APPLICATIONS OF NETWORK DATA ENVELOPMENT

ANALYSIS TO EXPLORE THE OPERATIONAL EFFICIENCIES OF

INTERNATIONAL FIVE-STAR HOTELS

Han Sheng Lei

Department of Business Administration, National Yunlin University of Science &
Technology, Yunlin, Taiwan, R.O.C.

Jui - Tu Liu
Department of Business Administration, National Yunlin University of Science &
Technology, Yunlin, Taiwan, R.O.C.
*Corresponding Author sls0114@[Link]

Abstract

Based on operating characteristics of the hotel industry, this study divided the service
processes of international five-star hotels into room service s and food & beverage
service, and used network data envelopment analysis to analyze the service efficien-
cies of the distinctive stages of the hotel industry. Study results show that interna-
tional five-star hotels that have distinctive customer sources and are located in distinc-
tive areas, indicating significant differences in overall efficiencies. In addition, if the
hotel is the international chain brand, the only significant difference is in room occu-
pancy efficiency. Finally, based on room occupancy efficiency and catering efficiency,
this study classified 46 DMU into four categories, and provided corresponding sug-
gestions for operational management and marketing activities.

Keywords: International Five-Star Hotels, Network Data Envelopment Analysis,

Room Service, Catering Service.

guru on management, once pointed out

Introduction that “the new economy is service econ-
omy”. The service industry has become
Due to fast economic developments an industry accompanied with global
and increase in income, countries world economy development, which grows
wide are gradually transforming from very fast in the 21st century (Fatma &
industrial society to service-oriented in- Timothy, 2005). According to statistics
dustrial structures. Peter F. Drucker, a from the United Nations World Tourism

