The Asian Journal of Shipping and Logistics 33(3) (2017) 141-153

Contents lists available at ScienceDirect

The Asian Journal of Shipping and Logistics

Journal homepage: w w w .elsevier.com/locate/ajsl

The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method:

A Case Study of Southern Thailand

Kraisee KOMCHORNRITa
a
Ph.D. Candidate, Department of Industrial Engineering, Mahidol University, Thailand, Email: kraisee.k@gmail.com

ARTICLE INFO ABSTRACT

Article history: An intermodal transport could be the potential solution for reduction of logistics cost, including an
Received 12 June 2017
acceleration of product flow, by way of dry port, where is regarded as a key component of
Received in revised form 31 August 2017
Accepted 10 September 2017 combined transport. To promote a dry port in southern Thailand, the aim of this study is to propose
an integrated method based on multi-criteria decision making (MCDM) for selecting its best
Keywords: location. Thus, a new hybrid CFA-MACBETH-PROMETHEE model is presented. Confirmatory
Intermodal Transport factor analysis (CFA) is applied to determine loads for sorting out the importance of criteria and
Dry Port also investigate the relationship of logistic policy and geographical determination of dry port
CFA
establishment. Later, measuring attractiveness by a categorical based evaluation technique
MCDM
MACBETH (MACBETH) is utilized to build weights of those criteria. Eventually, preference ranking
PROMETHEE organization method for enrichment of evaluations (PROMETHEE) is engaged to rank from the
most to least attractive alternatives of dry port. In this study, results indicate that Phatthalung
railway station is the most attractive location for being a new dry port.

Copyright © 2017 The Korean Association of Shipping and Logistics, Inc. Production and hosting by
Elsevier B.V. T h i s i s a n o p e n a c c e s s a r t i c l e u n d e r t h e C C B Y - N C - N D l i c e n s e
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Broadly recognized as one of the most important parts in the economic
system, the international trade is the source to generate revenue of each
country. Thailand is among of them where the nation has greatly driven
economy through export. Currently, an export has been accounted for 65
of GDP (Tradingeconomics, 2016). To carry out this, Lamchabang seaport
of Thailand, one of the world’s busiest container ports with massive
capacity more than six million twenty-foot equivalent units, is respected
as the crucial terminal to support super-post-Panamax ship, gateway for
Greater Mekong Subregion and logistics hub of Southeast Asia. With
goods carriage and storage along numerous supply chains, however, the
overall logistics cost per GDP of Thailand is rather high, standing at 14.2
(i.e. transport: 7.4%; warehousing and inventory: 5.5%; administration:
1.3%) due to relying largely on the road transport, where its proportion is
approximately 80% of all transport modes (National Economic and Social
Development Board, 2014). The intermodal transport could be an
alternative for decreasing that cost. In general, it uses at least two modes
to convey container(s) without opening or repacking over the entire route.
Reis et al. (2013) indicate that cost structure of mode combination is

