GaWC Research Bulletin 364

GaWC logo
  
 
  Gateways into GaWC

This Research Bulletin has been published in Global Networks

doi:10.1111/j.1471-0374.2011.00355.x

Please refer to the published version when quoting the paper


(Z)

The Worldwide Maritime Network of Container Shipping: Spatial Structure and Regional Dynamics

C. Ducruet* and T. Notteboom **

Abstract

Port and maritime studies dealing with containerization have observed traffic concentration and dispersion throughout the world. Globalization, intermodal transportation, and technological revolutions in the shipping industry have resulted in both network extension and rationalization. However, lack of precise data on inter-port relations prevent the application of wide network theories to global maritime container networks, which are often examined through case studies of specific firms or regions. This paper presents an analysis of the global liner shipping network in 1996 and 2006, a period of rapid change in port hierarchies and liner service configurations. While it refers to literature on port system development, shipping networks, and port selection, it is one of the only analyses of the properties of the global container shipping network. The paper analyzes the relative position of ports in the global network through indicators of centrality. The results reveal a certain level of robustness in the global shipping network. While transhipment hub flows and gateway flows might slightly shift among nodes in the network, the network properties remain rather stable in terms of the main nodes polarizing the network and the overall structure of the system. Additionally, mapping the changing centrality of ports confirms the impacts of global trade and logistics shifts on the port hierarchy and indicates that changes are predominantly geographic.

Key words: liner shipping, network analysis, nodal regions, port hierarchy, spatial change


INTRODUCTION

Maritime networks are among the oldest forms of spatial interaction. Port hierarchies and the spatial pattern of maritime linkages can be considered as illustrations of wider ongoing processes, such as the regionalization and globalization of trade patterns and business cycles, thus revealing a certain political economy of the world (Vigarié, 1995). Lewis and Wigen (1999) argue that the meta-geography of the world system would be better understood from the maritime looking glass of basins, seas, and oceans. Following decades of adaptation and diffusion since the emergence of containerization, the global maritime container shipping network has become a reality (Frémont, 2007; Rodrigue and Notteboom, 2010). The technological revolution of containerization has gradually produced new forms of relationships among countries, regions, and port cities, backed by a continuous pressure on transport costs (Limao and Venables, 2001) and an increasing power of shipping alliances and large carriers (Sys, 2009; Slack and Frémont, 2009). Investigating such changes would complement the lack of evidence about the spatial patterns of commodity chains (Leslie and Reimer, 1999), because ports compete not as individual places that handle ships but as crucial links within global supply chains (Notteboom and Winkelmans, 2001; Hall and Jacobs, 2010).

While the main shipping routes and ports are well described in a number of studies, the structure and evolution of the global maritime network itself has not been fully documented. More extensive is the research on global airline networks due to their closer overlap with systems of cities (Guimera et al., 2005; Choi et al., 2006; Derudder and Witlox, 2009). Despite the local dereliction of port-city linkages in recent decades, maritime transport remains absolutely necessary for globalization. Its crucial weight in world trade volumes (90%) makes it a useful looking glass for analyzing the global economy and its geographic architecture. In parallel, the spatial design of maritime transport not only follows trade demand but also possesses its own practical arrangements and network configurations, which also evolve over time. The concentration and regional polarization of flows by load centers and intermediate hubs toward other secondary ports are typical examples of such configurations. It is thus important to evaluate the respective influence of technological factors (e.g. carriers and infrastructures, industry changes) and territorial factors (e.g. geographic and trade proximities, socio-economic developments) in the formation of shipping networks, port hierarchies, and maritime regions.

The remainder of this paper is organized as follows. Section 2 introduces the concept of port system and reviews the mechanisms shaping port competition, port selection, and port concentration, while describing the specificity and complexity of liner service networks. In Section 3, data on vessel movements (1996 and 2006) and the methodology for analyzing the global liner service network are presented, together with some results on the structure and geographic coverage of this network. Section 4 provides a closer look at the port hierarchy based on centrality measures and the geographic pattern of nodal maritime regions. The paper ends with a discussion of the research outcomes for further analysis of the global economy and its networks.

PORT SYSTEMS AND MARITIME NETWORKS

Port Choice and the Hierarchy in Port Systems

Traffic flows through ports are a physical outcome of route and port selection by the relevant actors in the chain. The most relevant service-related and cost factors explaining port selection by the main players of the transport chain (e.g. shippers, ocean carriers, and forwarders) are identified in the scientific literature on port choice.1 Port choice thus becomes a function of the overall network cost and performance. Notteboom (2009b) groups the factors together in the demand profile of the port, the supply profile of the port, and the market profile of the port. Typical port choice criteria include factors such as:

  • Physical and technical infrastructure including the nautical accessibility (e.g. draft);
  • Terminal infrastructure and equipment, the hinterland accessibility and intermodal offer;
  • Geographical location vis-à-vis the main shipping lanes and the hinterland;
  • Port efficiency expressed as port turnaround time, terminal productivity and cost efficiency;
  • Interconnectivity of the port (sailing frequency of deep-sea and feeder shipping services);
  • Reliability, capacity, frequency and costs of inland transport services;
  • Quality and costs of auxiliary services such as pilotage, towage, customs, etc.;
  • Efficiency and costs of port management and administration (e.g. port dues);
  • Availability, quality and costs of logistic value-added activities (e.g. warehousing) and port community systems;
  • Port security/safety and environmental profile;
  • Port reputation.

The aggregate outcome of port choice and supply chain decisions leads to a specific distribution of cargo flows in port systems. The search for regularities in the development of port hierarchies has mostly been done from a continental perspective considering ports as heads of land-based transport corridors willing to extend their hinterland coverage. Early works provided spatial models (Taaffe et al., 1963; Rimmer, 1967; Ogundana, 1970) suggesting a trend towards an increasing level of cargo concentration in port systems. The concepts of maritime range (Vigarié, 1964) and port system (Robinson, 1976) originally comprised a set of adjacent seaports in close proximity that were interdependent through land and sea freight flows. However, most scholars have continued focusing primarily on hinterlands, due to the development of intermodalism and logistic chains around ports (Van Klink 1998; Robinson, 2002) and the higher cost of land transport versus sea transport (Notteboom, 2004). The nature and performance of traffics is often explained by the situation of ports within land-based transport and urban systems (Ducruet et al., 2010c).

Although the development of peripheral ports (Hayuth, 1981) and offshore hubs has a maritime purpose for cargo distribution toward secondary ports (Slack and Wang, 2002; Notteboom, 2005), their emergence has been interpreted from the hinterland perspective of a port regionalization process leading to the formation of a regional load center network (Notteboom and Rodrigue, 2005). There remain important local deviations from general models of port system development due to path dependency and contingency (Notteboom, 2006a, 2009a).

