O.R. Applications [625029]

O.R. Applications
Measuring the national competitiveness of Southeast
Asian countries
Chiang Kaoa,*, Wann-Yih Wub, Wen-Jen Hsiehc, Tai-Yue Wanga,
Chinho Lina, Liang-Hsuan Chena
aDepartment of Industrial and Information Management, National Cheng Kung University, Tainan 70101, Taiwan, ROC
bDepartment of Business Administration, National Cheng Kung University, Tainan 70101, Taiwan, ROC
cDepartment of Economics, National Cheng Kung University, Tainan 70101, Taiwan, ROC
Received 3 April 2006; accepted 27 March 2007
Available online 19 April 2007
Abstract
National competitiveness is a measure of the relative ability of a nation to create and maintain an environment in which
enterprises can compete so that the level of prosperity can be improved. This paper proposes a methodology for measuring
the national competitiveness and uses the 10 Southeast Asian countries for illustration. The basic idea is to deconstruct the
complicated concept of national competitiveness to measurable criteria. The observations (data) on the criteria are thenaggregated according to their importance to obtain an index of national competitiveness. For data collected from ques-
tionnaire surveys, a calibration technique has been devised to alleviate bias due to personal prejudice. In data aggregation,
the importance is expressed by both a priori weights and a posteriori weights. These two types of weights consistently show
that Singapore, Malaysia, and Thailand have the highest national competitiveness, while Myanmar, Cambodia, and Laosare the least competitive countries. The performance of each country in every criteria measured also provides directions for
these countries to make improvements and for investors to allocate resources.
/C2112007 Published by Elsevier B.V.
Keywords: National competitiveness; Multiple criteria; Composite indicator; A priori weight; A posteriori weight; Southeast Asia
1. Introduction
In an increasingly open and integrated world
economy, competitiveness has become a central pre-
occupation of both advanced and developing coun-tries ( Porter, 1990 ). Interestingly, there seems to be
no agreed definition of national competitiveness(Krugman, 1996 ). Scholars of different disciplinesusually look at the problem from different points
of view ( Buckley et al., 1988 ). Competitiveness at
the company level is clear, where companies com-
pete for markets, and it is measured by looking at
market shares or profitability. Competitiveness atthe country level has been assumed to be similar.Unfortunately, market shares fail to give insightsinto countries’ balance of trade and economicstrength through their failure to consider imports(Krugman and Hatsopoulus, 1987 ). Moreover,
market shares ignore sales arising from foreign
0377-2217/$ – see front matter /C2112007 Published by Elsevier B.V.
doi:10.1016/j.ejor.2007.03.029*Corresponding author. Tel.: +886 6 2753396.
E-mail address: [anonimizat] (C. Kao).Available online at www.sciencedirect.com
European Journal of Operational Research 187 (2008) 613–628
www.elsevier.com/locate/ejor

affiliates and foreign licensed sales, since only
exports are considered. Regarding profitability, thestudy of Blaine (1993) on Japanese and American
firms in the 1980s indicates that highly profitable
firms do not necessarily lead to highly competitive
industries or countries, and the opposite is also true.Many American firms have remained extremelyprofitable despite the declining strength of boththeir industries and the US economy as a whole;conversely, most Japanese firms have remainedrelatively unprofitable despite the growing com-petitiveness of the Japanese economy. For these
reasons, competitiveness at the country level cannot
be measured the same way as that at the companylevel.
How then do we define competitiveness at the
country level? The OECD’s (1996) definition is
‘‘the degree to which a country can, under freeand fair market conditions, produce goods and ser-vices which meet the test of international markets,
while simultaneously maintaining and expanding
the real income of its people over the longer term.’’Scott and Lodge (1985) refer to national competi-
tiveness as ‘‘a country’s ability to create, produce,distribute, and/or service products in internationaltrade while earning rising returns on its resources.’’The Institute for Management Development ( IMD,
2003) defines competitiveness of nations as ‘‘how
nations create and maintain an environment which
sustains the competitiveness of its enterprises.’’The definition from the World Economic Forum(WEF, 2003 ) is ‘‘the set of institutions and economic
policies supportive of high rates of economic growthin the medium term.’’ While these definitions are notexactly the same, they share a common spirit, thatis, creation of an environment conducive to improv-
ing the prosperity of a country. It is this broad def-
inition that has attracted considerable attentionfrom policy makers, enterprises, and the public,and rankings based on the spirit of this definitionregularly appear in policy statements and the media.
Many indicators, models, and indices have been
proposed to measure national competitiveness. Indi-cators such as the relative price or cost indices pub-
lished by the IMF and the OECD show various
levels of countries’ competitiveness. Boltho (1996)
believes the real exchange rate is a short-term mea-sure of competitiveness and trend productivitygrowth a long-term measure. The most appropriateindicator of competitiveness, according to Porter
(1990) , would be total productivity. Buckley et al.
(1988) study a wide variety of publications on com-petitiveness and conclude that single measures of
competitiveness do not capture all the elements ofthe concept. Useful measures should encompasscompetitive performance, its sustainability through
the generation of competitive potential, and the
management of the competitive process. Fagerberg
(1996) also finds that most analysts use a broader
definition of competitiveness and focus on struc-tural factors affecting medium to long-term eco-nomic performance: productivity, innovation,skills, and so on.
Porter (1990) develops the diamond model
through studying competitive performance among
10 countries. The model is based on four country-specific determinants: factor conditions, demandconditions, related and supporting industries, andfirm strategy, structure, and rivalry, and two exter-nal variables: chance and government. This modelhas been widely applied in studying the competitive-ness of different countries ( Bellak and Weiss, 1993;
Hodgetts, 1993 ). However, it is flawed in some
aspects. Dunning (1993) argues that this model
underestimates the significance of globalizationand markets for the competitive advantage ofnations. Grant (1991) finds that the breadth and rel-
evance of Porter’s analysis have been achieved at theexpenses of precision and determinancy – its empir-ical data are chosen selectively and interpreted sub-
jectively. In studying the Austrian economy, Bellak
and Weiss (1993) point out that Porter’s framework
of analysis on competitiveness has shortcomings forsmall, open economies.
In addition to Porter’s diamond model, there are
two leading indices that measure national competi-tiveness. One is prepared by the IMD and appearsin the World Competitiveness Yearbook , and the
other is contained in the Global Competitiveness
Report of the WEF. The former uses approximately
300 criteria to rank 60 countries, while the latteruses approximately 170 variables to rank 117 coun-tries. Note that the number of criteria for the twoindices differs from year to year and the numberof countries being ranked has been increasing overthe years. Both indices rely on evidence-based hard
data and opinion-based softdata. The major differ-
ence between these two indices is that the WEFplaces greater reliance on soft data (around two-thirds), while for the IMD this is reversed. Lall
(2001) points out that the Global Competitiveness
Report has deficiencies at several levels. It suffers
from several analytical, methodological, and quanti-tative weaknesses. For example, some of its implicit614 C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628