http://dx.doi.org/10.1016/j.ajsl.2017.09.004

2092-5212/© 2017 The Korean Association of Shipping and Logistics, Inc. Production and hosting by Elsevier B.V.
Peer review under responsibility of the Korean Association of Shipping and Logistics, Inc.
142 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand
significantly lower than the one of road mode when holding a long
distance. To extend this, an empirical work by Henttu and Multaharju
(2011) point out that the break-even points in cost of intermodal transport
(i.e. road and rail) against a single one (i.e. road) in transport network lie
in 157 and 255 km. for one and two additional transshipment, respectively.
Accordingly, a modal shift has appeared by utilizing a freight train on
railway as the major path, while a truck on road is served as a feeder of
rail transport. To connect between those two modes on purpose of goods
transshipment, a dry port is needed. The southern region of Thailand is
taken into consideration for establishment of dry port with the particular
reasons of peninsula (holding deep seaports on both coasts) along with a
long rail line, a possible increase of global trade in import/export and a
large volume of cross-border trade with such neighboring country as
Malaysia.
After all, a dry port may enhance the intermodal networks, which Thai
government has promoted through national policy in order to strengthen
the country’s overall logistics performance, whereas determinants of
transport node, transport link, production base and consumption point are
determined as factors of dry port establishment. Consequently, this study
will adopt CFA to examine the relationship between logistics policy and
geographical determination of dry port establishment, particularly factor
loadings of variables from the latter one for identifying their importance.
Then, the location analysis of selecting the most appropriate dry port
location is conducted by MCDM method. Commonly, each single method
of it has displayed advantage(s) and disadvantage(s) for any problem of
site selection. By heightening this, the integrated ones have continuously
been issued in the previous research, for instance combined MCDM
method, fuzzy set or others with MCDM one and so forth. Therefore, the
objective of this paper is to propose a new hybrid SEM-MACBETH-
PROMETHEE methodology for selecting the best dry port location in
southern Thailand. The new combined one could help in enhancing the
constraint of MACBETH and PROMETHEE. In order to sort out those
variables of geographical determination of dry port establishment, CFA is
firstly used to generate their loads. Subsequently, MACBETH is applied
to build weights of criteria, altered from those variables, via the pairwise
comparison computed by technique of linear programming. Eventually,
PROMETHEE I and II are manipulated to rank from the most to least
attractive sites.
2. Literature Review
2.1. Dry Port and Southern Thailand
With today’s highly intense competition in the domestic and
international markets, many firms have swiftly reacted to shift their
capabilities for retaining business growth. To cope with it, logistics
management is considerably viewed as a vital tool to succeed in cost
reduction as well as response of customer satisfaction. Yet, transport
sector has been widely regarded as the most essential part of logistics
owing to having the greatest impact upon logistics cost structure.
Alternatively, the combination of mode for goods movement from origin
to destination is likely to be the imperative choice to lower that cost. An
intermodal transport has consequently emerged, although some similar
terms of mode combination, e.g. multimodal or combined transport, can
be called interchangeably. Especially, it is defined as the movement of
goods in one and the same loading unit or road vehicle, which uses
successively two or more modes of transport without handling the goods
themselves in changing modes (UN, 2001). Hanaoka and Regmi (2011)
make clear that development of intermodalism requires the consideration
of three of its attributes: transport links (e.g. highway), transport nodes
(e.g. seaport) and the provision of efficient services (e.g. quality of trade).
As one of transport nodes, the dry port could play a major role in
promoting intermodal transport (Hanaoka and Regmi, 2011). Or else, a
dry port is a key component of intermodal transport (Hanaoka and Regmi,
2011; Henttu and Hilmola, 2011; Lättilä et al., 2013). In particular, a dry
port is an inland intermodal terminal directly connected to seaport(s) with
high capacity transport mean(s), where customers can leave/pick up their
standardized units as if directly to a seaport (Leveque and Roso, 2002). As
cited in Ng and Gujar (2009), CONCOR explains that a dry port supports
various needs along the supply chain, namely: aggregation and unitization
of cargoes, in-transit storage, custom clearance, issuance of bill of lading
in advance, relieving congestion in gateway seaports, assistance in
inventory management, as well as deference of duty payment for imports
stored in bonded warehouse.
To establish a dry port, the policy is considered as a major role.
According to ESCAP (Economic and Social Commission for Asia and the
Pacific) as cited by Hanaoka and Regmi (2011), the existing government
policies and regulations associated with dry ports influence their
development. In reference to Thailand’s policy of logistics aspect, the
government has hugely realized that logistics cost is important for the
country’s competitiveness, so some parts of the eleventh national
economic and social development plan (2012 – 2016) have concentrated
on development of logistics, where a dry port may be involved.
In the light of dry port location, the surrounding accessible places tied
up with a site of dry port on purpose of goods movement are required to
take into consideration as referred by the previous studies. Usually located
at strategic places near gateway seaports, industrial areas or along major
transportation axes, a dry port plays significant roles in optimizing all
activities to ensure cargoes can be delivered from one end to another in an
efficient manner (Juhel, 1999). An inland port must be linked with a port
terminal with a high capacity corridor where rail or barge dedicated links
are the best options (Rodrigue et al., 2010). Also, several dry ports have
an airport in proximity which can help support a variety of freight
activities (Rodrigue and Notteboom, 2012). In Asia, both road- and rail-
based intermodal dry ports are obviously nearing production, and
industrial centers are being developed with the aim of effectively
consolidating and distributing cargo (Hanaoka and Regmi, 2011). From
view of Rodrigue et al. (2010), an intermodal rail or barge terminal
enables to access from a port terminal to a regional production and
consumption market. Moreover, access to a large population base is of
importance since it will be linked to the level of import and export
activities handled by the inland port (Rodrigue and Notteboom, 2012).
Overall, the distance between possible sites of dry port and geographical
determinants of dry port establishment (transport node, transport link,
production base and consumption point) will be explored in order to rank
from the most to least attractive locations of dry port.
2.2. The Location Problem in Multi-criteria Decision Making
Nowadays, a large number of decisions have become a regular basis of
human life. However, with the radical change of development and
advancement in several issues (e.g. society, environment and technology)
it is even more complicated to make any decision correctly. For this
reason, the MCDM method seems to be an alternative tool to solve the
complex problem of those issues. In principle, it can be defined as the
evaluation of the alternatives for the purpose of selection or ranking, using
 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand
a number of qualitative and/or quantitative criteria that have different
measurement units (Özcan et al., 2011). The ultimate goal of MCDM
method is to investigate a number of alternatives in the light of multiple
criteria and conflicting objectives (Voogd, 1983). The location problem
for selection of the most appropriate site is regarded as the multi-criteria
decision because of taking different requirements into account.
Accordingly, this hints that the selection of location could be conducted
by MCDM technique. Since recognized by scholars as the considerable
aid, the MCDM process has been entailed in many problems of location
selection. Kovacic (2010) utilized PROMETHEE for selecting the site of
the nautical tourism port in the Northern Adriatic, and also her research
aimed at defining criteria and sub-criteria for optimizing the selection of
the location and facilities of nautical tourism port. Meanwhile, Athawale
and Chakraborty (2010) used the same technique to select the proper
facility location from a given set of alternatives on purpose to invest in the
place with cost minimization. Żak and Węgliński (2014) ranked regions
from the best to the worst one in terms of their suitability for locating the
logistics center by employing ELECTRE.
The newest trend in regard to MCDM method use is to combine two or
more methods to make up for shortcomings in any single particular
method (Velasquez and Hester, 2013). For instance, Yildirim and Önder
(2014) conducted the freight village analysis model by combining
between AHP and PROMETHEE. They collected opinions of experts in
order to turn into quantitative form by AHP, and then undertook
PROMETHEE to build the freight villages’ ratings. Sriniketha et al. (2014)
employed an integrated AHP and PROMETHEE methodology to solve a
problem of plant location selection, affecting performance of any
manufacturing firm; hence, this revealed that the appropriate location
resulted in higher economic benefits through an increased productivity
together with the good distribution network. In the research of Bu et al.
(2012), they arranged AHP-TOPSIS-GST approach for selecting optimal
city distribution reloading sites in urbanized areas.
Besides MCDM methods per se, some complements especially suitable
for decision making problems are cited such as fuzzy sets (Jato-Espino et
al., 2014). Chou et al. (2008) used the new fuzzy MCDM, or fuzzy simple
additive weighting system (FSAWS), for solving facility location
selection problems by utilizing objective and subjective attributes under
group decision-making conditions. Tabari et al. (2008) organized aid of
fuzzy AHP to consider the objective, critical, and subjective factors in
location analysis for any organization seeking a site for a new facility or a
relocation of the existing facility. Wang et al. (2014) proposed a fuzzy
AHP to select specific ports of call in order to provide information to port
operators that would enable them both to improve their management
strategies and to attract more cruise ships. The reason why fuzzy sets are
very present in hybrid approaches may lie in the distrust that these
methodologies generate concerning their capability to take into account
concepts like uncertainty and vagueness in data (Jato-Espino et al., 2014).
3. The Proposed CFA-MACBETH-PROMETHEE Method
Although presented as one of the most outstanding MCDM methods
applied in several disciplines, PROMETHEE has disclosed an obvious
drawback. It does not suggest how to properly generate weight for each
criterion. Macharis et al. (2004) explain that no specific guidelines are
provided to determine the weights. Likewise, it does not provide a clear
method by which to assign weights (Velasquez and Hester, 2013). By
removing that constraint, the merger between it and others is taken into
account. Extensively applied in decision environment, AHP is considered
143
as the full aggregation approach and the most popular one to build
weights of each criterion. Still, AHP has been criticized on its scale,
which pairwise comparison is utilized through eigenvalue method. The
fundamental scale from 1 to 9 is a ratio scale that associates a fixed
number with every semantic category, meaning that an extreme
importance of one option over another is related to a 9-fold higher
importance (Rietkötter, 2014). Also, AHP allows 10% inconsistency when
the consistency process is checked. On the contrary, MACBETH is likely
to outstrip AHP on those issues. MACBETH typically uses six semantic
categories on the ordinal scale by way of pairwise comparison relied upon
a technique of linear programming. When weights come into view, the
decision maker(s) can adjust them owing to occurrence of the certain
range between upper and lower limit. This identifies that weights of
MACBETH are more flexible than those of AHP. Besides, with regard to
a judgment matrix of MACBETH, if the matrix is sufficiently consistent,
the attractiveness can be calculated; otherwise the user is obliged to revise
judgments (Ishizaka and Nemery, 2013). This explicitly implies that
inconsistency is not accepted in MACBETH. With its constraint,
nonetheless, MACBETH clearly lacks in arrangement of criteria based on
their importance. Fortunately, CFA can enhance to define importance of
each criterion from factor loadings, derived from the consensus of a large
sample size.
Overall, CFA is used to sort out the important level of criteria, while
MACBETH is undertaken for determining weights of criteria; meanwhile
PROMETHEE is fulfilled for aggregation of those criteria and alternative
ranking. As a result, the new proposed CFA-MACBETH-PROMETHEE
method emerges to rank sites of dry port from the most to least
appropriate ones. Each method of the proposed one is described as follows.
3.1. Confirmatory Factor Analysis (CFA)
Principally considered as the measurement model of structural equation
modelling (SEM), CFA is employed to statistically test the significance of
a hypothesized factor model – that is whether the sample data confirm that
model (Schumacker and Lomax, 2010). Hatcher (2013) explains that CFA
produces indices that reflect the goodness of fit between the hypothesized
factor structure and the relationships that are actually observed in the
sample. It also produces factor loadings, inter-factor correlations, and
other results related to the psychometric properties of the model (Hatcher,
2013). Therefore, this implies that each observed variable could be put in
order from the most to least important ones in accordance with their
loadings.
In mathematical terms, Chou and Yeh (2013) present equations of the
measurement model of CFA based on regression analysis as follows:
X = Λxξ + δ (1)
This form can also be transformed into the following matrix:
ଵ Ɂଵ
൥ ‫ ڭ‬൩ GdGሾȦ ୶ ሿ୯ൈ୬ GRGሾɌሿ୬ൈଵGRG቎ ‫ ڭ‬቏ GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG(2)G
୯ ୯ൈଵ Ɂ୯
୯ൈଵ
As exposed by equation (1) and (2), X is the row vector for q observed
variables of ξ; Λx is the regression coefficient matrix (q×m) for exogenous
variables; ξ is the row vector for n exogenous variables; δ is the row
vector for the measurement error in q observed variables.
3.2. MACBETH
The MACBETH is an interactive approach that uses semantic
judgments about the differences in attractiveness of several stimuli to help
144 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand

