See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326435726

Transshipment Hub Port Competitiveness of the Port of Colombo against the

Major Southeast Asian Hub Ports*

Article  in  Asian Journal of Shipping and Logistics · June 2018

DOI: 10.1016/j.ajsl.2018.06.004

CITATIONS READS

11 692

3 authors:

Chathumi Ayanthi Kavirathna Tomoya Kawasaki

The University of Tokyo The University of Tokyo
17 PUBLICATIONS   37 CITATIONS    52 PUBLICATIONS   159 CITATIONS   

SEE PROFILE SEE PROFILE

Shinya Hanaoka
Tokyo Institute of Technology
124 PUBLICATIONS   1,218 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Global Scientist and Engineers Program View project

Emergency Preparedness and Response View project

All content following this page was uploaded by Chathumi Ayanthi Kavirathna on 04 March 2019.

The user has requested enhancement of the downloaded file.

 The Asian Journal of Shipping and Logistics 34(2) (2018) 071-082

Contents lists available at ScienceDirect

The Asian Journal of Shipping and Logistics

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

Transshipment Hub Port Competitiveness of the Port of Colombo against the Major
Southeast Asian Hub Ports*

Chathumi Ayanthi KAVIRATHNAa, Tomoya KAWASAKIb, Shinya HANAOKAc

a
PhD Student, Tokyo Institute of Technology, Japan, E-mail: chathumi.k.aa@m.titech.ac.jp
b
Assistant Professor, Tokyo Institute of Technology, Japan, Email: kawasaki@ide.titech.ac.jp
c
Professor, Tokyo Institute of Technology, Japan, Email: hanaoka@ide.titech.ac.jp

ARTICLE INFO ABSTRACT

Article history: The hub port competition has intensified beyond regional boundaries, resulting in the port of
Received 15 September 2017
Colombo competing with the Southeast Asian hub ports of Singapore, Klang, and Tanjung Pelepas
Received in revised form 15 February 2018
Accepted 31 May 2018 in the transshipment market. This study analyzes the competitiveness of Colombo as a
transshipment hub in “hub and spoke” and “relay” networks. Shipping lines evaluate the selection
Keywords: criteria for transshipment hub ports and the performance of competitive hub ports. The generalized
Transshipments cost approach together with a discrete choice model is used to assess port choice behavior by
Generalized Cost estimating the transshipment market share of hub ports and analyzing several scenarios. The results
Hub and Spoke
reveal Colombo’s lack of competitiveness in relay networks and most of the feeder ports in hub and
Relay
Port of Colombo spoke networks, mainly due to its high deviation from trunk sea routes and lower performance in
some non-quantitative criteria. Singapore has dominant market share in relay networks and for most
feeder ports in hub and spoke networks, mainly due to its high performance in non-quantitative
criteria, except Pipavav and Nhava Sheva, dominated by Colombo.

Copyright © 2018 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

As an economical transportation mode for international freight with multiple players and network structures (i.e., hub and spoke and
distribution, liner shipping is extensively used, despite its high complexity relay). Moreover, hub hopping behavior because of the availability of

* This work was supported by JSPS KAKENHI Grant Number 17H03327.

http://dx.doi.org/10.1016/j.ajsl.2018.06.004

2092-5212/© 2018 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.
72 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports
multiple hub ports in adjacent regions creates many alternative choices for
shipping lines. Transshipment refers to the shipment of goods or
containers to an intermediate destination before being taken to their final
destination (Soamiely et al., 2004), enabling shipping lines to maintain
minimum ports of call without restraining market coverage. The cascade
effect resulting from increasing vessel sizes, together with infrastructure
limitations in minor seaports, emphasizes the vital role of transshipment
operations.
The significant role of transshipment operations and the competitive
nature of the liner shipping industry synergistically create many
complications in the hub port selection process, while making it necessary
to consider a range of decision-making criteria. The situation becomes
more complex when competition rises among cross-regional hub ports for
serving overlapping origin/destination markets. The port of Colombo is
strategically located on the East-West main sea route as a major
transshipment hub in South Asia. Despite the location advantages,
currently Colombo competes with hub ports in Southeast Asia, such as
Singapore, Klang, and Tanjung Pelepas, as they develop their
transshipment market share beyond their regional boundaries (Szakonyi,
2015). Furthermore, Colombo is highly vulnerable, with serious
dependency on the Indian sub-continent feeder market, while not having a
significant role as a transshipment hub for relay networks (Szakonyi,
2015). Therefore, analysis of this competitive scenario creates an effective
contribution for Colombo to remain as a major transshipment hub port in
the region.
Since the study focuses on analyzing hub port competitiveness,
previous related studies were extensively reviewed. To identify
competitive dynamics, Yap et al. (2007) analyzed the market share,
growth of handling, shipping services, vessel sizes, and slot capacities
among East Asian ports. The results indicated that Chinese ports are
increasingly becoming attractive as direct calling ports, bypassing
Japanese and Taiwanese ports. According to the slot capacity analysis
carried out by Lam and Yap (2007), competition for Singapore port from
Port Klang and Tanjung Pelepas was highlighted, although Singapore was
consistently the premier transshipment hub in the region. These studies
indicate that, regardless of the dominant position of major hub ports, new
hub ports emerging with strong competitive advantages can have a huge
influence on shipping lines’ hub port selection decisions.
Moreover, structural changes in maritime markets can be influenced by
hub port competition. Ducruet et al.’s (2011) study on port competition
and network polarization in East Asia confirmed the progress of
secondary ports over their major competitors, while the overall network
structure tends to remain polarized by a few major hub ports that resist
internal and external threats. The effect of Chinese ports on Singapore
port was analyzed by Tongzon (2011), who indicated that, although the
Shanghai port has overtaken Singapore as the world’s busiest port, the
performance of the Singapore port has not been adversely affected, and
this complementary relationship might become competitive if the factors
of port choice change. This emphasizes the significance of analyzing hub
port selection factors as they change over time due to various market
influences. Yap and Lam (2006) studied the competitive dynamics among
East Asian container ports using port throughput data, while examining
the long-run relationships with co-integration tests. The
competition/cooperation between Shanghai and Ningbo-Zhoushan ports
was studied by Li and Oh (2010), while highlighting the impossibility of
cooperation, since each port has its own competitive advantages. Hence,
an endowment of competitive advantages is significant for hub ports to
withstand external market forces.
The competition between Busan and Kobe ports was analyzed by Ishii
et al. (2013) with game theory, considering the effect of port charges and
capacity. The significance of low port charges when there is high demand
elasticity was revealed, together with simultaneous port expansions of
competing ports. Hoshino (2010) studied the competition/collaboration
among Japanese ports for surviving major competition from Chinese and
Korean ports, and the significance of collaboration among Japanese minor
ports was highlighted. Hence, advance identification of competitive forces
is significant when determining strategies to maintain hub port
competitiveness.
These studies highlight the significance of analyzing hub port
competitiveness, as even major hub ports face strong competition in the
market. However, the competitiveness of hub ports may vary based on
different liner networks (i.e., hub and spoke and relay) although no
previous studies have focused on liner network types. In addition,
although previous studies confined their scope to analyzing competition
among hub ports from the same region, competition can be developed as
cross-regional matters, similar to the situation between Colombo and the
Southeast Asian hub ports. Therefore, the objectives of this study are to
identify the significant sources of competitiveness of the port of Colombo
for both hub and spoke and relay networks, compared to Singapore, Klang,
and Tanjung Pelepas, by estimating transshipment market shares, and to
analyze several practical scenarios to understand the potential implications
for the transshipment hub status of the port of Colombo.
As previous studies used different methodologies appropriate for
achieving their intended objectives, considering the objectives of this
study, the generalized cost approach is used, because both quantitative and
non-quantitative hub port performance measures can be incorporated with
generalized cost. The possibility of converting non-monetary performance
into monetary values with generalized cost enables identification of the
significant sources of competitiveness measured by high monetary values.
Since this study analyzes competitiveness via estimating market shares of
hub ports, generalized cost can be represented as the disutility of selecting
individual hub ports for transshipment operations.
2. Study Area
2.1. Study Area Selection
This study primarily focuses on the port of Colombo (CMB), while the
ports of Singapore (SIG), Klang (PKG), and Tanjung Pelepas (TPP) are
considered as competitive hub ports. The study area of the paper is
illustrated in Fig.1. The South Asia region is categorized into three
different feeder markets: the Indian East coast, South coast, and West
coast. The major feeder ports located in each feeder market and
competitive hub ports are introduced with relevant port codes in Fig. 1,
and are used throughout the study. Initially, to understand the competitive
situation, preliminary data analysis is carried out using container shipping
network data from 2013 provided by MDS Transmodal. Two different
indicators, annual service frequency and annual slot capacity of common
services, calling on both hub ports and feeder ports, respectively, are
considered, as shown in Table 1.


