ARTICLE IN PRESS K O'Connor/Joumal of Transport Geography xxx (2009)xxx-xxx showed some of its data referred to earlier years;in addition the Table 1 2008 publication added more cities and broadened the data base Shares of sea and air freight at different types of urban locations 2006. considerably,adding more variables,so it was potentially more Location Number of Share of Share of useful than the 2007 publication.It was decided to use the 2008 places air freight sea freight data as it is based on more data and includes more cities.The city Global city logistics region 44 48.8 58.4 ranking is based upon 72 indicators which are merged into seven Global city with airport only 29 15.5 separate dimensions.These are the legal and political framework; Non global cities economic stability;ease of doing business;financial flows;busi- Top 20 in sea or air traffic 20 18.2 19.4 ness centre;knowledge creation and information flow and livabil- Rest of cities in data bases 17.5 222 ity.The business centre dimension contributes 12%to the ranking Total 100 100 It is made up of six variables,four of which in fact measure the Total includes 952 airports,530 seaports logistics activity that is the focus of the current research.These variables are Port TEUs,Air Passenger and Airphone traffic,Air Car- go traffic and International Air Passenger traffic (Mastercard Worldwide,2008,p.13).Hence to use this as measure in a study Rotterdam and Amsterdam,as well as a separate city region in Bel- of logistics activity it was necessary to remove that dimension gium(Brussels and Liege for air and Zeebrugge and Antwerp for from the index.This was done and an Adjusted Mastercard Index sea)are further illustrations of the methodology.Likewise,an ur- for 2008 was used to re-rank the 75 global cities in the data base. ban region labeled London and South East UK stretched the idea by including Flexistowe and Southampton with some smaller ports 2.2.Identifying global city regions and a set of airports serving that region;the availability of road and rail links that allow freight movement across this area justified that The research then needed to identify the global city region of decision.In the case of Shanghai and Ningbo,the combination fol- these cities.Here the approach drew on that used in case studies lowed the research presented by Cullinane et al.(2005)and Wang carried out by Simmons and Hack(2000)and the conceptual think- and Oliver(2007b)and incorporated knowledge of the new bridge ing of Webster and Muller(2002)applied to a study of Bangkok reducing the distance between the two cities.A Dubai-Gulf region (Webster,2004).These approaches suggested global city regions was shaped following the research of Zaid Ashai et al.(2007). were areas up to 70 km from a central city.with both sea and air- In some cases national borders separating near-neighbours port and multi-lane road systems (and rail networks in some (Hong Kong-Shenzhen and Singapore-Tanjung Pelapas for exam- cases).Ideally the identification could be based upon measures of ple)were ignored to provide a regional perspective on known (or the capacity and efficiency of the region's multi-modal transport potential)movement of goods and the spread of logistics manage- systems;that information was not available in a consistent form ment skills between these two places(Wang and Olivier,2007a: in all urban regions across the globe.Hence the approach followed Tongzon,2006;Loo et al.,2005).These did not extend to large geo- O'Connor's(2003)identification of multiple airport cities;it used graphic scale regions(which could link Guangzhou in the Hong local maps to find the location of transport infrastructure within Kong Shenzen case,and Penang in the Malaysian case)as the dis- the 70 km radius from the central city of the places identified in tances were beyond the 70 km limit. the Mastercard list discussed above.For inclusion in the study a The requirement of co-incident sea and airport in the method- global city region had to have at least one sea port and an airport ology meant that several global cities were left out of the analysis: in data supplied from sources identified below These included Paris,Frankfurt,Madrid and Zurich in Europe and Chicago,Dallas,Atlanta and Toronto in North America.Some sig- 2.3.Identifying logistics activity in global city regions nificant seaports and airports have also been excluded as they were not in locations classified as global cities:Kaioshung.Qingdao- The third step involved the measurement of logistics activity. Yantai and Busan-Gwangyang are examples of seaports,while The initial approach aimed to identify both physical and service Memphis,Louisville and Anchorage are examples of air freight dimensions of logistics,but it quickly became apparent that the locations.Given these exclusions it might be expected that the measurement of the service side of logistics(explored via directo- concentration on global cities with sea and airports might provide ries and lists of the head offices of companies)was not going to be a limited view of the logistics scene.The results show otherwise. sufficiently