a decision maker quantify the relative attractiveness of each (Bana e Costa a is preferred over b. If di(a, b) < qi then alternative a and b are indifferent.
and Vansnick, 1999). When the judgments and consistency are obtained, Formally, this can be summarized as:
the following linear programing (or LP-MACBETH) will be solved to Pi(a, b) = 0 if di(a, b) < qi (alternative a is indifferent to b)
gain the basic MACBETH scale (Bana e Costa et al., 2005): Pi(a, b) = 1 if di(a, b) > pi (alternative a is preferred to b)
min x1; Generally, there are six different types of preference function – usual,
subject to U-shape, V-shape, level, linear and Gaussian one (Brans and Mareschal,
(1) xp – xr = 0 ‫( ׊‬ap, ar) ‫ א‬I with p < r 2013). Upon characteristic of criterion, that specific preference function is
(2) σi + ½ ≤ xp – xr ‫ ׊‬i, j ‫ א‬N1,Q with i ≤ j, ‫( ׊‬ap, ar) ‫ א‬Cij in turn preceded.
(3) xp – xr ≤ σj+1 – ½ ‫ ׊‬i, j ‫ א‬N1,Q-1 with i ≤ j, ‫( ׊‬ap, ar) ‫ א‬Cij
(4) σ1 = ½ 3.3.2 Aggregated Preference Functions
(5) σi-1 + 1 ≤ σi ‫ ׊‬i ‫ א‬N2,Q
(6) xi ≥ 0 ‫ ׊‬i ‫ א‬N1,n In order to evaluate how much the alternative a is preferred to b over all
(7) σi ≥ 0 ‫ ׊‬i ‫ א‬N1,Q criteria, the preference index π(a, b) is calculated with a weighted sum of
The x stands for the different scores assigned to the elements (different the preference degrees Pi(a, b). The weights wi represents the importance
options or criteria), the σ represents the difference in attractiveness (also of each criterion in the decision:
called thresholds or intervals) between two elements. The I stands for a π(a, b) = σ௡௜ୀଵ ܲ௜ (a, b)wi
binary relation of indifference between two elements, the N1,Q represents where, Pi(a, b): preference degree on criterion i
the range of non-negative natural numbers between 1 and Q, Q is the wi: weight of criterion i
number of semantic categories, n serves as the number of variables and Cij n: number of criteria.
stands for the semantic categories which apply in case of a dominance If π(a, b) ≈ 0, it implies a weak global preference or indifference of a
relation. over b;
The objective function of this linear program is the minimization of the π(a, b) ≈ 1, it implies a strong global preference of a over b.
score of the most preferred element x1. The first constraint (1) concerns
the scores in case of identity, which means that no difference in 3.3.3 Outranking Flows
attractiveness exists between two elements. In this case, the difference in
scores and the difference in attractiveness have to be zero. The second (2) As each alternative is compared with m – 1 other alternatives, two
and the third (3) equations stand for the dominance constraints. With flows can be defined.
Positive flow:
element p being preferred to element r, the order in the element rankings ଵ
߶ ା (a, b) = σ ߨሺܽǡ ‫ݔ‬ሻ
has to be preserved and the order between the thresholds must be ௠ିଵ ௫‫א‬஺

maintained. Constraint number four (4) acts as an anchor for the intervals where, m: number of alternatives
of the semantic categories, with 0.5 being arbitrarily assigned. Equation A: the set of the m alternatives.
number five (5) symbolizes the minimal sigma differences between the This score represents the global preference of alternative a in
intervals and finally constraints six (6) and seven (7) are non-negativity comparison to all the other alternatives. Indeed, this score has to be
maximized.
constraints for the scores of the elements and the thresholds of the
Negative flow:
semantic categories. ଵ
߶ ି (a, b) = σ ߨሺܽǡ ‫ݔ‬ሻ
௠ିଵ ௫‫א‬஺
3.3. PROMETHEE where, m: number of alternatives
A: the set of the m alternatives.
PROMETHEE is an outranking method for a finite set of alternatives to This score represents the global weakness of a in comparison to all the
be ranked and selected among criteria, which are often conflicting other alternatives. Indeed, this score has to be minimized.
(Behzadian et al., 2010). The alternatives are first pairwise compared on
each criterion according to the decision maker’s preferences, resulting in 3.3.4 Ranking
local scores (Ishizaka et al., 2013). These local scores are then aggregated
to a global score, which lead to the PROMETHEE I or II ranking (Brans Based on the positive and negative flows, the PROMETHEE I partial
and Vincke, 1985). The following procedure of PROMETHEE I and II is ranking is defined as follows:
described as follows (Ishizaka et al., 2013): a is preferred to b if:
߶ ା (a) ൒ ߶ ା (b) and ߶ ି (a) < ߶ ି (b) or
3.3.1 Information within a Criterion ߶ ା (a) ൐ ߶ ା (b) and ߶ ି (a) ≤ ߶ ି (b)
a is indifferent to b if:
For each criterion ci, and for each ordered pair of alternative, the ߶ ା (a) = ߶ ା (b) and ߶ ି(a) = ߶ ି(b)
decision maker expresses his preference by means of a preference degree. a is incomparable to b:
The preference degree Pi(a,b) indicates if an alternative a is preferred or otherwise
not to b on the criteria ci based on the difference between their evaluation However, these two flows are usually combined to obtain the net flows
di(a, b). This preference degree is obtained using the preference function defined as follows:
which may require different parameters such as the indifference threshold ߶(a) = ߶ ା(a) െ ߶ ି (a)
qi and the preference threshold pi. If the difference di(a, b) between the which leads to the complete ranking of PROMETHEE II. The higher
score of alternative a and b on criterion ci is higher than pi, the alternative the net flows, the better the rank of an alternative.
 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand 145

4. Numerical Illustration Table 1

Variables of logistics policy with codes used in CFA
4.1 Area of Study Latent variable Observed variable Source
Agriculture (LP1)
Focusing on Thailand’s southern region (Fig. 1.), it is selected to study Transport projects (LP2)
because of an outstanding geography of peninsula. Clearly, it is a long Railway transport (LP3)
distance from north to south and narrow shape from east to west, attached Airports and seaports (LP4)
Distribution center (LP5)
by Gulf of Thailand and Andaman Sea. As its railway is in the vertical
Human resources (LP6) The 11th national
line – approximately 1,140 km. from Bangkok, trucks can be acted as Logistics laws (LP7) economic and social
Logistics policy (LP)
feeders from the surrounding highways. This seems to be compatible with IT (LP8) development plan
the intermodal transport. Deep seaports on east and west coast are NSW system (LP9) from 2012 to 2016

gateways to other countries and can link to a dry port. Consequently, one Cross-border logistics (LP10)
Cross-border rules (LP11)
of the existing railway terminals/stations is chosen as the potential dry
Economic corridors (LP12)
port by expanding its area with the purpose of cost saving on construction. Economic zones (LP13)
As mentioned earlier, the transport node and link are much considerable to Gas emission (LP14)
drive the intermodal transport, where a dry port is explicitly involved.
Also, a dry port can access to source of production and consumption Table 2
Variables of geographical determination of dry port establishment with codes used in
(Rodrigue and Notteboom, 2012; Roso et al., 2013; Ng and Cetin, 2012).
CFA
Thus, transport node, transport link, production base and consumption
Latent
point are used in this study, referred as geographical determination of dry Determinant Observed variable Source
variable
port establishment. Seaport (DEP1)
Transport node
Airport (DEP2) (Hanaoka and
Geographical
Transport link Highway (DEP3) Regmi, 2011)
determination
Production base Industrial area (DEP4)
of dry port
Local market (DEP5)
establishment (Rodrigue and
Consumption Regional market (DEP6)
(DPE) Notteboom,
point Cross-border market
2012)
(DEP7)

Nonetheless, there has been no proof of the relationship between those

latent ones. Accordingly, they are investigated whether the logistic policy
is related to the geographical determination of dry port establishment in
southern Thailand by CFA.