64
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
Organization (UNWTO), in 2014, that can be used to compare the opera-
worldwide, there were over 1.1 billion tional efficiencies of distinctive tourist
trips taken abroad. Number of interna- hotels.
tional tourists has increased for five con-
secutive years since the financial crisis. The analysis methods used by pre-
Tourism has now become a primary vious studies on evaluations of hotel ef-
source of foreign exchange income for ficiency can be classified into three
many countries. The World Economic categories. The first category is ratio
Forum has just released a report on analysis, which was commonly adopted
global travel and tourism competitive- by earlier studies. This method calc-
ness. The report says that global GDP ulates the ratio between a single output
created by travel and tourism has grown and a single input to measure the opera-
at an average rate of 3.4% per year in tional performance of a hotel (Was-
the past three years, showing that the senaar & Stafford, 1991). The second
contribution from travel and tourism to category is the parametric approach.
the economy of every country is increas- This approach takes one output item as a
ing. dependent variable, and several input
items as independent variables, and uses
According to statistics from the a linear production function to evaluate
Tourism Bureau, Ministry of Transpor- the efficiency of the evaluated object
tation, & Communications, in 2014, 83 (Assaf & Magnini, 2012). The third
out of 100 visitors to Taiwan stayed in category is the data envelopment analy-
hotels, with sightseeing visitors spend- sis (DEA) which has been widely used
ing an average of 95 US dollars in hotels in recent years. However, this method
each day, approximately 40% of their doesn’t need to set a function in advance.
total spending (Bureau of Tourism, Min- Dividing multiple output items by mul-
istry of Transportation, & Communica- tiple input items, this method obtains
tions, 2015). Consequently, in order to production frontiers using mathematical
enhance tourism service qualities, the programming to evaluate the relative
satisfaction of tourists, and to shape the efficiencies of hotels. This analytical
tourism image of Taiwan, good opera- method is a kind of non-parametric ap-
tional management of tourist hotels is proach (Manasakis, Apostolakis, &
crucial in tourism. In June, 2015, there Datseris, 2013).
were 46 international five-star hotels in
Taiwan. Considering such fierce market The traditional DEA model (Char-
competition and in order to maintain nes, Cooper, & Rhodes, 1978; Banker,
sustainable operations, externally, hotels Charnes, & Cooper, 1984) can only pre-
should have acute responses to the mar- sent the overall operational efficiency,
ket, and internally, should properly use but cannot further present the transfor-
their limited resources to create maxi- mation processes of middle-stage re-
mum profit. In order to understand their sources. However, in practical operation,
advantages and weaknesses, and to the production activities of an organiza-
strengthen any shortcomings in their op- tion consist of multiple sub-activities
erational capacity, hotel owners should (Färe & Grosskopf, 2000; Tone &
conduct analyses of their competitors. Tsutsui, 2009). Besides the final outputs,
Therefore, this study aims to establish a each sub-production system may pro-
systematic efficiency evaluation model duce outputs that are used as inputs of
65
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
other sub-production systems. Therefore, 3. Establish a decision-making matrix
adopting a multistage network efficiency based on the room service efficiencies
evaluation model can help to obtain and food & beverage service efficien-
more transformational efficiencies of cies of hotels. The evaluated hotels
middle resources, which can be used as which are based on competitive posi-
important references for generating op- tions will be divided into four catego-
erational strategies. There have been ries: leading hotels, hotels with poten-
studies adopting multistage DEA to tial, followers, and hotels with little
measure operational efficiencies (Huang, potential in order to present the com-
Ho & Chiu, 2014; Hsieh & Lin, 2010; petitive positions of the hotels.
Yu & Lee, 2009).
Literature Adopting DEA to Evaluate
Previous studies have generally di- Hotel Efficiencies
vided hotel efficiency into efficiencies
of room service and food & beverage Reviewing the previous literature
service, but have failed to conduct fur- on hotel efficiency evaluation, evalua-
ther comparisons based on the character- tion models can be divided into either
istics of hotels, and also to understand single-stage DEA or multistage DEA
the competitive positions of hotels. To categories. Introductions to the two
summarize, this paper proposes the fol- categories are as follows.
lowing three study aims.
Single-stage DEA
1. Referring to the characteristics of the
production activities of hotels, this Earlier literature mostly adopted
study will adopt network data envel- static CCR and BCC models or SBM
opment analysis to establish a multi- DEA to evaluate hotels’ operational ef-
stage efficiency evaluation model that ficiencies for a specific year, or from a
can evaluate the room service effi- long-term perspective, adopted Malm-
ciencies, food & beverage service ef- quist Productivity Index (MPI) to dis-
ficiencies, and operational efficiencies cuss hotel efficiency changes over dif-
of hotels in Taiwan, and compare the ferent periods.
evaluation results with those obtained
by a traditional model to present the For example, Anderson, Fok &
differences, strengths and weaknesses Scott (2000) adopted CCR and BCC
of the network efficiency evaluation models to discuss the operational effi-
model. ciencies of 48 hotels in the USA. The
study results showed that poor perform-
2. Evaluate whether room service effi- ance of efficiency results from poor
ciencies, food & beverage service ef- technical efficiency and allocation effi-
ficiencies, and operational efficiencies ciency, and that efficient hotels allocate
of international five-star hotels in more resources to food & beverage
Taiwan are influenced by such factors management. Barros & Alves (2004)
as international chain brands, hotel used MPI to evaluate the operational
location, and types of tourists, and efficiencies of 42 hotels in Portugal. The
provide deeper implications for opera- study results showed that more than
tional management. 60% of the hotels had space for im-
provement in gross productivity. Hwang
66
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
& Chang (2003) adopted DEA to dis- fessional association to gather together
cuss the operational efficiencies of 45 the cohesive forces of the hotel industry
international hotels in 1998, and then to enhance overall market efficiency.
adopted MPI to compare the evaluated Manasakis, Apostolakis, & Datseris
hotels’ efficiency changes from 1994 to (2013) adopted DEA to compare the op-
1998. Based on relative efficiencies and erational efficiencies in year 2008 of 25
efficiency changes, the study divided the brand hotels with 25 independently op-
hotels into six groups, and proposed cor- erating hotels in Crete, Greece, finding
responding operational strategies for inefficient DMUs and proposing impli-
each group. Yang & Lu (2006) adopted cations for operational management. The
DEA to evaluate the operational effi- study results showed that domestic
ciencies in year 2002 of 56 international brand hotels were comparatively effi-
hotels in Taiwan, and further analyzed cient, and that international brand hotels
why some hotels failed to be efficient. had the best efficiencies. However, the
inefficiencies of inefficient hotels were
The study results show that 40% of mainly caused by the inequitable distri-
the international hotels fail to take full bution factor of input and output re-
advantage of their manpower, and fail to sources, and had nothing to do with the
make the best of the floorage of the food capabilities of their management teams.
& beverage department. Chiang, Tsai, & Chen, Lu, & Chung (2010) adopted
Wang (2004) adopted the BCC model to slacks-based measure (SBM) context-
discuss operational efficiencies of 25 dependent to analyze the operational ef-
international hotels of different operat- ficiencies of 34 international hotels in
ing types in Taiwan in year 2000. The Taiwan. The analysis results can assist
study results showed that the operational hotel operators to understand their com-
performance of hotels from international petitive advantages, find benchmarking
hotel chains is not necessarily higher hotels to learn from, and from which the
than independent hotels. generation of operating strategies can be
determined.
Wang, Shang, & Hung (2006)
added the perspective of quality element, Multistage DEA
and adopted MPI to measure service
qualities and operational efficiencies of A single-stage DEA can only pre-
29 international hotels in Taiwan. The sent overall operational efficiencies, but
studied period was the 11 years from cannot present the transformation proc-
