Port Economics

Citation preview

PORT ECONOMICS

i

RESEARCH IN TRANSPORTATION ECONOMICS Series Editor: Martin Dresner Volumes 1–6: Research in Transportation Economics – B. Starr McMullen Volume 7:

Railroad Bankruptcies and Mergers from Chicago West 1975–2001: Financial Analysis and Regulatory Critique – Michael Conant

Volume 8:

Economic Impacts of Intelligent Transportation Systems: Innovations and Case Studies – Evangelos Bekiaris and Yuko Nakanishi

Volume 9:

Road Pricing: Theory and Evidence – Georgina Santos

Volume 10:

Transportation Labor Issues and Regulatory Reform – James Peoples and Wayne K. Talley

Volume 11:

Interurban Road Charging for Trucks in Europe – Jose´ Manuel Viegas

Volume 12:

Shipping Economics – Kevin Cullinane

Volume 13:

Global Competition in Transportation Markets: Analysis and Policy Making – Katsuhiko Kuroda and Adib Kanafani

Volume 14:

Measuring the Marginal Social Cost of Transport – Christopher Nash and Bryan Matthews

Volume 15:

Procurement and Financing of Motorways in Europe – Giorgio Ragazzi and Werner Rothengatter ii

RESEARCH IN TRANSPORTATION ECONOMICS

VOLUME 16

PORT ECONOMICS

EDITED BY

KEVIN CULLINANE University of Newcastle, Newcastle upon Tyne, UK

WAYNE K. TALLEY Old Dominion University,Virginia, USA

Amsterdam – Boston – Heidelberg – London – New York – Oxford Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo iii

JAI Press is an imprint of Elsevier The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands 525 B Street, Suite 1900, San Diego, CA 92101-4495, USA First edition 2006 Copyright r 2006 Elsevier Ltd. All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email: [email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN-13: 978-0-7623-1198-9 ISBN-10: 0-7623-1198-3 ISSN: 0739-8859 (Series) For information on all JAI Press publications visit our website at books.elsevier.com Printed and bound in The Netherlands 06 07 08 09 10 10 9 8 7 6 5 4 3 2 1

iv

CONTENTS LIST OF CONTRIBUTORS

vii

INTRODUCTION Kevin Cullinane and Wayne K. Talley

1

THE EVOLUTION AND CHALLENGES OF PORT ECONOMICS Trevor Heaver

11

AN ECONOMIC THEORY OF THE PORT Wayne K. Talley

43

MULTIPLE OUTPUTS IN PORT COST FUNCTIONS Sergio R. Jara-Dı´az, Eduardo Martı´nez-Budrı´a and Juan Jose´ Dı´az-Herna´ndez ESTIMATING THE RELATIVE EFFICIENCY OF EUROPEAN CONTAINER PORTS: A STOCHASTIC FRONTIER ANALYSIS Kevin Cullinane and Dong-Wook Song THE IMPACT OF PORT CHARACTERISTICS ON INTERNATIONAL MARITIME TRANSPORT COSTS Gordon Wilmsmeier, Jan Hoffmann and Ricardo J. Sanchez

v

67

85

117

vi

CONTENTS

STRATEGIC POSITIONING ANALYSIS FOR SEAPORTS Elvira Haezendonck, Alain Verbeke and Chris Coeck

141

PORT INVESTMENT: PROFITABILITY, ECONOMIC IMPACT AND FINANCING Enrico Musso, Claudio Ferrari and Marco Benacchio

171

SHIPPING DEREGULATION’S WAGE EFFECT ON LOW AND HIGH WAGE DOCKWORKERS James Peoples, Wayne K. Talley and Pithoon Thanabordeekij

219

LIST OF CONTRIBUTORS Marco Benacchio

Autorita Garante della Concorrenza e del Mercato, Rome, Italy

Chris Coeck

Antwerp Port Authority, Antwerp, Belgium

Kevin Cullinane

University of Newcastle, Newcastle upon Tyne, UK

Juan Jose´ Dı´az-Herna´ndez

Universida`d de La Laguna, La Laguna, Spain

Claudio Ferrari

Universita` di Genova, Genova, Italy

Elvira Haezendonck

University of Antwerp, Antwerp, Belgium

Trevor Heaver

University of British Columbia, Vancouver, Canada

Jan Hoffmann

UNCTAD, Geneva, Switzerland

Sergio R. Jara-Dı´az

Universidad de Chile, Santiago, Chile

Eduardo Martı´nezBudrı´a

Universidad de La Laguna, La Laguna, Spain

Enrico Musso

Universita` di Genova, Genova, Italy

James Peoples

University of Wisconsin-Milwaukee, Milwaukee, WI, USA

Ricardo J. Sanchez

Austral University, Buenos Aires, Argentina

Dong-Wook Song

The University of Hong Kong, Hong Kong

Wayne K. Talley

Old Dominion University, Norfolk, VA, USA vii

viii

LIST OF CONTRIBUTORS

Pithoon Thanabordeekij

University of Wisconsin-Milwaukee, Milwaukee, WI, USA

Alain Verbeke

University of Calgary, Calgary, Canada

Gordon Wilmsmeier

University of Osnabru¨ck, Osnabru¨ck, Germany

INTRODUCTION Kevin Cullinane and Wayne K. Talley A port (or seaport) is a place that provides for the vessel transfer of cargo and passengers to and from waterways and shores. A port is a node in a transportation network – a spatial system of nodes and links over which the movement of cargo and passengers occurs. A port is also an economic unit that provides a (transfer) service as opposed to producing a physical product. The amount of this transfer service is referred to as the port’s throughput. In a competitive environment, ports not only compete on the basis of location and operational efficiency, but also on the basis of the fact that they are embedded in the supply chains of shippers (Robinson, 2002). Port economics is concerned with the study of the economics of port services. Users of port services are those that utilize the port as part of the transportation process of moving cargo and passengers to and from origin and destination locations. Users include transportation carriers such as shipping lines, railroads and trucking firms that perform these movements and shippers and individuals that provide the cargo and themselves as passengers to be transported. Port users demand port services, whereas port service providers supply port services to port users. The primary port-service provider is the port’s terminal operator that operates the port or one (or more) of its marine terminals. Other port-service providers include, for example, stevedores, ship agents, customs brokers, freight forwarders, ship pilots and towage, dredging and government customs-service providers. Port economics and shipping economics comprise the branch of economics known as maritime economics.

Port Economics Research in Transportation Economics, Volume 16, 1–10 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16001-1

1

2

KEVIN CULLINANE AND WAYNE K. TALLEY

This volume provides original contributions to the study of port economics. They consist of (1) the evolution of port economics; (2) economic theories of the port, port cost functions and port investment; and (3) empirical evidence on the relative efficiency of ports, the impact of ports on international maritime transport costs, the competitiveness of ports and the impact of deregulation on dockworker wages. The chapter by Trevor Heaver presents an historical overview of the emergence of port economics. It begins by discussing the early application of economics to transport, followed by the reasons for the slow application of economics to maritime transport. At the end of World II, books on the maritime industry remained descriptive – without an analytical economic framework. The book by Svendsen (1958) was the first to provide an economic treatment of shipping; a more detailed treatment followed in Thorburn (1960). However, the literature on port economics progressed slowly. Heaver observes that the port economics literature after 1973 is much greater in volume than the earlier literature. However, only one book has appeared with the title Port Economics (Jansson & Shneerson, 1982). Port economists have investigated the relationship between ports and ship costs, port costs and pricing, the industrial organization of ports (including public administration and strategic issues), port competition, port performance and such specific issues as port labor, economic rents and harbor tug services. Heaver also observes that the evolution of port economics has lagged that of transport and shipping economics. However, more sophisticated economic methods are now found in the port economics literature and the investigated topics reflect the port issues of the day more so than in the past. Furthermore, the port economist has greater opportunities to investigate port issues given the greater availability of data. Chapter 3 by Wayne Talley is an exposition of an economic theory of the port. A model is presented which disaggregates the demand for port services between bulk and container throughput by assuming that a port handles only both these two types of cargo and has the economic objective of maximizing annual throughput (of bulk and container cargoes) subject to a minimum profit constraint. The annual demand for the port’s bulk and container throughputs are functions of the generalized cost (money and time) associated with their handling in port. Thus, port charges and time costs incurred by ocean carriers, inland carriers and shippers are accounted for. Production functions for the provision of bulk and container throughputs are specified as well as the resource functions representing the relationship between the minimum amount of a given resource employed by a

Introduction

3

port and the levels of its operating options and amounts of cargo received. The choice variables in the optimization of the port’s economic objective are its port prices and operating options, with the latter representing the means by which a port can differentiate its service. The model caters for a situation where the port is owned and subsidized by government and, therefore, this profit constraint may be zero (where port revenue equals cost) or a maximum deficit (where port revenue is less than cost). The focus of the model lies with the generalized costs (port charges and time spent) incurred by ocean carriers, inland carriers and shippers. Potential uses of the model include investigations into the causes and/or effects of the time spent in port by ships, vehicles and cargoes, port resource utilization and port costs. Application of the model will help a port to differentiate its service (or operating options) by providing information on ship and vehicle loading/unloading service rates, channel and berth accessibility and reliability, entrance and departure gate reliability and damage and property losses to ships, vehicles and cargoes in port. Talley goes on to present further potential extensions of the model to incorporate port ship and vehicle congestion, port cost efficiency and port performance evaluation. Chapter 4 by Sergio Jara-Dı´ az, Eduardo Martı´ nez-Budrı´ a and Juan Dı´ azHerna´ndez presents a multi-output cost theory for a port. The authors show theoretically and empirically that the use of an aggregate output in a port’s cost function when distinct outputs exist causes erroneous conclusions with respect to the port’s marginal costs and the extent of economies of scale. The presence of economies of scope is captured as economies of scale when an aggregate output, rather than distinct outputs, is used in the port’s cost function, inducing biased estimates of the extent of economies of scale exhibited by the port. Jara-Dı´ az, Martı´ nez-Budrı´ a and Dı´ az-Herna´ndez also show that marginal cost differences are expected among various types of port cargo. The latter is supported by the estimation of a multi-output cost function for Spanish ports, where total annual costs for the ports consist of infrastructure, administration, labor, amortization and other costs. The cost function’s four distinct outputs are containerized general, non-containerized general, dry bulk and liquid bulk cargoes. An investigation of the relative efficiency of European container terminals is presented by Kevin Cullinane and Dong-Wook Song in Chapter 5. The 74 European container terminals included in the sample handle a significant proportion of Europe’s container volume. The relative production efficiency of these terminals is investigated using the analytical tool, stochastic frontier

4

KEVIN CULLINANE AND WAYNE K. TALLEY

analysis – under the assumption that the functional form of the production frontier is the log-linear Cobb–Douglas function. Terminal output is measured by annual throughput in TEUs. The three input variables include: (1) terminal quay length in meters, (2) terminal area in hectares and (3) the number of pieces of cargo handling equipment employed. For the three assumed distributions for which the model’s parameters were estimated, the average production efficiency level of the Italian port, Giao Tauro, is consistently high. Although with less consistency, the efficiency levels of the port of Felixstowe (United Kingdom), the port of La Spezia (Italy), the port of Tarragona (Spain), the Valencia Container Terminal (Spain) and the North Sea Terminal of the port of Bremen/ Bremerhaven (Germany) are also high. The container ports of Haydarpasa (Turkey), Gdansk (Poland), Flushing (Netherlands), Palermo (Italy) and the Umex Container Terminal (Rumania) are consistently the most inefficient ports in the sample. A primary finding by Cullinane and Song is that larger container ports or terminals tend to be more efficient than their smaller counterparts. With respect to the regions in the sample, the ports and terminals of the British Isles emerge as having by far the most efficient infrastructure usage – which is consistent with the chronic shortage of port container-handling capacity in this region. By contrast, Scandinavia and East Europe emerge as the regions with container ports and terminals yielding the lowest level of relative efficiency. Geographical location (being displaced from the mainline intercontinental container trades) and below average size are possible explanations for this result. The determinants of international maritime transport costs are analyzed in Chapter 6 by Gordon Wilmsmeier, Jan Hoffmann and Ricardo Sanchez. Since the vast majority of international trade is carried in ships, the motivation for their analysis lies with a desire to identify those cost elements in maritime transport where reductions can be achieved most easily and with greatest effect. Achieving such a reduction in the transaction costs of international trade has clear beneficial effects in promoting trade. The authors posit that most of the determinants of international maritime transport are beyond the control or influence of a nation’s policy makers, but that the most obvious exceptions to this relate to port characteristics. They argue that aspects such as port infrastructure and private sector participation in ports can indeed be influenced by governments. To test their hypothesis, regression analysis is applied to data on the containerizable trade between seven importing and sixteen exporting Latin American countries in 2002. The database for analysis comprises 75,928

Introduction

5

observations, relating to practically all maritime trade transactions on 105 intra-Latin American trade routes. As evidenced in the review by Tovar, Jara-Dı´ az, and Trujillo (2003), the regression models tested include the standard variables for explaining variation in international maritime transport costs, such as unit cargo value, volume per transaction, geographical distance, bilateral trade volume and trade balances. However, supplemental input variables relating to different port characteristics are also incorporated into the analysis as possible determinants of international maritime transport costs. The analysis finds clear evidence that port efficiency, port infrastructure, the degree of private sector participation in ports and inter-port connectivity all have a statistically significant and strong impact on international maritime transport costs. Indeed, the estimated elasticity for port efficiency is found to be the highest of all port-related variables; a finding that prompts the authors to rather graphically point out that a doubling of efficiency in the ports at each end of a trade has the same impact on international maritime transport costs as halving the distance between them. With the exception of customs delays, port improvements appear to have a stronger impact on the maritime freight charged on a country’s exports than on imports. The authors’ estimated models explain about 40–50% of the variation in maritime transport costs; a figure that can be significantly improved by segmenting the regressions by individual commodity groups. Yet further improvement could be secured by taking into account the specific time of the transaction within the year. The final conclusion is that improvements in ports are the most effective mechanism by which cost savings and increased trade competitiveness can be achieved and that it is important to recognize that this will have a beneficial impact beyond the direct effect on international maritime transport costs; on inter alia the cost of using other modes, the price of traded goods, a port’s scale of operation and its consequent unit cost. Port competitiveness is an issue that is currently at the forefront of economic enquiry (Hayuth, 1993; Heaver, 1995; Heaver, Meersman, & Van de Voorde, 2001; Song, 2003). There are a number of reasons why this is the case, not least, the increasing concentration of the container shipping industry (Heaver, Meersman, & Van de Voorde, 2000; Midoro, Musso, & Parola, 2005; Notteboom, 2004), the globalization and concentration of the container handling sector (Juhel, 2001; Notteboom, 2002), the role of ports in supply chains and the overlapping of port hinterlands (Heaver, 2002; Notteboom & Winkelmans, 2001; Robinson, 2002; Carbone & De Martino, 2003; Slack, 2003) and the deployment of ever-larger containerships

6

KEVIN CULLINANE AND WAYNE K. TALLEY

(Heaver, 1968; Heaver & Studer, 1972; Gilman, 1999; Cullinane & Khanna, 1999, 2000). Despite this, there exists no wholly acceptable or satisfactory metric for assessing port competitiveness. In Chapter 7, Elvira Haezendonck, Alain Verbeke and Chris Coeck propose the use of strategic positioning analysis (SPA) as an appropriate approach for doing just that. The advocacy of this approach is predicated on the need for port authorities, terminal operators and port users to have a conceptual understanding of the dynamics of international port competition and to perform strategic positioning analyses when engaging in strategic decision making. SPA can provide a quantitative evaluation of the competitive position of individual ports in terms of market share, market growth rates and diversification. The tool also allows for the weighting of individual categories of cargo traffic by the value-addition they create. Utilizing cross-sectional data, SPA produces an objective, integrated evaluation of the competitive position of individual ports, but with panel data the evolution of this position over time is also possible. Furthermore, it can be applied at the level of individual cargo categories as well as at the aggregate level of the port. The SPA tool is fully described and its use is illustrated with a comprehensive application to the nine most important seaports in the Hamburg–Le Havre range. The authors conclude by recognizing the limitations of the SPA approach, especially in terms of the sometimes contradictory outcomes and the inconsistency of the outcomes over different time periods. It is pointed out that, to make best use of SPA, it should be complemented by financial, managerial and other information yielded from, for example, social costbenefit analysis, environmental impact assessment and economic effect analysis. The primary purpose of this tool lies with strategy formulation and however problematic the findings might be, they do provide further insights into port competition and may still succeed in stimulating internal debate and facilitating the assessment of alternative scenarios or strategies for port development. Given its topicality as the focus of ongoing policy debates around the world – sometimes even reaching the pages of the popular press – investment in ports is an issue that is pivotal to modern port economics. The financing of infrastructure provides the basis upon which the potential costs and benefits that arise from that infrastructure are defined and assessed. It is clearly, therefore, a central consideration in the establishment of policies relating to the planning and development of port infrastructure. Chapter 8 on ‘Port Investment’ by Enrico Musso, Claudio Ferrari and Marco Benacchio begins by defining and evaluating the generic paradigms

Introduction

7

that characterize worldwide investments in the port industry. The different perspectives on an investment of the various potential stakeholders are graphically illustrated through a number of examples of the sorts of synergies and/or conflicts of motivation and of interest that may arise in practice. Most notable is the conflict that exists between the pursuit of private profit or maximum returns on an investment at one extreme and investments that are motivated by more socially oriented objectives, such as employment gains or local/regional economic welfare benefits. As the authors point out, this apparent dichotomy in perspectives is reflected in the literature on infrastructure investment. Historical analyses have typically revolved around either (1) a purely macroeconomic perspective, where investment in transport infrastructure is premised on the assumption that this is a necessary, but not sufficient, prerequisite for economic growth (Banister & Berechman, 2000) or (2) the microeconomic perspective of the firm as a decision-making unit that has as its objective the maximizing of some form of private benefit such as profits, return on investment, corporate growth, etc. Clearly, applying each of these different perspectives to the evaluation of the impact of port investment can lead to distinctly different judgments as to the costs and benefits involved and ultimately, therefore, to the overall desirability of making the investment. Rather than adopting either one of these individual perspectives, the authors develop a comprehensive theoretical model that provides an allencompassing and generally applicable framework for the analysis of port investment scenarios. The model framework has been specified on the basis of a detailed description and critical evaluation of the trade-offs, which define the public/private, management/economic and local/global perspectives. Particular aspects of a specific investment, and the value of variables related to it, can be treated as either inputs to, or initial conditions, for the model. As such, the specific detail of an investment under consideration can fit into a general scheme of interrelationships (the model) in order to identify the range of feasible outcomes, in the form of the output from the model. These outputs can inform all aspects of the specific investment decision and can contribute to the resolution of conflicts and to the creation of a synergy of interests. The authors suggest that they can also have an influence on policy; for example, by providing an insight into the most appropriate sources of finance, optimum pricing and taxation schemes, the efficacy of possible efficiency incentives and the implications for inter- and intra-port competition.

8

KEVIN CULLINANE AND WAYNE K. TALLEY

In Chapter 9, James Peoples, Wayne Talley and Pithoon Thanabordeekij employ a quantile regression model to examine the regional wage patterns of low- and high-wage dockworkers in the United States following deregulation. The initial findings reveal that dockworkers historically received significant wage premiums over workers in other industries, and that this wage advantage increased following the initial deregulation in 1984. For ports on the East and West coasts, there were significant wage premium increases for low-wage dockworkers following the 1984 deregulation. These premium gains not only surpass post-deregulation gains of high-wage dockworkers, but also were not matched by wage settlements for low-skilled union workers in other industries. Indeed, the mean wage advantage for low-wage dockworkers over low-wage workers in other industries is found to have increased from 6.77 to 128% following deregulation in 1984. In contrast, the mean wage advantage for high-wage dockworkers over high-wage workers in other industries is found to increase from 20.71 to only 52.53%. The raison d’eˆtre for this is attributed to the greater demand for East and West coast dockworker labor following deregulation and, in consequence, dockworker unions were better able to leverage on their enhanced negotiating position in order to attain lucrative wage settlements for relatively low-skill occupations. The pre-deregulation premium for dockworkers was found to vary by region and it was largest for workers residing in the South. A second major finding, however, is that deregulation is associated with larger increases in the wage premium of dockworkers residing in the Northeast and West. Following the initial deregulation in 1984, the authors find that dockworker premiums do not change significantly for individuals residing in the South; a finding that can be explained by evidence from collective bargaining settlements that suggests that the negotiating power of dockworker unions is undermined in the South, in part because of the competition from nonunion low-cost dockworker labor. In contrast to the post-1984 deregulation wage patterns, following passage of the Ocean Shipping Reform Act of 1998, high-wage dockworkers were found to have received larger premium gains compared to the gains for low-wage dockworkers. Findings comparing post-1998 deregulation wages to pre-1984 deregulation wages reveal that, for all regions, high-wage dockworkers received larger premium increases compared to the premium gains for low-wage dockworkers. Indeed, with this second phase of deregulation, the wage advantage of low-wage dockworkers falls to 18.08%, but remains above the pre-1984 deregulation level. In contrast, the wage advantage for high-wage dockworkers increases to 73.46%. This contrasting wage pattern

9

Introduction

for low- and high-wage dockworkers following the 1998 Act is consistent with the authors’ hypothesis that employers responded to the shortage of skilled equipment workers by agreeing to labor negotiations that increased wages and reclassified a day’s work.

REFERENCES Banister, D., & Berechman, J. (2000). Transport investment and economic development. London: UCL Press. Carbone, V., & De Martino, M. (2003). The changing role of ports in supply chain management: An empirical analysis. Maritime Policy and Management, 30(4), 305–320. Cullinane, K. P. B., & Khanna, M. (1999). Economies of scale in large container ships. Journal of Transport Economics and Policy, 33(2), 185–208. Cullinane, K. P. B., & Khanna, M. (2000). Economies of scale in large containerships: Optimal size and geographical implications. Journal of Transport Geography, 8, 181–195. Gilman, S. (1999). The size economies and network efficiency of large containerships. International Journal of Maritime Economics, 1(1), 39–59. Hayuth, Y. (1993). Port competition and regional cooperation. In: G. Blauwens, G. De Brabander & E. Van de Voorde (Eds), De dynamiek van een haven (pp. 210–226). De Nederlandsche Boekhandel: Uitgeverij Pelckmans. Heaver, T. D. (1968). The economics of vessel size: A study of shipping costs and their implication for port investment. Ottawa: National Harbours Board. Heaver, T. D. (1995). The implications of increased competition for port policy and management. Maritime Policy and Management, 22(2), 125–133. Heaver, T. D. (2002). Supply chain and logistics management: Implications for liner shipping. In: C. Th. Grammenos (Ed.), The handbook of maritime economics and business (pp. 375–396). London: Informa UK Ltd. Heaver, T. D., Meersman, H., & Van de Voorde, E. (2000). Do mergers and alliances influence European shipping and port competition? Maritime Policy and Management, 27(4), 363–373. Heaver, T. D., Meersman, H., & Van de Voorde, E. (2001). Co-operation and competition in container transport: Strategies for ports. Maritime Policy and Management, 28(3), 293–305. Heaver, T. D., & Studer, K. R. (1972). Ship size and turn-around time – some empirical evidence. Journal of Transport Economics and Policy, VI(1), 32–50. Jansson, J. O., & Shneerson, D. (1982). Port economics. Cambridge, MA: MIT Press. Juhel, M. H. (2001). Globalisation, privatisation and restructuring of ports. International Journal of Maritime Economics, 3(2), 139–174. Midoro, R., Musso, E., & Parola, F. (2005). Maritime liner shipping and the stevedoring industry: Market structure and competition strategies. Maritime Policy and Management, 32(2), 89–106. Notteboom, T. E. (2002). Consolidation and contestability in the European container handling industry. Maritime Policy and Management, 29(3), 257–269. Notteboom, T. E. (2004). Container shipping and ports: An overview. In: W. K. Talley, (Ed.), The industrial organization of shipping and ports, in a special issue of the journal Review of Network Economics, III(2), 86–106.

10

KEVIN CULLINANE AND WAYNE K. TALLEY

Notteboom, T. E., & Winkelmans, W. (2001). Structural changes in logistics: How will port authorities face the challenge? Maritime Policy and Management, 28(1), 78–91. Robinson, R. (2002). Ports as elements in value-driven chain systems: The new paradigm. Maritime Policy and Management, 29(3), 241–255. Slack, B. (2003). Pawns in the game: Ports in a global transportation system. Growth and Change, 24, 579–588. Song, D.-W. (2003). Port co-opetition in concept and practice. Maritime Policy and Management, 30(1), 29–44. Svendsen, A. S. (1958). Sea transport and shipping economics. Bremen: Institute for Shipping Research (Editor for Contributions in International Shipping Research, Gustav A. Theel). Thorburn, T. (1960). Supply and demand of water transport. Stockholm: The Business Research Institute at the Stockholm School of Economics. Tovar, B., Jara-Dı´ az, S., & Trujillo, L. (2003). Production and cost functions and their application to the port sector, a literature survey. World Bank Policy Research Working Paper 3123, August. Washington: World Bank.

THE EVOLUTION AND CHALLENGES OF PORT ECONOMICS Trevor Heaver As the correct solution to any problem depends primarily on a true understanding of what the problem really is, and wherein lies its difficulty, we may profitably pause upon the threshold of our subject to consider first, in a more general way, its real nature. (Wellington, 1887, p. 1)

I consider this chapter another in a line of papers reviewing maritime economics, although it has a narrower prime focus to port economics. Preceding papers are Metaxas (1983), Heaver (1993) and Goss (2002). The historical context, implied by ‘evolution,’ is different but useful orientation to the earlier papers by Metaxas and Heaver because it makes evident the burgeoning of transport economics and its more specialized components in the past 50 years. The growth shows the response of researchers to conditions of the time as research has reflected issues arising from current developments.

THEME AND OUTLINE OF THE CHAPTER The evolution of the study of shipping and ports reflects the history of the maritime industry. While the trade-offs and issues associated with ports are as old as maritime transport, analyses and decision-making associated with Port Economics Research in Transportation Economics, Volume 16, 11–41 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16002-3

11

12

TREVOR HEAVER

them reflect the conditions of the time. The merchants of ancient times were as aware as current logistics service providers of taking into account the cost of port services and overland transport, as well as the cost of maritime transport when selecting a port for the discharge or loading of cargo. Let us not claim too much new in our perspectives. However, the methods used to make decisions about the port facilities to be provided and the methods of operation have changed significantly over time. The study of the services provided in ports has become more sophisticated and demanding of economics. Yet the context for port studies remains the network of services to facilitate trade. The evolution traced in this chapter is based almost entirely on the English literature, but the trend internationally is similar. The review also accepts as ‘‘port economics’’ the range of topics and approaches typical of the meetings of the International Association of Maritime Economists (IAME) and its associated journals. The review of subject matter also relies heavily on academic publications. There is, in addition, an ever expanding literature in a range of conferences organized by associations and by professional firms. There is a substantial trade press of value to managers and economists whether corporate or academic. There is also a substantial body of reports produced by public organizations, such as the World Bank, United Nations organisations and the OECD. The growth of markets, as in the new terminal operating companies, has spurred the production of reports by maritime economic consulting firms. These firms owe their existence to the increased importance of economic intelligence to the maritime industry over the past 50 years. This is a large literature to have given virtually no attention in this review. The evolution of port economics is evidence of the increased attention given over time to the effects of ports on the transport networks of which ports are a part. The treatment of port economics can be considered first in the wider context of the place of transport in the early economic literature and, then, of its evolving presence in maritime economics. The application of economics in transport studies was, not surprisingly, a response to features of transport that raised new and important public policy issues.

THE EARLY APPLICATION OF ECONOMICS TO TRANSPORT Transport was not a part of the early work of economists. This is in spite of the writing of landowner J. H. von Thu¨nen, 1780–1850, who dealt with the

The Evolution and Challenges of Port Economics

13

role of transport in spatial economics. In his first treatise, The Isolated State (1826), he laid down the first serious treatment of spatial economics, relating distance and, consequently, the cost of transport, to the declining gradient of economic rent (land value) as distance from a single central market increased. Subsequently, transport played no part in the work of the early economists such as Alfred Marshall’s Principles of Political Economy, 1890 (Evans, 1993). This is in spite of Marshall noting, ‘‘The striking economic factor of our age is the revolution – not in production – but in transport’’ (Quoted by Svendsen, 1958). Evans quotes from George Stigler’s Essays in the History of Economics (1965): Again, perhaps the second most influential development (or a special form of the first) [i.e. the great technological advances, ignored by economists] in economic life in the nineteenth century was the improvement in transportation which never played a strategic and usually not even an explicit role in economic theory. (Evans, 1993, pp. 205–206)

The attention of some economists was drawn to transport early by the challenges raised by roads and railways for the allocation of costs among users (Winston, 1985). The railways as modern businesses of large size and wide geographic scale raised new challenges as a result of the market power that they soon exercised, their high costs for infrastructure and their ability and need to practice price discrimination. The interest of economists generated by the railways was not sustained. It was not until several years after World War II that transport again attracted their attention. Issues associated with needed investments in public investments in roads and other infrastructure led to a growth in the literature, for example, Mohring and Harwitz (1962); Beesley and Foster (1963). The growth of competition in land transport raised new challenges, such as transport costing, of relevance to private and public decisions. This was reflected in the literature, for example Meyer, Peck, Stenason, and Zwick (1959). The attention of more economists to transport, led to the evolution of active ‘‘communities’’ through the formation of organisations, such as the Transportation Research Forum in the USA in 1958 and the Canadian Transportation Research Forum in 1965. These provided an opportunity for public and private sector researchers to meet and exchange ideas, including those with an interest in maritime issues. The conferences of the organisations were dominantly national but international participation in the North American conferences stimulated support for an international conference of the College of Europe held in Bruges in 1972. This laid the foundation for the subsequent formation of the World Conference on Transport Research Society (it became a society after the Vancouver meeting in 1986). The

14

TREVOR HEAVER

growing interest in transport economics was reflected in the publication of the Journal of Transport Economics and Policy in 1967. Thus the recent evolution of transport economics goes back only about 50 years. It was in this period that more works in maritime economics started to appear. The following section traces this development but commences with a brief review of the treatment of maritime issues in earlier periods.

The Treatment of Maritime Issues The maritime sector was not without important challenges for companies and governments but these did not result in systematic analyses using economic principles. The challenges of ports were faced within the context of practices of the time and country. Until recently, that was only as a business for profit or as a public investment for a general public benefit. Maritime historians provide insights into the former practices. For example, Palmer (2003) reports on the role and practices of private dock companies in the private enterprise solution, adopted in London in the nineteenth century. Palmer (1990) notes about the North Sea ports used as nineteenth and twentieth centuries maritime history case studies, that public enterprise dominated because of the assumption that ‘‘port improvements could not for practical reasons, be a matter for private investment.’’ They were undertaken on the basis of general expectations for the public good. Into the twentieth century, the growth of trade and the increased use of larger steam vessels posed increased challenges for shipping companies and ports. However, there were few theoretical investigations. Often these are recorded in papers of proceedings of organizations. Thorburn (1960) gives the example of Biles (1923), who showed how transport costs per ton and mile varied depending on vessel size, speed and cargo loading and unloading rates. The type of information much needed by port authorities, then and 50 years later! The development of steam-powered ships led to the expansion of liner shipping and the evolution of the system of the cartels or shipping conferences. The challenges of investing in the right size and type of ship were matters for firms’ engineers and accountants. Investigations into practices of the conferences were dominated by the solicitation of views from shippers, shipping companies and others. While the focus of these investigations, such as the British 1909 Royal Commission on Shipping Rings, was on the case for and against anti-competitive practices of liner companies, interest in the study of liner conferences by economists did not emerge until later.

The Evolution and Challenges of Port Economics

15

During the first half of the twentieth century, books on the maritime industry appeared. These were largely descriptive of conditions, business structures and practices but commencing to be comprehensive. Owen, writing in 1914, notes segments of the maritime industry have long been ‘‘the subject of important treatises’’ but ‘‘the whole, collectively, have apparently never been dealt with at all’’ (Owen, 1914, p. v). His first chapter is essentially devoted to the description of the condition and organisation of ports. In commenting on the location of some new ports down river away from the communities that they serve, he notes that this ‘‘accords but ill with sound principles of port economy’’ (Owen, 1914, p. 9). He may have had his conclusion wrong but he was well aware of the trade-offs in land and ocean transport costs. He knew well, too, the gateway role of ports; the word port is after all derived from the Latin porta which means gate. He notes: On the Continent, they regard a port as a gateway for the country’s trade, and the wider open the gate, and the smoother the road, the greater, they consider, will be the tradegain to the country. (Owen, 1914, p. 17)1

The books typically have chapters on ports. The descriptions of ports are often by their traffic, location and general characteristics of their harbours, and according to the administrative structure. They pay attention to the types of facilities in ports, not only to the piers and wharfs providing berths for ships and facilities for handling cargoes but noting also the quality of the ocean and inland connections. For example, Hough (1924, p. 55) notes that in New York the ‘‘the congestion of traffic on Manhattan Island is extreme’’ because of ‘‘the lack of adequate rail-water co-ordination.’’ The general nature of problems does not change. However, the technical and institutional conditions in which they arise and the socio-economic environment that surround them do change. A development of note is the formation by port authorities of organisations to advance the management of ports. An early example is the American Association of Port Authorities (AAPA) formed in 1912 when public port administration was in its infancy. Though public port agencies existed in a number of American states, few, if any, actually owned or operated marine terminals. Commercial ports were for the most part dominated by powerful railroad corporations, which owned the terminals and controlled access to harbour areas. The AAPA believed ports so important to the nation that public not private enterprise should have the upper hand. To achieve its goals, the AAPA furthered the science of port administration and professionalism among its members. Although not dealing with port

16

TREVOR HEAVER

economics as such, it was a forerunner of organizations in the maritime sector that would remove the veils of secrecy and tradition.2 Reasons for the Slow Application of Economics to Issues in the Maritime Sector The maritime sector did not attract the attention of economists. As a result of the international nature of the industry and the high level of competition among shipping companies, maritime transport did not give rise to comparable problems of market power as the railways. Economic regulation was not necessary, as it was with the railways. Furthermore, the port business is more a collection of diverse service providers than an industry. It did not provide the ready focus for systematic study as the railways. Even ‘‘battles of the ports’’ were ‘‘commonly only another name for the rivalries of the railways.’’ This is a perspective by Owen (1914, p. 13) on conditions in England but the actions of railways to capture port hinterlands were also a concern in the USA, contributing to the specific regulation of railway import and export rates. Goss (2002) notes three reasons for limited development of a maritime economics literature prior to 1960. First, writers’ attention was focussed on new developments and the enterprises needed to bring about growth rather than on the economics involved. Economics was implicit in the engineering analyses, as in the cube rule determining the preference for large size in steam ships. Second, economics and, particularly, microeconomics was not a widely understood subject. The third reason was the defensiveness of shipowners. This was in keeping with the history of the industry. Commercial shipping evolved from merchants owning their own vessels and engaged in series of trade ventures. It was an environment in which secrecy about business was valued (Berglund, 1931). The shroud of secrecy pervading the industry did not lift easily. Berglund notes that in shipping ‘‘up-to-date business technique exists with an outlook distinctly archaic’’ (1931, p. 2). Nearly 30 years later, Thorburn (1960) notes that pricing port services ‘‘is largely bound by custom’’ (p. 156). Secrecy and custom were pervasive in the industry. One result was that well into the second half of the twentieth century, data on shipping were hard to obtain, unlike for the railways for which regulation in the nineteenth century soon resulted in public access to data (and, more recently, the airlines). Consequently, opportunities for quantitative analysis in maritime economics were limited. Books on shipping and the maritime industry were largely descriptive.

The Evolution and Challenges of Port Economics

17

THE EMERGENCE OF SHIPPING AND PORT ECONOMICS Even after the Second World War, general books on the maritime industry were essentially descriptive. No texts existed that provided an analytic economic framework. This was in spite of a rapidly changing environment for trade and shipping. However, change was afoot, evident in the role of new organisations and finally in the attention of economists to maritime issues. Openness Through the Role of International Organizations Greater openness whether required or voluntary was developing in the industry. Following the establishment of the United Nations in 1945, several countries proposed that a permanent international body should be established to promote maritime safety, more effectively. An international conference in Geneva in 1948 adopted a convention formally establishing the Inter-Governmental Maritime Consultative Organization (IMCO), whose name was changed in 1982 to the International Maritime Organization (IMO). IMCO was a new type of organisation in the maritime industry. Challenges of the industry other than safety were getting more widespread attention. There was rapid growth of trade in the post-war years when ports were still handicapped by under-investment or even by destruction in the war years. The result was that delays of ships in ports in the 1950s were longer than they had been in pre-war years (Svendsen, 1958, p. 99). The costs to shipping and trade were obvious. One consequence of this was the formation in 1952 of the International Cargo Handling Co-Ordination Association (ICHCA) to help remove some of the unsatisfactory conditions. In 1955, the International Association of Ports and Harbours (IAPH) was founded. One of the three elements in the IAPH mission is to collect, analyse, exchange and distribute information on developing trends in international trade, transportation, ports and the regulations of these industries. Acceptance of such organisations is a reflection of the gradual change that was taking place in maritime communities. But still books on the maritime industry were descriptive and almost devoid of economics, for example the books by McDowell and Gibbs (1954) and Metcalfe (1959) that were sometimes used as texts. Indeed, it was in engineering, exemplified by the excellent economic analyses of ships by Harry Benford, of the Naval Architecture and Marine Engineering faculty of the University of Michigan from 1948 to 1983, that the application of economic analysis was advanced

18

TREVOR HEAVER

effectively. However, analytic economic contributions appeared in the late 1950s when two Scandinavian economists, Arnljot S. Svendsen and Thomas Thorburn, wrote independently on shipping economics. Their work still deserves special examination not just because their contributions were first but because of their distinctive nature.

The Contributions of Svendsen and Thorburn Gustav A. Theel of the Institut fu¨r Schiffahrtsforschung Bremen, which had a role to further research in shipping, recognized a unique contribution in the Norwegian lecture notes of Arnljot S. Svendsen, then Associate Professor at the Norges Handelsho¨yskole Bergen. Theel notes in his foreword to Sea Transport and Shipping Economics (he had the notes translated into English and German for publication) that ‘‘there are no instructional books on shipping economics either in the German or the English language which at least treat in det[e]il the economics of shipping’’ (Svendsen, 1958). Svendsen’s book describes and sets out algebraically the relationships of various inputs with outputs for shipping and throughput for ports. It incorporates port time as a factor in ship costs. It does not have data for numeric examples. Svendsen approaches his subject recognizing that shipping economics is simply the application to sea transport of the same methods and analytic means that are used in the general study of economics. The book, widely regarded as the first on shipping economics, carries with it three important lessons. They are, the arbitrary nature of the field of port economics, the important role of institutions and the important role of individuals in the development of the subject. Elaboration on these lessons is provided later in the chapter. Svendsen’s book was followed 2 years later by the publication of Thomas Thorburn’s doctoral thesis (1960). While Svendsen’s book is the first economic treatment of shipping, Thorburn’s book is the more detailed and sophisticated. He examines price making under different demand and supply conditions for sectors of the shipping market and for port services. His aim was to clarify important relationships in the supply and demand of water transport, especially reflecting the character of ships and ports, both of which are taken into account when assessing the influence of distance in water-borne transport. Thorburn’s book was not only the first but, arguably, it remains the most comprehensive theoretical treatment of maritime economics. It deals with issues of the cost structure within ports and the consequences of ports (and the appraisal of port investments) for the total

The Evolution and Challenges of Port Economics

19

cost of transport services. The author’s and the book’s influence on the field may have been limited by the book’s limited circulation (it was a published thesis) and the fact that Thorburn, unlike Svendsen, did not work subsequently in maritime economics. Thorburn examines the nature and behaviour of ship and port costs in a sequential manner. For both ships and ports, he first uses simple models of single ships and single ports before moving to models of multiple ships and ports. His purpose was to show how the conditions of supply and demand work to determine prices. He suggests that the payments made by ships for port services amounted to some 30 per cent of their total costs and that the addition of port costs falling on the cargo raised port costs to 50 per cent. He notes that if the time costs of ships, goods and land vehicles in ports were included in the definition of port costs the percentage would be even greater. Whether defined as port costs or not, he recognized that they must be included in total costs. Thorburn raises issues about port costs that are still relevant today. He is concerned with the internal cost structure of ports. For example, he notes that the use of historic or present value of assets, the assumed value of the land on which the harbour is built and the public contributions for entrance channels are important matters in cost determination. He notes, ‘‘These questions, however, do not yet seem to have attracted much attention in harbour circles’’ (1960, p. 134). (Some of them still have not.) Beyond these matters, he notes that the level of monopoly power associated with ports (a matter of debate then and now, although competition among ports is generally greater now than in 1960) denies the opportunity to use revenues as a measure of economically rational investment. Port economists are still challenged to find effective measures of port efficiency. He also notes the challenge that harbour activities are distributed among many kinds of independent firms and authorities so that cost determination, let alone optimization, is difficult. The segmentation of the port activities among enterprises can result in sub-optimization. Port authorities are often ‘‘not in charge of the operations, and have not at their disposal the statistical information required to ensure that improvements will be made in the way that is best for the whole harbour’’ (1960, p. 138). Circumstances have not changed. Thorburn (1960) also has a valuable perspective on what composes the port. He notes: The objective suggested above [the economically rational performance for harbour activity as a whole] implies that the total activities are regarded as a unit y. We may imagine a ring drawn round the harbour area encircling the harbour basins, the quays, the warehouses and sheds and storage areas, as well as the railway tracks and roads

20

TREVOR HEAVER connecting them. All activities inside this ring in connection with the goods are regarded as a unit, and the total costs of the services with the same turnover of goods are to be kept as low as possible y. In certain complicated problems even the harbour will be too small an area. To attain co-ordination between the building of vessels and the design of harbours, for example, both shipping and harbour activities in several harbours should be included in the same calculation. (p. 139)

Thorburn’s description of a ring applied to a time when cargo storage and handling took place in an area close to the quays. Applied to today’s environment, it would encompass the drayage of containers to locations within an urban area for freight handling and container storage. Had Thorburn’s broad view achieved more currency, today’s challenges of port communities in urban container logistics might have been tackled differently. Thorburn also dealt with the external implications of ports for the costs of shipping. He acknowledged the consequences for the inland carriers. He applies his total-cost minimization model in the appendix of his book to a network of land, port and shipping services with data representative of previous conditions in Sweden. Neither Svendsen nor Thorburn deal with the industrial organization of ports, although Svendsen does deal with the economic organization of shipping companies. The absence of this treatment may be because European ports had simply remained under government jurisdiction. In other countries, such as the UK, Canada and the USA, significant changes had taken place through the assumption of public control. This took place to prevent financial collapse in the 1930s and to ensure protection of the perceived public interest in ports. The organization of ports remained a matter of political faith for many years before the relevance of economic arguments was considered. Characteristics of the Literature to 1973 The increased attention of economists to shipping in the 1960s is reflected also in a number of books. They include two books that focus on national interests in shipping. They are on American shipping policy (Ferguson et al., 1961) and on British shipping (Sturmey, 1962). The 1959 doctoral thesis of Zannetos (1966) subsequently was the foundation for a book on tanker rates. The first text on the economics of sea transport (O’Loughlin, 1967) was, as the author notes in the preface, ‘‘so long appearing’’ that no justification was necessary. The book includes three brief chapters on ports. However, the brevity of the book precluded it from making a contribution beyond being a text.

The Evolution and Challenges of Port Economics

21

Attention to ports was the result of post-war growth of trade, making evident the actual and potential bottlenecks in ports. The importance of making investment in ports to accommodate more trade and larger ships was obvious. The focus was on the ship/port interface as noted by the Rochdale Inquiry. While the development of policy towards ports must take account of domestic transport problems as a whole as well as shipping needs, the essential function of a port must be to serve the shipping industry. Moreover, the ship/port interface is both more critical and more important by its very nature than is the port/inland transport interface. (1970, p. 174)

This reality of the time reflected not just the growth of trade but also the rapid increase of vessel sizes, especially in the bulk trades. It resulted in attention of economists and others on the ship/port interface. Some of the studies examined the methodologies by which the economics of port investments should be judged, the framework evaluating the costs of ships in port and actual consequence of port time for ship costs. In spite of the promotion and use of cost benefit analysis for many years in water resource projects in the USA (Eckstein, 1958), its application to transport developed slowly (Beesley & Foster, 1963). Goss (1967a) presented cost benefit analysis as the method to follow to appraise ‘‘whether to invest, how to invest, when to invest and where to invest’’ in ports. Goss (1967b) presented calculations indicating the magnitude of savings resulting from reductions of port time for the typical closed shelter deck liner vessel. The significant savings were consistent with Thorburn’s expectation. In the following year, the Canadian National Harbours Board published a study by Heaver (1968) on actual bulk ship costs and their implications for port investments. The general absence of such public information led the British National Ports Council to publish a summary of the report (Heaver, 1970) and for Goss and Jones (1970) to complete a similar study in the English Board of Trade. The focus on the ship/port interface was reflected in a United Nations study (1967) and in studies of the conditions affecting the actual time that ships spent in port (Heaver & Studer, 1972; Robinson, 1978). The latter studies were on bulk ships. At the same time ‘‘the container revolution’’ was underway. It prompted many studies for ports and shipping lines into the propensity of cargo to be containerized and the implications for ports of new container ships. Two of the latter studies (McKinsey & Co., 1967; Arthur D. Little Ltd, 1970) were good sources of information about the characteristics and costs of container shipping. They examined the economics of large container ships and their

22

TREVOR HEAVER

implications for the number of ports of call in a port range. The need for more transport research led the UK government to fund four interdisciplinary research centres through the Science Research Council.3 This funding supported the start for the Marine Transport Centre at the University of Liverpool where Sydney Gilman led research focussed on ship choice, the likely evolution of ship itineraries and aspects of container terminal operations (Gilman, Maggs, & Ryder, 1977). These issues have resurfaced with the recent phase of trade expansion and increases in ship size. They reflect the central place of network analysis that unifies port and shipping issues in maritime economics. The increasing attention on ports is evident in two books on ports that appeared in 1971. James Bird is among the number of geographers who have made contributions relevant to the port economist. In the preface to his book Seaports and Seaport Terminals (1971), he warns ship lovers ‘‘For once, you encounter an author who is determined to make seaports the stars of the show.’’ He showed how the changes in shipping and freight handling were resulting in integrated logistics systems through specialized terminals and thus changing the competitive dynamics of trade in hinterlands and through ports. Johnson and Garnett (1971) took an approach even further removed from shipping. Their book arose from a study mainly concerned with the likely effects of containerization on the Glasgow conurbation. The book examines the nature and effects of containerization in advancing door-to-door transport services and it finds a need to advocate greater application of a total distribution cost approach by British managers. The authors, like Bird, note the implications for port competition. However, they do not anticipate the congestion problems now associated with the port/inland interface. That is for a later time. The divide in the evolution of port economics chosen here is 1973, the first year of publication for the journal Maritime Studies and Management, changed in 1976 to Maritime Policy and Management (MPM) to reflect the public and private sector readership.

PORT ECONOMICS SINCE 1973 There is a much greater public literature on port economics since 1973 than there was previously. This creates a challenge to capture the evolution over the 30+ years. In an earlier article (Heaver, 1993), I categorized the topics of articles written in maritime economics. Following the same approach in port economics is more difficult because finer decisions would be required

The Evolution and Challenges of Port Economics

23

about topics in order to categorize publications. Therefore, I have chosen the approach adopted by Winston (1985) in his review of transport economics; that is to follow a conceptual approach. This requires a different sort of judgement about the conceptual construct to follow in examining the literature. Since the orientation of books is often wide, they are treated separately from the rest of the literature. The Characteristics of Books Books written as texts in the area of maritime economics either do not have chapters on ports (Wijnolst & Wergeland, 1996; Stopford, 1988, 1997), or only have space for general coverage (McConville, 1999). Only one book has appeared with the title ‘‘Port Economics’’ (Jansson & Shneerson, 1982). The focus of this book is on optimal pricing and investment in ports. It deals at some length with production theory of port services and the application of queuing theory and congestion costs. It examines the practices and principles of pricing for port and stevedoring services. In these chapters, the book is dealing with particular aspects of port economics. The book does not attempt what might now be regarded as a comprehensive coverage of port economics. However, in the description of the functions and evolution of port activities, the authors identify the network nature of the multi-stage functions in ports and the shifting bottleneck that has occurred within the network of activities. The bottleneck theme is useful and appropriate in this network business. The authors note that the bottleneck has tended to move over time from stevedoring on the ship to crane and yard productivity. In keeping with this approach, it is argued here that in 2004, it was evident that the bottleneck has now moved to the terminal/ inland interface. Certainly keeping the network (and an expanding one at that) in balance was and remains a major challenge. Because of the multiple parties involved in port activities, several parties may need to be involved to remove a bottleneck. This poses an organizational challenge on top of the economic one, as noted by Thorburn (1960). Two other books warrant recognition here. They are the edited volumes by Grammenos (2002) and Leggate, McMconville, and Morvilllo (2005). The Handbook of Maritime Economics and Business is an excellent and substantial volume of 930 pages in which four papers provide very good coverage of major aspects of port economics. However, their contributions are best considered later under the concepts to which they relate. The same is true for the chapters in the book edited by Leggate et al. which deals with more selected perspectives.

24

TREVOR HEAVER

The Changing Orientation of Port Studies since 1973 If the underlying issues in port economics have not changed since ancient times, they have certainly not changed since 1973. Tracking these changes is aided by the presence of two main journals, MPM since 1999, and the International Journal of Maritime Economics, changed to Maritime Economics and Logistics (MEL) in 2003. Also, the formation of the IAME in 1992 has resulted in an increase in the number and size of specialized conferences. It is a challenge to review the resulting amount of material. Heaver (1993) notes that in selected general economic journals published between 1960 and 1987, just less than 22 per cent of articles on maritime economics were on port topics. Shipping conferences had most coverage, 36 per cent of the articles. When MPM first appeared, its coverage of port subjects was low, less than 12 per cent in the first 5 years. It was an interdisciplinary journal and covered a wide range of marine topics, for example, on fisheries and the then topical Law of the Sea. Between 1982 and 1991, articles on port topics averaged over 33 per cent of all maritime articles in the general economic journals (Heaver, 1993). In MPM and MEL, from 1999 to mid-2005, over 31 per cent of the articles have been port related. Thus, the share of articles on port economics has remained similar over more that 20 years, but the orientation of articles has shifted. Given the interdependence among aspects of port economics, identifying categories of topics involves arbitrary divisions. However, for convenience, developments and issues in port economics are discussed under six topics. They are: relationship of ports with ship costs; issues of port costs and pricing; industrial organization related to ports; the competitive relationship among ports; assessing port performance; and specialized studies. This review covers port economics topics that are being researched. I do not claim that articles cited have been scientifically selected although I have used general articles rather than regional or specific examples as much as possible. Occasionally, I point out topics that appear to be neglected.

The Relationship of Ports with Ship Costs In 1973, the increase in the size of bulk ships was already a fact; VLCC and ULCC tankers were coming from the yards as the oil crisis unfolded. The large bulk carriers could only operate at specially built terminals so that the uncertainty of port – ship cost trade-off was largely removed; but not the uncertainty of market product demand. It was in the container trade where

The Evolution and Challenges of Port Economics

25

the uncertainty of ship size and the economics of route structures and terminal design remained. Papers such as Gilman and Williams (1976) and Gilman (1999) captured the issues of how ship size affected the network of ports at which ships might call and the need for efficient port turnaround times. The introduction of post-Panamax sized container ships in 1988, unleashed a new round of uncertainty about the size of these ships and the consequences for ports. The anticipated size of the largest ships, 8 then, 9 or 12 thousand twenty-foot equivalent units (TEU), raises issues about the economics of extensive and complex network structures in which hub and transhipment ports may play a key role, see, for example, Baird (2002a) and Gilman (1999). The trade-off of ship and port costs in the total cost picture has not changed but its complexity has increased as the range of networks becomes more complex and time and reliability issues become more involved through possible multiple container handlings. Lirn et al. (2004) use, in part, a survey to explore the importance of service attributes for transhipment port selection. This research is on a current issue and the survey method is not one that would have been practical in the past; times have changed as the maritime industry has become more open.

Issues of Port Costs and Pricing The issue of the structure of costs and the appropriate level of charges for the use of port facilities and services have been consistent issues in port economics. This is reflected in many publications over time, for example, Heggie (1974), Bennathan and Walters (1979), Walters (1976), Jansson and Shneerson (1982), Goss and Stevens (2001) and Strandenes (2004). The issues remain but are affected by the changing industrial organization in ports, especially privatization and commercialization, considered below. Different divisions of responsibility now exist and create increased opportunities to reform port pricing (Goss & Stevens, 2001). Strandenes (2004) suggests greater attention to the effects of the level and structure of prices on the efficiency of shipping as well as the efficiency of ports by relating port charges to port costs and to time in terms of access to berths and service quality. The importance of the economics of time in various dimensions has become a more important feature in logistics and transport economics as supply chain management practices have demanded more time-responsive logistics systems. In spite of the changing organization structure in ports, papers addressing the costing and pricing topics still tend to do so under the broad heading of

26

TREVOR HEAVER

‘port’ rather than more specific functions. Cost recovery for navigation aids and dredging still remain challenging in economics and much influenced by political beliefs (Haralambides & Veenstra, 2002). Baird (2004) explicitly examines the concept of public goods, its role in public financing of ports and its possible market distorting effect. Concepts of costs, finance and pricing are very relevant to port policy issues, as are concepts about the nature and distribution of port benefits (Goss, 1999). Within the context of port costs, issues associated with the valuation of assets, a matter raised by Thorburn, get little attention. The use of historic or present values is an issue in all costing. The value of port land may be a matter conveniently ignored in public ports where no taxes are paid on land. This, of course, is a hidden benefit to those ports. Where taxes are paid but the land is restricted to port uses, the appropriate basis on which to value port land, like that of airports, public utilities and other public uses, can be important to the economics and competitiveness of the undertakings and is, therefore, contentious (Heaver, Fitzgerald, & Frounfelker, 2002; Heaver & Tretheway, 2002).