rigourous or comprehensive for the research.Hence the The full data base of cities,with their constituent seaports and air- focus fell back on the physical measures.The sources were Contain- ports is displayed in Appendix 1. erisation International,which provided the number of containers Table 1 provides an overview of the data base developed from moved through 530 ports in 2006;the same data was available this methodology,and a preliminary insight into the significance back to 1996 (though for a smaller number of ports)and Airport of the 44 global city regions that are the focus of the research.As Council International,which showed tonnes of air freight loaded shown in the table,this group accounted for 48.8%of global air at 952 airports in 2006.These data bases were scanned to find freight and 58.4%of global sea freight in 2006.Global cities in all sea and airports that could be assigned to the 70 km radius of the Mastercard ranking that did not have sea ports account for a each global logistics region.The condition that the global city re- further 15%of air freight.A group of the twenty busiest airports gion had to have an airport and a sea port for which data was avail- (led by Memphis,Louisville,Anchorage,and Luxembourg)and sea- able reduced the data set to 44 places which were labeled Global ports (such as Busan-Gwangyang,Kaohsiung.Quingdao and City Logistics Regions. Guangzhou)account for an additional one fifth of air and sea The data set includes some expected locations,like the ports of freight.The significant level of concentration of logistics activity Los Angeles-Long Beach along with airports in the Los Angeles ba- in these 44 global city logistics regions justifies further analysis. sin,while the incorporation of San Francisco (airport).Oakland (port and airport)and San Jose(airport)into one region is another 2.4.Measuring both sea and air freight obvious case.The New York-New Jersey network of airports and a seaport was extended to include Hartford following Bowen and The fourth step in the methodology involved a calculation to ex- Slack's(2007)observation on the latter's role.The location of ports press the scale of physical sea and air logistics activity.The aim and airports in the Tokyo-Yokohama region,the combination of was to find a way to compress the data to simplify the presentation Please cite this article in press as:O'Connor,K.Global city regions and the location of logistics activity.J.Transp.Geogr.(2009).doi:10.1016/ j.jtrangeo.2009.06.015showed some of its data referred to earlier years; in addition the 2008 publication added more cities and broadened the data base considerably, adding more variables, so it was potentially more useful than the 2007 publication. It was decided to use the 2008 data as it is based on more data and includes more cities. The city ranking is based upon 72 indicators which are merged into seven separate dimensions. These are the legal and political framework; economic stability; ease of doing business; financial flows; busi￾ness centre; knowledge creation and information flow and livabil￾ity. The business centre dimension contributes 12% to the ranking. It is made up of six variables, four of which in fact measure the logistics activity that is the focus of the current research. These variables are Port TEUs, Air Passenger and Airphone traffic, Air Car￾go traffic and International Air Passenger traffic (Mastercard Worldwide, 2008, p. 13). Hence to use this as measure in a study of logistics activity it was necessary to remove that dimension from the index. This was done and an Adjusted Mastercard Index for 2008 was used to re-rank the 75 global cities in the data base. 2.2. Identifying global city regions The research then needed to identify the global city region of these cities. Here the approach drew on that used in case studies carried out by Simmons and Hack (2000) and the conceptual think￾ing of Webster and Muller (2002) applied to a study of Bangkok (Webster, 2004). These approaches suggested global city regions were areas up to 70 km from a central city, with both sea and air￾port and multi-lane road systems (and rail networks in some cases). Ideally the identification could be based upon measures of the capacity and efficiency of the region’s multi-modal transport systems; that information was not available in a consistent form in all urban regions across the globe. Hence the approach followed O’Connor’s (2003) identification of multiple airport cities; it used local maps to find the location of transport infrastructure within the 70 km radius from the central city of the places identified in the Mastercard list discussed above. For inclusion in the study a global city region had to have at least one sea port and an airport in data supplied from sources identified below. 