premises are suited to advanced countries, rather
than equally applicable to less developed ones.The economic model that supposedly underlies ithas little to do with its empirical approach. And
there are many variables on which hard quantitative
data are available but questionnaire surveys areused instead. As a matter of fact, these weaknessesare also shared with the World Competitiveness
Yearbook . In addition to these shortcomings, both
publications concentrate on more developed coun-tries. For example, in World Competitiveness Year-
book there is a criterion of the number of Nobel
prize winners, in which is almost impossible for
underdeveloped countries to score above zero.Moreover, the complexity of the reports makes ithard to generate the model for those countries notselected ( Oral and Chabchoub, 1996; Lall, 2001 ).
The basic idea of these studies is to use composite
indicators to integrate large amount of informationinto easily understood formats for a general audi-
ence. A composite indicator is formed when individ-
ual indicators are complied into a single index onthe basis of an underlying model. It must be care-fully constructed to avoid a result which is mislead-ing. The Joint Research Center (JRC) of theEuropean Commission has an information server(JRC website ) which contains a multitude of refer-
ences regarding the issue of composite indicator
construction (e.g., Freudenberg, 2003; Munda and
Nardo, 2003; Nardo et al., 2005 ). In this paper,
we propose a methodology based on the idea ofcomposite indicator to calculate nationalcompetitiveness.
Southeast Asia has 10 countries: Brunei, Cambo-
dia, Indonesia, Laos, Malaysia, Myanmar, the Phil-ippines, Singapore, Thailand, and Vietnam, with a
total area of more than 4 million square kilometers
and a total population of more than 530 million.Because of foreign direct investment from the EastAsian countries, the area has experienced an ‘‘Asianmiracle’’ and emerged as a fast-growing region ofthe world ( Ku, 2002 ). This region has abundant nat-
ural resources and a sufficient and cheap laborforce. Since the creation of the ASEAN Free Trade
Area (AFTA), this region has experienced fast eco-
nomic development. Competitiveness has long beenconsidered a matter of national economic survivalby these countries ( Lall, 2001 ). Regrettably, not all
of these 10 countries have been included in eitherthe WEF or IMD indices. For these reasons, thispaper proposes a methodology for calculatingnational competitiveness, and data from the 10Southeast Asian countries are collected for
illustration.
In the followings, firstly, a methodology for cal-
culating national competitiveness for the Southeast
Asian countries is proposed. The methodology con-
tains three parts, criteria for measuring nationalcompetitiveness, calibration of data, and aggrega-tion of the performance on all criteria. Secondly,we describe how to collect data and calculate thenational competitiveness for the 10 Southeast Asiancountries. Finally, we compare the results withthose of IMD and WEF and draw conclusions from
the discussion.
2. Methodology
National competitiveness is an aggregate concept
which involves a variety of factors. As pointed outbyBuckley et al. (1988) , single measures are not
able to capture all the elements of the concept. In
the 2005 rankings, the World Competitiveness Year-
book utilized 314 criteria belonging to four factors:
economic performance, government efficiency, busi-ness efficiency, and infrastructure ( IMD website ).
The criteria considered in constructing the Global
Competitiveness Index of the WEF fall into nine,
what they call, pillars: institutions, infrastructure,macroeconomy, health and primary education,
higher education and training, market efficiency,
technological readiness, business sophistication,and innovation ( WEF website ). The viewpoint of
Doryan (1993) is that productivity increases, eco-
nomic conditions, socio-political stability, andenhancement of human resources are the four ele-ments that must be sustained for continued compet-itiveness. Consequently, he uses 123 indicators to
measure the competitive sustainability of five Cen-
tral American countries. All these studies have acommon philosophy, that is, the complicated systemof national competitiveness cannot be measureddirectly, it is necessary to deconstruct it into subsys-tems of manageable size. Deconstruction can bemade for more than one level, with the goal ofreaching a set of criteria which are measurable.
The measures of the criteria are then aggregated
to form an index which presumably is representativeof national competitiveness.
2.1. Factor deconstruction
Within this context, there are two major factors
to be noted, one is the criteria deconstructed fromC. Kao et al. / European Journal of Operational Research 187 (2008) 613–628 615

the concept of national competitiveness and the
other is the weights associated with the criteria incalculating the composite index. Similar to the stud-ies of IMD and WEF, we deconstruct the system of
national competitiveness into primary factors, sec-
ondary factors, and criteria. In determining themfor the Southeast Asian countries, scholars fromUniversity of Brunei, Royal University of PhnomPenh (Cambodia), University of Indonesia,National University of Laos, University of Malaya(Malaysia), University of Yangon (Myanmar), Uni-versity of the Philippines, Nanyang Technological
University (Singapore), Chulalongkorn University
(Thailand), and University of Economics-HCMC(Vietnam) were consulted to assure that the specialconditions of this region had been accounted for.In addition to communicating with the scholars ofthe 10 partner universities individually, two work-shops were held, one in Bangkok and another inKuala Lumpur, to have round-table discussions.
The structure and the criteria for measuring the
national competitiveness for the countries in thisregion were finalized in the 2002 Southeast Asia
Forum on Industrial Competitiveness held in Taiwan.
There are four primary factors at the first level:economy ,technology ,human resource , and manage-
ment. Each primary factor, in turn, is composed of
a number of secondary factors at the second level.
The structure is shown in Fig. 1 .
Under economy there are four secondary factors:
domestic economy, government, internationaltrade, and finance, whose performances are mea-sured by 8, 8, 8, and 9 criteria, respectively. Thesecriteria are listed in Appendix . Those with a solid
circle in front indicate that they are measured byhard data and those with a hollow circle are mea-
sured by soft data. Under technology there are also
four secondary factors: infrastructure, informationtechnology, research and development, and technol-ogy management. The numbers of criteria for mea-suring the performance of these secondary factorsare 6, 8, 11, and 7, respectively. The third primaryfactor, human resource , has three secondary factors:
quantity and quality, labor cost, and labor legisla-
tion, with 10, 6, and 6 criteria, respectively. The
fourth primary factor, management , has five second-
ary factors: managers’ competence, corporate cul-ture, industry integration, international operation,and productivity. They have 7, 8, 3, 5, and 6 criteria,respectively. For all four primary factors, there are116 criteria in total, of which 70 are measured byhard data and 46 by survey data. Note that the fac-tor with more secondary factors or the secondary
factor with more criteria does not mean that ele-ment is more important, it is just the result of logicalanalysis, and importance is determined by the
weights which will be discussed later. More criteria
to be considered obviously increases the reliabilityand robustness of the resulting measures. However,due to data availability, effort required in data col-lection, significance of the criteria, etc., many crite-ria can be or have to be ignored. For the latter case,appropriate proxies are used.
The criteria are of two types, cause and effect.
Good performance on cause-type criteria implies a
promising national competitiveness in the future.For example, education, as represented by illiteracyratio, school enrollment ratios, and ratios of pupilsto teachers (referring to the criteria listed in Appen-
dix), is a criterion of this type. Conversely, the
effect-type criteria are reflections of the level ofnational competitiveness in the past. They are the
result of some unknown cause-type criteria and
are used as proxies for those unknown cause-typecriteria. GDP is a criterion of this type. For somecriteria the distinction between cause and effect isvague, but this is not important. Mixed usage ofthese two types of criteria produces a more repre-sentative measure of the national competitiveness.
2.2. Data calibration
The hard data were collected from international
publications, databases, government reports, etc.,with the help of the partner universities. For softdata, opinions from experts were solicited. In eachcountry, 250 respondents were asked to provide per-sonal opinions, of which 100 were government offi-
cials, 100 were company managers, and 50 were
university professors. Each respondent was askedto choose a number between 0 and 10 to representthe performance of a criterion, where 0 indicatesthe lowest level and 10 the highest level. Since thesoft data are somewhat subjective, it is possible thatthe true level of performance of a country could bedistorted. For example, in the criterion ‘‘effective-
ness of government policy implementation,’’ even
though country Ais doing better than country B,
the respondents from country Amay give a lower
score than the respondents from country Bdo
because they have a higher standard. Therefore,supplementary information is required to adjustthe scores from different countries to make themcomparable. In this study, Singapore was chosen 616 C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628