Industrial Organization Related to Ports One of the major changes in ports and in port economics over the last 20 years has been in industrial organization, that is the structure and behaviour of organisations in and related to port activities. It has been driven by two separate but related developments. The first has been changes in the management of global supply chains and the role of logistics within them. Shippers and firms providing transport and related logistics services have undergone great change. Most firms have followed strategies to widen their services. For major companies that means growth to global scale and increased corporate market share through consolidation strategies that enable economies of traffic density on transport routes and in logistics markets. It has been associated with the development of expertise in terminal management in liner firms and in separate terminal management companies (Heaver, 2002; Midoro, Musso, & Parola, 2005; Notteboom, 2002, 2004). A new business model and a new economic market emerged in terminal management that is reflected in the port economics literature. The second development has been the devolution of public responsibility in ports through privatization and commercialization of activities (Brooks, 2004; Cullinane & Song, 2002). These changes in industrial organization have complex ramifications. They are captured well by Juhel (2001) and Robinson (2002), who describe

The Evolution and Challenges of Port Economics

27

the interactive changes in the individual components of the logistics systems, of which ports are one, and relate them to the changing values of shippers. Categorizing the changes within the literature on industrial organization when individual papers have their own specific purpose is a challenge. Some arbitrary decisions seem necessary. The approach adopted here is to make a separation between writings in which public policies appear to be central compared with those in which strategic decisions, whether by private or public corporations, are central issues.

The Public Administrative Organization of Ports The extent of privatization in ports varies widely among countries and ports. The business of some ports may have been entirely privatized, as in the UK, or, as is more general, concessions have been awarded to companies by landlord port authorities (Baird, 2002b). The result is a wide variety of governance structures among countries (Brooks, 2004). A considerable and diverse treatment of governance issues exists in the current port economics literature. The increased presence of private capital and management in ports has been a widespread and evolutionary process. Suykens (1985) noted that in the port of Antwerp the ownership of lifting capacity in quay cranes (excluding container gantry cranes) shifted from totally public in 1950 to twothird private in 1984. The full privatization of certain ports in the UK and commercialization and privatization in New Zealand were dramatic but exceptional policies. Baird (2000) covers the landmark sale to private interests of the ports of Associated British Ports in 1983. The shift from public to private investment in ports was not solely ideological or based on beliefs in the efficiency of private compared to public enterprises. It was affected too by the greater specialization of terminals in the logistics of trades, as noted by Bird (1971), and the opportunity to utilize private capital in face of a shortage of public capital. As terminal operating companies have grown, they have also been able to bring to ports an increased level of expertise as well as capital. The weighting of reasons for privatization and commercialization has varied from country to country (Goss, 2000). The different circumstances are covered in a number of papers that are country or region oriented, for example, Cullinane and Song (2001), Hoffman (2001), Ircha (2001) and Notteboom and Winkelmans (2001a). The regional aspect of port governance not only relates to groups of countries but also to levels of local government. For example, there are economic ramifications from the policies of provincial and local governments in China

28

TREVOR HEAVER

(Wang, 2004). In Europe and North America, also, policies of local governments about the support or taxation of ports (and airports) are issues. Policy issues for governments and assessed by port economists vary with the general regime under which the ports operate. In Europe, the EU is still grappling with its ports policy, (Pallis, 1997; Psaraftis, 2005). Issues about the efficacy of policies especially when faced with different policies among competing ports remain, as between Canada and the USA (Heaver, 1995; Ircha, 2001). The effectiveness of legislation in providing appropriate powers to port authorities remains an issue, for example, in Canada and Australia (Ircha, 2001; Everett, 2003). Gilman (2003) points to the problems of evaluating applications for port projects on a project basis, in this case on environmental grounds, without some central authority considering the consequences for the port and transport system as a whole. Gilman’s analysis not only draws attention again to the role of a central authority, but it is a reminder of today’s heightened reasons for public intervention on security and environmental grounds. Whether the grand scale model can be helpful in decision-making is debateable but, at least, it is appropriate for port economists to explore such an option (Luo & Grigalunas, 2003). Policy issues remain in port economics.

Strategic Issues in the New Industrial Organization The changes in industrial organization affect all the players and give rise to a wide range of issues for port economists. In light of the emergence of global terminal operating companies, the ever-larger shipping companies and the increased pressures on the port/inland interface, what strategies should port authorities and others adopt? Port authorities can affect the competitiveness and structure of logistics services through their policies and involvement in services. But how much involved they should be and what are the economic and business consequences (Heaver, Meersman, & Van de Voorde, 2000, 2001; Notteboom & Winkelmans, 2001b)? In particular, governments and port authorities are faced with different challenges to ensure adequate competition (De Souza et al., 2003; Flor & Defilippi, 2003; Notteboom, 2002). The increased competition among ports as parts of alternate routes is being offset by the increased dominance of large companies within a port range or even within a port. Competition policy has become a relevant consideration in ports. Although, perceptions on the economics of dedicated terminals have changed in favour of the strategy, questions are still appropriate (Haralambides, Cariou, & Benacchio, 2003; Turner, 2000).

The Evolution and Challenges of Port Economics

29

The greater integration of international logistics and the increased pressure on the port/inland interface is resulting in more attention to relationships in port communities. Research into the different roles of parties in enhancing levels of inland integration with ports is evident (Carbone & De Martins, 2003; De Langen, 2004; Martin & Thomas, 2001; Meersman, Van de Voorde, & Vanelslander, 2005) and may yield insights into market imperfections that need to be tackled.4 This is in an environment in which one of the consequences of improved through transport is the reduced impact per ton of cargo on local employment unless port community strategies are adopted to enhance local value-added services (Slack, 2003).

The Competitive Relationship Among Ports The description of competition among ports has changed from characterization as competition for common hinterlands (and forelands) to competition among alternate logistics systems of which ports are a part. In reality, the base of competition is the same; achieving the most efficient total service. But, today, there are more likely to be contractual relationships among participants in a logistics chain that solidify and, hopefully, improve the performance of the chain. Processes and language have changed to reflect the higher level of integration. Three broad categories of inter-port competition may be recognized. First, there is competition among ports in a port range as parts of logistics chains, as between ports in the ranges of northwest Europe or of the North American west coast. The success of the chain is recognized as dependent on the each of the parts working to provide an effective, reliable system. The second category of competition arises from the existence of routings through alternate ranges of ports, as in competition between Mediterranean ports and those in northwest Europe, or the competition among ports on the North American west, Gulf and Atlantic coasts. Finally, the advent of same day of the week liner services has sharpened the competition among ports on a route. For example, a line may substitute a port on one coast for a port on another if the profit contribution to a vessels’ route is enhanced by substituting one for the other within the cycle time available under the constraint of same-day service. Competition among ports may involve various activities beyond working with logistics partners to enhance the quality and cost of their own services; for example, collaborative marketing initiatives. Particularly for neighbouring

30

TREVOR HEAVER

port authorities, suggestions often arise about when and how they might cooperate rather than compete (Fleming & Baird, 1999; Hayuth, 1993; Song, 2003). Competition among ports has been the subject of some new approaches in port economics. Yap and Lam (2004) propose the use of indifference analysis as a means to examine port competition and complementarity. Nir, Lin, and Liang (2003) review methods common in shipper choice analysis and apply a multinomial logit model to container traffic through Taiwan’s ports. Malchow and Kanafani (2001) applied a similar approach on a shipment basis to four commodity types in American exports. These papers represent a different and more quantitative approach in port economics than had been common in the past, a reflection of the improved availability of data and the tools of researchers. The same is found in recent studies of performance and efficiency within ports. Assessing Port Performance The performance of ports and terminals is dealt with by a number of authors. Some papers are oriented to a variety of operational matters. Other papers deal with the more general matter of assessing port or terminal performance. This review concentrates on the latter topic. The heterogeneity of ports and, within them, terminals in terms of their physical characteristics, the mix of traffic handled, the size and volume of cargoes handled by ships and the lack of available data have all been factors limiting efforts to measure port or terminal performance in sophisticated ways in the past. There is greater standardization in container terminals but there is still considerable diversity. Nevertheless, the apparent or greater similarity among container terminals has led to an increased interest in performance measurement and comparison. Relatively simple but standard measures have been used, basically, to compare (‘‘benchmark’’) performance over time or in one port with another. However, the lack of clarity and uniformity of data often used led Dowd and Leschine (1990) to comment, ‘‘y the measurement of container productivity has more in common with a commercial art form than with science!’’ (p. 110) It is appropriate to comment on a successful programme using standard measures. In Australia, monitoring the effects of waterfront reform was the rationale for port performance measurement, at first by the Australian Productivity Commission (2003) and then on a regular basis by the Bureau of Transport and Regional Economics (BTRE – as the old BTE is now known)

The Evolution and Challenges of Port Economics

31

in the publication Waterline of (http://www.btre.gov.au/docs/waterline/ wline.aspx). Hamilton (1999) reports on the challenges and successes of the Australian experience. The experience has been positive because of the clear purpose of the data and consistency with which they have been collected, reported and, perhaps most important, the discussion of the data. Cullinane (2002) and Cullinane, Song, Ji, and Wang, (2004) describe the concepts of productivity and efficiency measurement for ports and terminals and review the evolution of approaches to performance measurement. The authors recognize a number of early studies that broadened analysis of terminal performance beyond simple measures, for example, Chang (1978), Kim and Sachish (1986), but it was not until well after Roll and Hayuth (1993) had explored the application of Data Envelopment Analysis (DEA) to ports that it was undertaken (Martinez-Budria, Diaz-Armas, NavarroIbanez, & Ravelo-Mesa, 1999). Now, as Cullinane makes clear, the application of DEA is an area in which port economics has progressed with the theoretical development of the techniques. The improvements in port productivity and efficiency assessment in all their forms are all important to ongoing efforts to improve the performance of maritime logistics systems. As Bichou and Gray (2004) note, expansion to frameworks that encompass value-added logistics services would be beneficial.

Specialized Studies There is commonly a need for an’other’ category. In this case, four topics fall into this category. They are, studies of labour wage rates; economic rents in the port context; the service of harbour tugs; and maritime security. The functioning of labour in ports is very important but has received less attention in the port economic literature than in the maritime history literature. Talley is an exception among port economists as he has published a number of papers on port labour – and has a chapter on the subject in this book. Talley (2004) shows that the wages of dockworkers have increased relative to those of workers in the deregulated periods for trucking and railroads in the USA. This conclusion raises some interesting questions about the consequences of the power of port unions. The subject of economic rents in ports has been raised by Goss (1999). In his paper, Goss takes a basic concept in economics and speculates on whether limits on competition among and within ports enables rent to be earned by participants and if so which ones. Of course, the ‘rent’ component of a return on labour or capital is not immediately evident. The higher

32

TREVOR HEAVER

earnings found by Talley for dockworkers than truck and railway workers since those modes were deregulated suggests that dockworkers may gain some rents in their wage levels. It opens the question of whether the rapid growth of terminal operating companies is because they have been able to capture some rents. The third topic is a study by Atkin and Rowlinson (2000) of harbour tug services. This is one of a number of necessary specialized services in ports that are usually not treated as a ‘market’ for analysis. Harbour tug services are provided locally and in markets that range in size from supporting one to a few firms. They are necessary in most ports. Therefore, this study of conditions in North European ports will be of wider interest than to those interested in those ports. Studies of this and other specialized services in ports are worthwhile to understand the full range of markets in ports and to a better understanding of resource allocation within ports. Finally, the terrorist attacks in the US in September 2001 added new dimensions to the ever increasing role of ports in the field of maritime safety and security. Concern for the safety of life, property and the environment contributed to the formation of the IMO in 1948. With the exercise of portstate control, the roles of ports in these activities have increased. However, post 9/11, measures to achieve more effective safety and security have led to new regimes which have logistics and economic consequences for trade, shipping and ports that will be subject to increasing study, for example, Bichou (2004).

LESSONS LEARNED FROM THE EVOLUTION OF PORT ECONOMICS This review of port economics gives more attention to ‘early’ – pre-1973 – literature than might have been expected. This has not been edited to a lesser amount for two reasons. The first reason is to make evident that the economic objective of ports and the systems of which they are a part have not changed. It is to provide value to shippers. (The extent to which governments ascribe other objectives, for example, local employment or sovereignty, is beyond economics although the objectives may have economic consequences worthy of economic analysis.) The second reason is to make clear just how young and dynamic the subject is and to provide a sense for individuals and institutions of the critical value of their current contributions. Examination of the evolution of port economics shows how the subject has grown with but lagged the development of transport and maritime

The Evolution and Challenges of Port Economics

33

economics. Greater attention within maritime economics has been paid to port issues as the growth of trade led to bottlenecks in ports. Subsequently, the evolution of port economics reflects the shifts that have occurred in the bottlenecks and obstacles to efficiency in the logistics of maritime trade and to the changing structure of the industry. It has also featured the application of more sophisticated economic methods, as has economics generally. However, the increased attention to specific aspects of port economics does not change the fundamental role of ports as a part of maritime, transport and wider logistics networks to meet the needs of shippers. Notable contributions to maritime and port economics have come from some individuals with a background other than economics, for example, H. Benford in engineering and J. Bird in geography.5 In part, this reflects the essentially overlapping nature of studies in industry sectors. It accounts for the collaborative relationship between the society of Naval Architects and Marine Engineers and IAME. Historians and particularly maritime historians provide insights through their research. However, in my limited reading of historians’ work, I have not found occasions in which the working of economic forces has been an identifiable theme in their studies. The evolution of port economics reveals the important roles of individuals and institutions in the development of the subject. The initiatives of Theel and Svendsen in the early days and Richard Goss in the development of IAME and support for and as the second editor of MPM, are examples spanning the years. Each programme has had its champion. Goss (2002) mentions those of selected programmes, some known internationally, others, less so. However, for many now involved in maritime economics, the leadership provided by Goss stands out. The role of individuals must be seen in the context of the institutions of which they were a part. Institutions provide a critical mass for achieving and giving visibility to work and to enabling leadership. They may also facilitate networking. The leadership shown in the programmes at Bremen and Bergen is early evidence. Subsequently, the establishment of university programmes for teaching and research, for example in the UK at Cardiff, Liverpool and Plymouth, were important in developing graduates with maritime economics training and in producing research papers. The prominence of individual programmes changes over time but, important for maritime economics, has been their increasing number, with programmes found in many countries and serving national and international students. The institutional development of the field was evident, also, in the evolution of research institutes supported by industry, such as the Japan Maritime Research Institute (JAMRI). Public organisations, such as the

34

TREVOR HEAVER

Committer on Shipping of the United Nations Conference on Trade and Development (UNCTAD), have also made important contributions, influenced as they have been by the issues of the time and their leadership. Private sector firms also identified needs and opportunities. Drewry Shipping Consultants, founded in 1970 was one of the first firms to appreciate that the changing nature of the maritime industry was associated with value in shipping market data and economic analyses. Many shipbrokers that previously had made data available freely, commenced to charge for their reports. A market for maritime data and intelligence developed. An important institutional development was the formation of IAME in 1992, the goals of which as set down in its constitution are to promote the development of maritime economics as a distinct discipline, to encourage rational and reasoned discussion within it and to facilitate the international exchange of ideas and research. IAME has been of great value in providing a networking opportunity for individuals and research programmes of modest or small size. It has also facilitated the success of the journals in the field. While the development of port economics, like any field, can highlight the roles of particular individuals and institutions, the effects of their contributions are cumulative. Ideas and methods advanced through publications stimulate others. Testing the validity of conclusions across different environments can be useful. These developments are evident in port economics. However, it is still easy to ignore writings of prior periods. Old phenomenon dressed in new terminology may sound so different as to be new. Changes do take place in technologies (containers) and in the size and nature of organizations but the underlying economic challenge of bringing the components of transport together efficiently has not changed. Our goal today, when using language about logistics integration, is not fundamentally different from the goal of Thorburn within his port ring or in his analysis of land and water routes in Sweden. Finding solutions to the same old problems require new technologies, new channel leaders and new attitudes. The port economist can provide the vision and the tools to bring this about. The environment for the port economist today has great opportunities and challenges. More data are available today than previously. Also, there is greater acceptability of surveys to gather data than formerly. The greater availability of data is one reason that the application of production frontier methods is possible. It is a big step towards the assessment of performance in ports. However, the scope of the challenges is more complex and wider. Major challenges are presented to the port economist by the shift of companies to logistics strategies and the growth of the port/landside interface as a

The Evolution and Challenges of Port Economics

35

bottleneck. Has anybody yet really tried to tackle the urban container logistics problem with simulation or any other tool? As the problem of port logistics has widened, so has the participation of the port authority and other port interests in off-dock and, in that sense, non-traditional activities. On the other hand, the role of private enterprise in the traditional port functions is now large. These changes raise new questions of competition policy and of the values that should be placed on assets used for port activities. The value shippers place on service attributes has changed as timesavings, reliability and flexibility have become more important. The criteria for the design and selection of optimal port and related services have become more complex. Examination of the evolution of port economics shows that the topics investigated reflect more the issues and challenges of the day rather than the investigation of issues in economics in the port environment. Adding the latter approach may lead to more attention to different issues or, at least, add new approaches to issues that have been given little attention. An example may be the question raised by Goss about economic rents in ports. Finally, although the development of port economics is consistent with the progressive specialization in the field of economics, there is still value to be found in studies that span across applications. I have not yet seen many attempts to examine port and airport problems in an integrated way.6

NOTES 1. Owens may have erred in referring only to a ‘‘country’s trade’’ as competition among ports to serve the Rhine hinterland grew as the colony-related trade became less significant. 2. National organizations were becoming common in maritime technology to advance ship design and safety, for example, the Royal Institution of Naval Architects in the UK formed in 1860 and the Maritime Research Institute Netherlands founded in 1926. Ship testing organisations were often the forerunner of research institutes, as with the National Maritime Research Institute of Japan, which has its origin in 1916. 3. I wish to thank Sydney Gilman for his comments on developments in maritime economics at this and more recent times. 4. The concept of clusters of economic activity can be thought of more broadly than just related to the port and related logistics activities. It can be applied widely to marine activities (De Langen, 2002). 5. As a one-time geographer myself, I know many geographers contributing to port economics. I hope they forgive me for only mentioning one. 6. There is a Port and Airport Research Institute in Japan which is an independent administrative institution. But its purpose is to contribute to the efficient and smooth construction of ports and airports.

36

TREVOR HEAVER

REFERENCES Artur D. Little Ltd. (1970). Transhipment in the seventies – a study of container transport. Report to the National Ports Council. Summarised in Research and Technical Bulleting, National Ports Council, 6, 259–275. Atkin, R., & Rowlinson, M. (2000). Competition in ship handling: A study of market turbulence in North European harbour towage. Maritime Policy and Management, 27(3), 267–281. Australian Productivity Commission. (2003). International Benchmarking of Container Stevedoring, Commission Research Paper, July. Baird, A. J. (2000). Port privatisation: Objectives, extent, process and the UK experience. International Journal of Maritime Economics, 2(2), 177–194. Baird, A. J. (2002a). The economics of transhipment. In: C. Th. Grammenos (Ed.), The handbook of maritime economics and business (pp. 832–859). London: Informa UK Ltd. Baird, A. J. (2002b). Privatization trends at the world’s top-100 container ports. Maritime Policy and Management, 29(3), 271–284. Baird, A. J. (2004). Public goods and the public financing of major European seaports. Maritime Policy and Management, 31(4), 375–391. Beesley, M. E., & Foster, C. D. (1963). Estimating the social benefit of constructing an underground railway in London. Journal of the Royal Statistical Society, Series A (General), 126(Part I), 46–93. Bennathan, E., & Walters, A. A. (1979). Port pricing and investment policy for developing countries. Oxford: Oxford University Press. Berglund, A. (1931). Ocean transportation. New York: Longmans, Green and Co. Bichou, K. (2004). The ISPS code and the cost of port compliance: An initial logistics and supply chain framework for port security assessment and management. Maritime Economics and Logistics, 6(4), 322–348. Bichou, K., & Gray, R. (2004). A logistics and supply chain a management approach to port performance measurement. Maritime Policy and Management, 31(1), 41–67. Biles, J. (1923). Proceedings of the Permanent International Association of Navigation Congresses. Paper No. 33. Bird, J. (1971). Seaports and seaport terminals. London: Hutchinson University Press. Brooks, M. R. (2004). The governance structure of ports. In: W. K. Talley, (Ed.), The industrial organization of shipping and ports. Review of Network Economics, Special Issue, III(2), 168–183. Carbone, V., & De Martins, M. (2003). The changing role of ports in supply chain management: An empirical analysis. Maritime Policy and Management, 30(4), 305–320. Chang, S. (1978). Production function, productivities and capacity utilisation in the port of Mobile. Maritime Policy and Management, 5(3), 297–305. Cullinane, K. (2002). The productivity and efficiency of ports and terminals: Methods and applications. In: C. Th. Grammenos (Ed.), The handbook of maritime economics and business (pp. 803–831). London: Informa UK Ltd. Cullinane, K., & Song, D.-W. (2001). The administrative and ownership structure of Asian container ports. International Journal of Maritime Economics, 3(2), 175–197. Cullinane, K., & Song, D.-W. (2002). Port privatisation: Principles and practice. Transport Reviews, 22, 55–75.

The Evolution and Challenges of Port Economics

37

Cullinane, K., Song, D.-W., Ji, P., & Wang, T.-F. (2004). An application of DEA windows analysis to container port production efficiency. In: W. K. Talley, (Ed.), The industrial organization of shipping and ports. Review of Network Economics (Special Issue), III(2), 184–206. De Langen, P. W. (2002). Clustering and performance: The case of maritime clustering in the Netherlands. Maritime Policy and Management, 29(3), 209–221. De Langen, P. W. (2004). Governance in seaport clusters. Maritime Economics and Logistics, 6(2), 141–156. De Souza, G. A., Beresford, A. K. C., & Petit, S. J. (2003). Liner shipping companies and terminal operators: Internationalisation or globalisation? Maritime Economics and Logistics, 5, 393–412. Dowd, T. J., & Leschine, T. M. (1990). Container terminal productivity: A perspective. Maritime and Policy Management, 17(2), 107–112. Eckstein, O. (1958). Water and resource development: The economics of project evaluation. Cambridge, MA: Harvard University Press. Evans, L. (1993). Transport, economics and economists: Adam Smith, George Stigler et al.,. International Journal of Maritime History, V(1), 203–219. Everett, S. (2003). Corporatization: A legislative framework for port inefficiencies. Maritime Policy and Management, 30(3), 211–219. Ferguson, A. R., Lerner, E. M., McGee, J. S., Oi, W. Y., Rapping, L. A., & Sobotka, S. P. (1961). The economic value of the United States merchant marine. Evanston, IL: Transportation Center, Northwestern University. Fleming, D. K., & Baird, A. J. (1999). Some reflections on port competition in the United States and Western Europe. Maritime Policy and Management, 26(4), 383–394. Flor, L., & Defilippi, E. (2003). Port infrastructure: An access model for the essential facility. International Journal of Maritime Economics, 5(2), 116–132. Gilman, S. (1999). The size economies and network efficiency of large containerships. International Journal of Maritime Economics, 1(1), 39–59. Gilman, S. (2003). Sustainability and national policy in UK port development. Maritime Policy and Management, 30(4), 275–292. Gilman, S., Maggs, R., & Ryder, S. C. (1977). Containers on the North Atlantic: An economic analysis of ships and routes. Liverpool: Marine Transport Centre, University of Liverpool. Gilman, S., & Williams, G. F. (1976). Multi port itineraries for container ships. Journal of Transport Economics and Policy, X(2), 137–149. Goss, R. O. (1967a). Towards an economic appraisal of port investments. Journal of Transport Economics and Policy, 1(3), 249–272. Goss, R. O. (1967b). The turnaround of cargo liners and its effect on sea transport costs. Journal of Transport Economics and Policy, 1(1), 75–89. Goss, R. O. (1999). On the distribution of economic rent in seaports. International Journal of Maritime Economics, 1(1), 1–10. Goss, R. O. (2000). Economic policies and seaports: (1) The economic functions of seaports. Maritime Policy and Management, 17(3), 207–219. (2) The diversity of port policies. Maritime Policy and Management, 17(3), 221–234. (3) Are port authorities necessary? Maritime Policy and Management, 17(3), 257–271. (4) Strategies for port authorities. Maritime Policy and Management, 17(3), 273–287. Goss, R. O. (2002). An early history of maritime economics. International Journal of Maritime Economics, 4, 390–404.

38

TREVOR HEAVER

Goss, G. O., & Jones, C. D. (1970). The economics of size in dry bulk carriers. London: HMSO. Goss, G. O., & Stevens, H. (2001). Marginal cost pricing in ports. International Journal of Maritime Economics, 3(2), 128–138. Grammenos, C. Th. (Ed.) (2002). The handbook of maritime economics and business. London: Informa UK Ltd. Hamilton, C. (1999). Measuring container port productivity: The Australian experience. The Australia Institute, Background Paper No. 17, p. 16. http://www.tai.org.au/Publications_Files/Papers&Sub_Files/BP17.pdf Haralambides, H. E., Cariou, P., & Benacchio, M. (2003). Costs, benefits and pricing of dedicated container terminals. Maritime Economics and Logistics, 4(1), 21–34. Haralambides, H. E., & Veenstra, A. W. (2002). Port pricing. In: C. Th. Grammenos (Ed.), The handbook of maritime economics and business (pp. 782–802). London: Informa UK Ltd. Hayuth, Y. (1993). Port competition and regional cooperation. In: G. Blauwens, G. De brabander & E. Van de Voorde (Eds), De dynamiek van een haven (pp. 210–226). Uitgeverij Pelckmans: De Nederlandsche Boekhandel. Heaver, T. D. (1968). The economics of vessel size: A study of shipping costs and their implication for port investment. Ottawa: National Harbours Board. Heaver, T. D. (1970). The cost of large vessels – an examination of the sensitivity of total vessel costs to certain operating conditions. Research and Technical Bulletin (Vol. 7, pp. 342–356). London: National Ports Council. Heaver, T. D. (1993). The many facets of maritime economics. Inaugural address to the association of maritime economists. Maritime Policy and Management, 20(2), 121–132. Heaver, T. D. (1995). The implications of increased competition for port policy and management. Maritime Policy and Management, 22(2), 125–133. Heaver, T. D. (2002). Supply chain and logistics management: Implications for liner shipping. In: C. Th. Grammenos (Ed.), The handbook of maritime economics and business (pp. 375–396). London: Informa UK Ltd. Heaver, T. D., Fitzgerald, R. H., & Frounfelker, T. (2002). The effects of local taxes and subsidies on port competition in the Pacific Northwest of North America, Proceedings of the international association of maritime economists, Panama, (p. 16). Heaver, T. D., Meersman, H., & Van de Voorde, E. (2000). Do mergers and alliances influence European shipping and port competition? Maritime Policy and Management, 27(4), 363–373. Heaver, T. D., Meersman, H., & Van de Voorde, E. (2001). Co-operation and competition in container transport: Strategies for ports. Maritime Policy and Management, 28(3), 293–305. Heaver, T. D., & Studer, K. R. (1972). Ship size and turn-around time – some empirical evidence. Journal of Transport Economics and Policy, VI(1), 32–50. Heaver, T. D., & Tretheway, M. W. (2002). The economics and politics of taxation and subsidies for ports and airports, Proceedings of the international association of maritime economists, Panama, (p. 12). Heggie, I. G. (1974). Charging for port facilities. Journal of Transport Economics and Policy, VIII(1), 3–25. Hoffman, J. (2001). Latin American ports: Results and determinants of private sector participation. International Journal of Maritime Economics, 3(2), 221–241. Hough, B. O. (1924). Ocean traffic and trade. Chicago: LaSalle Extension University. Ircha, M. C. (2001). North American port reform: The Canadian and American experience. International Journal of Maritime Economics, 3(2), 198–220. Jansson, J. O., & Shneerson, D. (1982). Port economics. Cambridge, MA: MIT Press.

The Evolution and Challenges of Port Economics

39

Juhel, M. H. (2001). Globalisation, privatisation and restructuring of ports. International Journal of Maritime Economics, 3(2), 139–174. Johnson, K. M., & Garnett, H. C. (1971). The economics of containerisation. London: George Allen & Unwin. Kim, M., & Sachish, A. (1986). The structure of production, technical change and productivity in a port. Journal of Industrial Economics, 35, 209–223. Leggate, H., McMconville, J., & Morvilllo, A. (2005). International maritime transport; perspectives. London: Routledge. Lirn, T. C., Thanopoulou, H. A., Beynon, M. J., & Beresford, A. K. C. (2004). An application of AHP on transhipment port selection: A global perspective. Maritime Economics and Logistics, 6(1), 70–91. Luo, M., & Grigalunas, T. A. (2003). A spatial-economic multimodal transportation simulation model for US coastal container ports. Maritime Economics and Logistics, 5(2), 158–178. Malchow, M., & Kanafani, A. (2001). A disaggregate analysis of factors influencing port selection. Maritime Policy and Management, 28(3), 265–277. Martin, J., & Thomas, B. J. (2001). The container community. Maritime Policy and Management, 28(3), 279–292. Martinez-Budria, E., Diaz-Armas, R., Navarro-Ibanez, M., & Ravelo-Mesa, T. (1999). A study of the efficiency of spanish port authorities using data envelopment analysis. International Journal of Transport Economics, XXVI, 37–53. McConville, J. (1999). The economics of maritime transport. London: Witherby & Co. Ltd. McDowell, C. E., & Gibbs, H. M. (1954). Ocean transportation. New York: McGraw-Hill. McKinsey, & Co. (1967). Containerization: The key to low-cost transportation. Report to the British Transport Docks Board. Meersman, H., Van de Voorde, E., & Vanelslander, T. (2005). Ports as hubs in the logistics chain. In: H. Leggate, J. McConville & A. Morvillo (Eds), International maritime transport: Perspectives (pp. 137–144). Oxford: Routledge. Metcalfe, L. V. (1959). The principles of ocean transportation. New York: Simmons-Boardman Publishing Corporation. Metaxas, B. N. (1983). Maritime economics: Problems and challenges. Maritime Policy and Management, 10(3), 145–164. Meyer, J. R., Peck, M. J., Stenason, J., & Zwick, C. (1959). The economics of competition in the transportation industries. Cambridge, MA: Harvard University Press. Midoro, R., Musso, E., & Parola, F. (2005). Maritime liner shipping and the stevedoring industry: Market structure and competition strategies. Maritime Policy and Management, 32(2), 89–106. Mohring, H., & Harwitz, M. (1962). Highway benefits: An analytical framework. Evanston, IL: Northwestern University Press. Nir, A.-S., Lin, K., & Liang, G.-S. (2003). Port choice behaviour – from the perspective of the shipper. Maritime Policy and Management, 33(2), 165–173. Notteboom, T. E. (2002). Consolidation and contestability in the European container handling industry. Maritime Policy and Management, 29(3), 257–269. Notteboom, T. E. (2004). Container shipping and ports: An overview. In: W. K. Talley (Ed.), The industrial organization of shipping and ports. Review of Network Economics, Special Issue, 3(2), 86–106. Notteboom, T. E., & Winkelmans, W. (2001a). Reassessing public sector involvement in European seaports. International Journal of Maritime Economics, 3(2), 242–259.

40

TREVOR HEAVER

Notteboom, T. E., & Winkelmans, W. (2001b). Structural changes in logistics: How will port authorities face the challenge? Maritime Policy and Management, 28(1), 78–91. O’Loughlin, C. (1967). The economics of sea transport. Oxford: Pergamon Press. Owen, D. (1914). Ocean trade and shipping. Cambridge, MA: University Press. Pallis, A. A. (1997). Towards a common ports policy? EU-proposals and the ports industry’s perceptions. Maritime Policy and Management, 24(4), 365–380. Palmer, S. (1990). Book Review of L. M., Akveld, & J. R. Bruijn (Eds), Shipping companies and port authorities in the nineteenth and twentieth centuries: Their common interests in the development of port facilities. The Hague: Nederlandse Vereniging voor Zeegeschiedenis, 1989; in International Journal of Maritime History, II(2) 266–269. Palmer, S. (2003). Port economics in an historical context: The nineteenth century port of London. International Journal of Maritime History, XV(1), 27–67. Psaraftis, H. N. (2005). EU ports policy: Where do we go from here? Maritime Economics and Logistics, 7(1), 73–82. Robinson, R. (1978). Size of vessels and turnaround time: Further evidence from the port of Hong Kong. Journal of Transport Economics and Policy, XII(2), 161–178. Robinson, R. (2002). Ports as elements in value-driven chain systems: The new paradigm. Maritime Policy and Management, 29(3), 241–255. Roll, Y., & Hayuth, Y. (1993). Port performance comparison applying data envelopment analysis (DEA). Maritime Policy and Management, 20(2), 153–161. Slack, B. (2003). Pawns in the game: Ports in a global transportation system. Growth and Change, 24, 579–588. Song, D.-W. (2003). Port co-opetition in concept and practice. Maritime Policy and Management, 30(1), 29–44. Stopford, M. (1988). Maritime economics (2nd edn., 1997). London: Unwin Hyman. Strandenes, S. P. (2004). Port pricing structures and ship efficiency. In: W. K. Talley, (Ed.), The industrial organization of shipping and ports. Review of Network Economic, Special Issue, III(2), 135–144. Sturmey, S. G. (1962). British shipping and world competition. London: The Athlone Press. Suykens, F. (1985). Administration and management at the port of Antwerp. Maritime Policy and Management, 12(3), 181–194. Svendsen, A. S. (1958). Sea transport and shipping economics. Bremen: Institute for Shipping Research (Editor for Contributions in International Shipping Research, Gustav, A. Theel). Talley, W. K. (2004). Wage differentials of intermodal transportation carriers and ports: Deregulation versus regulation. In: W. K. Talley (Ed.), The industrial organization of shipping and ports. Review of Network Economics, Special Issue, III(2), 207–227. Thorburn, T. (1960). Supply and demand of water transport. Stockholm: The Business Research Institute at the Stockholm School of Economics. Turner, H. S. (2000). Evaluating seaport policy alternatives: A simulation study of terminal leasing policy and system performance. Maritime Policy and Management, 27(3), 283–301. United Nations. (1967). The turn-around time of ships in port. New York: United Nations ST/ECA/67. Walters, A. A. (1976). Marginal cost pricing in ports. The Logistics and Transportation Review, 12(3), 99–144. Wang, J. J. (2004). Regional governance of port development in China: A case study of Shanghai international shipping center. Maritime Policy and Management, 31(4), 357–373.

The Evolution and Challenges of Port Economics

41

Wellington, A. M. (1887). The economic theory of the location of railways (6th ed., 1914). New York: Wiley. Wijnolst, N., & Wergeland, T. (1996). Shipping. Delft: Delft University Press. Winston, C. (1985). Conceptual developments in the economics of transportation: An interpretive survey. Journal of Economic Literature, XXIII, 57–94. Yap, W. Y., & Lam, J. S. L (2004). An interpretation of inter-container port relationships from the demand perspective. Maritime Policy and Management, 31(4), 337–355. Zannetos, Z. S. (1966). The theory of tankship rates: An economic analysis of tankship operations. Cambridge, MA: MIT Press.

This page intentionally left blank

42

AN ECONOMIC THEORY OF THE PORT Wayne K. Talley ABSTRACT This chapter presents an economic theory of the port that considers both the demand for and cost incurred for the two-cargo (bulk and container) throughput of the port. Port-generalized prices include port charges and time prices incurred by ocean carriers’, inland carriers’ and shippers’ ships, vehicles and cargoes, respectively. The theoretical port time, resource and cost functions may be used in empirical port studies to investigate determinants of and their effects on the times in port of ships, vehicles and cargoes; port resource utilization; and port costs. The means by which a port can differentiate its service (or operating options) include ship and vehicle loading/unloading service rates, channel and berth accessibility and reliability, entrance and departure gate reliability, and damage and property losses to ships, vehicles and cargoes in port. Alternatively, these operating options may be used as performance indicators to evaluate the performance of a port with respect to a port’s economic objective such as maximizing throughput subject to a minimum profit constraint.

Port Economics Research in Transportation Economics, Volume 16, 43–65 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16003-5

43

44

WAYNE K. TALLEY

1. INTRODUCTION A port is a place that provides for the transfer of cargo and/or passengers between waterways and shores. Alternatively, it is an intermodal node in the transportation network, where cargo and/or passengers change modes of transportation (e.g., from a ship to an inland transport mode and vice versa). Cargo ports handle general and bulk cargoes. General cargoes are either goods of various sizes and weights shipped as packaged cargo or goods of uniform sizes and weights shipped as loose (non-packaged) cargo. The former may be container or breakbulk cargo and the latter may be neobulk cargo. Container cargo is stored in standardized boxes (containers), generally 20 or 40 ft in length without wheels – i.e., as one twenty-foot equivalent unit (TEU) or as one forty-foot equivalent unit (FEU). Breakbulk cargo is general cargo that is packaged on pallets or in wire or rope slings for lifting on and off a ship. Neobulk cargo, for example, includes automobiles, steel and lumber. Bulk (dry and liquid) cargoes include goods neither packaged nor of uniform sizes and weights. Drybulk cargo, for example, includes coal and grains. Crude oil and refined petroleum products are examples of liquidbulk cargo. This chapter presents an economic theory of the cargo port. It is assumed that the port handles two types of cargo, bulk and container, and has the economic objective of maximizing annual throughput (of bulk and container cargoes) subject to a minimum profit constraint1. If the port is owned by government, this profit constraint may be zero (where port revenue equals cost) or a maximum deficit (where port revenue is less than cost) that is to be subsidized by government. In the following section, the basic economic model of the port is presented, followed by a discussion of port operating options. Then, a discussion of extending the model to incorporate port ship and vehicle congestion is presented, followed by a discussion of port cost efficiency. Then, the applicability of the theory to port performance evaluation and empirical research is discussed. Finally, a summary is presented.

2. THE BASIC PORT ECONOMIC MODEL Assume a cargo port that handles bulk and container cargoes and its economic objective is to maximize annual throughput subject to a minimum

45

An Economic Theory of the Port

profit constraint2. The annual demand for the port’s services in handling bulk cargo is represented by annual bulk throughput Nb, measured by the annual number of bulk cargo movements through the port. The annual demand for the port’s services in handling container cargo is represented by annual container throughput Nc, measured by the annual number of container cargo movements through the port. The annual demands for bulk and container throughputs are functions of their generalized prices (i.e., money and time prices).3 N b ¼ N b ðGPb Þ

(1)

N c ¼ N c ðGPc Þ

(2)

where GPb and GPc represent the port’s generalized prices per unit of bulk and container throughputs, respectively. The port’s generalized price is the sum of the port’s money prices (per unit of throughput) charged for various services rendered and port time prices incurred by ocean carriers, inland carriers and shippers per unit of throughput.4 Specifically, the port’s generalized price for bulk throughput (GPb) may be expressed as .X X XX GPb ¼ N sb Ppb þ V sb T sbj XX X X X þ V ib T ibj = N ib þ V hb T hb = N hb ð3Þ

where Ppb ¼ port charge per unit of bulk throughput for the pth port service. Vsb ¼ value of time per unit of time incurred in port by the sth ocean bulk carrier. Tsbj ¼ annual total time in port (i.e., turnaround time) incurred by the sth ocean bulk carrier during the jth port call. Nsb ¼ port annual bulk throughput incurred by the sth ocean bulk carrier. Vib ¼ value of time per unit of time incurred in port by the ith inland bulk carrier. Tibj ¼ annual total time in port (i.e., turnaround time) incurred by the ith inland bulk carrier during the jth port call. Nib ¼ port annual bulk throughput incurred by the ith inland bulk carrier. Vhb ¼ value of time per unit of time incurred in port by the bulk cargo of the hth shipper.5 Thb ¼ annual total time that the bulk cargo of the hth shipper is in port. Nhb ¼ port annual bulk throughput incurred by the hth bulk shipper.

46

WAYNE K. TALLEY

Similarly, the port’s generalized price for container throughput (GPc) may be expressed as X XX X GPc ¼ Ppc þ V sc T scj = N sc XX X X þ V ic T icj = N ic þ V hc T hc =N hc ð4Þ

The symbols are defined the same as above, but now with respect to container throughput. The first terms to the right of the equality sign in Eqs. (3) and (4) are money prices per unit of throughput for services rendered by the port. The second terms to the right of the equality sign are the annual time prices per unit of throughput incurred by ocean carriers in port. Similarly, the third and fourth terms to the right of the equality sign in both equations are the annual time prices per unit of throughput incurred by inland carriers and shippers in port, respectively. The time prices reflect the port opportunity costs of ocean carriers, inland carriers and shippers.6 The port’s minimum profit constraint equation may be expressed as X X X M¼ Ppb N b þ Ppc N c Rrb C r X X Rrc C r Rrbc C r r ¼ 1; 2; . . . ; U ð5Þ

where M ¼ annual minimum profit of the port. Rrb ¼ annual amount of the port’s rth resource utilized by bulk throughput. Rrc ¼ annual amount of the port’s rth resource utilized by container throughput. Rrbc ¼ annual amount of the port’s rth resource utilized (i.e., shared) by both bulk and container throughputs. Cr ¼ annual cost incurred by the port per unit of the rth resource. The remaining variables are the same as defined previously. The first and second terms of Eq. (5) to the right of the equality sign represent the annual total revenue of the port from bulk and container throughputs, respectively. The third and fourth terms are annual total direct costs (i.e., attributable to only one type of cargo) incurred by the port for bulk and container throughputs, respectively. The fifth term represents the annual indirect (i.e., attributable to both throughputs) costs incurred by the port, i.e., costs that are shared by both bulk and container throughputs. For

47

An Economic Theory of the Port

a private port, ‘‘M’’ is some positive value; for a public port, it is zero or negative (i.e., a deficit that is subsidized). The port is assumed to be technically efficient, i.e., its throughput is the maximum throughput that can be handled by its given level of resources.7 This relationship is exhibited by its production function.8 The port’s production function for bulk throughput may be expressed as N b ¼ f b ðRrb ; Rrbc Þ

r ¼ 1; 2; . . . ; U

(6)

The port’s production function for container throughput may be expressed as N c ¼ f c ðRrc ; Rrbc Þ

r ¼ 1; 2; . . . ; U

(7)

Substituting the production functions for Nb and Nc in profit constraint Eq. 5 and rewriting, the constraint equation becomes X X M¼ Ppb f b ðRrb ; Rrbc Þ þ Ppc f c ðRrc ; Rrbc Þ X X X Rrb C r Rrc C r Rrbc C r r ¼ 1; 2; . . . ; U ð8Þ For the purpose of optimization, the port’s economic objective of maximizing annual throughput (i.e., maximizing the demand for its throughput exhibited by demand functions 1 and 2) subject to the minimum profit constraint (Eq. 8) may be expressed as the following Lagrangean equation. hX L ¼ N b ðGPb Þ þ N c ðGPc Þ þ d Ppb f b ðRrb ; Rrbc Þ X X X þ Ppc f c ðRrc ; Rrbc Þ Rrb C r Rrc C r i X Rrbc C r M r ¼ 1; 2; . . . ; U ð9Þ

where d is the Lagrangean multiplier variable. Substituting Eqs. (3) and (4) for GPb and GPc in Eq. (9) and rewriting, the Lagrangean equation becomes  XX X XX X X X V sb T sbj = N sb þ V ib T ibj = N ib þ V hb T hb = N hb Ppb þ X  XX X XX X X X Ppc þ V sc T scj = N sc þ V ic T icj = N ic þ V hc T hc = N hc þ Nc hX X þd Ppb f b ðRrb ; Rrbc Þ þ Ppc f c ðRrc ; Rrbc Þ i X X X Rrb C r Rrc C r Rrbc C r M r ¼ 1; 2; . . . ; U ð10Þ

L ¼ Nb

X

What are the choice variables to be utilized by port management in the optimization of Lagrangean function (10)? For a variable to qualify as a choice variable, its value must be under the control of port management.

48

WAYNE K. TALLEY

Further, in order to optimize function (10), the choice variables must appear in both the objective function and constraint components of function (10). The variables that satisfy these two restrictions with respect to Lagrangean function (10) are the port’s prices (or charges) for bulk and container throughputs and its operating options.

3. PORT OPERATING OPTIONS A transportation carrier (e.g., shipping line or railroad) can differentiate the quality of its transportation service by varying the levels of its operating options (Talley, 1988). An operating option is the means by which a transportation carrier can differentiate its service. Carrier operating options include speed of movement, frequency of service, reliability of service, susceptibility to loss and damage and spatial accessibility of service. The greater the speed of movement for shipments, the higher the quality of transportation service, since shipments arrive at their destinations within a shorter period of time. Frequency of service is how often the service is provided; with higher frequencies, transportation service will be available more often (thereby the higher the quality of service). Reliability of service is the degree to which shipments arrive at their destinations at the specific time stated; the higher the reliability, the higher the quality of service. Susceptibility to loss and damage is the probability (or likelihood) that shipments will be damaged or lost by the carrier; the higher the probability, the lower the quality of service. Spatial accessibility of service is the spatial convenience of transportation service; the higher the degree of accessibility, the higher the quality of service. A port has similar operating options with which to differentiate the quality of its service. Operating options for the port that are analogous to that of speed of movement for a carrier are port loading and unloading service rates (i.e., cargo loaded and unloaded per unit of time) for ships of ocean carriers and vehicles of inland carriers. These rates with respect to bulk cargo may be expressed as SLb ¼ average ship loading (of bulk cargo) service rate for ships in port during the year (i.e., tons of bulk cargo loaded on ships per hour of loading time). SUb ¼ average ship unloading (of bulk cargo) service rate for ships in port during the year (i.e., tons of bulk cargo loaded from ships per hour of unloading time).

An Economic Theory of the Port

49

VLb ¼ average vehicle loading (of bulk cargo) service rate for vehicles of inland carriers in port during the year (i.e., tons of bulk cargo loaded on vehicles per hour of loading time). VUb ¼ average vehicle unloading (of bulk cargo) service rate for vehicles of inland carriers in port during the year (i.e., tons of bulk cargo loaded from vehicles per hour of unloading time). The corresponding service rates with respect to container cargo are SLc, SUc, VLc and VUc and are defined as above except with respect to container cargo.9 The annual total time in port incurred by the sth ocean carrier in the transport of bulk cargo during the jth port call (Tsbj) is affected by ship loading and unloading service rates SLb and SUb, as well as by the average ship arrival and departure waiting times SAWb and SDWb. SAWb ¼ average arrival waiting time (in hours) for bulk ships to be berthed in the port during the year – the arrival waiting time is the time interval between the point in time when a ship arrives at the entrance of the port’s channel and the point in time when it is finally berthed. SDWb ¼ average departure waiting time (in hours) for bulk ships to be unberthed in the port during the year – the departure waiting time is the time interval between the point in time when the ship seeks to begin the unberthing process and the point in time when the ship departs the port’s channel. The time Tsbj is also affected by port channel and berth operating options that are analogous to the spatial accessibility and reliability of service options of transportation carriers (discussed above) – i.e., port channel and berth accessibility (or spatial convenience) and reliability options PCA ¼ port channel accessibility – the average daily percent of time during the year that the port’s channel adheres to authorized depth and width dimensions. PBA ¼ port berth accessibility – the average daily percent of time during the year that the port’s berth adheres to authorized depth and width dimensions. PCR ¼ port channel reliability – the average daily percent of time during the year that the port’s channel is open to navigation. PBR ¼ port berth reliability – the average daily percent of time during the year that the port’s berth is open to the berthing of ships. Finally, the time Tsbj is affected by the port channel’s variability in water level.

50

WAYNE K. TALLEY

TIDE ¼ port channel variability in water level (or tide variability) – the average daily range in feet of water of the port’s channel during the year. Thus, the function for Tsbj may be expressed as T sbj ¼ T sbj ðSLb ; SUb ; SAWb ; PCA; PBA; PCR; PBR; TIDEÞ

(11)

Similarly, the Tscj time function for container cargo may be expressed as T scj ¼ T scj ðSLc ; SUc ; SAWc ; SDWc ; PCA; PBA; PCR; PBR; TIDEÞ

(12)

The symbols are defined the same as above except now with respect to container cargo. The average arrival waiting time for bulk ships (SAWb) in turn may be expressed as SAWb ¼ SAWb ðSARb ; SSRb Þ

(13)

where SARb ¼ average arrival rate (in ships per day) for bulk ships for the port during the year. SSRb ¼ average service rate (in ships per day) for bulk ships in the port during the year (i.e., the number of bulk ships that are serviced – cargo loaded and/or unloaded – each day at the port’s berths). Similarly, the average arrival waiting time for container ships (SAWc) may be expressed as SAWc ¼ SAWc ðSARc ; SSRc Þ

(14)

where SARc and SSRc are defined the same as above except now with respect to container cargo. The ship service rates (SSRb and SSRc) are affected by the port’s operating options – ship loading and unloading service rates, i.e., SSRb ¼ SSRb ðSLb ; SUb Þ

(15)

SSRc ¼ SSRc ðSLc ; SUc Þ

(16)

By similar reasoning, the ship departure waiting times SDWb and SDWc may be expressed as SDWb ¼ SDWb ðSDRb ; SSRb Þ

(17)

SDWc ¼ SDWc ðSDRc ; SSRc Þ

(18)

These functions differ from the above ship arrival waiting time functions in that the ship departure rates SDRb and SDRc replace the ship arrival rates.

An Economic Theory of the Port

51

SDRb and SDRc represent the average departure rates (in ships per day) for bulk and container ships, respectively, for the port during the year. By substituting function (15) in function (13) and then substituting in function (11) and also substituting function (15) in function (17) and then substituting in function (11) and rewriting, the function for Tsbj may now be expressed as T sbj ¼ F sbj ðSLb ; SUb ; PCA; PBA; PCR; PBR; SARb ; SDRb ; TIDEÞ

(19)

Note that Tsbj is now a function of port operating options and the SARb, SDRb and TIDE variables. Similarly, by substituting function (16) in function (14) and then substituting in function (12) and also substituting function (16) in function (18) and then substituting in function (12) and rewriting, the function for Tscj may now be expressed as T scj ¼ F scj ðSLc ; SUc ; PCA; PBA; PCR; PBR; SARc ; SDRc ; TIDEÞ

(20)

Here as well, Tscj is now a function of port operating options as well as the SARc, SDRc and TIDE variables. The annual total time in port incurred by the ith inland carrier in the transport of bulk cargo during the jth port call (Tibj) is affected by vehicle loading and unloading service rates VLb and VUb as well as by average vehicle arrival and departure waiting times VAWb and VDWb. VAWb ¼ average arrival waiting time (in hours) for bulk vehicles to enter the port during the year – the arrival waiting time is the time interval between the point in time when a vehicle arrives at the port’s entrance gate and the point in time when it finally enters the entrance gate. VDWb ¼ average departure waiting time (in hours) for bulk vehicles to depart the port during the year – the departure waiting time is the time interval between the point in time when the departure process for a vehicle begins and the point in time when it finally leaves the port through the departure gate. The time Tibj is also affected by the port’s operating options, entrance gate reliability and departure gate reliability. EGR ¼ entrance gate reliability – the average daily percent of time during the year that the port’s entrance gate is open for vehicles. DGR ¼ departure gate reliability – the average daily percent of time during the year that the port’s departure gate is open for vehicles.