2.3. Identifying logistics activity in global city regions The third step involved the measurement of logistics activity. The initial approach aimed to identify both physical and service dimensions of logistics, but it quickly became apparent that the measurement of the service side of logistics (explored via directo￾ries and lists of the head offices of companies) was not going to be sufficiently rigourous or comprehensive for the research. Hence the focus fell back on the physical measures. The sources were Contain￾erisation International, which provided the number of containers moved through 530 ports in 2006; the same data was available back to 1996 (though for a smaller number of ports) and Airport Council International, which showed tonnes of air freight loaded at 952 airports in 2006. These data bases were scanned to find all sea and airports that could be assigned to the 70 km radius of each global logistics region. The condition that the global city re￾gion had to have an airport and a sea port for which data was avail￾able reduced the data set to 44 places which were labeled Global City Logistics Regions. The data set includes some expected locations, like the ports of Los Angeles–Long Beach along with airports in the Los Angeles ba￾sin, while the incorporation of San Francisco (airport), Oakland (port and airport) and San Jose (airport) into one region is another obvious case. The New York–New Jersey network of airports and a seaport was extended to include Hartford following Bowen and Slack’s (2007) observation on the latter’s role. The location of ports and airports in the Tokyo–Yokohama region, the combination of Rotterdam and Amsterdam, as well as a separate city region in Bel￾gium (Brussels and Liege for air and Zeebrugge and Antwerp for sea) are further illustrations of the methodology. Likewise, an ur￾ban region labeled London and South East UK stretched the idea by including Flexistowe and Southampton with some smaller ports and a set of airports serving that region; the availability of road and rail links that allow freight movement across this area justified that decision. In the case of Shanghai and Ningbo, the combination fol￾lowed the research presented by Cullinane et al. (2005) and Wang and Oliver (2007b) and incorporated knowledge of the new bridge reducing the distance between the two cities. A Dubai-Gulf region was shaped following the research of Zaid Ashai et al. (2007). In some cases national borders separating near-neighbours (Hong Kong-Shenzhen and Singapore–Tanjung Pelapas for exam￾ple) were ignored to provide a regional perspective on known (or potential) movement of goods and the spread of logistics manage￾ment skills between these two places (Wang and Olivier, 2007a; Tongzon, 2006; Loo et al., 2005). These did not extend to large geo￾graphic scale regions (which could link Guangzhou in the Hong Kong Shenzen case, and Penang in the Malaysian case) as the dis￾tances were beyond the 70 km limit. The requirement of co-incident sea and airport in the method￾ology meant that several global cities were left out of the analysis: These included Paris, Frankfurt, Madrid and Zurich in Europe and Chicago, Dallas, Atlanta and Toronto in North America. Some sig￾nificant seaports and airports have also been excluded as they were not in locations classified as global cities: Kaioshung, Qingdao– Yantai and Busan–Gwangyang are examples of seaports, while Memphis, Louisville and Anchorage are examples of air freight locations. Given these exclusions it might be expected that the concentration on global cities with sea and airports might provide a limited view of the logistics scene. The results show otherwise. The full data base of cities, with their constituent seaports and air￾ports is displayed in Appendix 1. Table 1 provides an overview of the data base developed from this methodology, and a preliminary insight into the significance of the 44 global city regions that are the focus of the research. As shown in the table, this group accounted for 48.8% of global air freight and 58.4% of global sea freight in 2006. Global cities in the Mastercard ranking that did not have sea ports account for a further 15% of air freight. A group of the twenty busiest airports (led by Memphis, Louisville, Anchorage, and Luxembourg) and sea￾ports (such as Busan–Gwangyang, Kaohsiung, Quingdao and Guangzhou) account for an additional one fifth of air and sea freight. The significant level of concentration of logistics activity in these 44 global city logistics regions justifies further analysis. 2.4. Measuring both sea and air freight The fourth step in the methodology involved a calculation to ex￾press the scale of physical sea and air logistics activity. The aim was to find a way to compress the data to simplify the presentation Table 1 Shares of sea and air freight at different types of urban locations 2006. Location Number of places Share of air freight Share of sea freight Global city logistics region 44 48.8 58.4 Global city with airport only 29 15.5 Non global cities Top 20 in sea or air traffic 20 18.2 19.4 Rest of cities in data bases 17.5 22.2 Total 100 100 Total includes 952 airports, 530 seaports K. O’Connor / Journal of Transport Geography xxx (2009) xxx–xxx 3 ARTICLE IN PRESS Please cite this article in press as: O’Connor, K. Global city regions and the location of logistics activity. J. Transp. Geogr. (2009), doi:10.1016/ j.jtrangeo.2009.06.015