 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports 73

the advantage of using their own vessels, since transshipments take place
between mainline vessels. However, relay networks cannot be used
independently in many cases, if the ultimate origin and/or destination
ports are minor ports with infrastructure limitations.

Fig. 1. Study Area

Table 1 Fig. 2. Hub and Spoke and Relay Networks
Annual Service Frequency and Slot Capacities of Common Services
Feeder Feeder Annual Service Frequency Annual Slot Capacity (TEUs) Table 2
Market Ports SIG CMB PKG TPP SIG CMB PKG TPP
Transshipment Hub Port Selection Criteria Discussed in Previous Studies
KHI 407 338 303 156 1815820 1006564 1374964 276432
Indian Criteria References
MUN 159 260 147 0 692808 848198 594340 0
West
PAV 416 416 312 52 1917396 1718990 1406132 330148
Coast
NSA 562 588 562 104 2574810 2582584 2440172 562328 Chang et al. (2008), Chinonye et al. (2006), George and Hawa (2015),
Indian KOC 52 234 104 0 222976 447850 390312 0
Ishii et al. (2013), Koi (2006), Kurt et al. (1999), Lirn et al. (2003),
Cost
South
Saeed and Aaby (2012), Salem and El-Sakty (2014), Sedigheh
TUT 0 270 0 0 0 303482 0 0 (2015), Veldman and Buckmann (2003), Wang (2011)
Coast
MAA 412 305 364 52 984552 606992 939744 236288
Indian VTZ 140 149 104 52 284392 169048 256672 236288 Chang et al. (2008), Chinonye et al. (2006), George and Hawa (2015),
East HAL 130 52 173 0 120120 68692 177068 0
Location Koi (2006), Kurt et al. (1999), Lirn et al. (2003), Saeed and Aaby
Coast (2012), Salem and El-Sakty (2014), Sedigheh (2015), Wang (2011)
CCU 234 123 199 0 194896 86472 192876 0
CGP 520 260 381 0 782314 381524 587400 0
Chang et al. (2008), Chinonye et al. (2006), George and Hawa (2015),
Infrastructure/
Lirn et al. (2003), Koi (2006), Saeed and Aaby (2012), Salem and El-
Both indicators reveal the competition among SIG, CMB, and PKG Superstructure
Sakty (2014), Sedigheh (2015), Wang (2011)
from both the Indian East coast and West coast feeder markets, while TPP
Ducruet and Notteboom (2012), George and Hawa (2015), Kurt et al.
showed the lowest performance. However, since the relatively dominant Time
(1999), Koi (2006), Sedigheh (2015), Veldman and Buckmann (2003)
performance of CMB is highlighted from KOC and TUT feeder ports in
the Indian South coast market, hub port competition in this feeder market Productivity/Efficienc
Chinonye et al. (2006), George and Hawa (2015), Koi (2006), Kurt et
y/Loading
is not considered in this study. Therefore, the study analyzes competition al. (1999), Saeed and Aaby (2012), Sedigheh (2015), Wang (2011)
Discharging rates
individually, focusing on nine selected feeder ports that are located in the
Ownership/Managem George and Hawa, (2015), Koi (2006), Kurt et al. (1999), Lirn et al.
Indian West coast and East coast feeder markets and relay networks ent/Political Stability/ (2003), Salem and El-Sakty, (2014), Saeed and Aaby (2012),
related to the East-West trunk sea route. Regulations Sedigheh (2015)
Chinonye et al. (2006), Koi (2006), Li and Oh (2010), Saeed and
Service Quality
Aaby (2012), Tongzon (2011), Veldman and Buckmann (2003)
Ishii et al. (2013), Kurt et al. (1999), Koi (2006), Saeed and Aaby
2.2. Liner Network Types Capacity/ Congestion
(2012), Sedigheh (2015)

The study analyzes the competition related to two liner networks, Connectivity/Frequen Chang et al. (2008), Chinonye et al. (2006), Kurt et al. (1999), Saeed
cy of Ship visits and Aaby (2012), Veldman and Buckmann (2003), Wang (2011)
namely “hub and spoke” and “relay”, defined by Ducruet and Notteboom
(2012) as illustrated in Fig. 2. A hub and spoke network attempts service Cargo Availability/ Chang et al. (2008), George and Hawa (2015), Saeed and Aaby
optimization by integrating both mainline and feeder services between Handling Volume (2012), Tongzon (2011), Wang (2011)

ultimate origin and destination ports. It helps overcome the infrastructure Information
Koi (2006), Salem and El-Sakty (2014), Wang (2011)
constraints of minor ports by accommodating larger vessels only on the Technology
Chinonye et al. (2006), Ducruet and Notteboom (2012), Lirn et al.
trunk route, using small vessels for the feeder segments. The relay Damages/Safety
(2003), Koi (2006), Sedigheh (2015)
network optimizes the configuration by integrating multiple mainline
Logistics Services Koi (2006), Saeed and Aaby (2012), Sedigheh (2015)
services to maintain minimum ports of call without restraining market
Chinonye et al. (2006), Koi (2006), Lirn et al. (2003), Sedigheh
coverage. A network configuration allows cargo acceptance to/from Delays
(2015)
different ports that are not directly called by a liner’s own vessels since Chinonye et al (2006), D’Este and Meyrick (1992), Koi (2006),
Flexibility
the transshipment is undertaken with a connecting mainline vessel with a Sedigheh (2015)
different port calling pattern. Personal Contacts/
Although hub and spoke networks still dominate, shipping lines Marketing/Profession Saeed and Aaby (2012), Koi (2006)
currently attempt to minimize transshipments because of the high feeder al Personal

cost of third party feeder operators. In relay networks, shipping lines have
74 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports

3. Methodology region (Appendix A). After identifying the hub port selection criteria from
Table 2 and interview survey, all criteria were grouped into five categories
3.1. Generalized Cost Approach for Market Share Estimation as shown in Table 3, and further classified as quantitative and non-
quantitative. Since Table 2 highlights criteria related to cost, location,
The competitiveness of hub ports must be analyzed by incorporating
time, and operation as the most frequently used criteria for hub port
both quantitative and non-quantitative performance in an integrated selection, these four categories are initially identified. A liner-related
approach. The transshipment market share of hub ports is an effective category is included as a result of the interviews. Here, “H & S only”
indicator to reveal the hub port’s competitiveness, as high market share indicates criteria associated only with hub and spoke networks.
often results from strong competitiveness. Therefore, market share is Thereafter, the generalized cost function for hub port choice is defined
estimated, incorporating generalized cost with a multinomial logit model
with equation (1), such that each criterion in Table 3 is associated with
as a proven methodology for analyzing choice behaviors. However,
one component of the generalized cost function.
generalized cost commonly focuses on quantitative criteria rather than
incorporating benefits or costs of non-quantitative criteria. Thus, the
generalized cost function of this study considers both quantitative and