52

WAYNE K. TALLEY

Thus, the function for Tibj may be expressed as T ibj ¼ T ibj ðVLb ; VUb ; VAWb ; VDWb ; EGR; DGRÞ

(21)

Similarly, the Ticj time function for the ith inland container carrier may be expressed as T icj ¼ T icj ðVLc ; VUc ; VAWc ; VDWc ; EGR; DGRÞ

(22)

The symbols are defined the same as above except now with respect to container cargo. The average arrival waiting time for bulk vehicles may be expressed as VAWb ¼ VAWb ðVARb ; VSRb Þ

(23)

where VARb ¼ average arrival rate (in vehicles per day) for bulk vehicles to the port during the year. VSRb ¼ average service rate (in vehicles per day) for bulk vehicles in the port during the year (i.e., the number of bulk vehicles that are serviced – cargo loaded and/or unloaded – each day in the port). Similarly, VAWc may be expressed as VAWc ¼ VAWc ðVARc ; VSRc Þ

(24)

where VARc and VSRc are defined the same as above except now with respect to container cargo. Further, the vehicle service rates, VSRb and VSRc, are affected by the port’s operating options – vehicle loading and unloading service rates, i.e., VSRb ¼ VSRb ðVLb ; VUb Þ

(25)

VSRc ¼ VSRc ðVLc ; VUc Þ

(26)

By similar reasoning, the vehicle departure waiting times VDWb and VDWc may be expressed as VDWb ¼ VDWb ðVDRb ; VSRb Þ

(27)

VDWc ¼ VDWc ðVDRc ; VSRc Þ

(28)

where VDRb and VDRc represent the port’s average vehicle departure rates (in vehicles per day) for bulk and container vehicles, respectively, during the year. By substituting function (25) in function (23) and then substituting in function (21) and substituting function (25) in function (27) and then

An Economic Theory of the Port

53

substituting in function (21) and rewriting, the function for Tibj may now be expressed as T ibj ¼ F ibj ðVLb ; VUb ; EGR; DGR; VARb ; VDRb Þ

(29)

Note that Tibj is now a function of port operating options and the VARb and VDRb variables. Similarly, by substituting function (26) in function (24) and then substituting in function (22) and substituting function (26) in function (28) and then substituting in function (22) and rewriting, the function Ticj may now be expressed as T icj ¼ F icj ðVLc ; VUc ; EGR; DGR; VARc ; VDRc Þ

(30)

Here as well, Ticj is now a function of port operating options and the VARc and VDRc variables. The annual total time Thb that bulk cargo of the hth shipper is in port is the sum of the time TShb it is aboard a ship in port, the time TVhb it is aboard an inland-carrier’s vehicle in port, the time TOhb it is in storage in port and the transit time TRhb it incurs to and from storage in port. The ship-related time of bulk cargo in port should be a function of the same variables that appear in the bulk ship’s port time function (19). Thus, the ship-related time TShb in port for bulk cargo shipped by the hth shipper may be expressed as TShb ¼ TShb ðSLb ; SUb ; PCA; PBA; PCR; PBR; SARb ; SDRb ; TIDEÞ (31) The vehicle-related time of bulk cargo in port should be a function of the same variables that appear in the bulk vehicle’s port time functions (29). Thus, the vehicle-related time TVhb in port for bulk cargo shipped by the hth shipper may be expressed as TVhb ¼ TVhb ðVLb ; VUb ; EGR; DGR; VARb ; VDRb Þ

(32)

The storage time in port for bulk cargo should be a function of ship and vehicle loading/unloading service rates and ship and vehicle arrival and departure rates. Thus, the time in storage TOhb in port for bulk cargo shipped by the hth shipper may be expressed as TOhb ¼ TOhb ðSLb ; SUb ; SARb ; SDRb ; VLb ; VUb ; VARb ; VDRb Þ

(33)

The transit time to and from storage in port for bulk cargo should be a function of ship and vehicle loading/unloading service rates. Thus, the transit time TRhb to and from storage in port for bulk cargo shipped by the hth shipper may be expressed as TRhb ¼ TRhb ðSLb ; SUb ; VLb ; VUb Þ

(34)

54

WAYNE K. TALLEY

The annual total time Thb that bulk cargo of the hth shipper is in port may be expressed as T hb ¼ TShb þ TVhb þ TOhb þTRhb

(35)

Similarly, the annual total time Thc that container cargo of the hth shipper is in port may be expressed as T hc ¼ TShc þ TVhc þ TOhc þ TRhcb

(36)

Port operating options are also expected to affect the values of time incurred in port by ocean carriers, inland carriers and shippers. Since these values reflect their opportunity costs in port, they will be affected by the port’s operating options of susceptibility of ocean carriers, inland carriers and shippers to damage and loss in port, i.e.,10 SDAM ¼ probability of damage (e.g., due to an accident) to ships in port for ocean carriers. SLOS ¼ probability of loss of ship property aboard ships in port for ocean carriers. VDAM ¼ probability of damage to vehicles in port for inland carriers. VLOS ¼ probability of loss of vehicles or vehicle property in port for inland carriers. CDAM ¼ probability of damage to cargo in port for shippers. CLOS ¼ probability of loss of cargo in port for shippers. The values of time incurred by bulk and container ocean carriers, inland carriers and shippers may be expressed as V sb ¼ V sb ðSDAM; SLOSÞ

(37)

V sc ¼ V sc ðSDAM; SLOSÞ

(38)

V ib ¼ V ib ðVDAM; VLOSÞ

(39)

V ic ¼ V ic ðVDAM; VLOSÞ

(40)

V hb ¼ V hb ðCDAM; CLOSÞ

(41)

V hc ¼ V hc ðCDAM; CLOSÞ

(42)

In addition to affecting the time and the value of time in port incurred by ocean carriers, inland carriers and shippers, the operating options of the port will also affect the amounts of resources utilized by the port. A resource function is defined as the relationship between the minimum amount of

55

An Economic Theory of the Port

a given resource employed by a port and the levels of its operating options and amounts of types of cargo received by the port (see Talley, 1988, p. 46). The resource function for the rth resource for handling bulk cargo (Rrb) may be expressed as ! SLb ; SUb ; VLb ; VUb ; PCA; PBA; PCR; PBR; EGR; DGR; Rrb ¼ Rrb SDAM; SLOS; VDAM; VLOS; CDAM; CLOS; BULK (43) where BULK is the tons of bulk cargo that the port receives during the year. The resource function for the rth resource for handling container cargo may be expressed as ! SLc ; SUc ; VLc ; VUc ; PCA; PBA; PCR; PBR; EGR; DGR; Rrc ¼ Rrc SDAM; SLOS; VDAM; VLOS; CDAM; CLOS; CONT (44) where CONT is the number of TEUs of container cargo that the port receives during the year. The resource function for the rth resource that is used in handling both bulk and container cargoes may be expressed as Rrbc ¼ Rrbc

SLb ; SUb ; VLb ; VUb ; SLc ; SUc ; VLc ; VUc ; PCA; PBA; PCR; PBR; EGR; DGR; SDAM; SLOS; VDAM; VLOS; CDAM; CLOS; BULK; CONT

!

(45) The port’s operating options that have appeared in this section are summarized in Table 1. In addition, the directions of change for the operating options to have a positive impact on the port’s economic objective of maximizing throughput subject to a minimum profit constraint are presented.

4. PORT CONGESTION Port congestion arises when port users interfere with one another in the utilization of port resources, thereby increasing their time in port. Heretofore, it was implicitly assumed that port berth congestion does not arise from bulk ships interfering with container ships (and vice versa) at the port’s berth(s) and port gate congestion does not arise from bulk vehicles interfering with container vehicles (and vice versa) at the port’s gate(s). Port congestion may be intentional or unintentional. Intentional port congestion may arise from preemptive priority, e.g., when a port grants

56

WAYNE K. TALLEY

Table 1.

Port Operating Options and Directional Changes for Maximizing its Economic Objective.

Operating Option

Directional Change

Loading Service Rate for Bulk Ships (SLb) Unloading Service Rate for Bulk Ships (SUb) Loading Service Rate for Container Ships (SLc) Unloading Service Rate for Container Ships (SUc) Loading Service Rate for Bulk Vehicles (VLb) Unloading Service Rate for Bulk Vehicles (VUb) Loading Service Rate for Container Vehicles (VLc) Unloading Service Rate for Container Vehicles (VUc) Port Channel Accessibility (PCA) Port Channel Reliability (PCR) Port Berth Accessibility (PBA) Port Berth Reliability (PBR) Entrance Gate Reliability (EGR) Departure Gate Reliability (DGR) Probability of Ship Damage (SDAM) Probability of Ship Property Loss (SLOS) Probability of Vehicle Damage (VDAM) Probability of Vehicle Property Loss (VLOS) Probability of Port Cargo Damage (CDAM) Probability of Port Cargo Loss (CLOS)

Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Decrease Decrease Decrease Decrease Decrease Decrease

priority to ships or vehicles transporting a certain type of cargo over ships or vehicles transporting another type of cargo. The berthing (unberthing) of a bulk ship may be delayed to allow for the berthing (unberthing) of an incoming (departing) containership. If so, the average arrival and departure waiting times of the delayed ships will increase and therefore their total time in port. Unintentional interference arises in the normal utilization of port resources. Port queuing (or waiting-time) costs are extreme congestion costs that arise when the demand for use of a port resource exceeds its supply.11

4.1. Ship Berth Congestion Bulk ships and containerships are not expected to use the same berths. However, if they seek to be berthed or unberthed at approximately the same time, port resources such as tugs and pilots may not be available in sufficient number for the berthing and unberthing. If so, one ship will have to wait until resources are deployed from servicing another ship. Alternatively, tugs

An Economic Theory of the Port

57

and pilots may be sufficient in number, but the width of the port’s channel is not sufficient to be traversed by more than one ship at a time. The possibility of ship berth congestion that arises when containerships and bulk ships interfere with one another in berthing may be incorporated into our model by including the average arrival (SARb) and service (SSRb) rates for bulk ships in the average arrival waiting time (SAWc) function (14) for containerships and including the average arrival (SARc) and service (SSRc) rates for containerships in the average arrival waiting time (SAWb) function (13) for bulk ships. The possibility of berth congestion that arises when containerships and bulk ships interfere with one another in unberthing may be incorporated into our model by including the average departure (SDRb) and service (SSRb) rates for bulk ships in the average departure waiting time (SDWc) function (18) for containerships and including the average departure (SDRc) and service (SSRc) rates for containerships in the average departure waiting time (SDWb) function (17) for bulk ships.

4.2. Ship Work Congestion In addition to berthing and unberthing of ships, port resources (e.g., labor and equipment) also may not be sufficient in number to prevent ship work congestion, i.e., congestion that arises, for example, when a bulk ship has to wait to be loaded/unloaded (or worked) until a containership has been loaded/unloaded. This situation is more likely to arise with labor than equipment, since equipment tends to be specialized in handling container and bulk cargoes. The possibility of ship work congestion may be incorporated into our model by including the average loading (SLb) and unloading (SUb) service rates of bulk ships in the containership total time function (12) and including the average loading (SLc) and unloading (SUc) service rates of containerships in the bulk ship total time function (11).

4.3. Vehicle Gate Congestion Unlike the situation for berths and ships, vehicles (truck and rail) transporting bulk and container cargoes may utilize the same entrance and departure gates. Trucks (railcars) transporting containers may have to wait for trucks (railcars) transporting bulk cargo to traverse the gates and vice versa, thus increasing their waiting times in port. This possibility may be incorporated into our model by including the average arrival (VARb) and service

58

WAYNE K. TALLEY

(VSRb) rates for bulk vehicles in the average arrival waiting time (VAWc) function (24) for container vehicles and including the average arrival (VARc) and service (VSRc) rates for container vehicles in the average arrival waiting time (VAWb) function (23) for bulk vehicles. For departures, the average departure (VDRb) and service (VSRb) rates for bulk vehicles are to be included in the average departure waiting time (VDWc) function (28) for container vehicles and the average departure (VDRc) and service (VSRc) rates for container vehicles are to be included in the average departure waiting time (VDWb) function (27) for bulk vehicles.

4.4. Vehicle Work Congestion In addition to vehicle gate congestion, vehicle congestion may also arise in the loading/unloading of vehicles. Specifically, port resources such as labor and equipment may not be of sufficient number for the simultaneous loading/unloading of container and bulk vehicles. If so, a bulk vehicle will have to wait until the container vehicle is loaded/unloaded and vice versa. This situation (as for ships) is more likely to arise with labor than with equipment (which tends to be cargo specific). This possibility may be incorporated into our model by including the average arrival (VAWb) and departure (VDWb) waiting times and the average loading (VLb) and unloading (VUb) service rates of bulk vehicles in the time function (22) for a container vehicle in port; and including the average arrival (VAWc) and departure (VDWc) waiting times and the average loading (VLc) and unloading (VUc) service rates of container vehicles in the time function (21) for a bulk vehicle in port.

5. COST EFFICIENCY In the basic model section, the port was assumed to be technically efficient, exhibited by production functions (6) and (7) for bulk and container throughputs, respectively. When the port’s economic objective is optimized (represented by Lagrangean Eq. 10), it must follow that the port is also cost efficient, since cost efficiency is a necessary condition for optimization of Eq. (10). If the port were cost inefficient, the optimization could not occur; if the port were cost inefficient and then becomes cost efficient, its profit would increase for a given throughput; hence, it could increase its throughput while adhering to its minimum profit constraint by becoming cost efficient.

59

An Economic Theory of the Port

The port’s long-run cost function (representing cost efficiency) can be derived from minimizing its costs (C) subject to throughput productionfunction constraints, i.e., X X X Minimize C ¼ Rrb C r þ Rrc C r þ Rrbc C r subject to N b ¼ f b ðRrb ; Rrbc Þ and N c ¼ f c ðRrc; Rrbc Þ

ð46Þ

Optimizing (46) via the operating options (or choice variables) found in resource functions (43), (44) and (45), it follows that the port’s long-run cost function may be expressed as12 C ¼ CðC r ; N b ; N c ; BULK; CONTÞ

r ¼ 1; 2; . . . ; U

(47)

Note that the output of the port is represented by the number of bulk and container cargo movements through the port and by the tons of bulk cargo and TEUs of container cargo received by the port. Port cost function (47) reveals that a port can incur separate costs with respect to the movement of cargo via Nb and Nc and the cargo itself via BULK and CONT (e.g., storage cost). Alternatively, the product of Nb and BULK (or tons of bulk cargo throughput) may be used as a measure of bulk cargo throughput and the product of Nc and CONT (or TEUs of container cargo throughput) may be used as a measure of container throughput. The product of the number of bulk (container) movements through the port and the tons (TEUs) of bulk (container) cargo received by the port, i.e., NbBULK (NcCONT), is analogous to the output measure, the ton-mile, used in surface transportation – i.e., the product of vehicle miles provided by a carrier in moving the tons of cargo provided by the shipper. Note that a transportation carrier cannot force a shipper to provide it with cargo to be shipped. Hence, in order for a transportation service to occur, e.g., tonmiles, both the carrier and the shipper must be willing participants, i.e., the carrier must be willing to provide a vehicle and thus vehicle miles to transport the shipper’s cargo and the shipper must be willing to provide the tons of cargo to be shipped by the carrier. By similar reasoning, a port cannot force a shipper, ocean carrier or inland carrier to use its port, i.e., to provide it with cargo. Hence, in order for a port service to occur, e.g., TEUs of container cargo throughput, the shipper, ocean carrier or inland carrier must be willing to provide TEUs of container cargo to the port and the port must be willing to provide facilities for movement of the cargo through the port.13 The port’s long-run cost function consisting of the throughputs, NbBULK and NcCONT, may be expressed as C ¼ C 0 ðC r ; N b BULK; N c CONTÞ

r ¼ 1; 2; . . . ; U

(48)

60

WAYNE K. TALLEY

The existence of port overall economies of scale based upon cost function (48) can be investigated by using the following equation: 

 S ¼ C ðN b BULKÞ @C=@N b BULK þ ðN c CONTÞ @C=@N c CONT (49)

If S4 1, the port exhibits overall economies of scale with respect to tons of bulk cargo and TEUs of container cargo throughputs and constant returns to and diseconomies of scale if equal to or less than one. If economies of scale exist, a proportional increase in bulk and container cargo throughputs will lead to a less than proportional increase in port costs. If constant returns to scale (diseconomies of scale) exist, the port’s costs increase in proportion (greater than in proportion) to the increase in bulk and container cargo throughputs. The existence of port product-specific economies of scale with respect to tons of bulk cargo throughput can be investigated by using the following equation: S b ¼ AICb =MCb

(50)

where AICb is the long-run average incremental cost of tons of bulk cargo throughput and MCb is the long-run marginal cost of tons of bulk cargo throughput. Based upon cost function (48), AICb and MCb may be formally stated as AICb ¼ ½C 0 ðC r ; N b BULK; N c CONTÞ MCb ¼ @C 0 =@N b BULK

C 0 ðC r ; 0; N c CONTފ=N b BULK ð51Þ

If Sb41, the port exhibits economies of scale with respect to tons of bulk cargo throughput and constant returns to and diseconomies of scale if equal to or less than one. If economies of scale exist, the port’s costs do not increase in proportion to the increase in tons of bulk cargo throughput. If constant returns to scale (diseconomies of scale) exist, the port’s costs increase in proportion (greater than in proportion) to the increase in tons of bulk cargo throughput. Similarly, the existence of port product-specific economies of scale with respect to TEUs of container cargo throughput can be investigated by using the following equation: S c ¼ AICc =MCc

(52)

The port will exhibit economies of scope in the provision of tons of bulk cargo and TEUs of container cargo throughputs if it can provide both throughputs more cheaply than if each level of a given throughput were provided separately. Based upon cost function (48), economies of scope will

An Economic Theory of the Port

61

exist if C 0 ðC r ; N b BULK; N c CONTÞoC 0 ðC r ; 0; N c CONTÞ þ C 0 ðC r ; N b BULK; 0Þ (53) If the inequality sign in (53) were reversed, diseconomies of scale would exist.14

6. APPLICABILITY OF THE THEORY The choice variables (i.e., variables whose values are under the control of port management) for optimizing the port’s economic objective have also been referred to as the port’s individual performance indicators (Talley, 1994). In this chapter, the specific choice variables (or individual performance indicators), port prices and operating options, were deduced for a port with the economic objective of maximizing throughput subject to a minimum profit constraint. The theory can also be used to deduce performance indicators for ports with alternative economic objectives. The values of individual performance indicators that optimize the port’s economic objective are the indicators’ standards (or benchmarks). If the actual values of the indicators approach (depart from) their prospective standards over time, the port’s performance with respect to its given economic objective has improved (deteriorated) over time. If the indicator standards are not known, the port’s performance can still be evaluated by using the actual values of its performance indicators: If the direction of movement in indicator values (see Table 1) over time moves the port nearer to (away from) achieving its economic objective, the conclusion is that the port’s performance is improving (or deteriorating) over time. For one indicator, a rising trend over time in its actual values may move the port nearer to achieving its economic objective; for another indicator, it may be a declining trend in its actual values. An advantage to port management in having individual performance indicators (as found in Table 1) for evaluating port performance over time is that it can evaluate the performance of various services and service areas (e.g., the dock, entrance and departure gates and the port channel) of the port, thereby detecting areas where performance is improving or declining. However, a disadvantage is how to evaluate the net impact of changes in these indicators on the port’s overall performance, i.e., when changes in some indicators improve performance and changes in other indicators negatively affect performance. What is needed is an overall (or single) port

62

WAYNE K. TALLEY

performance indicator. In our model (where the port’s economic objective is to maximize annual throughput subject to a minimum profit constraint), this overall indicator is the Lagrangean multiplier (or the shadow choice) variable (d) in function (10). Its performance standard is the maximum annual port throughput per profit dollar (given the port’s profit constraint). Thus, if the port’s annual throughput per profit dollar is rising (declining) over time, it follows that its performance is improving (declining) over time – furthermore, the net impact of changes over time in the individual performance indicators on port performance has been positive (negative). The economic theory of the port found in this chapter includes port time, resource, production and cost functions that may used in empirical studies of the port. Specifically, the time functions Tsbj, Tscj, Tibj, Ticj, Thb and Thc may be estimated to investigate determinants of, and their effects on, the times in port of ships, vehicles and cargoes. Also, the resource functions (43), (44) and (45) may be estimated to investigate determinants of, and their effects on, the utilization of port resources. Since a resource function represents the relationship between the minimum amount of a given resource employed by a port and the port’s levels of operating options and amounts of cargo received, these functions may be used in frontier statistical models such as data envelopment analysis (DEA) models to investigate the relative efficiency of resource utilization among ports.15 Estimates of the relative production (or technical) efficiency of ports based upon port production functions such as functions (6) and (7) and DEA models are discussed in Cullinane (2002). The cost functions (47) and (48) may be estimated to investigate whether a multi-throughput port exhibits economies of scope and scale. Jara-Diaz, Martinez-Budria, Cortes, and Varagas (1997, 2002) in estimations of a cost function for Spanish multi-throughput ports providing liquid bulk, solid bulk and general cargo throughputs found the presence of economies of scope. In Jara-Diaz, Tovar de la Fe, and Trujillo (2005), an estimated cost function reveals the existence of overall and product-specific economies of scale and economies of scope for port multi-throughput terminals providing container, general cargo and RO-RO cargo throughput.

7. SUMMARY This chapter has presented an economic theory of a two-cargo port. The port’s economic objective is to maximize annual bulk/container throughput subject to a minimum profit constraint. The annual demands for the port’s

63

An Economic Theory of the Port

bulk and container throughputs are functions of their generalized (money and time) prices – port charges and time prices incurred by ocean carriers’, inland carriers’ and shippers’ ships, vehicles and cargoes, respectively, while in port. Production functions in the provision of bulk and container throughputs were specified as well as resource functions – representing the relationship between the minimum amount of a given resource employed by a port and the levels of its operating options and amounts of cargo received. The choice variables in the optimization of the port’s economic objective are its port prices and operating options. An operating option is the means by which a port can differentiate its service. The port’s operating options include its ship and vehicle loading/unloading service rates, channel and berth accessibility and reliability, entrance and departure gate reliability and damage and property losses to ships, vehicles and cargoes in port. Port congestion arises when port users interfere with one another in the utilization of port resources, thereby increasing their time in port. Bulk ships and containerships may experience ship berth congestion (when seeking to berth/unberth at the same time) and ship work congestion (when seeking to be loaded/unloaded at the same time). Bulk and container vehicles may experience vehicle gate congestion (when seeking to enter/depart at the same time) and vehicle work congestion (when seeking to be loaded/unloaded at the same time). A port is cost efficient when its throughputs are provided at minimum cost. This relationship is exhibited by the port’s cost function. A two-cargo port exhibits economies of scale for a given type of throughput when the port’s cost does not increase in proportion to the increase in the throughput. The port will exhibit economies of scope in the provision of two types of throughput if the cost of providing both throughputs is less than the sum of the costs of providing the throughputs (at the same levels) separately. The choice variables, port prices and operating options, may be used by the port as individual performance indicators for evaluating its performance. The theoretical port time, resource and cost functions may be used in empirical studies of the port to investigate determinants of, and their effects on: the times in port of ships, vehicles and cargoes; port resource utilization; and port costs.

NOTES 1. The port’s economic objective may also be to maximize profits. 2. Ports may have noncommercial goals, such as develop marine commerce; generate jobs; and support national, regional and local interests with respect to

64

WAYNE K. TALLEY

promotion of maritime related commerce, fisheries, recreation, industrial and commercial activities (Maritime Administration, 1994, pp. 52–53). 3. In port modeling studies by Bobrovitch (1982) and Goodman (1984), the term ‘‘port user cost per unit’’ is used rather than the term the ‘‘generalized price’’ of the port. The latter is also referred to as the ‘‘total port price’’ in Talley (1994). 4. Since shippers are incurring port time prices, it is implicitly assumed that shippers have ownership of cargo until it reaches its final destination. If ownership is transferred to the consignee at the time of pickup, the consignee incurs the port time price for the cargo rather than the shipper. 5. The ‘‘value of time’’ for transportation users is discussed in Talley (1983). 6. Opportunity costs are costs incurred in foregoing the next best alternative. 7. The technically efficient assumption is made, since technical efficiency is a necessary condition for cost efficiency (i.e., for a port to minimize cost in the provision of throughput). Cost efficiency, in turn, is a necessary condition for a port to maximize annual throughput subject to a minimum profit constraint. 8. For a discussion of production functions for transportation carriers, see Talley (1988). 9. If vehicle loading and unloading service rates for a given type of cargo vary by type of vehicle, then the model may be reconstructed to have a distinct VL and a distinct VU for type of cargo and type of vehicle. A vehicle may load to and from ships as well as to and from port storage areas such as yards and warehouses. 10. For investigations of ship and cargo damage severity of ship accidents, see Loeb, Talley, and Zlatoper (1994), Talley (1995, 1996, 2002) and Talley, Jin, and Kite-Powell (2004). 11. Preemptive-priority queueing models for arrivals express the average waiting time for high-priority arrivals as a function of their arrival and service rates; the average waiting time for low-priority arrivals is expressed as a function of their arrival and service rates as well as a function of the arrival and service rates of highpriority arrivals (see Gross & Harris, 1974). An application of preemptive-priority queueing functions to a mixed-cargo port is found in Wood (1982). For other applications of queueing functions in port planning, see Chadwin, Pope, and Talley (1990), Frankel (1987) and Jansson and Shneerson (1982). 12. Cost function (47) is a long-run cost function in that there are no fixed costs, i.e., all resources are variable. 13. For further discussion, see Talley (1988, pp. 63–65). 14. For a discussion of short-run cost functions and economies of density for transportation carriers, see Talley (1988). 15. For a discussion of DEA models, see Cullinane (2002).

REFERENCES Bobrovitch, D. (1982). Decentralised planning and competition in a national multi-port system. Journal of Transport Economics and Policy, 16, 31–42. Chadwin, M. L., Pope, J. A., & Talley, W. K. (1990). Ocean container transportation: An operational perspective. New York: Taylor & Francis.

An Economic Theory of the Port

65

Cullinane, K. P. B. (2002). The productivity and efficiency of ports and terminals: Methods and applications. In: C. Grammenos (Ed.), The handbook of maritime economics and business (pp. 426–442). London: Lloyds of London Press. Frankel, E. G. (1987). Port planning and development. New York: Wiley. Goodman, A. C. (1984). Port planning and financing for bulk cargo ships: Theory and a North American example. Journal of Transport Economics and Policy, 18, 237–252. Gross, D., & Harris, C. M. (1974). Fundamentals of queueing theory. New York: Wiley. Jansson, J. O., & Shneerson, D. (1982). Port economics. Cambridge, MA: The MIT Press. Jara-Diaz, S., Martinez-Budria, E., Cortes, C., & Varagas, A. (1997). Marginal costs and scale economies in Spanish ports. European Transport Form: Proceedings, 25, 137–147. Jara-Diaz, S., Martinez-Budria, E., Cortes, C., & Basso, L. (2002). A multi-output cost function for the services of Spanish ports’ infrastructure. Transportation, 29, 419–437. Jara-Diaz, S., Tovar de la Fe, B., & Trujillo, L. (2005). Multioutput analysis of cargo handling firms: An application to a Spanish port. Transportation, 32, 275–292. Loeb, P. D., Talley, W. K., & Zlatoper, T. J. (1994). Causes and deterrents of transportation accidents: An analysis by mode. Westport, CT: Quorum Books. Maritime Administration. (1994). Public port financing in the United States. Washington, DC: U.S. Department of Transportation. Talley, W. K. (1983). Introduction to transportation. Cincinnati, OH: South-Western Publishing Company. Talley, W. K. (1988). Transport carrier costing. New York: Gordon and Breach Science Publishers. Talley, W. K. (1994). Performance indicators and port performance evaluation. The Logistics and Transportation Review, 30, 339–352. Talley, W. K. (1995). Vessel damage severity of tanker accidents. The Logistics and Transportation Review, 31, 191–207. Talley, W. K. (1996). Determinants of cargo damage risk and severity: The case of containership accidents. The Logistics and Transportation Review, 32, 377–388. Talley, W. K. (2002). Vessel damage cost differentials: Bulk, container and tanker accidents. International Journal of Maritime Economics, 4, 307–322. Talley, W. K., Jin, D., & Kite-Powell, H. (2004). Post OPA-90 vessel oil spill differentials: Transfers versus vessel accidents. Maritime Policy and Management, 31, 225–240. Wood, T. W. (1982). The economics of mixed cargo and cruise ship traffic in a port. Journal of Transport Economics and Policy, 16, 43–53.

This page intentionally left blank

66

MULTIPLE OUTPUTS IN PORT COST FUNCTIONS Sergio R. Jara-Dı´ az, Eduardo Martı´ nez-Budrı´ a and Juan Jose´ Dı´ az-Herna´ndez ABSTRACT By describing the technical process, in this article we argue that cargo that arrives in many different forms are distinct outputs of a port, whose services require inputs that can be grossly defined as labor, space, facilities, and equipment. Then we show theoretically and empirically that the use of aggregate output in the presence of distinct outputs causes erroneous conclusions on marginal costs, on economies of scale, and on optimal industry structure.

1. INTRODUCTION Port activity is complex and involves a large number of economic stakeholders. It is worth highlighting that what is known as port operation really encompasses a large number of smaller operations, most of which form the successive links of a chain in which the weakest link is the one that determines the strength of the chain as a whole. The fact that there are a large number of stakeholders and operations means that coordination becomes one of the essential keys to port efficiency. Port Economics Research in Transportation Economics, Volume 16, 67–84 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16004-7

67

68

SERGIO R. JARA-DI´AZ ET AL.

Port operations can be divided into three stages: ship-oriented services, cargo-oriented services and inter-modal connection. There are, therefore, two main users of the ports: the shipping lines on the waterside and the shippers on the landside. Among the services provided in ports, there are those provided by the facilities (lighthouses and beacons, docks and quays, security of ships and cargo, etc.), cargo-handling services (loading–unloading ships, loading– unloading inland transport and goods planning in cargo terminals), other ship-oriented services (pilot guidance, towage, berthing), other ship- and/or cargo-oriented services (repairs, marking, forwarding and bulk breaking, storage), and other services like inspection, customs, police, etc. In general, these functions are exercised by a range of both public and private bodies. Port authorities are the competent agency for facilities, the stevedore companies are the competent agents for cargo handling, ship- and cargo-oriented services (except pilot services) are usually provided by private companies, while pilot services, inspection, and the like are generally provided by the administration. The task of coordinating the different functions and companies usually befalls to the port authorities within each port. The main private agencies operating in ports are the shipbrokers that coordinate most of the services required by ships, stevedore firms that take care of the cargo-handling operations, and freight forwarder companies that coordinate all the cargo services for shippers. They usually act as intermediaries between their clients, the owners of the cargo or shipping companies, and the service companies. However, the present concept of a port as concentrating traffic or as a logistics activity zone, is leading to the new figure of the logistics operator, with a global view of the entire transport operation from the cargo’s point of origin to its destination. In this case, they charge for a complete service, very often door to door; therefore, the boundaries between the service providers and the services they offer are becoming increasingly blurred. For synthesis, viewed as a whole, ports are factories designed to receive and dispatch cargo that arrives in many different forms. These movements are the outputs of a port, whose services require inputs that can be grossly defined as labor, space, facilities, and equipment. The technical process can be described as the search for the optimal combination of inputs to be able to move different combinations of outputs, which can be described either through a production (transformation) function, or through a cost function that represents the minimum expenditure necessary to produce those movements. In this chapter, we expose the reasons why cost functions estimated to describe port activities should be specified in a multiple output form

Multiple Outputs in Port Cost Functions

69

(Jara-Dı´ az, 1983) as opposed to the aggregate output approach, usually total tons moved or variations, e.g. Kim and Sachish (1986), Reker, Connell, and Ross (1990), Tongzon (1993), and Martı´ nez-Budrı´ a (1996). In the following section we present the basic elements of multioutput theory as developed by Baumol, Panzar, and Willig (1982) and a theoretical formulation of the problems caused by output aggregation in the context of ports. Then the technical analysis of input usage to produce the movements of an output mix in ports is presented, providing a case for a multioutput analysis instead of an aggregate output one. In the fourth section, some evidence from the literature is presented to verify the theoretical predictions. The final section concludes the chapter.

2. MULTIOUTPUT THEORY AND AGGREGATION The cost function C(W,Y) represents the minimum expenditure needed to produce outputs in Y, at factor prices W, which have been suppressed below for simplicity. The marginal cost of producing i keeping all other products constant is calculated as @C ¼ Cmi @yi

(1)

The multioutput degree of economies of scale, S, is a technical property of the productive process, defined in the transformation or production functions. However, dual relations allow the calculation of S directly from the cost function as (Panzar & Willig, 1977) S¼

CðY Þ Y ry CðY Þ

(2)

S represents the maximum growth rate that the product vector can reach when productive factors increase by the same proportion. Therefore, the presence of increasing returns to scale (S41) implies that a proportional growth of all products induces a less than proportional growth of costs, i.e. a production expansion exhibits advantages from the point of view of costs. It is easy to verify that in the single output case S is given by the ratio between average cost C/Y and marginal cost and that increasing returns are present when average cost is decreasing with output. The incremental cost of product i, ICi, is defined as the cost of adding that product to the line of production. This corresponds to IC i ¼ Cðy1 ; y2 ; . . . ; yn Þ

Cðy1 ; y2 ; . . . ; yi 1 ; 0; yiþ1 ; . . . ; yn Þ

(3)

SERGIO R. JARA-DI´AZ ET AL.

70

This concept can be extended to a subset of products R, generating ICR(Y). The degree of economies of scale specific to subset R is defined as SR ðY Þ ¼ P

j2R

IC R ðY Þ IC R ðY Þ ¼P yj @CðY Þ=@yi yj Cmi ðY Þ

(4)

j2R

The interpretation of SR(Y) is similar to that of S. Two products are said to exhibit cost complementarity when the marginal cost of one of them diminishes as the other product increases, suggesting some form of advantage in joint production. Formally, this means that C ij ðY Þ ¼

@2 CðY Þ 0 @yi yj

(5)

Economies of scope measures the relative cost increase that would result from the division of the production of Y into two different production lines T and N T. The degree of economies of scope SCT of subset of products T with relation to its complementary subset N T is defined as SC T ðY Þ ¼

1 ½CðY T Þ þ CðY N CðY Þ

TÞ

CðY ފ

(6)

If SCT(Y) 40, economies of scope are said to exist and it is cheaper to produce vector Y jointly than to produce vectors YT and YN T separately. In other words, it is not advisable to specialize but to diversify production. It is easy to see that 1o SCo1. It can be shown that scale and scope are related by S N ðY Þ ¼

aT ST ðY Þ þ ð1 aT ÞS N 1 SC T ðY Þ

T ðY Þ

(7)

with . yj @CðY Þ @yj j2T . aT ¼ P yj @CðY Þ @yj P

(8)

j2N

This means that, in the absence of economies of scope (SC ¼ 0), S would be a weighted average of the specific economies of scale of each subset. The existence of economies of scope (SC40) favors the presence of overall economies of scale. Now we can examine the case in which a multioutput production process in ports is viewed or analyzed by means of a cost function C~ that has been

71

Multiple Outputs in Port Cost Functions

estimated describing product as a single (aggregated) output Y~ ; total tons moved per period. As shown in the next section, output in ports should distinguish the different types of cargo that are served. Let us define yi as the volume P of cargo type i. Then, Y is the output vector containing all yi and Y~ ¼ yi : ~ Y~ Þ is an implicit represenUnder these circumstances, the estimated Cð tation C^ of C(Y) because (Jara-Dı´ az & Corte´s, 1996)   ~ Y~ Þ  C~ Y~ ðY Þ  CðY ^ Þ Cð (9) ~ Y~ Þ only quantities like (1) and This is a very important relation, as from Cð (2) can be calculated for different values of Y~ ; unlike C(Y) from which all elements (1)–(7) can be investigated. This has implications when it comes to calculate marginal costs and scale economies. First, and in a very evident way, the marginal cost estimate for each product would be @C^ @C~ @Y~ @C~ ¼ ¼ @yi @Y~ @yi @Y~

(10)

which implies that the value of the additional resources necessary to move one additional volume unit of cargo i is equal for all cargo types. This suggests that the multioutput approach should be used whenever there are reasons to believe that marginal costs differ across products. What is less evident, though, is the way in which the presence of economies of scope between the products in Y is ‘‘captured’’ when Y~ is used, which is expressed in the following Theorem. Theorem. The existence of economies of scope between products in T and ~ Y~ Þ (single output scale econN T induces decreasing average costs in Cð omies). Proof. First note that Y~ T  Y~ ðY T Þ ¼

X

yi and Y~ N

T

 Y~ ðY N

TÞ

i2T

¼

X

yi

(11)

ieT

which implies that Y~ T ¼ yT Y~ and Y~ N

T

¼ ð1

yT ÞY~

(12)

where yT A (0,1) is the proportion of the volume of products in T with respect to the total. From Eqs. (6) and (12) the existence of economies of scope between products in T and N T implies that   ^ N T Þ ¼ Cðy ~ T Y~ Þ þ C~ ð1 yT ÞY~ 4Cð ~ Y~ Þ ^ T Þ þ CðY (13) CðY

72

SERGIO R. JARA-DI´AZ ET AL.

Without losing generality we assume that   ~ T Y~ Þ C~ ð1 yT ÞY~ Cðy 4 ð1 yT ÞY~ yT Y~

(14)

As known, if a/b4c/d with a, b, c, and d positive, then a/b4a+c/b+d. Applying this and imposing (13) we get     ~ T Y~ Þ þ C~ ð1 yT ÞY~ ~ T Y~ Þ þ C~ ð1 yT ÞY~ ~ T Y~ Þ Cðy ~ Y~ Þ Cðy Cðy Cð 4 ¼ 4 ~ ~ ~ ~ yT Y þ ð1 yT ÞY Y yT Y Y~ (15) As yT A (0,1), this shows that average costs are decreasing. Note that condition (14)  is general,  as if inequality was the opposite the proof is similar for C~ ð1 yT ÞY~ =ð1 yT ÞY~ : Q.E.D. This is quite an important result, as the presence of scope in C(Y) would ~ Y~ Þ; overestimating true economies of turn into scale when analyzed from Cð scale and reinforcing the identification of a port activity as a natural monopoly.

3. PORT PRODUCTION 3.1. Port Operation Sea transport, like all productive processes, has undergone profound technological changes over recent decades, aimed at increasing productivity. This has been attained by reducing the cost of the operation that, like any transport operation, it can be divided into two parts: the monetary tariffs and the cost of the time of the operation, which in the case of cargo, can be objectively quantified. This technological advance could not occur solely in a single link of the transport chain; it has to be accompanied by all the other links. Hence, the new kinds of specialized ships have meant that new, more efficient port terminals and cargo-handling equipment had to be designed to make port operations faster. To this end, cargoes have to be transported in standard formats so that it can be handled by standard equipment used in all ports. This cargo standardization process means that the cost of the process is more dependent on how the cargo is packed than on the nature of the product. The container and the pallet are the best-known forms of standardizing cargo-handling operations.

Multiple Outputs in Port Cost Functions

73

The productive factors necessary for the port operation are a sheltered dock, mechanical equipment for loading (unloading) the ship, physical space and mechanical equipment for handling and distributing the cargo on land after the ship has been unloaded (before loading), and equipment for loading (unloading) the cargo onto terrestrial transport and personnel, both for managing the necessary infrastructure and mechanical resources used and for operating this equipment. In summary, physical capital is required in infrastructure for ships, cargo, and terrestrial transport; in cranes and other forms of unloading ships, in forklift trucks, and other means of moving the cargo in the terminal, in computer and communication systems; and personnel to manage the infrastructure and cranes and stevedores for the cargo-handling equipment. There is also a need for a range of intermediate inputs like electricity, fuel, water, communication services, office material, spare parts, etc. These are heterogeneous, rather than a single uniform process, each of which will combine these factors in a different manner depending on the cargo carried by the ship. The most widely accepted classification of cargoes is into liquid bulk, solid bulk, and general cargo. But this classification is too general to explain the complex operations of modern ports. So we need a sub-classification for each of these categories and a brief description of the operation, with the emphasis on both the specialized and the multipurpose inputs.

3.2. Port Cargoes 3.2.1. Liquid Bulks: Oil and Derivatives, Liquid Gases Loading and unloading is done by pipes running directly between the tank and the ship’s hold. This is the kind of cargo that uses the least amount of port inputs, as these are limited to a sheltered dock (sometimes, they even use sheltered waters, rather than a dock), the pipes, which are connected by the employees of the shipper or the receiving company, and the storage tanks, which may be public or private. 3.2.2. Solid Bulk: Cereals, Minerals, Cement Clinker, etc. There are different techniques for handling solid bulks, which can be divided into two groups: those that use specialized facilities and those that use multipurpose facilities and equipment. For the former group, we have specialized facilities like pneumatic loader and conveyor belts and, on the other hand, multipurpose cranes. The pneumatic loader is used for loading

74

SERGIO R. JARA-DI´AZ ET AL.

(unloading) cereals and cement clinker, and conveyor belts are used for unloading (loading) minerals. In both cases, there is a direct connection between the places where the cargo is to be stored and the ship, requiring specialized docks and, for minerals, large dedicated areas. Stevedores play a relatively small part in operations of this kind. The multipurpose cranes are used for unloading (loading) cereals and clinker, coupling a special device called a bucket, to the crane, to scoop up the cargo. In this case, the infrastructure requirement is the same (depending solely on the ship) but the demand for stevedore work is notably greater as workers have to pile up the bulk in the ship’s holds. If the cargo first has to be unloaded onto the dock as well, then this demand is greater still as it will then require another operation to load the said cargo onto a truck, train, etc. With regard to handling bulk cargoes on land, there are two options. First of all, the solid bulk is placed in a hopper and transferred to the trucks that will take it to the prepared silos. This way, the speed of the operation will depend on the availability of inland transport and the capacity of those receiving the grain. At other times, the crane will unload the bulk directly onto the dock, from where it will later be loaded onto trucks as these arrive. In this latter case, cranes are used less, as their job is finished once the bulk cargo is unloaded from the ship, although more personnel is required for handling the bulk from the dockside to the truck. Here, of course, the efficiency of the operation will depend on how fast the port workers act and the reception capacity of the client the cargo is being delivered to. The decision to use special facilities or multi-purpose equipment will depend on the volume of traffic. There is a minimum threshold of solid bulk that will pay back the investment with a reasonable return. So, the volume of bulks and the solution chosen for the problem of coordinating between sea and land transport will vary the demand for factors. 3.2.3. General Cargo in Containers and Carried on Container Ships Container loading and unloading operations are very different from other cargoes. For a port to handle large volumes of containers, it needs specific facilities: port terminals with large storage spaces, gantry cranes for loading (unloading) the container onto the ship, and equipment for moving the containers within the terminal and for loading them on to modes of land transport. The port hinterland trade structure can also determine the need for space and other inputs. If there is a pronounced imbalance between the arrival and departure of cargo in the hinterland, there will be a need for large flows of empty containers that, in turn, will determine the productive technique used.

Multiple Outputs in Port Cost Functions

75

Availability of storage area will determine the height to which containers can be stacked and there is a sustainable ratio between the input of area and the input of labor, mechanical equipment on the ground, and energy to power these. So, if the price of the area of land is relatively low in comparison with the other inputs, then the goods can be stacked to a lower height and, in this case only a few containers with have to be moved (in average value) to dispatch the container required. On the other hand, if the price of the land is relatively high in comparison with the other inputs, this will change the optimum demand, forcing managers to stack to a greater height and, therefore, face higher costs for the inputs of personnel, equipment to move the containers on land and intermediate inputs. Furthermore, managing a container terminal requires a costly specialized computer system. Another option for container handling is to transport them to the shipside on a rolling platform (or from the ship to the storage area) using tractors or mobile platforms. In this case, more multipurpose equipment can be used as these tractors or mobile platforms can also be used for handling containers to be transported in Roll-on Roll-off ships and also for moving general, non-containerized cargo. 3.2.4. General Cargo in Containers and Transported in Roll-on Roll-off Ships There are two techniques for moving containers in Roll-on Roll-off ships. In one of them, the container is placed on a platform that is towed onto the ship with a tractor. In the port of destination, the operation is reversed. The other way is to drive the truck straight onto the ship with the container. The former method allows greater flexibility and the use of multipurpose mechanical equipment as the tractors used are the same ones as those described in the handling of containers to be carried on container ships and for moving general, non-containerized cargo, on pallets for example, which are loaded on the same mobile platform as the ones used for carrying containers. 3.2.5. General Non-Containerized Cargo There are different forms for shipping general non-containerized cargo: in individual units like rolls of paper; with a pre-slung system like iron or wood; on pallets (the most widespread form) and, particularly, cars. In all these cases, handling requires the use of medium-sized cranes for loading (unloading) the ship and sometimes mobile, wheeled cranes that, although they are not specifically devoted to port work, as they are also designed for building roads, or building works in general, they do give the investment in cranes greater flexibility and adapt it more to the volumes of cargo. Forklift

76

SERGIO R. JARA-DI´AZ ET AL.

trucks are also needed for moving cargo on the dockside, apart from the stevedores. But there are differences in handling techniques, depending on the cargo to be handled, although the ship’s port infrastructure requirements are the same for the techniques described above and will depend exclusively on the technical specifications of the ship in question. If the cargo is transported in individual units, such as rolls of paper, specific fittings are used both on the cranes and on the forklift trucks that handle the merchandise on land, because the cargo is fragile, and they also require a labor force with a certain degree of specialist training. These two kinds of mechanical equipment are the same as those used for other kinds of non-containerized cargoes, i.e. medium capacity cranes. Docks with nearby cold store facilities are usually used for handling perishable goods. Another form of transporting merchandise is the pre-slung system, designed for putting units of cargo (like wood and iron) together in bundles or batches to facilitate their handling, to hook them up to the cranes and to move them from one place to another. In the case of cars carried as cargo, generally on special ships, they are unloaded by the stevedores by driving the cars off the ship and performance and efficiency will depend on the availability of parking areas and how close these are to the ship.

3.3. Common and Specialized Inputs As we have seen, there are inputs that are used for all kinds of goods and there are others that are good specific, although this is related with total and relative volumes as well. In multipurpose terminals, the inputs are common to the different kinds of outputs that are moved, that is, the same infrastructure, the same cranes and the same stevedores are used whatever the cargo: containers or pallets, individual units, pre-slung and Roll-on Roll-off. They can even use the same facilities for loading solid bulk cargoes. This kind of terminal is a good option for small ports, where the volume of traffic does not justify specialized terminals. Large volumes make a case for a greater degree of inputs’ specialization, but this is never absolute because, even in large ports with highly specialized terminals, there are inputs that are used generally for all kinds of cargo. These resources that are common to all terminals, regardless of the kind of goods handled and the volume of traffic, include, at least the following: lighthouses and shipping services, breakwaters, road networks, buildings, infrastructure, mechanical equipment and loading/unloading equipment

Multiple Outputs in Port Cost Functions

77

management personnel, most of the cargo-handling personnel and the inspection, customs, port security services, etc. Furthermore, there are resources that are used for several different kinds of cargo. For example, the general cargo docks are generally used for loading (unloading) both pallets and loose or pre-slung goods. In this case, the stevedores, cranes, and landside resources are all common. The first two are used for handling solid bulk if there are no special facilities. Tractors and platforms are used for most road traffic, whether it is transported on pallets or in containers and, in some ports, they are also used for moving containers on the dockside before they are loaded (unloaded) by crane. On the other hand, we have specialized inputs used for certain goods. In the case of general perishable goods, with temperature and/or humidity requirements, the only specific input will be the cold storage facilities. The most significant example of specialized inputs, however, is that of container terminals, where, first of all, the berths are specific, depending on the draught and length needed by the container ship. There also require spacious and dedicated areas close to the docks for handling and storage. Moreover, the cranes for loading (unloading) the ship are specialized cranes, normally gantry cranes that are more powerful and faster than the cranes used for handling other cargoes. The mechanical equipment for handling and dispatching (receiving) containers on land is also specialist equipment, such as trastainers and more versatile machinery like the reach stacker and the spreader forklift. These mechanical resources are specially designed for handling containers, thus making the operation faster and more efficient, which, at the same time, guarantees the security of the operation. The operation also has to be carried out by employees specially trained in the use of this kind of machinery, both to attain the best possible efficiency and because the machinery is complex. Solid bulk is handled in special facilities when large volumes are involved. These include specific dock and surface infrastructures and mechanical equipment for transporting the bulk from ship to shore. The specific inputs for handling cereals are the pneumatic loaders, the technical specifications of which will depend on the volume of bulk to be handled, and the Silos on the dockside. Minerals, on the other hand, require large areas on the dockside and specially designed conveyor belts. For relatively small volumes of bulk, hoppers are used for receiving and dispatching, and these are used solely for a specific kind of good. Finally, liquid bulk is transferred with underground pipelines connecting the ship’s tanks with the shoreside tanks that are used exclusively for storing

78

SERGIO R. JARA-DI´AZ ET AL.

the goods safely. One example of this is the tanks and pipelines used for oil products.

4. EMPIRICAL EVIDENCE For synthesis, the technical elements provided in the previous section show that there are technical reasons to expect different marginal costs for distinct types of cargo moved within a port, which, as argued earlier, are a cause for a multioutput analysis of port production. In what follows we present empirical evidence regarding the type of bias induced by aggregate output approaches in port activities, including one example on infrastructure port services and another on cargo-handling activities. Infrastructure port services related to planning and management of port facilities in Spanish ports were analyzed by Martı´ nez-Budrı´ a (1996) using a Cobb–Douglas specification within an aggregate output approach, obtaining significantly large economies of scale ðS^ ¼ 3:47Þ: Data covers 26 Spanish ports during a period of 5 years (1985–1989), including information from the port authorities. Using an expanded data set for the same ports (11 years, 1985–1995) Jara-Dı´ az, Martı´ nez-Budrı´ a, Cortes, and Basso (2002) examined the same sector under a multioutput view. The dependent variable, total annual cost (TC) for infrastructure and its administration, includes labor (GL), amortization (GK), and other expenses (GI), directly obtained from port reports. The explanatory variables include four kinds of cargo in the multioutput case, and three indices for input prices. The output components represent port activities, and they include the movement of containerized general cargo (CGC), non-containerized general cargo (NCGC), dry bulk (DB), and liquid bulk (LB). The multioutput function was estimated using a quadratic specification, which allowed for scope analysis. As the aggregate output cost function reported by Martı´ nez-Budrı´ a (1996) has less information and used a more restricted form than the multioutput model presented in Jara-Dı´ az et al. (2002), we estimated a new version of the aggregate output cost function using the same quadratic specification and data base used in the latter (see Appendix A). Results of the two directly comparable models are shown in Table 1. At the mean of the observations, the multioutput approach revealed different marginal costs for each of the four products (minimum and maximum averages by port in parenthesis): 427 (278 647) for CGC, 500 (335 613) for NCGC, 125 (75 220) for LB, and 183 (143–246) for DB, all in pesetas/ton.

79

Multiple Outputs in Port Cost Functions

Table 1. Cost Structure in the Multioutput and Aggregate Output Analysis of Infrastructure Services of Spanish Portsa. Multioutputb

Approach Output Marginal cost (pesetas/ton) Scale economies

CGC 427

NCGC 500 1.69

Aggregate Output

DB

LB

Total tons

183

125

255 3.09

a

Quadratic cost function and input expenditure equations used in both models estimated using the SURE procedure. Period 1985–1995, 26 ports authorities, and 286 observations. b Source: Jara-Dı´ az et al. (2002).

It is fairly evident that the relative order of the marginal costs figures by product reflects the technical discussion presented in Section 3: the lowest for liquid bulk and the largest for NCGC. The aggregate output analysis gave a single value of 255 pesetas/ton using the quadratic specification. When it comes to analyzing scale economies, the importance of the theorem proved in Section 2 becomes apparent. Jara-Dı´ az et al. (2002) obtained a figure of 1.69 for the degree of scale economies at the mean. They also studied economies of scope calculating SC using Eq. (6) for three partitions of output: liquid bulk versus the rest (SCLB), general cargo from liquid and dry bulk (SCGC), and CGC versus the rest (SCCGC). The three positive values indicated the presence of economies of scope between the corresponding subsets, reflecting the convenience of using common infrastructure if all port services are to be produced. We have estimated a value of 3.09 for the degree of economies of scale economies at the mean using the aggregate output quadratic cost function. These values are according to the theorem. A second country-wise study that can be used for comparison of the two approaches is Martı´ nez-Budrı´ a and Dı´ az-Herna´ndez (2006) that analyzed cargo handling in Spain using a multiproduct approach. The outputs analyzed in this study were defined according to how the merchandise is handled: CGC, NCGC, and solid (dry) bulk cargoes that are handled without special facilities (DB). Likewise the two factors considered are labor and cranes. The data from the Spanish Ports’ Annual Reports have been used to get the quantities of cargo moved by each port and year included in the sample. Other data sources were a questionnaire presented to the SEEDs (Sociedad Estatal de Estiba y Desestiba) that provided information on the labor factor, basically concerned with labor costs and hours worked by stevedores. The Annual Reports of each port and the information received from crane operators companies in ports provided the hours worked by

SERGIO R. JARA-DI´AZ ET AL.

80

Table 2. Cost Structure and Technical Change in the Multioutput and Aggregate Output Analysis of Cargo Handling in Spanish Portsa. Multioutputb

Approach Output Marginal cost (pesetas/ton) Scale economies Average technical change (%)

Aggregate Output

CGC

NCGC

DB

Total tons

488

593 1.071 3.66

292

403 1.34 2.53

a

Normalized quadratic cost function and input expenditure equations used in both models estimated using the SURE procedure. Period 1990–1998, 21 cargo-handling firms, 158 observations. b Source: Martı´ nez-Budrı´ a and Dı´ az-Herna´ndez (2006).

cranes and permitted the calculation of total expenses on this item. Twentyone Spanish ports observed from 1990 to 1998 were included in this study. However, as some SEEDs were created during the study period, the number of observations for each port varies. The above-mentioned sources were used to build a data panel with 158 annual observations. The normalized quadratic cost system was used to estimate the multioutput cost function. Table 2 shows the results including the calculation of the rate of technical change.1 Using the same data and specification, we estimated an aggregate output cost function, shown in Appendix B. Marginal costs and scale economies are shown in Table 2 for the two models. These results again show differences for marginal costs across products, which justify the multiproduct approach, and the relative order fits the intuition suggested by our discussion in the previous section. The overall degree of scale economies shows nearly constant returns. On the other hand, economies of scope exist for all partitions involving single output production, which means that it is less costly to produce the three products jointly than to do it with two firms, one producing either of the three outputs and the other producing the other two (at the mean). The marginal cost figure at the mean for the aggregate output approach was 403 pesetas/ton and the estimated degree of scale economies was S ¼ 1.34, i.e. increasing returns were found. According to the theorem in Section 2, this is again explained by the presence of economies of scope detected when the process is correctly analyzed. It is worth noting that the estimate for the average rate of technical change drops to 2.53 with the aggregate output view, underestimating the effect of technology. Finally, a very specific analysis made by Jara-Dı´ az, Tovar, and Trujillo (2005) on three cargo-handling firms in the port of Las Palmas show the

81

Multiple Outputs in Port Cost Functions

Table 3. Cost Structure in the Multioutput and Aggregate Output Analysis of Cargo Handling Firms in Las Palmas Porta. Multioutputb

Approach Output Marginal cost (pesetas/ton) Scale economies

Aggregate Outputc

CGC

NCGC

RR

Total tons

745

1974 1.64

1056

715.9 1.96

a

Quadratic cost function and input demand equations estimated with the SURE procedure. Period 1991–1999 (monthly), three firms, and 264 observations. b Source: Jara-Dı´ az et al. (2005). c Source: Jara-Dı´ az et al. (2006).

same qualitative effects as those shown here. In that study, other cost items were included in addition to labor and equipment (intermediate inputs, land, and other capital items). Different marginal costs were detected among containers, Roll-on Roll-off cargo (RR), and general cargo. These are shown in Table 3, along with increasing returns that were detected at a disaggregated level (S ¼ 1.64). The presence of economies of scope for all partitions once again turns into an overestimated value S ¼ 1.96 when the aggregate analysis was used (Jara-Dı´ az, Tovar, & Trujillo, 2006). Besides illustrating empirically the bias in scale economies induced by the aggregate output analysis under the presence of economies of scope, proved in our theorem, there is another point that merits attention. The statistical results from the aggregate analysis are quite satisfactory (see Appendices A and B), which might induce the acceptance of these results if there was no alternative specification, with all the implications on optimal industry structure and potential regulatory policies.