4.3 Determination of Criteria and Alternatives in MCDM

According to Table 2, those observed variables are viewed as the main

Fig. 1. Geography of southern Thailand
Source: thailandtravelbag.com
4.2 Determination of Variables in CFA
Regarding a policy to develop a dry port from ESCAP (Hanaoka and
Regmi, 2011) and the Thailand’s eleventh national economic and social
development plan (National Economic and Social Development Board,
2012), the evidence is likely to identify that logistics policy is considered
as the essential issue. As seen in Table 1, that policy is served as a latent
variable, comprising 14 observed variables, whereas a latent one of
geographical determination of dry port establishment with seven observed
ones is required to take into account as exhibited in Table 2.
criteria, but each of them may encompass with one criterion or more.
Those main ones can be later divided as three schemes. Firstly, criteria of
seaport, highway, industrial area and regional market are identified. Port
of Songkhla and Phuket are regarded as the major deep seaports
(Transport, 2014), where have dominated over the seaborne freight in the
southern region. The main highway in southern Thailand is Highway 4
(Phetkasem road) and 41, parts of Asian Highway (AH2), so those
alternatives of dry port can choose to reach only one of them. For
production base, the southern region industrial estate is located in Hatyai.
From view of Rodrigue and Notteboom (2012), they state that a large
population site is crucial for emergence of import and export. So, the most
important regional market is Bangkok, where the freight trains with
containers have now served between the existing southern railway
terminals and capital city. Secondly, criteria of airport and cross-border
market are selected by the 80/20 rule. In general, this principal asserts that
a minority of causes, inputs or effort usually lead to a majority of the
results, outputs or rewards (Koch, 1988). Phuket and Hatyai international
airport are picked out owing to proportional airfreight of 90.41%. Sadao
and Padang Besar are exposed as the places of cross-border trade with
occupancy of 98.84%. Finally, in terms of local market, the city
municipality, specifically over 50,000 people, is taken into account owing
to a large number of people. This is the same reason of the populous city,
where is described by Rodrigue and Notteboom (2012). So, Hatyai, Surat
146 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand

Thani and Nakhon Si Thammarat are chosen because those of them bear 4.4 Data Collection
the population over half (59.76%) out of total ones. The major criteria
with their criteria can be summarized in Table 3. In terms of accessible Based on Table 1 and 2, those observed variables, potentially having a
mode, road transport is directly related to potential dry ports and relationship with latent ones, were constructed as questions held in the
geographical determinants of dry port establishment. The reason behind first issue of questionnaire. The purpose is to collect all data of observed
this is that the rail accessibility between those above-mentioned points has ones bound to latent variables at once. All questions were inquired from
not existed, so the road mode should be replaced. If a position of dry port the sample size of 236, involving in operations at ICD Ladkrabang, to
location is precisely identified, the railway construction between those of assess scales ranging from one (“strongly disagree”) to five (“strongly
them will emerge. For instance, when ICD (inland container depot) agree”). The quantity of sample size were computed by Chi-square test in
Ladkrabang, only dry port of Thailand, had been found, the rail network G*Power software. When all data were completely preceded, the second
between it and Laemchabang seaport, the largest port of country, was later issue would be generated from response of the first one, particularly
built. loadings of variables in geographical determination of dry port
establishment. This enables to sort out the important levels of them. In the
Table 3 second one, 11 experts in field of logistics were invited to compare each
Main criteria and criteria for dry port location
pair of those criteria for weight building, including opinions of indifferent
Main criterion Criterion and preference threshold for distance between a dry port and those criteria.
Port of Songkhla
Seaport
Port of Phuket 4.5 Results
Phuket international airport
Airport
Hatyai international airport
Highway Highway 4 or 41 4.5.1 The Measurement Model
Industrial area Southern region industrial estate
City municipality of Hatyai There are two latent variables with 21 observed variables in the
Local market City municipality of Surat Thani measurement model as illustrated in Fig. 3. The first one is logistics
City municipality of Nakhon Si Thammarat policy (LP), encompassing 14 observed variables (LP1 – LP14). The other
Regional market Bangkok
one is geographical determination of dry port establishment (DPE) based
Sadao
Cross-border market on the distance between dry port and strategic points expressed by
Padang Besar
observed ones (DPE1 – DPE7). A two-headed arrow between LP and
DPE represents the correlation of them, and e values (e1 – e21) with
In regard to alternatives, current railway terminals/stations are assessed arrows along the left and right of observed ones are defined as errors. To
to develop as a new dry port in southern area. Five existing railway carry out this analysis, AMOS software is applied to compute the
terminals for container conveyance (Mapammarit, Banthungpho, Surat measurement model.
Thani, Thungsong and Hatyai) are chosen. Then, 15 first-class railway
stations, where population is dense, are initially in the spotlight, but five e1 LP1
of them (Chumphon, Bansong, Nakhon Si Thammarat, Trang, and e2 LP2
Phatthalung) are picked because their districts have population density
e3 LP3
higher on average than the southern ones (131.38 people per square km).
Also, there has been no threat from the terrorist attack, in particular the e4 LP4

destruction of railway. Therefore, those 10 possible positions of dry port e5 LP5 DPE1 e15

are depicted in Fig. 2. e6 LP6 DPE2 e16

e7 LP7 DPE3 e17

1
6 e8 LP8 DPE4 e18
LP DPE
e9 LP9 DPE5 e19

e10 LP10 DPE6 e20

2
3 e11 LP11 DPE7 e21

e12 LP12
8
7 e13 LP13

10 e14 LP14
4
Fig. 3. Diagram of the measurement model
9
With reference to the goodness-of-fit indices after modification of the
5
measurement model, seen in Table 4, χ2, where p of 0.71 is higher than
0.05, is considered as an excellent criterion and no statistical significance,
Fig. 2. The potential dry port positions of southern Thailand while χ2/df is at 0.94, indicating a good shape of criterion. RMR and
Source: Adapted from thailandtrains.com
Remark: 1 – 10 stand for Mapammarit, Banthungpho, Surat Thani, Thungsong,
RMSEA is 0.05 and 0.00, respectively, staying in the designated criterion.
Hatyai, Chumphon, Bansong, Nakhon Si Thammarat, Trang and Phatthalung, NFI is 0.90, falling on the acceptable criterion, while CFI is 1.00,
respectively exceeding the recommended level of 0.97. Finally, GFI and AGFI are
 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand 147