‡‡”ƒŽ‹œ‡†‘•– ሺ୦ሻ ൌ ‘‡–ƒ”›ሺ୦ሻ ൅ ‘—”‡›‘•– ሺ୦ሻ ൅
non-quantitative criteria to estimate the overall generalized cost for each
‹‡‘•– ሺ୦ሻ െ ሺ୦ሻ  ‫(Š׊‬1)
hub port choice, following a similar approach used by Koi (2006).
Initially, to identify hub port selection criteria, the criteria used in several
As per equation (1), generalized cost consists of the monetary values of
previous studies are summarized in Table 2, which highlights the criteria
port charges (MonetaryPC(h)), journey cost, time cost, and value of non-
most frequently used in studies. Although different studies have focused
quantitative criteria (VNQC) related to hub port h. In Table 3, both port
on different criteria, Table 2 indicates the scarcity of studies that
charge (PC) and journey cost related criteria are described in the monetary
considered a range of criteria within the same study.
category, and time cost related criteria are described in the time category.
Moreover, both monetary and time categories are grouped as quantitative,
Table 3
Hub Port Selection Criteria since those performances are evaluated with related quantitative data,
while the remaining three categories are grouped as non-quantitative
Type Category Criteria
categories. VNQC is defined as the monetary value perceived by shipping
Port related Port Charges (PC)
Monetary Journey Cost Deviation Cost
lines due to high/low performance of hub ports related to criteria in three
related Feeder Link Cost [H & S only] non-quantitative categories (i.e., location, operation, and liner-related) in
Quantitative Deviation Time Table 3. High VNQC indicates high performance and low VNQC
Vessel Turnaround Time indicates low performance of hub ports. Therefore, VNQC, which
Time
Waiting Time
increases the satisfaction/utility of the shipping lines, is introduced with a
Feeder Link Time [H & S only]
Location with Other Hub Ports negative sign in equation (1), as it must be deducted from other cost
Location Hub Port Accessibility components of the generalized cost function. The calculation of each
Location with Feeder Markets [H & S only] component of generalized cost and market share estimation are explained
Port Capacity
in the following subsections. The study calculates the generalized cost
Berth Availability
incorporating the perception of shipping lines.
Frequency of Delays
Records of Damages
Port Authority/Custom Policies/Regulations 3.1.1. Port Charges
Port Infrastructures “Port charges (PC)” are a significant component of the generalized cost
Port Superstructures
function. However, overall charges are difficult to obtain, since each port
IT and Advanced Technology
Operation Logistics Facilities has a different tariff structure (Meersman et al., 2014). Therefore,
Efficiency of Navigational Services container handling charges for an entire transshipment operation per TEU
Non-
Quantitative
Efficiency of Husbandry Services is directly obtained from individual port tariffs, and two times this value is
Professional Employees
considered as the monetary values for PC of hub port h (MonetaryPC(h)),
Marketing Efforts
as interviews with shipping lines indicated that container handling charges
Port’s Flexibility on Shipping Line Requests
Financial Clearance Capability represent around 50% of overall port charges.
Frequency of Ship Visits
Number of Services Calling at Port 3.1.2. Journey Cost Calculation
Availability of Dedicated/Own Terminal
Since it represents cost related to the journey, the journey cost
Personal Contacts
Special Preferences on Shipping Lines calculation consists of the two criteria listed under the monetary category
Liner-
Availability of Customer/Captive Cargo in Table 3; these are “deviation cost,” which represents the journey cost
related
Availability of Feeder Services [H & S only] when deviating from trunk sea routes to access a hub port, and “feeder
Opinions/ Preferences of Shippers and Forwarders
link cost,” which represents the journey cost between hub ports and feeder
Location of Hub Port with Shipping Line’s services
ports. Therefore, the journey cost calculation incorporates these two
criteria using the unit distance cost (UDC) approach. UDC (j, h) is defined
In addition to the literature review, several important criteria related to
as the journey cost for one unit of distance (cost per one nautical mile) of
shipping lines and network configurations were identified during the
the jth criteria for hub port h. The calculation of UDC(j, h) is done separately
interview and questionnaire survey conducted with 17 decision makers of
for deviation cost and feeder link cost, using a common method as
12 mainline shipping companies and 1 feeder shipping company in the
explained by equation (2).
 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports 75

౉౥౤౛౪౗౨౯ౌి
ሺ౞ሻ
ቆ
ఽ౬౏౏ሺౌిሻ
ቇ‫כ‬୅୴ୗୗሺౠሻ  3.1.3. Time Cost Calculation
ሺ୨ǡ୦ሻ ൌ ‫Œ׊‬ǡ ‫Š׊‬ሺʹሻ According to Table 3, the time category consists of “deviation time,”
ሺσ౞సర
౞సభ ୈሺౠǡ౞ሻൗସሻ
“vessel turnaround time,” “waiting time”, and “feeder link time”.
j indicates the criteria related to the journey cost, j= {deviation cost, feeder link Different calculation methods are used to calculate each time performance
cost} quantitatively, as shown in Table 4. Besides the time cost calculation, the
quantitative estimation of time performance, such as vessel turnaround
Where time, with reliable data sources is a significant contribution, since they can
z j indicates the criteria related to the journey cost be used straightforwardly to benchmark hub port performance.
z h represents any of four competitive hub ports (SIG, CMB, PKG, TPP) After estimating quantitative time values, time cost is calculated by
z D(j, h) refers to the journey distances (nautical miles) associated with the incorporating the value of time (VOT) approach. VOT(t, h) is defined as the
jth criteria for hub port (h). As an example, when j=deviation cost, cost of a unit of time (cost per hour) of the t th criteria for hub port h. The
D(deviation cost,h) represents the quantitative journey distance value for calculation of VOT(t, h) is done separately for each time related criteria,
calculating “deviation cost” of hub port h. Distance values are obtained using a common method, explained in equation (4).
using similar methods as the related time values mentioned in Table (4)
Table 4
z MonetaryPC(h) refers to the port charge of hub port (h) in monetary
Calculation Methods and Data Sources for Time Related Criteria
terms (US$).
Criteria Calculation Methods Data source
AvSS(j) indicates the “Average Significance Score” of the jth criteria to
Estimate average vessel turnaround
reflect its level of significance. Similarly, AvSS(PC) is the average Vessel
time considering container vessels Japanese Maritime
Turnaround
significance score of “port charge.” called between 1st September to 31st Institute
Time
A limitation of this study is that it calculates UDC assuming cardinal December
scale properties with AvSS, as the respondents were instructed to assume Deviation
Estimate average deviation time
AIS Marine Traffic
the cardinal scale when assigning scores for the significance of different considering the vessel
Time Data
access/departure routes of hub port
criteria. While assuming cardinal scale properties with AvSS, PC is used
Port Websites, IHS
as the base for monetization owing to the availability of real port charges
Waiting Time Directly obtained from data source Maritime and Trade,
(MonetaryPC(h)). Therefore, the monetary value of port charge October 2015
(MonetaryPC(h)) in US$ is divided by AvSS(PC) which indicates the Feeder Link AXS Marine distance
Directly obtained from data source
monetary value of one significance score multiplied by the AvSS(j) to Time data base
calculate the monetary value related to jth criteria.
౉౥౤౛౪౗౨౯ౌి
ሺ౞ሻ
Since the AvSS calculation is significant for each remaining section of ቆ
ఽ౬౏౏ሺౌిሻ
ቇ‫כ‬୅୴ୗୗሺ౪ሻ 
the methodology, a detailed discussion is provided as follows. To ሺ୲ǡ୦ሻ ൌ ‫–׊‬ǡ ‫(Š׊‬4)
ሺσ౞సర
౞సభ ୘ሺ౪ǡ౞ሻ ൗସሻ
calculate AvSS, the significance of each criterion is evaluated with a
questionnaire survey; while a sample of reputed shipping lines is selected t indicates the criteria related to time, t= {deviation time, vessel turnaround
using the snowballing sampling method, resulting in thirteen respondents. time, waiting time, feeder link time}
Due to the oligopolistic nature of the industry, the sample represents 63.09%
of the world’s liner shipping market, and 75.26% of transshipments in Where
focused feeder ports in terms of TEUs handled, according to MDS
z t indicates the criteria related to time
Transmodal data. The evaluation of hub port selection criteria is done by
z T(t,h) refers to the quantitative time value (hours) associated with the tth
incorporating a significance score with a scale ranging from 0 to 5 (0: No criteria for hub port (h) as calculated in Table 4. For example, when t =
significance at all, 1: Nearly no significance, 2: Little significance, 3: deviation time, T(deviation time,h) indicates the quantitative “deviation time”
Some significance, 4: Significant, 5: Very significant) as proposed by Koi of hub port h.
(2006). Thereafter, AvSS is calculated as the average score from the entire z Like the journey cost calculation, MonetaryPC(h) refers to the port
sample, to reflect the significance of individual criteria.
charge (US$) of hub port (h) and monetization is done considering the
After calculating UDCs of both deviation cost and feeder link cost with
relative significance of the tth criteria compared to “port charges;”
equation (2), total journey cost of hub port (h) is calculated using equation AvSS(t) indicates the average significance score of the tth criteria
(3).
After calculating the VOT of each time criteria using equation (4), total
‘—”‡›‘•– ሺ୦ሻ ൌ ൣ ‫ כ‬ሺୢୣ୴୧ୟ୲୧୭୬ୡ୭ୱ୲ǡ୦ሻ ‫ כ‬ሺୢୣ୴୧ୟ୲୧୭୬ୡ୭ୱ୲ǡ୦ሻ ൧ ൅ time cost of hub port (h) is calculated using equation (5).
ൣ ‫ כ‬ሺ୤ୣୣୢୣ୰୪୧୬୩ୡ୭ୱ୲ǡ୦ሻ ‫ כ‬ሺ୤ୣୣୢୣ୰୪୧୬୩ୡ୭ୱ୲ǡ୦ሻ ൧ ‫( Š׊‬3)

