data envelopment
analysis to measure hotel managerial efficiency change in Taiwan.
Tourism Management, 24(4), 357-369.
22. Johns, N., Howcroft, B. & Drake, L. (1997). The use of data
envelopment analysis to monitor hotel productivity. Progress in
Tourism and Hospitality Research, 3(2), 119-127.
23. Joppe, M. & Li, X. P. (2016). Productivity measurement in tourism:
The need for better tools. Journal of Tourism Research, 55(2), 139-
149.
24. Kosmaczewska, J. (2014). Tourism interest and the efficiency of its
utilisation based on the example of the EU countries. Acta
Scientiarum Polonorum Oeconomia, 13(1), 77-90.
25. Kulshreshtha, M. & Parikh, J. K. (2002). Study of efficiency and
productivity growth in opencast and underground coal mining in
India: a DEA analysis. Energy Economics, 24(5), 439-453.
26. Oukil, A., Channouf, N. & Al-Zaidi, A. (2016). Performance
evaluation of the hotel industry in an emerging tourism destination:

48
European Scientific Journal April 2017 edition Vol.13, No.10 ISSN: 1857 – 7881 (Print) e - ISSN 1857- 7431

The case of Oman. Journal of Hospitality and Tourism Management,


29, 60-68.
27. Peypoch, N. (2007). On measuring tourism productivity. Asia Pacific
Journal of Tourism Research, 12(3), 237-244.
28. Porter, M. E. (1990). The competitive advantage of nations. Harvard
Business Review, March-April, 73-91.
29. Raab, R. L. & Lichty, R. W. (2002). Identifying subareas that
comprise a greater metropolitan area: The criterion of county relative
efficiency. Journal of Regional Science, 42(3), 579-594.
30. Sherman, H. D. & Zhu, J. (2006). Service productivity management:
Improving service performance using data envelopment analysis.
Boston, MA: Springer.
31. UNWTO (2016). Tourism highlights 2016 Edition.
32. Yi, T. & Liang M. (2015). Evolutional model of tourism efficiency
based on the DEA method: A case study of cities in Guangdong
Province, China. Asia Pacific Journal of Tourism Research, 20(7),
789-806.

49

You might also like