The definition of port systems has often been limited to coastal morphology, geographic proximity, and administrative boundaries (Ducruet et al., 2009a, 2009b). Never have port systems been defined and delineated from the maritime perspective of inter-port linkages. This raises the question of whether physical factors and geographic proximity still play a role in the current spatial patterns of container shipping circulations. The concepts of maritime region and port region, which remain rather descriptive and vague in the literature (Ducruet, 2009), may benefit from the application of similar methods used by studies of other global networks (see Derudder and Taylor, 2005), allowing for the definition of coherent groups of ports as well as the identification of leader ports. A close look at the current organization of liner shipping networks is necessary before applying specific network analytical tools.

Design and Operation of Liner Service Networks

The development of liner shipping in the last 30 years has exceeded the growth of world trade volumes. The activity of this very dynamic branch of maritime transport is measured in Figure 1 based on annual container port throughputs. Besides continuous growth in throughput volumes, we also observe a parallel increase in the concentration in the global port system, notwithstanding slight decreases in recent years, notably after the 2008 financial crisis that directly affected traffic volumes and distributions. Despite those conjectural changes, liner shipping remains built on a series of specific network configurations.

Figure 1: World port throughput and concentration, 1970-2009


Source: own elaboration based on Containerisation International data

Container shipping features a complex combination of end-to-end services, line-bundling services, and pendulum services, which are connected to form extensive shipping networks. Port hierarchy in the container business is intrinsically linked to shipping lines’ design of these liner service networks in terms of service variables such as service frequency, vessel capacity, fleet mix, vessel speed, and the number and order of port calls (Fagerholt 2004; Notteboom 2006c). Liner service design is a function not only of carrier-specific operational factors (i.e. lower costs) but also of shippers’ needs (e.g. transit time) and willingness to pay for a better service.

In the last two decades, increased cargo availability has led carriers and strategic alliances among them to reshape their liner shipping networks through the introduction of new types of liner services on the main east-west trade lanes (see Figure 2). The largest ships operate on multi-port itineraries calling at a limited number of ports. The Europe–Far East trade provides a good example. Most mainline operators and alliances running services from the Far East to North Europe stick to line bundling itineraries with direct calls scheduled in each of the main markets. Notwithstanding diversity in calling patterns on the observed routes, carriers select up to five regional ports of call per loop. Shipping lines have significantly increased average vessel sizes deployed on the route from around 4500 TEU in 2000 to over 7500 TEU in early 2010. These scale increases in vessel size have put downward pressure on the average number of port calls per loop on the Far East–North Europe trade: 4.9 ports of call in 1989, 3.84 in 1998, 3.77 in October 2000, 3.68 in February 2006, and 3.35 in December 2009.

Maersk Line, MSC, and CMA-CGM are among the truly global liner operators with a strong presence in secondary routes. Their networks are based on traffic circulation through specific hubs. Productivity has been improved through the use of larger ships,2 new operational patterns, and cooperation between shipping lines. Container shipping lines have been very active in securing (semi)dedicated terminal capacity in the strategic locations within their liner service networks. Figure 3 gives an overview of the strategic ports in the worldwide liner network of Maersk Line. Shipping lines also rely on horizontal integration through operating agreements (e.g. vessel sharing agreements, slot chartering agreements, consortia and strategic alliances) and mergers and acquisitions. Alliance structures (cf. Grand Alliance, New World Alliance, and CYKH) provide its members easy access to more loops or services with relatively low-cost implications and allow them to share terminals.

Figure 2: Typical examples of liner services on trade routes in relation to Europe


Source: shipping lines’ websites

Figure 3: The main strategic ports in the liner service network of Maersk Line


Note: Relay/Interlining involves trade route based transhipment at key network ports between deep-sea vessel strings. The aim is to transfer containers between mainline services, thereby adding new service options.
Source: based on liner service data from Maersk Line

In the last few decades, extensive hub-feeder container systems and short-sea shipping networks came into existence to cope with increasing volumes and to connect to other port ranges (Rodrigue and Notteboom, 2010). The economics of transhipment and relay/interlining have resulted in the establishment of intermediate hubs with terminals owned, in whole or in part, by carriers or port operators. In some cases, intermediate hubs were developed within offshore location often on small islands with an implicit local cargo base (Rodrigue and Notteboom, 2010). The development of offshore hubs did not exclude transhipment activities at traditional gateway ports such as in the Western Mediterranean port system, where the distinction between hub ports and gateway ports has become blurred (Gouvernal et al., 2005). The position of pure transhipment hubs is generally more unstable than that of pure gateway ports: once traffic volumes for the gateway ports are sufficient, hubs are bypassed and might even become redundant (Wilmsmeier and Notteboom, 2010). The location of transhipment hubs remains important, because they lower the deviation distance to/from main trunk lines (Zohil and Prijon, 1999). There remains a subtle combination between centrality (proximity to origin/destination markets) and intermediacy (insertion in carrier networks) in nearly every port (Fleming and Hayuth, 1994).

METHODOLOGY AND LINER SHIPPING NETWORK CHARACTERISTICS

In their recent review of the scientific literature on maritime network analysis, Ducruet et al. (2010a) particularly stress the scarcity and fragmentation of empirical studies, which may be categorized among four main approaches:

  • Geographic coverage of carrier networks: regional or global distribution of the port networks for individual shipping companies based on service data (e.g. Coscon, Maersk) revealing their strategic choices (Rimmer and Comtois 2005; Frémont, 2007; Bergantino and Veenstra, 2007);

  • Network connectivity: characteristics of a given network based on its topology, with reference to spatial analysis and graph theory, such as the pioneer study of Joly (1999) showing the tripolar organisation of the global maritime system based on Reeds zones, and other works on a regional level (McCalla, 2004; Ducruet et al. 2010b);

  • Network efficiency: modeling of port selection processes and search for the optimal location, for instance, of a transhipment hub lowering overall shipping costs (Zeng and Yang, 2002; Song et al., 2005; Tai, 2005);

  • Complex networks: description of the network’ hierarchical structure on a global level comparing its properties with general models of small-world and scale-free networks (Deng et al., 2009; Hu and Zhu, 2009; Kaluza et al., 2010).

This paper wishes to further the interpretation of network structure, port hierarchy, and the dynamics influencing them. It gives paramount importance to the visualization of the network as a whole and of emerging regional patterns. This is based on a rarely used data source on daily vessel movements, which is more precise than service data and therefore more representative of the reality and complexity of liner shipping.