‹‡‘•– ሺ୦ሻ ൌ
Where
 ‫ כ‬ൣ൫ሺୢୣ୴୧ୟ୲୧୭୬୲୧୫ୣǡ୦ሻ ‫ כ‬ሺୢୣ୴୧ୟ୲୧୭୬୲୧୫ୣǡ୦ሻ൯ ൅ ൫ሺ୴ୣୱୱୣ୪୲୳୰୬ୟ୰୭୳୬ୢ୲୧୫ୣǡ୦ሻ ‫כ‬
z Journey Cost(h) is the total journey cost associated with hub port (h)
ሺ୴ୣୱୱୣ୪୲୳୰୬ୟ୰୭୳୬ୢ୲୧୫ୣǡ୦ሻ൯ ൅ ൫ሺ୵ୟ୧୲୧୬୥୲୧୫ୣǡ୦ሻ ‫ כ‬ሺ୵ୟ୧୲୧୬୥୲୧୫ୣǡ୦ሻ൯൧ ൅  ‫כ‬
z m is a dummy variable where m=1 for hub and spoke and m=2 for
ൣሺ୤ୣୣୢୣ୰୪୧୬୩୲୧୫ୣǡ୦ሻ ‫ כ‬ሺ୤ୣୣୢୣ୰୪୧୬୩୲୧୫ୣǡ୦ሻ൧ ‫( Š׊‬5)
relay, since the transshipments in relay networks occur between two
mainline vessels; hence two mainline vessels must be deviated from
Where
trunk sea routes to access hub port (h).
z Time Cost (h) is the total time cost related to hub port (h)
k is a dummy variable where k=1 for hub and spoke and k=0 for relay,
m and k are dummy variables with similar expressions as in the journey
as feeder link cost is not associated with relay networks
cost calculations
76 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports

୐୉୍ሺ౞ሻ ୓୉୍ሺ౞ሻ
3.1.4. Value of Non-Quantitative Criteria (VNQC) Calculation ሺ୦ሻ ൌ ሾ‘‡–ƒ”›ሺ୦ሻ ‫כ‬ ሿ ൅ ሾ‘‡–ƒ”›ሺ୦ሻ ‫כ‬ ሿ൅
୅୴ୗୗሺౌిሻ ୅୴ୗୗሺౌిሻ
Since the cost related to quantitative categories (i.e., monetary and time) ሾ‘‡–ƒ”›ሺ୦ሻ ‫כ‬
୐ୖ୉୍ሺ౞ሻ
ሿ ‫Š׊‬ሺͻሻ
୅୴ୗୗሺౌిሻ
were estimated in previous sections, this section considers incorporating
non-quantitative categories in generalized cost. For this, VNQC is “MonetaryPC(h)*LEI(h) /AvSS(PC)” indicates the overall monetary value
calculated assuming high VNQC resulting from the higher performance of from hub port (h)’s performance in the location category. Since LEI
hub ports significantly reduces the generalized cost of shipping lines than represents the performance of all criteria in the location category,
lower performance (Koi, 2006). specifically considering single criterion (i), this monetary value can be
To calculate VNQC, initially the efficiency indexes of individual expressed as “PC(h)* AvAS(location(i),h) * AvSS (location (i)) /AvSS(PC).”
categories are measured with equations (6), (7), and (8) Accordingly, monetization is done considering the ratio between the
AvSS of the ith criteria and the AvSS of PC, since PC are available in
monetary terms (MonetaryPC), while multiplication with AvAS is done to
ሺ୦ሻ ൌ ቂσ୧ୀ୬
୧ୀଵ ˜൫୪୭ୡୟ୲୧୭୬ሺ౟ሻ ǡ୦൯ ‫˜ כ‬൫୪୭ୡୟ୲୧୭୬ሺ౟ሻ൯ ቃ Ȁ ‫Š׊‬ሺ͸ሻ
appreciate the performance of individual hub ports. Since the
monetization is done based on PC, VNQC reflects the monetary value of
ሺ୦ሻ ൌ ቂσ୧ୀ୬
୧ୀଵ ˜൫୭୮ୣ୰ୟ୲୧୭୬ሺ౟ሻ ǡ୦൯ ‫˜ כ‬൫୭୮ୣ୰ୟ୲୧୭୬ሺ౟ሻ ൯ ቃ Ȁ ‫Š׊‬ሺ͹ሻ non-quantitative criteria of hub ports compared to port charges paid by
shipping lines; VNQC cannot be compared among ports as they are
ሺ୦ሻ ൌ ቂσ୧ୀ୬
୧ୀଵ ˜൫୪୧୬ୣ୰ି୰ୣ୪ୟ୲ୣୢሺ౟ሻǡ୦൯ ‫˜ כ‬൫୪୧୬ୣ୰ି୰ୣ୪ୟ୲ୣୢሺ౟ሻ ൯ ቃ Ȁ ‫Š׊‬ሺͺሻ
Where
z i generally refers to any criteria in the non-quantitative categories,
while location(i), operation(i), and liner-related(i) indicate the criteria
specifically related to location, operation, and liner-related categories,
respectively.
z LEI(h), OEI(h), and LREI(h) indicate “Location,” “Operation”, and
“Liner-related” efficiency indexes of hub port (h).
z “n” indicates the total number of criteria in each category; for example,
n=3 in equation (6), since the location category has three criteria.
z AvSS(location(i)) in equation (6) refers to the AvSS of the ith criteria in the
location category, with similar expressions in operation and liner-
related categories in equations (7) and (8). The calculation of AvSS
values is similar to the explanation in Section 3.1.2.
z AvAS(location(i),h) in equation (6) indicates the “Average Appreciation
Score” of hub port (h) in terms of the ith criteria in the location category,
and similar expressions in operation and liner-related categories in
equations (7) and (8), respectively.
AvAS is calculated to represent the average performances of individual
hub ports in terms of each criterion in the non-quantitative categories.
Before calculating VNQC, it is important to understand the non-
quantitative performances represented with AvAS values. AvAS for each
criterion is calculated as follows.
To calculate AvAS, the performances of hub ports related to each
criterion in the non-quantitative categories (i.e., location, operation, and
liner-related) are evaluated using the same questionnaire survey,
mentioned previously in AvSS calculations. The appreciation score, with
a scale ranging from -3 to +3 (-3: Very negative, -2: Negative, -1:
Somewhat negative, 0: Neutral, 1: Somewhat positive, 2: Positive, 3: Very
positive), is used for the evaluation, as proposed by Koi (2006).
Respondents were requested to assign an appropriate appreciation score
for hub ports in terms of individual criteria. Thereafter, AvAS is
calculated as the average score given by the entire sample, indicating the
level of performance of the hub port for individual non-quantitative
criteria.
After calculating efficiency indexes with equations (6), (7), and (8),
they are converted into monetary values, using equation (9) to calculate
VNQC. “Port charge” is used for the monetization assuming linearly
substitutable relationships among the criteria.
associated with their own PC values.
3.1.5. Estimating Market Share
After calculating each component, overall generalized cost is estimated
with equation (1) assuming the main ocean crossing portion of the journey
has no impact on the hub port selection decision. Thereafter, calculated
generalized cost values are converted into utilities with equation (10),
since utilities reflect the attractiveness of hub ports.
ଵ
–‹Ž‹–›ሺ୦ሻ ൌ  ృ౛౤౛౨౗ౢ౟౰౛ౚి౥౩౪ሺ౞ሻ ‫( Š׊‬10)