as the base country for comparison since most peo-
ple are familiar with this country. The respondentsof every country were requested to evaluate Singa-pore in addition to their country. Ideally, therespondents from Singapore should evaluate not
only their own country but also the other nine coun-
tries, so a countercheck can be conducted. Since it isdifficult to find people in Singapore who are familiarwith all 10 Southeast Asian countries, some rectifi-cation is required to solve this problem.
The basic idea is to ask experts to evaluate three
countries, rather than 10 countries. This will reducethe amount of imprecise or even incorrect informa-
tion provided by the respondents due to insufficient
knowledge. The dissenting opinion regarding acountry evaluated by different groups of experts iscompromised based on a mathematical relationshipto be discussed later. In this research we divide the10 countries into two groups: peninsular countriesand island countries. The former are those on theIndo-China peninsular: Vietnam, Laos, Cambodia,Thailand, and Myanmar, and the latter include
the Philippines, Brunei, Malaysia, Singapore, andIndonesia. Most people of the peninsular countriesare familiar with Thailand, in addition to Singa-pore, and most people of the island countries are
familiar with Malaysia. In the survey, the res-
pondents from each peninsular country other thanThailand were asked to evaluate their own coun-try, Thailand, and Singapore. The respondentsfrom Thailand were asked to evaluate their owncountry, Malaysia, and Singapore. The samemethod of survey was applied to the island coun-tries. Specifically, the respondents from Malaysia
were asked to evaluate their own country, Thailand,
and Singapore, and the respondents from the coun-tries other than Malaysia and Singapore were askedto evaluate their own country, Malaysia, and Singa-pore. The respondents from Singapore evaluatedtheir own country, Thailand, and Malaysia. Thus,every respondent evaluated three countries. Thisform of evaluation is depicted in Fig. 2 , where PNational
CompetitivenessDomestic economy
Economy Government
International trade
Finance
Infrastructure
Technology Information technology
Research and development
Technology management
Quantity and quality
Human Resource Labor cost
Labor legislation
Managers’ competence
ManagementCorporate culture
Industry integration
International operation
Productivity
Fig. 1. Structure of national competitiveness: primary and secondary factors.C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628 617

stands for a peninsular country, Ifor an island
country, Tfor Thailand, Mfor Malaysia, and S
for Singapore. Each arrow indicates that the respon-dents from the country at the tail of the arrow madean evaluation for the country at the head of thearrow.
For each criterion of soft data, the 250 responses
from the same country are averaged to form the
score of that country. Let S
k,Tk, and Mkdenote
the scores evaluated by country kfor Singapore,
Thailand, and Malaysia, respectively. The average
S¼X10
k¼1Sk=10
is the score for Singapore. The ratio TT/STis the rel-
ative score of Thailand as compared to Singaporefrom the viewpoint of the respondents of ThailandandT
S/SSis the relative score from the viewpoint
of the respondents of Singapore. Their average elim-inates the possible prejudice from the respondentsof the two countries. When multiplied by the score
of Singapore, we obtain the score for Thailand
T¼0:5ðTT=STțTS=SSȚS:
Similarly, the score for Malaysia is
M¼0:5ðMM=SMțMS=SSȚS:
For other countries we use the average of two
measures to represent their performance. Let Xk
denote the score evaluated by the respondents of
country kfor country X. One measure for country
kis the relative score of this country as compared
to Singapore multiplied by the score of Singapore:(Kk=SkȚS. The relation is represented by arc PS for
peninsular countries and arc IS for island countriesinFig. 2 . Another measure is derived from the inter-
mediate country Thailand for peninsular countries
and from Malaysia for island countries. For a pen-
insular country it is the relative score of this countryas compared to Thailand multiplied by the score ofThailand, ( K
k=TkȚT. This relation is represented by
path PTS in Fig. 2 . Similarly, the measure for island
countries is ðKk=MkȚMvia path IMS in Fig. 2 . The
score for country kis the average of the two mea-
sures, that is
P¼0:5Kk
Sk/C18/C19
SțKk
Tk/C18/C19
T/C20/C21
for peninsular countries ;
I¼0:5Kk
Sk/C18/C19
SțKk
Mk/C18/C19
M/C20/C21
for island countries :
Note that the scores TMandMT, i.e., the perfor-
mance of Thailand evaluated by the respondentsfrom Malaysia and the performance of Malaysiaevaluated by the respondents from Thailand, havenot been used in the above calculations. Their func-tion is to let the respondents from Thailand andMalaysia provide more accurate scores for the rela-tive performance of Singapore, Thailand, and
Malaysia so that the scores
TandMused for calcu-
lating the performance of other countries will alsobe more accurate.
The score for soft data is in the range of 0 and 10.
For hard data, however, it can be any units of anyscale. For example, GDP is in the scale of billionsof dollars while unemployment rate is usually smal-ler than 10%. To make all hard data comparable
with soft data, a standardization treatment is
required. In the literature, several techniques forstandardizing data have been discussed ( Freuden-
berg, 2003 ). Of which rankings, z-score standardiza-
tion, and re-scaling are probably the most prevailingones. Since ranks lose the information on absolutelevels and z-score standardization may produce neg-
ative values and unexpected range of values, a re-
scaling technique is used in this study.
LetX
kdenote the score for country kon a crite-
rion, Xmax= max. { Xk,k=1 , …,10}, and Xmin=
min. { Xk,k=1 , …,10}. The standardized score
for country kis
Favorable criteria :10Xk
Xmax/C18/C19
;
Unfavorable criteria :10 1 /C0Xk
XmaxțXmin
Xmax/C18/C19
:TPS
I
M
Fig. 2. Form of soft data collection for Singapore (S), peninsular
countries (P), Thailand (T), Malaysia (M), and island countries(I).618 C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628