5. CONCLUSIONS The main contribution of this paper to the literature is to show that there are various important reasons to prefer the multioutput analysis to the aggregate output view as the most appropriate approach to study the supply side of port production. First, we have proved theoretically that the presence of scope is captured as scale when no distinction among different products is made, inducing biased estimates of scale and erroneous conclusions on optimal industry structure when the aggregate output view is adopted. Second, we have provided technical reasons to expect relevant marginal cost differences among the various types of cargo moved within a port, which is a cause

SERGIO R. JARA-DI´AZ ET AL.

82

for multioutput analysis. Finally, empirical evidence was presented to verify the theoretical properties. This also shows that good statistical properties of an aggregate output cost function are no guarantee of a correct analysis of industry structure. The technical aspects should be understood as well.

NOTES 

1. The rate of technical change is given by T ¼ 1=C@CðW ; Y ; tÞ=@t; where W is the input price vector, Y the output vector, and t the time trend introduced to capture the technical change.

ACKNOWLEDGMENTS Support by Fondecyt, Chile, grant 1050643, the Millennium Nucleus in Complex Engineering Systems, and the Spanish Ministry of Public Works, call 2002 for Research Grants in the Area of Transport is gratefully acknowledged.

REFERENCES Baumol, W., Panzar, J., & Willig, R. (1982). Contestable markets and the theory of industry structure. New York: Harcourt, Bruce and Jovanovich, Inc. Jara-Dı´ az, S. (1983). Freight transportation multioutput analysis. Transportation Research, 17A, 429–438. Jara-Dı´ az, S., & Corte´s, C. (1996). On the calculation of scale economies from transport cost functions. Journal of Transport Economics and Policy, 30, 157–170. Jara-Dı´ az, S., Martı´ nez-Budrı´ a, E., Cortes, C., & Basso, L. (2002). A multioutput cost function for the services of Spanish ports’ infrastructure. Transportation, 29, 419–437. Jara-Dı´ az, S. R., Tovar, B., & Trujillo, L. (2005). Multioutput analysis of cargo handling firms: An application to a Spanish port. Transportation, 32, 275–291. Jara-Dı´ az, S. R., Tovar, B., & Trujillo, L. (2006). The effect of using aggregated output in the economic analysis of cargo handling operations. In: P. Coto-Milla´n & V. Inglada (Eds), Essays in transport economics. Heidelberg: Springer. Kim, M., & Sachish, A. (1986). The structure of production, technical change and productivity in a port. Journal of Industrial Economics, 35, 209–223. Martı´ nez-Budrı´ a, E. (1996). Un estudio econome´trico de los costes del sistema portuario espan˜ol (An econometric study of the Spanish ports system costs). Revista Asturiana de Economia, 5, 135–149. Martı´ nez-Budrı´ a, E., & Dı´ az-Herna´ndez, J. J. (2006). Multioutput analysis of the cost and productivity of cargo handling in Spanish ports. In: P. Coto-Milla´n & V. Inglada (Eds), Essays in transport economics. Heidelberg: Springer. Panzar, J. C., & Willig, R. D. (1977). Economies of scale in multioutput production. Quarterly journal of Economics, 91(August), 481–493.

83

Multiple Outputs in Port Cost Functions

Reker, R. A., Connell, D., & Ross, D. I. (1990). The development of a production function for a container terminal in the port of Melbourne. Papers of the Australasian Transport Research Forum, 15, 209–218. Tongzon, J. L. (1993). The port of Melbourne authority’s pricing policy: Its efficiency and distribution implications. Maritime Policy and Management, 20, 197–203.

APPENDIX A. PARAMETER ESTIMATES OF THE AGGREGATE OUTPUT COST FUNCTION FOR INFRASTRUCTURE SERVICES IN SPANISH PORTS Parameter Aggregate cargo (C) Labor price (WL) Capital price (WK) Intermediate input (WI) C*C WL*WL WK*WK WI*WI WL*WK WL*WI WK*WI C*WL C*WK C*WI dALGEGIRAS dALICANTE dALMERIA dBARCELONA dBILBAO dCADIZ dCARTAGENA dCASTELLON dCEUTA dEL FERROL dGIJON dHUELVA dCORUN˜A dLAS PALMAS

Estimate 255.7 136.9 2729.9 2.61E+06 1.31E 03 0.037 0.038 260259 0.3623 47.246 373.71 0.0263 0.4022 26.837 3.27E+0.6 2.93E+0.6 2.54E+0.6 0.99E+0.5 0.77E+0.5 4.16E+0.6 2.40E+0.6 2.56E+0.6 2.35E+0.6 2.01E+0.6 5.75E+0.6 4.03E+0.6 2.50E+0.6 4.65E+0.6

t-Statistic 9.142 4.410 15.672 23.829 10.392 5.301 5.628 23.873 2.038 2.608 4.280 4.332 7.255 13.412 21.354 17.373 13.991 61.764 34.483 27.441 13.532 13.322 16.395 11.046 27.191 26.793 14.113 31.233

SERGIO R. JARA-DI´AZ ET AL.

84

APPENDIX A. (Continued ) Parameter dMALAGA dMELILLA dPMALLORCA dPASAJES dPONTEVEDRA dTENERIFE dSANTANDER dSEVILLA dTARRAGONA dVALENCIA dVIGO dVILLAGARCIA

Estimate

t-Statistic

3.36E+0.6 2.41E+0.6 3.76E+0.6 3.83E+0.6 2.34E+0.6 4.22E+0.6 3.82E+0.6 4.06E+0.6 4.07E+0.6 5.31E+0.6 3.74E+0.6 1.84E+0.6

20.852 11.857 27.241 26.126 11.778 30.972 25.331 22.424 28.292 31.493 23.485 8.859

Notes: The variable dALGECIRAS and those that follow are binary variables denoting port; all are included as no constant was used. For discussion of the estimation procedure, see Jara-Diaz et al. (2002); 286 observations, R2 cost function: 0.978, R2 labor expenditure equation: 0.817, R2 intermediate input expenditure equation: 0.855, and R2 capital expenditure equation: 0.801.

APPENDIX B. PARAMETER ESTIMATES OF THE AGGREGATE OUTPUT COST FUNCTION FOR CARGO HANDLING IN SPANISH PORTS Parameter Labor price (WL) Aggregate cargo (C) Trend (t) WL*WL WL*C t*t WL*t C*t C*C Constant

Estimate

t-Statistic

159.75 11.49 0.88 28.18 48.80 0.26 6.08 0.56 0.17 43.33

7.11 14.28 9.04 8.26 5.03 2.27 3.99 3.67 2.72 14.73

Notes: R2 quadratic cost function: 0.884 and R2 labor expenditure equation: 0.908.

ESTIMATING THE RELATIVE EFFICIENCY OF EUROPEAN CONTAINER PORTS: A STOCHASTIC FRONTIER ANALYSIS Kevin Cullinane and Dong-Wook Song ABSTRACT This paper estimates the relative technical efficiency of a sample of European container ports using the cross-sectional version of the ‘stochastic frontier model’ under the assumption that the functional form of the production frontier is the log-linear Cobb-Douglas function. The estimated efficiency measures are broadly similar for the three assumed error distributions that were tested. From the results of the analysis, it is concluded that the size of a port or terminal is closely correlated with its efficiency. Ports in the United Kingdom were found to have the most efficient infrastructure usage; a finding consistent with the shortage of containerhandling capacity. Scandinavian and Eastern European container terminals yielded the lowest estimates of relative efficiency. Geographical location (being displaced from the mainline intercontinental container trades) and below average size are possible explanations for this result.

Port Economics Research in Transportation Economics, Volume 16, 85–115 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16005-9

85

86

KEVIN CULLINANE AND DONG-WOOK SONG

1. INTRODUCTION The globalisation of production and consumption and the impact that this has had upon international trade patterns has enhanced the importance of container transportation. This is largely because of the numerous technical and economic advantages that this method of transporting cargoes offers in comparison to more traditional methods. As critical nodes and potential bottlenecks in international logistical networks, container ports play a pivotal role in the container transportation process, particularly in their role as a conduit for facilitating intermodal transfers at the interface of sea and inland transportation systems. One of the distinguishing features of today’s container port industry is that competition is more intense than has previously been the case. Because of the immovable geographical locations of ports and the concentration of cargoes at these locations, port markets have in the past been perceived as monopolistic in nature. However, the rapid development and adoption of container and intermodal technologies have radically altered the structure of port markets from one of monopoly to one where fierce competition prevails in many parts of the world. As such, container ports can no longer take for granted the handling of cargoes from within their hinterland. Instead, a port’s hinterland increasingly overlaps with those of rival ports, in a situation where those that exercise the port choice decision now have a much greater range of viable alternatives to select from (Cullinane & Khanna, 2000). These characteristics of the contemporary container port industry are nowhere more acutely observed than in Europe. As an obvious corollary of burgeoning international trade and the rapid increase in the volume of seaborne cargoes associated with this, container ports provide the platform upon which national and regional economic development within, and of, the European Union is founded (e.g. Winkelmans, 2004; Martine, Perez, San Juan, & Suarez, 2004). In addition, the fierce competition that exists among Europe’s container ports has now reached such a level of intensity that it has prompted numerous analyses of the competitive environment (e.g. Notteboom, 1997; Wang & Cullinane, 2004). This intense competition has also stimulated an overt interest in the efficiency with which the sector utilizes its resources. Given that individual container port (or terminal) performance makes a significant contribution to the prospects for survival, and the competitiveness, of the industry as a whole and the players that comprise it, the raison d’eˆtre behind this interest is obvious. By conducting a formal analysis of efficiency, not only is port management provided with a rational basis for decision making, but also an

Estimating the Relative Efficiency of European Container Ports

87

important input is derived for informing regional and national port planning and operations. It is important to emphasise that the analysis contained within this chapter is limited solely to deriving and comparing various estimates of the relative efficiency of the major ports and terminals within the European container port industry. Justification for such a study comes from the recognition that a significant level of competition exists within this industry and that the outcome of that competition is likely to be influenced, inter alia, by the levels of efficiency displayed by the various competing entities. It is certainly not the case, however, that any direct causal relationship is being hypothesised between changing levels of efficiency over time and the degree of competition within the container port industry. Although such a hypothesis is definitely worthy of testing, particularly because of the potential impact of market contestability (Baumol, Panzar, & Willig, 1982), it is one of such complexity and potential importance that it merits a separate study in its own right. For the purposes of this study, therefore, the primary focus is the analysis of cross-sectional data for the derivation of efficiency estimates for the European container port industry. A number of approaches have been utilised for analysing the performance of ports. In the past, these have included: the calculation of cargo-handling productivity at berth (Bendall & Stent, 1987; Tabernacle, 1995; Ashar, 1997), the measurement of single factor productivity (De Monie, 1987), the comparison of actual with optimum throughput over a specific time period (Talley, 1988), the estimation of a port cost function (De Neufville & Tsunokawa, 1981), the calculation of the total factor productivity of a port (Kim & Sachish, 1986) and the establishment of a port performance and efficiency model using multiple regression analysis (Tongzon, 1995). In recent years, however, significant progress has been made in the area of measuring the degree of relative efficiency in productive activities. To this end, two more complex, yet more appropriately holistic approaches, data envelopment analysis (DEA) and stochastic frontier analysis (SFA), have been increasingly deployed in efforts to analyse port production and performance. This paper adopts SFA as an appropriate analytical tool for measuring the relative efficiency of European container terminals. The ensuing section of this chapter provides an exposition of the SFA methodology. Section 3 deals with the operationalisation of the SFA methodology by precisely defining the input and output variables that are collected for driving the model that is implicit to the analysis. This section also outlines the nature, form and sources for the data collection exercise and goes on to present the model specification to be estimated and the assumptions that underpin the ensuing

88

KEVIN CULLINANE AND DONG-WOOK SONG

analysis. The results of applying SFA to the sample data are presented in Section 4 and the implications of the analysis, conclusions, limitations and further research opportunities are presented in Section 5.

2. METHODOLOGY: STOCHASTIC FRONTIER ANALYSIS A number of methods for estimating efficiency based on the concept of a frontier have been proposed over the past decade. Under such methods, efficient units are deemed to be those operating on the cost or production frontier, while inefficient units are those that, in the case of a production frontier, operate below the frontier or, in the case of a cost frontier, operate above the frontier. There are a number of reasons explaining why the use of frontier models is becoming increasingly widespread (Bauer, 1990).  An approach based on a production or cost frontier is consistent with the underlying economic theory of optimising behaviour.  Any deviation in operating location away from a frontier has a natural interpretation as a measure of the efficiency with which economic units pursue their technical or behavioural objectives.  Information about the structure of the frontier and about the relative efficiency of economic units has many policy applications. The greater use of efficiency models revolving around the frontier concept is motivated by a number of interests; in the structure of efficient production technology, in the divergence between observed and frontier operations and also by the concept of economic efficiency itself. The literature on frontier models originates in the seminal work of Farrell (1957), who proposed what has subsequently become a widely accepted framework for analysing economic efficiency in terms of realised deviations from an idealised frontier isoquant. In order to assess the extent to which an individual unit’s actual operation deviates from an efficient frontier, it is obviously an important first stage to be able to describe or define the precise location of that frontier. In fact, a clear dichotomy exists in the taxonomy for classifying the methods that may be employed to derive the specification of a frontier model. At the first level of this taxonomy, such models can be estimated through the application of either statistical or non-statistical methods. While the former body of methods necessarily involve making

Estimating the Relative Efficiency of European Container Ports

89

assumptions about the stochastic properties of the data, the latter does not. Another differentiation in taxonomy concerns the issue of whether the chosen method is parametric or non-parametric. In this case, the former body of methods imposes a particular functional form, while the latter does not. In fact, the non-parametric approach revolves around mathematical programming techniques, the most commonly applied of which is generically referred to as DEA.1 The parametric approach, on the other hand, involves the application of econometric techniques where efficiency is measured relative to a statistically estimated frontier production function. Econometric approaches have a strong policy orientation, especially in terms of assessing alternative industrial organisation or governance structures and in evaluating the efficiency of government and other public agencies. Mathematical programming approaches, on the other hand, have a much greater managerial decision-making orientation (Aigner & Schmidt, 1980; Fare, Grosskopt, & Lovell, 1994; Lovell, 1995). The policy orientation of the econometric approaches more closely supports the purpose of this paper, especially since they have a more solid grounding in economic theory (Forsund, Lovell, & Schmidt, 1980; Pitt & Lee, 1981; Bauer, 1990). In addition, several studies (e.g. Gong & Sickles, 1992; Oum & Waters, 1996) have compared the performance of alternative methods for measuring efficiency, focusing on the econometric method (in particular, the stochastic frontier model) and the mathematical programming method. As measured by the correlation coefficients and rank correlation coefficients between the true and estimated relative efficiencies, the results show that when the functional form of the econometric model is well specified, the stochastic frontier approach generally produces better estimates of efficiency than the latter approach, especially when measuring firm-specific efficiency where panel data are available. Most poignantly, Cullinane, Ji, and Wang (2005) have also found this to be the case when comparing the results from the application of both programming and econometric approaches to data from the container port industry. The econometric approach involves the specification of a parametric representation of technology that, at a subsidiary level of taxonomy, can be divided into two different forms of model, either deterministic or stochastic frontiers may be specified depending upon what assumptions are made concerning the underlying data. The early parametric frontier models (e.g. Aigner & Chu, 1968; Afriat, 1972) were classified as deterministic as a consequence of the fact that all economic units being analysed were assumed to share a common fixed form of frontier. Although there is an obvious need for some form of simplifying assumption to underpin any econometric

90

KEVIN CULLINANE AND DONG-WOOK SONG

analysis, this was widely held to be too much of an oversimplification that compromised the validity of the results obtained from the analysis. Fundamentally, this assumption ignores the real possibility that the observed performance of an economic unit under study may be affected by exogenous (i.e. random shock), as well as endogenous (i.e. inefficiency), factors. Whether these factors exert a favourable or an unfavourable influence or whether they are within or beyond the control of the economic unit itself, it is clearly a dubious and imprecise generalisation to attribute all these factors to a single disturbance term and to refer to their combined impact as ‘inefficiency’. As an alternative, the development and use of the stochastic frontier model has been motivated by the concept that deviations from a production or cost frontier might not be entirely under the control of the economic unit being studied (Greene, 1993). Working independently of each other, Aigner, Amemiya, and Poirier (1976), Aigner, Lovell, and Schmidt (1977) and Meeusen and van den Broeck (1977) all sought to construct a more logical and acceptable error structure than one that was purely one-sided, as was the case with the deterministic model form. The outcome of this body of research was the specification of a linear model for the frontier production function given by Y it ¼ f ðX it ; bÞ expðnit

uit Þ;

i ¼ 1; 2; . . . ; N; t ¼ 1; 2; . . . ; T

(1)

where Yit denotes the appropriate form of output for the ith firm at time t, Xit is a vector of inputs associated with the ith firm at time t and b is a vector of input coefficients for the associated independent variables in the production function. The component vit in the disturbance term represents a symmetric disturbance term that accounts for the random variation of the production function across economic units. This may be due, for example, not only to the effects of measurement and specification error, but also to the effects of exogenous shocks that occur both sporadically and randomly and which are beyond the control of the economic unit (e.g. the influences of weather conditions, geography or machine performance). The other component within the disturbance term, uit (X0), is a one-sided disturbance term and represents an economic unit’s ‘productive inefficiency’ relative to the stochastic production function. That the disturbance term uit is non-negative reflects the fact that output must either lie on or below its production frontier. Any observation may deviate from the deterministic kernel of the stochastic production function (Eq. 1). When it does, this arises from two sources: (i) symmetric random variation of the deterministic kernel f (Xit; b) across observations that is captured by the component vit and (ii) asymmetric variation (or productive

Estimating the Relative Efficiency of European Container Ports

91

inefficiency) captured by the component uit. The term uit measures productive inefficiency in the sense that it measures the shortfall of output Yit from that implied by its maximum frontier given by f (Xit; b) exp(vit). Deriving an estimate of the level of efficiency of an economic unit is undertaken by calculating the following: Y it f ðX it ; bÞexpðvit Þ

(2)

relative to the stochastic frontier f (X; b) exp(v). Irrespective of the outcome of this calculation, any estimate of a firm’s level of efficiency will not be consistent. This is because it will contain elements relating to statistical noise as well as to productive inefficiency. In addition, stochastic frontier models suffer from two other difficulties. One is the requirement for specific assumptions about the distributions underlying productive inefficiency and statistical noise (usually assumed to have a normal distribution). The most frequently defined distribution for the uit is the half-normal (i.e. uit|N(0, su2)|), though other distributional assumptions for the uit terms have been proposed. For example, the exponential (Aigner et al., 1977), the truncated normal (Stevenson, 1980) and the gamma distribution (Greene, 1980). However, as Bauer (1990, p. 41) suggests: ‘Stronger assumptions generate stronger results, but they strain one’s conscience more y. The appropriate structure to impose can only be determined by a careful consideration of the data and the characteristics of the industry under study’. The other potential criticism of an approach based on a stochastic frontier model is the required assumption that regressors (the input variables contained in the vector X) and productive inefficiency are independent. Since any firm or other economic unit that becomes aware of its level of inefficiency is then likely to take action (affecting its input choices) to ameliorate the source of such problems, the potential clearly exists for this to prove to be rather an unrealistic assumption.

3. OPERATIONALISING THE MODEL 3.1. Definition of Variables Ports come in a range of different sizes and can face great variety in the traffic mix, which they handle. In consequence, Braeutigam, Daughety, and Turnquist (1984) have suggested that the use of cross-sectional, time-series

92

KEVIN CULLINANE AND DONG-WOOK SONG

or even panel data may still fail to identify even the most basic of differences between ports. Because of this, it is possible that an analysis of port efficiency that is based on such data may lead to a misjudgement as to each port’s level of performance. Kim and Sachish (1986) propose, therefore, that to minimise the chances of this happening, the structure of production in ports should be estimated econometrically at the level of the single port or terminal, using appropriate data that has been derived from as homogeneous a sample as is feasible. In attempting to derive a port production function and to use this as a baseline for the measurement of port performance, the analysis conducted by Chang (1978) focused on the handling solely of general cargo. The study was founded on the assumption that port operations follow the conventional Cobb-Douglas case as expressed by: Y ¼ AK a Lb egT=L

(3)

where Y denotes annual gross earnings (in real terms), K is the real value of net assets in the port, L is the number of labourers utilised per year and the average number of employees per month each year and eg(T/L) is a proxy for technological improvement in which (T/L) shows the tonnage handled per unit of labour. Chang (1978) argued that in order to estimate a production function of the form shown in Eq. (3), the output of a port should be measured in terms of either total tonnage handled at the port or its gross earnings. With the objective of aiding policy makers in their assessment of the merits of different ship types, the model presented in Eq. (3) has been improved upon by Bendall and Stent (1987). A more elaborate method of estimating the production function of a port was later developed and applied by Liu (1995) who, making certain underpinning assumptions, econometrically estimated the production function of UK ports by employing frontier models such as that defined in Eq. (3). It has been argued by Dowd and Leschine (1990), among others, that the productivity of a container port or terminal crucially depends on the efficient use of factor inputs – specifically, labour, land and equipment (as the most important form of capital investment in a port). The measurement of terminal productivity, therefore, and its relative assessment against those of homogeneous units within the sample, provides a means of quantifying the efficiency with which these three resources are utilised. Bernard (1991) questioned the efficacy of focusing on the total tonnage handled at a container port or terminal as a measure of the output from its production process. His logic rests with the fact that the use of total tonnage handled would appear to be an illogical metric for the assessment of output

93

Estimating the Relative Efficiency of European Container Ports

at a container terminal or port. One reason for this is that the basic unit handled can conform to only a few possible standards that apply to containers. Another more compelling reason, however, is that irrespective of a container’s size and, more importantly, its weight, the production inputs required for the movement of any single container are more or less the same. This is so even when comparing full containers with empty containers. In consequence, the obvious solution to representing the output of a container port or terminal that is often applied in practice is to measure the throughput in terms of the number of container movements across the quayside or, alternatively, in terms of the monetary value of these movements as indicated by the revenue associated solely with this, as opposed to any other, aspect of a port’s operations or revenue base. In terms of a conventional categorisation of the inputs to any type of port, a typical expenditure structure over a given period of time is illustrated in Fig. 1. As a proxy for the capital input variable, the combined values of buildings and equipment (mainly cargo-handling equipment) accounts for 42% of total expenditure. Thus, the labour and capital costs of a port or terminal together comprise 95% of the total cost structure of a port or terminal operation. It seems a reasonable assumption that this is a sufficiently high proportion to describe virtually the whole cost account. For this reason, it was the original intention to fulfil the data requirements of this study by collecting values pertaining to the basic economic inputs of capital and labour. This was the approach adopted in Chang (1978), Bendall and Stent (1987) and Liu (1995), as well as in Song, Cullinane, and Roe (2001), where the focus was the estimation of the productive efficiency of the main Korean and UK container terminals. Unfortunately, across the European region in general, this sort of cost data proved impossible to collect from secondary sources. Unsurprisingly, there are inevitable and

equipment 15%

other 5%

labour 53% building s 27%

Fig. 1.

A Port/Terminal Expenditure Structure. Source: Derived from Sachish (1996, p. 347).

94

KEVIN CULLINANE AND DONG-WOOK SONG

numerous problems associated with overcoming the variations in, and consequent incomparability of, accounting statements in the different nations that together comprise Europe. In addition to this difficulty and in stark contrast to the situation in the UK, for example, accounting standards and conventions within the region exhibit a general tendency not to require the publication of costs at a high enough level of detail to allow their individual identification in the manner that would be necessary and desirable for this analysis. Difficulties in assessing the labour input to the port production process is a problem that has often been found, not only because of differences in accounting standards as previously described, but also due to the prevailing social policies and political environment within different countries (Cullinane & Khanna, 1999; Valentine & Gray, 2001). Instead, and in an effort to circumvent these problems, an alternative approach was adopted, which utilised certain key physical characteristics of the terminals as the required input data. These variables related to terminal quay length (X1), terminal area in hectares (X2) and the number of pieces (X3) of cargo-handling equipment (including gantry cranes, ship–shore gantries, yard cranes, mobile cranes, etc.). While these variables can clearly be seen as accounting for the important factor inputs to the port production process of land and capital, the labour factor appears to be ignored. In fact, in terms of the physical quantity of labour involved in the container port or terminal production process, there is virtually a fixed relationship between the amount of equipment deployed and the labour factor input employed in production. This corresponds directly to the approach adopted in a longitudinal study of U.S. East Coast port authorities by De Neufville and Tsunokawa (1981), where the inputs utilized in their analysis were simply quay length and the number of cranes employed. The authors argued that these inputs are actually proxies for all other inputs in that the labour input to the production process (both from dock workers and administrative or management personnel) is supplied in proportion to the number of cranes employed. Similarly, quay length is argued to be proportional to the amount of land dedicated to the marshalling and storage of containers. Turner, Windle, and Dresner (2004) also adopted this approach in their analysis of North American containerport productivity by restricting inputs solely to physical measures of containerport infrastructure and outputs to those produced by that infrastructure. Utilising DEA, the inputs to their model were total terminal land dedicated to container operations, total quayside container gantry cranes and total container berth length. They justified this approach by arguing that not only was there a generally applicable relationship between physical amounts of infrastructure and labour,

Estimating the Relative Efficiency of European Container Ports

95

but also that in North America differences in labour productivity between ports are minimal because of standardized collective bargaining agreements that establish the size of longshore labour gangs and related work rules. Similar assumptions on the existence of a relationship between the labour and capital inputs to the port production process have also been made by Tongzon (1995), Notteboom, Coeck, and van den Broeck (2000) and, probably most persuasively, in a recent World Bank working paper by Gonza´lez and Trujillo (2005). Thus, in common with many other studies of this type, the analysis reported herein assumes a high degree of correlation between the physical quantities of capital and labour factor inputs and, thereby, the omission of the labour input from the analysis is, to some extent, justified. From a pragmatic point of view, this overall approach has the distinct advantage that the data on these measures of physical container terminal capacities are readily available within the public domain and precedents do exist where they have been used for just this sort of analysis. Data on the productive output from each of the container terminals in the sample are also required in order to conduct an econometric estimation of efficiency using the stochastic frontier model. In Song et al. (2001), terminal output (Y) was defined as being solely the turnover derived from the delivery of container terminal services. Most critically, this obviously excludes revenue generated on the basis of, for example, property sales, a most lucrative source of European port revenue in the 1980s and 1990s. Again, due to the wide diversity of accounting systems employed by the nations in Europe and, therefore, by the sample of terminal operators, it proved to be a wholly intractable problem to consistently separate out the revenue attributable to different sources. In parallel with the solution proposed to fulfil the data requirements for production inputs, the readily accessible, physical measure of annual container throughput in TEUs2 was adopted as the basis for measuring the productive output of container terminals (Y). This approach also has its precedents (e.g. Bernard, 1991; Notteboom et al., 2000) though it would be preferable to use (but impossible to collect information on) the actual number of boxes handled.

3.2. Data Sources The 74 European container terminals selected for inclusion in the sample account for a significant proportion of European container traffic and represent a wide range of different ownership attributes. The data for 2002 were

96

KEVIN CULLINANE AND DONG-WOOK SONG

Table 1. Variablesa Y X1 X2 X3

Statistical Properties of the Variables.

Mean

Median

Min

Max

S.D.

398,835.00 946.70 37.28 37.05

210,742.00 730.00 20.00 27.50

11,286.00 100.00 1.50 3.00

2,896,835.00 3100.00 200.00 127.00

569,081.00 684.60 42.73 30.69

a

Y is defined as the terminal output as measured by annual container throughput in TEUs, X1 is defined as the terminal quay length in metres, X2 is defined as the terminal area in hectares, X3 is defined as the number of pieces of cargo handling equipment employed.

mainly collected from the Containerisation International Yearbook and Lloyds Ports of the World, but were validated and, in certain instances, supplemented by approaching each of the terminals directly. Variable definitions and descriptive statistics for the data are shown in Table 1. 3.3. Model Specification and Assumptions As is the case with all econometric analyses, certain assumptions have to be made that underpin the analysis and influence the interpretation of the results obtained. The primary objective of terminal operators is assumed to be the maximisation of the profits derived from container operations. In other words, a terminal operating company is regarded as a profit-maximiser that is assumed to be a price-taker in its input markets. The corollary of this assumption is that input prices may be treated as exogenous to the model. Another assumption necessary for operationalising the model is that the production function that is estimated relates to a single form of output. This is justified on the basis that the main operational function of container terminals and the main issue of policy interest is container handling. Thus, earnings from sources such as the sales of terminal property are not classified as output and do not affect the production frontier. The further assumption is made that the log-linear Cobb-Douglas function is an appropriate structure for the model. This, or some form of, assumption on the functional form of the production frontier is a necessary prerequisite to the estimation of relative terminal operator efficiency. The logarithmic stochastic frontier model specified for the container terminal operating sector in the cross-sectional case is, therefore, defined by In Y it ¼ In f ðX 1it ; X 2it ; X 3it ; bÞ þ nit

uit

i ¼ 1; 2; . . . ; 15; t ¼ 1; 2; . . . ; T (4)

Estimating the Relative Efficiency of European Container Ports

97

where Yit represents the output of the ith container terminal operator at time t. X1it, X2it and X3it denote the respective input variables associated with the ith terminal operator at time t and b is a vector of input coefficients associated with the independent variables in the model and is the object of estimation. The disturbance term vit represents the symmetric (statistical noise) component. Finally, uit ( X 0 ) is the one-sided (inefficiency) component. Since observations on the sample have been taken in only a single year (2002), the subscript t relating to time can be ignored in this instance. However, more generally, the cross-sectional case assumes the independence of observations of a single port or terminal in different time periods and treats all such observations as completely separate and actually as if they were part of a set of cross-sectional data. That is, just as if observations on a single port or terminal in different years were actually observations on different ports or terminals. Finally, the concept of the average terminal frontier is applied as the definition of the frontier. The estimation of terminal efficiency is conducted by applying the David-Fletcher–Powell algorithm (Fletcher, 1980; Greene, 1997) using the LIMDEP 8.0 econometric software package (Greene, 1995). Since the maximum likelihood method is a large-sample estimation procedure (Maddala, 1992), it is required that an asymptotic test statistic be used. Thus, since it is one of the general large-sample tests based upon maximum likelihood estimation (MLE), the likelihood ratio test statistic (LR) is applied to test whether or not model coefficients are significantly different from zero. Under general conditions, the LR has a w2 distribution with degrees of freedom equal to the number of restrictions imposed and can be expressed as follows (Engle, 1984):   LR LR ¼ 2 In (5) LU ywhere LR denotes the ‘restricted’ likelihood function and LU the ‘unrestricted’ likelihood function.

4. RESULTS The first step to take in applying the procedure for estimating a stochastic frontier model is to check the sign of the third moment, relating to the skewness of the ordinary least squares (OLS) residuals associated with the sample data (Waldman, 1982). The third moment for the terminal frontier model estimated on the basis of the sample data is 0.71. The negative sign

98

KEVIN CULLINANE AND DONG-WOOK SONG Histogram of the Residuals (response is ln Y) 25

Frequency

20

15

10

5

0

-3

-2

Fig. 2.

-1 Residual

0

1

Histogram of the OLS Residuals.

implies that the residuals from the model estimated for the sample data possess the correct pattern for the implementation of the MLE procedure. The degree of negative skewness can be clearly seen in the graphical depiction of the histogram of the residuals shown in Fig. 2. The estimation procedures that have been applied yield merely the residuals e, rather than the inefficiency term u. In fact, this term in the model must be observed indirectly (Greene, 1993). In the case of the cross-sectional model that is being applied herein and which is shown generically in Eq. (4), Jondrow, Lovell, Materov, and Schmidt (1982) suggest that the conditional expectation of uit, conditioned on the realised value of the error term eit ¼ (vit uit), is an appropriate estimator of uit. In other words, E ½uit jit Š is taken to be the ‘mean productive inefficiency’ for the ith terminal operator in the industry at any time t. Since there are three assumed possible distributional forms for the inefficiency term in the model, what this means is that under each of these individual cases:  For the half-normal model " fðitslÞ sl E½uit jit Š ¼ ð1 þ l2 Þ Fð itslÞ

it l s

#

(6)

99

Estimating the Relative Efficiency of European Container Ports

 For the exponential model

E ½uit jit Š ¼ it

h i ð ys2 Þ sv f it sv v
ys2v þ h i ð ys2 Þ F it sv v

(7)

 For the truncated normal model the inefficiency term is obtained merely by replacing ðit lÞ=s in Eq. (6) with:   it l m þ s sl

(8)

Based on the sample cross-sectional data, the OLS estimates and the MLEs for each of the three assumed distributions of the inefficiency term in the frontier model (4) are shown in Table 2.

Table 2.

Estimation Results of Frontier Production Functions.

Variables/ Parameters

OLS

MLE Half-normal

Constant ln X1 ln X2 ln X3 l s2v s2u y m Log-likelihood R2 F-value LR

8.478* (7.28) 0.111 (0.48) 0.302* (2.14) 0.621* (3.22) – – – – – – 0.514 24.71

9.689* (14.713) 0.028 (0.189) 0.310* (2.988) 0.670* (5.735) 3.640 0.058 0.765 – – 118.096 – –

Exponential 11.447* (18.905) 0.324 ( 1.880) 0.100 (0.803) 1.235* (7.073) – 0.000 0.729 1.171 – 108.202 – – 166.58

Truncated Normal 9.745* (28.187) 0.000 ( 0.009) 0.312* (4.602) 0.698* (7.778) 53.627 0.039 111.005 – 210.582 199.130 – –

Notes: (1) *Not significant at the 1% level. (2) Figures in parentheses indicate t-ratios.

100

KEVIN CULLINANE AND DONG-WOOK SONG

The estimated OLS coefficients are only of limited value, but they do provide a starting point for the MLE process. The goodness of fit of the estimated regression equation evaluated by R2 for the least squares method looks mediocre at 0.514. This implies that the four inputs to the model satisfactorily explain approximately 51% of the variation in the model output. In addition, the F-statistic of 24.71 reveals that the relationship between mean squares due to the regression (endogenous factors) and due to the error (exogenous factors) is significant at the 1% level. The results also show that only the first input (X1), relating to the total terminal quay length in metres, as well as X2 under the exponential form of the model, are statistically significant. All the other variables are not statistically significant. To some extent, this may reflect the relatively high degree of multicollinearity between the independent variables. Another outcome is that the MLEs under the three alternative inefficiency distributions yield parameters that are comparatively close to each other. The signs of the parameters, particularly relating to the significant variable X1, do not conform to a priori expectations, but this again is probably due to multicollinearity. Another interesting point is that, except for the estimates produced from the exponential form of the model, the MLEs differ only marginally from the OLS estimates. This is to be expected since both methods are consistent. Overall, however, the likelihood ratio test statistic of 166.58 reveals a high degree of significance beyond the 1% level, thus leading to the rejection of the null hypothesis that the coefficients are equal to zero.3 The parameter l (as calculated by su /sv) provides some insight into the relative variance of the two composite errors that make up the total variation in the structural disturbance term. The two variances of the two error components, in addition to l, indicate that the inefficiency component u varies more widely than the uncontrollable random exogenous component v. This means that the productive inefficiency u makes a more important contribution to the variability of the total error in the cross-sectional frontier model. Having reviewed the diagnostics associated with the model’s overall estimation, the individual estimates of relative efficiency for each port or terminal within the sample are shown in Table 3. Specific relative efficiency estimates for the ports or terminals within the sample that have been identified as most and least efficient, under the assumption of a half-normal distribution, are shown in Figs. 3 and 4. Under all three assumed distributions for which the model parameters have been estimated, the average efficiency level of Giao Tauro in Italy is consistently highest. Although with less consistency under each of the three

101

Estimating the Relative Efficiency of European Container Ports

Table 3.

SFA Productive Efficiency Estimates (%).

Container Port/Terminal Code 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

Half-Normal

Exponential

Truncated Normal

41.14 67.15 25.83 73.76 76.64 31.67 38.84 37.25 30.91 25.48 71.62 37.40 25.48 63.17 34.28 32.98 25.48 66.99 77.41 89.29 79.14 44.13 76.64 70.08 39.69 41.36 67.27 45.18 37.70 74.60 90.90 86.45 59.09 25.48 25.48 25.48 30.02 43.77 52.58 60.73 33.69 25.48 87.08 25.48

38.64 37.23 16.45 78.11 43.88 20.01 19.43 33.40 19.90 1.21 27.80 19.43 8.52 53.92 12.02 23.66 1.71 44.50 48.75 68.11 46.80 13.31 46.18 36.78 13.01 29.20 53.56 18.89 11.75 42.49 95.80 51.43 26.29 9.83 14.39 1.45 15.31 16.23 37.89 25.81 23.26 11.61 97.09 1.16

45.45 73.91 31.38 76.05 82.91 32.81 40.59 39.71 31.38 31.38 79.06 39.24 31.38 70.01 35.16 34.94 31.38 73.45 83.28 93.00 84.89 45.46 83.44 76.15 40.62 44.30 74.10 47.74 38.34 80.51 94.23 90.98 63.43 31.38 31.38 31.38 31.38 46.00 56.62 64.99 34.76 31.38 91.90 31.38

102

KEVIN CULLINANE AND DONG-WOOK SONG

Table 3. (Continued ) Container Port/Terminal Code 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74

Half-Normal

Exponential

Truncated Normal

69.35 69.75 25.48 31.69 78.52 46.43 53.24 36.52 25.48 57.56 32.31 38.31 80.10 55.88 54.65 78.91 32.08 59.46 89.21 27.59 25.48 36.68 69.34 84.13 68.34 74.89 78.17 80.10 78.54 36.94

30.57 42.74 8.56 15.08 52.18 24.37 31.06 19.99 9.92 22.44 20.03 23.34 51.62 34.94 31.82 70.62 11.16 60.90 76.07 14.24 5.57 10.85 35.48 55.63 15.67 43.06 38.37 40.80 24.18 16.46

74.50 76.09 31.38 32.17 85.02 49.42 58.15 38.14 31.38 61.19 33.38 40.10 85.98 60.31 59.51 86.30 32.54 66.43 93.11 31.38 31.38 36.47 76.12 89.10 72.36 80.78 83.69 85.68 83.62 38.34

Notes: Figures are calculated by converting the inefficiency estimates using exp( u).

distributions, Felixstowe in the UK, La Spezia in Italy, Tarragona in Spain, the Valencia Container Terminal in Spain and the North Sea Terminal of the port of Bremen/Bremerhaven in Germany all tended to appear in the top few ranking positions. The container ports of Haydarpasa in Turkey, Gdansk in Poland, Flushing in the Netherlands, Palermo in Italy and the Umex Container Terminal in Constantza in Rumania are consistently found to be among the most inefficient ports in the sample, even though again their precise rankings do not correlate perfectly across all the distributional assumptions imposed.

103

Estimating the Relative Efficiency of European Container Ports 100 Relative Efficiency (%)

90 80 70 60 50 40 30 20 10 Bremen/Bremerhaven 1

Southampton

Leixoes

Tilbury

Tarragona

Hamburg 1

Thamesport

Algeciras

Felixstowe

La Spezia

Valencia 3

Bremen/Bremerhaven 2

Giao Tauro

0

Port or Terminal

Fig. 3.

Ports/Terminals with the Highest Efficiency Estimates (Half-Normal Distribution).

When analysing the consistency of efficiency estimates yielded by applying different specifications of the stochastic frontier model, most previous research has found no statistically significant differences between estimates (e.g. Cummins & Zi, 1998), while a few have (e.g. De Borger & Kerstens, 1996; Kim & Schmidt, 2000). However, when the rankings of efficiency estimates derived from different SFA specifications are considered, the degree of correlation is invariably overwhelmingly high (e.g. Gong & Sickles, 1992). Differences in estimates have been found much more likely to be statistically significant when comparing the estimates derived from applying SFA and other methodologies (e.g. Bravo-Ureta & Rieger, 1990; De Borger & Kerstens, 1996; Sharma, Leung, & Zaleski, 1997; Cummins & Zi, 1998; Wadud & White, 2000; Bru¨mmer, 2001). Overall, the efficiency estimates derived under the three distributional assumptions utilised in this analysis appear to be quite highly correlated. As one might expect, there is a very high degree of correlation between the estimates derived from the half-normal and truncated normal distributions, with a correlation coefficient of 0.996 (with a 95% confidence interval of

104

KEVIN CULLINANE AND DONG-WOOK SONG 100

Relative Efficiency (%)

90 80 70 60 50 40 30 20

Haydarpasa

Constantza 2

Gdansk

Flushing (Vlissingen)

Klaipeda

Palermo

Leghorn 3

Leghorn 2

Sete

Le Havre 2

Turku

Varna

Stockholm

Ravenna

0

Kotka

10

Port or Terminal

Fig. 4.

Ports/Terminals with the Lowest Efficiency Estimates (Half-Normal Distribution).

0.994 0.997). Between the half-normal and exponential, there is a correlation of 0.846 (with a 95% confidence interval of 0.760 0.903) and between the truncated normal and the exponential the correlation coefficient is 0.838 (with a 95% confidence interval of 0.748 0.897). All these seem to suggest that the efficiency estimates derived from the application of the stochastic frontier model are relatively robust to the distributional assumptions made, even though the estimations from the exponential distribution appear to be quite low at the bottom end of the efficiency scale compared to those yielded under the other two distributional assumptions.4 The robustness of the results under different distributional assumptions is even more evident when considering the rank order of efficiency estimates under the three distributions. The Spearman’s rank order correlation coefficient (r) for the ranked estimates produced under the half-normal distribution compared to those derived under the exponential assumption is 0.888 (with a 95% confidence interval of 0.823 0.930). The same calculated statistic for the half-normal distribution compared to the truncated normal distribution is 0.995 (with a

105

Estimating the Relative Efficiency of European Container Ports

95% confidence interval of 0.992 0.996) and the same statistic for the truncated normal distribution versus the exponential distribution is 0.905 (with a 95% confidence interval of 0.849 0.940). This finding has important implications for the relevance of any benchmarking exercise, which any given economic unit from within the sample might engage in following the estimation of its level of relative efficiency. In order to assess the impact of the general approach that might be taken to the derivation of the production frontier and, ultimately therefore, to the efficiency estimates, the results obtained for the analysis undertaken within this study are compared to the results obtained from adopting the programming approach of DEA. The approach and findings from applying DEA to the same database as analysed within this study are fully reported in Cullinane and Wang (2005). Comparing the results obtained from applying the two standard DEA model forms with those obtained herein from applying three distributional assumptions, the degree of correlation is found to be highly significant (see Table 4). Because the degree of correlation between the relative efficiency rankings derived from applying the two DEA models in Cullinane and Wang (2005) was virtually perfect (i.e. approaching unity), for the purpose of assessing the degree of correlation between the efficiency rankings produced by both SFA and DEA approaches, it is necessary to take into account only the rankings produced by any one of the two DEA models. The Spearman’s rank order correlation coefficient between the rankings produced by the assumption of a half-normal distribution under SFA and the rankings that result from a DEA model is 0.799 (with a 95% confidence interval of 0.691– 0.872). The equivalent statistic for the exponential distribution versus the Table 4.

Correlation Coefficients between SFA and DEA Efficiency Estimates.

DEA Model

SFA Distributional Assumption Half-normal

DEA–CCR

DEA–BCC

0.753 (0.625 (0.576 0.759 (0.634 (0.585

 95% confidence interval.  99% confidence interval.

0.841) 0.862) 0.845) 0.866)

Exponential 0.593 (0.410 (0.344 0.602 (0.422 (0.356

0.730) 0.764) 0.736) 0.770)

Truncated Normal 0.742 (0.610 (0.559 0.749 (0.620 (0.570

0.834) 0.856) 0.838) 0.860)

106

KEVIN CULLINANE AND DONG-WOOK SONG

DEA model is 0.687 (with a 95% confidence interval of 0.534–0.796) and for the truncated normal distribution versus the DEA model it is 0.797 (with a 95% confidence interval of 0.688–0.871). Again, this provides some support for the robustness of the analysis, at least in rankings if not in levels of relative efficiency, irrespective of even the general modelling approach that is adopted. A further intriguing implication that can be drawn from the results is that the estimated relative efficiency of a container port or terminal is found to be significantly and positively correlated to its size, as measured in terms of container throughput (r ¼ 0.57, with a 95% confidence interval of: 0.381– 0.713). In other words, larger container ports or terminals have a tendency to be more efficient than their smaller counterparts. This finding reinforces previous empirical work where the same finding emerged (for example, Cullinane & Khanna, 1999; Cullinane & Wang, 2005). The closeness of the relationship can be inferred from Fig. 5. By combining and averaging the estimated efficiencies of the individual ports or terminals that together comprise a particular part of Europe, an evaluation can be made of any regional disparities that may exist within it. The outcome of such an exercise is presented in Fig. 6, where the high degree of correlation between the results derived from the different distributional assumptions underpinning the SFA can be quite clearly seen. Given the 100 90 Relative Efficiency (%)

80 70 60 50 40 30 20 10 0 0.0

0.5

1.0

1.5 2.0 2.5 Throughput (million TEU)

3.0

3.5

Fig. 5. The Relationship between Estimated Relative Efficiency and Throughput.

107

Estimating the Relative Efficiency of European Container Ports 100.00 Average Relative Efficiency (%)

90.00 Half-Normal

Exponential

Truncated Normal

Overall

80.00 70.00 60.00 50.00 40.00 30.00 20.00 10.00 0.00 West Europe

South east Europe

South Europe

Central Europe

East Europe

Scandinavia

British Isles

Region of Europe

Fig. 6.

Average Estimated Relative Efficiency by Region of Europe.

already reported significant level of correlation between the efficiency estimates derived within this analysis and those of Cullinane and Wang (2005) using DEA applied to the same database, it should come as no surprise that these regional averages are also very highly correlated (r ¼ 0.90, with a 95% confidence interval of: 0.842–0.937) with similar measures in Cullinane and Wang (2005). The British Isles emerges as having by far the most efficient infrastructure usage; a finding that could be explained by the fact that England, in particular, has in recent years been experiencing a chronic shortage of container-handling capacity in ports. While the demand has increased from both ocean and short-sea shipping, potential sites for the expansion of existing facilities are in short supply. At the same time, some major development projects have failed to receive the required government approval on environmental grounds. Many British container ports and terminals are, therefore, operating at close to, or sometimes beyond, their design capacity. Irrespective of current levels of productive efficiency, this is clearly a feature that will inevitably (if it has not already done so) impact upon congestion levels, customer (shipping line) perception and, therefore, future levels of demand. It is important to recognise that the effect of such circumstances on efficiency estimates derived from applying SFA is likely to be greatly exacerbated when the analysis is based on cross-sectional data as herein.

108

KEVIN CULLINANE AND DONG-WOOK SONG

At the same time, Scandinavia and East Europe emerge as the regions with ports or terminals yielding the lowest average level of relative efficiency. However, given some of the inferences that have already been drawn, it is difficult to attribute this simply to a sub-optimal governance structure or to an inferior level of management ability. Geographical location itself may play an important part in explaining this result. As has already been shown, the results of this analysis suggest that there exists a significant relationship between size and efficiency and ports in these two regions tend to be below the average size for Europe as a whole. However, it could be conjectured that one reason for this is that these areas are geographically displaced from the mainline intercontinental (east–west) container trades and that inevitably, therefore, this has an impact on their throughput levels (size).

5. CONCLUSIONS, LIMITATIONS AND FURTHER RESEARCH The results achieved from this application of the stochastic frontier model provide evidence of the levels of relative efficiency of the major container ports and terminals within Europe. These results suggest that, at least in 2002, significant inefficiency pervaded the European container-handling industry. In addition, the average levels of efficiency differ quite considerably across different parts of Europe, with Scandinavia and East Europe performing least well. It is difficult to conceive that, in aggregate, this situation would have dramatically altered during the interim period, especially since relative efficiency would appear to be influenced by the related factors of production scale and geographical proximity to mainline container trades. The detailed findings suggest that the most efficient ports or terminals generally tend to exhibit the common characteristic of having a relatively large throughput compared to others within the sample and that this level of operational throughput equates very closely to design capacity. In the face of physical or regulatory constraints on a port or terminal’s propensity to increase supply in response to consistently rising demand over recent years, this implies that many ports or terminals in Europe are rising to this challenge by reaping efficiency gains in the utilisation of resources, particularly land and infrastructure. Nevertheless, the empirical finding that efficiency is related positively to scale is likely to further reinforce the dominant market positions of certain ports and terminals in the region. From a methodological perspective, a comparison of the efficiency estimates derived in this study across a range of distributional assumptions

Estimating the Relative Efficiency of European Container Ports

109

seems to suggest that SFA produces robust results. This is also the case when the results are compared to those derived from applying DEA, an alternative methodology based on programming techniques, and which are reported in Cullinane and Wang (2005). One more potentially intervening variable that could have influenced the results achieved within this study is the level of private sector involvement in, and market regulation of, container terminal operations. Simultaneously controlling for each of the individual effects of both these factors will allow greater fine-tuning in policy assessment and/or formulation. This analysis has assumed that the amount of labour as a factor input to the port production process is very closely correlated with the amount of equipment. This simplifying assumption may well be very valid and certainly renders the problem more analytically tractable than otherwise would be the case. However, even if the quantity of labour input per unit of equipment is more or less constant in practice, the unit price of that labour across Europe most certainly is not. Similarly, although this study has implicitly assumed that there is a standard unit of equipment, in practice the quality of equipment is a critical determinant in port efficiency. The level of capital investment in, or maintenance expenditure on, equipment over a given period of time would provide a reasonable proxy for its quality. The monetarisation of capital and labour inputs would constitute, therefore, a potential fruitful avenue for extending this research, although it would be difficult to collect the required data and make it comparable between countries. An important limitation of this research is the cross-sectional nature of the data. The analysis of this form of data means that only a snapshot of relative efficiency can be obtained. Efficiency changes over time and, in consequence, there is a need to implement some form of dynamic analysis using panel data. Most critically, the lumpy nature of investment in port infrastructure means that inefficiency will occur immediately following investment in facilities that is intended to cater for future growth in their use. Thus, recent or imminent investments are likely to have a significant deleterious impact on measures of relative efficiency that are based on crosssectional data. Apart from the natural smoothing over time of the impact of lumpy investments in the port industry, Hausman and Taylor (1981), Baltagi (1995) and Blundell (1996) identify a number of other attractions to using estimation techniques that are based on panel, as opposed to crosssectional, data.  Using panel data allows for the control of individual effects, which may be correlated with other variables included in the specification of an

110

KEVIN CULLINANE AND DONG-WOOK SONG

economic relationship; a possibility that complicates any analysis undertaken on the basis of cross-sectional data.  Consistent estimates of the productive efficiency of an economic unit can be obtained as the number of time periods tends to infinity. As a result, strong distributional assumptions are not necessary.  Model parameters and levels of relative efficiency can be estimated without assuming that the input variables are uncorrelated with productive inefficiency. Therefore, as Schmidt and Sickles (1984) note, a variety of different estimates will be considered, depending on what assumptions are made about the distribution of productive inefficiency and its potential correlation with the regressors.  Reinforcing the benefits of extending this analysis by utilising panel data, time-series variation in the sample will facilitate the assessment of the dynamic impacts of changes in port policies. For example, the effects of greater deregulation and/or private sector involvement. Other avenues for future research lie with  developing appropriate clusters of ports in the sample that can be benchmarked against one another in order to identify sources of inefficiency and measures for its amelioration;  expanding the dataset of input variables to take into account environmental or instrumental (contextual) variables, such as geographical proximity to mainline trades, accessibility to other ports via the liner shipping network (Wang & Cullinane, 2006) and hinterland locations via the general transport network serving the port, etc.;  identifying the causal factors that affect port efficiency. In this respect, precedent has been set by Turner et al. (2004) in approaching this problem by utilizing Tobit regression analysis; and  utilising port efficiency estimates, in tandem with a number of other variables, to explain port choice within a discrete choice modelling methodology. The ultimate aim of this would be to explain and predict port competitiveness and market share.

NOTES 1. For a general overview of the DEA approach, see Charnes, Cooper, and Rhodes (1978), Banker, Charnes, and Cooper (1984) and Thanassoulis (2001). In recent years, there have been a number of applications of the DEA programming approach to the port sector. These include, but are not limited to, Martinez-Budria, Diaz-Armas, Navarro-Ibanez, and Ravelo-Mesa (1999), Tongzon (2001), Valentine

Estimating the Relative Efficiency of European Container Ports

111

and Gray (2001), Itoh (2002), Barros (2003a, 2003b), Barros and Athanassiou (2004), Turner et al. (2004), Park and De (2004), Bonilla, Casasus, Medal, and Sala (2004), Cullinane, Song, Ji, and Wang (2004), Wang (2004), Cullinane, Ji, and Wang, (2005) and Cullinane et al.(2005), Wang, Cullinane, and Song (2005), Tongzon and Heng (2005), Cullinane, Song, and Wang (2006). 2. Twenty-foot Equivalent Unit; a standard size of container used for denoting the container carrying capacity of container ships. 3. The Likelihood Ratio test statistic is calculated as LR ¼ 2( 187.56+ 104.27) ¼ 166.58. 4. It can also be seen that the lowest estimate of efficiency stabilizes at 25.48 and 31.38, respectively, under the half-normal and truncated normal distributions. While it is not entirely clear why this should be the case, it is probably due to some asymptotic property associated with the solution algorithm employed within the estimation software.