ృ౛౤౛౨౗ౢ౟౰౛ౚి౥౩౪ሺై౛౗౩౪ሻ
Where
z Utility(h) represents the utility of hub port (h)
z Generalized Cost(Least) represents the generalized cost of the best
performing hub port among the four hub ports.
Thereafter, a multinomial logit model is used to estimate the
transshipment market share of hub ports with equation (11).
౑౪౟ౢ౟౪౯ሺ౞ሻ
ୣ
ሺ୦ሻ ൌ ౑౪౟ౢ౟౪౯ሺ౞ሻ ‫Š׊‬ሺͳͳሻ
σ౞సౡ
౞సభ ୣ
Where
z EMS(h) indicates the estimated market share of hub port (h)
z ’k’ indicates the number of competitive hub ports related to the analysis
After estimating market share, the results are compared with the
predicted market share from a regression analysis. The relationships
between real market share and significant explanatory variables are
identified, with 38 container ports with high transshipment ratios (the ratio
between transshipment TEUs and local cargo TEUs is greater than 0.5)
using the linear regression method. The best fitting model is then used to
predict the transshipment market share from the feeder ports. Although
this approach cannot be considered a perfect validation of the model, as
both market shares are predicted values rather than real values, the author
adopts this analysis due to the unavailability of data on port to port
transshipment volume. Despite this limitation, since all hub ports are
treated equally in both methods, this comparison helps identify the
similarities or differences between the results of the model and the
regression analysis.
 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports 77

4. Results and Discussion

45
R² = 0.7198

Model Market Share (%)

4.1. Estimated Market Share of Each Hub Port 40

35
Market share is estimated as explained in the methodology section, and R² = 0.7993 SIG
30 CMB
the results are summarized in Fig. 3. In terms of hub and spoke, for the
Indian East coast feeder market, especially close to the Bay of Bengal, 25 PKG
R² = 0.4338
TPP
CMB has a relatively lower market share than SIG and PKG. CMB has
20 R² = 0.4573
advantages in the Indian West coast, with the highest market share in
15
PAV and NSA feeder ports. PKG demonstrates competitive performance,
15 20 25 30 35 40
while SIG has the dominant market share in most feeder ports. Regression Market Share (%)
Considering this, CMB should employ various strategies to maintain its
competitive performance, especially in the East coast market, due to the Fig. 4. Comparison between Model and Regression Market Shares
relatively high market shares of SIG and PKG.
4.1.2. Port Charge, Journey Cost, and Time Cost of Hub Ports
Since PC, journey cost, and time cost are significant components of
generalized cost functions, we can assume they may have major influence
on the estimated market share. The derived PC, journey cost, and time
cost values for all nine feeder ports and relay cases are summarized in Fig.
5. According to PC values, the highest and lowest PC belong to SIG and
CMB, respectively, highlighting that SIG could obtain a high market
share even with the highest port charge. According to the calculated
journey cost values, although PKG had the lowest journey cost, relatively
low journey costs from feeder ports are obtained by CMB, due to its
proximity to feeder ports, although CMB has a high journey cost in the
relay case.

Fig. 3. Estimated Market Shares for Hub and Spoke and Relay Cases Port Charge Journey Cost Time Cost
2500

TPP has the lowest market share in many feeder ports. Since TPP 2000 1440

1438
currently does not compete in transshipment traffic related to some feeder

1431

1421
1415

1393
1407
1374

1372
1355
1181

ports (i.e., CGP, CCU, HAL, MUN), in the preliminary data analysis, TPP
1117

1108

1101

1100
1500

1094

1086
1082
1075
1046
1049

1048
1035
Cost ($)

is excluded from the transshipment market share estimations related to

1133
964

1511
851

852
1076

852
851
1009
1006

927

898
998

860
840

1135
those ports. In the case of relay, the lowest market share is held by CMB, 1000

963
while the highest belongs to SIG. Therefore, CMB faces competitive
778

776

769

758
752
744
708

707
689

865
757

648

647

641

212 633
724

212 628
723

212 621
212 593

212 592
212 579

164 660

164 621

164 613
164 600

164 597
164 585

pressure not only from SIG and PKG, but also from TPP in the case of 500
168 494

168 492

168 488
168 488
168 477
168 458

168 438
168 438
168 426

272158

212208
168196
relay.
272

272

272

272

272

272

272

272

272
212

212

212
164

164

164

164
0
The estimated market share in Fig. 3 indicates the lack of
TPP

TPP

TPP

TPP

TPP

TPP

TPP

TPP

TPP

TPP
CMB

CMB

CMB

CMB

CMB

CMB

CMB

CMB

CMB

CMB
SIG

SIG

SIG

SIG

SIG

SIG

SIG

SIG

SIG

SIG
PKG

PKG

PKG

PKG

PKG

PKG

PKG

PKG

PKG

competitiveness of CMB, due to a relatively lower market share than SIG
and PKG in many feeder ports and relay cases. It is important to
understand the reasons behind this, considering each component of the
generalized cost function related to quantitative and non-quantitative
performance, since market share is estimated using the generalized cost
function.
4.1.1. Comparison Between Model and Regression Market Shares
After estimating market share, the results are compared with the
predicted results from the regression analysis as discussed in the
methodology section. Since the “numbers of vessels deployed in services”
was the best fitted model with transshipment volume in the simple linear
regression model (adjusted R2: 0.7902, p-value: 0.000, coefficients:
9395.27, intercept: 584273), the estimation of transshipment volume is
done considering “numbers of vessels deployed in common services,”
calling on both the respective hub port and feeder port, as the explanatory
variable. Hence, this comparison is limited to only the hub and spoke case.
Fig. 4 illustrates the relationship between model market shares and
regression market shares, where strong relationships were derived for both
SIG and PKG individual results.
PKG
CGP HAL VTZ CCU MAA NSA KHI PAV MUN
Hub and Spoke Relay
Fig. 5. PC, Journey Cost, and Time Cost of Hub Ports
In terms of time cost, quantitative estimation of time performances
indicates the least deviation time, vessel turnaround time, and waiting
time at SIG, PKG, and CMB respectively. The highest deviation time,
vessel turnaround time, and waiting time are found at CMB, SIG, and
PKG, respectively, whereas feeder link time values vary based on the
respective feeder ports. After converting quantitative time performances
into monetary values, the resulting time cost in Fig. 5 shows the highest
time cost for SIG could be due to long feeder time, while CMB has
considerably higher time cost in the relay case than in hub and spoke.
Since the calculations of costs related to quantitative performance were
made by incorporating AvSS values of the related criteria, the differences
in AvSS values can influence the magnitude of cost components in the
generalized cost function. Considering the criteria in the quantitative
categories, AvSS of PC for hub and spoke and relay networks were
calculated as 4.46 and 4.31, highlighting its high significance for hub and
78 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports

spoke networks. Two components of journey cost, deviation cost and
feeder link cost, have AvSS for hub and spoke of 4.46 and 4.69,
respectively, showing high significance for the feeder segment. In the case
of relay, the AvSS of deviation cost is 4.31, and there is no AvSS for
feeder link cost in relay as feeder segments are not associated with relay
networks. When considering time cost related criteria, deviation time,
vessel turnaround time, waiting time, and feeder link time, the AvSS for
hub and spoke are 4.69, 4.77, 4.46, and 4.77, respectively. In the relay
case, deviation time, vessel turnaround time, and waiting time have AvSS
as 4.31, 4.54, and 4.31, while feeder link time is not related to relay
networks. Derived AvSS values demonstrate different significance levels
for different criteria, while generating various influences on the
generalized cost function. Since individual hub ports indicate different
levels of quantitative performances for these criteria, the calculated cost
components vary among the four hub ports, although they use the same
AvSS associated with each criterion. AvSS of the criteria in quantitative
categories is more highly significant for hub and spoke than for relay, and
this could be one reason for generating higher cost for hub and spoke than
categories were derived as summarized in Appendix (B), which indicates
different AvSSs even for the same criteria when considering hub and
spoke and relay, due to different significance levels (Kavirathna et al.,
2018). Accordingly, the differences in VNQC among hub ports are
explained by the various significance levels of individual criteria reflected
in their AvSS values, together with differences in AvAS obtained by
individual hub ports as mentioned in the following section.
The hub port performances were evaluated for the criteria mentioned in
the non-quantitative categories (i.e., location, operation, and liner-related)
as in Appendix (C), highlighting the AvAS obtained by each hub port for
individual criteria, while results of the analysis of variance (ANOVA) and
special remarks are mentioned in the last two columns (Kavirathna et.al.,
2018). Referring to that, when analyzing the competitiveness of hub ports,
it is important to understand the different levels of performance indicated
by each hub port in terms of individual criteria.
40.00
Location- Indian East Coast Location- Indian West Coast
Location-Relay Operation