1992 to 2002. The study results showed esses of middle resources. Consequently,
that when employees felt that they were Seiford & Zhu (1999) first proposed a
not respected, the operational efficien- two-stage DEA, using profitability and
cies and service qualities of the hotels marketability to measure the operational
would decrease. Barros & Dieke (2008) efficiencies of banks, which can then
used CCR and BCC models to discuss present the sub-production activities of
the operational efficiencies of 12 hotels the organizations. As soon as the con-
in Rwanda during the seven year period cept was proposed, it was applied by
from 2000 to 2006. The study results scholars to different fields. Two-stage
show that only one hotel demonstrated DEA assumes that the inputs of the sec-
the scale effect. The study suggested that ond stage are the outputs of the first
the government should establish an pro- stage, while in practice, production ac-
67
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
tivities have multiple stages, and inputs the efficient level. Yu & Lee (2009)
and outputs are not necessarily consis- used network DEA to analyze the effi-
tent. Consequently, Färe & Grosskopf ciencies of international hotels in Tai-
(2000) proposed the model of network wan, and divided service into the two
DEA to measure the efficiencies of stages of production process and market-
healthcare organizations in Sweden, ing process. The study results show that
covering middle efficiencies inside the the service qualities of hotel industry
organizations and resource allocation have been receiving more and more at-
conditions, which was close to practical tention in recent years, and the model
production processes. Lewis & Sexton for network DEA can clearly present the
(2004) took Major League baseball efficiencies of resource transformation
teams as the study objects, using the processes, and how the self-positioning
status of team funds as an input item, of a hotel can also influence its produc-
and results of games as an output item, tion efficiency or marketing efficiency.
dividing the operational efficiencies of
the teams into administrative efficien- To summarize, literature that
cies, immediate efficiencies, and overall adopted network DEA often divided the
efficiencies. The study compared the operational efficiencies of hotels into
analysis results of the network DEA and room service efficiencies and food &
the traditional DEA, and discovered that beverage service efficiencies when con-
the network DEA could provide more ducting a discussion. However, these
detailed information. Yu & Lin (2008) literatures failed to consider the relation-
adopted the model of network DEA, and ships between tourist guests and the
took the number of trains, railway lines, food & beverage service. Tourist guests
and number of employees as input items, often prefer to eat at the hotel they are
and transporting distance as an output staying at. Therefore, the number of
item to evaluate the passenger and goods tourist guests will influence the quality
transportation efficiencies of 20 railway of the food & beverage service. This
companies worldwide. study adds the number of tourist guests
as a middle input item at the second
Certain studies applied multi-stage stage.
DEA model of the hotel efficiencies in
the past, such as Huang, Ho, & Chiu Research Methods
(2014) adopted two-stage DEA to ana-
lyze the production efficiencies, room Analysis procedures
service efficiencies, and food & bever- This study took 46 international
age service efficiencies of 58 interna- five-star hotels in Taiwan as the research
tional hotels in Taiwan. The study re- objects, and established a network DEA
sults show that multistage DEA can pre- efficiency evaluation model to analyze
sent the middle efficiencies of all DMUs. food & beverage service efficiencies,
Hsieh & Lin (2010) divided the efficien- room service efficiencies, and the over-
cies of international hotels in Taiwan all operational efficiencies of hotels. It
into service efficiencies and catering ef- also compared the efficiency values of
ficiencies, and combining them with dif- DMUs to confirm whether there were
ferential variable analysis, provided significant differences due to the factors
suggestions on how to improve resource of international hotel chain brand, hotel
allocation to DMUs that failed to reach
68
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
location, and target customer groups. vice efficiency and room service effi-
Then, based on room service efficiency ciency as the coordinate axes to ascer-
and food & beverage service efficiency, tain the competitive positions of the
this study classified the competitive po- tourist hotels, which could provide
sitions of the international five-star ho- more detailed information for hotel
tels into four categories. The analysis operators, before concluding manage-
consisted of four steps, and the explana- rial implications.
tions are as follows.
Evaluation framework
1. The selection of input and output
items: this study referred to evaluation DEA uses production frontiers es-
indices used by previous studies, and tablished by the input and output data of
considered the availability of the data. all evaluated units to analyze the relative
The input and output data on the tour- efficiency of each DMU. This study re-
ist hotels comes from the 2013 Op- ferred to related previous studies, con-
eration Analysis Report of Interna- sidered practical production activities of
tional Hotels in Taiwan. tourist hotel industry, representativeness
2. The establishment of an efficiency of indices, and the availability of data,
evaluation model: based on the input before dividing the operational process
and output items selected, this study of the evaluated units into food & bever-
firstly used correlation analysis to ver- age service efficiency, room services'
ify whether the indices conformed to efficiency, and operational efficiency, as
the isotonicity hypothesis before es- shown in Figure 1. A total of nine indi-
tablishing a network efficiency ces were adopted, and detailed explana-
evaluation framework to evaluate the tions are as follows.
food & beverage service efficiencies,
(1) The number of guest rooms I1
room service efficiencies, and opera-
(room): the number of guest rooms
tional efficiencies of tourist hotels.
provided by each hotel can not only
3. Analysis of efficiency influencing fac- reflect the operational conditions of
tors: using the Kruskal-Wallis χ2 of the hotel industry, but also represent
nonparametric statistics, this study the flourishing conditions of the
took food & beverage service effi- tourism industry. Consequently, the
ciency, room service efficiency, and number of guest rooms belongs to
operational efficiency as the depend- the fixed costs of the production fac-
ent variables so as to find out factors tor (Wang, Lee, & Wong, 2007; Sun
that would significantly influence the & Lu, 2005; Wang, Hung, & Shang,
efficiencies of international five-star 2004).
hotels, including international hotel
chain brand, location, and countries of (2) Guest room employee number I2
major customer source, which can be (persons): formal employees hired
used as important references for hotel by a hotel to work in the rooms de-
operators when making decisions. partment. The hotel industry is a la-
4. Efficiency analysis results and con- bor-intensive industry and depends
clusions: based on the efficiency heavily on manpower to provide ser-
evaluation results of tourist hotels, vices. If the employee turnover rate
this study took food & beverage ser- is high, it will influence the opera-
69
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 tional management of the hotel. The (Wang, Shang, & Huang, 2006;
number belongs to the variable cost Wang, Hung, & Shang, 2004;
of production factor (Sun & Lu, Hwang & Chang, 2003).
2005; Wang, Hung, & Shang, 2004,
Hwang & Chang, 2003). (8) Food & beverage income O2 (thou-
sand NTD): practical income ob-
(3) Revenue expenditure I3 (thousand tained from Chinese and western-
NTD): the total revenue expenditure style food, desserts, wines, and
includes salaries and related ex- drinks provided by each hotel
penses, electricity and water and fuel (Wang, Lee, & Wong, 2007; Wang,
costs, depreciation expenses, renova- Hung, & Shang, 2004; Hwang &
tion and maintenance costs, and Chang, 2003).
laundry costs. It is a necessary input
for maintaining the normal operating (9) Tourist guest number L: the number
activities of a hotel (Wang, Lee, & of tourist guests for each hotel, in-
Wong, 2007; Sun & Lu, 2005; cluding domestic and foreign guests.
Hwang & Chang, 2003). international five-star hotels in Tai-
wan generally provide breakfast.
(4) Floorage of the food & beverage Therefore, the number of guests will
department I4 (square meters): the influence food & beverage service
floorage of the food & beverage efficiency (Huang, Ho, & Chiu,
space of a hotel represents the ac- 2014; Manasakis, Apostolakis, &
commodation capacity in the hotel Datseris, 2013; Wang, Shang, &
dining hall (Wang, Lee, & Wong, Huang, 2006; Barros & Alves, 2004).
2007; Sun & Lu, 2005; Wang, Hung,
& Shang, 2004).