Data Overview

The methodology used for building the global liner network defines an inter-port connection by the circulation of vessels between the ports through a 365-day sequence of port calls. Thus, nodes (vertices) in the network are the ports, and links (edges) in the network are the connections realized by vessel movements (Table 1). The years 1996 and 2006 were chosen, because 1996 marked the emergence of post-panamax vessels (e.g. the Regina Maersk of 6140 TEU was introduced in 1996) and the start of strategic alliance formation among shipping lines; 2006 saw the introduction of the first 10,000+ TEU vessels in a period of rapid container growth mainly triggered by the China effect in the world economy. Data was obtained from Lloyd’s Marine Intelligence Unit (LMIU)3 that ensures most of the world fleet for all types of vessels. The obtained database covers approximately 92% and 98% of the world’s fleet of container vessels in 1996 and 2006, respectively. Interestingly, the capacity and size of the fleet as well as the number of vessel movements have grown faster than the number of ports and operators, while the average vessel capacity has grown from 1906 TEU to 2413 TEU. Such evidence confirms the observed limitations for ports accommodating ever-growing vessels and traffic, which remain in the hands of horizontally and vertically integrated companies.

Table 1: Overview of the database on vessel movements, 1996-2006

 

1996

2006

2006/1996

No. Ports

975

1,240

1.27

No. Vessel movements

176,439

390,740

2.21

No. Vessels

1,759

3,973

2.26

No. Operators

497

720

1.46

Total slot capacity (TEUs)

3,352,849

9,590,309

2.86

Share world fleet (% TEUs)

92.15

97.91

+7.75

Source: own elaboration based on LMIU data

The global network was modeled based on vessel characteristics, ports of call, and vessel movements. The first result is a global network composed of weighted and non-directed links between ports, which can be analyzed in two different ways. On the one hand, vessel circulations create a graph of direct linkages (GDL) based on the successive ports of calls (i.e. from port A to port B and from port B to port C). On the other hand, it can be argued that two ports are also connected if they belong to the same liner service or loop, although they are not adjacent calls; a graph of all linkages (GAL) thus adds indirect linkages (i.e. from port A to port C). In the GDL, Le Havre and Tokyo are never connected by a direct link, whereas, in the GAL, this connection might occur inside a pendulum or round-the-world service. The GAL is the overlap of all individual complete graphs created by the circulation of each vessel. These two dimensions of the same reality (GDL and GAL) may exhibit distinct features in terms of network structure and port hierarchy. In order to reveal the structural properties of the two graphs for each year of observation, we apply conventional measures derived from graph theory, which were originally applied to transport networks by Kansky (1963), Haggett and Chorley (1972), and Garrison and Marble (1974), and from complex systems theory, referring to the works of Barabasi and Albert (1999) and Watts and Strogatz (1998). This set of measures provides clear evidence about the nature of the network based on topological properties.

One limitation of the data is that it ignores how many full or empty containers were truly handled by ships and ports. In reality, some vessels may not be fully loaded, since their passage in a port does not always include stevedoring activities (e.g. a port visit in the framework of bunkering activities). However, with reference to Joly (1999), the linear correlation between vessel traffic and port throughput4  is very significant: about 88% and 87% of total variance is explained by the regression in 1996 and 2006, respectively. This verifies the good fit and quality of the LMIU data source with official port statistics for analyzing container ports and their position in liner shipping networks.

Network Structure

Table 2 highlights important differences between the GDL and GAL approaches and between the two years of observation. In terms of network size, the GDL has fewer links than the GAL, which includes numerous indirect connections among ports, thus making it about 5 times larger (edges) and 12 to 13 times longer  than the GDL for the same number of ports (vertices). In the GDL, the most central port in terms of maximum degree value connects about 18 to 19% of all ports; in the GAL, it connects 48 to 51%. Such differences in size have a strong influence on other network properties. Indeed, the GAL has about 5 to 6 times greater density, connectivity, and lattice degree compared to the GDL.

More robust measures proposed by physics complement such findings by revealing the polarized or scale-free structure of the GDL with power-law exponents higher than one (−1.35 in 1996 and −1.29 in 2006): few ports concentrate a large number of links (high degree centrality), while most ports have a limited number of links with other ports. Due to its higher density, the GAL is more likely to be a small-world network: higher average clustering coefficients (0.74 and 0.73), higher transitivity (0.40 and 0.43), lower power-law exponents (−0.62 and −0.65), and smaller diameters (4 and 5) than the GDL indicate the tendency for a given port to have its direct neighbors connected to each other, thus forming tightly connected communities. Thus, the GDL is more representative of hub ports dominating secondary ports, whereas the GAL represents densely connected maritime regions. Consequently, the GAL is more efficient than the GDL, because the inclusion of indirect links facilitates the circulation of flows in the graph, as reflected by its shorter average path length. Our results are similar to those of Hu and Zhu (2009) based on 2006 service data, both for the GDL (power-law exponent of −1.7, average clustering coefficient of 0.4) and the GAL (average clustering coefficient of 0.7).

Table 2: Topological properties of the global maritime network

Index

Measure

Graph of direct linkages (GDL)

Graph of all linkages (GAL)

1996

2006

1996

2006

Network size

No. vertices

910

1205

910

1205

No. edges

5,666

9,829

28,510

51,057

Max. degree

165

226

437

610

Avg. degree

12.787

17.027

64.178

87.521

Total length (000s km)

5,159

10,813

71,835

130,927

Traffic density (TEU/km)

331

407

125

183

Max. edge length (km)

10,012

10,018

10,018

10,018

Avg. edge length (km)

1,008

1,227

2,900

2,997

Diameter

9

8

4

5

Cycles

Cyclomatic number

4,757

8,625

27,601

49,853

Lattice degree

Alpha

0.005

0.012

0.033

0.069

Complexity

Beta

6.226

8.156

31.329

42.370

Connectivity

Gamma

0.014

0.014

0.069

0.070

Scale-free

Power-law coefficient

-1.351

-1.293

-0.624

-0.647

Small-world

Avg. clustering coefficient (local)

0.540

0.545

0.744

0.734

Transitivity (global)

0.266

0.266

0.404

0.435

Efficiency

Avg. path length

3.253

3.189

2.230

2.219

Source: realized by authors based on LMIU data

Despite their fundamental differences in size and structure, the two networks share similar evolutionary paths. Network structure has remained somewhat resilient to the aforementioned industry changes (and their spatial consequences), as seen with the stable connectivity (gamma index) and clustering coefficients. However, both networks have become more complex (cf. higher values for alpha and beta indices) due to the multiplication of nodes and edges, resulting in better efficiency as illustrated by the decreased average path length. One important trend that is only visible in the GDL is the decrease of the power-law coefficient, which seems to contradict the higher polarization of shipping networks for individual shipping companies as a result of service rationalization and a reduction of port calls per liner service.