Efficiency Index
5.87 Liner Related
relay in Fig. 5 for all hub ports except CMB. In CMB, relay has high cost 8.69 5.09 3.98
20.00
due to the absence of advantages related to feeder segments, and 4.38 3.42
8.78 8.48 3.07
7.72 3.37 4.58
comparatively high deviation cost of CMB compared to other hub ports, 4.27
3.09 3.73 2.80
4.67 4.46 4.66 3.63
which could be one reason for the lowest market share for CMB in the 9.63 10.9 8.25 7.41 5.74 5.61 5.09 5.74
0.00
relay case. H&S Relay H&S Relay H&S Relay H&S Relay
As a summary, the overall results from cost calculations in quantitative SIG CMB PKG TPP
performances indicated advantages of CMB for hub and spoke, and Fig. 6. Efficiency Indicators of Non-Quantitative Categories
disadvantages for relay. The high market share of SIG may not directly
result from quantitative performance, as SIG has the highest port charge

1.55
2.00
1.47

1.42
and considerably higher journey cost and time cost values. Therefore, it is
VNQC ( Thousands $)

SIG
important to consider cost associated with non-quantitative performance 1.50
CMB
as well.
0.61

0.60
0.59

0.57
0.56

0.54
1.00
0.52

0.51
0.48
PKG
0.50
4.1.3. Value of Non-Quantitative Criteria of Hub Ports TPP
0.00
To estimate the VNQC, efficiency indexes were initially calculated as Indian East Coast Feeder Indian West Coast Relay Network Case
in Fig. 6. For the convenience of representation, “H & S” indicates hub Market (H & S) Feeder Market (H & S)

and spoke which has two location efficiency indexes representing the two
feeder markets (i.e., Indian east coast and Indian west coast). The derived
efficiency indexes can be used to benchmark port performances since they
incorporate both AvSS and AvAS. Fig. 6 highlights the highest and lowest
performances of SIG and TPP, respectively, for both networks. In
operation and liner-related categories, efficiency indexes of hub and spoke
are higher than relay, which could be due to the lower significance of the
criteria in relay.
Subsequently, VNQC were calculated by converting the derived
efficiency indexes into monetary terms, as seen in Fig. 7. Although
VNQC cannot be compared among ports, since monetization was done
based on port charges, the resulting VNQC indicates the monetary value
perceived by shipping lines compared to the cost paid as container
handling charges. All positive VNQC indicate that shipping lines perceive
satisfactory performances from all hub ports for non-quantitative criteria.
Since the VNQC is calculated based on the performance indexes derived
previously, SIG has advantages with high VNQC for both networks, and
this could be one main reason for generating a high market share for SIG.
However, CMB’s lower market share compared to that of PKG could
result mainly from the quantitative performance in the previous section,
although CMB has a comparatively higher VNQC than PKG.
Since the VNQC was calculated incorporating both AvSS and AvAS
values related to criteria in non-quantitative categories, it is important to
discuss their values as well. AvSSs of criteria in three non-quantitative
Fig. 7. Calculated Value of Non-Quantitative Criteria
As a summary of non-quantitative performances, SIG has advantages
due to the high AvAS obtained for most criteria, while generating high
VNQC. CMB has advantages over PKG and TPP in terms of non-
quantitative performance related to a few different criteria. This could be
the reason that CMB generates higher VNQC values in hub and spoke
networks. The high VNQC of both SIG and TPP in the relay case could be
the reason for generating their significant market shares. Due to the lower
significance level of criteria in relay (Appendix B), CMB has lower
VNQC for relay than hub and spoke, supporting its lower market share as
indicated in Fig. 3. Since monetization is done based on “PC,” the
difference in the AvSS of PC between hub and spoke and relay have a
significant influence on the different VNQC values of the two networks.
4.1.4. Generalized Cost
Finally, considering the overall generalized cost values in Fig. 8, the
highest market share of CMB for PAV and NSA, and the highest market
share of SIG for the remaining feeder ports and relay cases are reflected
by the significantly low generalized cost values related to them. Moreover,
significantly high generalized costs of CMB resulting from high costs in
quantitative categories, and lower VNQC in non-quantitative categories
has generated the lowest market share for CMB in the relay case.
 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports 79

2000
SIG CMB PKG TPP than in Scenario-1, with nearly similar variation between the two feeder
1800
markets. However, a significant difference was found between the two

1,848.93
1600
Generalized Cost ($)

1400 networks, with less impact on the relay case. Therefore, just being located
1,491.40

1,413.42

1,402.13
1,395.65
1,392.40

in proximity to the feeder market cannot generate a large market share,

1,386.56
1,385.46

1,377.06
1200

1,343.56
1,289.45

1,287.52
1,261.42

1,259.42
1000 because shipping lines consider multiple criteria simultaneously. Focusing

1,116.98

1,063.44

1,048.85
1,044.45

1,037.75
1,028.74
1,021.74
1,016.47
1,015.16

1,008.41
800

956.05
948.23

1,260.42
1,185.05

on this, operation and liner-related criteria are especially important for

879.07

876.55

1,111.94
1,097.94

1,086.80
1,082.16

1,081.03
841.56

1,038.84
1,019.02
600

813.45
400 CMB when benchmarking with the best performing ports. The higher

394.30
200 market share gain from scenario-2 than scenario-1 could be because the
0 current low performances of CMB (Appendix C) cannot be improved with
CGP HAL VTZ CCU MAA NSA KHI PAV MUN
just an infrastructure development project in scenario-1. However, despite
Hub and Spoke Relay
Case a considerable market share loss of PKG in the hub and spoke case,
Fig. 8. Calculated Generalized Cost scenario-2 indicates a slight market share gain for PKG in the relay case.

10
SIG CMB PKG TPP
Importantly, considering the overall results of this section, since the

% change of market share

results in Fig. 3 were derived incorporating a range of criteria, the derived 5
market shares of each hub port have different sources of competitiveness.
The lower handling charges and location advantages of CMB cannot 0
guarantee the highest market share, even for hub and spoke networks, as S-1 S-2 S-1 S-2 S-1 S-2 S-1 S-2 S-1 S-2 S-1 S-2 S-1 S-2 S-1 S-2 S-1 S-2 S-1 S-2

shipping lines consider multiple criteria simultaneously. As an example, CGP CCU HAL VTZ MAA NSA PAV MUN KHI Relay
-5 Case
from interviews, shipping lines appreciate the high operational
performance of SIG, regardless of high handling charges, due to the
competitive nature of the liner shipping industry. -10