(5) Food & beverage department em-

ployee number I5 (persons): formal
employees hired by a hotel to work
in the food & beverage department
(Hsieh & Lin, 2010).

(6) Food & beverage expenditure I6

(thousand NTD): direct raw material
costs and transportation miscellane-
ous costs of the food & beverage
service of each hotel (Anderson,
Fork, & Scott, 2000).

(7) Guest room income O1 (thousand

NTD): practical income of guest
rooms after deducting discounts.
Guests include individual guests,
group guests, contract guests and
long-term guests. This revenue is the
main income source of a hotel
70
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 I1
I2
I3 Rooms
•I Service • O
O1
1 1
•I • O
2 2
•I L
3
•I
4
•I I4 Food & O2
5 I5 Beverage
•I I6 Service
6

Operational Efficiency

Figure 1. Evaluation Framework

71
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 Editors Note: the following section is in single column format to facilitate easier
reading of the formulas.)

Network data envelopment analysis

Services of the tourist hotel industry contain the two aspects of rooms and food
& beverage. In order to highlight the characteristics of the hotel industry, this study
adopted the non-oriented model of SBM Network DEA which can simultaneously
consider the input difference and output difference when inquiry food & beverage
service efficiency, room service efficiency, and overall operational efficiency. As
*
shown in formula (1), when Ek = 1, it indicates that the evaluated unit k reaches the
*
efficient state, otherwise, when Ek <1, it indicates that the evaluated unit k fails to
reach the efficient state, and the closer the value is to 0, the poorer the efficiency.
 1  md sikd − 
D d
1 − w
  d 
　*
Min Ek = min 　 　
d =1

D d
 md  i =1 xik 

d +  (1)
1  rd srk 
 w 1 +
d 


d =1
 rd  i =1 y rk 

　　x = X λ + s
s.t. d
k
d d d−

　　　y = Y λ − s
d
k
d d d+

　　　λ , s , s ≥ 0, 　d = 1,2
d d− d+

　w
D

 w d = 1, d
≥ 0
Among: wd is the weight of stage of d,
d =1

md is the number of input of stage of d.

rd is the number of output of stage of d.

s ikd − is the amount of difference in the ith input item of evaluated unit k in

stage of d.
th
d + is the amount of difference in the r output item of evaluated unit k in
s rk
stage of d.
th
y rkd is the amount of input in the i input of evaluated unit k in stage of d.
th
xikd is the amount of output in the r output of evaluated unit k in stage of d.

xkd is the amount of input of evaluated unit k in stage of d.

ykd is the amount of output of evaluated unit k in stage of d.

d− 72
Thes International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 is the overall difference amount of stage of d.

s d + is the overall of amount of excess of stage of d.

X d is the total input amount of stage of d. X = ( x1 ,..., xn ) ∈ R d

d d d m ×n

Y d is the total output amount of stage of d. Y = ( y1 ,..., yn ) ∈ R d

d d d r ×n

λd is the nonnegative number of stage of dither efficiency values for food

& beverage service and room service can be obtained through formula (2).When
E kd
= 1, it indicates that the evaluated unit k at stage d reaches an efficient state;
d
when E k <1, it indicates that the evaluated unit k at stage d fails to reach an efficient
state, and the closer the efficiency value is to 0, the poorer the efficiency indicated.

1  md sikd − 
1−  d 
md  i =1 x 
d  ik  (2)
Ek =
1  rd s d +
1 +   rkd 
rd  i =1 y rk 
This study used the number of hotel guests as the inputs items of intermediate
outputs. It is a variable value. Therefore, this study added a new limit type, shown in
formula (3).

Z ( k ,h ) λh = Z ( k ,h ) λk (∀(k , h)) (3)