As an example, consider five observations col-
lected from five countries: [50, 20, 10, 5, 1]. Themaximum and minimum values are 50 and 1,respectively. If this set of observations belongs toa favorable criterion, then the standardized scoreswill be [10, 4, 2, 1, 0.2]. On the other hand, if itbelongs to an unfavorable criterion, then the stan-dardized scores are [0.2, 6.2, 8.2, 9.2, 10]. Fig. 3
depicts the concept of this standardization, which
should be self-explanatory.
2.3. Score aggregation
Several methods for aggregating the component
performance have been reported ( Munda and
Nardo, 2003 ). Without further information, the
method of weighted average is the most popular
and widely accepted one ( Freudenberg, 2003 ). The
performance of each criterion under a secondaryfactor is considered as a proxy of that secondaryfactor. While the performance of the secondary fac-tor cannot be measured directly, we use the perfor-mance of the criterion to represent the performanceof that secondary factor. If there is more than one
criterion for a secondary factor, their average is
used. In this setting, the importance of every crite-rion is considered the same. Statistically, more crite-ria for a secondary factor implies a more reliableand robust measure for that secondary factor.
When the performances of the secondary factors
under the same primary factor are aggregated torepresent the performance of the primary factor,
the importance, as expressed by weight, of each sec-
ondary factor, however, is allowed to be different.The weights of the factors can be determined eithersubjectively by experts or objectively by data them-selves. The former is opinion driven. They are a pri-
oriweights since they are determined before data are
collected. The latter is data driven. They are a pos-
teriori weights because they are determined afterdata are collected. This idea is proposed by Kao
and Hung (2005) and has been applied to calculate
the management performance of the manufacturingfirms in Taiwan ( Kao and Hung, 2007 ). The a priori
weights were solicited from the same experts via sur-
vey, in that the experts gave scores in the range of 0and 100 to all secondary factors under the same pri-mary factor. The ratio of the score of a secondaryfactor to the total score of all secondary factors rep-resents the weight of that secondary factor.
The a posteriori weights are determined in two
stages. In the first stage, the ideal index, or the max-
imal index, of a country is determined by applying
the most favorable weights to the factors. Theweights selected by each country can be different.The a posteriori weights are then determined by
minimizing the total squared difference betweenthe ideal index and that calculated from the a poste-
riori weights of the 10 countries. At this stage the
determined weights are common to all countries.
LetS
ijdenote the score of country ion the jth sec-
ondary factor, j=1 , …,m. Mathematically, the
ideal index of country iis determined via the follow-
ing linear program:
Max Ii¼Xm
j¼1wijSij
s:t:Xm
j¼1wij¼1;
lj6wij6uj;j¼1;… ;m;ð1Ț
where ljandujare the smallest weight and largest
weight, respectively, collected from the experts forthejth factor. The most favorable weights are re-
quired to lie within the bounds and have to sumto 1.
Model (1)is essentially a DEA model without
inputs. To see the equivalence, one notes that aDEA model without inputs can be formulated as
(Kao, 1994 )
E
i¼MaxXm
j¼1wijSij
s:t:Xm
j¼1wijSkj610;k¼1;… ;n;
wijP0;j¼1;… ;m;ð2Ț
where nis the number of countries. Since all Skjare
less than or equal to 10, it is necessary to havePm
j¼1w/C3
ijP1, where w/C3
ijare the optimal weights
solved from (2), in order for Eito approach the0
maxk
XX1Xmin Xk Xmax
Observations
1
maxmin
maxk
XX
XX1 + − 0
maxmin
XX(a)
(b)
Fig. 3. Data standardization for (a) favorable and (b) unfavor-
able criteria.C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628 619

upper bound 10 as close as possible. DenotePm
j¼1w/C3
ij¼Wand ^wij¼wij=W, the above DEA
model is equivalent to
Ei=W¼MaxXm
j¼1^wijSij
s:t:Xm
j¼1^wijSkj610=W;k¼1;… ;n;
^wijP0;j¼1;… ;m:
ð3Ț
If one requiresPm
j¼1^wij¼1, then it is clear thatPm
j¼1^wijSkj610=Wwill hold. In other words, the
latter constraint can be replaced by the former. Byadding bound constraints to the weights, we obtain
the following model, which is exactly the same as
Model (1):
MaxX
m
j¼1^wijSij
s:t:Xm
j¼1^wij¼1;
lj6^wij6uj;j¼1;… ;m:ð4Ț
Other types of DEA models have also been used for
constructing composite indicators (e.g., Cherchye
et al., 2006; Despotis, 2005; Lovell et al., 1995 ).
Thea posteriori weights wj,j=1 , …,mare deter-
mined in the second stage via the following qua-
dratic program:
MinXn
i¼1I/C3
i/C0Xm
j¼1wjSij !2
s:t:Xm
j¼1wj¼1;
lj6wj6uj;j¼1;… ;m:ð5Ț
The lower bound ljand upper bound ujare the same
as those used in Model (1). Conceptually, a regres-
sion linePm
j¼1wjSijwhich explains the trend of the
observations I/C3
ithe best is sought. Thus, in minimiz-
ing the objective function, larger weights would be
assigned to the subjects that are more favorable tothe countries.
In fitting the regression lineP
m
j¼1wjSij, one could
apply a weighted regression method, which assumesunequal variance among observations. Since weonly have 10 observations, one for each country, itis impossible to obtain a reliable estimation of the
variance for different observations. Therefore, thesimple least-squares method, which assumes equal
variance, is used. The performances of the four pri-mary factors are aggregated to form a measure ofnational competitiveness in a similar manner.
3. Empirical results
In the beginning of 2003, a pilot study was con-
ducted and the results of the 10 countries were pub-lished in Asia Pacific Management Review (Kao,
2004). The pilot study mainly focused on criterion
selection, design of the questionnaire, and estima-
tion of the variance. National competitiveness was
not calculated for each country. After the pilotstudy, some modifications were made to the ques-tionnaires, and a full-scale survey was conductedat the end of 2003.
For hard data, we collected three years, 1999,
2000, and 2001, and calculated the average to makethe results more reliable and representative for that
period of time. For soft data, the pilot study shows
that most criteria have a standard deviation of 1.2for the scores collected from each country, and thelargest standard deviation is approximately 1.6.Assume the scores are normally distributed, we have(Cochran, 1963 ):Prfj
X/C0ljPtn/C01;1/C0a=2s=ffiffiffinpg¼a,
where Xis the average and sis the standard
deviation of the scores collected from the respon-
dents, lis the population mean, nis the sample size,
1/C0ais the confidence level, and tn/C01;1/C0a=2is the
upper 1 /C0a/2 critical point for the tdistribution
with n/C01 degrees of freedom. Suppose we want
the margin error d¼tn/C01;1/C0a=2s=ffiffiffinpto be smaller
than 0.2. Since the sample size nis not known yet,
we use the critical point of the standard normal dis-tribution to approximate t
n/C01;1/C0a=2. Then the sample
sizenfor the worst case at a= 0.05, can be calcu-
lated as: n¼ðtn/C01;1/C0a=2s=dȚ2¼ð1:96/C21:6=0:2Ț2¼
246. Therefore, 250 experts from each country weresurveyed and their responses were averaged to rep-resent the general opinion of each country.
The calibration procedure stated in the preceding
section was applied to standardize the results.Table 1 shows the results of the 10 countries for
economy . Recall that each secondary factor has sev-
eral criteria associated with it. Their standardizedscores were averaged to represent the performanceof the corresponding secondary factor. Singaporehas the best performance for all four secondary fac-tors. Brunei also performs well on government, witha score close to that of Singapore. On the other side,Philippines, Cambodia, Laos, and Indonesia per-620 C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628