Fig. 9. Results of Scenarios 1 and 2

5. Scenario Analysis 5.2. Scenarios Considering Performance Improvements in SIG

This section analyzes four different scenarios to understand the impacts As the previous two scenarios were based on a potential market share
of several practical implications. gain of CMB, Scenarios 3 and 4 are designed to analyze the potential
market share loss of CMB, considering different implications related to
5.1. Scenarios Considering Performance Improvements in CMB SIG as the main competitor of CMB.
Scenario-3 considered the potential market share loss of CMB from the
Since a lack of competitiveness of CMB was indicated from its derived SINGA (Sustainable Integrated Next Generation Advanced) project in
market share, the first and second scenarios focus on the potential market SIG. Although the project has multiple impacts on SIG, the scenario
share gains of CMB considering two different implications for CMB. focuses only on improvements in the time related criteria, assuming that
For Scenario-1, the potential market share gain from the Colombo port after completion of the project, vessel turnaround time and waiting time at
expansion project was considered, after identifying relationships between SIG will be reduced to match the current best performing port, since SIG
major project components and the hub port selection criteria in this study, currently shows relatively low time performance.
assuming a 50% increase in hub port performance (AvAS) in the affected Since SIG currently has the highest port charges among the four hub
criteria after the completion of the project. This increase was determined ports, scenario-4 analyzed the potential market share loss of CMB with a
because the development project has significant impacts on the reduction in SIG’s port charges, if port charges decrease to $204.00,
performance of CMB. which is the average charge among the four hub ports.
Scenario-2 focuses on the potential market share gain of CMB by The results of scenarios 3 and 4 are summarized in Fig.10. Scenario-3
improving performance up to the level of the current best performing port (S-3) indicates considerable market share loss from PKG and CMB in hub
in individual criterion, except those that are unrealistic, such as location or and spoke networks. Importantly, relay networks show a higher impact
journey cost. As an example, vessel turnaround time of CMB is than hub and spoke, highlighting that time performance improvements in
considered to be equal to that of PKG, which currently has the lowest SIG have considerable advantages in relay networks, while both PKG and
vessel turnaround time. Therefore, scenario-2 considers the significance of TPP have significant market share losses. However, the SINGA project
benchmarking with the best performing ports. will create various challenges for CMB, because although scenario-3 only
The results of scenarios 1 and 2 are summarized in Fig. 9. Scenario- considered time performance, the project can have a range of performance
1(S-1) indicates significant differences between the two feeder markets, improvements for SIG. In terms of scenario-4 (S-4) results, a considerable
with fewer market share changes in the Indian East coast feeder ports. market share gain for SIG was indicated, while both PKG and CMB are
Both SIG and PKG have almost similar market share losses in many negatively affected. Although relay networks have considerably more
feeder ports. The fewer overall impacts of scenario-1, especially in the impact from scenario-4 than scenarios 1 and 2, the impact is still less than
Indian East coast, reveal that development project components may not be the hub and spoke networks. However, scenarios 3 and 4 highlight the
capable of attracting a large market share from other hub ports. Moreover, significance of improving the time performance of SIG to attract a greater
the relay case has lower sensitivity to scenario-1 than hub and spoke. market share, especially in relay networks. Both scenarios generate
Results of scenario-2 (S-2) indicate a higher market share gain of CMB considerable market share losses for CMB in hub and spoke networks,
80 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports
although scenario-4 indicates a slight market share gain for CMB in the
relay case. Unlike the results in Fig. 9, there is no significant difference
between the results of two feeder markets in Fig.10.
15
SIG CMB PKG TPP
10
% change of market share
5
0 study. The Colombo port management is currently privatizing new
S-3 S-4 S-3 S-4 S-3 S-4 S-3 S-4 S-3 S-4 S-3 S-4 S-3 S-4 S-3 S-4 S-3 S-4 S-3 S-4
-5 CGP CCU HAL VTZ MAA NSA PAV
-10
-15
Fig. 10. Results of Scenario 3 and 4
6. Conclusion
This study has important industrial implications through its
identification of significant hub port selection criteria for hub and spoke
and relay networks, evaluation of competitiveness of hub ports, namely
Colombo, Singapore, Klang, and Tanjung Pelepas with market shares
estimation, and scenarios analysis. An important contribution for policy
makers is identifying the sources of competitiveness for each hub port,
which is critical for improving their performance. Owing to intense
competition, it is very important for port authorities and terminal
operators to understand the exact requirements of shipping lines, which
are the main customers of port facilities. This study contributes to the
literature by analyzing the competitive scenario among these four hub
ports considering each feeder ports individually, as previous studies do
not consider these four hub ports simultaneously, despite their competitive
scenario.
As the main findings, the dominant performance of Singapore was
highlighted for many criteria, while Port Klang and Colombo showed
competitive performances as well. The estimated market share of
Colombo was relatively higher in the Indian West coast feeder market
than in the East coast feeder market, although Singapore held the
dominant market share for most feeder ports, except Pipavav and Nhava
Sheva, which were dominated by Colombo. In the case of relay, hub ports
can be arranged as Singapore, Klang, Tanjung Pelepas, and finally
Colombo, based on descending order of market share. The scenario
analysis indicated considerable market share changes, especially for the
hub and spoke case.
There is significant focus on the Port of Colombo due to the existing
competitive scenario and Colombo’s high dependency on the
transshipment market in the Indian sub-continent. As per discussions with
the Sri Lankan Port Authority, while identifying the risk of being
seriously dependent on hub and spoke networks related to the Indian
feeder market, Colombo port management attempts to enhance relay
transshipment operations not only by creating better infrastructure but also
by reducing overall cost, especially bunkering cost, which is currently
comparatively higher in Colombo than in the South East Asian
competitive hub ports. Furthermore, port management is currently
planning advanced IT systems to streamline container operations, which
will eventually enhance shipping lines’ satisfaction, as this study also
reveals the lowest performance of Colombo in terms of IT and advanced
technology (Appendix C). Furthermore, based on the scenario analysis, it
is not possible to improve the current low performance of Colombo with
only infrastructure development; therefore, it is necessary to consider the
development of other aspects. The Colombo International Container
Terminal, which was opened in late 2014 in the port of Colombo and is
capable of handling the largest (18,000+TEUs) vessels, will help attract
more transshipments in relay networks (Szakonyi, 2015), as the current
low performance of Colombo in relay networks was also revealed in this
container terminals to enhance operation efficiency and market
MUN KHI Relay
Case
opportunities with extended networks similar to global terminal operators.
One limitation of this study is the evaluation of non-quantitative hub
port performance through perception-based data, although respondents
were selected from senior management in network planning divisions of
shipping lines. Moreover, this study uses a combination of two types of
data: ordinal data and ratio data, in both equations 2 and 4, as
monetization is done assuming cardinal scale properties with the average
significant scores of criteria, which is a limitation, although respondents
were provided with appropriate instructions on cardinal scale properties
beforehand. The calculation results obtained using ordinal data will be
subject to the scale used (e.g., 0-5 to 0-9). As the methodology treats all
hub ports equally throughout the calculation process, we assume that the
conclusion would not be significantly different even if another scale were
used, as the study considers the relative values between hub ports.
Validation of the results was not possible owing to the unavailability of
real transshipment volume data. For further research, the accuracy of the
model can be improved by developing a validation method, while
including new criteria and excluding non-significant criteria.
References
CHANG, Y., LEE, S. and JOSE, L. T. (2008), “Port Selection Factors by
Shipping Lines: Different Perspectives between Trunk Liners and Feeder
Service Providers”, Marine Policy Vol. 32, pp. 877–885.
CHINONYE, U., OGOCHUKWU, U. and INNOCENT, C. O. (2006), “An
Analytic Hierarchy Process (AHP) Approach to Port Selection Decisions-
Empirical Evidence from Nigerian Ports”, Maritime Economics & Logistics,
Vol.8, pp. 251–266.
D'ESTE, G. M., and S. MEYRICK. (1992), "Carrier selection in a RO/RO
ferry trade Part 1. Decision factors and attitudes", Maritime Policy &
Management Vol.19, pp. 115–126.
DUCRUET, C., and NOTTEBOOM, T. (2012), “Developing Liner Service
Networks in Container Shipping”, Maritime Logistics: A complete guide to
effective shipping and port management. Kogan Page, London, ISBN 978 0
7494 6369 4, pp. 77–100.
DUCRUET, C., SUNG-WOO, L. and ADOLF K.Y.N. (2011), “Port
Competition and Network Polarization in the East Asian Maritime Corridor”,
Territoire en mouvement Revue de géographie et aménagement [En ligne], 10 |
2011, mis en ligne le 01 juin 2013, consulté le 20 septembre 2017. URL:
http://tem.revues.org/1327; DOI: 10.4000/tem.1327
GEORGE, K.V.D. and HAWA, M.I. (2015). “Multi-Criteria Evaluation of
 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports 81

Port Competitiveness in West Africa Using Analytic Hierarchy Process
(AHP)”, American Journal of Industrial and Business Management, Vol. 5, pp.
432–446.
HOSHINO, H. (2010), “Competition and Collaboration among Container
Ports”, The Asian Journal of Shipping and Logistics, Vol. 26 (1), pp. 031–048.
ISHII, M., LEE, P.T., TEZUKA, K. and CHANG, Y. (2013), “A game
theoretical analysis of port competition”, Transportation Research Part E:
Logistics and Transportation Review, Vol. 49, pp. 92–106.
KAVIRATHNA, C.A., KAWASAKI, T., HANAOKA, S. and MATSUDA,
T. (2018), “Transshipment hub port selection criteria by shipping lines: the
case of hub ports around the bay of Bengal”, Journal of Shipping and Trade,
Vol.3 (4).
KOI, Y, A. (2006), “Theory and structure of port competition”, University
of Oxford.
KURT, I., BOULOUGOURIS, E. and TURAN, O. (1999), “An AHP
Decision Support Model for the Hub Port Choice of the Shipping Liners on the
Mediterranean Region”, University of Strathclyde.
LAM, J. S. L. and YAP, W. Y. (2007), “Competition for Transshipment
Containers by Major, Ports in Southeast Asia: Slot Capacity Analysis”,
International Conference on Logistics, Shipping and Port Management.
the United States”, U.S. International Trade Commission.
SZAKONYI, M. (2015, October 20), “Colombo breaks through as South
Asia’s next big transshipment port”, http://www.joc.com/port-news/asian-
ports/port-colombo
TONGZON, J. L. (2011), “The Rise of Chinese Ports and Its Impact on the
Port of Singapore”, Annual International Workshop on Port Economics and
Policy.
VELDMAN, S.J. and BUCKMANN, E.H. (2003), “A Model on Container
Port Competition: An Application for the West European Container Hub-
Ports”, Maritime Economics and Logistics, Vol. 5, pp. 3–22.
WANG, L. (2011), “Container seaport selection criteria for shipping lines in
a global supply chain perspective: implications for regional port competition”,
Erasmus University Rotterdam.
YAP, W.Y. and LAM, J.S.L. (2006), “Competition dynamics between
containers ports in East Asia”, Transportation Research Part A: Policy and
Practice, Vol. 40, pp. 35–51.
YAP, W. Y., LAM, J.S.L. and NOTTEBOOM, T. (2007), “Developments in
Container Port Competition in East Asia”, Transport Reviews, Vol. 26 (2), pp.
167–188.