t
( k ,h ) × n
Among: Z ( k ,h ) = ( z1( k ,h ) ,⋅ ⋅ ⋅, z n( k ,h ) ) ∈ R

Research Results sion performance, food & beverage per-

formance, and operational performance.
Efficiency evaluation results The descriptive statistics of the raw data
is shown in Table 3.
This study used Network SBM
DEA proposed by Tone & Tsutsui (2009) (Editor's Note: Please see all the rest of
as research method, and adopted nine the Tables and the Appendix at the end
input and output indices to analyze the of this article.)
performance of 46 international five-star
hotels in Taiwan, including rooms divi-
73
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 In general, due to the differences in guests from Mainland China, while the
location and scale, the 46 DMUs have majority of the guests at the Caesar Park
large differences in resource inputs with Hotel Taipei, also located in Taipei,
regards to the rooms department and the come from Japan. The difference in the
food & beverage department, with the number of guests for all DMUs can
numbers of employees working in the reach 206,000.
rooms department varying from 46 to
257, and an average of 109. The Lalu This section presents the analysis
Hotel has 96 rooms and is the smallest. results of the relative performance of the
The Grand Hyatt Taipei Hotel located in 46 DMUs, including rooms' division
a flourishing area of Taipei has 865 performance, food & beverage perform-
rooms. The Le Meridien Taipei and ance, and operating performance. Table
Sheraton Grande Taipei Hotel with 692 2 shows the performance evaluation re-
rooms has the largest revenue expendi- sults of international five-star hotels in
ture, reaching NTD 2.13 billion, and the Taiwan. A total of eight DMUs have an
Hotel Royal Chihpen has the least reve- operational efficiency of 1, with their
nue expenditure of NTD 196 million. rooms' service and food & beverage ser-
vice also reaching an efficient state. This
The average number of employees group of DMUs can be divided into two
in food & beverage departments across categories. One can effectively refer to
the 46 five-star hotels is 181, with an the scale economies effect, using more
average food & beverage floorage of input resources than the industrial stan-
4780 square meters. The Westin Taipei dard to create benefits that shareholders
has the largest food & beverage floorage, are satisfied with. This category contains
and the Hotel Royal Chiaohsi has the H4, H7, and H8. The other category can
smallest, with the floorage difference use fewer resources than the industrial
between the two hotels reaching 30,000 standard to create the maximum benefits.
square meters. The difference is also re- This category contains H12, H27, H29,
flected by the food & beverage costs, H31, and H32. The three DMUs with the
with the standard deviation reaching poorest performance are H17(0.5283),
NTD137 million. H37(0.5326), and H23(0.5517). When
exploring the reasons for poor perform-
Different DMUs have different ance, it was discovered that these three
market positions, and the benefits cre- international five-star hotels have com-
ated are different. The Le Meridien paratively poor rooms' efficiencies, al-
Taipei and Sheraton Grande Taipei Ho- though their food &beverage perform-
tel has the largest guest room income, ance is just slightly under the average of
while the Regent Taipei Hotel has the the 46 DMUs. Therefore, in order to en-
largest food & beverage income. How- hance their overall operational efficien-
ever, on the whole, the average food & cies, they need to start by improving
beverage income of the five-star hotels their room service.
is 46 million NTD higher than the aver-
age guest room income. With respect to With respect to rooms' efficiency,
guest sources and differences of location eight DMUs reached an efficient state.
and transportation, there are differences The average rooms' income (529 million
in guest demographics. For example, the NTD) of the DMUs of this group was
Grand Hotel has comparatively more higher than the average amount (333
74
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
million NTD) of all the other DMUs. all DMUs. Depending on the input and
Depending on input levels, this group output levels, this group can further be
can further be divided into two catego- divided into two categories. The first
ries. The first category has a resource category has a resource input smaller
input lower than the average for the in- than the average for the industry. This
dustry. This category contains H27, H29, category contains H27, H29, H31, H32,
H31, and H32. These international ho- H26, H45, and H33. These international
tels can effectively use limited guest hotels can effectively use limited food &
rooms, employees, and revenue expendi- beverage resources to create high food
ture for the rooms departments whilst & beverage incomes. The second cate-
creating maximum guest room income. gory has higher food & beverage in-
The second category has higher output comes than the average for the industry.
benefits than the average for the industry. This category contains H4, H7, H8, H12,
This category contains H4, H7, H8, and H10, and H2. These international five-
H12. These international hotels aim to star hotels can create high food & bever-
create the highest occupancy rates. Ex- age incomes. Although these hotels in-
cept for H12, the other three DMUs put more food & beverage resources
have invested more in guest room re- than the average standard, the benefits
sources than the industrial standard, but they create are higher than the average
in so doing they have created more standard of the industry. The minimal
benefits, even three times as many bene- three DMUs in food & beverage per-
fits when compared to the average for formance are H41(0.5187), H44(0.6005),
the industry. The minimal three DMUs and H6(0.6281). Of these, the food &
for rooms division performance are H22 beverage service resources input by H16
(0.3899), H37 (0.3906), and H17 are more than the average industry stan-
(0.3978). As well as three international dard. However, H16 fails to effectively
five-star hotels having fewer inputs in respond to the effects of scale economy.
rooms' resources than the average indus- The other two DMUs input fewer re-
try standard, the output benefits created sources than the average standard for the
by them are also far lower than the in- industry, but the output benefits they
dustry standard. This might be related to have created are far lower than the aver-
limited rooms' resources, including age industry standard. They have limited
manpower and the number of guest food & beverage resources, including
rooms. Among them, the number of manpower and food & beverage floor-
guest visits for H60 is only half of the age. In order to enhance food & bever-
average industry standard. In order to age performance, hotels need to provide
improve rooms' division performance, special food to increase the number of
hotels should adopt effective marketing guests who receive the food & beverage
and promotion methods to enhance guest service and improve the guest turnover
numbers and guest room income. rate.

With respect to food & beverage Comparison of the efficiency results

performance, the performance values for
15 DMUs were 1. The average food & This study used the method pro-
beverage income (530 million NTD) of posed by Tone & Tsutsui (2009) to dis-
this group of DMUs is higher than the cuss whether the efficiencies of room
average income (379 million NTD) of service and food & beverage service
75
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
have a linking effect. It divides the DMUs have a performance efficiency of
original values of the middle input and 1. Using a single performance value
output items by the targeted improve- cannot fully explore the competitive
ment value. If the result is 1, it means strengths and weaknesses of a DMU.
the middle input and output items have Efficiency information obtained by
reached the optimal allocation. The lar- original SBM DEA has poor discrimina-
ger the value, the higher the improve- tion power. Therefore, it is necessary to
ment margin. The results are shown in use the network structure to discuss the
Figure 2. performance of middle efficiencies.

The middle input and output items Figure 4. shows the comparisons
for this study are the number of guests. between rooms' division performance,
The results show that the top three for food & beverage performance, and op-
the linking effect are H37(2.305), erational performance. The results show
H35(2), and H9(1.786). It means that that dividing the overall performance of
these three international five-star hotels the 46 DMUs into three aspects can help
need to further explore the position of to highlight the competitive strengths
insufficient guests. They can consider and weaknesses of each international
enhancing the attractions of the hotels five-star hotels, and obtain more infor-
by aligning themselves with other indus- mation on resource allocation and opera-
tries or combining with local cultural tional management. For example, H22
activities, thus increasing the number of has good resource applying abilities in
guests. This study used the method pro- the food & beverage department, and
posed by Tone & Tsutsui (2009), to serves special meals. Therefore, its food
compare the overall efficiency values & beverage performance is the best.
obtained by single-stage SBM DEA to However, its occupancy rate is compara-
those obtained by SBM Network DEA. tively low. The hotel could consider
The analysis results were compared for launching a guest room and food & bev-
rooms' division performance, food & erage package to increase occupancy
beverage performance, and operational rates and guest room income. The bene-
performance to confirm whether these fit for H6 mainly comes from guest
performance values experience the net- rooms because the hotel has a good loca-
work effect, before their differences tion and high-quality guest rooms and
were compared. Single-stage SBM DEA facilities. Its average annual occupancy
has six input items, namely the number rate is higher than 70%. However, the
of employees in the rooms department, hotel still has space for improvement in
the number of guest rooms, operating its food & beverage service.
costs, the number of employees in the
food & beverage department, food & This study further compared the
beverage costs, and food & beverage efficiency evaluation results of the 46
floorage. The output items were food & international five-star hotels depending
beverage income, guest room income, on whether a hotel chain brand, the loca-
and the number of guests. Figure 3 tion, and the nationalities of the guests.
shows the comparison results of the per- The results are shown in Table 3. The
formance values obtained by SBM DEA results of the non-parametric statistics
and by SBM Network DEA. It shows test show that international five-star
that by means of SBM DEA, up to 22
76
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 Figure 2. Comparison of the Linking Effect