0.95 and 0.92, respectively. Both values are in excellent position of their of variables, the construct reliability (ρc) and average variance extracted
criteria. Concisely, those values are fit enough to support the measurement (ρv) for each latent variable are taken into account. Construct reliability
model. means that a set of latent indicators of construct are consistent in their
Also, such statistical value as factor loading, standard error (S.E.) and measurement (Lu and Yang, 2007). The highly reliable constructs are
critical ratio (C.R.) is tabulated in Table 4. Firstly, a factor loading those in which the indicators are highly intercorrelated, indicating that
represents the relationship between a latent variable and an observed one, they are all measuring the same latent construct (Deng et al., 2013).
or how weight observed one is loaded onto a latent one. Focusing on a Outcomes in Table 5 show that ρc of LP and DPE are at 0.90 and 0.79,
variable of logistics policy, L9 is the highest factor loading of 0.84; on the respectively, exceeding the level of 0.6 where regarded as the good value
other hand LP14 is the lowest one of - 0.07. For geographical (Bagozzi and Yi, 1988; Hair et al., 2009; Sanchez-Rodriguez et al., 2005).
determination of dry port establishment, DPE1 peaks at 0.74 of loading, An average variance extracted value measures the amount of variance in
but DPE5 bottoms out at loading of 0.02. Secondly, S.E. is the standard the specified indicators accounted for by the latent construct and is a
deviation in the sampling distribution for a parameter (Blunch, 2013). In complementary measure to the construct reliability value (Lu and Yang,
brief, S.E. values of observed variable are rather similar. Lastly, C.R., or t- 2007). Particularly, higher variance extracted values occur when the
value, is the estimate divided by its standard error (Blunch, 2013). A t- indicators are truly representative of the latent construct (Deng et al.,
value greater than 1.96 or smaller than - 1.96 implies statistical 2013). The value of ρv should be more than 0.5 for a latent one (Hair et al.,
significance at a level of 0.05 (Byrne, 2001). Results demonstrate that 2009; Bagozzi and Yi, 1988). As demonstrated in Table 5, ρv of LP and
LP11, LP12, LP14, DPE5, DPE6 and DPE7 are less than 1.96. This DPE are equally at 0.5, judged as acceptable values.
means that those of them are not significant for being the parts of LP and
Table 5
DPE. Furthermore, the correlation between LP and DPE is at 0.02,
Results of reliability for logistics policy and geographical determination of dry port
interpreting as the low relationship. establishment
Latent variable Observed variable R2 ρca ρvb
Table 4
CFA results for logistics policy and geographical determination of dry port LP 0.90 0.50
establishment LP1 0.46
Latent Observed Unstandardized Standardized LP2 0.23
S.E.a C.R.b
variable variable factor loading factor loading LP3 0.36
LP1 1.00 0.68 -c - LP4 0.15
LP2 0.77 0.48 0.10 7.66 LP5 0.27
LP3 0.99 0.60 0.11 8.73 LP6 0.36
LP4 0.65 0.38 0.12 5.57 LP7 0.61
LP5 0.79 0.52 0.10 7.67 LP8 0.59
LP6 0.89 0.60 0.10 8.80 LP9 0.71
LP7 1.24 0.78 0.12 10.54 LP10 0.20
LP
LP8 1.19 0.77 0.10 11.00 LP11 0.00
LP9 1.29 0.84 0.12 11.24 LP12 0.00
LP10 0.83 0.44 0.13 6.60 LP13 0.02
LP11 -0.06 -0.03 0.14 -0.41 LP14 0.00
LP12 -0.10 -0.05 0.14 -0.69 DPE 0.79 0.50
LP13 0.35 0.15 0.15 2.30 DPE1 0.55
LP14 -0.15 -0.07 0.15 -1.00 DPE2 0.10
DPE1 1.00 0.74 - - DPE3 0.50
DPE2 0.33 0.32 0.09 3.95 DPE4 0.02
DPE3 1.00 0.71 0.19 5.23 DPE5 0.00
DPE DPE4 0.23 0.15 0.11 2.04 DPE6 0.00
DPE5 0.03 0.02 0.10 0.27 DPE7 0.01
a
DPE6 0.08 0.06 0.10 0.80 ρc = (sum of standardized loadings)2 / [(sum of standardized loadings)2 + (sum of
DPE7 0.13 0.10 0.10 1.26 indicator measurement error)]
b
Fit indices: χ2 = 151.46 (df = 162, p = 0.71), χ2/df = 0.94, RMR = 0.05, RMSEA = ρv = sum of squared standardized loadings / [(sum of squared standardized loadings)
+ (sum of indicator measurement error)]
0.00, NFI = 0.90, CFI = 1.00, GFI = 0.95, AGFI = 0.92
a
S.E. is an estimate of the standard error of the covariance
b
C.R. is the critical ratio obtained by dividing the estimate of the covariance by its 4.5.2 Weights of Criteria of Dry Port Location
standard error. A value exceeding 1.96 represents a level of significance of 0.05.
c
Parameter fixed at 1.00 in the original solution. Those observed variables in a latent variable of geographical
determination of dry port establishment can be sorted out from the most to
According to Table 5 in the matter of reliability of observed variable, least important levels as exhibited by standardized factor loadings in
R2 is used to explain variance of a latent variable. Or, R2 can be used to Table 4. Afterwards, the observed ones are designated as the main criteria
measure the reliability of a particular observed variable (Koufteros, 1999). of MCDM, which MACBETH is concerned with weight construction.
LP9 and DPE1 reach the highest reliability for measurement of LP and Therefore, order of those main criteria with their factor loadings in
DPE at 71 and 55%, respectively. In contrast, LP11, LP12 and LP14, parenthesis is ranked as follows: seaport (0.74), highway (0.71), airport
along with DPE5 and DPE6 hit a trough with 0%. With respect to stability (0.32), industrial area (0.15), cross-border market (0.10), regional market
148 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand

(0.06) and local market (0.02), respectively. Later, logistics experts were Table 8
The overall weights of criteria
asked to make their judgment by comparing each pair of main criteria in
terms of how attractive of the one over another one based on distance Main criterion Criterion Weight (%)

from a dry port. Those main criteria are subsequently put into the matrix Port of Songkhla 18.56
Seaport
of judgment by ranking them in order of importance as presented in Table 6. Port of Phuket 3.10
Highway Highway 4 or 41 20.00
Phuket international airport 15.71
Airport
Table 6 Hatyai international airport 2.62
Matrix of judgments of main criteria Industrial area Southern region industrial estate 16.67
Sadao 12.00
Cross-border market
Padang Besar 3.00
Regional market Bangkok 7.78
City municipality of Hatyai 0.34
City municipality of Surat Thani 0.17
Local market
City municipality of Nakhon Si 0.05
Thammarat

4.5.3 Ranking of Alternatives for Dry Port Location

Remark: S is seaport; H is highway; A is airport; I is industrial area; C is cross-
border market; R is regional market; L is local market.
The hierarchical structure of criteria in PROMETHEE is firstly
classified into three types of cluster, group and criterion. As depicted in
The procedure of MACBETH in order to gain weights of main criteria
is performed through software of M-MACBETH. The first step is to Fig. 4, the structure is composed of four, seven and 12 clusters, groups
check the consistency of judgments. However, there are four points to and criteria, respectively.
modify due to inconsistent judgments. It suggests that the judgment
between airport and cross-border market should be moderately attractive,
while those judgments from each pair of industrial area and local market,
cross-border and local market and regional and local market are strongly
attractive. After that, weights of those criteria are executed by linear
optimization, where is used to minimize the score of the most attractive
criterion (Ishizaka and Nemery, 2013). The fixed weights with their
interval in % are tabulated in Table 7. Experts can adjust those weights in
that interval. Nonetheless, the outputs of those fixed weights satisfy them.