LI, J. and OH, Y. (2010), “A Research on Competition and Cooperation Appendix A: Study Sample
between Shanghai Port and Ningbo-Zhoushan Port”, The Asian Journal of Questionnaire
Organization Interviews
Shipping and Logistics, Vol. 26, pp. 067–092. Survey
Hyundai Merchant Marine (HMM) Yes 㻌
Mediterranean Shipping Company (MSC) Yes 㻌
LIRN, T.C., THANOPOULOU, H.A. and BERESFORD A.K.C. (2003),
Maersk line Yes Yes
“Transshipment Port Selection and Decision-making Behavior: Analyzing the Orient Overseas Container Line (OOCL) Yes Yes
Taiwanese Case”, International Journal of Logistics: Research and CMA-CGM Group Yes 㻌
Applications, Vol. 6 (4). Nippon Yusen Kaisha (NYK) line Yes Yes
Evergreen Shipping Yes 㻌
Hapag-Lloyd Yes 㻌
MDS Transmodal Inc. MDS containership databank. (2013)
"K" LINE Kawasaki Kisen Kaisha, Ltd. Yes 㻌
Mitsui O.S.K. Lines (MOL) Yes Yes
MEERSMAN, H., STRANDENES, S.P., and VOORDE, E.V. (2014), Port Hanjin Shipping Yes Yes
Pricing: Principles, Structure and Models. Handbook of Research Methods and American President Lines (APL) 㻌 Yes
Applications in Transport Economics and Policy, Edward Elgar, Cheltenham, Bengal Tiger Line (BTL) Yes Yes
forthcoming. Sri Lankan port authority (SLPA) 㻌 Yes

SAEED, N. and AABY, B. C. (2012), “An Analysis of Factors Contributing Appendix B: Average Significance Scores (AvSS) of Criteria in Non-
as Selection Criteria for Users of European Container Terminals”, Molde Quantitative Categories
University College. Hub and
Category Criteria Relay
Spoke
SALEM, I.I. and EL-SAKTY, K.G. (2014), “Port Selection Criteria and Its Location with Other Hub Ports 4 3.85
Location Hub Port Accessibility 4.23 4.39
Impact on Port Competitiveness”, International Journal of Humanities and
Location with Feeder Markets 4.15 N/A
Social Sciences (IJHSS), ISSN(P):2319-393X, ISSN(E): 2319-3948, Vol. 3,
Port Capacity 4.31 4.23
Issue 6, pp. 29–36. Berth Availability 5 4.85
Frequency of Delays 4.69 4.77
SEDIGHEH, Z. (2015), “The Key Factors in Shipping Company’s Port Records of Damages 3.54 3.62
Operation Port Authority/Customs Policies/Regulations 4 3.85
Selection for Providing Their Supplies”, International Journal of Social,
Category Port Infrastructures 4.15 4.08
Behavioral, Educational, Economic, Business and Industrial Engineering, Vol. Port Superstructures 3.92 3.92
9, 4. IT and Advanced Technology 4 3.69
Logistics Facilities 4.08 3.62
SOAMIELY,A., HUGH, A. and MICHAEL, J.F. (2004), “Transshipment in Efficiency of Navigational Services 4.31 4.08
82 Transshipment Hub Port Competitiveness of the Port of Colombo against the Major Southeast Asian Hub Ports

Efficiency of Husbandry Services 3 3.15

Professional Employees 4 3.92
Marketing Efforts 3.23 3
Port Flexibility on Shipping Line Requests 4.15 4.15
Financial Clearance Capability 3.62 3.62
Frequency of Ship Visits 3.92 3.85
Number of Services Calling at Port 3.92 3.92
Availability of Dedicated/Own Terminal 3.08 3.23
Personal Contacts 3.23 3.15
Special Preferences on Shipping Lines 3.23 3
Availability of Customers/Captive Cargo 3.77 3.62
Liner-related
Availability of Feeder Services 3.77 N/A
Category
Opinions/Preferences of Shipper and
3.62 3.54
Forwarders
Location of Hub Port with Shipping Line’s
3.69 3.62
Services

Appendix C: Average Appreciation Scores (AvAS) of Hub ports

Criteria SIG CMB PKG TPP Pr(>F) Remarks
Location with Other Hub SIG>PKG,
2.62 1.92 1.08 1.23 0.001**
Ports SIG>TPP
Hub Port Accessibility 2.69 1.69 1.62 1.54 0.077


Location with Indian East 0.001 CMB>PK,
1.69 2.39 1.46 0.92
coast Feeder Market ** CMB>TPP
Location with Indian West 0.001 CMB>PK,
1.08 2 0.69 0.62
coast Feeder Market ** CMB>TPP
Port Capacity 2.54 1.69 1.39 1.08 0.024 * SIG>TPP
0.009 SIG>PKG,
Berth Availability 2.31 1.31 0.92 1
** SIG>TPP
0.009 SIG>PKG,
Frequency of Delays 1.85 0.77 0.54 0.62
** SIG>TPP
Records of Damages 1.62 1 1.08 1.08 0.573

Policies and Regulations 2 0.85 1.23 0.92 0.082

SIG>CMB,
Port Infrastructures 2.46 1.23 1.46 1.15 0.010 *
SIG>TPP
SIG>CMB,
Port Superstructures 2.39 1.23 1.31 1 0.010 *
SIG>TPP
IT and Advanced SIG>CMB,
2.31 0.85 1.39 1.08 0.011 *
Technology SIG>TPP
0.009 SIG>CMB,
Logistics Facilities 2.62 1.31 1.46 1.08
** SIG>TPP
Efficiency of Navigational
2.15 1.31 1.46 1.08 0.107


Services
Efficiency of Husbandry
2.39 0.85 1.15 0.62 0.107


Services
Professional Employees 2.15 1.15 1.23 1 0.047 * SIG>TPP

Marketing Efforts 1.77 0.85 1.08 1.08 0.235

Port’s Flexibility 1.69 0.85 0.62 0.69 0.163

Financial Clearance SIG>CMB,
1.92 0.77 1.08 0.77 0.029 *
Capability SIG>TPP
SIG>PKG,
Frequency of Ship Visits 2.54 1.39 1.31 1 0.004** SIG>CMB,
SIG>TPP
SIG>PKG,
No. of Services Calling at 0.000
2.23 1.08 0.92 0.62 SIG>CMB,
Port **
SIG>TPP
Availability of
1.15 1.08 0.85 1.08 0.938


Dedicated/Own Terminal
Personal Contacts 1.62 1 0.77 0.69 0.064


Special Preferences on
1.08 0.85 0.77 0.77 0.88


Shipping Lines
Availability of Customers/
1.54 1.08 1.31 0.92 0.674


Captive Cargo
Availability of Feeder 0.002 SIG>PKG,
2.62 1.92 1.31 1.15
Services ** SIG>TPP
Opinion/ Preferences of 0.008 SIG>PKG,
2.08 1.23 0.77 0.77
Shipper and Forwarders ** SIG>TPP
Location of Hub Port with
1.54 0.77 1 0.77 0.29


Shipping Line’s services
Note: **and *represent significance at the 1% and 5% levels, respectively. SIG
> PKG means that the mean score of SIG is significantly higher than that of
PKG.

View publication stats