This decrease may be interpreted as the combined influence of bottom-up and top-down retroactions. Bottom-up phenomena include congestion issues at the port-urban interface and regional integration processes. On a local level, large ports face important limitations in terms of lack and cost of available land for further expansion as well as congestion and bottleneck effects at terminals situated within dense urban environments. Port-city separation and the shift of modern terminals outside urban areas may be avoided in some cases through efficient planning policies (Lee et al., 2008). On a regional level, trade growth has multiplied the number of intra-regional shipping connections, thus making the network denser and more evenly distributed. This is particularly true in emerging economies where maritime transport plays a crucial role (e.g. China, India, Brazil, and Middle East).

Top-down retroactions are found at the level of the competition among shipping lines. A number of shipping lines seek differentiation and competitive advantage by fully or partially controlling (semi)dedicated terminal facilities. However, this spatial concentration at the company level does not necessarily result in higher cargo concentration at port system level since individual shipping lines often opt for different locations to set up their hub ports (Cullinane and Khana, 1999; Frémont and Soppé 2007). Traffic thus becomes relatively more balanced among several hubs rather than one mega-hub. Even when dedicated hubs are developed, shipping lines can still follow a risk-spreading strategy over different ports in view of offering more routing options to shippers. However, there are important variations in the position of individual routes and ports, as demonstrated in the next sections.

Geographic Coverage of the Network

The interplay between distance and flow intensity is depicted in Figure 4, where most traffic occurs across relatively short distances. This trend is more obvious in the GDL approach, where 78 to 79% of worldwide vessel traffic occurs over links of 500km or less. Links of 100km or less support more than half of worldwide traffic in both years. Strong traffic links are likely to occur among adjacent seaports serving shared hinterlands (e.g. Antwerp/Rotterdam in the Benelux area) or acting as dual hubs (e.g. Busan/Gwangyang in South Korea), which often receive multiple calls for the same vessels or liner services. The share of links shorter than 500km is much lower in the GAL, because it includes many long-distance and high-density indirect maritime links between world ports such as Le Havre-Tokyo and New York-Singapore. There is a noticeable increase in the traffic share of the 3000 to 5000km edges, caused mainly by the growing importance of trans-Pacific relations. Continued globalization and technological progress in the maritime industry are likely to be responsible for the increased share of the longest links (over 5000km) from 7% in 1996 to almost 10% in 2006. The increased share of shortest links (from 51% in 1996 to 55% in 2006) illustrate that long-distance links remain inferior to the number and weight of intra-regional linkages (i.e. short-sea shipping or hub-and-feeder services with a high sailing frequency).

Figure 4: Edge and traffic distribution over distance in 1996 and 2006


Source: own elaboration based on LMIU data

Table 3 illustrates the application of the gravity model for estimating traffic flows by links and by ports.5 At the link level, the gravity model is able to estimate a non-negligible proportion of observed flows, although a majority of them cannot be explained by simple kilometric distance and port size. This result indicates that maritime networks are not completely disconnected from spatial matters, despite their low transport cost and high geographic flexibility. Obviously, important distortions, ranging from the aforementioned port selection factors to wider issues of geopolitics and natural barriers, remain important. More interestingly, results differ according to the type of graph and to the variable used to measure port traffic. The GAL always provides better results, with 35 to 40% of actual flows being explained by the model, compared with 28 to 31% for the GDL. Port throughputs are less relevant than vessel traffic for explaining the flows between ports. Still, almost one-third of the variance is explained by port throughputs in 2006 for the GAL.

At the port level, we summed the estimated traffic of the links for every port, thus generating a third measure of port traffic (i.e. estimated weighted degree). Higher coefficients are observed in line with the aforementioned high correlation between weighted degree (vessel traffic) and port throughput. The results for port throughput are closer to the estimated traffics than the results for weighted degree (62% in 2006 in GDL and GAL), meaning that a large proportion of port activity may be estimated based on available port statistics and Euclidian distances between ports.

Table 3: Estimated maritime traffic flows based on gravity model

Unit

Type of observed traffic

Direct linkages (GDL)

All linkages (GAL)

1996

2006

1996

2006

Links

Port throughput

0.145

0.184

0.259

0.308

Weighted degree

0.285

0.309

0.397

0.354

Ports

Port throughput

0.533

0.620

0.570

0.615

Weighted degree

0.458

0.514

0.414

0.687

N.B. values represent determination coefficients (%) of the power-law lines

A look at the spatial distribution of the heaviest direct links provides interesting findings regarding the network’s geographic coverage (Figure 5). The top 100 links represent 52% and 39% of total worldwide vessel traffic in 1996 and 2006, respectively. They connect primarily neighboring ports and remain intra-regional rather than interregional.6 There is a clear dominance of three main poles: Asia, Europe, and North America. In each pole, a small number of ports constitute the backbone (i.e. East Asian corridor, North European range, and US East and West coasts). In 1996, only Buenos Aires, Santos, Jeddah, and Colombo stand out as main ports outside of these poles. The strongest inter-regional links run between Asia and the two other large poles, with Japan (i.e. Tokyo) and Singapore acting as turntables across the Pacific and the Indian Ocean, respectively. Most other inter-regional links generate less traffic, while some regions remain isolated (e.g. South Africa and Australia). The pattern in 2006 is similar, but there is an intensification of intra-regional links at the expense of inter-regional links. Busan has taken over as the key bridge between East Asia and North America, and Trans-Atlantic links has disappeared from the top 100 list. Such changes in the network structure and geographic coverage should also be analyzed from the perspective of port hierarchies.

Figure 5: Top hundred direct maritime links in 1996 and 2006


Source: own elaboration based on LMIU data and Philcarto software9

CHANGING PORT HIERARCHIES

Centrality of Ports in the Network

The centrality of ports in the network can be approached at the local and global levels. Degree centrality is a local level measure counting for each port the number of connections to other ports. Betweenness centrality is a global level measure summing for each port the number of its positions on the shortest possible paths within the entire network. Degree centrality is a measure of connectivity, while betweenness centrality can be regarded as a measure of accessibility. The hypothesis is that hub ports will have both a high degree centrality and a high betweenness centrality, due to their role as inter-regional pivots in the global network. As defined by Fleming and Hayuth (1994), hub ports are those that welcome mother vessels for redistributing cargoes to satellite (and often secondary) ports via feeder vessel services.

Figure 6 visualizes the GDL and the port hierarchy. At first sight, the geography of the network appears similar over time, with Asia-Pacific centered on the Singapore-Busan axis and Europe-Atlantic with the Le Havre-Hamburg range. Surprisingly, large North American and Japanese ports are poorly represented despite their traffic volume due to their lack of hub/feeder activities. Inherent to the data, gateway (hinterland) functions of seaports are not included in the analysis. Results indicate that Singapore is the most central port of the global system, which echoes its rank at the top of throughput hierarchy in official statistics.7 The very high centrality of the Suez and Panama canals underlines the strong vulnerability of the global network, but they are not taken into account in the following analyses. 