REFERENCES Afriat, S. (1972). Efficiency estimation of production functions. International Economic Review, 13(3), 568–598. Aigner, D., Amemiya, T., & Poirier, D. (1976). On the estimation of production frontiers: Maximum likelihood estimation of the parameters of a discontinuous density function. International Economic Review, 17(2), 377–396. Aigner, D., & Chu, S. (1968). On estimating the industry production function. American Economic Review, 58(4), 826–839. Aigner, D., Lovell, C., & Schmidt, P. (1977). Formulation and estimation of stochastic frontier production function models. Journal of Econometrics, 6(1), 21–37. Aigner, D., & Schmidt, P. (1980). Specification and estimation of frontier production, profit and cost functions, editors’ introduction. Journal of Econometrics, 13(Suppl.), 1–3. Ashar, A. (1997). Counting the moves. Port development international, 13(November), 25–29. Baltagi, B. (1995). Econometric analysis of panel data. Chichester: Wiley. Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models for estimating technical and scale inefficiencies in data envelopment analysis. Management Science, 30, 1078–1092. Barros, C. P. (2003a). Incentive regulation and efficiency of Portuguese Port Authorities. Maritime Economics and Logistics, 5(1), 55–69. Barros, C. P. (2003b). The measurement of efficiency of Portuguese Sea Port Authorities with DEA. International Journal of Transport Economics, XXX(3), 335–354. Barros, C. P., & Athanassiou, M. (2004). Efficiency in European Seaports with DEA: Evidence from Greece and Portugal. Maritime Economics and Logistics, 6(2), 122–140. Bauer, P. (1990). Recent developments in the econometric estimation of frontiers. Journal of Econometrics, 46, 39–56. Baumol, W., Panzar, J., & Willig, R. (1982). Contestable markets and the theory of industrial structure. San Diego: Harcourt Brace Jovanovich. Bendall, H., & Stent, A. (1987). On measuring cargo handling productivity. Maritime Policy and Management, 14(4), 337–343.

112

KEVIN CULLINANE AND DONG-WOOK SONG

Bernard, J. (1991). European deep-sea container terminals: Locational and operational perspectives, unpublished doctoral Thesis, University of Liverpool, Liverpool. Blundell, R. (1996). Microeconometrics. In: D. Greenaway, M. Bleaney & I. Stewart (Eds), A Guide to Modern Economics (pp. 508–536). London: Routledge. Bonilla, M., Casasus, T., Medal, A., & Sala, R. (2004). An efficiency analysis of the Spanish Port System. International Journal of Transport Economics, XXXI(3), 379–400. Braeutigam, R., Daughety, A., & Turnquist, M. (1984). A firm specific analysis of economies of density in the US railroad industry. Journal of Industrial Economics, 33, 3–20. Bravo-Ureta, B. E., & Rieger, L. (1990). Alternative production frontier methodologies and dairy farm efficiency. Journal of Agricultural Economics, 41, 215–226. Bru¨mmer, B. (2001). Estimating confidence intervals for technical efficiency: The case of private farms in Slovenia. European Review of Agricultural Economics, 28, 285–306. Chang, S. (1978). Production function, productivities and capacity utilisation of the port of mobile. Maritime Policy and Management, 5, 297–305. Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of Operational Research, 2, 429–444. Cullinane, K. P. B., Ji, P., & Wang, T.-F. (2005). The relationship between privatization and DEA estimates of efficiency in the container port industry. Journal of Economics and Business, 57(5), 433–462. Cullinane, K. P. B., & Khanna, M. (1999). Economies of scale in large container ships. Journal of Transport Economics and Policy, 33(2), 185–208. Cullinane, K. P. B., & Khanna, M. (2000). Economies of scale in large containerships: Optimal size and geographical implications. Journal of Transport Geography, 8, 181–195. Cullinane, K. P. B., Song, D.-W., Ji, P., & Wang, T.-F. (2004). An application of DEA windows analysis to container port production efficiency. In: W. K. Talley (Ed.), The Industrial Organization of Shipping and Ports, Review of Network Economics (special issue), III(2), 186–208. Cullinane, K. P. B., Song, D.-W., & Wang, T.-F. (2006). The application of mathematical programming approaches to estimating container port production. Journal of Productivity Analysis, 24(1), 73–92. Cullinane, K. P. B., & Wang, T.-F. (2005). The efficiency of European container ports: A crosssectional data envelopment analysis. International Journal of Logistics: Research and Applications, 9(1), 19–31. Cullinane, K. P. B., Wang, T.-F., Song, D.-W., & Ji, P. (2005). A comparative analysis of DEA and SFA approaches to estimating the technical efficiency of container ports. Transportation Research A: Policy and Practice, 40(4), 354–374. Cummins, J. D., & Zi, H. (1998). Comparison of frontier efficiency methods: An application to the US life insurance industry. Journal of Productivity Analysis, 10, 131–152. De Borger, B., & Kerstens, K. (1996). Cost efficiency of Belgian Local Governments: A comparative analysis of FDH, DEA and econometric approaches. Regional Science and Urban Economics, 26, 145–170. De Monie, G. (1987). Measuring and evaluating port performance and productivity. Geneva: UNCTAD Monograph No. 6 on Port Management. De Neufville, R., & Tsunokawa, K. (1981). Productivity and returns to scale of container ports. Maritime Policy and Management, 8(2), 121–129. Dowd, T., & Leschine, T. (1990). Container terminal productivity: A perspective. Maritime Policy and Management, 17(2), 107–112.

Estimating the Relative Efficiency of European Container Ports

113

Engle, R. (1984). Wald, likelihood ratio and Lagrange multiplier tests in econometrics. In: Z. Griliches & M. Intriligator (Eds), Handbook of Econometrics (Vol. II, pp. 775–826). Amsterdam: Elsevier Science Publishers. Fare, R., Grosskopf, S., & Lovell, C. (1994). Production frontiers. Cambridge: Cambridge University Press. Farrell, M. (1957). The measurement of productive efficiency. Journal of the Royal Statistical Society, 120(3), 253–290. Fletcher, R. (1980). Practical methods of optimisation. New York: Wiley. Forsund, F., Lovell, C., & Schmidt, P. (1980). A survey of frontier production functions and of their relationship to efficiency measurement. Journal of Econometrics, 13, 5–25. Gong, B. H., & Sickles, R. (1992). Finite sample evidence on the performance of stochastic frontiers and data envelopment analysis using panel data. Journal of Econometrics, 51, 259–284. Gonza´lez, M. M., & Trujillo, L. (2005). Reforms and infrastructure efficiency in Spain’s container ports. World Bank Policy Research Working Paper 3515, February. Greene, W. (1980). Maximum likelihood estimation of econometric frontier functions. Journal of Econometrics, 13, 27–56. Greene, W. (1993). The econometric approach to efficiency analysis. In: H. Fried, C. Lovell & S. Schmidt (Eds), The measurement of productive efficiency: Techniques and applications (pp. 68–119). New York: Oxford University Press. Greene, W. (1995). LIMDEP user’s manual. Castle Hill, Australia: Econometric Software, Inc. Greene, W. (1997). Econometric analysis (3rd ed.). New Jersey: Prentice-Hall International. Hausman, J., & Taylor, W. (1981). Panel data and unobserved individual effects. Econometrica, 49, 1377–1398. Itoh, H. (2002). Efficiency changes at major container ports in Japan: A window application of data envelopment analysis. Review of Urban and regional Development Studies, 14(2), 133–152. Jondrow, J., Lovell, C., Materov, I., & Schmidt, P. (1982). On the estimation of technical inefficiency in the stochastic frontier production model. Journal of Econometrics, 19, 233–238. Kim, M., & Sachish, A. (1986). The structure of production, technical change and productivity in a port. Journal of Industrial Economics, 35(2), 209–223. Kim, Y., & Schmidt, P. (2000). A review and empirical comparison of Bayesian and classical approaches to inference on efficiency levels in stochastic frontier models with panel data. Journal of Productivity Analysis, 14(2), 91–118. Liu, Z. (1995). The comparative performance of public and private enterprises: The case of British ports. Journal of Transport Economics and Policy, 29(3), 263–274. Lovell, C. (1995). Econometric efficiency analysis: A policy-oriented review. European Journal of Operational Research, 80(3), 452–462. Maddala, G. (1992). Introduction to econometrics (2nd ed.). New York: Macmillan Publishing Company. Martine, Z. I., Perez, G. E., San Juan, L. E., & Suarez, B. C. (2004). How important are transport costs for international trade? An empirical study for Spanish exporting sectors. International Association of Maritime Economists Annual Conference, Izmir, 30 June – 2 July. Martinez-Budria, E., Diaz-Armas, R., Navarro-Ibanez, M., & Ravelo-Mesa, T. (1999). A study of the efficiency of Spanish port authorities using data envelopment analysis. International Journal of Transport Economics, XXVI(2), 237–253.

114

KEVIN CULLINANE AND DONG-WOOK SONG

Meeusen, W., & van den Broeck, J. (1977). Efficiency estimation from Cobb-Douglas production functions with composed error. International Economic Review, 18(2), 435–444. Notteboom, T. (1997). Concentration and load centre development in the European container port system. Journal of Transport Geography, 5(2), 99–115. Notteboom, T., Coeck, C., & van den Broeck, J. (2000). Measuring and explaining the relative efficiency of container terminals by means of Bayesian stochastic frontier models. International Journal of Maritime Economics, 2(2), 83–106. Oum, T. H., & Waters, W. (1996). A survey of recent development in transportation cost function research. Logistics and Transportation Review, 32(4), 423–463. Park, R.-K., & De, P. (2004). An alternative approach to efficiency measurement of seaports. Maritime Economics and Logistics, 6, 53–69. Pitt, M., & Lee, L. (1981). The measurement and sources of technical efficiency in the Indonesian weaving industry. Journal of Development Economics, 9, 43–64. Sachish, A. (1996). Productivity functions as a managerial tool in Israeli Ports. Maritime Policy and Management, 23(4), 341–369. Schmidt, P., & Sickles, R. (1984). Production frontiers and panel data. Journal of Business and Economic Statistics, 2(4), 367–374. Sharma, K. R., Leung, P., & Zaleski, H. M. (1997). Productive efficiency of the swine industry in Hawaii: Stochastic frontier vs data envelopment analysis. Journal of Productivity Analysis, 8, 447–459. Song, D.-W., Cullinane, K. P. B., & Roe, M. (2001). The productive efficiency of container terminals: An application to Korea and the UK. London: Ashgate. Stevenson, R. (1980). Likelihood functions for generalised stochastic frontier estimation. Journal of Econometrics, 13, 57–66. Tabernacle, J. B. (1995). A study of the changes in performance of quayside container cranes. Maritime Policy and Management, 22(2), 115–124. Talley, W. K. (1988). Optimum throughput and performance evaluation of marine terminals. Maritime Policy and Management, 15(4), 327–331. Thanassoulis, E. (2001). Introduction to theory and application of data envelopment analysis. Norwell, Mass: Kluwer Academic Publishers. Tongzon, J. (1995). Determinants of port performance and efficiency. Transportation Research A: Policy and Practice, 29(3), 245–252. Tongzon, J. (2001). Efficiency measurement of selected Australian and other international ports using data envelopment analysis. Transportation Research A: Policy and Practice, 35(2), 113–128. Tongzon, J., & Heng, W. (2005). Port privatization, efficiency and competitiveness: Some empirical evidence from container ports (terminals). Transportation Research A: Policy and Practice, 39(5), 405–424. Turner, H., Windle, R., & Dresner, M. (2004). North American containerport productivity: 1984–1997. Transportation Research E: Logistics and Transportation Review, 40, 339–356. Valentine, V. F., & Gray, R. (2001). The measurement of port efficiency using data envelopment analysis. Proceedings of the 9th World Conference on Transport Research, 22–27 July, Seoul, South Korea. Wadud, A., & White, B. (2000). Farm household efficiency in Bangladesh: A comparison of stochastic frontier and DEA methods. Applied Economics, 32, 1665–1673. Waldman, D. (1982). A stationary point for the stochastic frontier likelihood. Journal of Econometrics, 18, 275–279.

Estimating the Relative Efficiency of European Container Ports

115

Wang, T.-F. (2004). Analysis of the container port industry using efficiency measurement: A comparison of China with its international counterparts. Unpublished doctoral thesis. The Hong Kong Polytechnic University, Hong Kong. Wang, T.-F., & Cullinane, K. P. B. (2004). Industrial Concentration in Container Ports. International Association of Maritime Economists Annual Conference, Izmir, 30 June–2 July. Wang, T.-F., Cullinane, K. P. B., & Song, D.-W. (2005). Container port production and economic efficiency. Basingstoke: Palgrave–Macmillan. Wang, Y., & Cullinane, K. P. B. (2006). Inter-port competition and measures of individual container port accessibility. International Association of Maritime Economists Annual Conference, Melbourne, 12–14 July. Winkelmans, W. (2004). The socio-economic value of seaports in the European Union. The European Sea Ports Conference 2004: European Sea Ports in a Dynamic Market – Ports and the EU Agenda, Rotterdam, 17–18 June.

This page intentionally left blank

116

THE IMPACT OF PORT CHARACTERISTICS ON INTERNATIONAL MARITIME TRANSPORT COSTS Gordon Wilmsmeier, Jan Hoffmann and Ricardo J. Sanchez ABSTRACT The chapter provides empirical evidence that indicators for different port characteristics have a statistically significant and strong impact on international maritime transport costs. It reports on empirical work on trade among 16 Latin-American countries. The database incorporates 75,928 observations, which comprise practically all maritime trade transactions in containerizable goods on most intra-Latin-American trade routes for the year 2002. The regressions incorporate the main classical explanatory variables of maritime transport costs, such as unit cargo value, volume per transaction, geographical distance, bilateral trade volume, and trade balances. It further looks at six indicators for different port characteristics as possible additional determinants of international transport costs. It is found that indicators for port efficiency, port infrastructure, private sector participation, and inter-port connectivity have significant impacts on international maritime transport costs. The estimated elasticity for port Port Economics Research in Transportation Economics, Volume 16, 117–140 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16006-0

117

118

GORDON WILMSMEIER ET AL.

efficiency is the highest of all port-related variables; doubling port efficiency in a pair of ports has the same impact on international transport costs as halving the distance between them would have.

1. BACKGROUND Determinants of international transport costs are the topic of a growing recent literature. Interest in the topic arises from the desire to better explain economic development and international trade patterns, as well as to identify possibilities to reduce transaction costs. Most international trade continues to be transported by sea, and ports are crucial nodes in global shipping networks. Transport costs are a major component of overall ‘‘trade costs’’. Anderson and van Wincoop (2004) provide an extensive review of trade costs, which are estimated to amount to a 170% ad valorem tax-equivalent, including all transport, border-related, and local distribution costs from the foreign producer to the domestic user. Initial work on the determinants of international transport costs, for example by Radelet and Sachs (1998), uses mainly explanatory variables that are related to distance and geographical characteristics, such as if countries are land locked, or if trading partners are neighbours, and to country characteristics such as GDP per capita. Martinez-Zarzoso, Garcia Menendez, and Suarez-Burguet (2003) suggest that greater distance and poor trade partner infrastructure notably increases maritime transport costs. Inclusion of infrastructure measures improves the fit of the regression, corroborating the importance of infrastructure in determining transport costs. Hummels (1999, 2000, 2001) assesses whether international transport costs have declined, and introduces time as a trade barrier. Wilson, Mann, and Otsuki (2003) find that port efficiency has a strong and significant impact on bilateral trade flows in the Asia-Pacific region. This positive impact of port efficiency on trade flows is most likely due to both its effects on the quality of maritime transport services and, also, on the international maritime transport costs. The present chapter is about the role of port characteristics as determinants of international maritime transport costs. It follows up the work of Fuchsluger (1999), Hoffmann (2002), Kumar and Hoffmann (2002), Sanchez et al. (2003), Wilmsmeier (2003), and Ma´rquez Ramos, Martı´ nez Zarzoso, Pe´rez Garcı´ a, and Wilmsmeier (2006). We analyse international ‘‘freight’’ charges as captured by customs declarations. For each maritime trade transaction, the CIF (‘‘Cost, Insurance, Freight’’) value declared

International Maritime Transport Costs

119

to customs is the sum of the cargo’s FOB (‘‘Free on Board’’) value, the insurance costs, and the freight charges. It is these ‘‘freight’’ charges alone, i.e. not including the insurance, which we use for the empirical research presented in this chapter. Also, it is important to emphasize that we do not look at average CIF/FOB ratios, as have been used in some early cross-country studies (e.g. Gallup, Sachs, & Mellinger, 1998; Radelet & Sachs, 1998; Limao & Venables, 2000), but rather at data for individual transactions. We present empirical results from trade between 7 importing and 16 exporting Latin-American countries. The database incorporates 75,928 observations. They comprise practically all maritime trade transactions on 105 intra-Latin-American trade routes for containerizable goods in the year 2002; ‘‘containerizable’’ meaning here a high likelihood of being containerized (see Annex C). The presented research incorporates the main classical explanatory variables of cargo value, volume per transaction, geographical distance, bilateral trade volume, and trade balances. It generally confirms previous results as regards the impact of these variables. It further looks at six different indicators for port characteristics as possible additional determinants of international transport costs; the indicators are for port infrastructure, port efficiency, port privatization, general transport infrastructure, customs delay, and port connectivity. The relationships between such port characteristics, port costs, and international transport costs are not at all straightforward (see, e.g. Tovar, Jara-Dı´ az, & Trujillo, 2003 for an overview of the literature on cost functions in the port sector; Cullinane & Song, 2002 on private sector participation in ports; Hoffmann, 2001 on ports in Latin America; de Langen, 2004 on maritime clusters and seaports; Beresford & Dubey, 1990 on the competitiveness of trade corridors; Bichou & Gray, 2005 on port terminology). Better port infrastructure may improve efficiency, but this may be at a cost, i.e. it might actually increase port charges and consequently, also the overall transport costs. Port privatization may lead to new investment, but it may also coincide with reduced public subsidies, leading to higher charges to port users. Shippers may be prepared to pay more for a faster and more reliable service, because overall transaction costs are not identical to international transport costs. In spite of these diverse relationships, the empirical results presented in this chapter are quite clear and straightforward: increases in port efficiency, port infrastructure, private sector participation, and inter-port connectivity all help to reduce the overall international maritime transport costs.

120

GORDON WILMSMEIER ET AL.

Our results confirm those presented by Sanchez et al. (2003), who develop a complex measure for port efficiency based on quantitative port performance indicators. Their work provides evidence that port efficiency has the equivalent impact on international maritime transport costs as geographical distance. Sanchez et al. used data for US imports from LatinAmerican countries. They concluded, ‘‘Given the large pertaining differences in productivity among Latin-American ports, these conclusions are relevant for policy makers, for the ports, and for researchers. Unlike distance, economies of scale, and most other determinants of transport costs, port efficiency is within the scope of national policies’’. This chapter attempts to analyse if different port characteristics have a measurable impact on international maritime transport costs, and to quantify these impacts. The different indicators for these characteristics themselves are not being discussed, and neither do we attempt to provide an in-depth analysis of the specific mechanisms through which they might influence maritime transport costs.

2. MODEL The log of the maritime freight costs (FREIGHTij) per ton of import cargo to country i from country j is assumed to depend on     

the type and value of the commodity, distance, volume, the trade balance, and port characteristics.

These basic variables are chosen because they have been shown to be relevant in the previous research mentioned above. Some other variables, such as ‘‘being land-locked’’, that have proven to be significant in other work, are not relevant for our group of countries. Again other variables that have been included in previous research were excluded here because they did not appear to have a significant impact; examples are the flag of the vessel or whether the trading countries belong to the same political block. Many other variables with an impact on transport costs, such as fuel prices or vessel charter rates, are not relevant for our analysis because they vary over time and do not depend on the chosen port or trade route. As has been common practice in the prevailing literature, the log was chosen for most non-binary variables; this has been shown to result in better

International Maritime Transport Costs

121

econometric fits, and it also allows for the interpretation of the results as elasticities.  In order to capture different types and values of the commodity, we include different constants for different commodity groups, and we include the USD value per ton of cargo.  In order to capture the effect of distance, we include the distance in kilometer between the two main ports of the importing and the exporting country.  In order to capture economies of scale we include the volume (tons) of each individual transaction as well as the total volume of the containerizable trade between the two countries.  In order to capture the trade balance, we include the balance of containerizable trade between the two countries.  In order to capture port characteristics, we include six different indicators of which port infrastructure, port efficiency, overall transport infrastructure, and private sector participation are qualitative; and average customs delay and port connectivity are quantitative indicators. The data is described in more detail in Section 3. The resulting model is given in Eq. (1). FREIGHTi;j;c;k ¼ b0;c þ b1 TONSk þ b2 VALUEPERTONk þ b3 DISTANCEij þ b4 BILATERSLVOLUMEij þ b5 BALENCEROUTEij þ b6 PORTINFRAi þ b7 PORTINFRAj þ b8 PORTEFFICi þ b9 PORTEFFICj þ b10 TRANSPORTINFRAi þ b11 TRANSPORTINEFRAj þ b12 PORTPRIVATi þ b13 PORTPRIVATj þ b14 CUSTOMSDELAYi þ b15 CUSTOMSDELAYj þ b16 PORTCNNECTij ð1Þ ywhere b0,c is the constant term, which is different for each commodity group c, TONSk the total weight in tons of the individual trade transaction k in natural logarithm, VALUEPERTONk the US dollar value of the individual trade transaction k in natural logarithm, DISTANCEij the

122

GORDON WILMSMEIER ET AL.

distance in kilometer between the main ports of country i and country j in natural logarithm, BILATERALVOLUMEij the total volume of containerizable trade between country i and country j in natural logarithm, BALANCEROUTEij the coefficient of the imports of containerizable cargo of country i from country j divided by the exports of containerizable cargo from country i to country j, PORTINFRAi an indicator for port infrastructure in the importing country i, PORTINFRAj the equivalent for the exporting country j, PORTINFRAij the log of the sum of the two indicators, PORTEFFICi an indicator for port efficiency in the importing country i, PORTEFFICj the equivalent for the exporting country j, PORTEFFICij the natural log of the sum of the two indicators, TRANSPORTINFRAi an indicator for general transport infrastructure in the importing country i, TRANSPORTINFRAj the equivalent for the exporting country j, PORTPRIVATi an indicator for successful advances with private sector participation in the importing country’s main common user port, PORTPRIVATj the equivalent for the exporting country’s main common user port, CUSTOMSDELAYi the average delay of customs clearance in the importing country in natural logarithm, CUSTOMSDELAYj the equivalent for the exporting country j, and PORTCONNECTij the monthly frequency of direct liner services between the ports of country i and country j in natural logarithm.

3. DATA 3.1. Observations and Variables After filtering out observations with incomplete or extreme data and selecting only commodity groups that are containerizable, the database includes n ¼ 75,928 observations. Each observation corresponds to a transaction k; hence there are 75,928 values for the variables FREIGHTk, TONSk, and VALUEPERTONk. There are seven importing countries i which lead to seven different values for PORTEFFICi, PORTPRIVATi, CUSTOMSDELAYi, PORTINFRAi, and TRANSPORTINFRAi. There are 16 exporting countries j, which lead to 16 different values for PORTEFFICj, PORTPRIVATj, CUSTOMSDELAYj, PORTINFRAj, and TRANSPORTINFRAj. There are 7 times 16 minus 7 pairs of countries (the 16 exporting countries include the 7 importing countries), which lead to 105 different values for DISTANCEij, BILATERALVOLUMEij, BALANCEROUTEij, and PORTCONNECTij.

International Maritime Transport Costs

123

3.2. The Dependent Variable FREIGHT The international maritime transport costs are those recorded by the customs authorities of seven Latin-American importing countries, as reported in the International Trade Data Base (BTI), which is maintained by the United Nations Economic Commission for Latin America and the Caribbean (ECLAC).1 FREIGHT is the log of the maritime transport costs, without insurance costs, of one trade transaction. For this chapter, we use the data for all imports of containerizable cargo of Argentina, Brazil, Chile, Colombia, Ecuador, Peru, and Uruguay coming from the exporting countries of Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela. The BTI distinguishes between the country of ‘‘origin’’, which is where the good is made, and the country of ‘‘departure’’, which is the country from where the good is exported during this particular trade transaction. We use the country of departure, which is of relevance for transport costs. The country of ‘‘origin’’ would be of relevance if, for example, the level of customs duties had to be determined. The BTI does not provide information on whether the cargo was actually containerized or not. For the purposes of this chapter, we selected a group of Standardized International Trade Classification (SITC) codes at the three-digit level that are assumed to be in principle containerizable. Above all, those commodities that are usually transported as liquid or dry bulk are thus excluded from this research. See Annex C for the list of SITC codes included in the regressions. See also Annex A for a description of the data with respect to the number of observations, means, maximum and minimum values, and the standard deviation.

3.3. Explanatory Variables 3.3.1. Type and Value of the Commodity The type of commodity and its value per ton might, in theory, not be related to the pure freight rate. It could be assumed that for a container-shipping operator it is irrelevant what is inside the box, as it does not affect his costs, and, hence, neither the freight rate nor the box-handling charge. In practice, however, traditional ‘‘tariffs’’ of liner-shipping companies do strongly distinguish between different commodities. In particular, if goods are of very high value, the freight may effectively be charged irrespective of weight and

124

GORDON WILMSMEIER ET AL.

measurement on an ad valorem basis. Different charges are also applied to ‘‘less-than-container-loads’’ (LCL) and ‘‘full-container-loads’’ (FCL). In the former case, the rates are usually the same as those charged for noncontainerized shipments. For FCL containers, charges are either for ‘‘freight-all-kinds’’ or the carrier may apply ‘‘commodity box rates’’. Apart from possibly applying different freight rates for different types of commodities, there may also exist different price elasticities for different commodities. In particular, for higher valued goods, a shipper is likely to be prepared to pay a higher freight. It must also be noted that our data does not distinguish between different types of services. Some shippers may be willing to pay a premium for a direct fast service, whereas others might choose to pay less, accepting perhaps a later delivery, transshipment or a less reliable itinerary. A higher FREIGHT may in this case simply be an indicator of a better service. We attempt to capture the possible effect of different commodities and unit values by, first, introducing different CONSTANTs for different commodities, and second, by including the value per ton of cargo VALUEPERTON as an explanatory variable in the regressions. CONSTANT b0,c assumes different values for the different codes of the SITC system at the one-digit level. Goods belonging to SITC codes 3, 4, and 9 are excluded from our research, which does not cover bulk cargo. Hence, we have six different constant terms, reflecting the SITC codes 1, 2, 5, 6, 7, and 8. The variable VALUEPERTON is the log of the value in USD per ton of cargo. It is computed by the authors based on the customs declarations as regards the FOB value and the weight of the traded goods. The mean value per ton is USD 11,048, with a standard deviation of 51,870. 3.3.2. Distance Ceteris paribus, freight increases with distance as it implies more fuel and use of vessels and working hours. The variable DISTANCE is the log of the maritime distance in kilometers. The mean distance is 4,874 km and the standard deviation is 3,162. The source for the distances is Fairplay Ports Guide and www.distances.com. 3.3.3. Economies of Scale Maritime transport is a traditional prime example of economies of scale. ‘‘A ship’s carrying power varies as the cube of her dimensions, while the resistance offered by the water increases only a little faster than the square of her dimensions’’ (Marshall, 1890). Economies of scale can be found in ports as well as in shipping. We attempt to capture the effect of economies of scale on

International Maritime Transport Costs

125

the freight by including variables for the total volume of bilateral trade between the two countries and the volume of the individual transaction. The variable TONS is the log of the volume of the individual trade transaction for which the freight is being paid, as reported by the BTI. The mean for the volume is 3.49 ton, with a standard deviation of 6.35. The variable BILATERALVOLUME is the log of the total annual containerizable trade between the two countries in 2002. It is calculated from the BTI data. The mean total volume of the containerizable trade is 9,58,587 metric tons, with a standard deviation of 1,570,857. 3.3.4. Trade Balance If a ship or a container has to return empty from the importing country, the freight paid for this import cargo will also have to bear the repositioning costs. We attempt to capture this effect by including the trade balance of containerizable goods between the two countries, based on the BTI data. BALANCEROUTE is calculated by dividing the volume of imports of country i from country j by the volume of exports from country i to country j. The mean value of the balance (imports/exports) is 9.27, with a standard deviation of 25.68. 3.3.5. Port Characteristics A port’s efficiency, its private sector participation, delays at customs clearance, the port infrastructure, the country’s general transport infrastructure, and inter-port connectivity may possibly have an impact on the international maritime transport costs. Inter-port connectivity, i.e. liner-shipping services that connect two ports, itself will depend strongly on other port characteristics that affect the services provided to the shipping lines, as well as on trade volumes and inter-modal connections. PORTINFRAi and PORTINFRAj are the indices for the perceived quality of the importing and exporting countries’ port infrastructure in 2002. The means and standard deviations of the indices are 3.09 and 0.64, respectively, for the importing, and 3.79 and 0.96 for the exporting country. PORTINFRAij is the log of the sum of the two indices. The source for the indices is the World Economic Forum (2004). PORTEFFICi and PORTEFFICj are the indices for the perceived efficiency of the importing and exporting countries’ main ports. The mean and standard deviation are 3.39 and 0.57 for the importing country, and 3.74 and 0.80 for exporting country, respectively. PORTEFFICij is the log of the sum of the two indicators. The source for the indices is the World Economic Forum (2004).

126

GORDON WILMSMEIER ET AL.

TRANSPORTINFRAi and TRANSPORTINFRAj are the indices for the perceived quality of the importing and exporting countries’ general transport infrastructure. The means and standard deviations of the indices are 3.28 and 0.70, and 3.71 and 0.49 for the importing and exporting country, respectively. The source for the indices is the World Economic Forum (2004). PORTPRIVATi and PORTPRIVATj are the indices for the perceived success of private sector participation in common user ports for the importing country and for the exporting country, respectively. Data is taken from Hoffmann (2001) and reflects the results of a poll among Latin-American port specialists who attached values between 1 (very poor) and 10 (highly successful) for the introduction of private sector participation in the main common user port of each of the 16 non-Caribbean Latin-American countries in 2000. The highest index was computed for Panama (8.4) and the lowest for El Salvador (1.9). The mean and standard deviation for PORTPRIVATi and PORTPRIVATj are 4.79 and 1.81, and 6.25 and 1.75, respectively. CUSTOMSDELAYi and CUSTOMSDELAYj are the logs of the average delay in customs clearance for the importing and the exporting country. The mean customs delay for the importing country is 6.08 days, with a standard deviation of 1.63. The mean customs delay for the exporting country is 6.56 days, with a standard deviation of 2.70. Note that the delay refers to clearance of imports; it is included here also for the exporting country as a possible indicator of the efficiency of customs as regards export procedures. The source of the data is The World Bank (2003). PORTCONNECTij is the log of the number of direct liner services per month between the two countries’ ports, i.e. an indicator of inter-port connectivity.2 The mean of the number of services per month is 68.15, with a standard deviation of 103.11. Source for the data is Containerization International on-line, July 2002. Inter-port connectivity is not so much a characteristic of a single port, but rather an indicator for the level of services, and possibly liner-shipping competition, between a pair of ports.

4. EMPIRICAL RESULTS 4.1. The Basic Model In the basic model, we introduce five variables considered to reflect the major determinants of international maritime transport costs, i.e. TONS, VALUEPERTON, DISTANCE, BILATERALVOLUME, and BALANCEROUTE. We differentiate between three groups of cargo. Models 1 and 4

International Maritime Transport Costs

127

include all goods that are considered ‘‘containerizable’’; Models 2 and 5 include only those goods with a medium to high likelihood of containerization; and Models 3 and 6 only those with a high likelihood of containerization (see Annex C for the list of SITC codes). The observations included in Model 2 are thus a sub-set of those included in Model 1; and the observations included in Model 3 are a sub-set of those included in Model 2. We further differentiate between the case where different constants are included for the main commodity groups by one-digit SITC code (Models 1, 2, and 3), and the case where one single constant is included (Models 4, 5, and 6). The results are presented in Table 1. For all six models, the estimated parameters have the expected signs and are statistically significant at the 95% level (except for BILATERALVOLUME in Model 3). The estimated parameter values for TONS, VALUEPERTON, and DISTANCE are very stable, with double-digit t-values. The estimated parameter values for BILATERALVOLUME and BALANCEROUTE are slightly less stable, with single-digit t-values. All six models can be considered adequate to serve as a basic model, upon which to build and expand the analysis to include additional variables. We chose to continue with Model 1 for two reasons. First, it provides for a larger number of observations and a larger variance among the explanatory variables. Second, it allows for different constants for different commodity groups, given that different commodities are often traded by different countries and on different routes. An F-test confirms the hypothesis that the CONSTANTSITC are the same has to be rejected. In any case, most of the subsequent results have been tested against the other models, and no significant change occurred as regards the sign and magnitude of the estimated parameters. In Table 2, we present further empirical results, based on Model 1, i.e. including all containerizable cargo, and allowing for different constants for different SITC code commodities. Given that a high correlation exists between most of the variables that aim to measure different port characteristics, we present only the results with one or two port variables included at a time (See Partial Correlation Coefficients Annex B).

5. INTERPRETATION OF RESULTS 5.1. The Base Model 5.1.1. Type and Value of the Commodity Despite looking only at containerizable cargo, different types and values of commodities continue to lead to a significant variation of freight rates. The

128

Table 1. Variable Observations

Regression Results, Basic Model.

Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

N ¼ 75,928. All Containerizable Cargoes

N ¼ 73,536. Medium+High Containerization

N ¼ 72,319. Only High Containerization

N ¼ 75,928. All Containerizable Cargoes

N ¼ 73536. Medium+High Containerization

N ¼ 72319. Only High Containerization

0.5317

0.5260

0.5121

0.7849 0.6389 0.7020 0.6196 0.4815 0.4710 0.0847 ( 56.71) 0.3408 (128.38) 0.3716 (97.67) 0.0065 ( 3.31) 0.00049 (4.38)

0.8522 0.6179 0.7046 0.6144 0.4835 0.4708 0.0837 ( 55.22) 0.3392 (125.75) 0.3710 (96.08) 0.0048 ( 2.45) 0.00042 (3.65)

0.8530 0.6143 0.6968 0.5926 0.4725 0.4589 0.0823 ( 53.85) 0.3362 (123.49) 0.3700 (95.18) 0.0029 ( 1.47) 0.00040 (3.47)

0.0948 ( 64.76) 0.3586 (139.75) 0.3623 (94.96) 0.0161 ( 8.38) 0.00069 (6.15)

0.0942 ( 63.30) 0.3563 (135.84) 0.3618 (93.38) 0.0145 (7.46) 0.00063 (5.47)

0.0933 ( 62.25) 0.3533 (133.42) 0.3609 (92.49) 0.0128 ( 6.51) 0.00062 (5.33)

Adjusted R2 F

0.431 5760

0.424 5403

0.417 5176

0.423 11132

0.414 10409

0.408 9955

Note: t-Values in parentheses.

GORDON WILMSMEIER ET AL.

CONSTANT CONSTANT SITC1 CONSTANT SITC2 CONSTANT SITC5 CONSTANT SITC6 CONSTANT SITC7 CONSTANT SITC8 TONSk VALUEPERTONk DISTANCEij BILATERALVOLUMEij BALANCEROUTEij

Variable Observations TONSk VALUEPERTONk DISTANCEij BILATERALVOLUMEij BALANCEROUTEij PORTINFRAi PORTINFRAj PORTINFRAij PORTEFICij TRANSPORTINFRAi TRANSPORTINFRAj PORTPRIVATi PORTPRIVATj CUSTOMSDELAYi CUSTOMSDELAYj PORTCONNECTij Adjusted R F

2

Regression Results, Expanded Model with Port Characteristics.

Model 7

Model 8

Model 9

Model 10

Model 11

Model 12

Model 13

N ¼ 75,928

N ¼ 75,928

N ¼ 75,928

N ¼ 75,928

N ¼ 75,928

N ¼ 35,438

N ¼ 73,818

0.0863 ( 57.65) 0.3422 (128.74) 0.3716 (95.80) 0.0100 ( 4.46) 0.00020 (1.73) 0.0333 ( 9.92) 0.0497 ( 10.76)

0.0863 ( 57.67) 0.3416 (128.82) 0.3698 (97.26) 0.0109 ( 5.53) 0.00027 (2.40)

0.0869 ( 58.11) 0.3416 (128.94) 0.3542 (90.31) 0.0161 ( 7.97) 0.00047 (4.25)

0.0846 ( 56.51) 0.3408 (128.38) 0.3716 (92.47) 0.0075 ( 3.31) 0.00051 (4.31)

0.0874 ( 58.85) 0.3374 (127.73) 0.3890 (96.81) 0.0322 ( 13.70) 0.00022 ( 1.80)

0.0632 ( 29.15) 0.4665 (113.19) 0.3380 (55.36) 0.0794 ( 23.74) 0.00082 (5.06)

0.0857 ( 57.00) 0.3447 (129.16) 0.1769 (30.28) 0.0256 (10.91) 0.00228 (14.31)

International Maritime Transport Costs

Table 2.

0.2444 ( 13.51) 0.3835 ( 17.65)

0.3786 ( 17.03) 0.0056 ( 1.19) 0.0011(0.19) 0.0038 (2.00) 0.0562 ( 32.00) 0.0512 (4.32) 0.0074 (0.80) 0.1129 ( 32.60)

0.433 4832

0.433 5265

0.434 5286

0.432 5160

0.439 4953

0.501 2971

0.445 4933

Notes: t-Values in parentheses. Constants not reported. The number of observations for Model 12 is smaller, because information about average customs delays was not available for all countries in the sample.

129

130

GORDON WILMSMEIER ET AL.

estimated elasticity for VALUEPERTON is 0.34 (Model 1), i.e. a 1 per cent increase in the unit value of the goods leads to an increase of 0.34 per cent in the freight charged. Given the high variance of this variable, the overall impact on the variance of the freight is the highest among all variables taken into account in our model. An increase in the value per ton by 469 per cent (this is equivalent to the standard deviation divided by the mean) leads to an increase of the freight per ton by 80.91 per cent. Or, to take a simpler example, doubling the unit value leads to an increase in the freight charged of 26.6 per cent. Note that our data does not include payments for insurance by the shipper. 5.1.2. Distance The estimated elasticity for DISTANCE coincides with the results of other research. An increase of the distance by 1 per cent leads to an increase of the freight by 0.37 per cent. Although this is a high elasticity if compared to other variables, it is actually quite low if compared to the traditional assumption made in classical gravity trade models that distance could be used as a proxy for transport costs. Distance is certainly not proportional to transport costs. Doubling the distance does not double the freight, but leads to an increase of just 29.4 per cent, and an increase of the distance of 65 per cent (i.e. the standard deviation in our sample) increases the freight by only 20.4 per cent (Model 1). 5.1.3. Economies of Scale The elasticity for TONS is 0.0847, i.e. an increase in the volume of a transaction of 1 per cent leads to a reduction of the freight by 0.0847 per cent. Although this may not seem high, it makes an important contribution to the variation of FREIGHT, because TONS itself has a high variance. If, by way of example, we ship 1,000 ton from country j to country i with one single transaction, instead of shipping the same goods with 10 shipments of 100 ton each, we will, on average, achieve a saving of 8.04 per cent on the international maritime transport costs (Model 1). The impact of BILATERALVOLUME has the expected negative sign, and is statistically significant in most regressions. The estimated parameter value is quite low, however. An increase of the bilateral containerizable trade of 1 per cent leads to a reduction of the freight charges by only 0.0065 per cent (Model 1). If, by way of example, two countries have bilateral trade of 10 million tons instead of 1 million tons, the FREIGHT (per ton) for this bilateral trade will be 1.5 per cent lower.

International Maritime Transport Costs

131

As regards the specification of FREIGHT as the dependent variable, and the volume of trade as an explanatory variable, it could be argued that trade volumes are also being explained by transport costs. FREIGHT is basically a price, which depends on supply and demand, and it would have to be estimated by using, for example, instrumental variables. However, for our case, on a given route, the total volume of bilateral trade per year can be assumed to be ‘‘given’’ for a transport service provider, who adjusts their price for a given transaction at short notice in view of costs and the market environment. In fact, we believe that the effect of economies of scale on freight rates may be stronger than the effect of lower transport costs on trade volumes. The elasticities for transport costs as a determinant of trade volumes as estimated, for example, by Limao and Venables (2000) may be too high. The dynamic relation between transport costs and trade volumes will require further research.

5.1.4. Trade Balance BALANCEROUTE has the expected positive sign, i.e. if a country imports more than what it exports, the FREIGHT for the imports will go up. Each increase of the coefficient imports/exports by one point will lead to an increase in the freight costs by 0.00049 per cent. If a country i imports twice as much from country j as it exports to country j, its freight will go up by 0.034 per cent (Model 1). Although the parameter is statistically significant in all models, and has the expected sign, the estimated parameter value is far too low and does not reflect the real impact of trade imbalances on freight rates. As any container-shipping company knows, on many major liner-shipping routes, the freight rates in one direction may be twice as high as in the other direction, and one main reason for the difference is the unbalanced trade. The variables BILATERALTRADE and BALANCEROUTE are both computed only for the trade between countries i and j. This appears to be inadequate to capture the effect of economies of scale and trade imbalances, both of which need to be applied to broader trade routes. By way of example, the bilateral trade volumes and the trade imbalances between Guatemala and Chile have only a minor effect on the freight rates between these two countries. What really matters is the total trade volume along the South American Pacific coast and South America’s trade balance with North America and Asia. Future research will have to attempt to capture the broader trade volumes and imbalances.

132

GORDON WILMSMEIER ET AL.

5.2. Expanded Models Incorporating Port Characteristics 5.2.1. Port Infrastructure The index PORTINFRAi for the importing country’s port infrastructure has a negative impact on FREIGHT, i.e. it leads to a reduction of transport costs. If an importing country with the lowest index of the sample (2.3) could improve its port infrastructure to the level of the best index of the sample (4.6), it would be expected to reduce the maritime transport costs for its imports by 7.4 per cent (Model 7). As regards the index PORTINFRAj for the exporting country’s port infrastructure, the impact on FREIGHT is larger than for the importing country’s port infrastructure. PORTINFRAj has a larger variation, and the value of the estimated parameter is higher. If a country with the worst index (1.4) could improve its port infrastructure to the level of the best index (5.4), it would be expected to reduce the maritime transport costs for its exports by 18 per cent (Model 7). In a different approach of including port infrastructure into our model, we generated the log of the sum of the two indices for the importing and exporting countries PORTINFRAij. This allows for an easier interpretation of the estimated elasticity, i.e. a 1 per cent increase of the combined port infrastructure index leads to a reduction of the freight by 0.24 per cent (Model 8). If the two countries of the sample with the worst port infrastructure improved theirs to the level of the two countries with the best port infrastructure, the maritime transport costs on the route between them would be expected to decrease by 21.6 per cent. 5.2.2. Port Efficiency The combined port efficiency of the importing and exporting countries’ ports PORTEFFICij has the highest estimated elasticity of all variables included in our regressions. Increasing the indicator for port efficiency by 1 per cent reduces freight charges by 0.38 per cent (Model 9). If the two countries of the sample with the lowest port efficiency improved their efficiency to the level of the two countries of the sample with the highest indices, the freight charges on the route between them would be expected to decrease by 25.9 per cent. 5.2.3. General Transport Infrastructure The general transport infrastructure of a country has practically no bearing on the international maritime freight (Model 10). The estimated parameter values for TRANSPORTINFRAi and TRANSPORTINFRAj are

International Maritime Transport Costs

133

statistically not significant. This does not mean that general transport infrastructure would not be relevant for overall trade efficiency, just that it has no effect on the international maritime portion of the trade costs. 5.2.4. Port Privatization The private sector participation in the main container ports of the countries of the sample, as measured by an index derived from a poll taken in 2000, leads to somewhat ambiguous results. The impact of PORTPRIVATi for the importing country is very small, and positive, i.e., it leads to a minor increase of the freight (Model 11). The difference between the best and the worst case of our sample leads to a difference in the freight of less than 2 per cent. For the private sector participation on the exporting country’s side, PORTPRIVATj, the impact is far stronger. The difference between the best and the worst case of our sample leads to a difference in the freight of 30.6 per cent (Model 11). In other words, if the country with the lowest indicator had advanced as much as the country with the highest indicator, the maritime freights for its exports would be expected to be 30.6 per cent lower. 5.2.5. Customs Delay The delay of cargo during customs procedures has a minor, positive, impact on freight. On the importing country’s side, CUSTOMSDELAYi, a 1 per cent reduction of the time it takes to clear customs implies a reduction of the maritime freight of 0.051 per cent (Model 12). For the exporting country, CUSTOMSDELAYj is statistically not significant. The speed of customs operations may be correlated to other aspects of port efficiency and thus just be an indicator of the latter. It may also be that carriers charge higher freights if their containers are expected to spend more time in the importing country due to delayed customs clearance. 5.2.6. Inter-Port Connectivity Increasing the frequency of liner services between a pair of ports by 1 per cent leads to a reduction of freight by 0.113 per cent (Model 13). Given the high variability of this variable, the impact on the freight is quite large. If two ports increase their connectivity by 150 per cent (i.e. the standard deviation in our sample), the freight between them can be expected to go down by almost 10 per cent. PORTCONNECT is closely correlated with BILATERALVOLUME; ships follow the cargo. In fact, the estimated parameter for BILATERALVOLUME in Model 13 becomes positive, suggesting a different interpretation

134

GORDON WILMSMEIER ET AL.

of the parameters. The number of liner services per ton of cargo could be interpreted as an indicator of competition as shown in Eq. (2). LINERCOMPETITION ¼ PORTCONNECT

BILATERALVOLUME (2)

Note that the variables are defined as logarithms, i.e. LINERCOMPETITION is the logarithm of the coefficient (number of liner services)/(total bilateral trade volume). Hence, a reformulation of Model 13, where PORTCONNECT is replaced with LINERCOMPETITION would yield the estimated parameters 0.1129 for LINERCOMPETITION and 0.0873 for BILATERALVOLUME ( 0.0873 ¼ 0.0256–0.1129). The interpretation of these parameters would be as follows: A 1 per cent increase in the level of competition between liner services (number of services per ton of cargo) leads to a decrease of freight by 0.1129 per cent. At the same time, an increase in the volume of bilateral trade leads to a decrease of freight by 0.0873 per cent (Model 13).

6. DISCUSSION AND CONCLUSIONS A more efficient port does not necessarily need to be less expensive. On the contrary, it may charge higher prices to the shipper and the carrier if it provides faster and more reliable services, or if it allows the shipper or the carrier to achieve savings elsewhere. Installing ship-to-shore gantries, for example, may well lead to higher port charges to the shipping line. The line may still achieve an overall saving, because its ships spends less time in the port, or because it can change from geared to gearless vessels. This, in turn, will also lead to lower freight rates. The empirical results of our research suggest that this is effectively the case. We do not know if port improvements lead to lower freights because of lower port costs charged to the carrier, better services provided to the carrier, or both. What is clear, however, is that there is a clear measurable impact on international maritime transport costs. Increases in port infrastructure and private sector participation, too, lead to reduced maritime transport costs. Inter-port connectivity, too, reduces transport costs, most likely because it allows for economies of scale, and also more competition among carriers. The elasticity for port efficiency is higher than the elasticity for distance; in fact, it is the highest of all the variables included in our research. Unlike distance, port efficiency can be influenced by policy

International Maritime Transport Costs

135

makers. Doubling port efficiency at both ends has the same effect on international maritime transport costs as would a ‘‘move’’ of the two ports 50 per cent closer to each other, i.e. reducing the distance between them by half. Port improvements appear to have a stronger impact on the maritime freight of a country’s exports than on the freight of its imports. The exception is average customs delay, which – as might be expected – has a stronger bearing on the maritime freight charged on imports. The general land transport infrastructure has – as expected – no significant bearing on maritime transport costs. Our models explain between around 40 and 50 per cent of the variance of FREIGHT. The remaining part of the variance may partly be due to the fluctuations of freight rates throughout a year (see also Stopford, 2002; Sanchez, 2004; Hoffmann, 2005). The BTI does not tell us in which month a transaction took place and the aspect of time could thus not be incorporated into our model. It also appears that additional or different measures to cover economies of scale as well as trade imbalances might further improve the regression fit. Finally, the R2 can be improved significantly if regressions are undertaken for individual commodity groups, reaching values of up to 0.8. The main results regarding port characteristics as presented in this chapter, however, remain unchanged. The overall impact of port efficiency on trade costs goes beyond the measurable impact on international maritime transport costs. Almost all trade uses more than one mode of transport, and not all port costs are charged to the maritime transport operator. Some port costs may be charged to the trader prior to determining the good’s FOB value, and others may be charged to the trader after the CIF value has been determined and declared to customs. In addition, port improvements will not only lead to lower freight rates, but by providing better services ports can also attract additional liner services and additional cargo. Both – more liner services and higher cargo volumes – lead to a further reduction of freight rates. Lower transport costs, in turn, will stimulate increased trade volumes, which lead to further economies of scale and lower freight charges. These dynamic effects of port improvements will thus lead to further reductions of transport costs that go beyond those measured by our research. The international leg of most international trade transactions continues to be maritime, and most determinants of international maritime transport costs are beyond the control of policy makers. It is through improvements in the ports that cost savings and increased trade competitiveness can be achieved.

136

GORDON WILMSMEIER ET AL.

NOTES 1. The International Transport Data Base (BTI, ‘‘Base de datos de Transporte Internacional’’) was created by the United Nations Economic Commission for Latin America and the Caribbean (ECLAC) in 2000 in order to facilitate research in the areas of trade and international transport. It was the result of a research project (Fuchsluger, 1999) and is described in more detail in Hoffmann, Pe´rez, and Wilmsmeier (2002). It contains trade data for 11 Latin American countries. In addition to the typical trade data that is commonly published for example by COMTRADE (http://unstats.un.org/unsd/comtrade/), the BTI includes, inter alia, information about the mode of transport, the country of departure, the freight, and the insurance paid for international transport. Further information is available on the ECLAC Maritime Profile www.eclac.cl/transporte/perfil/bti.asp. 2. Given that between some pairs of countries there are no direct liner services, we added one to all observations. This avoids the problem of having to take logarithm of zero values, and it can be justified to represent the option of using an indirect service, with transshipment.

REFERENCES Anderson, J. E., & van Wincoop, E. (2004). Trade costs. Journal of Economic Literature, XLII(September), 691–751. Beresford, A., & Dubey, R. (1990). Handbook on the management and operations of dry ports. UNCTAD, RDP/LDC/7, Geneva. Bichou, K., & Gray, R. (2005). A critical review of conventional terminology for classifying seaports. Transportation Research Part A, 39, 75–92. Cullinane, K. P. B., & Song, D.-W. (2002). Port privatisation policy and practice. Transport Reviews, 22(1), 55–75. de Langen, P. (2004). Governance in seaport clusters. Maritime Economics and Logistics, 4, 141–156. Fuchsluger, J. (1999). Analysis of maritime transport costs in South America. Unpublished masters thesis. University of Karlsruhe, Germany. Gallup, J., Sachs, J., & Mellinger, A. (1998). Geography and economic development. Presented at the annual bank conference on development economics, World Bank, April. Hoffmann, J. (2001). Latin American ports: Results and determinants of private sector participation. International Journal of Maritime Economics, 3, 221–241. Hoffmann, J. (2002). El costo del transporte internacional, y la integracio´n y competitividad de Ame´rica Latina y el Caribe. Boletı´n Fal No. 191, July. Santiago: United Nations ECLAC. Hoffmann, J. (2005). The determinants and fluctuations of maritime freight and charter rates. IAME Cyprus 2005 conference proceedings, Cyprus. Hoffmann, J., Pe´rez, G., & Wilmsmeier, G. (2002). International trade and transport profiles of Latin American countries, year 2000. Series Manuals No. 19. Santiago: United Nations ECLAC. Hummels, D. (1999). Toward a geography of trade costs. Chicago, USA: Mimeo, University of Chicago.