form the worst in domestic economy, government,
international trade, and finance, respectively. Toaggregate the performances of the secondary factorsto form the performance of economy , both the a pri-
oriweights, which are the averages of the weights
solicited from the 2500 experts, and the a posteriori
weights, which were determined from the two-stage
approach, were applied to calculate the weighted
sums. The weights at the bottom of Table 1 indicate
that finance is a little less important than the otherthree secondary factors as conceived by the experts.
The a posteriori weight for government is obvi-
ously larger and for international trade is smallerthan the corresponding a priori weight. This is sim-
ply because the scores of government are relatively
higher and the scores of international trade are rel-
atively lower than those of other secondary factors,assigning larger and smaller weights, respectively,will result in larger weighted sums.
The last two columns of Table 1 show the aggre-
gate performances on economy calculated from the
two sets of weights for each country. In terms ofrankings, these two sets of weights produce the same
result, only the indices calculated from the a poste-
riori weights are in general higher than those calcu-
lated from the a priori weights. Since Singapore has
performed the best for all secondary factors, it is notsurprising that it has the best aggregate perfor-mance. Malaysia and Thailand are the second andthird best, respectively. The next three countries,Brunei, Vietnam, and Myanmar, have close scores.The last four countries, Cambodia, Indonesia, Phil-
ippines, and Laos, also have close scores. Notably,the Philippines, a once prosperous country in Asia,has declined.
The same procedure is applied to calculate the
performance of technology and the results are sum-
marized in Table 2 . Of the four secondary factors,
Singapore performs the best for all of them. Con-
versely, Laos has the worst performance on infra-structure, R&D, and technology management, andMyanmar has the worst performance on infrastruc-ture (shared with Laos) and information technol-ogy. Surprisingly, a rich country, Brunei, performsthe second worst on infrastructure. The a priori
weights at the bottom indicate that infrastructure,
with a weight of 0.2733, is a little more important
and technology management, with a weight of0.2217, is a little less important than the others ascompared to the average weight of 0.25. The a pos-
teriori weights, on the other hand, are quite even
with each other. However, they produce weightedsums of the same rankings. The aggregate perfor-mances in the last two columns of Table 2 consis-
tently indicate that Singapore is far ahead of the
other countries. Malaysia is the second, followedby Thailand and Indonesia. Brunei, Philippines,Vietnam, and Cambodia have similar performance.Myanmar and Laos are the last two.
The results of human resource , as shown in
Table 3 , are quite interesting. Brunei performs the
best on quantitative and quality and PhilippinesTable 1
Economic performance of the 10 countries
Country Domestic economy Government International trade Finance Aggregate performance
A priori A posteriori
Brunei 3.8092 8.1007 3.4018 4.2931 4.9570 5.2798
Cambodia 4.3430 4.0545 3.1358 4.6742 4.0385 4.1402Indonesia 4.5237 4.7099 4.1485 2.3926 3.9979 4.0016Laos 4.2369 5.0238 0.9963 4.1635 3.6171 3.9705
Malaysia 5.6769 6.2907 6.9993 6.5415 6.3668 6.2994
Myanmar 5.3636 5.6851 2.2256 5.1380 4.6114 4.9224Philippines 3.7217 4.1877 2.3789 5.0180 3.8016 3.9841
Singapore 7.6246 8.1843 8.7140 7.8561 8.0988 8.0369
Thailand 5.9312 5.8043 5.0046 6.4743 5.7878 5.8779Vietnam 5.3184 5.6644 3.1108 5.3187 4.8570 5.0867
Weight
Lower bound 0.1912 0.2328 0.1500 0.2000 0.2459
Upper bound 0.4000 0.3298 0.3050 0.2565 0.3618
A priori 0.2577 0.2658 0.2507 0.2258 0.2760
A posteriori 0.2673 0.3298 0.1567 0.2462 0.2459
*Bold-face numbers are weights associated with the primary factor economy .C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628 621

performs the worst; however, their difference is not
much. Regarding labor cost, the rich countries Bru-nei and Singapore obviously have higher cost thanthe poor countries, such as Laos and Myanmar.Note that this secondary factor is an unfavorableone, where higher cost implies worse performance,
and that the highest scores appear at Myanmar
and Laos and the lowest at Brunei and Singapore.For the third secondary factor, labor legislation,Singapore has the best performance while Myanmarhas the worst. The weights for quantitative andquality, labor cost, and labor legislation are
0.4225, 0.3203, and 0.2572, respectively, as shownat the bottom of Table 3 . The weight of 0.4225 is
26.76% higher than the average weight of 0.3333,and the weight of 0.2572 is 22.83% lower than theaverage weight. This makes the total difference
between these two secondary factors amount to
49.59%. The a posteriori weights have expanded
the difference from 0.4225 to 0.6198 for quantityand quality and 0.2572 to 0.1250 for labor legisla-tion, where the former is four times larger thanTable 2
Technology performance of the 10 countries
Country Infrastructure Information
technologyR&D Technology
managementAggregate performance
A priori A posteriori
Brunei 0.9683 4.5425 1.5603 6.2161 3.2186 3.3670
Cambodia 2.1416 0.5959 1.3530 6.2838 2.4617 2.5991Indonesia 3.8862 1.5685 4.3836 6.6790 4.0128 4.1189Laos 0.8547 1.3264 0.9208 3.9200 1.6747 1.7700
Malaysia 4.5142 4.4538 4.5546 8.3749 5.3639 5.4876
Myanmar 0.8547 0.0791 1.3694 5.6894 1.8456 2.0074Philippines 1.2155 2.0706 3.5163 6.0313 3.0631 3.2239
Singapore 9.7401 8.7172 9.3954 10.000 9.4445 9.4572
Thailand 4.5962 2.3583 4.4018 7.0828 4.5094 4.6016Vietnam 2.5131 1.7927 1.0693 5.3810 2.6109 2.6991
Weight
Lower bound 0.2422 0.2128 0.1591 0.1399 0.1708
Upper bound 0.3439 0.3571 0.3366 0.2536 0.2598
A priori 0.2733 0.2642 0.2408 0.2217 0.2195
A posteriori 0.2422 0.2580 0.2462 0.2536 0.1708
*Bold-face numbers are weights associated with the primary factor technology .
Table 3
Human resource performance of the 10 countries
Country Quantity and
qualityLabor cost Labor
legislationAggregate performance
A priori A posteriori
Brunei 9.4226 3.9688 7.6680 7.2245 7.8115
Cambodia 7.3725 8.5129 5.3077 7.2067 7.4054Indonesia 8.8882 7.1740 8.4805 8.2343 8.3998
Laos 7.8968 9.0318 5.2895 7.5897 7.8605
Malaysia 8.9783 6.0000 8.8474 7.9907 8.2019Myanmar 7.6152 10.000 4.7738 7.6482 7.8686Philippines 7.3233 9.0150 5.5165 7.4004 7.5292
Singapore 8.8281 4.0234 9.8967 7.5640 7.7355
Thailand 9.0304 6.8624 8.2876 8.1449 8.3843Vietnam 7.9173 8.9989 7.7507 8.2209 8.1725
Weight
Lower bound 0.3410 0.2500 0.1250 0.2251
Upper bound 0.6250 0.3672 0.3227 0.3185
A priori 0.4225 0.3203 0.2572 0.2664
A posteriori 0.6198 0.2552 0.1250 0.3153
*Bold-face numbers are weights associated with the primary factor human resource .622 C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628