Figure 6: Visualization of the global liner shipping network in 1996 and 2006


Source: own elaboration based on LMIU data and TULIP software10

Figure 7 reveals noticeable changes between 1996 and 2006. The lowered centrality of Houston and Port Everglades to Kingston, Jamaica in the Caribbean is a good example of the impact of hub-and-spoke strategies. Several ports have strengthened their positions based on their gateway functions, such as Santos, Brazil and Shanghai, China. In East Asia and the Mediterranean, an increasing number of ports have high connectivity (e.g. Gwangyang, Port Klang, Xiamen, Shenzhen in Asia; Marsaxlokk, Gioia Tauro in the Mediterranean), but this growth has not altered the established position of established pivotal hubs (e.g. Singapore, Busan, Algeciras, Gioia Tauro) and gateway ports (e.g. Barcelona, Valencia in Spain). Conversely, the position of some formerly central ports has lowered significantly, as in the cases of Los Angeles, Houston, New York, Melbourne, Bilbao, North European range ports, Tokyo-Yokohama, Kaohsiung, and even Singapore. In contrast with recent literature (Yap, 2010), Hong Kong has maintained and even increased its position in the network. This trend is thus a good illustration of the globalization process with the shift of production from mature to emerging economies. Changes in betweenness centrality scores follow a similar geographic pattern with more drastic gaps among ports. We clearly see the strong effects of hub strategies at Kingston, Gioia Tauro, Dubai, and Busan as well as the emergence of large load centers in South Brazil and China. There is a clear North–South divide illustrating emerging economies and differentiating ports according to local and global changes in trade routes and port selection.

Figure 7: Changes in throughput and centrality, 1996-2006


Source: own elaboration based on LMIU data, TULIP and Philcarto softwares

Polarization and Nodal Regions

The method applied originally by Nystuen and Dacey (1961) to telephone flows among Washington State cities in the U.S. has been extensively applied to many transport networks (Cattan, 1995; Grubesic et al., 2008; Van Nuffel et al., 2010), but this is the first time that it has been applied to maritime transport. The method allows for delimiting so-called nodal regions by focusing on the strongest associations between city pairs, which are believed to reflect the hierarchy of central places in which subordinate nodes (satellites) are under the influence of independent (dominant) nodes through a transitive principle: an independent city also dominates the satellites of its satellites. Due to its higher average clustering coefficient (see Table 2), this algorithm is applied to the GAL to reduce the likelihood that geographic proximity is the main explanatory factor behind the delimitation of nodal regions.

A small number of nodal regions form the world maritime system and tend to merge or to split with each other over time (see Figure 8). Indeed, the global network is highly polarized by a few large entities concentrating 58% and 69% of all ports in 1996 and 2006, respectively. Singapore and Hong Kong’s combined nodal regions include 39% in 1996 and 50% in 2006 of all ports, while Hamburg and Rotterdam maintain their 19% share. This complements the sole indicators of centrality and better highlights the increasing influence of Asia on the world economy.

Hong Kong is the independent port of the largest region centred on Asia due to its role as a gateway for South China and centrally located hub in East Asia. Hong Kong’s ramifications8 remain focused on East Asia in 1996, except the link with Los Angeles, but they extend much further in 2006 with the inclusion of important Caribbean and Mediterranean ports. Despite their traffic size, other large Northeast Asian ports remain under Hong Kong’s influence due to double calls and hub dependence (Yap, 2010). Some of them have also extended their influence, such as Busan and Shanghai, while Taiwanese and Japanese ports have seen a significant reduction in their influence.

Comparatively, Singapore possessed a widely diversified tributary area in 1996 due to its pivotal role between Europe and Asia, as reflected by its ramifications covering Southeast Asia, South Asia, the Middle East, and large parts of the Mediterranean. This pattern did not change in 2006, notwithstanding the increase in the number of ports under its influence. In fact, a number of large subordinate ports such as Incheon, Surabaya, and Port Klang extended their own tributary areas in response to the overwhelming dominance of Singapore and Hong Kong. The extensions of the latter two ports toward the Mediterranean reflect the importance of the Europe–Asia trade link with a continuous and regular alignment of transhipment hubs during the last two decades. The port hierarchy does not always overlap traffic volumes due the dominant gateway function of some ports compared with their limited transhipment activities (e.g. Antwerp, Bremerhaven, Le Havre, Shenzhen, and Tokyo).

Figure 8: Nodal maritime regions of the world in 1996 and 2006


Source: own elaboration based on LMIU data and Philcarto software

Although European ports appear as subordinates of Asian ports in 1996, this is no longer the case in 2006, as the core region has split in two between Europe and Asia. This change can be interpreted in two ways. On the one hand, each region has reinforced its internal connectivity, making it a distinct entity stemming from regional integration forces. On the other hand, the so-called global shift (especially in the manufacturing sector) has placed Asia at the forefront of the global scene, thus relegating Europe and the rest of the world to the periphery.

A closer look at the geographic coverage of Rotterdam and Hamburg, the two main ports of the European nodal region, reveals their respective specialization. In 1996, Rotterdam  primarily covered the British Isles, Iceland, the Iberian Peninsula, and the Canary Islands, while Hamburg turned toward Scandinavia and the Baltic regions, with Antwerp as a large subordinate. Despite the stability in the number of subordinates and geographic coverage in 2006, Rotterdam extended to Africa and the Black Sea, while Hamburg reached across the Atlantic through Antwerp, Liverpool, and Le Havre.

Geographic proximity and regional integration seem to explain the delineation of most nodal regions, except for the giant Asian region, and despite the absence of a comparable transatlantic region. Physical geography that reinforces the internal connectivity of basins may also be responsible for the limited North–South linkages across Europe. Several secondary nodal regions remained in 2006, such as the one centered on New York, which integrated the region comprising Kingston and Rio Haina. Another important region expanded in 2006: the one from Santos primarily bound to Brazil ports but embracing Venezuela and some Caribbean ports. One may interpret such changes under the context of NAFTA and MERCOSUR arrangements due to growing North–South trade among the Americas.

Some nodal regions have become detached from large ones, such as the region of Lisbon, with strong links to the Azores, the region of Constantza in the Black Sea, and the region of Izmir and Ambarli in Turkey. Others have remained rather stable in their size and geographic distribution, such as the West Mediterranean range polarized by Barcelona and the West African range polarized by Abidjan, despite a drop in the number of its subordinates that were caught by Algeciras on one side and the Asian region on the other. The independent port has shifted in some regions: Melbourne and Buenaventura have replaced Sydney and Callao (i.e. Oceania and the Latin American West Coast). Conversely, some formerly detached regions have been integrated within a larger one, such as the East Mediterranean range (Piraeus) and the region of Cape Town (Southern Africa) shifting to the Asian region as well as the Belem and Puerto Cabello regions shifting under the influence of Santos (Brazil). Such phenomena depict the expansion of ocean carrier networks, making the world system increasingly interconnected.