Figure 3. Comparisons between SBM DEA and SBM Network DEA

77
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 Figure 4. Comparisons of Three Performance Values

Figure 5. The Decision-making Matrix of the 46 International Five-star Hotels

78
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
hotels belonging to international hotel products to launch special foods,
chain brands demonstrate better per- launch limited foods depending on
formance in three kinds of efficiencies different seasons, create a special
than those hotels that are not part of an dining environment, or hire famous
international chain brand. The differ- chefs to improve the standards of
ences in occupancy rates are especially their food & beverage. They are the
significant. The operational efficien- followers of the five-star hotel in-
cies, rooms efficiencies, and food & dustry.
beverage efficiencies in seven areas all
show significant differences. On the 3. Low rooms' efficiency but high food
whole, international five-star hotels & beverage efficiency: this group
located at scenic area have the best ef- includes the eight DMUs of H1, H20,
ficiencies. With respect to guest H22, H25, H30, H33, H35, and H40.
sources, the performance of the three They have outstanding food & bev-
efficiencies all show significant differ- erage service capabilities, but their
ences. International five-star hotels guest room service capabilities need
with the main guest source from North to be improved. They are hotels with
America have the best efficiencies. potential in the five-star hotel indus-
try. In order to increase rooms' effi-
The decision-making matrix ciency, these DMUs could launch a
guest room and food & beverage
1. High rooms' efficiency and high package, actively participate in na-
food & beverage efficiency: DMUs in tional traveling exhibition, launch
this area include H2, H4, H7, H8, H10, guest room discounts irregularly,
H11, H12, H26, H27, H28, H29, H31, cooperate with local businesses, and
H32, and H45. The room service effi- create local events to attract more
ciencies and food & beverage service guests.
efficiencies of these 14 DMUs are
higher than the industry standard. They 4. Low rooms' efficiency and low
are the leading hotels of Taiwan’s five- food & beverage efficiency: a total
star hotel industry. In order to maintain of 16 DMUs fall into this group.
their competitive status, the DMUs of Their overall service capacities need
this group need to maintain their atten- to be strengthened. They are the ho-
tion to service quality, stabilize their tels with little potential in the five-
core competitiveness, and monitor en- star hotel industry. The DMUs of
vironmental changes at all times in or- this group need to re-examine the al-
der to adjust their operating strategies location and usage conditions of
corresponding to market trends. their tangible resources, such as
manpower, funds, and site, or invite
2. High rooms' efficiency but low food an external consulting group to un-
& beverage efficiency: this group dertake an examination, and find the
includes the eight DMUs of H3, H5, competitive weakness and operating
H6, H9, H36, H42, H43, and H44. problems of the hotels, thus enhanc-
They have outstanding room service ing their overall service ability.
capabilities, but their food & bever- Conclusions and Suggestions
age services need to be improved.
For example, they could use local
79
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 This study took international This study has discovered that,
five-star hotels in Taiwan as DMUs, due to the brand enchantment and
adopted the SBM Network DEA pro- abundant operating experience of an
posed by Tone & Tsutsui (2009), and international team, international five-
established a multistage efficiency star hotels belonging to an interna-
evaluation model to enable evaluation tional hotel chain brand demonstrate
of the rooms' efficiencies, food & bev- better performance in room service
erage service efficiencies, and overall efficiency, food & beverage service
operational efficiencies of the DMUs. efficiency, and overall operational effi-
It also compared the efficiency value ciency than those DMUs that are not
differences of the DMUs with regards connected to an international hotel
to international hotel chain brand, loca- chain brand. International five-star ho-
tion, and the nationalities of the main tels located at scenic spots utilize the
guests. Aims and implementation sightseeing features of the locality
methods are explained as follows. when conducting their brand marketing.
As these hotels have fewer rooms and
1. The network framework can high- higher prices, they can provide high-
light the transformation process of the quality services to customers. In addi-
middle resources. tion, international five-star hotels
whose main guest source is from North
During practical service proce- America have more than 800 guest
dures, from resource input to substan- rooms, distinctive styles of dining en-
tial output, there will be several sub- vironment, and special foods. They
production activities. A traditional ef- give full attention to the scale economy
ficiency evaluation model can only effect.
present the performance of the overall
operating efficiency, and the efficien- To summarize, this study pro-
cies of internal sub-production activi- poses the following three suggestions
ties cannot be presented. Establishing for future studies.
an efficiency evaluation model for the
network framework can not only pre- (1) The network efficiency evaluation
sent the resource transformation proc- framework is not limited to the ho-
esses of a DMU at different stages, but tel industry. Other industries can
also present the competitive strengths consider the processes behind their
and weaknesses of each DMU. The production activities and the char-
model can enhance the discriminating acteristics of evaluated units to es-
power of the analysis results, and pro- tablish multistage efficiency
vide improvement directions as refer- evaluation models, and present the
ences for decision makers when mak- efficiency performances at multiple
ing resource allocation decisions and stages of the evaluated units.
generating management strategies.
(2) This study collected official sec-
2. There are significant differences in ondary data, and adopted a non-
the efficiencies of the DMUs of differ- oriented efficiency evaluation
ent operating types. framework that can simultaneously
consider the input and output
spread. This study suggests that fu-
80
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 ture studies should use question- ductivity in a Portuguese Retail
naire surveys to acquire satisfac- Chain using the Malmquist In-
tion information that cannot be pre- dex. Journal of Retailing and
sented by the secondary data, and Consumer Services, Vol.5, 269-
use it as an output item of the 278.
DMU.
Barros, C. P. & Dieke, P. U. C. (2008)
(3) This study aims to establish a mul- Technical Efficiency of African
tistage network efficiency evalua- Hotels. International Journal of
tion model. This study used one- Hospitality Management, Vol.
year static data as analysis exam- 27, 438-447.
ples. Further studies can use the
Malmquist Productivity Index to Bureau of Tourism, Ministry of Trans-
analyze the long-term efficiency portation & Communications.
changes of the DMUs, and use (2015) Operation Analysis Re-
them as important references for port of International Hotels in
establishing operational targets and Taiwan. Tourism Service Center,
generating operating strategies. Bureau of Tourism, Ministry of
Transportation and Communica-
tions.
References
Charnes, A., Cooper, W. W. & Rho-
Anderson, R. I., Fork, R. & Scott, J. des, E. (1978). Measuring the Ef-
(2000). Hotel Industry Effi- ficiency of Decision Making
ciency: an Advanced Linear Pro- Units. European Journal of Op-
gramming Examination. Ameri- erational Research,Vol.2, 429-
can Business Review, Vol.18, 444.
40-48.
Chen, H., Lu, Y. C. & Chung, J. T.
Assaf, A. G. & Magnini, V. (2012). (2010). Improved Slack-Based
Accounting for Customer Satis- Context-Dependent DEA-A
faction in Measuring Hotel Effi- Study of International Tourist
ciency: Evidence from the US Hotels in Taiwan. Expert Sys-
hotel industry. International tems with Applications, Vol.37,
Journal of Hospitality Manage- 6452-6458.
ment, Vol.31, 642-647.
Chiang, W. E., Tsai, M. H. & Wang, L.
Banker, R. D., Charnes, A. & Cooper, S. M. (2004). Research Note: A
W. W. (1984). Some Models for DEA Evaluation of Taipei Ho-
Estimating Technical and Scale tels. Annals of Tourism Re-
Inefficiencies in Data Envelop- search, Vol.31, 712-715.
ment Analysis. Management Sci-
ence, Vol.30, 1078-1092. Färe, R. & Grosskopf, S. (2000). Net-
work DEA. Socio-Economic
Barros C. P., & Alves, C. A. (2004). Planning Sciences, Vol.34, 35-
An Empirical Analysis of Pro- 49.