the latter. However, the average aggregate perfor-
mances calculated from these two sets of weightsdo not differ by much, 7.7224 versus 7.9369. Sincethe 10 countries perform similarly on this factor,
8.2343 for the highest and 7.2067 for the lowest
from the a priori weights and 8.3998 for the highest
and 7.4054 for the lowest from the a posteriori
weights, the rankings are of little importance.
Under management there are five secondary fac-
tors and their performances for each country areshown in Table 4 . Singapore performs the best in
all secondary factors, except productivity, where
Brunei is the best. Laos performs the worst in man-
agers’ competence, industry integration, and inter-national operation. Myanmar and Vietnamperform the worst in corporate culture and produc-tivity, respectively. All five secondary factors havesimilar a priori weights. The a posteriori weights,
on the other hand, show that a higher weight formanagers’ competence (0.2364) and a lower weight
for productivity (0.1658) produce more favorable
aggregate performance indices. The weighted sumscalculated from the two sets of weights have thesame rankings, only Myanmar and Vietnam arereversed, where Singapore is the best, Malaysiaand Philippines are the second and third best,respectively. Then we have Thailand, Indonesia,Brunei, Cambodia, in sequence. Laos performs the
worst.
The performances of the four primary factors are
finally aggregated to represent the national compet-
itiveness. The weights used are also of two types, a
priori and a posteriori. The former were solicited
from the same 2500 experts, which are 0.2760,0.2195, 0.2664, and 0.2381 for economy ,technology ,
human resource ,a n d management , respectively, as
shown in the lower right corner of Tables 1–4 in
bold-face. They are not far away from the averageof 0.25. The largest difference is 10.40%, occurring
ateconomy . The second column of Table 5 shows
the national competitiveness of the 10 countries indescending order. This order more or less followsthe order of each primary factor. Singapore hasthe highest score. This is not surprising since it per-forms the best for three factors. Malaysia and Thai-land are the second and third best, respectively.Their scores are significantly smaller than those of
Singapore. The remaining seven countries, in
sequence, are Indonesia, Philippines, Brunei, Viet-nam, Myanmar, Cambodia, and Laos. In contrastto Singapore, Laos performs the worst on threefactors.
When the two-stage approach is applied, the a
posteriori weights obtained are 0.2459, 0.1708,
Table 4
Management performance of the 10 countries
Country Managers’
competenceCorporate
cultureIndustry
integrationInternational
operationProductivity Aggregate
performance
A
prioriA
posteriori
Brunei 5.9640 5.5278 5.2379 3.0139 8.3312 5.5634 5.5644
Cambodia 5.6905 4.7575 5.4750 2.9300 6.6819 5.0745 5.1078Indonesia 6.9501 5.9564 7.0872 5.6115 6.0328 6.3325 6.3795Laos 4.2313 4.4215 4.3430 2.0330 6.2321 4.2091 4.2137
Malaysia 7.9662 6.7625 7.5972 7.2534 6.9960 7.3242 7.3546
Myanmar 5.5976 4.2876 5.6029 2.7076 6.9361 4.9817 5.0202Philippines 8.5370 6.7385 7.9206 6.2897 5.7475 7.0844 7.1663
Singapore 9.8234 9.8605 10.000 10.000 5.4855 9.1183 9.1820
Thailand 8.0115 6.1448 7.4559 6.0240 5.3079 6.6244 6.7030Vietnam 6.1204 5.2312 5.4680 3.1574 4.8249 4.9759 5.0367Weight
Lower
bound0.2004 0.1727 0.1680 0.1869 0.1293 0.1745
Upper
bound0.2364 0.2417 0.2122 0.2290 0.2018 0.2680
A priori 0.2156 0.2092 0.1926 0.2022 0.1804 0.2381
A posteriori 0.2364 0.1987 0.2122 0.1869 0.1658 0.2680
*Bold-face numbers are weights associated with the primary factor management .C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628 623

0.3153, and 0.2680 for economy ,technology ,human
resource , and management , respectively, where
0.1708 is much smaller than and 0.3153 is muchgreater than the average of 0.25. The bounds forthese weights as solicited from the experts are shown
in the lower right corner of Tables 1–4 in bold face.
The weighted sums calculated from the a posteriori
weights are shown in the third column of Table 5 .
Their rankings are the same as those obtained fromthea priori weights, only the rankings of Philippines
and Brunei are reversed. As expected, the average ofthe weighted sums calculated from the a posteriori
weights, 6.0975, is greater than that calculated from
thea priori weights, 5.7388.
In the Global Competitiveness Report 2005 (WEF
website ), 117 countries have been ranked, which
include seven Southeast Asian countries: Cambodia,Indonesia, Malaysia, Philippines, Singapore, Thai-land, and Vietnam. The report contains two indices,Growth Competitiveness Index (GCI), which
attempts to measure economic conditions that lead
to high GDP per capita growth, and Business Com-
petitiveness Index (BCI), which concentrates more
on microeconomic factors to measure the currentlevel of competitiveness. The order of these sevencountries for both of these two indices is the sameas that of this research under both a priori weights
and a posteriori weights. The fourth column of
Table 5 shows the GCI and their world ranks (in
parentheses) of these seven countries. For BCI only
the ranks have been reported, they are listed in thefifth column of Table 5 .
The results of the World Competitiveness Year-
book 2005 (IMD website ) are somewhat different
from ours. For the 60 countries being ranked, fiveSoutheast Asian countries, Indonesia, Malaysia,Philippines, Singapore, and Thailand, are included.Their scores and world ranks are listed in the last
column of Table 5 . The order is: Singapore, Thai-
land, Malaysia, Philippines, and Indonesia, whereThailand is better than Malaysia by one rank andPhilippines is better than Indonesia by 10 ranks.
Notably, in their 2004 report Malaysia is better than
Thailand by 13 ranks.
Different studies on national competitiveness use
different approaches to calculate competitivenessindices. Consequently, different results would beexpected. The results of this study is similar to thoseofGlobal Competitiveness Report andWorld Com-
petitiveness Yearbook , indicating that the rankings
obtained from these three approaches are quite rep-
resentative. Since the criteria adopted and theweights used for performance aggregation in thisstudy are solicited from the experts in this region,its results should be more convincing and acceptableto the Southeast Asian countries. Moreover, thecountries which have not appeared in the two abovementioned studies are those that were ranked rela-
tively lower in this research. This confirms the state-
ment at the beginning of this paper that the existingstudies only focus on more developed countrieswhen measuring the national competitiveness.
4. Conclusions
National competitiveness is a complicated con-
cept. It involves many aspects in measurement and
requires much effort in data collection. This paperproposes a framework for measuring the nationalcompetitiveness of the Southeast Asian countriesvia thorough discussions with university professorsin this region. The basic idea is to deconstruct theconcept of national competitiveness into measurablecriteria through a logical analysis. At the first level,Table 5
National competitiveness of the 10 countries
Country This research Global competitiveness report World Competitiveness
Yearbook 2005A priori A posteriori GCI BCI
1. Singapore 8.4945 8.4914 5.48 (6) (5) 89.679 (3)
2. Malaysia 6.8072 7.0434 4.90 (24) (23) 65.844 (28)3. Thailand 6.3343 6.6713 4.50 (36) (37) 66.012 (27)4. Indonesia 5.6856 6.0457 3.53 (74) (59) 33.811 (59)
5. Philippines 5.3799 5.8249 3.47 (77) (69) 51.103 (49)
6. Brunei 5.3239 5.8276 – – –7. Vietnam 5.2884 5.6384 3.37 (81) (80) –
8. Myanmar 4.9015 5.3797 – – –
9. Cambodia 4.7831 5.1658 2.82 (112) (109) –10. Laos 4.3900 4.8864 – – –624 C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628