CONCLUSION

This paper presents an analysis of the global liner shipping network in 1996 and 2006, a period of rapid change in port hierarchies and liner service configurations. While it refers to a wide literature on port system development, shipping networks, and port selection, it is one of the only analyses of the properties of the global container shipping network. The paper examines the network structure and the relative position of ports based on daily vessel movement data covering all of the world’s container fleets. The application of graph theory and complex network analysis provide a number of important and rather novel results about ports and liner shipping networks.

Although such networks are highly dynamic due to changes made by market players  in port and hub selection and the changing geography of container demand, we observe a certain level of robustness in the network structure. While transhipment hub flows and gateway flows might slightly shift among nodes, topological properties remain rather stable. The increasing size and complexity of the network occur in parallel with its decreasing spikiness caused by simultaneous bottom-up and top-down retroactions.

The analysis confirms the strong influence of geography and distance on the distribution of traffic, showing the dominance of intraregional links and demonstrating good applicability of the gravity model for estimating inter-port traffic. As in previous analyses of other global inter-city networks, maritime linkages retain an important regional dimension (Derudder and Taylor, 2005), but there is a striking absence of strong transatlantic linkages compared with the global pattern of airline networks (Cattan, 2004). The overarching importance of the Asia–Pacific area in the maritime network is best illustrated by delineating the ramifications of nodal regions. This was also made evident when mapping the changing centrality of ports. While the Old World (Atlantic, Northern hemisphere) versus the New World (Asia-Pacific, Southern hemisphere) would be a too simplistic interpretation of our results, the role of a changing world geography cannot be ignored. As such, analyzing the global liner shipping network provides a useful and necessary complement to the study of globalization and regionalization processes, which are often approached through other types of global networks.

Further research in this field may benefit from the inclusion of land-based networks (e.g. road, rail) as a means of considering hinterland accessibility. Currently, a worldwide database of vessel movements over the contemporary period (1946-2010) for all types of vessels is being built to expand the analysis of the global maritime network’s dynamics.

ACKNOWLEDGEMENTS

Authors are grateful to Liliane Lizzi and Igor Lugo at Géographie-Cités for their help on the cartography and data calculation respectively. This research benefited from the financial support of Marie Curie EIF-FP6 and the Reintegration Grant (ERG).

REFERENCES

Barabasi, A.L., Albert, R. (1999) 'Emergence of scaling in random networks', Science 286(5439), 509-512.

Barros, C.P., Athanasiou, M. (2004) 'Efficiency in European seaports with DEA: Evidence from Greece and Portugal', Maritime Economics and Logistics 6(2), 123-140.

Bergantino, A.S., Veenstra, A.W. (2002) 'Interconnection and co-ordination: An application of network theory to liner shipping', International Journal of Maritime Economics 4(3), 231-248.

Cattan, N. (1995) 'Barrier effects: The case of air and rail flows', International Political Science Review 16(3), 237-248.

Cattan, N. (2004) 'Le monde au prisme des réseaux aériens', Flux 58, 32-43.

Choi, J.H., Barnett, G.A., B.S. Chon (2006) 'Comparing world city networks: A network analysis of Internet backbone and air transport intercity linkages', Global Networks 6(1), 81-99.

Chou, C.C., Chu, C.W., Liang, G.S. (2003) 'Comparison of two models for port choice', Maritime Quarterly 12(3), 45-62.

Cullinane, K., Khanna, M. (1999) 'Economies of scale in large container ships', Journal of Transport Economics and Policy 33(2), 185.208.

Deng, W.B., Long, G., Wei, L., Xu, C. (2009) 'Worldwide marine transportation network: Efficiency and container throughput', Chinese Physics Letters 26(11), 118901.

Derudder, B., Taylor, P. (2005) 'The cliquishness of world cities', Global Networks 5(1), 71-91.

Ducruet, C. (2009) 'Port regions and globalization', in T.E. Notteboom, C. Ducruet and P.W. De Langen (eds), Ports in proximity: Competition and coordination among adjacent seaports, Aldershot: Ashgate, 41-54.

Ducruet, C., Koster, R.A., Van der Beek, D.J. (2010c) 'Commodity variety and seaport performance', Regional Studies (forthcoming), DOI: 10.1080/00343400903167904

Ducruet, C., Lee, S.W., Ng, K.Y.A. (2010b) 'Centrality and vulnerability in liner shipping networks: Revisiting the Northeast Asian port hierarchy', Maritime Policy and Management 37(1), 17-36.

Ducruet, C., Notteboom, T.E., De Langen, P. (2009a) 'Revisiting inter-port relationships under the New Economic Geography research framework', in T.E. Notteboom, C. Ducruet and P.W. De Langen (eds), Ports in proximity: Competition and coordination among adjacent seaports, Aldershot: Ashgate, 11-28.

Ducruet, C., Roussin, S., Jo, J.C. (2009b) 'Going West? Spatial polarization of the North Korean port system', Journal of Transport Geography 17(5), 357-368.

Ducruet, C., Rozenblat, C., Zaidi, F. (2010a) 'Ports in multi-level maritime networks: Evidence from the Atlantic (1996-2006) ', Journal of Transport Geography 18(4), 508-518.

Fagerholt, K. (2004) 'Designing optimal routes in a liner shipping problem', Maritime Policy and Management 31(4), 259-268.

Fleming, D.K., Hayuth, Y. (1994) 'Spatial characteristics of transportation hubs: Centrality and intermediacy', Journal of Transport Geography 2(1), 3-18.

Frémont, A. (2007) 'Global maritime networks: The case of Maersk', Journal of Transport Geography 15(6), 431-442.

Frémont, A., Soppé, M. (2007) 'Northern European range: Shipping line concentration and port hierarchy', in J.J. Wang, T.E. Notteboom, D. Olivier and B. Slack (eds), Ports, cities, and global supply chains, Aldershot: Ashgate, 105-120.

Garrison, W.L., Marble, D.F. (1974) 'Graph theoretic concepts', in M.E. Eliot Hurst (ed), Transportation geography: Comments and readings, New York: McGraw, 58-80.

Gouvernal, E., Debrie, J., Slack, B. (2005) 'Dynamics of change in the port system of the western Mediterranean', Maritime Policy and Management 32(2), 107-121.

Grubesic, T.H., Matisziw, T.C., Zook, M.A. (2008) 'Global airline networks and nodal regions', Geojournal 71(1), 53-66.