International Maritime Transport Costs

137

Hummels, D. (2000). Time as a Trade Barrier. Mimeo, Purdue University, October. Hummels, D. (2001). Have international transport costs declined? Journal of International Economics, 54(1), 75–96. Kumar, S., & Hoffmann, J. (2002). Globalization, the maritime nexus. Handbook of maritime economics and business, London: LLP. Limao, N., & Venables, A. (2000). Infrastructure, geographical disadvantage, transport costs and trade. World Bank Policy Research Paper 2257. Washington, DC: World Bank. Ma´rquez Ramos, L., Martı´ nez Zarzoso, I., Pe´rez Garcı´ a E., & Wilmsmeier, G. (2006). Determinantes de los Costes de Transporte Marı´ timos: El Caso de las Exportaciones Espan˜olas. Monogra´fico del ICE. Jaume, Spain (forthcoming). Marshall, A. (1890). Principles of economics. Book Four: The Agents of Production: Land, Labour, and Capital and Organization. Chapter 11, Industrial organization: Production on a large scale. London: Macmillan. Martinez-Zarzoso, I., Garcia Menendez, L., & Suarez-Burguet, C. (2003). The impact of transport costs on international trade: The case of Spanish ceramic exports. Maritime Economics and Logistics, 5, 179–198. Radelet, S., & Sachs, J. (1998). Shipping costs, manufactured exports, and economic growth. Mimeo. Paper presented at the American Economic Association Meetings, Harvard University. Sanchez, R. (2004). Puertos y transporte marı´timo en Ame´rica Latina y el Caribe: Un ana´lisis de su desempen˜o reciente. Serie Recursos Naturales e fraestructura No. 82, LC/L.2227-P/E. United Nations, ECLAC, Santiago de Chile. Sanchez, R., Hoffmann, J., Micco, A., Pizzolitto, G., Sgut, M., & Wilmsmeier, G. (2003). Port efficiency and international trade: Port efficiency as a determinant of maritime transport cost. Maritime Economics and Logistics, 5, 199–218. Stopford, M. (2002). Shipping markets cycles. Handbook of maritime economics and business, London: LLP. The World Bank. (2003). Global economic prospects, 2004. Washington: World Bank. Tovar, B., Jara-Dı´ az, S., & Trujillo, L. (2003). Production and cost functions and their application to the port sector, a literature survey. World Bank Policy Research Working Paper 3123, August. Washington, DC: World Bank. Wilmsmeier, G. (2003). Modal choice in South American freight transport: Analysis of constraint variables and a perspective for diversified modal participation in South America. Unpublished masters thesis. Dresden: Technische Universita¨t Dresden. Wilson, J., Mann, C., & Otsuki, T. (2003). Trade facilitation and economic development. World Bank Policy Research Working Paper 2988, Washington, DC: World Bank. World Economic Forum. (2004). World competitiveness report 2003–2004. London: Palgrave.

138

GORDON WILMSMEIER ET AL.

ANNEX A. DESCRIPTION OF DATA Variable

Obs.

BALANCEROUTEij BILATERALVOLUMEij CUSTOMSDELAYi CUSTOMSDELAYj DISTANCEij FREIGHT PORTCONNECTij PORTEFFICi PORTEFFICj PORTEFFICij PORTINFRAi PORTINFRAj PORTINFRAij PORTPRIVATi PORTPRIVATj TONS TRANSPORTINFRAi VALUEPERTON

75953 75953 35518 75857 75954 75929 73843 75954 75954 75954 75954 75954 75954 75954 75954 75954 75954 75954

Mean

Std. Dev.

9.272659 25.68117 12.75741 1.478283 1.755342 0.3437897 1.798404 0.410935 8.23266 0.7561608 5.282098 1.008404 3.363312 1.433481 3.390797 0.5738745 3.737947 0.8003441 1.950333 0.1398182 3.090664 0.6368681 1.301688 0.2498696 1.916256 0.1604536 4.786635 1.813817 6.249979 1.750989 0.5805068 2.269712 3.284014 0.7008314 8.297586 1.289257

Min.

Max.

0.0000787 518.24 6.536532 15.52339 1.098612 1.94591 1.098612 2.70805 5.47 9.39 0.97 9.98 0.3074847 6.089476 2.5 4.3 1.8 5 1.45 2.23 2.3 4.6 0.3364722 1.686399 1.308333 2.302585 2.7 7.5 1.9 8.4 5.3 3.6 2.5 4.8 3.066657 15.14848

ANNEX B. PARTIAL CORRELATION COEFFICIENTS BETWEEN THE VARIABLES FREIGHT TONS VALUE DISTANCE BILATERAL BALANCE PORT PORT PORT PORT PORT TRANS PORT PORT CUSTOMS CUSTOMS PORT PERTON VOLUME ROUTE INFRAi INFRAj EFFICi EFFICj EFFICij PORTINFj PRIVATi PRIVATj DELAYi DELAYj CONNECTij FREIGHT TONS VALUEPERTON DISTANCE BILATERALVOLUME BALANCEROUTE PORTINFRAi PORTINFRAj PORTEFFICi PORTEFFICj PORTEFFICij TRANSPORTINFj PORTPRIVATi PORTPRIVATj CUSTOMSDELAYi CUSTOMSDELAYj PORTCONNECTij

1 0.53 0.64 0.34 0.26 0.13 0.10 0.07 0.17 0.09 0.10 0.08 0.12 0.15 0.04 0.05 0.38

0.53 1 0.58 0.02 0.08 0.03 0.02 0.07 0.03 0.07 0.06 0.02 0.01 0.08 0.02 0.05 0.04

0.64 0.58 1 0.08 0.02 0.02 0.04 0.00 0.04 0.00 0.04 0.02 0.01 0.07 0.03 0.01 0.07

0.34 0.02 0.08 1 0.38 0.31 0.03 0.03 0.19 0.01 0.15 0.13 0.09 0.29 0.18 0.06 0.76

0.26 0.08 0.02 0.38 1 0.20 0.34 0.29 0.49 0.39 0.23 0.09 0.39 0.30 0.04 0.34 0.59

0.13 0.03 0.02 0.31 0.20 1 0.29 0.08 0.33 0.29 0.11 0.15 0.06 0.19 0.24 0.06 0.41

0.10 0.02 0.04 0.03 0.32 0.29 1 0.17 0.88 0.16 0.69 0.25 0.35 0.01 0.84 0.05 0.17

0.07 0.07 0.00 0.03 0.29 0.08 0.17 1 0.13 0.89 0.41 0.79 0.13 0.43 0.14 0.41 0.00

0.17 0.03 0.04 0.19 0.49 0.33 0.88 0.13 1 0.13 0.81 0.18 0.26 0.06 0.56 0.13 0.41

0.09 0.07 0.00 0.01 0.39 0.29 0.16 0.89 0.13 1 0.47 0.60 0.15 0.41 0.11 0.36 0.11

0.10 0.06 0.02 0.13 0.23 0.11 0.79 0.41 0.81 0.47 1 0.22 0.14 0.19 0.43 0.11 0.30

0.08 0.02 0.03 0.00 0.09 0.25 0.99 0.18 0.18 0.60 0.22 1 0.06 0.33 0.27 0.31 0.01

0.12 0.01 0.01 0.09 0.39 0.06 0.35 0.13 0.26 0.15 0.14 0.06 1 0.11 0.02 0.09 0.15

0.15 0.08 0.07 0.29 0.30 0.19 0.01 0.43 0.06 0.41 0.19 0.33 0.11 1 0.08 0.71 0.15

0.04 0.02 0.03 0.18 0.04 0.24 0.84 0.14 0.56 0.11 0.43 0.27 0.02 0.08 1 0.06 0.13

0.05 0.05 0.01 0.06 0.34 0.06 0.05 0.41 0.13 0.36 0.11 0.31 0.09 0.71 0.06 1 0.06

0.37 0.04 0.07 0.76 0.59 0.41 0.17 0.00 0.41 0.11 0.30 0.01 0.16 0.15 0.13 0.06 1

140

GORDON WILMSMEIER ET AL.

ANNEX C. LIST OF SITC CODES INCLUDED IN THE DATA The following commodities as defined by the United Nations Standard International Trade Classification, Revision 3, code (SITC rev. 3) are included in the empirical research. See http://unstats.un.org/unsd/cr/registry/ regcst.asp?Cl=14 for more details on the individual codes. High probability of containerizations. 111, 112, 12, 12, 121, 122, 16, 17, 212, 22, 261, 263, 264, 266, 267, 268, 289, 35, 37, 48, 515, 525, 531, 532, 533, 541, 542, 551, 553, 554, 56, 57, 571, 572, 573, 574, 575, 58, 581, 582, 583, 59, 593, 597, 598, 611, 612, 613, 62, 621, 625, 629, 633, 64, 641, 642, 651, 652, 653, 654, 655, 656, 657, 658, 659, 664, 665, 666, 667, 681, 683, 684, 685, 686, 687, 689, 694, 695, 696, 697, 733, 735, 737, 74, 74, 741, 742, 743, 744, 745, 746, 747, 748, 749, 75, 751, 752, 759, 76, 761, 762, 763, 764, 77, 771, 772, 773, 774, 775, 776, 778, 784, 785, 811, 812, 813, 821, 831, 841, 842, 843, 844, 845, 846, 848, 851, 871, 872, 873, 874, 881, 882, 883, 884, 885, 891, 892, 893, 894, 895, 896, 897, 898, 899, 98. Medium probability of containerization. 211, 222, 223, 231, 232, 244, 245, 265, 269, 277, 284, 285, 286, 287, 288, 291, 292, 42, 431, 46, 47, 512, 513, 514, 516, 522, 523, 524, 591, 592, 634, 635, 663, 675, 676, 678, 679, 692, 699, 711, 712, 713, 714, 716, 718, 72, 721, 723, 724, 725, 726, 727, 728. Low probability of containerization. 511, 579, 662, 671, 672, 673, 674, 677, 691, 693, 722, 731, 791, 792, 793. All other SITC codes are considered not to be containerizable and are excluded from the regressions.

STRATEGIC POSITIONING ANALYSIS FOR SEAPORTS Elvira Haezendonck, Alain Verbeke and Chris Coeck ABSTRACT In this chapter, a Strategic positioning analysis (SPA) is developed as a specific analytical approach consisting of a product portfolio analysis, a shift-share analysis and a diversification analysis. The SPA describes the performance of ports and traffic categories within ports in terms of market share, growth rate, diversification and value added. The SPA needs to be used taking into account the port’s position with regard to valueadded created by the different traffic categories. By using this integrative instrument, indications on the overall strategic position of ports are provided and will benefit strategy formulation and decision-making on port development.

1. INTRODUCTION When engaging in strategic decision-making, port authorities, terminal operators and port users must build upon a conceptual understanding of the dynamics of international port competition and perform strategicpositioning analyses. Moreover, they should take into account the best available information on the port industry. This chapter presents a tool that Port Economics Research in Transportation Economics, Volume 16, 141–169 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16007-2

141

142

ELVIRA HAEZENDONCK ET AL.

describes the evolution of a port’s competitive position in terms of growth, market share and diversification and also includes aspects of value added. This tool is applied to the positioning of the nine most important seaports in the Hamburg – Le Havre range (i.e. Amsterdam, Antwerp, Bremen, Dunkirk, Ghent, Hamburg, Le Havre, Rotterdam and Zeebrugge). The chapter builds upon a specific analytical approach, namely the strategic positioning analysis (SPA) and growth-share matrix, originally introduced by the Boston Consulting Group (BCG) in 1968, in the context of large, diversified business firms, with multiple business units (Henderson, 1979). Even in case the results obtained from the SPA are not directly used for decision-making on large investment projects, the particular way in which information on the evolution of the competitive position of the ports considered is collected, processed and presented, can improve the capabilities of decision makers to assess strategic alternatives (Hax & Majluf, 1983; Hambrick & MacMillan, 1982; Bettis & Hall, 1981). At the same time, it should be recognized from the outset that the use of SPA as a strategic management tool also has its limitations. Both the merits and the limitations of this type of analysis will be discussed in this chapter. An SPA describes quantitatively the ‘performance’ of individual ports in terms of market shares, growth rates and diversification. An SPA can provide comparative, micro-level information on both the port as a whole as well as on individual traffic units. However, the portfolio concept, as will be explained in detail in the next section, is not a substitute for managerial judgment as to the exact significance of analytical results (Brown, 1991). In this context, Miles (1986) has argued that using SPA as the basis for strategy formulation runs the risk of oversimplifying what it takes to succeed. In order to gain insights into the sustainability of an observed competitive position, SPA should be combined with, inter alia, a resource-based investigation of underlying ‘port specific advantages’ (Haezendonck, 2001). Nevertheless, the SPA may represent a useful component of a broader, more comprehensive appraisal of port competitiveness.

2. CONCEPTUAL FOUNDATIONS 2.1. Strategic Positioning Analysis 2.1.1. Introduction Verbeke (1992), Winkelmans and Coeck (1993), Notteboom and Coeck (1994) and Verbeke, Peeters, and Declercq (1995) have argued that the main

Strategic Positioning Analysis for Seaports

143

purposes of SPA for a seaport vis-a`-vis its main competitors, apart from its obvious descriptive value, are threefold: 1. to process and present statistical information on the recent evolution or change in the competitive position of different seaports; 2. to help assessing the future economic potential of a seaport, given anticipated future developments; and 3. to prepare options and alternatives in the context of strategic decisionmaking. The SPA developed in this chapter consists of three interrelated analytical approaches to determine the competitive position of a port as compared to the other ports in the port range considered. These three approaches are (1) product portfolio analysis (PPA), (2) shift-share analysis (SSA), and (3) product diversification analysis (PDA). By combining the results of these three techniques, relevant conclusions as to the recent competitive performance and potential of the most important seaports in the Hamburg – Le Havre range can be drawn, as shown in the empirical analysis in this chapter. One of the main advantages of this approach is that no confidential financial or marketing data needs to be included in the analysis. The SPAtechnique allows drawing relevant conclusions based on easily obtainable data or information, i.e. the actual traffic flows in the different ports under consideration. However, the decisions on which ports to include in the range, which commodity groups or traffic flows to study and on the observation period to use in the analysis, are critical and need to be taken prior to conducting the actual SPA. The choice of ports, traffic categories and observation periods considered will be discussed in the empirical part of this chapter. 2.1.2. Product Portfolio Analysis Dyson (1990) lists a number of analytical techniques, among others the experience curve, threats, opportunities, weaknesses and strengths (TOWS) analysis, the profitability impact of marketing strategies (PIMS) model and the ‘growth-share matrix’, each with specific advantages and disadvantages that allow a comparative or competitive positioning of businesses or business units. These models could in principle be applied to the port sector. In this chapter, portfolio analysis, or the use of the ‘growth-share matrix’, is selected as the most appropriate tool for determining the competitive

144

ELVIRA HAEZENDONCK ET AL.

position of seaports. This choice resulted from the basic characteristics of this technique.  The technique allows ‘insight-creating visualization’.  All data required for its application can be found in publicly available data sources.  It can be usefully applied, irrespective of the presence of profit-maximizing goals pursued by the organizations studied. Moreover, the introduction of the ‘value-added’ concept (Haezendonck, Coeck, & Verbeke, 2000) can be easily incorporated into the portfolio analysis. Extensive empirical research on port competition, e.g. De Lombaerde and Verbeke (1989), has indeed led to the conclusion that seaport authorities in the Hamburg – Le Havre range are primarily interested in: (a) increasing their port’s market share vis-a`-vis rivals in the range and (b) diversifying their traffic structure, in order to achieve sustainable growth, irrespective of the evolution of individual traffic categories. The PPA-technique, which was originally developed in the field of strategic business planning by BCG, allows interpreting the performance of businesses and business units using two variables, i.e. market share and growth in a growth-share matrix. The main contribution of this portfolio analysis approach consists of determining the present position of specific businesses vis-a`-vis rivals and their potential for increasing their market share (Day, 1977; Wind & Mahajan, 1981). The original ‘growth-share matrix’ combines the measurement of actual industry market share for each of the firm’s businesses with related growth rates. These growth rates, when compared with industry averages, impact upon future market shares. The present study does not assume that a particular structure of a seaport’s traffic portfolio should be subject to the same normative guidelines as applied in a company setting, i.e. attempting to maximize its profitability by means of a number of investment and divestment decisions and aiming to use cross-financing with cash flows earned in specific businesses. Nevertheless, obtaining high market shares and increasing these market shares for traffic categories with a high value added to the port, is viewed as critical by many port authorities and port operators, irrespective of whether this maximizes financial performance. A large-scale consulting study conducted for the European Commission (DG TREN – Transport and Energy) and including 13 large European seaports (including Antwerp, Hamburg and Rotterdam) has led to the conclusion that all seaport authorities consider overall cost recovery at the level of the entire port

Strategic Positioning Analysis for Seaports

145

HIGH

‘Question Marks’

‘Stars’

LOW

GROWTH RATE

PRODUCT MARKET

as important. However, profit maximization did not appear to be a critical goal for most port authorities, see ATENCO (1999). According to the BCG, a firm’s set of strategic business units (SBUs) or product market entities can be positioned in a decision matrix, as a function of their market share and relative growth (Henderson, 1979). The basic concepts of the BCG-matrix can easily be translated in terms of ports. The SBUs of the BCG-matrix, when applied to the port sector can be conceived as the different traffic categories under consideration, e.g. liquid bulk, dry bulk, containers, Ro-Ro (roll on, roll off) and conventional cargo, and, therefore, could be considered as ‘strategic traffic units’ (STUs). Here, every relevant traffic flow for each port is described in terms of relative market share and traffic growth. The BCG-matrix distinguishes among four distinct market positions. According to the positioning within this matrix, ‘question marks’, ‘stars’, ‘cash cows’ and ‘dogs’ can be observed, see Fig. 1. The arrows in the matrix indicate the ‘normal’ evolution of a business activity, in this case traffic flows of a seaport. In a conventional business context, ‘Question marks’1 (high growth rate and small market share) are SBUs usually demanding important investments in order to gain market share, due to their high growth rate. When an important market share is combined with an above average growth rate, a ‘Star’ position is obtained. These SBUs represent the ‘success stories’ of a company. ‘Cash cows’ (large market share and small growth rate) generate financial resources, needed to reinvest in, e.g. ‘question marks’. ‘Dogs’ have little intrinsic economic potential, as they are unable to generate sufficient cash flows as a result of

‘Dogs’

‘Cash Cows’

LOW

HIGH RELATIVE MARKET SHARE

Fig. 1.

Boston Consulting Group-Matrix. Source: Dibb, Simkin, Pride, and Ferrell (1991).

146

ELVIRA HAEZENDONCK ET AL.

a small market share combined with a less than average growth rate. An excellent in-depth analysis of the strengths, possible extensions and limitations of portfolio analysis, albeit mainly in the context of conventional business activities, has been provided by McKiernan (1992). In any case, it should be emphasized that when applying portfolio analysis to port traffic structures, the conventional, simple normative implications of the tool, for example generating cash flow or demanding investments to gain market share, are unlikely to be valid. Nevertheless, portfolio analysis still allows both port authorities and port operators to gain useful insights into the structure of the port’s traffic flows as compared to rivals. It allows both a comparative description of past performance and provides a starting point for the strategic planning of future resource allocations that impact upon the traffic portfolio. The portfolio analysis approach has been applied to the traffic structure of ports in a particular range at four ‘levels’, see Verbeke (1992). These ‘levels’ need to be considered as different types of analysis, rather than as a hierarchy of importance between the different ways of data analysis. Indeed, the levels are complementary and provide additional information. The notion of ‘level’ is only used to indicate the versatility of the PPA instrument and to summarize the possibilities of the strategic analysis. However, a hierarchy between the STUs can be considered, starting with an analysis of the entire port system, then to a port and then to an analysis of individual traffic categories. As mentioned, no priority is given to a specific ‘level’ of analysis. At a first ‘level’, the PPA compares overall market shares and total growth rates of the ports under consideration (external positioning analysis). In this situation, the range itself is conceptualized as a ‘portfolio of ports’, without a distinction being made between relevant commodity groups or traffic categories. As a result, this type of portfolio analysis provides a classification of ports according to their overall traffic evolution. In accordance with the original BCG-matrix, the average growth ratio per year and the average market share are represented respectively vertically and horizontally. At the second ‘level’, the PPA investigates the traffic structure of each individual seaport in the range. In contrast with the previous PPA, the share as well as the growth rate of each traffic category in the port’s total traffic is described (internal positioning analysis). Therefore, each individual seaport is considered as a portfolio of traffic categories. A positioning of traffic flows within each seaport is the result of this analysis. At the third ‘level’, each commodity group in the range is itself viewed as a ‘portfolio of the included seaports’, i.e. as a total traffic volume that can be

Strategic Positioning Analysis for Seaports

147

decomposed among these seaports. As a result, for each traffic category a classification of ports will result according to their market share in the range and their growth rate for that specific traffic category. Finally, the fourth ‘level’ differs from the third ‘level’ in such a way that here the X-axis represents the share of a specific category within a port, rather than the share of this category in the range, i.e. each considered traffic category is being examined in relation to the share of this category within the seaport. This level introduces an additional dimension to the portfolio analysis: a circular shape with a surface proportional to the absolute traffic volume of the port considered in the total range. The advantage of this ‘level’ is that for each seaport in the range the position (of a traffic category in the commodity structure of a single port), the size (of the considered commodity in relation to the size of the same commodity in the other ports in the range) and the growth rate (of the commodity for the port in question) can be presented simultaneously in one figure (joint internal and external positioning analysis). From an analytical point of view, the fourth level of the analysis is the most relevant one; yet, the combination of all four PPAs provides an eclectic set of perspectives on the competitive position of each seaport in the range. In this chapter’s empirical analysis, each of the four levels will be shown graphically by one figure. In addition to this PPA at four ‘levels’, a dynamic analysis over different time periods can be presented. In such a dynamic analysis, the positioning in different time periods is determined. This will allow drawing more specific conclusions on changes in competitive position over time. It should be noted that when conducting applications to the port context, one can be dissatisfied with the conventional portfolio analysis terminology. Therefore, new terms to describe the four possible positions in the portfolio matrix can be proposed. First, the ‘Star’ concept can remain, but slightly renamed as ‘Star Performer’ to make the important point that the long-term sustainability of this favorable position is far from guaranteed; a port or traffic category may be a ‘Star Performer’ today but rapid environmental change or an unsatisfactory internal performance may rapidly alter this situation. Second, the ‘Cash Cow’ concept needs to be eliminated, as a high market share and slow growth rate in the port context are entirely unrelated to actual cash flows accruing to either the port authority or the port operators concerned. A port or traffic category in this position can be more usefully renamed as ‘Mature Leader’, in accordance with the direct significance of the two variables used to determine this position. A similar conclusion holds for the ‘Dog’ concept: there is nothing intrinsically ‘bad’

148

ELVIRA HAEZENDONCK ET AL.

associated with a low market share and below average growth, in contrast to what the ‘Dog’ concept suggests. These elements merely point to the fact that a port or traffic category should be viewed as a ‘Minor Performer’ when compared to other ports or traffic categories. Finally, the ‘Question Mark’ concept is wholly inappropriate because such a position does not at all raise more questions than a position in any of the other three quadrants. A ‘high growth – low market share’ position merely suggests that an individual port or traffic category constitutes a ‘High Potential’, in the sense that a continuation of above average growth over time could also position the port or traffic category in a higher than average market share position. 2.1.3. Shift-share Analysis As mentioned, the SPA also includes a shift-share analysis (SSA). Originally developed in regional economics, De Lombaerde and Verbeke (1989) pioneered the application of the an SSA-technique to international port competition by analyzing the composition and evolution of port traffic flows. The technique allows an unbundling of the overall evolution (growth or decline) of a port’s traffic volume into various components and thereby provides a useful complement to PPA. More specifically, SSA allows decomposing the growth or decline of a variable, i.e. a traffic flow, into three relevant elements:  a share-effect,  a commodity-shift, and  a competitiveness-shift. A negative result for a specific element indicates an unfavorable position, whereas a positive figure reflects a good performance. Moreover, the higher the positive figure, the better the situation can be considered. However, it needs to be recognized that the results of an SSA can be substantially influenced by the choice of ports, the relevant traffic categories, the base year and the observation period. The share-effect indicates the hypothetical estimated growth of traffic in a port, assuming a constant market share of this port in the range. This implies that the share-effect describes the change in traffic volume that would have occurred in a case where all commodity types would have evolved in the same way as the average evolution of traffic in the port range. The difference between the actual recorded growth and the computed shareeffect reflects an increase or decrease in actual recorded market share and is represented by the shift-effect. This effect indicates that a port has not

Strategic Positioning Analysis for Seaports

149

evolved as was expected based on the share-effect and is divided into a commodity-shift and a competitiveness-shift. Given a particular traffic structure, the commodity-shift reflects the degree of specialization of a port in the best performing traffic categories. As a result, this effect takes into account the impact of the unique traffic structure of each port. A positive commodity-shift reflects the fact that a port is specialized in the fastest growing traffic categories, i.e. it has developed a favorable traffic structure. A negative commodity-shift reflects an unfavorable traffic structure. The calculation of a commodity-shift assumes that all traffic categories maintain their initial share in the individual port’s traffic. Therefore, a port that is specialized in the fastest growing traffic categories in the range should also see its overall market share in the range grow; it should grow as a result of the divergence in growth rates of the various individual commodities in the range and the port’s favorable commodity composition. Given a specific (favorable or unfavorable) commodity structure of a port’s traffic, a positive competitiveness-shift implies that the port did ‘better than expected’, i.e. it outperformed its rivals in the traffic categories in which it is specialized. In other words, this shift is a general indicator of the overall improvement or deterioration of the port’s market share in different traffic categories. The composition of the actual growth of the traffic volume in a seaport in the period t to t+x can be represented by the following expression, see, e.g. De Lombaerde and Verbeke (1989): Actual growth ¼ share-effect+commodity-shift+competitiveness-shift  X
t X tþx t tþx t tþx tþx tþx Ptþx P P ¼ p P þ p p þ p p (1) Ptij j j j i ij ij i i

ij

where

Ptj ¼ total traffic in port j in year t. Ptþx ¼ total traffic in port j in year t+x. j Pt ¼ actual growth of traffic volume of traffic category i in seaport j. Ptþx Pij tþx ij t Pij Þ ¼ actual growth of traffic volume in seaport j. i ðPij ptþx Ptij ¼ share-effect of traffic category i in seaport j. P tþx t Pij Þ ¼ share-effect in seaport j. i ðp tþx ðptþx p ÞPtij ¼ commodity-effect of traffic category i in seaport j. Pi tþx ptþx ÞPtij ¼ commodity-effect in seaport j. i ðpi t ðptþx ptþx ij i ÞPij ¼ competitiveness-effect of traffic category i in seaport j.

150

ELVIRA HAEZENDONCK ET AL.

P

tþx pitþx ÞPtij ¼ competitiveness-effect in seaport j. i ðpij tþx Pij ¼ traffic in port j of traffic category i in year t+x. Ptij ¼ traffic in port j of traffic category in year t. tþx

p ¼ relative growth of traffic in the seaport range in year t+x. ¼ relative growth of traffic category i in the seaport range in year t+x. ptþx i ¼ relative growth of traffic category i in port j in year t+x. ptþx ij The approach described above will be applied in the fourth section of this chapter to the traffic figures of the seaports in the North Western European Hamburg – Le Havre range. An important limitation of the conventional SSA-method is the static and merely numerical representation of the SSA-calculations, which makes the clear visualization of results and the formulation of conclusions extremely difficult. In order to remedy this weakness, a graphical representation of SSA-results can be suggested as shown in Fig. 2. In Fig. 2, both the commodity- and competitiveness-shifts are included. The X-axis represents the commodity-shift: a positive commodity-shift indicates a favorable traffic structure, whereas a negative shift indicates an unfavorable traffic structure. The Y-axis in Fig. 2 represents the competitiveness-shift: a positive competitiveness-shift of a seaport points at a strengthening of this port’s competitive position, whereas a negative figure indicates a weakening of this position. Both axes (i.e. the commodity- and competitiveness-shifts) intersect at zero. Indeed, in accordance with the COMPETITIVENESS-SHIFT (INDEX)

Joker

Envied achiever COMMODITY-SHIFT (INDEX)

Waning idler

Fig. 2.

Sleeping beauty

Graphical Representation of SSA-Results. Source: Developed by Haezendonck, Coeck, and Verbeke.

Strategic Positioning Analysis for Seaports

151

algebraic expression above, the average commodity-shift and the average competitiveness-shift for the seaport range always amount to zero. Given that both axes intersect, four quadrants can be distinguished. This situation is similar to the quadrants used when conducting a ‘first level’ portfolio analysis, see supra. The visual representation was inspired by the representation of the strategic role of national subsidiaries in a multinational corporation, suggested by Bartlett and Ghoshal (1992). The latter classification made a distinction between an endogenous variable (internal subsidiary strengths) and a largely exogenous one (country attractiveness) determining the subsidiary role in the firm. This is similar to the approach adopted here. Whereas the basic decision variables in a portfolio analysis are market share and traffic growth, the commodity- and competitiveness-shifts constitute the key parameters in the proposed shift-share visualization. The former variable should be viewed as exogenous in the short-run, as a port’s commodity structure cannot be changed overnight by a port authority or set of port operators, even if they would like to do this, and it is usually associated with several ‘assets’ with little alternative use, including a specific infrastructure and networking capabilities (for example, related to the failure of some ports to become major deep-sea container handlers during the past few decades, in spite of extensive efforts to this effect by both port authorities and a number of port operators). The latter variable represents an endogenous parameter, as it represents actual port performance in existing areas of specialization, compared to other ports. In a conventional portfolio analysis, only one parameter, namely the growth rate, reflects the ‘dynamics’ of port competition. In contrast, with a SSA-visualization the rivalry dynamics are reflected in both parameters. By combining these two elements, four different market positions can be distinguished. According to the positioning within this matrix, ‘Jokers’, ‘Envied Achievers’, ‘Sleeping Beauties’ and ‘Waning Idlers’ can be identified, see Fig. 2. The arrows in the matrix indicate an improvement in the port’s competitive position. Assuming that the matrix positions ports, the port with on the one hand a weaker competitive position in period x than in the previous observation period (x 1), and on the other a favorable traffic structure is regarded as a ‘Sleeping Beauty’; although specialization in the fastest growing traffic categories should in principle confer a long-term advantage vis-a`-vis rivals (‘Beauty’), the port has not performed well with this favorable traffic structure (indicated by the term ‘Sleeping’). If, however, the port is able to improve its market position in these favorable niches with a fast range growth, its position will improve so as to become an ‘Envied

152

ELVIRA HAEZENDONCK ET AL.

Achiever’; ‘envied’ by rivals because of its attractive initial traffic mix, and an ‘achiever’ because of its effective leveraging of this initial position. ‘Jokers’ are ports characterized by an ameliorating competitive position but they are faced with an unfavorable traffic structure. Because of this unfavorable composition of traffic, the sustainability of a good market position is ambiguous, unless the port succeeds at becoming the ‘Mature Leader’ in declining traffic categories (e.g. Antwerp’s specialization in conventional cargo; Haezendonck, 2001). In case an unfavorable competitive performance is combined with an unfavorable traffic structure, a ‘Waning Idler’ is obtained. Ports situated in this quadrant should be viewed as declining, unless corrective (managerial) action is undertaken. This may include improving performance in existing niches in the short-run (creating a positive competitiveness shift) and/or changing the traffic structure itself in the longer-run (attempting a shift to a more positive commodity structure). Finally, it should be mentioned that an extra dimension (share-effect) can be added to the representation. A three dimensional representation allows us to show graphically the total SSA-effect of seaports. In addition, the SSA can be performed for a sequence of time periods considered relevant. The consideration of different time periods is relevant as we can conclude that the relative position of the ports in terms of shift-effects will evolve over time. An analysis over different time periods will provide evidence of this evolution. 2.1.4. Product Diversification Analysis The third and final part of a SPA consists of analyzing the diversification of the port’s traffic for a specific period (De Lombaerde & Verbeke, 1989). The traffic diversification index determines the relative importance of the different traffic categories in the total seaport traffic volume and assesses the composition of this traffic. Here, the Hirshman–Herfindahl index is used to determine the level of diversification of a seaport and, as a result, to indicate the relative importance of the different traffic categories within the total traffic structure of each seaport. The algebraic expression of the diversification-index is (De Lombaerde & Verbeke, 1989)

Dj ¼ 

n P

i¼1 n P

i¼1

P2ij

Pij

2  1

(2)

Strategic Positioning Analysis for Seaports

153

where Dj ¼ diversification-index for port j Pij ¼ traffic volume i of port j The index reflects the degree of concentration of the different traffic categories within the considered port. In a case where the index attains a value of 1, the total traffic structure of the studied port is dominated by one specific traffic category. An equal division of the total traffic volume over the studied categories would result in the lowest possible diversification index of 1 over n, with n representing the number of traffic categories under consideration. As a result, a high diversification index reflects a high degree of inequality (which indicates specialization in one niche), whereas a lower diversification index reflects a balance in volume among traffic categories. 2.2. Synthesis of the Extensions to Conventional SPA 2.2.1. The Necessity of a Weighted Analysis In an SPA, ‘unweighted’ as well as ‘weighted’ traffic structures should be considered. The rationale for a ‘weighted’ analysis is the existence of differences in value added among traffic categories. In order to obtain ‘weighted’ traffic figures, ‘weighing’ coefficients need to be used. Thinking in terms of the differential value added created by various traffic categories, allows the obtaining of information on the seaport’s success in attracting cargo, which generates high value added, i.e. flows such as containers and conventional cargo. A ‘weighted’ analysis provides a positioning of ports, taking into account their performance in terms of ‘intrinsic cargo handling tons’. The ‘weighing’ of traffic data allows us to focus on the value added or welfare created in terms of the contribution to a city’s, region’s or nation’s gross product and it also allows the establishment of links to employment, production and government revenues (Verbeke & Debisschop, 1996). The conventional rules to ‘weigh’ traffic categories include the so-called ‘Bremen Rule’ and ‘Rotterdam Rule’. Haezendonck et al. (2000) demonstrated that both rules lack transparency and have only limited validity. Therefore, a ‘Range Rule’ was developed that allows an appropriate use of the value-added concept in comparative traffic studies, which include, in this case, all considered ports in the Hamburg – Le Havre range. The ‘Range Rule’ reflects the differences in value-added among different traffic categories, i.e. the value of one ton of conventional cargo equals that of approximately one ton of Ro-Ro, three tons of containers, five tons of dry bulk and 13 tons of liquid bulk.2

154

ELVIRA HAEZENDONCK ET AL.

2.2.2. An Extension of the SSA Technique It has been argued above that the SPA approach can be appropriately used to assess the competitive position of ports in a range. Both the portfolio analysis and the diversification analysis are characterized by little operational or interpretative problems. The conventional SSA is associated with two limitations. First, the traffic growth is only divided into three general components. Second, it allows only a static and numerical representation. These limitations can be eliminated by developing a further decomposition of the SSA-results (see infra) and by a graphical representation of the computed SSA-results (see Fig. 2). 2.2.3. Integration of Loaded and Discharged Traffic Figures in the SPA Analysis Separate data on loaded and discharged volumes can be included as an additional element to the conventional SPA. The positioning of ports in the ‘weighted’ version can, as a result, use data on incoming and outgoing flows. This distinction, combined with the results of a SPA using total traffic flows, may provide a more in-depth description of the relative position of ports in the considered range. Indeed, a number of ports may show a relatively high share in loaded tons compared to other ports in the range. It should be mentioned that in the case that the former category of ports performs particularly well in high value traffic flows, such as containers and conventional cargo, this observation would suggest that the presence of a high share of loaded tons and a relatively important volume of high valueadded goods may be interrelated, which provides a dual advantage to the relevant ports. However, for some ports the loaded/discharged ratio in, e.g., container traffic may be relatively high (e.g. larger than 2), but the positive effect causing this high ratio can be minor, owing to an extremely low share of the port’s container traffic in the range’s total container traffic. Hence, these ratios should be interpreted with caution: only in the case that the absolute traffic volumes in a port are sufficiently large, may a high loaded/discharged ratio be interpreted as the reflection of a strong ‘cargo-generating effect’.3

3. EMPIRICAL ANALYSIS 3.1. Introduction In this part of the chapter, the traffic flows of the nine most important seaports in the Hamburg – Le Havre range are analyzed. The Hamburg – Le

Strategic Positioning Analysis for Seaports

155

Havre range includes seaports in North Western Europe between Hamburg and Le Havre, i.e. Antwerp, Ghent and Zeebrugge in Belgium, Rotterdam and Amsterdam in the Netherlands, Hamburg and Bremen in Germany and Le Havre and Dunkirk in France. In the figures included in this section, ports are abbreviated as follows: Antwerp (A), Ghent (G), Zeebrugge (Z), Rotterdam (R), Amsterdam (M), Hamburg (H), Bremen (B), Le Havre (L) and Dunkirk (D). For all nine considered seaports, five different traffic categories are distinguished, i.e. liquid bulk, dry bulk, containers, Ro-Ro and conventional cargo. These traffic categories are abbreviated as follows: liquid bulk (LB), dry bulk (DB), containers (CONT), Ro-Ro (RO-RO) and conventional cargo (CC). Data was collected and analyses conducted for each year over a period of 20 years (1985–2004). By distinguishing four consecutive periods and combining these into single visual representations, interesting insights into the evolution of traffic flows were obtained. The ‘weighted’ traffic figures are the basis of the analyses. Here, ports and traffic categories are positioned and analyzed in terms of three economic variables, namely market share, growth rate and also the value added they create. In order to effectively apply the SSA-instrument to cargo flows in the Hamburg – Le Havre range, the most reliable traffic figures were collected. The data used are port statistics, rather than customs statistics, as the former data are more in accordance with the traffic categories defined in the study. Moreover, customs statistics often underestimate traffic figures since they are characterized by considerable delays in registration, see Coeck, Notteboom, Verbeke, and Winkelmans (1995). However, port statistics are not entirely reliable either, due to differences in definitions adopted by the various ports (e.g. in the port of Rotterdam the container statistics contain the ‘weight’ of the container itself; in the Antwerp statistics this ‘weight’ is not included). Fortunately, this limitation is not critical, as this study focuses primarily on the relative position of seaports (rather than on their absolute traffic volumes) and on the evolution of this position over time. This chapter does not present the entire set of analyses that were actually carried out. Instead, by way of example, only the most important results with respect to total and container traffic in the range, and the port of Antwerp in particular, are included in the figures in this section. 3.2. Portfolio Analysis Applied to the Hamburg – Le Havre Range The main results of the PPA are shown graphically by a number of figures, indicating the different ‘levels’ of the portfolio analysis. In these diagrams, a

156

ELVIRA HAEZENDONCK ET AL.

bold horizontal line and a bold vertical line are included. The latter represents the average market share, i.e. depending on the diagram, the market share in the range or in the individual port’s traffic. The former bold line allows a distinction to be made between relatively fast growing and slower growing seaports or traffic categories, as it indicates the average annual growth rate of the port range or the traffic category in the period 1985–2004. 3.2.1. Portfolio of Ports for Total Traffic (Level 1) In Fig. 3, the port range is considered as a portfolio of ports: the different ports are positioned in the growth-share matrix according to their average market share in the total port range and the average annual growth rate of their traffic volume in the observation period 1985–2004. Here, no description is provided of individual commodity groups or traffic categories. When considering the ‘weighted’ analysis, both Rotterdam and Antwerp grew faster than the range average and positioned themselves in the ‘Star Performer’ quadrants. The position of the port of Hamburg on the contrary, is less pronounced as this port is situated at the intersection of the ‘High Potential’ and ‘Star Performer’ quadrants due to an average market share. It should also be mentioned that, when considering ‘value tons’, the French seaports outperformed the growth rates of all their competitors, except for Portfolio Analysis - Total Traffic Structure of Total Traffic - weighted - period 1985-2004 3.5% Z L D

Average annual growth rate

3.0% 2.5%

H R

2.0% B

A

M

1.5% 1.0% 0.5% 0.0% G -0.5% 0%

5%

10%

15%

20%

25%

30%

35%

Average market share in range

Fig. 3. Portfolio of Ports with their Total Traffic in the Hamburg–Le Havre Range (1985–2004) – Weighted Analysis. Source: Calculations Based on Port Statistics.

Strategic Positioning Analysis for Seaports

157

Zeebrugge. The port of Ghent seems a pronounced ‘Minor Performer’ for its total traffic in the considered period and range. 3.2.2. Portfolio of Traffic Categories for Individual Seaports (Level 2) Fig. 4 shows the PPA of the individual port traffic structure of the port of Antwerp for ‘weighted’ total traffic data. Here, the traffic volume of the individual seaports is considered as a portfolio of five traffic categories, i.e. liquid bulk, dry bulk, containers, Ro-Ro and conventional cargo. The relative share of each category in the total traffic of the selected ports (X-axis) is related to their respective growth rate (Y-axis). The diagram in Fig. 4 represents the total ‘weighted’ version of this type of portfolio for the port of Antwerp during the period 1985–2004. Antwerp has clearly been fast growing for containers. An average annual growth rate of 10% has been recorded for Antwerp in the case where containers are analyzed. Conventional cargo as well as dry bulk, however, appears to lose internal market share in the port, due to a negative average growth rate of respectively 1.35% and 0.56%. Within Antwerp, only container traffic positions itself as a ‘Star Performer’. Liquid bulk and Ro-Ro traffic for the port of Antwerp constitute ‘High Potential’ categories, whereas conventional cargo has a relatively high share in port traffic but is not growing and can therefore be qualified as a ‘Mature Leader’. Portfolio Analysis - Total Traffic Traffic structure of the port of Antwerp - weighted - period 1985-2004 12% CONT

Average annual growth rate

10%

8%

6% RoRo

4% LB 2%

0% DB CC -2% 0%

Fig. 4.

10%

20%

30% 40% Average share in port traffic

50%

60%

Traffic Structure of the Port of Antwerp (1985–2004) – Weighted Analysis. Source: Calculations Based on Port Statistics.

158

ELVIRA HAEZENDONCK ET AL.

3.2.3. Portfolio of Ports for Individual Traffic Categories (Level 3) In Fig. 5, the ‘third level’ of the PPA is shown graphically. The Hamburg – Le Havre range is now analyzed as a portfolio of ports for a specific commodity group. Here, the individual ports’ competitive position in the range is assessed for one traffic category. Container traffic was selected for further description in this chapter. This category has had the highest growth rates in several important ports in the range during the 20-year period that is considered. Container traffic is also aggressively pursued by most seaports in the range (Notteboom, 1997; Baird, 1996; Slack, Comtois, & Sletmo, 1996; Slack, 1985). In this context, Miles (1986) has suggested that portfolio analysis is particularly interesting to identify and further investigate those areas where businesses have the highest yield or potential yield. He has argued that companies should focus on those business units that can gain and hold an edge over competitors and that are valued by customers, see Miles (1986). When translated to the port context, this implies that Rotterdam, Antwerp, Hamburg and Bremen indeed need to focus their strategy analysis especially on containers, given these ports’ favorable range position in this area. As regards total container4 traffic, see Fig. 5, Antwerp and Hamburg are definitely ‘Star Performers’. Their market share and growth rate exceed the average position in the range, i.e. a market share of 11.11% and an average Portfolio Analysis - Total Traffic Structure of Container Traffic - weighted - period 1985-2004 13% 12%

H

Average annual growth rate

11%

Z

10%

A

9% 8% 7%

L

G

B R

6% 5%

D

4% 3% 2% 1% M

0% -1% 0%

5%

10%

15%

20%

25%

30%

35%

40%

Average market share in range

Fig. 5.

Structure of Container Traffic in the H–LH Range (1985–2004) – Weighted Analysis. Source: Calculations Based on Port Statistics.

Strategic Positioning Analysis for Seaports

159

annual growth rate of approximately 8%. Rotterdam still outperforms the other ports with respect to market share (close to 35%), but has grown more slowly than the ‘average port’ in the range and as a result cannot be considered a ‘Star Performer’. The importance of Zeebrugge for container traffic is limited (market share of less than 5%). Nevertheless, it is fast growing (10% average growth rate over 20 years) and can therefore be qualified as having a ‘High Potential’ for container traffic. Bremen has the fourth most important market share, but is growing too slowly to be considered a ‘Star Performer’. Together with Rotterdam it is positioned in the ‘Mature Leader’ quadrant. The French ports and Amsterdam are performing poorly and are therefore ‘Minor Performers’ for container traffic. Le Havre and Ghent come close to the average growth rate but have a small to negligible market share in this traffic category when compared to Rotterdam, Antwerp and the German ports. 3.2.4. Portfolio of Ports for Individual Traffic Categories, Related to these Categories’ Share in Port Traffic and Share in the Range (Level 4) The ‘fourth level’ of the static approach of this portfolio analysis is represented by Fig. 6. The difference with the ‘third level’ of PPA is that here the X-axis represents the share of a specific category within a port rather than the share of this category in the range, i.e. each traffic category considered is being examined in relation to the share of this category within

Average annual growth rate

Portfolio Analysis - Total Traffic Containers vs port traffic - weighted - period 1985-2004 13% 12% 11% 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% -1% -2% -3% -10%

A

H

Z

G

B L R

D

M

0%

10%

20%

30%

40%

50%

60%

Average share in port traffic

Fig. 6. Analysis of Total Container Traffic (Versus Total Port Traffic) of the H–LH Range (1985–2004) – Weighted Analysis. Source: Calculations Based on Port Statistics.

160

ELVIRA HAEZENDONCK ET AL.

the seaport considered. This level also introduces an additional dimension to the portfolio analysis: a circular shape with a surface proportional to the absolute traffic volume of the port considered in the total range. The center of each circle represents the coordinates of growth rate and market share. The main advantage of this graphical presentation is that, for each seaport in the range, the position (of a traffic category in the commodity structure of a single port), the size (of the considered commodity in relation to the size of the same commodity in the other ports in the range) and the growth rate (of the commodity for the port in question) are presented simultaneously. In Fig. 6, the bold vertical line now represents the average share of the relevant traffic category in all ports within the range for 1985–2004, whereas the bold horizontal line reflects the average annual growth rate of this traffic category in the range, and as such is identical to that in the diagram of ‘level three’. In Fig. 6, container traffic is shown graphically. Here, two ports can be considered as a ‘Star Performer’, i.e. Hamburg and, less pronounced, the port of Antwerp. Rotterdam, Le Havre and Bremen also have a relatively large share of containers in their port traffic as well as in the range, but show a lower than average growth rate and are ‘Mature Leaders’. Zeebrugge reveals substantial potential in this traffic category with a growth rate that is 2% higher than the range average. As a result, this port can be considered as being ‘High Potential’. The ports of Ghent, Amsterdam and Dunkirk reveal less potential in the area of container traffic. 3.2.5. Dynamic Portfolio of Ports A dynamic version of a PPA requires comparing various data series, related to different time periods. In a dynamic portfolio analysis the portfolio technique is applied a number of times, i.e. for different time periods, and the results are integrated into a single diagram. The main objective of a dynamic analysis is to reveal the evolution between the different time periods, in order to evaluate possibilities for future developments. In the dynamic analyses undertaken, four time periods have been selected, i.e. 1985–1990 (first period), 1990–1995 (second period), 1995–2000 (third period) and 2000–2004 (fourth period). In Fig. 7, a dynamic positioning of the container traffic structure of four selected seaports is represented. The ports of Rotterdam, Hamburg and Bremen were selected based on their competitive position with regard to the Antwerp port in terms of container traffic. A dynamic portfolio analysis of the container traffic is shown graphically in Fig. 7. The lines in Fig. 7 represent the chronological order between the time periods. With respect to containers, only Antwerp and Hamburg have

Strategic Positioning Analysis for Seaports

161

Dynamic Portfolio Analysis - Total Traffic Structure of Container Traffic for 4 most important container ports in H - LH range (weighted) 18% 1985-1990 1990-1995 1995-2000 2000-2004

H 16%

Average annual growth rate

14% 12% 10% A

8%

R 6% B 4% 2% 0% -2% 0%

5%

10%

15%

20% 25% 30% Average market share in range

35%

40%

45%

Fig. 7. Dynamic Analysis of the Container Traffic for the Ports of Antwerp, Bremen, Hamburg and Rotterdam (1985–2004 in Four Time Periods) – Weighted Analysis. Source: Calculations Based on Port Statistics.

succeeded in maintaining their ‘Star Performer’ position over the full 20year period. Rotterdam has lost market share (but still holds over 35% market share in the range), while its growth rate has been decreasing, but has grown strongly in the last period considered (from 0 to over 12%). It has therefore shifted from a ‘Mature Leader’ to a ‘Star Performer’. Because of its slower growth rates in previous periods, it has close to 8% market share in the range. Bremen, which is widely considered as a container port, was in fact a ‘Mature Leader’ in the first and second period considered, while substantially increasing its growth rate in the third period, and then again moving to a ‘Mature Leader’ position in the last period. 3.3. SSA Applied to the Hamburg––Le Havre Range In this section, SSA is applied to analyze the traffic flows within the Hamburg–Le Havre range for the period 1985–2004. Again, four periods are distinguished, i.e. 1985–1990, 1990–1995, 1995–2000 and 2000–2004. Only the results of the last two periods will be graphically represented, see Figs. 8 and 9.

162

ELVIRA HAEZENDONCK ET AL. Shift-Share Analysis: Total traffic Hamburg - Le Havre range - period 1995-2000 4000 Z

2000

M

B

Competitiveness-shift

G A -2500

-2000

-1500

-1000

-500

0 0

L

500

1000

-2000

D

1500

2000

2500

H

-4000 -6000 -8000 R

-10000 -12000

Commodity-shift

Fig. 8. Shift-Share Analysis: Visualization of Shift-Effect of Total Traffic (1995– 2000) in the Hamburg–Le Havre Range (Weighted Analysis). Source: Calculations Based upon Port Statistics.

Shift-Share Analysis: Total traffic Hamburg - Le Havre range - period 2000-2004 8000

Competitiveness-shift

D

6000 4000 R 2000

M -3000

-2000

A

G -1000

0 0 -2000

1000 L

2000

3000

4000H

5000

B

-4000 -6000 Z -8000

Commodity-shift

Fig. 9. Shift-Share Analysis: Visualization of Share-Effect of Total Traffic (2000– 2004) in the Hamburg–Le Havre Range (Weighted Analysis). Source: Calculations Based upon Port Statistics.

Strategic Positioning Analysis for Seaports

163

A conventional SSA on total traffic volumes was conducted. SSA tables distinguish seven elements: (1) the observation period, (2) the absolute recorded traffic growth, (3) the annual growth during the observation period in percentage terms, (4) the share-effect, (5) the commodity-shift, (6) the competitiveness-shift and (7) the total shift-effect. 3.3.1. Share-effect As regards total traffic volume, all seaports in the range have been characterized by a positive share-effect in all four observation periods, but especially in the last period (2000–2004), due to a substantial overall traffic growth in the range. 3.3.2. Shift-effect In the conceptual part of this chapter, we propose using a graphical representation of the shift-effect to facilitate the formulation of results. Figs. 8 and 9 show the most important results of the nine seaports considered in the Hamburg–Le Havre range as regards their total weighted traffic for two periods in time, respectively 1995–2000 and 2000–2004. The axes in Figs. 8 and 9 represent the values of respectively the competitive and the commodity shift. These figures do not directly refer to the realized traffic figures but are the result of the calculations based upon the SSA-formula described in the theoretical part of the chapter. In the first period (1985–1990), the Belgian ports of Antwerp and Zeebrugge, and to a lesser extent the ports of Rotterdam and Dunkirk recorded a positive competitiveness-shift, which in the case of Antwerp, was unfortunately overcompensated by a very negative commodity-shift (see Fig. 8). In contrast, the port of Zeebrugge was characterized by a very positive traffic structure and was, together with Rotterdam, a pronounced ‘Envied Achiever’ in the range for that period. The negative commodityshift for the port of Antwerp led to a positioning in the ‘Joker’ quadrant. The ports of Bremen and Le Havre recorded a somewhat positive traffic structure, but their competitive position weakened and, as a result, these ports were ‘Sleeping Beauties’ in 1985–1990. The other ports in the range combined a negative traffic structure with a weakening competitive position, thus being ‘Waning Idlers’. During the period 1990–1995 (see Fig. 9), the ports of Zeebrugge and Hamburg succeeded in reaching the ‘Envied Achiever’ status. Antwerp maintained ‘Joker’ position, but again with a very negative commodity shift ( 2,888.28). Amsterdam became a ‘Joker’. The Rotterdam, Bremen and Le Havre seaports were able to combine a positive commodity-shift with a

164

ELVIRA HAEZENDONCK ET AL.

negative competitiveness shift and, as a result, Rotterdam joined the other two ports as a ‘Sleeping Beauty’. The port of Ghent remained a ‘Waning Idler’, confirming its unfavorable competitive position. Fig. 8 shows that during the period 1995–2000, Zeebrugge lost most of its large shift-effect, composed of a substantial competitiveness-effect, but also a very negative commodity-shift. In the same period, the port of Hamburg recorded a slightly negative shift-effect and joined Rotterdam and, to a lesser extent, Le Havre in the ‘Sleeping Beauty’ quadrant. The Antwerp seaport was still a ‘Joker’, but reduced the size of its negative commodity shift-effect. The ports of Amsterdam and Ghent obtained a positive competitive position but still had an unfavorable traffic structure. Dunkirk noted an overall negative shift and was a ‘Waning Idler’. The last period (2000–2004), graphically represented in Fig. 9, shows some remarkable results with respect to the competitive position and the commodity structure of the seaports in the Hamburg–Le Havre range, as compared to the previous periods. Not one port was able to maintain the position of ‘Envied Achiever’. As regards the total shift, the Dutch ports have been characterized by a very favorable competitive position (the port of Rotterdam, in particular, worked successfully on its competitive position), but do not appear to have been able to focus more on the traffic categories characterized by the highest growth rates. Most especially, the Amsterdam seaport recorded an even more negative commodity structure. Zeebrugge’s competitive position was seriously weakened (from +2,000 to 7,000) and became a ‘Waining Idler’. The other Belgian ports of Antwerp and Ghent lost their slightly positive competitive position and joined Zeebrugge as ‘Waning Idlers’. Bremen lost its positive competitive position and joined Le Havre and Hamburg as a ‘Sleeping Beauty’. It is important to note that the two largest competitors in the range in terms of total value tons generated in the last 20 years, i.e. the ports of Rotterdam and Antwerp, have rarely succeeded in combining their favorable competitive performance in their areas of specialization with a more positive overall traffic structure, indicating that ports indeed carry with them an important ‘heritage’ from the past, that acts as a mobility barrier (Ghemawat, 1999) but sometimes can be beneficial when a ‘total service package’ is pursued. 3.4. Diversification Analysis Applied to the Hamburg––Le Havre Range: Diversification Index The diversification indices were computed for all nine considered ports in the Hamburg–Le Havre range for the period 1985–2004. Moreover, a

Strategic Positioning Analysis for Seaports

Antwerp Ghent Zeebrugge Rotterdam Amsterdam Hamburg Bremen Le Havre Dunkirk

Fig. 10.