81
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
Fatma, P. & Timothy, H. (2005). Pa- Seiford, L. M. & Zhu, J. (1999). Prof-
tient Satisfaction in a Preopera- itability and Marketability of the
tive Assessment Clinic: An Top 55 US Commercial Banks.
Analysis Using SERVQUAL Management Science, Vol.45,
Dimensions. Total Quality Man- 1270-1288.
agement & Business Excellence,
Vol.16, 15-20. Sun, S. & Lu, W. M. (2005). Evaluat-
ing the Performance of the Tai-
Hsieh, L. F. & Lin, L. H. (2010).A Per- wanese Hotel Industry using a
formance Evaluation Model for Weight Slacks-based Measure.
International Tourist Hotels in Asia-Pacific Journal of Opera-
Taiwan-An Application of the tional Research, Vol.22, 487-512.
Relational Network DEA. Inter-
national Journal of Hospitality Tone, K. & Tsutsui, M. (2009). Net-
Management, Vol.29, 14-24. work DEA: A Slacks-based
Measure Approach. European
Huang, C. W., Ho, F. N. & Chiu, Y. H. Journal of Operational Research,
(2014) Measurement of Tourist Vol.197, 243-252.
Hotels’ Productive Efficiency,
Occupancy and Catering Service Wang, F. C., Hung, W. T. & Shang, J.
Effectiveness using a Modified K. (2004). Measuring the Effi-
Two-stage DEA Model in Tai- ciency of China and Non-chain
wan. OMEGA, The International International Tourist Hotel in
Journal of Management Science, Taiwan. Asia-Pacific Economic
Vol.48, 49-59. and Management Review, Vol.7,
109-123.
Hwang, S. N. & Chang, T. Y. (2003).
Using Data Envelopment Analy- Wang, F. C., Shang J. K. & Hung, W.
sis to Measurement Hotel Mana- T. (2006). Productivity and Ser-
gerial Efficiency Change in Tai- vice Quality Changes in Interna-
wan. Tourism Management, tional Hotels in Taiwan. Annals
Vol.24, 357-369. of Tourism Research, Vol.33,
571-574.
Lewis, H. F. & Sexton, T. R. (2004).
Network DEA: Efficiency Wang, Y. H., Lee, W. F. & Wong, C.
Analysis of Organization with C. (2007). Productivity and Effi-
Complex Internal Structure. ciency Analysis of International
Computers & Operations Re- Tourist Hotels in Taiwan: an Ap-
search, Vol.31, 1365-1410. plication of the Stochastic Fron-
tier Approach. Taiwan Economic
Manasakis, C., Apostolakis, A. & Review, Vol.35, 55-86.
Datseris, G. (2013). Using Data
Envelopment Analysis to Meas- Wassenaar, K. & Stafford, E. R.
ure Hotel Efficiency in Crete. In- (1991). The Lodging Index: An
ternational Journal of Contempo- Economic Indicator for the Ho-
rary Hospitality Management, tel/Motel Industry. Journal of
Vol.25, 510-535. Travel Research, Vol.30, 18-21.
82
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
Yang, C. & Lu, W. M. (2006). Per-
formance Benchmarking for
Taiwan’s International Tourist
Hotels. INFOR, Vol.44, 229-245.

Yu, M. M. & Lee, B. C. (2009). Effi-

ciency and Effectiveness of Ser-
vice Business: Evidence from In-
ternational Tourist Hotels in
Taiwan. Tourism Management,
Vol.30, 571-580.

Yu, M. M. & Lin, E. J. T. (2008). Effi-

ciency and Effectiveness in
Railway Performance Using a
Multi-Activity Network DEA
Model. International Journal of
Management Science, Vol.36,
1005-1017.