Guy, E., Urli, B. (2006) 'Port selection and multicriteria analysis: An application to the Montreal-New York alternative', Maritime Economics and Logistics 8, 169-186.

Haggett P., Chorley R. (1972) 'Network analysis in geography', London: Arnold.

Hall, P.V., Jacobs, W. (2010) 'Shifting proximities: The maritime ports sector in an era of global supply chains', Regional Studies (in press).

Hayuth, Y. (1981) 'Containerization and the load center concept', Economic Geography 57(2), 160-176.

Hayuth, Y., Fleming, D.F. (1994) 'Concepts of strategic commercial location: The case of container ports', Maritime Policy and Management 21, 187-193.

Hu, Y., Zhu, D. (2009) 'Empirical analysis of the worldwide maritime transportation network', Physica A 388(10), 2061-2071.

Joly, O. (1999) 'La structuration des réseaux de circulation maritime', Unpublished PhD Dissertation in Territorial Planning, Le Havre: Le Havre University.

Kaluza, P., Kölzsch, A., Gastner, M.T., Blasius, B. (2010) 'The complex network of global cargo ship movements', Journal of the Royal Society Interface 7(48), 1093-1103.

Kansky, K. (1963) 'The structure of transportation networks', Research Paper 84, Chicago: University of Chicago.

Lee, S.W., Song, D.W., Ducruet, C. (2008) 'A tale of Asia’s world ports: The spatial evolution in global hub port cities', Geoforum, 39(1), 372-395.

Lewis, M.W., Wigen, K. (1999) 'A maritime response to the crisis in area studies', The Geographical Review 89(2), 161-168.

Limao, N., Venables, A.J. (2001) 'Infrastructure, geographical disadvantage, transport costs, and trade', The World Bank Economic Review 15(3), 451-479.

Malchow, M., Kanafani, A. (2001) 'A disaggregate analysis of factors influencing port selection', Maritime Policy and Management 28, 265-277.

McCalla, R.J. (2004) 'Hierarchical network structure as seen in container shipping liner services in the Caribbean basin', Belgeo 4, 407-417.

Murphy, P., Daley, J., Dalenberg, D. (1992) 'Port selection criteria: An application of a transportation research framework', Logistics and Transportation Review 28, 237-255.

Murphy, P., Daley, J. (1994) 'A comparative analysis of port selection factors', Transportation Journal 3, 15-21.

Nir, A.S., Kuang, L., Gin-Shun, L. (2003) 'Port choice behaviour - from the perspective of the shipper', Maritime Policy and Management 30, 165-173.

Notteboom, T.E. (1997) 'Concentration and load centre development in the European container port system', Journal of Transport Geography 5(2), 99-115.

Notteboom, T.E. (2004) 'Container shipping and ports: An overview', Review of Network Economics 3(2), 86-106.

Notteboom, T.E. (2005) 'The peripheral port challenge in container port systems', in H. Leggate, J. McConville, and A. Morvillo (eds), International Maritime Transport: Perspectives, London: Routledge, 173-188.

Notteboom, T.E. (2006a) 'Traffic inequality in seaport systems revisited', Journal of Transport Geography 14(2), 95-108.

Notteboom, T.E. (2006b) 'Container throughput dynamics in the East Asian container port system', Journal of International Logistics and Trade 4(1), 31-52.

Notteboom, T.E. (2006c) 'The time factor in liner shipping services', Maritime Economics and Logistics 8(1), 19-39.

Notteboom, T.E. (2009a) 'Path dependency and contingency in the development of multi-gateway port regions and multi-hub port regions', in T.E. Notteboom, C. Ducruet and P.W. De Langen (eds), Ports in proximity: Competition and cooperation among adjacent seaports, Aldershot: Ashgate, 55-72.

Notteboom, T.E. (2009b) 'Complementarity and substitutability among adjacent gateway ports', Environment and Planning A 41(3), 743-762.

Notteboom, T.E., Rodrigue, J.P. (2005) 'Port regionalization: Towards a new phase in port development', Maritime Policy and Management 32(3), 297-313.

Notteboom, T.E., Winkelmans, W. (2001) ‘Structural changes in logistics: how will port authorities face the challenge?’, Maritime Policy and Management 28(1), 71–89.

Nystuen, J.D., Dacey, M.F. (1961) 'A graph theory interpretation of nodal regions', Papers in Regional Science 7(1), 29-42.

Ogundana, B. (1970) 'Patterns and problems of seaport evolution in Nigeria', in B.S. Hoyle and D. Hilling (eds), Seaports and development in Tropical Africa, London: Macmillan, 67-182.

Rimmer, P.J. (1967) 'The search for spatial regularities in the development of Australian seaports 1861-1961/2', Geografiska Annaler 49B, 42-54.

Rimmer, P.J., Comtois, C. (2005) 'China’s extra- and intra-Asian liner shipping connections', 1990-2000, Journal of International Logistics and Trade 3(1), 75-97.

Robinson, R. (1976) 'Modelling the port as an operational system: A perspective for research', Economic Geography 52, 71-86.

Robinson, R. (2002) 'Ports as elements in value-driven chain systems: The new paradigm', Maritime Policy and Management 29(3), 241-255.

Rodrigue, J.P., Notteboom, T. (2010) 'Foreland-based regionalization: Integrating intermediate hubs with port hinterlands', Research in Transportation Economics 27(1), 19-29.

Slack, B., Frémont A. (2009) 'Fifty years of organisational change in container shipping: Regional shifts and the role of family firms', Geojournal 74(1), 23-34.

Slack, B., Wang, J.J. (2002) 'The challenge of peripheral ports: An Asian perspective', Geojournal 65(2), 159-166.

Song, D.W., Yeo, K.T. (2004) 'A competitive analysis of Chinese container ports using the analytic hierarchy process', Maritime Economics and Logistics 6, 34-52.

Sys, C. (2009) 'Is the container liner shipping industry an oligopoly? ' Transport Policy 16(5), 259-270.

Taaffe, E.J., Morrill, R.L., Gould, P.R. (1963) 'Transport expansion in underdeveloped countries: A comparative analysis', Geographical Review 53, 503-529.

Tai, H. (2005) 'Analysis of hub port choice for container trunk lines in East Asia', Journal of the Eastern Asia Society for Transportation Studies 6, 907-919.

Tiwari, P., Itoh, H., Doi, M. (2003) 'Shippers’ port and carrier selection behavior in China: A discrete choice analysis', Maritime Economics and Logistics, 5, 23-39.

Van Klink, H.A. (1998) 'The port network as a new stage in port development: The case of Rotterdam', Environment and Planning A 30(1), 143-160.

Van Nuffel, N., Derudder, B., Witlox, F. (2010) 'Even important connections are not always meaningful: On the use of a polarization measure in a typology of European cities in air transport networks', Tijdschrift voor Economische en Sociale Geografie 101(3), 333-348.