1985 0.47 0.43 0.69 0.23 0.40 0.33 0.35 0.23 0.28

1990 0.39 0.38 0.58 0.22 0.35 0.27 0.29 0.25 0.35

165 1995 0.31 0.38 0.56 0.22 0.39 0.30 0.29 0.29 0.31

2000 0.32 0.35 0.57 0.22 0.41 0.45 0.36 0.28 0.33

2004 0.32 0.35 0.49 0.24 0.41 0.57 0.40 0.36 0.35

Diversification Analysis Index (Minimum Index is 0.20). Source: Calculations Based upon Port Statistics.

distinction was made among total, loaded and discharged traffic. As five traffic categories are distinguished in the study (i.e. liquid bulk, dry bulk, containers, Ro-Ro and conventional cargo), the minimum value of the index is 0.20, indicating the highest possible level of diversification in terms of traffic volume. The diversification analysis is based upon weighted traffic figures derived from the value-added created by the different commodities and not merely upon absolute traffic volumes transshipped (Fig. 10). Considering the diversification indices of total ‘weighted’ traffic for 2004, the ports of Rotterdam and Antwerp, followed by the French ports considered and the port of Ghent, show the most favorable diversification of port traffic. This implies that these ports have highly diversified total traffic structures in value terms as compared to the other ports considered in the Hamburg–Le Havre range. From a dynamic perspective, an important decrease in the diversification of Hamburg, Bremen and Le Havre, and an important increase in the diversification of the port of Zeebrugge can be noted, when compared to the reference year 1985. The ports of Zeebrugge, Amsterdam, Bremen and Hamburg appear to be specialized in specific cargoes; as already shown in the PPA conducted in this chapter, Zeebrugge can be considered a Ro-Ro port, whereas Amsterdam is to be viewed as a bulk port. To a lesser extent, the ports of Bremen and Hamburg are conventional cargo ports. As regards the total ‘weighted’ traffic structure, only Rotterdam records a diversification that is as favorable as the range’s overall diversification in 2004. However, it should be emphasized that all ports in the range show a rather diversified ‘weighted’ traffic structure, as their indices appear much closer to the maximum diversification possible (index ¼ 0.20) than to the maximum degree of specialization (index ¼ 1). This is an interesting observation, because it may imply that, at least in Western Europe, large ports can only be sustained in the longer run, if they are well diversified.

166

ELVIRA HAEZENDONCK ET AL.

The Flemish ports, in particular, realize a favorable result. The diversification index of e.g. Hamburg records a negative evolution over the years.

4. CONCLUSION In this chapter, we have developed a SPA tool to provide indications of the overall competitive position of individual ports and specific traffic categories in these ports. The term ‘competitive’ is relevant, as the instrument does not merely provide an evaluation of the comparative positioning of seaports but also adds a ‘value-added’ dimension. By integrating this element through the use of weighted traffic figures, the proposed SPA analysis extends well beyond traditional comparative analysis. SPA can be useful both for internal and external positioning purposes. A number of simple tools, i.e. portfolio analysis, SSA and diversification analysis – originally developed in the field of strategic management – were adapted to the seaport context, and applied to the study of traffic flows in the Hamburg–Le Havre range. The SPA combines these instruments and allows an integrative description of the competitive position of individual ports, as well as the evolution of this position. The innovative aspect of an SPA is that no choice needs to be made between instruments to perform the analysis. The SPA is an integrative instrument in such a way that no preference is necessary for an SSA or a PPA; the different instruments provide complementary information on port activities. From a strategic management perspective, seaport authorities (and in many cases port operators) perceive improvements in their port’s market share, its growth rate and its degree of traffic diversification as major goals to be pursued. To the extent that seaport authorities and other stakeholders are indeed primarily interested in (a) increasing their port’s market share visa`-vis rivals in the range and (b) a stronger diversification of their traffic structure, information on the evolution of these two parameters in different competing seaports should be considered as extremely valuable. However, viewing market share and growth rates of maritime traffic flows expressed in ‘nominal’ tons as the prime indicators of a port’s ‘effectiveness’, is not necessarily conducive to achieving economic efficiency (Goss, 1990). In order to approximate a focus on ‘efficiency’ goals, the ‘value added’ concept was introduced in the SPA in this chapter. A focus on ‘value added’ allows decision makers to think in terms of ‘value tons’ rather than nominal tons, i.e. in terms of contribution to local, regional or national gross product.

Strategic Positioning Analysis for Seaports

167

Mainstream port economics scholars could of course argue that the SPA as developed in this chapter has only limited value as a guide to decision makers, since the different subcalculations performed here sometimes lead to contradictory outcomes in terms of individual port performance and range positioning, and since the position of individual ports also appears to fluctuate rather strongly over different time periods. However, it should be kept in mind that the SPA tool intends to aid in strategy formulation and that the, sometimes contradictory, outcomes, e.g., depending upon the level of portfolio analysis performed, may instigate debate and in-depth reflection on what port strategy to be pursued. Here, the contribution of this chapter to economic literature needs to be made explicit. The unique feature of the SPA is that it provides an integration of conventional instruments used in different parts of economic science and is based upon freely available information. However, the valuable information provided by the SPA still needs to be complemented with financial, managerial and societal aspects (based upon social cost-benefit analysis, environmental impact assessment and economic effect analysis), in order to be able to select, e.g. the most appropriate way of developing port activities. The comprehensive way in which the available information is processed and shown graphically substantially improves the thinking and the capabilities of the decision maker. Therefore, the use of an SSA is advisable in order to gain further insights into port competition and to be able to assess alternative scenarios for port development.

NOTES 1. ‘Question marks’ are sometimes also referred to as ‘Wild cats’ or ‘Problem children’ in the relevant literature. 2. In Haezendonck et al. (2000), a difference was made between crude oil (coefficient 18) and other liquid bulk (coefficient 5). The collected data for the empirical analysis unfortunately do not allow a clear differentiation between these two categories of liquid bulk. Therefore, these two categories are merged, taking into account their relative importance (in tons, %) in the considered range. 3. The presence of manufacturing industries in the Antwerp area and its vast hinterland enhance the ‘cargo-generating effect’. The well-developed hinterland connections and the fact that the Antwerp seaport is a frequent port-of-call for major shipping alliances, largely explain the traffic volume attracted to the port of Antwerp (Blomme, Foulon, & Uyttenhove, 1997; Notteboom, 1997). 4. No important differences can be observed among total, loaded and discharged analyses for container traffic.

168

ELVIRA HAEZENDONCK ET AL.

REFERENCES ATENCO. (1999). Analysis of the cost structures of the main ten ports. Final report. Technum Flanders Engineering NV, Antwerp. Baird, A. J. (1996). Containerization and the decline of the upstream urban port in Europe. Maritime Management and Policy, 23(2), 145–150. Bartlett, C. A., & Ghoshal, S. (1992). Transnational management. Homewood: Irwin. Bettis, R., & Hall, W. (1981). Strategic portfolio management in the multibusiness firm. California Management Review, 24(1), 23–38. Blomme, J., Foulon, A., & Uyttenhove, J. (1997). Duidelijk ge(s)teld: Het containervervoer in Antwerpen. Hinterland, 174, 4–9. Brown, R. (1991). Making the product portfolio a basis for action. Long Range Planning 24(1), 102–110. Coeck, C., Notteboom, T., Verbeke, A., & Winkelmans, W. (1995). The unreliability of maritime trade statistics: An extension of results. International Journal of Transport Economics, 22(2), 217–224. Day, G. S. (1977). Diagnosing the product portfolio. Journal of Marketing, 41(2), 29–38. De Lombaerde, P., & Verbeke, A. (1989). Assessing international seaport competition: A tool for strategic decision making. International Journal of Transport Economics, 16(2), 175–192. Dibb, S., Simkin, L., Pride, W. M., & Ferrell, O. C. (1991). Marketing: Concepts and strategies. Boston: Houghton Mifflin. Dyson, R. G. (1990). Strategic planning: Models and analytical techniques. England: Wiley. Ghemawat, P. (1999). Strategy and the business landscape. Reading: Addison-Wesley. Goss, R. O. (1990). Economic policies and seaports: The economic functions of ports. Maritime Policy and Management, 17, 207–220. Haezendonck, E. (2001). Essays on strategy analysis for seaports. Louvain: Garant Publishing. Haezendonck, E., Coeck, C., & Verbeke, A. (2000). The competitive position of seaports: Introduction of the value added concept. International Journal of Maritime Economics, 2(2), 107–118. Hambrick, D. C., & MacMillan, I. C. (1982). The product portfolio and man’s best friend. California Management Review, 25(1), 84–95. Hax, A. C., & Majluf, N. S. (1983). The use of the growth-share matrix in the strategic planning. Interfaces, 13(1), 46–60. Henderson, B. D. (1979). Henderson on corporate strategy. Cambridge, MA: Abt Books. McKiernan, P. (1992). Strategies for growth. London: Routledge. Miles, A. W. (1986). Renaissance of the portfolio. Perspectives. San Francisco: The Boston Consulting Group. Notteboom, T. (1997). Concentration and load centre development in the European container port system. Journal of Transport Geography, 5(2), 99–115. Notteboom, T., & Coeck, C. (1994). Strategische positionering binnen het Belgische goederenvervoer. Tijdschrift Vervoerswetenschap, 2, 85–110. Slack, B. (1985). Containerisation, inter-port competition and port selection. Maritime Management and Policy, 12, 293–303. Slack, B., Comtois, C., & Sletmo, G. (1996). Shipping lines as agents of change in the port industry. Maritime Policy and Management, 23(3), 289–300. Verbeke, A. (1992). Een strategische positiebepaling van Noordnatie binnen de range Hamburg – Le Havre. Final report. Policy Research Corporation NV, Antwerp.

Strategic Positioning Analysis for Seaports

169

Verbeke, A., & Debisschop, K. (1996). A note on the use of port economic impact studies for the evaluation of large scale port projects. International Journal of Transport Economics, 23(3), 247–266. Verbeke, A., Peeters, C., & Declercq, E. (1995). De toepassing van de produkt portfolio methode in functie van een zeehaven-strategie. Tijdschrift Vervoerwetenschap, 3, 231–242. Wind, Y., & Mahajan, V. (1981). Designing product and business portfolios. Harvard Business Review, 59(January–February), 155–165. Winkelmans, W., & Coeck, C. (1993). Strategic positioning analysis as an evaluation instrument for effective port policy. Planologisch Nieuws, 13(3), 263–270.

This page intentionally left blank

170

PORT INVESTMENT: PROFITABILITY, ECONOMIC IMPACT AND FINANCING Enrico Musso, Claudio Ferrari and Marco Benacchio ABSTRACT Port investment is a key issue in modern port economics with respect to planning port development, financing and assessing the return on investment. This chapter addresses some of the features related to port investment, starting from the evaluation of the main paradigms that characterize the port industry from a global point of view, and focusing on the relations, synergies and conflicts between the numerous stakeholders actually involved. Profitability, economic impact and financing are seen as the most critical nodes in the complex chain of port investment decisions.The chapter builds up a comprehensive scenario where single aspects and variables related to port investments can fit into a general scheme of interrelations that identifies feasible outcomes. The foreseeable outputs in terms of demand and supply provide insights for possible incentives to efficiency to be improved upon by decision-makers at different levels, promoting the reduction of conflicts and a synergy of interests.

Port Economics Research in Transportation Economics, Volume 16, 171–218 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16008-4

171

172

ENRICO MUSSO ET AL.

1. INTRODUCTION As far as investing in port assets is concerned, there are two ways, almost in contrast with one another, of regarding the port:  The port may be considered a public service that is generally useful to the economy, justifying the tax system being utilised for the purpose of funding the investments required.  The port may be considered a business system that operates within a highly competitive market and requires investment projects to be selected with efficiency. The line drawn between these two functions changes, depending on the country, environment, business, social and political culture, period and political trends. In most institutional models, the large infrastructures that either provide access to a port or are used for general purposes attract public investments, while terminal superstructures are instead invested in by the terminal company itself. Halfway between these two spheres, a terminal’s special-purpose infrastructures may be the focus of either side. The difficulties encountered when attempting to identify optimal criteria for investment decisions and the funding of investments stem from the particular intensity and complexity of relationships between the port, the transport system and the economy. On the one hand, a port investment usually involves sizeable external costs, both direct (caused by the port infrastructure itself) and indirect (incurred by transport activities generated by the port). On the other hand, it leads to strong external economies – be they ‘‘microeconomic’’, involving operators positioned downstream within the productive process (carriers, manufacturers and consumers of the goods transported), or ‘‘macroeconomic’’. The latter actually concerns ‘‘catalytic’’ effects, such as attracting businesses and creating jobs and income within the area of the port, as well as effects whereby income is multiplied (in a Keynesian sense) and investment is accelerated (in a neoclassical sense). Externalities (both positive and negative) give place to a typical ‘‘market failure’’, accompanied by others: the possible natural monopoly situation and the public or club nature of a number of the capital goods making up the port development. Market failures and positive impacts on the local economy and the hinterland have by tradition led to heavy public involvement in investment decisions and funding, leading to the coexistence of different investment

Port Investment

173

evaluation criteria (direct private profitability and collective benefits/costs). The source of funding has an influence on the investment selection process. Private funding meets solely the criterion of risk being remunerated through expected profits, and ignores possible external factors that may have a positive effect on the economy. Public financing, on the other hand, frees investments from the burden of direct profitability, but in the process introduces potential inefficiencies, leads to the ‘‘crowding-out’’ of private capital and alters competition among ports. Possible distortions on competition also suggest a public role in market regulation for the protection of market contestability. On the other hand, the strong public role that has historically characterised the growth of ‘‘capital’’ (for both strategic or military reasons) and the productive capacity of the majority of ports have, in turn, given rise to the several so-called ‘‘government failures’’ that over the last few decades have caused a growing number of countries (including the European Union) to attempt to reinforce market mechanisms within the context of port operations. All of these lead to an intrinsically mixed model, where the effects (be they direct or external) of investments, investor strategies and public administration bodies play an essential role and modify the ‘‘general’’ aspects of investment theories. These peculiarities have repercussions on funding, pricing and tax systems, and therefore on the institutional and management models adopted for the port as well. In this chapter, Section 2 will deal with the key features of port investments, in terms of both ‘‘direct’’ profitability and social desirability. Section 3 recreates the ‘‘chain’’ of effects of a port investment and the financial flows stemming from it. Section 4 comes up with a simple model that enables the effects of investments – and in particular, the direct benefits (profits) and indirect benefits – to be regarded comprehensively and in relation to one another. Section 5 looks at investment as a crucial component within the competitive game played by port businesses, expanding upon their ability to limit the number of new entrants into the market. Section 6 examines the issue of inter-port competition and the possibility that, in a number of cases, port investment may ultimately distort the competitive game played by businesses, outlining the actions involved in a policy agenda targeting this particular matter. To end, Section 7 takes its cue from the conclusions drawn by the model for considerations regarding the funding of investments, through pricing and tax revenues, identifying in the process a number of policy guidelines.

174

ENRICO MUSSO ET AL.

2. PORT INVESTMENTS, PROFITABILITY, SOCIAL DESIRABILITY Investment is a variation of the total stock of capital goods used in productive activities. In the port sector this is necessarily a variation in instrumental assets, as the product – the throughput – is a service and therefore cannot be stocked. Investment is carried out by a port business in order to have the desired level of throughput capacity at its disposal. Investing in ports, therefore, has a direct impact on overall port capacity and supply. Neoclassical production theory expresses investment as a variation over time of the level of capital used by a business. It usually hypothesises a standard (Cobb–Douglas) production function as Qt ¼ L1t b K bt

(1)

where L and K are, respectively, the amounts of labour and capital employed over a period of time, investment is the variation in capital levels DK, which takes place between one period and the next.1 According to neoclassical economists, the investment decision is a direct function of the amount of capital needed to produce the level of output Q deemed optimal by a business (for example, the amount needed to maximise its profits), and an inverse function of the interest rate, which is the cost of the investment. The investment, as a variation in the level of capital, will be equal to I t ¼ Kðit ; Qt Þ

Kðit 1 ; Qt 1 Þ

(2)

According to Keynesian theory, investment takes place if the marginal efficiency of capital2 is higher than the market interest rate, which represents the return of the other possible uses of the resources employed. It is also normally considered that the rate of profit expected from the investment should be greater than the interest rate plus a spread (to ‘‘reward’’ the risk of the profit achieved proving to be lower than that originally expected). Since the potential investor will rank possible investment projects starting from those with the highest marginal efficiency, the well-known inverse relationship results between (cumulated) investment and the market interest rate. In the port industry, the product is throughput, and the investment is the creation of throughput capacity. Port investments are those increases in capital goods that allow greater throughput via an increased efficiency in

Port Investment

175

using the production factors (Wiegmans, Ubbels, Rietveld, & Nijkamp, 2002). These include the following:  infrastructures, such as breakwaters, dams and lock systems that enable access along canals and rivers, the excavation or dredging of riverbeds and the construction of new piers, wharfs, yards, etc.;  terminal ‘‘superstructures’’ (cranes, means of transport, buildings used for storage or port services); and  other assets useful for the production of port services. Most port investments – particularly infrastructural investments – bear the following features:  their profitability is at least in part indirect, since they are part of collective capital, which acts as a location factor for business activities and generates positive externalities;  they also generate environmental costs and negative externalities;  the construction of infrastructures brings with it significant indivisibilities, owing to economies of scale, financial requirements and network economies;  they require considerable time to be accomplished, including a lengthy planning and design period, and subsequently boast an extremely long economic life. As a result, there is a hefty time lag between costs (incurred primarily before the port comes into operation) and revenues, and a long payback period for the investment itself;  high risk and high uncertainty of expected profit, due in part to the difficulty of estimating costs;  in the case of ‘‘general purpose’’ assets (such as dams, canals and basins), cost cannot be imputed to individual users,3 while the benefit for each user cannot be quantified either; and  infrastructure costs are ‘‘sunk’’ (i.e. lost whenever the investor decides to withdraw from the market), and therefore act as ‘‘exit barriers’’ that jeopardise the market’s contestability and create the risk of a monopoly. Some capital goods are characterised by non-rivalry, and more rarely by non-excludability. In cases of non-excludability and non-rivalry (e.g. a lighthouse, or an asset relating to safety within port facilities), we are dealing with a ‘‘public good’’ that, being subject to free riding, is not profitable for a private operator. If the capital good is just non-rival, then it is referred to as a ‘‘club good’’, with different consequences on pricing criteria (see Section 4). Traditionally, and although some would disagree that costs cannot be fully recovered from users, these features have mainly burdened the public

176

ENRICO MUSSO ET AL.

sector with the cost of port investments, which includes the realisation and maintenance of:  general-purpose infrastructures that provide access to the port by land (connections to land-based infrastructural networks) and by sea (access channels, locks, breakwaters, dams, port basins, lighthouses, etc.), or in any event for general use (safety-oriented infrastructures, signalling, telecommunications);  infrastructures used specifically by a single terminal (yards, specific transport infrastructure, reclamation work carried out prior to the realisation of superstructures); and  terminal superstructures, including transportation systems, cranes, warehouses, office buildings, etc., and other instrumental assets used to produce saleable services. Usually, large general-purpose infrastructures attract public investments, while terminal superstructures are invested in by the terminal company itself, and terminal infrastructure may either be funded by public or private players. In many ports, however, companies operating terminals are corporatised but publicly owned bodies, more or less separate from the port authority; while even where terminals are leased out or franchised to private operators, public financing sometimes covers (at least in part) the costs of the superstructures involved. Where public financing is restricted entirely, private projects for new terminals or ports (particularly in developing countries) may be funded in full with private capital. In recent years, large shipping groups trying to realise a dedicated terminal or logistics platform have generally plumped for substantial autonomy in their investment decisions. On the other hand, recently developed countries with abundant public resources (as has been the case for oil-exploring countries) fund new terminals or new ports entirely, precisely because they regard them as development drivers, above all in terms of their ability to boost accessibility (catalytic impact). The relationship between port investment, on one side, and ownership, financing and pricing criteria, on the other side, is an extremely complex one: we shall return to this relationship in Sections 4 and 5. A port investment thus represents a potential increase in throughput, which is actually achieved where there is demand for it. The investment allows, and normally causes, an increase in production and in the demand for factors (capital, labour, enterprise, space, social capital, skills, etc.). In this way it gives rise to  a direct profit for the investor, which we usually assume is also the manager of the port asset realised with the investment;

Port Investment

177

 external microeconomic benefits caused by a reduction in generalised cost and the consequent surplus transferred to the port users (the carriers), and through them to their own clients (shippers and consumers in the hinterland);  external macroeconomic benefits resulting from the attraction of firms in the area, the increase in employment and earnings within the local economy (catalytic effect), and Keynesian multiplier effects; these effects are the result of a higher accessibility and/or of an increased demand for goods and services generated directly or indirectly by the port (portrelated industries); and  a negative macroeconomic effect, arising mainly from environmental costs and other (e.g. social) external costs, and their negative location effect on certain ‘‘port-rejected’’ industries (such as tourism). Both microeconomic and macroeconomic external benefits will give rise to effects such as increases in employment, production, earnings and consumption, as well as produce long-term effects in terms of accessibility, the efficiency of the productive–logistic system and possible innovation capabilities. Direct benefits depend upon throughput, while external benefits depend partly upon throughput and partly upon the impact of a port on the location patterns and consumption of firms. The nature of these effects is complex, and their boundaries are not always clear-cut. A key point is the level of excludability from these effects: an investment brings direct benefits to operators that may be excluded if they do not pay for it (stevedores, carriers), and indirect benefits (identified herein as ‘‘macroeconomic’’) to third parties that cannot be prevented from enjoying them (e.g. owners of land close to the port). There is therefore the need for a mixed decisionmaking process, both private and public, which should occur at both local and national level. The costs of the investment are the opportunity costs of input used to realise the asset, input that may be privately (money, labour, space) or publicly (tax receipts, public space, social capital) owned. Direct costs are made up of the monetary capital needed to build nautical structures, infrastructures and systems or to acquire other assets, as well as to acquire any private land on which the port facility may be built. External costs are deterioration or consumption of environment, collective space and capital stock (namely infrastructure). The decision to carry out an investment may therefore concern  a private operator that invests if the profit expected (marginal efficiency of capital) is greater than the market interest rate (to which a risk

178

ENRICO MUSSO ET AL.

premium is added) pe  i þ rp

(3)

 a public operator that, in principle, should invest if external utility is greater than external costs X NPV ¼ ½ðB CÞt ð1 þ iÞ t Š  0 (4)

In the first case, if the return turns out to be below that expected, the private investor may try and obtain a public subsidy to cover the difference, giving rise to different forms of mixed investment. In both cases, the public administration should invest (or encourage or allow private investment) if the total utility (private+external) is higher than total cost (private+external). According to some (Bonnafous & Jensen, 2005), a public investment takes place where the socio-economic internal rate of return (SE-IRR) is above a certain benchmark, this being a condition of collective profitability. However, if the investment is entirely public, the ranking between the various investment options may quite simply be provided by a decreasing SE-IRR, while budgetary restrictions will instead determine whether projects actually go ahead or not. Comparing direct benefits and costs evaluates the port investment’s profitability, while comparing external benefits with costs evaluates its social desirability or usefulness. The former analysis is undertaken by the (potential) investor, while the latter analysis is undertaken by those appointed to represent the collective interest (central or local government, port authority). As for the former, any decision to carry out the investment will depend on expectations regarding future demand trends, the behaviour expected from competitors and the investor’s risk aversion. The decision will also depend on the degree of costless reversibility and on the degree of uncertainty (see below, Section 5). As for the latter, any decision to carry out the investment will depend on the evaluations of the political decision-maker, which will in turn be based on an analysis of the benefits and costs of the investment for the collectivity. These analyses (such as the costs/benefits analysis) enable us to estimate the net present value (NPV), i.e. the present value of the investment’s net benefits, or to establish its IRR, i.e. the rate at which the present value of the net benefits expected is equal to the cost of the investment. In principle, there is public interest in the investment if NPV is positive (or above a predetermined standard), or if IRR is higher than the market interest rate.

Port Investment

179

b c

private profitability

Social utility

a

Fig. 1.

Direct Profitability and Social Utility of Investment.

Comparing direct usefulness (profitability), be it positive or negative, with external usefulness (be it positive or negative) produces four possible combinations, shown in Fig. 1 as a Cartesian graph, where direct profitability (profit forecast) is shown along the abscissa and social utility (net benefit) is shown along the ordinate. Assuming that the coordinates at the origin of the axes are, respectively, market interest rate (to which a risk premium may be added), and 0 (or, alternatively, the above-mentioned ‘‘standard’’ socioeconomic internal rate of return), the bisector of quadrants II–IV separates the situations bearing total (direct+external) positive utility, above the bisector, from those bearing total negative utility, below the bisector. Private profitability normally stems from the private nature of benefits (port services or assets are ‘‘private goods’’, featuring excludability and rivalry between users): in the port arena, this can be the case for services (both to goods or to ships), superstructures (cranes) and, to a lesser extent, terminal infrastructure. Public profitability stems from the existence of longterm external benefits, such as hinterland accessibility, ‘‘public’’ or ‘‘club’’ goods such as nautical assets (dredgings, breakwaters, locks, etc.), landbased networks and general local accessibility. Fig. 1 shows the situations that may then arise. Quadrant I contains those situations where investment is driven by private profitability and also implies a public benefit. It is promoted by the market and there is no reason for it to be halted by the public administration

180

ENRICO MUSSO ET AL.

(although it may be regulated in order to enhance public benefit). On the opposite side, quadrant III clearly shows investment projects that appear neither profitable nor socially desirable and, therefore, should never be promoted. Quadrant II features investments deemed ‘‘socially useful’’ (external economies, accessibility, etc.) but with little or no direct profitability. If the balance is positive (i.e. above the bisector), investments should be promoted by adopting the appropriate policies, which might include grants and public–private partnerships (PPPs), capable to shift profitability (as represented by vector b) even if the offsetting costs reduces overall utility, and therefore moves the point closer to the bisector. What is an unprofitable investment for private capital may, nevertheless, be regarded as socially desirable (for example, as a driver of regional economic development). Ports have often been regarded, be it rightly or wrongly, as drivers of regional development as well as a source of considerable external benefits (on this debate see, among others, Goss, 1990a, b, c, d; Gripaios & Gripaios, 1995). Nowadays the ‘‘local’’ net external benefit is less certain, although ports are regarded – more than before – as essential gateways for the competitiveness of the hinterland. This may drive forward an investment even with no private profitability. The investment can be entirely public, or (if public resources are scarce) publicly co-financed in order to supplement private profitability and push it above the threshold that is critical for the private investor (i.e. interest rate+risk premium). Yet, the risk is to promote investments that are actually below the bisector.4 If we assume that any compensation policy shifting benefits/costs from one sector to another does have a cost, then no policy can make the point shift from below to above the bisector. Quadrant IV shows investments that are profitable for the investor, but a source of net external costs. This situation is common nowadays, and increases in port capacity required by terminal and logistics companies often gives rise to conflict and opposition at a local level, due to there being no (or very few) external benefits in comparison with external costs. An investment should nevertheless be encouraged for projects placed above the bisector, through the offsetting and reduction of external costs to ‘‘shift’’ the investment towards quadrant I, as represented by vector a (even if offsetting costs reduces the investment’s overall utility, and again it moves the point closer to the bisector). Investments where social disutility of external costs exceeds direct profitability (below the bisector) must instead be prevented by way of appropriate bans and restrictions, etc. A pure market economy would promote all – and only – investments in quadrants I and IV (where direct profitability is higher than the market

Port Investment

181

interest rate), while a centralised economy should promote all – and only – investments in quadrants I and II. If market failures are taken into account instead, the focus should be on promoting investments ranging above bisector II–IV. Indeed, area A of quadrant IV shows situations where the social effect of directly profitable investments needs to be mitigated through reductions and restrictions (even at the risk of limiting their profitability). In area B of quadrant II, private investment that would otherwise be uneconomical needs to be encouraged through incentives or through funding that is seen to stem directly from public investment. Only in quadrant I, are investments privately and socially profitable, although regulations and governance-oriented measures may be taken to amend the ‘‘profitability mix’’ (as represented, for example, by vector c). This chart leads us to consider a potential investment based on the answers to the following two questions:  Where does a port investment project tend to fit into the diagram, and upon what do direct profitability and social profitability depend? (type of cargo, technological features, etc.).  Is it appropriate (and when) to ‘‘shift’’ it, i.e. amend the profitability mix, and with what tools? This conceptual framework needs further elaboration (which will be addressed in Section 5), concerning measurability (namely of external effects), uncertainty (namely for private profitability, due to the long economic life of most capital assets) and irreversibility (which can cause a lack of market contestability). Some additional remarks must be added here. First, once a port investment (usually irreversible) has been made, the competitive environment (i.e. competition with other ports or terminals) can push the price down to something close to the opportunity cost, which has in the meantime fallen to marginal cost. Marginal cost in turn falls precisely because of the investment’s irreversibility. This triggers the risk of direct profitability falling vertically due to the market structure and the bargaining power of demand (due to concentration in the shipping market). All other things being equal, the profitability mix may then fall below the bisector representing the critical threshold. We will come back to this in Section 5. Second, the port industry has traditionally been characterised by an industrial ‘‘maturity’’ process: from the labour-intensive merchant ports before the Industrial Revolution, ports progressed to become manufacturing sites, moving vast quantities of commodities in bulk thanks to their becoming increasingly large and operating with increasingly expensive equipment. With the containerisation era, and the growing specialisation of ships

182

ENRICO MUSSO ET AL.

and terminals, a capital-intensification process made ports a typically capital-intensive industry. Ultimately, the dramatic fall in the cost of both land transport and maritime transport caused the hinterland around ports to overlap with one another, thus boosting inter-port competition. This significantly reduces the profit margins of the port industry, and accelerates its horizontal and vertical (towards the transport industry) integration. Consequently, investment projects are likely to have approached the bisector, and are therefore more likely to ‘‘fall’’ into quadrant II or IV, albeit remaining above the bisector. This means that over the decades, the need for governance in respect of port-investment decisions and processes has increased, despite the extra momentum gained by market-enhancing port policies.

3. TOWARDS A PORT INVESTMENT MODEL The port capacity installed by the investor (be it public or private) may be exploited by the investor itself, if it acts as the asset’s manager and charges the carrier for use. Alternatively, it may be leased to a stevedore, which manages it and charges the carrier. Moreover, the carrier and the terminal operator may be vertically integrated (the so-called dedicated terminals) and in some cases, the carrier may also overlap with the shipper, which may manage its own ships and sometimes its own terminal(s) as well. However, if only business functions are considered, the port investment ‘‘chain’’ involves the following players: (i) the investor investing in the port facility; (ii) the terminal operator; (iii) the carrier using the port, or its representatives; and (iv) the shipper, or its representatives. The investment’s return is determined by the stevedoring industry’s profits, which in turn influence those of the shipping industry, logistics and eventually the profits of the manufacturers/shippers and the utility of consumers. This section investigates the effects generated by the port investment, so as to highlight significant relationships between players, as well as the implications for investment decisions and for the funding of investments. It focuses on the ‘‘microeconomic’’ effects of an investment, disregarding any macroeconomic benefits to employment, earnings and their distribution, any environmental benefits–costs (both direct environmental impact and the balance between the environmental impact of maritime transport and that of alternative transport). These macroeconomic effects are rather difficult to measure, while the environmental effects are uncertain, since the development of maritime transport through port investments leads to an increase

Port Investment

183

in the environmental costs associated with a port and maritime transport, but on the other side it encourages a modal split with a more sustainable environmental impact. A port investment may be ‘‘extensive’’, if its aim is to increase productive capacity while average costs remain unchanged, or ‘‘intensive’’, if its aim is to increase productivity and reduce unitary costs. From a theoretical perspective, the notion of a purely extensive investment may be viable when, for example, a terminal operator – having to meet sharp rises in demand – decides to increase its throughput capacity by adding new infrastructures that offer the same productivity as those already in use. When, on the other hand, demand is stagnant or competition from other operators is already fierce or on the rise, a terminal operator may well plump for a purely intensive investment, aimed at increasing productivity. Actually, though, it is very likely that in the former situation the new assets would be more productive than those already in place, thanks to technological improvements that are likely to have been introduced. As a result, an increase in quantity also translates into an increase in average productivity. In the latter situation, a rise in productivity is normally achieved, thanks to the reduced time per unit of throughput, and the consequent increase in throughput per unit of time. It is therefore very realistic to assume that between these two ‘‘theoretical’’ extremes, the effects of the investment will, in practice, be distributed between an increase in quantity and a reduction in costs. In a market of perfect competition this cost reduction would turn into a correspondent reduction in price (or in generalised cost) without increase in profit. On the other hand, in a monopoly situation, or if demand is extremely inelastic, it could lead purely to a rise in profit, without any benefit being enjoyed by the user (with a reduction in price approaching or equal to zero). In any intermediate situation, the effect will be distributed, depending upon the elasticity of demand and the position of the cost curves, between an increase in profit and a reduction in price, accompanied by an increase in throughput. From a microeconomic viewpoint, then, an investment in a port asset normally causes an increase in the level of throughput (total and per unit of time) as well as an improvement in the level of service. This causes a reduction in the generalised cost Cg of the port service (equivalent to a reduction in price) and/or an increase in the profits of the stevedore.5 The decrease in generalised cost will cause throughput to increase, at a rate that will be directly correlated to the degree of competition within the port services market: the greater the competition, the greater the reduction in price and the increase in throughput; the lower the competition, the greater

184

ENRICO MUSSO ET AL.

the profits netted by the manager (unless demand is completely inelastic). The increase in throughput leads in turn to an increase in the stevedore’s profits, and usually to increasing returns to scale as well, thereby triggering a further fall in the cost of production, generalised cost, price and potentially a further increase in profits. Moreover, this reduction in generalised cost/price leads to a decrease in the generalised cost (price) of the whole transport cycle, triggering within the transport industry the same kind of effects: lower prices, higher volumes and higher profits. Again, an increasing return to scale is likely to occur, and these effects can therefore build up to become stronger. Finally, the same kind of effect will also be seen for shippers (and possibly for intermediate operators such as logistic operators, forwarders, etc.,): a lower generalised cost causes both volumes and profits to rise, with possible further increases due to economies of scale. The final decrease in prices for transported goods can eventually benefit final consumers. There is therefore a ‘‘chain’’ running from port investors, to port operators, carriers, forwarders or logistic operators, all the way through to shippers and consumers, as shown in Fig. 2. Fig. 3a hows the investment’s microeconomic impact assuming a noninfinite elasticity of demand. Since it increases capacity by reducing production costs, the marginal cost function moves from MC to MC0 , while the average cost function moves from AC to AC0 . Assuming that the terminal operator is profit-maximising, the throughput shifts from OQ to OQ0 , the fare from OB to OB0 and the average cost from OC to OC0 . The investment has caused the profit to shift from EBCK to E0 B0 C0 K0 , while the users’ surplus to shift from ABE to AB0 E0 . Profit may increase or decrease, as per the trend shown in Fig. 3b, while the user’s surplus continues to grow, providing that the investment is accompanied by a reduction in price and an increase in throughput (since area ABE continues to increase as price falls and throughput rises). Port investment will then cause an increase in the capacity offered and/or in throughput, as well as a reduction in prices and/or in costs, and/or an increase in profits. These effects will take place to a varying extent depending on the market’s structure and on the functions of demand and cost. Increasing returns to scale reducing the unitary cost of cargo handling are also likely. Reductions in prices (or generalised costs) turn into reductions in cost for downstream sectors and an increase in the ‘‘user’s surplus’’. It goes without saying that this situation implies a downward-sloping demand curve, a profit-maximising approach and some degree of long-term

Port Investment

185

Private investment

Port industry

stevedores

carriers

p Inv

C s.l.

Th GC

GC

forwarders shippers

GC

GC

Port pricing

Local economy

Private tax environment land

Employment External economies External costs

Fig. 2. Port Investment and its Microeconomic and Macroeconomic Consequences. Note: Inv ¼ investment, C ¼ cost, GC ¼ generalised cost, s.l. ¼ service level, p ¼ price, Th ¼ throughput, P ¼ profit.

stability (entry barriers). That in fact means a monopoly or quasi-monopoly situation, which in the port sector is possible, although it is not the only possibility. While a situation of perfect competition is indeed very unlikely, whenever there are several ports (or several terminals) serving the same hinterland, or when capacity can be at least established afterwards, the producer is likely to reduce prices in order to increase throughput. This is done with a view to boost its market share while reducing extra-profits (which is the potential incentive offered to possible newcomers). In this case, the price would have fallen to the average cost, with extra-profits being cancelled out and consumer’s surplus increasing to ADF. Thanks to the new investment, the price would decrease with surplus profits remaining at zero and user’s surplus rising to AD0 F0 . It should be noted that the new investment will determine the employment of more production factors due to a larger facility being operated. The envelope of short-run average cost curves will then reflect the various possible sizes of the facility and give rise to the long-run average cost (LRAC) curve (Fig. 4a), which can be matched with a long run total profits curve (Fig. 4b).

186

ENRICO MUSSO ET AL.

MC

A

A B’ C’ D’

MC’ AC

E

B C

K

F

E’ K’

AC’ F’

AR MR

(a)

q

TR

T

T ’

(b)

q

Fig. 3.

Effects of Investment.

The investor will still be interested in making a new investment as long as it generates a rate of return higher than market interest rate (plus risk premium). The path traced by the profit function should place this limit in the growing part of the long run profit curve, LRTP in Fig. 4b. From a public standpoint on the other hand, there is a growing benefit (user’s surplus ABE in Fig. 3), which theoretically speaking could represent the upper limit of a levy imposed on the port investment, or on a public financing under a PPP. This co-financing could of course enable private IRR to reach a rate considered profitable (interest rate plus risk premium). Such public interest should justify a wider capacity, and therefore an investment able to generate this greater quantity at the lowest possible cost.

Port Investment

187

(a)

(b)

Fig. 4.

Costs, Revenues and Profit over the Long Term.

It is worthwhile noting the variations of profits and external benefits further to an increase in investments. In Fig. 4b, total revenues, total costs and total profits reflect the trends of demand and average/marginal costs over the long term (Fig. 4a). The behaviour of total revenues is a

188

ENRICO MUSSO ET AL.

function of the demand curve (that is to say, of average revenues) and of marginal revenues. The total cost is a function of the investments made, with the pattern followed by total profit varying in relation to the investment made (i.e. it identifies the highest profit realisable for each facility size), and is initially upward (with progressively smaller increases being registered) and subsequently downward. Taking its cue from these considerations, Fig. 5 shows the users’ surplus curve US (i.e. the surface ABE in Fig. 3a) in relation to the various possible price/quantity combinations. Assuming a linear demand function, such as p¼a

bq

(5)

for every possible position that may be taken by E, then: US ¼ q½a

ða

bqފ=2 ¼ 1=2 bq2

(6)

This is a parabola on an ever-upward slope in the first quadrant, which may be shown as the curve US in Fig. 5. The revenue function is then TR ¼ aq

bq2

(7)

US

Fig. 5.

Users’ Surplus and Total Profit.

Port Investment

189

and the function of LRACs is straight, expressed that is to say by the function TC ¼ cq

(8)

(as assumed, e.g., in Haralambides & Veenstra, 2002; if economies of scale occur, this function would instead follow a downward slope, but the argument does not change anyway), total profit can therefore be expressed as TP ¼ TR

TC ¼ ða

cÞq

bq2

(9)

The profit-maximising quantity is given by d TP=dq ¼ 0

(10)

from which we obtain a

c

q ¼ ða

2bq ¼ 0

(11)

cÞ=2b

(12)

The quantity that maximises the sum of profit and consumer’s surplus, considering thus both direct and external profitability, is then dðTP þ USÞ=dq ¼ 0

(13)

a

(14)

from which we obtain c

2bq þ bq ¼ 0

with the optimal quantity emerging thus q ¼ ða

cÞ=b

(15)

As a result, the quantity that maximises the sum of profit and consumer surplus Eq. (15) is double the quantity that maximises profit Eq. (12). This result can now be traced back to Fig. 1 (as presented in Section 2). The reduction in costs afforded by the investment causes an increase in user’s surplus BB0 EE0 , and a shift in the operator’s profit from KEBC to K0 E0 B0 C0 . With reference to Fig. 1, three different cases may actually emerge  profit increases, as does user’s surplus (quadrant I);  K0 E0 B0 C0 is smaller than KEBC, and this decrease in profit is smaller than the rise in user’s surplus (BB0 EE0 ); in this case, the investment project features in quadrant II, above the bisector; and

190

ENRICO MUSSO ET AL.

 K0 E0 B0 C0 is smaller than KEBC, and the decrease is bigger than the rise in user’s surplus (BB0 EE0 ); the investment project features in quadrant II, below the bisector. If we do not take into account external effects of an environmental nature, the investment remains in quadrants I or II. If environmental effects were to be included, then a negative balance could result between the reduction in external costs caused by the boost of maritime transport (replacing land transport) and the increase caused by the environmental impact of port infrastructures or of their use. This could put the investment in quadrant III (neither profitable for a private investor nor socially useful) or in quadrant IV (profitable for the investor, but its social impact is negative). Above the bisector of quadrant IV, the two components would produce a positive balance, which would therefore suggest seeking a set-off to mitigate the variances triggered by the private investment. Below the bisector, the balance would be negative, indicating that the investment should be avoided.

4. INVESTMENT, PROFITABILITY, PRICING, PRIVATE AND PUBLIC FINANCING These results suggest some remarks on the financing of port investment. Port investment may produce both direct and indirect benefits. Direct benefits provide a funding channel by way of the pricing applied for the use of infrastructure, revenues and the consequent profit for the company that builds and/or manages the terminal (if two different companies are involved, the profit of the terminal operator will be used to pay the charge to the company that owns the port facility). Net public benefits justify the utilisation of fiscal resources instead. So far, as mentioned in Section 2, port investments have very often attracted public investment, due to the very features of the infrastructures and systems associated with ports. However, there has been no proper criterion in place to determine – if only theoretically – the extent to which the public taxation system should be involved in a port infrastructure. This particular issue is closely linked to the price charged for using infrastructure, for two reasons: (i) the pricing applied to, and the payment made for, the utilisation of a port asset generates a level of private profit that is complementary to the public taxation system (the greater the resources obtainable from pricing, the lower the resources required from taxation, and vice versa); and (ii) the pricing criterion itself may reflect not only the port

Port Investment

191

operators’ profit-maximising strategies (or just its market strategies), but also the purpose of maximising the welfare generated by the investment. This issue clearly reports to the ‘‘port pricing’’ debate, and specifically to whether it is better to set pricing at marginal cost or at average cost. The former criterion, by definition, maximises efficiency in resources allocation; but since marginal costs are normally lower than average costs, the investment cannot be recovered. The latter enables investments to be recovered in full, provided that there is sufficient demand willing to pay (which might not happen due to the growing competition between terminals). In the long term, all input being variable (including the infrastructure itself), there exists – for each possible level of traffic – a size that allows traffic to operate at minimum cost. This gives rise to the LRAC curve, indicating the minimum infrastructure cost for each level of traffic. Its slope depends on whether the infrastructure contains economies or diseconomies of scale. It is coupled to a function of long run marginal cost (LRMC), positioned below the LRAC while it is decreasing, in line with it when it is constant, and above the LRAC when increasing (diseconomies of scale) (Haralambides & Veenstra, 2002). Different objectives are implicit in the different criteria that may be adopted when determining the pricing of port infrastructures, mainly involving  the maximisation of net benefits (surplus), be they private or social (i.e. including externalities) stemming from the use of the infrastructure;  the complete recovery of the infrastructural investment, in keeping with the so-called ‘‘user pays’’ principle; and  the maximisation of an investment’s profitability, typical of a monopolistic port market.6 The goal of maximising the collective benefit is tackled by setting price at (private or social) marginal cost, i.e. the higher costs owing to the additional ship: maintenance works or replacement, additional congestion or pollution, etc. The infrastructure is used as long as the benefits are higher than marginal costs. ‘‘Social’’ marginal cost allows to charge the user for external costs as well, guaranteeing the highest possible overall surplus, although with a higher price and a consequent lower throughput. However, unless the infrastructure is congested, price set at marginal cost will not recover the cost of the investment. This criterion therefore best suits the public financing of port facilities. If the main objective is instead to cover investment costs (because the port is privately owned, or because the value added is incorporated into the

192

ENRICO MUSSO ET AL.

transported good whose price should therefore cover all costs, or in general when social objectives are not relevant) other pricing criteria apply. At most, a monopolist producer may set a price corresponding to the traffic for which marginal costs equal marginal revenues, thus achieving a maximum extra-profit extracted from port users, in clear contrast to the social benefit expected from the infrastructure being used. Fortunately, however, it is becoming increasingly unlikely for a port terminal to operate as a monopolist – unless it is an island port that is not a transhipment port, or unless it handles very specific categories of goods, such as liquid natural gas – and the competition from outside will force prices down. In standard situations, though, when infrastructure costs have to be fully covered, a price equal to average cost needs to be applied, whereby average costs and revenues (and total costs and revenues) are equal. Of course, this is impossible when the demand function is constantly below average costs: in this case, a public grant is always necessary. Other solutions also allow the variable price to be used ‘‘strategically’’ (strategic pricing) in relation to the targets being pursued. If the port infrastructure is a ‘‘club good’’, a two-part price may be applied: one covering the marginal cost, and one – fixed – as a ‘‘membership fee’’ to join the club of users. The latter is a fraction of the difference between cost and marginal cost: where the number of users (i.e. club members) is ‘‘n’’, it is equal to 1/n of this difference, thus enabling the cost of the investment to be recovered. This part has therefore a lower incidence on the user’s unitary costs the more frequently the infrastructure is used: the economy of density provided by frequent use is thus transferred in part to the user. This logic of club goods applies specifically to those port infrastructures with a limited number of users that are satisfying a private requirement (and social usefulness is negligible). One variant of two-part prices is provided by the so-called ‘‘Ramsey prices’’ (Ramsey, 1927), which enable the operator to charge ‘‘whatever the traffic can bear’’. To this end, the portion of price exceeding marginal cost is inversely proportional to the elasticity of demand (i.e. price is higher for an operator whose demand is more inelastic). By comparing the various pricing criteria, it may be seen that  price set at marginal cost enables the collective benefit to be maximised and resources to be used as efficiently as possible (if they are equal to social marginal costs, then they enable external costs for the use of the infrastructure to be contained as well), but does not enable the costs of the infrastructural investment to be covered;

Port Investment

193

 prices equal to average costs enable the investment to be recovered (and consequently encourage investments and help to reduce congestion), but fail in maximising the collective benefit;  two-part pricing helps the cost of the investment to be recovered and establishes a relationship with the marginal cost, while also allowing collective net benefits to be increased; in their specific environment, Ramsey prices minimise the ‘‘loss of benefits’’ in relation to the marginal cost criterion, but otherwise lead to price discrimination; and  monopoly price, wherever it can be applied, maximises the profitability of private investment, but minimises the collective benefit (and facilitates price discrimination). The pricing of infrastructures – and, symmetrically, the funding of investments – rank among the most significant differences among port models. While some heavily use public funding, others (such as private ports that are not granted public funding) need to recover infrastructure costs in full. Intermediate models (such as ‘‘landlord ports’’, where the public authority acts as both the developer and owner of infrastructures and space) adopt solutions that combine public funding and partial funding from the market. These differences are crucial in the way they influence the competition between ports and, as a result, the competition between logistics chains, firms and local economies. This clearly involves complex transition problems: for example, shifting from publicly funded ports to a cost-coverage logic leads to the risk that imbalances created by the previous public investments are consolidated. In this case, the privatisation of public infrastructures at a price lower than the market price is tantamount to a public subsidy to the private company.

5. INVESTMENTS, MARKET FORMS AND BUSINESS STRATEGIES Two attributes of the investments that characterise – even if not exclusively – companies operating within ports are the degree to which investments are reversible and uncertainty, which is typical of every decision that has anything to do with the future. The first of these two attributes would appear to be of considerable importance in our case, since a growing number of private firms or PPP are being asked to invest not only in port superstructures, but also in actual

194

ENRICO MUSSO ET AL.

infrastructures (with regard to the growing use of PPP to fund transport infrastructures, see Bonnafous & Jensen, 2005). Suffice it to consider, for example, the many dedicated terminals – typical of the container transport sector, and similarly the cruise transport sector – in which the transport company is vertically integrated to become a terminal company as well, thereby participating in the cost of the terminal investment proportionally to its share in the venture. In these situations, it is clear how at least part of the investment should be regarded as irreversible, making it interesting therefore to ascertain how this circumstance, together with the uncertainty as to how the operating environment will evolve, may cause the company to accumulate surplus, or insufficient, capital. According to Abel and Eberly (1999) irreversibility and uncertainty of investments involve two types of effect  the so-called ‘‘user-cost’’ effect, which leads firms to under-invest. This is because entrepreneurs are more reluctant to invest, given that their inability to disinvest results in a higher user-cost of capital in relation to current investment decisions; and  the so-called ‘‘hangover’’ effect, indicating the reliance of current capital stock on past behaviour, which leads firms to over-invest in the presence of irreversibility and uncertainty. These two effects lead to contrasting consequences, so ‘‘which effect becomes relatively stronger depends on the characteristics of the firm and its environment’’ (Abel & Eberly, 1999). Chirinko and Schaller (2001) found that ‘‘it is firms with very low investment that will be affected most by the possibility of bumping up against the irreversibility constraint’’. Moreover, all these authors are keen to point out that features peculiar to the industry, market and environment in which firms operate may influence investment decisions differently. The features of the port industry are therefore crucial when it comes to companies investing in it. The port industry may be characterised as an oligopoly: an oligopoly, however, that features high competition together with a trend towards concentration, at least in some port functions such as those related to containers. Concentration implies greater technological lumpiness and discontinuity in technological substitution rates for production factors. A number of large terminal groups (suffice it to consider the major international stevedoring groups in the container transport sector and monitored by Drewry in its annual reports) operating alongside a sizeable gathering of small/medium-sized firms at final destination ports leads the

Port Investment

195

market to take on a structure reminiscent of Stackelberg oligopoly (1934), where large groups lead the market and where other terminal operators act as their followers. Plumping for a Stackelberg-type oligopoly does not lie in the negligible role played by price in the choice that a ship makes between one terminal and another as much as in the critical availability at those terminals of the space and equipment needed to serve the ship (Ferrari & Benacchio, 2003). Indeed, the ship tends to require an oversized port terminal (Musso, Ferrari, & Benacchio, 1999), since the greater a port terminal’s unutilised capacity, the lower the chances of it being unable to operate the ship as soon as it docks at the terminal. In other words, the ship’s costs tend to grow at a considerably faster rate compared with the time spent at the port, thus forcing shipping companies to look for oversized infrastructures. Furthermore, it may be demonstrated that price-oriented competition leads firms to collude with one another, while quantity-oriented competition fuels competition among firms. This brings the interpretative model very close to a port business founded upon several regional markets accommodating a number of distinct companies (‘‘distinct’’ in view of the variety of different services offered, as well as their geographical location), where large port companies operate alongside smaller operators and competition is at times especially fierce – even when the number of operators tends to decrease due to merger and acquisition processes.

5.1. Stackelberg Equilibrium It is worthwhile remembering that the Stackelberg duopoly considers two firms – known as ‘‘L’’ and ‘‘F’’ (leader and follower) – that need to decide (not at the same time) how much capital to employ. The function of profits for these two firms may be expressed as PL ðK L ; K F Þ ¼ K L ð1 F

P ðK L ; K F Þ ¼ K F ð1

KL

KF Þ

KL

KF Þ

(16)

This situation leads firm L (the firm to decide first, since it is the first to introduce a new technology or to enter a particular market) to select the amount of capital in such a way as to maximise its own profit function, while taking into account the reaction curve of firm F. This means that where K F ¼ RF ðK L Þ ¼

1

KL 2

(17)

196

ENRICO MUSSO ET AL.

the equilibrium – which is different from the Cournot equilibrium, based on companies making their choices at the same time – makes the levels of capital employed equal to K L ¼ 1=2 K F ¼ 1=4

(18)

Profits and the ratio between them therefore emerge as PL ¼ 18 1 PF ¼ 16 PL PF

(19)

¼2

The firm investing first therefore accumulates twice as much capital as the other, while also netting a profit that is double that realised by the follower. If at a later stage a rise in demand is expected, the model shows how both firms will increase their productive capacity to a similar extent, so that the ratio between the respective profits of the two firms remains unchanged. We are thus witnessing a game that repeats itself by the same procedures. In other words, this duopoly leads the firms to cover three-quarters of the amount that would be exchanged in a market of perfect competition, leaving unchanged their respective market share as leader and follower, with one twice that of the other. Compared with monopolistic equilibrium, this model guarantees that a higher amount is exchanged (50% more in fact). This gives rise to the capacity surplus seen in these types of oligopoly. If a fixed market-entry cost and/or the cost of indivisibilities – a typical feature of investments made by port companies – is inserted into the Stackelberg–Spence–Dixit model and dynamic equilibriums are analysed (see Tirole, 1988), then different conclusions are reached. Optimal behaviour on the part of the incumbent would be, therefore, to hinder the entrance of new operators by making the most of its capacity, that is to say investments and the accumulation of capital. 5.2. Application to Port Companies As already mentioned, this form of oligopoly leaves companies with overproductive capacity, which may be interpreted as a kind of entry barrier and also has the effect of lowering the rates of return of other operators’ investment projects (actual and potential).