national competitiveness is deconstructed into four
primary factors: economy ,technology ,human
resource ,a n d management . Each primary factor is
further deconstructed into 3–5 secondary factors
at the second level, with a total of 16 secondary fac-
tors. Then a set of criteria is adopted to measure theperformance of each secondary factor. In total,there are 116 criteria being used. The scores of thecriteria at the bottom level are aggregated to forma measure for national competitiveness.
The most difficult task in this type of research is
data collection. With the assistance from the partner
universities in those 10 countries we were able to
acquire hard data from publications and soft datafrom questionnaire surveys. Since it is almost impos-sible to find experts who are familiar with the statusof all 10 countries to be able to obtain consistentevaluation, this paper proposes a methodologywhich only requires knowledge of three countriesfrom the respondents, including their own country,
Singapore, and either Thailand or Malaysia depend-
ing on the country type of the respondents is of pen-insular or island. A standardization technique isdevised to make the hard data from favorable andunfavorable criteria comparable with the soft data.Another difficulty is the determination of the weightsfor aggregating the performances of the secondaryfactors to the performances of the primary factors
and subsequently the national competitiveness. Usu-
ally, experts’ opinion or prior information is col-lected to determine a set of a priori weights to use.
When these are lacking, then a set of a posteriori
weights obtained from the data themselves is sought.This paper applies both of these two ideas to calcu-late the national competitiveness.
Our results show that Singapore has the highest
national competitiveness and is a distance ahead of
other Southeast Asian countries. Malaysia and Thai-land, two fast-growing countries in this region, arethe second and third best, respectively. The largestcountry, Indonesia, is ranked the fourth. Philippines,a country with a glorious past, is ranked fifth. Next isthe oil-rich country Brunei. Notably, for the a poste-
riori weights the rankings of Philippines and Brunei
are reversed. Then we have Vietnam, a country liber-
alizing its economy and opening its markets afterdecades of communism. Philippines, Brunei, andVietnam have very close scores for national compet-itiveness. The relatively closed countries Myanmarand Laos are ranked eighth and tenth, respectively,and Cambodia, a country which has suffered frominternal turmoil for a long time, is ranked ninth. Ofthe 10 countries, Brunei, Myanmar, and Laos have
never appeared in other studies. The inclusion ofthese three countries in this study gives the readersan insight as to how these countries are performing
as compared to other countries in the region. Fur-
ther, the performances of the 10 countries on everyprimary factor and secondary factor provide usefulinformation for policy making by the governmentand resource allocation between countries for inves-tors. For example, Vietnam is ranked lower than thePhilippines and Brunei. One reason is due to its lesssatisfactory performance on technology , especially
on R&D. Hence, if the Vietnam government can
require its enterprises to increase expenditures onR&D, then the performance of Vietnam on technol-
ogycan be improved, and consequently its national
competitiveness will be raised.
The eventual goal of this research is to improve
the quality of life in Southeast Asian countries.Although the focus of this paper is on Southeast
Asia, the methodology is applicable to other regions
in the world.
Acknowledgements
This research is supported by the National
Science Council of the Republic of China underContracts NSC90-2416-H-006-041, NSC91-2416-H-006-027, and NSC92-2416-H-006-001. The assis-
tance from University of Brunei, Royal University
of Phnom Penh, University of Indonesia, NationalUniversity of Laos, University of Malaya, Univer-sity of Yangon, University of the Philippines, Nany-ang Technological University, ChulalongkornUniversity, and University of Economics-HCMCin data collection is gratefully acknowledged. Theauthors also thank Professor S.T. Hung for compu-
tational works.
Appendix. Criteria for measuring national
competitiveness
A. Economy
a. Domestic economy
•Gross domestic product (GDP)
•Growth of GDP
•Unemployment rate
•GDP per capita
•Growth of GDP per capita
•Gross capital formation in percentage of GDP
•Gross domestic savings in percentage of GDP
/C14Resilience of economyC. Kao et al. / European Journal of Operational Research 187 (2008) 613–628 625

b. Government
•Government revenue in percentage of GDP
•Government expenditure in percentage of GDP
•Government fiscal balance in percentage of GDP
/C14Political stability
/C14Effectiveness of government policy implemen-
tation
/C14Adequate policy for promoting country’s com-
petitiveness
/C14Obedience of companies to legal regulation
/C14Adaptiveness of government policies to changes
in economic environment
c. International trade
•Exports of goods and services
•Imports of goods and services
•Growth of exports of goods and services
•Growth of imports of goods and services
•Balance of trade
•Balance of current account
•Degree of openness
/C14Government assistance to companies in accessing
foreign markets
d. Finance
•Number of registered companies
•Inflation rate
•Lending rate
•Foreign reserves
•Total debt outstanding
•Ratio of total debt service to exports
•Net inflows of foreign direct investment
/C14Availability of local financial market
/C14Soundness of the central bank’s economic devel-
opment policy
B. Technology
a. Infrastructure
•Road density
•Railroad density
•Number of air transportation passengers
•Electricity cost for industry
•Domestic energy production as a percentage of
total energy consumption
•Energy imports as a percentage of total exports
b. Information technology
•Investment in telecommunication as a percentage
of GDP
•Number of computers per 1000 people
•Number of host connections to internet per 1000
people•Number of accesses to internet per 1000 people
•Bandwidth connected abroad
•Number of telephone lines per 1000 people
•Number of cellular phones per 1000 people
•Cost of international calls to USA
c. Research and development
•Total R&D expenditures
•Total R&D expenditures as a percentage of GDP
•R&D expenditures financed by enterprise
•R&D expenditures financed by enterprise as a
percentage of sales
•Number of R&D staff per 10,000 staff
•Number of R&D staff with college degrees or
higher
•Number of domestic patents granted to residents
•Number of foreign patents obtained by residents
•Export value of technology
•Import value of technology
/C14Effect of basic research on technological develop-
ment
d. Technology management
/C14Technology cooperation between companies
/C14Technology cooperation between universities and
companies
/C14Technology cooperation between research insti-
tutes and companies
/C14Technology transfer from universities to
companies
/C14Effect of insufficient financial resources on tech-
nological development
/C14Attitude of students to majoring in technology-
related areas
/C14Enforcement of copyright protection
C. Human resource
a. Quantity and quality
•Labor force participation rate
•Weekly working hours
•Ratio of skilled labor to total labor
•Illiteracy ratio
•Elementary school enrollment ratio
•Secondary school enrollment ratio
•Higher education enrollment ratio
•Ratio of pupils to teachers in elementary school
•Ratio of pupils to teachers in secondary school
•Ratio of pupils to teachers in higher education
b. Labor cost
•Average wage rate
•Average skilled labor/technician wage rate626 C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628