Vigarié, A. (1964) Les grands ports de commerce de la Seine au Rhin, Paris: SABRI.

Vigarié, A. (1995) La mer et la géostratégie des nations, Paris: Economica.

Watts, D.J. (1999) Small worlds: The dynamics of networks between order and randomness, Princeton: Princeton University Press.

Wiegmans, B., Van Der Hoest, A., Notteboom, T. (2008) 'Port and terminal selection by deep-sea container operators', Maritime Policy and Management, 35(6), 517-534.

Wilmsmeier, G., Notteboom, T. (2010) 'Determinants of liner shipping network configuration: A two region comparison', Geojournal, in press, doi:10.1007/s10708-009-9333-2.

Yap, W.Y. (2010) Container shipping services and their impact on container port competitiveness, Antwerp: University Press Antwerp (UPA).

Zohil, J., Prijon, M. (1999) 'The MED rule: The interdependence of container throughput and transhipment volumes in the Mediterranean ports', Maritime Policy and Management 26(2), 175-193.

 


Appendix

Appendix 1: Top 25 most central ports in 1996 and 200611

Graph of direct links (GDL)

1996

2006

Port

BC

DC

Port

BC

DC

Singapore

150,240

165

Singapore

174,516

226

Rotterdam

97,875

140

Rotterdam

146,454

167

Hamburg

90,978

124

Hamburg

127,733

150

Hong Kong

61,839

126

Hong Kong

117,675

203

Antwerp

50,513

112

Busan

96,257

190

Busan

39,943

105

Shanghai

92,838

193

Le Havre

34,593

90

Bremerhaven

56,219

105

Houston

32,841

71

Antwerp

53,766

137

New York

32,536

70

Port Klang

52,191

148

Yokohama

31,090

83

Gioia Tauro

47,971

120

Los Angeles

30,726

66

Marsaxlokk

45,183

120

Felixstowe

27,606

88

Surabaya

39,030

50

Kaohsiung

27,551

82

Kingston(JAM)

37,495

104

Piraeus

24,827

71

Algeciras

36,846

130

Melbourne

22,516

44

Valencia

33,688

120

Philadelphia

21,867

44

Miami

32,963

83

Bremerhaven

21,661

56

Barcelona

32,462

118

Algeciras

20,373

72

Le Havre

31,623

98

Port Klang

19,782

58

Kaohsiung

31,419

125

Bilbao

19,549

60

New York

30,607

93

Valencia

17,380

78

Jebel Ali

28,785

97

Port Everglades

16,176

67

Felixstowe

28,216

92

Colombo

16,043

62

Durban

27,708

82

Izmir

14,854

55

Santos

26,306

92

Shanghai

14,719

59

Shenzhen

25,582

107

Graph of all links (GAL)

1996

2006

Port

BC

DC

Port

BC

DC

Hamburg

45,925

439

Rotterdam

83,246

610

Rotterdam

35,935

427

Shanghai

41,642

556

Antwerp

29,876

414

Antwerp

37,526

536

Busan

18,977

327

Busan

32,718

492

Bremerhaven

17,156

320

Bremerhaven

28,364

424

Yokohama

16,365

316

Port Klang

25,248

464

Felixstowe

15,944

366

Jakarta

22,876

325

Le Havre

13,190

334

Qingdao

19,312

436

Kaohsiung

12,551

291

Shenzhen

16,664

449

Aalborg

12,299

53

Ningbo

15,238

426

Kobe

11,448

277

Ambarli

14,526

347

Port Klang

9,959

264

Felixstowe

14,228

407

Shanghai

9,526

235

Leghorn

13,663

378

Kingston(JAM)

9,152

229

Barcelona

13,137

399

Los Angeles

8,543

241

Colombo

12,347

372

Houston

8,458

238

Durban

11,778

368

Izmir

8,225

270

Xiamen

11,442

383

New York

8,128

298

Bilbao

11,187

304

Fos

7,825

245

Le Havre

9,789

397

Aarhus

7,693

207

Genoa

9,547

371

Port Everglades

7,623

187

Osaka

9,437

330

Genoa

7,372

288

Wismar

9,249

117

Osaka

6,649

226

Kobe

8,998

345

Jeddah

6,465

254

Port Everglades

8,947

279

Bilbao

6,234

227

Leixoes

8,836

288

N.B. BC = betweenness centrality (no. of positions on possible shortest paths); DC = degree centrality (no. of ports connected)

Appendix 2: Graph visualisation of main nodal maritime regions12


The main nodal maritime region in 1996

 


The Asian nodal maritime region in 2006


The European nodal maritime region in 2006

N.B. figures are drawn using a GEM-Frick algorithm in TULIP software that positions most central nodes in the centre of the figure and least central nodes to its periphery


NOTES

* César Ducruet, Sorbonne University / UMR 8504 Géographie-Cités, CNRS / Paris (France), email: ducruet@parisgeo.cnrs.fr

** Theo Notteboom, ITMMA / University of Antwerp / Antwerp Maritime Academy, Antwerp (Belgium), email: theo.notteboom@ua.ac.be

1. See Murphy et al.(1992), Murphy and Daley (1994), Malchow and Kanafani (2001), Tiwari et al. (2003), Nir et al. (2003), Chou et al. (2003), Song and Yeo (2004), Barros and Athanassiou (2004), Guy and Urli (2006) and Wiegmans et al. (2008)

2. The average vessel size increased from 1,155 TEU in 1987 to 1,581 TEU ten years later,  2,417 TEU in 2007 and 2,618 TEU in 2009 (UNCTAD, 2009). In 2006, Maersk Line introduced the Emma Maersk of around 13,500 TEU capacity, the first vessel to move far beyond the 10,000 TEU mark. The total fleet in late 2009 counted 39 vessels in the range of 10,000-15,500 TEU and another 168 vessels of above 10,000 TEU unit capacity were on order (Source: Alphaliner, www.alphaliner.com)

3. http://www.seasearcher.com/lmiu/index.htm (Accessed October 2010)

4. Source: Containerisation International

5. The dis tance parameter of the gravity model is the orthodromic distance in kilometres (i.e. taking into account the sphericity of the Earth) between ports.

6. We have calculated at the level of 25 maritime LMIU regions the share of intra-regional traffic versus total traffic. It has increased from 53% to 55% in the GDL and from 27% to 31% in the GAL.

7. Appendix 1 provides detailed centrality scores for the top 25 ports.11

8. Appendix 2 provides a graph visualization of the largest regions where the ramifications of individual ports are more readable.12

9. http://philcarto.free.fr/ (Accessed October 2010)

10. http://tulip.labri.fr/TulipDrupal/ (Accessed October 2010)

 


Edited and posted on the web on 22nd October 2010