Table 7
Weights of main criteria
Main criterion Fixed weight (%) Interval of weight (%)

Seaport 21.66 20.35 – 22.08

Highway 20.00 18.65 – 21.29
Airport 18.33 16.96 – 19.66
Industrial area 16.67 15.27 – 18.02
Cross-border market 15.00 14.77 – 16.38
Regional market 7.78 7.27 – 8.28
Local market 0.56 0.02 – 1.10
Fig. 4. Hierarchical diagram of criteria in PROMETHEE
In Table 3, those main criteria may comprise one, two or three criteria.
Therefore, weights of two more criteria, particularly main ones of seaport, Secondly, the following preference parameters are involved. The
airport, cross-border market and local market must be adjusted. After minimum distance between criteria and dry port is preferred on purpose to
computation, the judgments in each pair of seaport, airport and cross- reduce logistics cost, whereas weights of criteria are previously
border market are consistent, while weights of them in % are as follows:
constructed by MACBETH. In terms of the preference function, average
Port of Songkhla and Phuket are 85.71 and 14.29%, respectively.
of q (indifference threshold) and p (preference threshold) from 11 experts’
Likewise, Phuket and Hatyai international airport are the same weight as
response are obtained (Table 9). This reveals that the linear preference
those ports. The cross-border points of Sadao and Padang Besar are 80
and 20%, respectively. On the contrary, the judgment between Hatyai and function, as depicted in Fig. 5, is chosen because the distance of each
Surat Thani is modified by adjusting to be strong attractiveness, while the criterion is regarded as the numeric data. The linear preference function is
others are consistent. Also, weights of Hatyai, Surat Thani and Nakhon Si the best choice for most quantitative criteria (Brans and Mareschal, 2013).
Thammarat are fallen on 61.53, 30.77 and 7.70%, respectively. The final So, the values of q and p are later existed. Specifically, if the difference
step is to convert weights of those criteria as displayed in Table 8. between the evaluations on a criterion is smaller than the indifference
threshold, then no difference can be perceived between two alternatives;
in contrast, if the difference is higher than the preference threshold, then
the preference is strong (Ishizaka and Nemery, 2013).
 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand 149

Table 9 the best one on the left-side bar, and reaching 0.0 is the best one on the
Average of indifferent threshold (q) and preference threshold (p) for all criteria (unit:
right-side bar. In Fig. 6, the ranking of alternatives on Phi+ from the best
km.)
to worst one is Hatyai, Phatthalung, Trang, Thungsong, Bansong, Nakhon
Criterion q p
Si Thammarat, Surat Thani (= Banthungpho), Chumphon and
Port of Songkhla Mapammarit, respectively. On the other hand, the ranking of those
9 26
Port of Phuket alternatives on Phi- is Phatthalung where is on the top, followed by Trang,
Phuket international airport Hatyai (=Thungsong), Bansong, Banthungpho, Surat Thani, Nakhon Si
5 13
Hatyai international airport Thammarat, Chumphon and Mapammarit, respectively. From
Highway 4 or 41 8 19
PROMETHEE I, it may be inferred that Hatyai is incomparable with
Southern regional industry estate 9 24
Phatthalung because it has a better score on Phi+ and a worse one on Phi-.
City municipality of Hatyai
Also, the incomparability is included Trang, Banthungpho and Nakhon Si
City municipality of Surat Thani 25 44
Thammarat. In contrast, those alternatives of Thungsong, Bansong, Surat
City municipality of Nakhon Si Thammarat
Bangkok 22 37
Thani, Chumphon and Mapammarit are obvious in the same order of Phi+
Sadao and Phi-. Additionally, the unicriterion net flows, positive minus negative
8 21 flow, are exhibited for each criterion of those alternatives in Table 11.
Padang Besar

Fig. 5. Graph of linear preference function (Ishizaka et al., 2013)

Remark: Pi(a,b) is preference degree; qi is the indifference threshold; pi is the
preference threshold; di(a, b) is the different between the score of alternative a and b
on criterion ci.

Thirdly, distance along the road between 12 criteria and 10 alternatives

of railway terminal/station are measured using Google Map as a tool. Fig. 6. PROMETHEE I partial ranking of alternatives
Those distances are summarized in Table 10. Finally, several results of
PROMETHEE I and II, after calculation via software of Visual Table 11
Unicriterion net flows of criteria
PROMETHEE, exist.

Table 10
Distance between criteria and alternatives (unit: km.)

For PROMETHEE II complete ranking, it is based on the net flows

Remark: PS is Port of Songkhla; PP is Port of Phuket; PIA is Phuket international
airport; HIA is Hatyai international airport; H4/41 is Highway 4 or 41; SRIA is only and leads to a complete ranking of the alternatives, and the
southern regional industry estate; CMH is city municipality of Hatyai; CMS is city incomparable status is not existed; those alternatives can thus be ordered
municipality of Surat Thani; CMN is city municipality of Nakhon Si Thammarat; B from the best to the worst (Ishizaka and Nemery, 2013). According to Fig.
is Bangkok; S is Sadao; P is Padang Besar; MA is Mapammarit; BA is Banthungpho;
7 and Table 12, the highest score of Phi belongs to Phatthalung.
SU is Surat Thani; TH is Thungsong; HA is Hatyai; CH is Chumphon; BAN is
Bansong; NA is Nakhon Si Thammarat; TR is Trang; PH is Phatthalung. Obviously, Phatthalung, Hatyai, Trang, Thungsong and Bansong are on
the top in positive scores. Banthungpho and Surat Thani fall on negative
Regarding PROMETHEE I partial ranking, it is based on the positive score and are very close to zero. Nakhon Si Thammarat and Chumphon
and negative flows (Ishizaka and Nemery, 2013). Typically, Phi+ are also in the negative side, while Mapammarit is at the bottom and
(positive flow) is a measure of strength, representing on the left-side bar shows the lowest score of net flow.
with the best value at the top of the bar and the worst one at the bottom,
while Phi- (negative flow) is a measure of weakness, representing on the
right-side bar with the best value at the top of the bar and the worst one at
the bottom (Brans and Mareschal, 2013). In other words, reaching 1.0 is
150 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand

city municipality of Hatyai, Sadao, Highway 4/41, city municipality of

Nakhon Si Thammarat and Padang Besar.

Fig. 7. PROMETHEE II complete ranking of alternatives

Table 12 Fig. 8. GAIA plane

Multicriteria positive, negative and net flows
Alternative Phi+ Phi- Phi 5. Conclusions
Mapammarit 0.1089 0.8752 -0.7662
Banthungpho 0.3625 0.3797 -0.0172 According to the previous literature, the logistics policy is regarded as
Surat Thani 0.3625 0.3912 -0.0287
an essential driver to found a dry port, but the CFA result indicates that
Thungsong 0.5308 0.2936 0.2372
Hatyai 0.5731 0.2936 0.3173 the relationship between it and geographical determination of dry port
Chumphon 0.2152 0.6257 -0.4105 establishment is very low, implying that Thailand’s logistics policy has no
Bansong 0.5010 0.3272 0.1738 significance on transport node, transport link, production base and
Nakhon Si Thammarat 0.4198 0.5402 -0.1204 consumption point of rail-based dry port establishment in the country’s
Trang 0.5545 0.2591 0.2954 southern region. Obviously, this is the same as southern Thailand’s
Phatthalung 0.5624 0.2429 0.3195
potential project of land bridge, where is regarded as the strategic route to
link sea transport between Pacific (east) and Indian (west) Ocean through
As a result, the evidence seems to indicate that Phatthalung is the best
railway in order to reduce travelling time of passing Strait of Malacca and
alternative of dry port site, followed by Hatyai, Trang, Thungsong and
transport cost. In other words, this zone may become the important hub to
Bansong, where Phi is positive. support regional and global trade in the future. Unfortunately, since
Moreover, PROMETHEE could reinforce with GAIA plane, which is introduced and held in the national master plan of logistics over 10 years,
described as a two-dimensional representation of a decision problem the project has been in the slow progress due to the main concern of
(Ishizaka and Nemery, 2013). The GAIA plane is a descriptive environmental impact, tourism industry and loss of traditional occupations
complement to the PROMETHEE rankings, and contains three types of of local people.
information – alternative represented by point, criterion represented by Furthermore, results suggest that variables of cross-border rule,
axis and weight of criterion represented by the decision axis (Brans and economic corridor, economic zone and gas emission should not be
Mareschal, 2013). In Fig. 8, there are four groups of alternative. included as parts of Thailand’s logistics policy in the future study because
Mapammarit and Chumphon are the first one, where both of them are in their values of factor loading and R2 are too low to take into account.
There are some reasons behind those variables from the sample size. The
the same province. Similarly, Banthungpho, Surat Thani and Bansong are
trade negotiation between two nations may be time consuming, while the
the second one and in province of Surat Thani. Nakhon Si Thammarat,
high investment cost and cooperation among several domestic and
Thungsong and Trang are the third one, where the first two stations are in international parties are the most concerned issues for the emergence
province of Nakhon Si Thammarat, but the latter one is close to those two economic corridor and zone. In the meantime, an issue of gas emission is
stations. Finally, the fourth group is Phatthalung and Hatyai, where both complex as the environment problem and will be affecting health of
of them are close to each other. people in the near future. Also, importance of a distance between a
Criteria of Port of Songkhla, Hatyai international airport, southern potential dry port and such determinant as industrial area, local market,
regional industry estate, city municipality of Hatyai and Sadao are at the regional market and cross-border market may be minimized for dry port
same point. Those of them together with Highway 4/41, city municipality establishment of Thailand due to the same reason of small ones of factor
of Nakhon Si Thammarat and Padang Besar are explicitly strong on loading and R2. Some of professionals point out that there is only one
alternatives of Nakhon Si Thammarat, Thungsong, Trang, Phatthalung industrial park with light industry, whereas a volume of goods for local,
regional and cross-border market is too little to carry by freight trains.
and Hatyai. Banthungpho, Surat Thani and Bansong are correlated with
The minimization of distance has been the only purpose to identify a
criteria of Port of Phuket, Phuket international airport and city
location of dry port via geographical determinants of dry port
municipality of Surat Thani; meanwhile, Mapammarit and Chumphon are
establishment, where the important levels of them are constructed by CFA
associated with criterion of Bangkok. Also, the decision axis, carrying through factor loadings. Nonetheless, the location selection is frequently
most weights of criteria, represents eight out of 12 criteria, which are Port considered as MCDM discipline, which the ultimate goal of it is to
of Songkhla, Hatyai international airport, southern regional industry estate, investigate a number of alternatives in the light of multiple criteria and
conflicting objectives (Voogd, 1983). Besides, the location selection is a
 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand 151