83
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 Table 1. Description of Raw Data

Index Max. Min. Avg. SD

The Number of
865 96 325 161
Guest Rooms I1

Guest Room
Employee Number 257 46 109 55
I2

Revenue
2,130,638,474 195,688,361 611,154,667 426,069,641
Expenditure I3

Floorage of the
Food & Beverage 32,045 586 4,780 5,011
Department I4

Food & Beverage

Department
516 32 181 118
Employee number
I5

Food & Beverage

613,686,250 27,409,718 151,731,750 136,708,317
Expenditure I6

Guest Room
1,066,689,012 86,201,435 332,870,503 208,835,868
Income O1

Food & Beverage

1,564,468,209 69,770,691 379,175,047 328,049,084
Income O2

Tourist Guest
1,471,149 56,928 171,123 205,941
Number L

84
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 Table 2. Performance Evaluation Results for International Hotels in Taiwan

DMU Operation Rooms Ef- Food & DM Operation Rooms Effi- Food &
Efficiency ficiency Beverage U Efficiency ciency Beverage
Efficiency Efficiency
H1 0.7275 0.5327 0.9027 H24 0.6979 0.5891 0.7047
H2 0.8881 0.7708 1 H25 0.8721 0.5731 0.9841
H3 0.7188 0.705 0.71 H26 0.9779 0.9567 1
H4 1 1 1 H27 1 1 1
H5 0.8572 0.8428 0.8332 H28 0.9506 0.9431 0.8855
H6 0.7995 0.7322 0.6281 H29 1 1 1
H7 1 1 1 H30 0.8489 0.6761 0.8393
H8 1 1 1 H31 1 1 1
H9 0.8732 0.8542 0.7642 H32 1 1 1
H10 0.968 0.8606 1 H33 0.8248 0.6506 1
H11 0.9149 0.7791 0.9828 H34 0.6799 0.5465 0.7924
H12 1 1 1 H35 0.7154 0.5581 0.8499
H13 0.5688 0.403 0.7277 H36 0.7975 0.7196 0.825
H14 0.7096 0.6123 0.763 H37 0.5326 0.3906 0.7882
H15 0.6258 0.4662 0.765 H38 0.6992 0.6009 0.7028
H16 0.6711 0.5363 0.6705 H39 0.5803 0.4839 0.7041
H17 0.5283 0.3978 0.7111 H40 0.7055 0.524 1
H18 0.6147 0.5388 0.6653 H41 0.5953 0.5626 0.5187
H19 0.5614 0.4187 0.6302 H42 0.8005 0.7398 0.657
H20 0.8192 0.573 0.9639 H43 0.7565 0.6922 0.7119
H21 0.5801 0.5204 0.6895 H44 0.7197 0.6988 0.6005
H22 0.6398 0.3899 1 H45 0.9872 0.8141 1
H23 0.5517 0.4264 0.7394 H46 0.5748 0.4427 0.6433

85
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 Table 3. Analysis of Performance Differences of International
Hotels of Different Types

Type Item Avg. Avg. Rooms Avg. Food &

Operational Efficiency Beverage
Efficiency Efficiency
International Hotel
Chain Brand
Yes 0.8383 0.7809 0.8534

No 0.7587 0.6476 0.8260

2
Kruskal-Wallis χ 2.489 4.171** 0.536
Taipei 0.8956 0.8398 0.9018
Taoyuan,
Hsinchu, and
Location Miaoli 0.7100 0.5731 0.8511
Taichung 0.6501 0.4755 0.8209
Kaohsiung 0.6197 0.4924 0.7171
Hualien 0.7072 0.5295 0.8094
Others 0.7132 0.6177 0.7265
Scenic Spots 0.9682 0.9394 0.9607
Kruskal-Wallis χ2 28.138*** 29.504*** 16.211**
Japan 0.8948 0.8644 0.8816
North America 1.0000 1.0000 1.0000
Main Guests Taiwan 0.7348 0.6314 0.7969
Asia 0.9026 0.7634 0.9943
Mainland
China 0.8574 0.6734 0.9514
2
Kruskal-Wallis χ 10.789** 10.927** 7.943*
Note: *** represents P<0.01, **represents P<0.05, and *represents P<0.1

86
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
 Appendix

DMU Name DMU Name

H1 The Grand Hotel H24 Parkview Hotel
H2 Ambassador Hotel Taipei H25 Hualien FarGlory Hotel
H3 The Landis Taipei Hotel H26 Hotel Royal Chiaohsi
H4 Sheraton Grande Taipei Hotel H27 The Lalu Hotel
H5 Hotel Royal-Nikko Taipei H28 Fleur De Chine Hotel
H6 The Howard Plaza Hotel Taipei H29 Caesar Park Hotel Taipei
H7 Grand Hyatt Taipei H30 Howard Beach Resort Kenting
H8 Regent Taipei Hotel H31 Hotel Royal Chihpen
H9 The Sherwood Taipei H32 Silks Place Taroko Hotel
Shangri-La's Far Eastern Plaza Monarch Skyline International Ho-
H10 H33
Hotel Taipei tel
Hotel Novotel Taipei Taoyuan In-
H11 The Westin Taipei H34
ternational Airport
H12 Le Méridien Taipei H35 Sheraton Hsinchu Hotel
H13 Ambassador Hotel Kaohsiung H36 Hotel Royal Hsinchu
H14 The Grand Hi Lai Hotel) H37 Ambassador Hotel Hsinchu
The Howard Plaza Hotel Kaoh- Fullon Hotel Tamsui Fishermen's
H15 H38
siung Wharf
H16 85 Sky Tower Hotel H39 Tayih Landis Hotel Tainan
H17 Han-Hsien International Hotel H40 Evergreen Plaza Hotel Tainan
Shangri-La's Far Eastern Plaza Ho-
H18 The Lees Hotel H41
tel
H19 Hotel National H42 E-Da Royal Hotel
H20 Evergreen Laurel Hotel Taichung H43 Silks Place Yilan Hotel
H21 Howard Prince Hotel Taichung H44 Evergreen Resort Hotel Jiaosi
H22 The Splendor Hotel Taichung H45 Formosan Naruwan Hotel Taitung
H23 Chateau de Chine Hualien H46 Nice Prince Hotel

87
The International Journal of Organizational Innovation Vol 10 Num 3 January 2018
Copyright of International Journal of Organizational Innovation is the property of Frederick
L. Dembowski and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.