/C14Benefit level of employees
/C14Recruitment cost
/C14Training cost
/C14Severance payment
c. Labor legislation
•Number of disputes between the management
and labor
•Working days lost
/C14Enforcement of labor legislation
/C14Completeness of labor laws
/C14Power of the union
/C14Labor’s existing power over foreign companies
D.Management
a. Managers’ competence
/C14Inclination toward risk taking
/C14Prestige of managers
/C14Availability of competent senior managers
/C14Competitiveness
/C14Harmony with employees
/C14Loyalty to company
/C14Attitude to employee training
b. Corporate culture
/C14Entrepreneurship
/C14Capability of corporate boards to prevent impro-
per practices
/C14Core value of shareholders
/C14Social responsibility
/C14Emphasis on customer satisfaction
/C14Tax evasion
/C14Business ethics
/C14Bribing
c. Industry integration
/C14Integration between suppliers and manufacturers
/C14Integration between manufacturers and channel-
providers
/C14Vertical integration of industries
d. International operation
•Expenditures on advertising
•Ratio of corporate tax to profit
/C14Price/quality ratio compared to imported
products
/C14Experience of international business
/C14Management competence in global operations
e. Productivity
•Total factor productivity
•Productivity in agriculture•Productivity in mining
•Productivity in industry
•Productivity in construction
•Productivity in services
•Hard data to be collected from publications and
reports.
/C14Soft data to be surveyed from experts.
References
Bellak, C.J., Weiss, A., 1993. A note on the Austrian ‘‘diamond’’.
Management International Review 33 (2), 109–118.
Blaine, M., 1993. Profitability and competitiveness: Lessons from
Japanese and American firms in the 1980s. CaliforniaManagement Review 36 (1), 48–58.
Boltho, A., 1996. The assessment: International competitiveness.
Oxford Review of Economic Policy 12 (3), 1–16.
Buckley, P.J., Pass, C.L., Prescott, K., 1988. Measures of
international competitiveness: A critical survey. Journal of
Marketing Management 4 (2), 175–200.
Cherchye, L., Moesen, W., Rogge, N., van Puyenbroeck, T.,
Saisana, M., Saltelli, A., Liska, R., Tarantola, S., 2006.Creating Composite Indicators with DEA and Robustness
Analysis: The Case of the Technology Achievement Index.
Center for Economic Studies, Catholic University of Leuven,Mimeo.
Cochran, W.G., 1963. Sampling Techniques, second ed. John
Wiley and Sons, NY.
Despotis, D.K., 2005. A reassessment of the human development
index via data envelopment analysis. Journal of OperationalResearch Society 56, 969–980.
Doryan, E.A., 1993. An institutional perspective of competitive-
ness and industrial restructuring policies in developing coun-tries. Journal of Economic Issues 27 (2), 451–458.
Dunning, J.H., 1993. Internationalizing Porter’s diamond. Man-
agement International Review 33 (2), 7–15.
Fagerberg, J., 1996. Technology and competitiveness. Oxford
Review of Economic Policy 12 (3), 39–51.
Freudenberg, M., 2003. Composite Indicators of Country Per-
formance: A Critical Assessment. OECD DSTI/IND(2003)5.
Grant, R.M., 1991. Porter’s ‘‘Competitive Advantage of
Nations’’: An assessment. Strategic Management Journal 12
(7), 535–548.
Hodgetts, R.M., 1993. Porter’s diamond framework in a Mexican
context. Management International Review 33 (2), 41–54.
IMD, 2003. World Competitiveness Yearbook 2003. Institute for
Management Development, Lausanne, Switzerland.
IMD website, <http://www.imd.ch/wcy/> , January 24, 2006
visited.
JRC website, <http://farmweb.jrc.cec.eu.int/CI/> , December 9,
2006 visited.
Kao, C., 1994. Evaluation of junior colleges of technology: The
Taiwan case. European Journal of Operational Research 72,
43–51.
Kao, C. (Ed.), 2004. National competitiveness of the Southeast
Asian countries. Asia Pacific Management Review 9 (2)
(special issue).
Kao, C., Hung, H.T., 2005. Data envelopment analysis with
common weights: The compromise solution approach. Jour-nal of Operational Research Society 56, 1196–1203.C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628 627

Kao, C., Hung, H.T., 2007. Management performance: An
empirical study of the manufacturing companies in Taiwan.Omega 35, 152–160.
Krugman, P.R., 1996. Making sense of the competitiveness
debate. Oxford Review of Economic Policy 12 (3), 17–25.
Krugman, P.R., Hatsopoulus, G.N., 1987. The problem of US
competitiveness in manufacturing. New England Economic
Review (January–February), 18–29.
Ku, S.C.Y., 2002. The political economy of regime transforma-
tion: Taiwan and Southeast Asia. World Affairs 165 (2), 59–78.
Lall, S., 2001. Competitiveness indices and developing countries:
An economic evaluation of the Global CompetitivenessReport. World Development 29 (9), 1501–1525.
Lovell, C.A.K., Pastor, J.T., Turner, J.A., 1995. Measuring
macroeconomic performance in the OECD: A comparison of
European and non-European countries. European Journal ofOperational Research 87, 507–518.
Munda, G., Nardo, M., 2003. On the Methodological Founda-
tions of Composite Indicators Used for Ranking Countries.
Universitat Autonoma de Barcelona, Mimeo.Nardo, M., Saisana, M., Saltelli, A., Tarantola, S., Hoffman, A.,
Giovannini, E., 2005. Handbook on Constructing CompositeIndicators: Methodology and User Guide. OECD Statistics
Working Paper STD/DOC(2005)3.
OECD, 1996. Industrial Competitiveness: Benchmarking Busi-
ness Environment in the Global Economy. OECD, Paris,
France.
Oral, M., Chabchoub, H., 1996. On the methodology of the
World Competitiveness Report. European Journal of Oper-ational Research 90, 514–535.
Porter, M.E., 1990. The Competitive Advantage of Nations.
Macmillan, London.
Scott, B.R., Lodge, G.C., 1985. US Competitiveness in the
World Economy. Harvard Business School Press, Boston,
Mass.
WEF, 2003. The Global Competitiveness Report 2003.
Oxford University Press for World Economic Forum, NewYork.
WEF website, <http://www.weforum.org/> , January 24, 2006
visited.628 C. Kao et al. / European Journal of Operational Research 187 (2008) 613–628