multi-criteria decision because it requires to take into consideration both

qualitative and quantitative factors (Cinar, 2009). As a consequence,
MCDM method is applied in the problem of site selection. While a
constraint of a single MCDM one has appeared, the combined method
becomes more widely recognized. In this study, the hybrid CFA-
MACBETH-PROMETHEE method is proposed. The objective is to fulfill
a limitation of two latter ones. MACBETH is employed because it is more
flexible to construct weights on ordinal scales and rejects inconsistency of
criteria, but displaying a disadvantage on sorting of criteria in order of
importance. By solving this, CFA is in turn used. PROMETHEE has been
extensively recognized as the solid method for ranking of alternatives
from the best to worst ones, but lacking of how to properly generate
weights of criteria.
With 12 criteria and 10 alternatives, the result reveals that Phatthalung
railway station, situated in city district, is the most attractive location for
being a prospective dry port in southern Thailand. In terms of geography,
the province of Phatthalung is located in the middle of southern region
and is a regional center of road transport between Gulf of Thailand (east)
and Andaman Sea (west) through Highway 4 to upper and lower boundary.
Moreover, the railway of it lies from north to south and connects most of
its districts. According to the provincial development plan of Phatthalung
between 2018 and 2021 (Governor, 2016) in Strategy IV of infrastructure
and logistics development, the spaces nearby of railway station will be
developed for supporting a track-doubling project. By doing this, the
freight transshipment will be shipped from the rail to road mode, and a
rail-based distribution center will be built. On this basis it may be inferred
that the expansion of Phatthalung railway station as a dry port conforms to
the above-mentioned development plan.
In regard to the future research, firstly, such factor as operations cost,
trade volume, economic level, community and environment should be
taken into account based on both quantitative and qualitative data in order
for catching sight of complete aspects of dry port establishment. Secondly,
criteria are specifically limited on Thailand, so the comparison with other
territories is suggested in the future investigation. Next, results of this
study are required to perform sensitivity analysis by comparing with other
single and combined MCDM methods, for example AHP, TOPSIS,
ELECTRE, VIKOR and integrated ones with fuzzy set theory. Finally, the
new proposed CFA-MACBETH-PROMETHEE method should apply to
other location problems, for example port, airport, distribution center,
freight village and so on.
References
BEHZADIAN, M., KAZEMZADEH, R. B., ALBADVI, A. and AGHDASI,
M. (2010), “PROMETHEE: A comprehensive literature review on
methodologies and applications”, European Journal of Operational Research,
Vol. 200, No. 1, pp. 198-215.
BLUNCH, N. J. (2013), Introduction to Structural Equation Modeling Using
IBM SPSS Statistics and AMOS, SAGE Publications.
BRANS, J. P. and MARESCHAL, B. (2013), “Visual PROMETHEE 1.4
Manual”.
BRANS, J. P. and VINCKE, P. (1985), “A PREFERENCE RANKING
ORGANISATION METHOD”, Management Science, Vol. 31, No. 6, pp. 647-
656.
BU, L., VAN DUIN, J. H. R., WIEGMANS, B., LUO, Z. and YIN, C.
(2012), “Selection of City Distribution Locations in Urbanized Areas”,
Procedia - Social and Behavioral Sciences, Vol. 39, pp. 556-567.
BYRNE, B. M. (2001), Structural Equation Modeling with AMOS: Basic
Concepts, Applications, and Programming, Lawrence Erlbaum Associates,
New Jersey.
CHOU, J.-S. and YEH, C.-P. (2013), “Influential constructs, mediating
effects, and moderating effects on operations performance of high speed rail
from passenger perspective”, Transport Policy, Vol. 30, pp. 207-219.
CHOU, S.-Y., CHANG, Y.-H. and SHEN, C.-Y. (2008), “A fuzzy simple
additive weighting system under group decision-making for facility location
selection with objective/subjective attributes”, European Journal of
Operational Research, Vol. 189, No. 1, pp. 132-145.
CINAR, N. (2009), “A Decision Support Model for Bank Branch Location
Selection”, International Journal of Mechanical, Aerospace, Industrial and
Mechatronics Engineering, Vol. 3, No. 12, pp. 18-23.
DENG, P., LU, S. and XIAO, H. (2013), “Evaluation of the relevance
measure between ports and regional economy using structural equation
modeling”, Transport Policy, Vol. 27, pp. 123-133.

ATHAWALE, V. M. and CHAKRABORTY, S. (2010), “Facility Location

GOVERNOR, O. 'Provincial development plan of Phatthalung 2018-2021'
Selection using PROMETHEE II Method”, Proceedings of the 2010
[online] http://www.phatthalung.go.th/develop_plan (Accessed May 28 2017).
International Conference on Industrial Engineering and Operations
Management, Dhaka, Bangladesh, pp. 59-64.
HAIR, J. F., BLACK, W. C., BABIN, B. J. and ANDERSON, R. E. (2009),
Multivariate Data Analysis, Prentice Hall.
BAGOZZI, R. P. and YI, Y. (1988), “On the evaluation of structural
equation models”, Academy of Marketing Science, Vol. 16, No. 1, pp. 74-93.
HANAOKA, S. and REGMI, M. B. (2011), “Promoting intermodal freight
transport through the development of dry ports in Asia: An environmental
BANA E COSTA, C. A., DE CORTE, J.-M. and VANSNICK, J.-C. (2005),
perspective”, IATSS Research, Vol. 35, No. 1, pp. 16-23.
“On the Mathematical Foundations of MACBETH”, in FIGUEIRA, J.,
GRECO, S. and EHRGOTT, M. (Eds.) Multiple Criteria Decision Analysis:
HATCHER, L. (2013), Advanced Statistics in Research: Reading,
State of the Art Surveys, Springer, USA, pp. 409-442.
Understanding, and Writing Up Data Analysis Results, Shadow Finch Media,
Michigan, US.
BANA E COSTA, C. A. and VANSNICK, J.-C. (1999), “The MACBETH
approach: Basic ideas, software and an application”, in MESKENS, N. and
HENTTU, V. and HILMOLA, O.-P. (2011), “Financial and environmental
ROUBENS, M. (Eds.) Advances in Decision Analysis, Kluwer Academic
impacts of hypothetical Finnish dry port structure”, Research in Transportation
Publishers, Dordrecht, pp. 131-157.
Economics, Vol. 33, No. 1, pp. 35-41.
152 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand
HENTTU, V. and MULTAHARJU, S. (2011), Transshipment Costs of
Intermodal Transport in Finnish Context. Finland.
ISHIZAKA, A. and NEMERY, P. (2013), Multi-Criteria Decision Analysis:
Methods and Software, John Wiley & Sons, Ltd.
ISHIZAKA, A., NEMERY, P. and LIDOUH, K. (2013), “Location selection
for the construction of a casino in the Greater London region: A triple multi-
criteria approach”, Tourism Management, Vol. 34, pp. 211-220.
JATO-ESPINO, D., CASTILLO-LOPEZ, E., RODRIGUEZ-HERNANDEZ,
J. and CANTERAS-JORDANA, J. C. (2014), “A review of application of
multi-criteria decision making methods in construction”, Automation in
Construction, Vol. 45, pp. 151-162.
JUHEL, M. H. (1999), “The role of logistics in stimulating economic
development”, China Logistics Seminar, Beijing.
KOCH, R. (1988), The 80/20 Principle, Nicholas Brealey Publishing, UK.
KOUFTEROS, X. A. (1999), “Testing a model of pull production: a
paradigm for manufacturing research using structural equation modeling”,
Journal of Operations Management, Vol. 17, pp. 467-488.
KOVACIC, M. (2010), “Selecting the location of a nautical tourism port by
applying PROMETHEE and GAIA methods: case study – Croatian Northern
Adriatic”, Promet – Traffic&Transportation, Vol. 22, No. 5, pp. 341-351.
LÄTTILÄ, L., HENTTU, V. and HILMOLA, O.-P. (2013), “Hinterland
operations of sea ports do matter: Dry port usage effects on transportation costs
and CO2 emissions”, Transportation Research Part E: Logistics and
Transportation Review, Vol. 55, pp. 23-42.
LEVEQUE, P. and ROSO, V. (2002), Dry Port concept for seaport inland
access with intermodal solutions. Chalmers University of Technology.
LU, C.-S. and YANG, C.-C. (2007), “An evaluation of the investment
environment in international logistics zones: A Taiwanese manufacturer's
perspective”, International Journal of Production Economics, Vol. 107, No. 1,
pp. 279-300.
MACHARIS, C., SPRINGAEL, J., DE BRUCKER, K. and VERBEKE, A.
(2004), “PROMETHEE and AHP: The design of operational synergies in
multicriteria analysis”, European Journal of Operational Research, Vol. 153,
No. 2, pp. 307-317.
NATIONAL ECONOMIC AND SOCIAL DEVELOPMENT BOARD, O.
(2012), The Eleventh National Economic and Social Development Plan.
NESDB, Bangkok, Thailand.
NATIONAL ECONOMIC AND SOCIAL DEVELOPMENT BOARD, O.
(2014), Thailand’s Logistics Report 2014. NESDB, Thailand.
NG, A. K. Y. and CETIN, I. B. (2012), “Locational Characteristics of Dry
Ports in Developing Economies: Some Lessons from Northern India”,
Regional Studies, Vol. 46, No. 6, pp. 757-773.
NG, A. K. Y. and GUJAR, G. C. (2009), “Government policies, efficiency
and competitiveness: The case of dry ports in India”, Transport Policy, Vol. 16,
No. 5, pp. 232-239.
ÖZCAN, T., ÇELEBI, N. and ESNAF, Ş. (2011), “Comparative analysis of
multi-criteria decision making methodologies and implementation of a
warehouse location selection problem”, Expert Systems with Applications, Vol.
38, No. 8, pp. 9773-9779.
REIS, V., FABIAN MEIER, J., PACE, G. and PALACIN, R. (2013), “Rail
and multi-modal transport”, Research in Transportation Economics, Vol. 41,
No. 1, pp. 17-30.
RIETKöTTER, L. (2014), The development and evaluation of guidelines for
the use of MACBETH in multi-criteria group decision making for the
assessment of new medical products. University of Twente.
RODRIGUE, J.-P., DEBRIE, J., FREMONT, A. and GOUVERNAL, E.
(2010), “Functions and actors of inland ports: European and North American
dynamics”, Journal of Transport Geography, Vol. 18, No. 4, pp. 519-529.
RODRIGUE, J.-P. and NOTTEBOOM, T. (2012), “Dry ports in European
and North American intermodal rail systems: Two of a kind?”, Research in
Transportation Business & Management, Vol. 5, pp. 4-15.
ROSO, V., RUSSELL, D., RUAMSOOK, K. and STEFANSSON, G. (2013),
“Connecting inland ports and seaports via intermodal transportation: A process
evaluation”, 13th WCTR, Rio de Janeiro, Brazil, pp. 1-20.
SANCHEZ-RODRIGUEZ, C., HEMSWORTH, D. and MARTINEZ-
LORENTE, A. R. (2005), “The effect of supplier development initiatives on
purchasing performance: a structural model”, Supply Chain Management: An
International Journal, Vol. 10, No. 4, pp. 289-301.
SCHUMACKER, R. E. and LOMAX, R. G. (2010), A Beginner’s Guide to
Structural Equation Modeling, Taylor and Francis Group, New York.
SRINIKETHA, D., REDDY, V. D. and PHANEENDRA, A. N. (2014),
“Plant location selection by using MCDM methods”, International Journal of
Engineering Research and Applications, Vol. 4, No. 12 (Part 1), pp. 110-116.
TABARI, M., KABOLI, A., ARYANEZHAD, M. B., SHAHANAGHI, K.
and SIADAT, A. (2008), “A new method for location selection: A hybrid
analysis”, Applied Mathematics and Computation, Vol. 206, No. 2, pp. 598-
606.
TRADINGECONOMICS “Thailand Exports” [online]
http://www.tradingeconomics.com/thailand/exports (Accessed April 6 2016).
TRANSPORT, M. “Statistics of Transport” [online]
http://www.mot.go.th/statmot.html?id=11 (Accessed August 6 2016).
UN (2001), Terminology on combined transport. NATIONS, U., Geneva,
Switzerland.
VELASQUEZ, M. and HESTER, P. T. (2013), “An Analysis of Multi-
Criteria Decision Making Methods”, International Journal of Operations
Research, Vol. 10, No. 2, pp. 56-66.
 The Selection of Dry Port Location by a Hybrid CFA-MACBETH-PROMETHEE Method: A Case Study of Southern Thailand 153

VOOGD, H. (1983), Multicriteria evaluation for urban and regional
planning, Pion, London.
WANG, Y., JUNG, K.-A., YEO, G.-T. and CHOU, C.-C. (2014), “Selecting
a cruise port of call location using the fuzzy-AHP method: A case study in East
Asia”, Tourism Management, Vol. 42, pp. 262-270.
YILDIRIM, B. F. and ÖNDER, E. (2014), “Evaluating Potential Freight
Villages in Istanbul Using Multi Criteria Decision Making Techniques”,
Journal of Logistics Management, Vol. 3, No. 1, pp. 1-10.
ŻAK, J. and WĘGLIŃSKI, S. (2014), “The Selection of the Logistics Center
Location Based on MCDM/A Methodology”, Transportation Research
Procedia, Vol. 3, pp. 555-564.