Port Investment

197

An increase of operating capacity by leaders therefore serves to regulate the level of profits and the number of new market entrants. For instance, Drewry Shipping Consultants Ltd (2005) forecasts an average annual rate of increase (until 2010) in port capacity of 5.7% for global terminal operators, while the other private operators will face a rate of increase of 4.8%. Moreover, in the same report, Drewry states: ‘‘given the high entry cost (y) there is limited scope for new entrants to break into the global terminal operators club’’. Indeed, in the presence of a production function with increasing returns, an increase in productive capacity reduces marginal cost, which will in turn reduce the level of productive capacity offered and the profits achieved by the rival firm, going as far as to render the new operator’s entry to the market completely useless (since profits would be zero). However, major terminal companies will not be inclined to expand their investments until they completely eliminate the ability of new operators to enter the market, since it is also in their interest to maintain a sufficiently large number of nodes (ports) that the network economies of the maritime transport sector may be boosted as higher profits are concentrated within large terminals. If we abandon the notion of just the two firms (i.e. leader and follower) in order to move closer to business reality, the investment made by one of the incumbents – which is needed to limit new entrants – takes on the features of a ‘‘public good’’. Indeed, other operators that already have a market presence will also reap the benefit of this behaviour, even without bearing any of the costs. Business literature that deals with this very issue (summarised in Tirole, 1988) seems to lead to a conclusion for a level of investments by incumbents that is lower than that sought, on the back of the far from co-operative behaviour of the businesses involved. Free-riding approaches are thus witnessed. However, in the case of ports, the opposite solution seems to be the most likely. This firm belief emerges from the fact that, to be more precise, the surplus capacity needed to limit the entry of new operators stems, above all, from port infrastructures – in the majority of cases funded in full or in part by a public body – being equipped with bigger and better facilities. From a theoretical perspective, similar conclusions are in fact reached by Gilbert and Vives (1986) (assuming that the profit realised by operators is proportional to the productive capacity of their systems), meaning that these operators will be inclined to make as great a contribution as possible in putting this operating capacity in place in order to limit new entries. Applying these conclusions to the case of port terminals, while also bearing in mind what was said earlier regarding the charge for infrastructures being

198

ENRICO MUSSO ET AL.

borne by the public body and the increasing role played by PPPs in the funding of transport infrastructures, the tendency to create oversized ports would seem to be well founded. In other words, we might say that terminal companies seek to boost their own market power by making the most of investments and by passing a part of the financial charge involved on to the collectivity (see Musso et al., 1999). To support the notion of there being a tendency to equip ports with an infrastructural overcapacity (for the most part funded in full or in part by a public body), reasons need to be sought as to why the public body agrees to go along with this, i.e. makes investments that (as a general tendency) are higher than the level required from time to time. An initial answer lies in the difficulty innate in every investment process when defining the optimal amount of new capacity that needs to be channelled into the productive process: so much so that this capacity cannot be defined through a continuous function, due to the indivisibilities characterising all transport infrastructures. This difficulty, which is typical of every investment, becomes yet more marked in the case of service companies whose output, if once realised and made available finds no buyer, goes to waste. Another issue concerns the difficulty encountered when trying to express a definitive opinion about the quality of the investment effected. Indeed, the long useful life of a port investment – above all in the case of port infrastructures, marine projects and the construction of dams, spurs, breakwaters, etc. – makes the infrastructure’s start-up period (when the plant is not yet fully up and running) an extremely long one. This in turn makes it difficult to judge whether the original investment actually exceeded demand. So much so that to date, the demand for transport has always been on the rise, even though it may have experienced periods of accelerated growth and periods of less notable progress, such as to render investments geared to boost port capacity only rarely useless. Similar to this is the ‘‘hiding hand’’ effect referred to by Hirschman, which plays a part in several infrastructural projects, not exclusively in developing economies.7 In other words, we are dealing with investors underestimating all the difficulties (and/or risks) inherent in the initiative and, above all, the possible solutions to the difficulties that the venture will encounter during its lifetime. While difficulties and risks might be foreseen and may lead to the project being rejected, the possible solutions to them – including new uses for the infrastructures realised – cannot be evaluated ex-ante, since they are the fruit of the inventiveness put into play by the need to overcome an obstacle that has been subsequently created. To end this section, we should not overlook the positive external economies stemming from port activities either: here we are referring to the

Port Investment

199

effects in terms of employment, the creation of value added, the reduction in internal prices, which port activities are able to generate across the region in which they are located. Effects that, as seen previously, spread across an area that very often passes over the boundaries of maritime/port regions and similarly those of the same countries, but that undoubtedly for the most part have an impact within the boundaries of the country summoned through public action to contribute to the investment. Owing to the development and innovations experienced by the land transport sector, the competition between the various areas is becoming increasingly fierce, meaning that leadership is no longer being fought for in a play-off between the port operators and the shipping companies but also includes those public bodies responsible for looking after the port business. Indeed, each one is only too aware that declining an investment or delaying the approval process that it needs to go through or deferring its realisation may cause some or all traffic to choose another port – possibly in a neighbouring country – consequently causing all positive external effects to be lost. Examples of the inclination of public bodies to invest in a port business competing with neighbouring countries have been witnessed several times over the last two decades. Suffice it to consider the competition that initially emerged between Japanese and Korean ports and the way in which Chinese ports subsequently joined in, or the fierce competition seen in Europe between ports, municipalities and the states hosting the ports of the Northern Range. The entire collection of these reasons – to which in certain cases we might also add the inefficiency that often characterises the public body involved – appears more than enough to explain how the companies operating terminals can make the most of those (partly or completely) public investments that exceed the levels actually necessary.

6. INVESTMENTS AND COMPETITION BETWEEN PORTS 6.1. A Double-Sided Relationship A dual-sided link of interaction exists between the port investment function and the competitive dynamics between ports  on the one hand, investment decisions, the level of capital invested – but also the sharing of investment costs between public capital and private

200

ENRICO MUSSO ET AL.

capital – significantly influence the paradigms of competition between competing ports (and port areas). So, competitive dynamics may be regarded as a function of port investments; and  on the other, the degree of intensity and the means by which the competition between ports is expressed actually influence the amount and type of port investments made, and sometimes end up influencing the public or private nature of the capital financing projects as well. So, port investments may be regarded as a function of the competitive environment. An example based on the opposite situation may actually simplify the way in which these two relationships are presented. In the absence of competition – e.g. a situation where there is just one port on an island served exclusively by the island’s own basin – a port investment will only impact the individual port’s ability to serve outgoing and incoming traffic (for example, the number and type of ships) and/or the efficiency of the port service’s output in terms of costs and time. As an example, any failure to realise an investment facilitating the dredging of access waters will limit access to ships with a bigger draught (or will influence the way in which the maritime service is organised by imposing transhipments on smaller vessels). On the other hand, the realisation of a new terminal will enable demand that would otherwise remain unsatisfied to be satisfied, or help improving the productivity of port operations. To summarise, investment decisions, influenced exclusively by evaluations that analyse the appropriateness of investments in relation to the business system at which they are targeted, will affect solely an individual port’s production function (in terms of output and the cost of the various factors involved). Pricing and taxation mechanisms will thus determine the way in which the cost of an investment (or a failed investment) is shared between the public sector and the private sector, between port users and contributors. Indeed, in the specific case put forward, by definition there are no actual or potential interactions with other ports in terms of competition for the same hinterland. In case of competition, the picture becomes a more complex one. The amount and actual purpose of investments channelled into a port (new terminals, surrounding areas, land transport infrastructures) indeed influences the latter’s competitive advantage over other ports competing for a certain type of traffic (be it geared to a market area being vied for or made up of transhipment traffic, which due to its very nature is not tied to the hinterland from where the goods originate and the hinterland to which they are headed). The ratios of strength between some ports are therefore amended, with a share of traffic consequently won or lost. The investment

Port Investment

201

decision is not therefore a neutral one for other ports: so much so that underlying investment decisions are not only the evaluations carried out in respect of an investment’s appropriateness and the profitability it is expected to generate in relation to the incremental demand forecasts for traffic and/or a political willingness to achieve the port’s beneficial externalities, but also strategic assessments regarding the port’s position in relation to rival ports (investments aimed at taking away traffic or limiting the expansion of other ports).

6.2. Changes in the Competitive Environment and the Repercussions on Investments The technological and organisational advances enjoyed by the transport system over the last 30 years have brought the traditional paradigm of ports up for discussion. The said paradigm, founded upon a port development model, was based on (i) market areas that tend to be protected (limiting the freedom of circulation afforded to goods through customs barriers, the protection of domestic borders and the lack of adequate land transport infrastructures) and (ii) the role of the port as a tool of regional policy aimed at pursuing economic impact and growth objectives set for the local and domestic system. Against this backdrop, it is easy to understand why the problems associated with recovering the cost of a port investment and with splitting this cost between the public sector and the private sector were not top priorities. Especially given that the ports were mainly being financed through the tax system by virtue of the beneficial externalities offered to the contributor–consumer. Generally speaking, improving accessibility to the sea and, above all, to the land surrounding the ports of the most economically developed regions has led to a gradual reduction of captive, incontestable market areas, thus causing a gradual overlap among hinterlands – including on an international scale – that may be served by different ports depending on their impact on the generalised cost of transport from the place of origin to the final destination.8 The improved efficiency of maritime transport, achieved with the unitization of cargo, the specialisation of ships, and the consequent ability to achieve economies of scale has further widened the choice available to carriers, thereby reducing a cargo’s loyalty to a particular port and helping to erode the traditional rent accruing to ports. The actual location of a port does indeed tend to become less important than its ability to offer services and connections that meet the needs of carriers and shippers. In

202

ENRICO MUSSO ET AL.

a global economy, ports have increasingly become critical transit ‘‘nodes’’ that come under increasing pressure from the various players active within the logistics chain (shipping companies first and foremost) and, at the same time, as elements that become less and less indispensable and easily replaceable on an individual basis within the various international transport networks. The changes outlined end up fuelling competition between ports, for which investment decisions become increasingly crucial elements – not only for development purposes, but often for the survival of ports in an increasingly fierce competitive environment as well (Ferrari & Benacchio, 2003).9 Heavy investments (almost entirely public) have enabled many ports to build up a strong market position and have made it rather easy for them to advocate the need for further market-driven investments. These two effects act as a limit pricing policy aimed at raising the rival’s cost and deterring the market entry of competitors in other competing ports (see previous paragraph). The evolution of maritime transport (in the sense that ships have gotten larger and in terms of the co-operation projects and mergers in container transport) is also influencing the size of port terminals, which are tending to increase their scale. The economies of scale attainable by a terminal/ port thanks to a new investment often depend on the size of a terminal when it starts up, as well as on the traffic generated: a mechanism for which the competitive gap between the larger terminals/ports and smaller operations (or new entrants) increases as the accumulation of capital is also heightened. The emergence of large operators (mega-alliances), a large portion of whose traffic is concentrated on specific trades, at the same time increases (i) the need for a single port to invest above the optimal level needed to improve the quality of the service delivered (for example, in terms of turnaround time for ships) and to attract/retain these ‘‘essential’’ clients and (ii) the risk of productive capacity being excessive and remaining unused should the ships decide to switch to rival ports thus shifting sizeable amounts of traffic (this being the so-called ‘‘volatility’’ of shipping companies). Significant imbalances in the ratios of strength between these operators may influence the investment decisions made at individual ports (be they of a public or private nature) and have an impact on the oversizing phenomenon seen at ports. The increasingly widespread phenomenon of terminals that are partly or entirely dedicated to serving one or more shipping companies may be interpreted as an important form of investment that seeks to maintain a balance between the contrasting needs of the two operators involved in

Port Investment

203

the exchange of goods, the ship and the port. The dedicated infrastructure reflects the port’s commitment towards premium clients (in addition to the cost of realising or quite simply laying out and organising the new terminal, it represents the opportunity cost of limiting the space assigned to common user facilities that may be used by a greater number of carriers), whose risk is guaranteed by the involvement of shipping companies in the realisation and/or management of the dedicated infrastructure, and therefore by an implicit reduction in their mobility as regards selecting particular hub ports (Haralambides, Cariou, & Benacchio, 2002). As regards the port players themselves, the drive towards beefing up investments in response to the more competitive dynamics triggered between ports (made even fiercer by the tendency to cooperate in the container transport sector, while traffic at the same time is concentrated around an increasingly small number of operators) is clearly leading to the convergence of the interests of private operators (terminal companies) and the port authorities, united in their effort to compete with other ports in order to pursue their respective profit targets and make their economic impact felt across the area in which they operate. If by increasing a terminal’s traffic capacity, it also increases its projected profitability, any proposed investment will indeed be in keeping with the profit maximisation objective of the port operator, which will promote its realisation. Similarly, positive effects (in terms of increased taxes, port duties, lease rents, direct and subsequent employment and the development of port-related activities) will justify – where greater than the investment’s actual costs – the interest of the port authority, whether it is directly involved or not in producing and selling the service. This sharing of interests does not always lead to the cost of the investment itself consequently being shared between the public operator and the private operator, though, since – as already predicted – the features of the port investment are traditionally founded on public intervention. This is despite the increasingly widespread solutions of Public–Private Partnerships (at least as far as those projects where higher levels of profitability are forecast are concerned), which are increasing the private sector’s involvement in the funding and development of the port operations involved (Estache & Serebrisky, 2004). In keeping with the aim of encouraging private involvement in infrastructural investment enterprises is the development, in the case of terminals realised from scratch, of BOT (build, operate and transfer) contracts, which get the market to participate in the realisation of the infrastructure through the terminal company, thereby implicitly solving the problem of determining the franchise charge to be applied.

204

ENRICO MUSSO ET AL.

6.3. Public and Private Funding and the Repercussions on Competition between Ports Since investments take a far from neutral stance in developments when it comes to the competition arising between the different ports, it is important to understand that the issue of financing might change competition-driven relationships between ports. Indeed, public and private capital intervene at different percentages and take on different forms, depending on the ports and countries (in the sense of legal/institutional systems) in which they are located. In an environment where there is competition between ports on an international scale, the varied nature of both rules and situations may distort the competitive game being played. By meeting criteria that are not market-driven and by not considering profitability as a benchmark of top priority when selecting an investment project, financial intervention from a public body (be it a port authority or a local or central government body), does indeed change the competition witnessed between ports in those areas where its presence is more or less significant. The market may indeed see those port operators that have to reward the capital invested in infrastructures entirely with the sales revenues generated from terminal services come face to face with other operators, which by utilising public funds to finance their infrastructures, only need to cover the production cost incurred for the service and the investments in superstructures. Taking it to the limit, it will not be worth it for a port, up against other ports obtaining public funding for free, to borrow from the capital markets, even with the prospect of its investment being profitable. This is due to the competitive disadvantage represented by the cost of the loan taken out or the commercial unfeasibility of its passing on the cost in the selling price of the service provided (the so-called ‘‘crowding-out’’ effect of private capital). Furthermore, inverting the logic underlying the decision-making process that oversees investment decisions, where private and public funded ports are competing, increases the risk of the investment’s size being far from optimal and the wrong projects being selected. Indeed, in the case of a private investment, realising an investment is subject to being able to find a form of borrowing whose interest rate is lower than the rate of projected profitability. In the case of a public investment, funding (above all where granted by a Central Government body) becomes the determining prerequisite for the investment (see Section 7). The availability of public loans may put the focus on growth-oriented choices that favour the consolidation of activities, even if such choices may also lead to the resources invested being inefficiently allocated.

Port Investment

205

This distorting element is magnified by the evidence that at some ports (in Northern Europe, for example) terminal companies are actually state-owned and more or less formally separated by the government-controlled port authority. As a result, decision-making processes in respect of the running and management of port operations therefore end up mixed with little transparency.

6.4. Distorted Competition and Practical Measures – Charging Framework vs. State Aid Law: European Perspective The issue regarding the way in which competition becomes distorted, due to funding situations varying widely from one port to another is one of the hottest topics within the field of European port management policy. This may be considered a prime example of a government overseeing individual ports and individual countries attempting to achieve some kind of regulation, albeit minimal. Two important papers take pride of place within this argument: the Green Paper on ‘‘Sea Ports and Maritime Infrastructure’’ (European Commission, 1997) and the White Paper for ‘‘Fair Payment for Infrastructure Use’’ (European Commission, 1998). Both papers focus on passing infrastructure costs onto users (the so-called ‘‘user pays’’ principle) as a means of encouraging investment projects to be selected on the basis of their profitability and thus encouraging only those investments that can be remunerated by the market to be realised (Haralambides, Verbeke, Musso, & Benacchio, 2001). It should be pointed out that the across-the-board adoption of this stance, from a specific moment in time and for all operations indiscriminately (regardless of their nature, size, etc.) – as well as presenting implementation problems of a technical nature (concerning the determination of the charging framework for correct and efficient infrastructure use) and raising doubts as to whether or not the cost of the infrastructure should actually be included in the charge to use it – could end up creating additional imbalances, leaving the infrastructural gap wide open. There is indeed the risk that those ports that have based their development on public funding end up building up stronger market positions over the less developed ports that are still growing and would be prevented from resorting to public capital as a means of funding. Moreover, where the port infrastructure has been developed with public resources above a determined level, any private investment subsequently made requires lower amounts and carries less risk than that inherent in the project’s overall realisation.

206

ENRICO MUSSO ET AL.

In the presence of the possible risks of failure thus intimated for a form of intervention aimed at harmonising the game’s rules, what other tools may be used to protect the public good represented by undistorted port competition, waged properly on a ‘‘level playing field’’, so to speak, by the individual ports involved, notwithstanding the different conditions from which they all start competing and despite the rules not being standardised nationwide? Applying legislation regarding state aid provided to ports – which consider enterprises as ‘‘any entity engaged in economic activities of a commercial value, notwithstanding its public or private nature’’ (including therefore those port authorities that are active in the terminal services market) may prove to be a useful tool when it comes to highlighting and correcting the distortions caused by an unequal use of public funds to support investments.10 Having a proper scheme in place for the recovery of invested capital would not even be necessary in a system that effectively controls state aid, since the fees charged to use the infrastructure would be determined by market investment principles (and by taking into account the actual cost of the infrastructure concerned). Indeed the notion of state aid that is unlawful in the eyes of European legislation involves the following criteria being met: (i) public resources must be transferred, (ii) an economic advantage is to be provided by the transfer, (iii) selectivity is to be displayed in allocating aid to one or more enterprises or areas (which makes a distinction between state aid and the ‘‘general measures’’ applied indiscriminately to all enterprises or the entire country) and (iv) effects on competition and trade between EU member states (European Commission, 2002). A great deal of ambiguity is not normally encountered when verifying that requirements (i) and (iv) are duly fulfilled, with such procedures producing an analysis of a technical/accounting nature and an assessment of the level of competition between ports. Implicitly, public financing that is repaid in full through the charges applied to use the infrastructure should not be considered ‘‘state aid’’, since it would not affect competition between ports, as the port receiving support would be competing with higher prices. As for the other requirements, standing out among the various principles established within several cases of Community law regarding the ‘‘economic advantage’’ transferred is the complex differentiation between  the public funding of port infrastructures, which is ‘‘necessary for, and directly related to, the undertaking of public authority functions’’ and does not usually involve state aid;

Port Investment

207

 the public funding of port infrastructures, which is ‘‘required for the provision of port services, which always confer an economic advantage to the service provider as beneficiary’’ (which may even be the body managing the port). Drawing a line between public and private functions as seen in the market may indeed prove to be not that straightforward a task, given the somewhat confusing assortment of interests held by just the one player (at the same time both landlord and private operator) and the transparency often lacking in the assignment of responsibilities, as well as in accounting processes. In this regard, criteria for a split accounting system or for a corporate distinction between the body managing the port and the terminal company may prove to be completely ineffective in the presence of a ‘‘joint economic interest’’ justifying a ‘‘legal, administrative and structural relation’’ between the companies concerned. This issue is quite typical of liberalised sectors with a strong infrastructural component (e.g. railways, airports), where there is no corporate separation between the infrastructure manager (public or private monopolist) and the rival providing the service. The nodal point concerning ‘‘selectivity’’ would instead appear to be what distinguishes the funding of ‘‘public (general) port infrastructures, open to all users’’ from the funding of ‘‘user-specific port infrastructures’’, representative per se of an advantage attributed solely to one or more enterprises. In general, ‘‘public (general) port infrastructures’’ include maritime access and maintenance (such as dykes, breakwaters, locks, navigable channels, dredging, navigation aids, lights and buoys), public land transport facilities within the port area, and short connecting links to the national transport networks, etc. ‘‘User-specific infrastructures’’ instead include ‘‘yards, jetties, pipes and cables for utilities on the terminal sites, filling of harbour basins, rough levelling, and, in general, infrastructures that are not open to all users on a non-discriminatory basis, but are dedicated to one or more specific users as roads, tracks etc’’. Fundamental differences remain, however, as regards the way in which different EU member states define some port infrastructures, such differences being enough to draw a question mark over the possibility of having a level playing field between competing ports. For example, investments in docks and quay walls for certain aspects – and quite rightly so – fall into the category of ‘‘access and defence projects’’, but may also be dedicated to specific categories of users. For instance, the dredging of an access channel with a depth of up to 15 m seemingly constitutes a ‘‘public investment open to all users’’, even though it is difficult to believe that such a depth is

208

ENRICO MUSSO ET AL.

indispensable for all vessels arriving at the port other than those with a deeper draught. If this channel provides access to just the one terminal, then it will benefit the terminal’s franchisee first and foremost. It is therefore clear how the ‘‘open to all users’’ notion, which is discriminating for the purpose of evaluating the lawfulness of a financing in a fair competition environment, still needs looking into in greater depth from a legal standpoint, in the absence of any guidelines laying down the evaluation criteria to be adopted. This need would appear to have been acknowledged, since the new proposal for a ‘‘Directive of the European Parliament and of the council in market access to port services’’ presented by the Commission (COM(2004) 654 final) and in the process of being evaluated by the European Parliament. This requires common guidelines for funding granted to ports by member states to indicate what funding to ports is compatible with the internal market, with said guidelines to be drawn up no later than one year from the date on which the Directive comes into force.

7. CONSIDERATIONS REGARDING THE FUNDING OF PORT INVESTMENTS The funding of port investment, along with the organisation models adopted for port planning, plays a crucial role when it comes to devising the institutional and management models used for ports and is the ultimate essence of port management policy. The joint presence of public and private funding implies a combination of fiscal charges (port charges, maritime duties, mooring fees and every form of public taxation devised to compensate for the use of publicly funded infrastructure) and fees payable to the terminal operator, which remunerate the production of the terminal’s services (including the return on investment realised by the company). Once more, we find the two roles blurring into one another: a public function and saleable services. On the one hand, the notion of a port as a merit good, if not a public good in the strictest sense,11 justifies taxation. On the other hand, the notion of a port as a system of businesses in a highly competitive environment means that it is absolutely vital that resources are efficiently allocated. In the selection and planning of investment the decision-making process must be as close as possible to the market, and it is crucial that equal and fair conditions are maintained between operators.

Port Investment

209

As mentioned in Section 2, the line drawn between these two functions changes, depending on the country, environment, business, social and political cultures and trends. In most port models at least major infrastructures providing access to the port and used for general purposes are publicly funded, and at least terminal superstructures are privately funded, while the special-purpose infrastructures used within a single terminal may be the focus of either side. However, a general interest may, in fact, be identified at best for general projects: sea access, dredging, dams, spurs and breakwaters, safety, etc. Terminal infrastructure, as well as the production of terminal services and most other port services, are saleable goods and services, in respect of which private bodies may operate in a market environment (although a public control over charges raised by private operators is justified if geographical situations rule out competition). Therefore, public intervention seems rather the fruit of political willingness to boost the port’s external benefits (accessibility and the so-called ‘‘catalytic impact’’) or the consequent Leontievian benefits (supply of goods/services) and Keynesian multiplier benefits. These political needs can be more or less defendable: among the former, the social need to protect employment no longer justified by the market; among the latter, the attempt to extend or maintain political control over the economy. In port investment funding we therefore frequently witness hybrid solutions, whereby public and private capital intervene at different percentages and take on different forms. This situation crucially influences not only the funding, but the very selection, of investment projects. Public funding meets not so much market criteria but rather political evaluation of the port’s development, the accessibility of the hinterland and the growth of direct and induced employment, while private funding meets criteria for the return on investment and the rewarding of risk. Public funding has actually three relevant effects.  It releases investments from having their profitability evaluated, thereby introducing an element of potential inefficiency in investment selection and resource allocation.  It favours a ‘‘crowding-out’’ of private capital, which may no longer be used for profitable projects since its cost would imply a competitive disadvantage for the port. The crowding-out of private capital arises because terminal operators that have to reward the capital invested (including that invested in infrastructures) with the revenues generated from terminal services face other operators (be they public or private) that only need to cover the marginal cost of producing the service and of other related

210

ENRICO MUSSO ET AL.

investments. A port whose rivals obtain loans for which they do not have to pay anything for, will not be able to borrow from the capital markets even if the investment is potentially profitable.  It troubles competition between ports, and in theory between maritime transport and other modes – which are usually, however, funded much more than maritime transport is. This obviously raises the complex issue of equalisation among the levels of subsidy applied to the different transport modes. The current imbalance in favour of road transport (a particularly serious problem in the European Union) is due not only to its higher external costs but also to the greater level of explicit or implicit subsidisation (on the lack of equalisation between maritime and land transport, affecting the development of shortsea shipping, see Musso & Marchese, 2002). Moreover, the effect of public funding is different if it is obtained ‘‘locally’’ (regional or council government) or at a central government level. A centralised taxation may offer some advantages:  it allows a focus on well balanced and harmonious growth development of the country’s ports and regions.  it allows the concentration of significant resources on top-priority projects.  it enables a better co-ordination and optimal dimensions of port facilities.  it fosters optimal utilisation of the whole port network (by creating, for example, a network of terminals dedicated to shortsea shipping, thus encouraging a more balanced modal split, etc.).  it helps to better evaluate the investment projects on a national basis (whatever the criterion of preference adopted). On the other hand, the risks of centralisation of port fiscal policies are also crucial. First, there is a kind of ‘‘logical inversion’’ of the decision making: state funding is somehow ‘‘free’’ (since its fiscal cost has in any case already been paid) and its being obtained is therefore in any case a benefit, whatever the project it is used for. It therefore becomes a benefit and not a cost, and the effect – paradoxical as it might seem – is heightened by the fact that, by obtaining it, it is removed from other ports that may be direct competitors. The investment selection process may then become completely unreliable. The aim is the obtaining of funding per se, since it can then be used without any control by neither the market (since no cost needs to be paid to a lender) nor the public lender (since the funding has actually been approved

Port Investment

211

in advance). This lowers the responsibility of port authorities/bodies in charge of the investment. Furthermore, the selection criteria used by central governments are often far from objective if not highly discretionary (for example, in selecting projects with the greatest macroeconomic effects expected). They may lack significance (such as those based on historical traffic flows of each port, or on the amount of funding provided in previous years), or lack transparency (the higher or lower electoral clout of each port city, or their being represented to a greater or lower extent in government or parliament). These criteria may even be met by logrolling among members of the government or political parties. A decentralised public funding system clearly bears symmetrical advantages and disadvantages. However, having the necessary management and financial leverage fully available at a local level usually makes it easier to cooperate with private operators and to interact with them more promptly as well. It also enhances the responsibleness of local policy-makers as they decide how the resources made available to them should be best employed. Over the last two or three decades, the general tendency has been towards a greater role of the market, and the expansion of private funding, implying a complete remuneration of port investments. This trend has set in as the result of  the widespread restrictions of public finance in several countries (e.g. Eurozone countries are required to comply with the ‘‘Stability and Growth Pact’’ included in the Maastricht Treaty);  the growing cost of infrastructure projects, due above all to ships’ increasing size;  the growing cost of superstructures, due to the levels of specialisation and automation attained by goods handling and processing techniques;  increased competition and traffic volatility, as well as the faster evolution (and therefore obsolescence) of goods handling and processing technologies, in turn boosting the financial importance of superstructures over infrastructures and leading to far faster depreciation.12 Once this trend has become established, investments have been selected more for their profitability, and therefore by considering the (microeconomic) efficiency, rather than by evaluating their macroeconomic or social desirability. This helped boost productivity, with the most efficient investments being selected and levels of traffic being set in keeping with capital remuneration targets.

212

ENRICO MUSSO ET AL.

Similarly, the general move – within the world of public finance – from a central taxation system towards a decentralised one has largely similar effects on the market, above all when the institutional level around which the taxation system is based (revenues from port taxes and decisions regarding infrastructural investments) is that of the port authority itself. Indeed, where this is the case, traffic revenues must cover the entire cost of infrastructural investments, thus requiring a level of profitability similar to that required by private capital. Although no clear evidence exists on the matter, in several countries the tax yield ploughed into port investment probably exceeds the specific tax yield generated by the ports themselves. This would mean that the increasing competition in the transport and logistics industry, and as a result in the seaports, has ended up passing on some of the infrastructural costs to the general taxation system. The transport industry thus externalises some of its own costs, just as it also does with environmental costs. Against this backdrop, the complete market placement of port finance – providing that it is across-the-board, and not limited to certain countries – would pursue the additional benefit of allocating to the transport system all costs (including the infrastructural investment) relevant to it, without any potential inefficiency due to an excessive growth of international trade and maritime transport (which might have become excessive precisely owing to the fact that several costs are being externalised). What this trend actually implies in political approaches is clear, though: ports are no longer seen mainly as catalysers or facilitators of exchanges, factors of hinterland accessibility, capable of attracting businesses, and induce economic development. Rather, they are entrepreneurial entities that should be managed entirely by private operators. Their cost must therefore shift from taxpayers to users, and the ports themselves must become completely responsible for selecting investments, which must be recovered in full through the market. Indeed, the prevalence of the entrepreneurial and commercial role appears more justified in developed countries where a high level of accessibility normally exists and from which additional benefits can stem, rather than from the heightened competition allowed by the port. Although this view can hardly apply to all other countries, given that in a number of them the inadequacy of port infrastructure significantly hampers economic development and/or causes monopolies, so that mere privatisation would just worsen things. The European Union also focuses on passing on infrastructural costs to users (the ‘‘user pays’’ principle), in order to select investment projects based

Port Investment

213

on their profitability and therefore only those remunerated by the market. This opens up a road of opportunity to private equity rather than public capital for the financing of infrastructures, as stated in the European Commission’s Green Paper on Seaports and maritime infrastructure (European Commission, 1997) and White Paper on Fair Payment for Infrastructure Use (European Commission, 1998). Yet, the generalised adoption of this stance within the EU could end up creating additional imbalances, since it would leave the infrastructural gap between ports (primarily based on public funds) wide open and consequently transform privatisation processes into a surreptitious form of state aid for businesses. This is precisely the opinion expressed by the associations for Europe’s port authorities and terminal companies, ESPO (European Sea Port Organisation) and FEPORT (Federation of European Private Port Operators), in a joint paper that offers up an alternative proposal, whereby the State remains responsible for realising general-use public infrastructure projects (ESPO-FEPORT, 1999). The association for Italy’s port authorities (Assoporti) pronounced itself against the ‘‘the user pays’’ principle, warning that it would have crystallised the imbalance built up until this point between Mediterranean ports and Northern European countries. It should be nonetheless pointed out that in the same period the ports of Northern Europe questioned the actual lawfulness of the sizeable public financing obtained by the Port of Gioia Tauro, and the competitive imbalances triggered as a result (Baird, 2002). The mixed public–private scheme could be adequate, but (only) by shifting fiscal and financial leverage to a local level (or by keeping it there), and if possible to the port authority itself. The results should be similar to a completely private financial model. Indeed, while the presence of a public taxation system centred upon the port authority may well be maintained, its complete autonomy from central and regional governments would force the port development to fully recover the cost of infrastructural investments. At that point, choosing between self financing, an equity capital increase or borrowing would respond to the same financial logic as those adopted by a private operator. A question mark still hangs, however, over whether the level of tax yield generated by the ports is actually able to balance out the investments made in their favour. The effectiveness of port investment relies heavily on the funding mechanism used, and specifically on whether the yield generated from the use of the infrastructure and the decision-making process can be traced back to the same player. Specifically, if port investments are to be funded (increasingly) via the financial markets and free of any state subsidisation, then the investment needs to be profitable.13 On the contrary, centralised taxation led

214

ENRICO MUSSO ET AL.

to a profit distribution driven by neither the effectiveness of the port services themselves nor the investment’s profitability. On the other hand, a private investment can quite comfortably fit into a decentralised public finance system, where making the port authority or another operator responsible makes considering and evaluating the actual need for the investment unavoidable. It would therefore seem appropriate to guide the decision-making level for investments and their funding towards the local level and towards the world of private funding, the aim being to  keep responsibility on the decision-maker, thereby forcing them to consider benefits in relation to the investment’s actual cost (in this logic, incentives linked to the level of throughput may also be hypothesised);  link funding to economic results and make public fund flows more transparent;  bring back to a local/regional level the decisions concerning the level of ‘‘involvement’’ in the port industry;  enhance competition in the seaport sector, by replacing public operators with private operators wherever possible;  face up to the growing size of investments required, in a period when public funding is generally restricted; and  increase flexibility and speed in investment decisions, in response to the increasingly rapid transformations of the market.

8. CONCLUSION Port investment is a key issue in modern port economics with regard to planning port development, financing and assessing the return on investment. In the literature, the topic of infrastructure investment has been historically tackled either from a pure macro-economic perspective (i.e. investment in links and nodes of the network as a fundamental condition for economic growth: see Banister & Berechman, 2000) or from the mere firm’s point of view (the managerial decision process related to port investment). These approaches focus mainly on the macro-economic costs and benefits of the port industry and, on the other side, on the economic efficiency of the port function for port users. This paper overcomes that kind of segmentation. It addresses some of the features related to port investment starting from the evaluation of the main paradigms that characterize the port industry from a global point of view,

Port Investment

215

and focuses on the relations, synergies and conflicts between the numerous stakeholders actually involved. Profitability, economic impact and financing are seen as the most critical nodes in the complex chain of port investment decisions. Port investment has been described as the result of the equilibrium of several interactions between different forces and interests, where the most relevant aspects are (i) the public/private combination, which imprints the port industry and (ii) the geographical scale of evaluation. The mix between public and private interests, and the specific role of public bodies, may in fact be seen as the core of a specific port investment theory, which evaluates direct and indirect effects as well as uncertainty in returns. The different perspectives of evaluating the impact of port investment can lead to a different evaluation of the costs and benefits involved, and their desirability. The framework of the paper has been primarily based on the description and critical evaluation of the public/private and local/global tradeoffs, which in turn affect the assessment of port impacts, the development of funding, pricing and tax systems, the competitive scenario and distortions, which are likely to occur in inter-port and intra-port competitions. The main contribution of the paper may be seen in the effort of building up a comprehensive scenario where single aspects and variables related to port investments can fit into a general scheme of interrelationships, which identify feasible outcomes. The foreseeable outputs in terms of demand and supply provide insights for possible incentives to efficiency to be improved by decision-makers at different levels, promoting the reduction of conflicts and the synergies of interests. Although the topic has clearly practical implications, the work follows a theoretical approach rather than an empirical one. The proposal is, in fact, to develop an overall framework of analysis with a certain degree of originality in comparison with consolidated fields of the past literature, limiting at the same time the risk of a rapidly non-updated decision-support tool. The implementation of a number of outlined policy guidelines can be considered as an implicit agenda for future research.

NOTES 1. We refer here to gross investment, which includes the depreciation of existing capital over the period covered (net investment is instead equal to gross investment less the depreciation, and reflects the actual variation in productive capacity). 2. The marginal efficiency of capital (or internal rate of return) is the discount rate for which the present value of revenues expected from the investment is equal to the

216

ENRICO MUSSO ET AL.

present value of its costs. It is a subjective benchmark, since it includes the investor’s expectations. 3. The costs of other infrastructures are to be imputed to a group of users (for example, the costs of a single wharf or landing place to those utilising the terminal), while others are to be imputed with precision to each single use (cranes and other superstructures). 4. Bonnafous and Jensen (2005) point out that the level of support required increases more than proportionately with the target set for IRR, and show that in this instance the most efficient investment selection criterion is not in fact based on a higher NPV, but rather on a higher ratio between NPV and the level of public support. 5. We assume that the operator carrying out the investment is the same entity that manages the terminal and sells services to the carrier. This happens when the port is entirely public or private. In the case of a landlord port, it holds true for those investments made directly by the terminal manager (e.g. infrastructures and terminal superstructures). If instead the investor is different from the terminal operator, the increase in capacity will lead to an increase in the manager’s potential profit, which may or may not lead to an increase in the rent charged for the asset. 6. Other economic, political or social objectives can also occur, such as: promoting and encouraging maritime transport for a lower environmental impact, boosting competitiveness in a particular shipping route or port range in order to encourage the development of a region. Then, transport (and port) prices can be used to pursue objectives in other sectors. 7. Li Donni (1998) uses the ‘‘hiding hand’’ principle to study the case of the port of Gioia Tauro, which was realised in the 1970s as an iron and steel port but was never brought on-stream due to market changes and which in the 1990s instead became the Mediterranean’s main hub port. Had that erroneous investment not been made by the public body concerned, then today Italy’s ports would probably not be competing for leadership in the Mediterranean with Spain’s own ports. 8. As an example within Europe, Austria, the Czech Republic, Hungary and Switzerland are all usually served through Belgian, Dutch, German, Italian, Polish and Slovenian ports, which compared with that hinterland may be considered to be competing with one another. 9. We should point out, however, that even limiting our observations to the development of a terminal (e.g. containers), infrastructure investments may vary dramatically, depending on the type of work foreseen and the project’s term and financial burden. Here we move from one extreme involving the realisation of a new terminal from a greenfield situation (among the various realisation costs is the opportunity cost of allocating a coastal area to the port operation, which should be computed by the public body), to a situation whereby an existing terminal is expanded or its current use converted so that it may be used for a different function or other type of traffic. 10. In the absence of any specific Guidelines, one stance taken by the European Commission is defined in the Commission Staff Working Document ‘‘Vademecum on Community rules on State Aid and the financing of the construction of seaport infrastructures’’ (15.01.2002). 11. The aforementioned features of non-excludability and non-rivalry only recur for some port service inputs (shallows and port access points, lighthouses and

Port Investment

217

navigation safety projects, etc.), while the port service itself – including terminal services and nautical services – is rather a saleable service. Therefore, a free-riding problem does not arise, rather there is a political decision and opinion about the desirability of port services being widely utilised, owing to benefits in terms of market accessibility, economic development, environmentally friendly modal split, etc. 12. Stevens (1999) underlines how, while in the first few decades after World War II, depreciating infrastructures over 40 years or more was fairly standard practice, these days the depreciation period has been reduced to around 25 years, with the depreciation period used for superstructures peaking at 10–15 years. 13. An approach with results that are not that different from project financing, which over the last few years has spread out and developed as a tool used to attract private capital for public investments, thanks to the income generated from the management of the infrastructure.

ACKNOWLEDGMENTS The paper results from the close co-operation of the authors. However, Sections 1–4, 7 are by E. Musso, Section 5 is by C. Ferrari and Sections 6 and 8 are by M. Benacchio.

REFERENCES Abel, A. B., & Eberly, J. C. (1999). The effects of irreversibility and uncertainty on capital accumulation. Journal of Monetary Economics, 44(3), 339–377. Baird, A. (2002). Privatisation trends at port in the world’s top-100 container ports. Maritime Policy and Management, 29(3), 267–280. Banister, D., & Berechman, J. (2000). Transport investment and economic development. London: UCL Press. Bonnafous, A., & Jensen, P. (2005). Ranking transport projects by their socioeconomic value or financial internal rate of return? Transport Policy, 12, 131–136. Chirinko, R. S., & Schaller, H. (2001). The irreversibility premium. Paper presented at the ESRC Conference, April, London. Drewry Shipping Consultants Ltd. (2005). Annual review of global container terminal operators. London: Drewry Publications. ESPO-FEPORT (European Sea Ports Organization-Federation of European Private Ports Operators). (1999). State aid in the seaport sector: Analysis and policy recommendations. Bruxelles: Available on www.feport.be Estache, A., & Serebrisky, T. (2004). Where do we stand on transport infrastructure deregulation and public–private partnership? World Bank Policy Research Working Paper No. 3356. European Commission. (1997). Green Paper on ‘‘Seaports and maritime infrastructure’’, COM(1997) 678 final.

218

ENRICO MUSSO ET AL.

European Commission. (1998). White Paper on ‘‘Fair Payment for Infrastructure Use: A phased approach to a common transport infrastructure charging framework in the EU, COM(1998) 466 final. European Commission. (2002). Vademecum on Community rules on State Aid and the financing of the construction of seaport infrastructures, Working document. Ferrari, C., & Benacchio, M. (2003). Recent trends in the market structure of terminal services: Which way to integration? Pomorski Zbornik – Annals of Maritime Studies, 40, 153–176. Gilbert, R., & Vives, X. (1986). Entry deterrence and the free rider problem. Review of Economic Studies, 53(1), 71–83. Goss, R. O. (1990a). Economic policies and seaports: 1. The economic function of seaports. Maritime Policy and Management, 17(3), 207–219. Goss, R. O. (1990b). Economic policies and seaports: 2. The diversity of port policies. Maritime Policy and Management, 17(3), 221–234. Goss, R. O. (1990c). Economic policies and seaports: 3. Are port authorities necessary? Maritime Policy and Management, 17(3), 257–271. Goss, R. O. (1990d). Economic policies and seaports: 4. Strategies for port authorities. Maritime Policy and Management, 17(4), 273–288. Gripaios, P., & Gripaios, R. (1995). The impact of a port on its local economy: The case of Plymouth. Maritime Policy & Management, 22(1), 13–24. Haralambides, H. E., Cariou, P., & Benacchio, M. (2002). Costs, benefits and pricing of dedicated container terminals. International Journal of Maritime Economics, 4(1), 21–34. Haralambides, H., & Veenstra, A. (2002). Port pricing. In: C. Th. Grammenos (Ed.), The handbook of maritime economics and business (pp. 782–802). London/Hong Kong: Lloyds of London Press. Haralambides, H. E., Verbeke, A., Musso, E., & Benacchio, M. (2001). Port financing and pricing in the European Union: Theory, politics and reality. International Journal of Maritime Economics, 3(4), 368–386. Li Donni, V. (1998). La ‘‘mano che nasconde’’ e il ruolo delle esternalita` nei progetti di infrastrutture e trasporti. In Esternalita` e Trasporti IV Scientific Meeting of Italian Transport Economists’ Society, Trieste, 171–178. Musso, E., Ferrari, C., & Benacchio, M. (1999). On the global optimum size of port terminals. International Journal of Transport Economics, XXVI(3), 415–437. Musso, E., & Marchese, U. (2002). Economics of shortsea shipping. In: C. Th. Gramenos (Ed.), The handbook of maritime economics and business (pp. 280–304). London/Hong Kong: Lloyd’s of London Press. Ramsey, F. (1927). A contribution to the theory of taxation. Economic Journal, 37, 47–61. Stevens, H. (1999). The institutional position of seaports: An international comparison. Dordrecht: Kluwer Academic Publishers. Tirole, J. (1988). The theory of industrial organization. Cambridge (MA): MIT Press. von Stackelberg, H. (1934). Marktform und Gleichgewicht. Vienna: J. Springer. Wiegmans, B. W., Ubbels, B., Rietveld, P., & Nijkamp, P. (2002). Investment in container terminals: Public private partnership in Europe. International Journal of Maritime Economics, 4(1), 1–20.

SHIPPING DEREGULATION’S WAGE EFFECT ON LOW AND HIGH WAGE DOCKWORKERS James Peoples, Wayne K. Talley and Pithoon Thanabordeekij ABSTRACT This paper examines regional wage patterns of low- and high-wage dockworkers following deregulation. Findings reveal significant wage premium increases for low-wage dockworkers residing in the East and West Coasts following the initial deregulation in 1984. These premium gains surpass post-deregulation gains of high-wage dockworkers. Dockworker premiums do not change significantly for individuals residing in the southern U.S. These findings suggest that low-wage nonunion competition played a key role suppressing wages in the southern U.S. Unparallel union bargaining power in the East and West Coasts contributed to the high-relative wage gains of low-skilled dockworkers. In contrast to the post-1984 deregulation wage patterns, high-wage dockworkers received larger premium gains compared to the gains for low-wage dockworkers following passage of the Ocean Shipping Reform Act of 1998. These post1998 findings support the hypothesis that the challenges in employing high-skilled dockworkers in the late 1990s contributed to their relativewage gains. Port Economics Research in Transportation Economics, Volume 16, 219–249 Copyright r 2006 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 0739-8859/doi:10.1016/S0739-8859(06)16009-6

219

220

JAMES PEOPLES ET AL.

1. INTRODUCTION Ocean transport is a key contributor to economic growth in the U.S. This industry has also traditionally provided good pay for union work. The growing dependence on foreign products by U.S. industries in the latter half of the 20th century created the potential for increasing high-paying job opportunities for dockworkers. Even though the volume of cargo handled by U.S. ports has grown appreciably, the introduction of new transportation technology, such as containerization eroded job security for dockworkers (Chadwin, Pope, & Talley, 1990). Job losses were especially acute for dockworkers with low-skill content job responsibilities. However, the stimulus in U.S. trade following the passage of the Shipping Act of 1984 (which deregulated shipping in U.S. waters) has resulted in a net increase in the demand for dockworkers, i.e., this stimulus in the demand for dockworkers has more than offset the decrease in demand from technological improvements. The economic deregulation of shipping lines may not only affect the wages of their employees, e.g., seafarers, but also the wages of dockworkers in the related port industry. Talley (2002) found that the real hourly and weekly wages of U.S. union dockworkers increased 14.3 percent and 15.3 percent, respectively, in the post-1984 deregulation period. These results, however, are in contrast to those found in the transportation carrier literature, where the wages of the employees of the deregulated carriers have declined in the deregulation period. For example, the real weekly wages of union railroad engineers have declined 11.7 percent (Talley & SchwarzMiller, 1998) and the real hourly wages of union and nonunion for-hire truck carrier drivers have declined 23 percent and 10 percent (Hirsch & Macpherson, 1998) under railroad and trucking deregulation. ‘‘The primary reason for the widening in the wage gaps among the occupations in the deregulation period appears to be the change in the relative bargaining power among the occupations. For both truck carriers and railroads, there has been a shift in the balance of power in wage negotiations from unions to management. For dockworkers, the shift has been from management to unions’’ (Talley, 2004, p. 222). Shipping deregulation has enhanced wage premiums for union dockworkers employed by ports located in the East and West Coasts of the U.S. (Talley, 2002). Furthermore, this premium growth is also likely to differ by wage levels. Wage gains of low-cost dockworker labor (LCDL) could easily outpace that of low-wage workers in other industries given the unparallel

Shipping Deregulation’s Wage Effect

221

negotiation advantage of dockworker unions. In contrast, the inelastic demand for high-wage union workers in other industries should make it more difficult for high-wage union dockworkers to receive wage gains that significantly outpace the gains of these other high-wage union workers. This study contributes to the literature on dockworker pay by using a quantile regression technique to examine whether deregulation’s wage effect differs by wage level. In addition, wage patterns are examined by geographic region to account for regional unions adopting different approaches to addressing the threat of competition from nonunion low-wage dockworkers. Such an analysis is significant in part because of the increasing demand for low-wage workers in this industry and because it contributes to our understanding of the conditions in which stepped-up competition contributes to higher relative earnings for low-wage workers. The remainder of this study is structured as follows. Section 2 reviews regulatory reform in the shipping industry with the purpose of highlighting the industry’s changing business environment. Section 3 reviews the labor history of dockworkers. Such a review helps explain why deregulation’s wage effect on union dockworkers should vary by region and by worker wage level. Data and the empirical approach used by this study are presented in Section 4. The results from estimating dockworker wage equations are presented in Section 5. Concluding remarks are presented in Section 6.

2. SHIPPING REGULATION AND DEREGULATION A pattern of federal regulation of transportation services started with passage of the U.S. Shipping Act of 1916. This Act created the U.S. Shipping Board, which was renamed the Federal Maritime Commission (FMC) in 1961. The Shipping Board was given ‘‘jurisdiction over common carriers by water operating in interstate or foreign commerce on the high seas and upon the Great Lakes’’ (Locklin, 1972, p. 746). The Act, however, did not include Board jurisdiction over inland waterways.1 The rationale for enacting this legislation was to protect carriers from ruinous competition. The risk of foreclosure associated with competitive pricing arose because the industry was characterized by economies of scale and high sunk cost. The Act created a business environment that helped carriers avoid ruinous competition by providing anti-trust immunity to shipping liner conference agreements. Liner conferences are shipping line cartels that provide scheduled vessel

222

JAMES PEOPLES ET AL.

service over specific trade routes and collectively discuss and set rates, usually only port-to-port rates.2 The Act also sought to promote the U.S. maritime industry by authorizing the organization of the Emergency Fleet Corporation to carry out a shipbuilding and ownership program to support the World War I effort.3 The 1916 Act was amended with passage of the Shipping Act of 1984. The 1984 Act reduced U.S. economic regulation, thus providing for economic deregulation, of shipping lines or carriers.4 It also sought to improve the intermodal transportation efficiency of foreign commerce and the negotiating powers of shippers. Specifically, the Act permitted service contracts between shippers and carriers/conferences and the ‘‘right to independent action’’ by conference carriers. Service contracts require shippers to commit to a minimum amount of cargo for a given length of time and carriers/conferences to commit to agreed upon rate schedules and service levels.5 The 1984 Act also permitted door-to-door as opposed to just port-to-port rates for shipping lines involved in U.S. foreign commerce. Door-to-door rates for cargo enabled container shipping lines to develop more costefficient ocean transportation networks – leaving the choice of port-of-call to the shipping line which incurred cost savings (from economies of ship size) in using larger container ships to call at load-center ports, where containers were accumulated. Since cargo to and from regions via the nearest port could no longer be guaranteed, U.S. container ports found themselves competing not only with neighboring ports, but also against ports hundreds of miles away. The door-to-door rates also allowed shipping lines to obtain, given their large volumes of container cargo, lower rates from inland carriers for the inland transportation of containers than could have been obtained by individual shippers.6 An example of the transition to more efficient transportation networks is the business model used by the U.S. carrier American President Line (APL). In 1984 APL, a U.S. flag-container shipping line, began offering landbridge service7 – i.e., its ships began calling at ports along the U.S. West Coast, where containers were unloaded and put on rail cars for the trip to East, as opposed to all-water service through the Panama Canal to the East Coast. APL contracted with railroads to operate double-stack trains over their rail lines. Double-stack trains consist of platform rail cars capable of moving containers stacked two high and have a cost advantage over conventional container-on-flat car (COFC) trains – for slightly more locomotive power, the same labor and slightly more fuel, 200 containers can be transported on a double-stack train as opposed to 100 containers on a COFC train. By the

Shipping Deregulation’s Wage Effect

223

late 1980s, the majority of the container cargo from Asia bound for the U.S. East Coast did not arrive by ship, but rather was discharged on the West Coast, and hauled by rail across the continent, thereby placing West Coast ports in competition with East Coast ports. Today, double-stack trains operate to and from ports on both the West and East Coasts. Movement toward stepped-up competition among ports and the ‘‘independent rate action’’ and the ‘‘intermodal rate-making’’ provisions of the Shipping Act of 1984 clearly reduced the ability of U.S. liner conferences to set noncompetitive rates. The Ocean Shipping Reform Act of 1998, which amended the 1984 Act and took effect May 1, 1999, reduced this ability even further and contributed to the further decline of conferences. Its ‘‘confidential contract’’ provision allows, for the first time, confidential one-on-one contracts by shipping lines, but not conferences, with their customers. Other major provisions of the 1998 Act include: (a) shippers remain subject to standard U.S. anti-trust law and ocean carriers are still subject to FMC regulation; (b) individual carrier tariff-filing requirements with the FMC have been eliminated, but carriers are required to publish rates via the Internet or other media; (c) contracts must be filed with the FMC for agency oversight; and (d) ocean carriers engaged in confidential arrangements with big shippers must disclose contractual information regarding specific dock and port movements to longshore unions. Between May 1, 1999 and May 31, 2000, 46,035 new service contracts and 95,627 contract amendments were filed with the FMC. During 1999, the number of active conference agreements on file at the FMC dropped from 33 to 22. The 1998 Act continued the trend toward shipping deregulation established by the 1984 Act. Its principal goal is to ‘‘provide shippers and ocean carriers greater choice and flexibility in entering into contractual relationships with shippers for ocean transportation and intermodal services’’ (Lewis & Vellenga, 2000, p. 29). Shippers have clearly benefited from the 1998 Act: (1) shippers have been able to tailor ocean/intermodal transportation rates and services to individual requirements and circumstances and (2) even smaller shippers have been able to gain leverage to reduce rates (Leach, 2005). Also, confidential contracts provide U.S. exporters protection from disclosure to foreign competitors of contractual relationships. Lower shipping rates and improved services for ocean-containerized cargo under shipping deregulation have contributed to the significant growth in U.S. containerized trade. However, container port competition has also significantly increased, placing these ports under pressure to improve productivity without disenfranchising dockworkers.

224

JAMES PEOPLES ET AL.

3. DOCKWORKERS Dockworkers are workers involved in the movement of cargo within a port or marine terminal. The two primary U.S. dockworker unions are the International Longshoremen’s Association (ILA) and the International Longshore and Warehouse Union (ILWU), East Coast and West Coast unions, respectively. The ILWU was formed in 1937 when West Coast union dockworkers broke away from the ILA. Dockworkers are also members of the Teamster union at Gulf Coast ports and at some East Coast ports where they and the ILA have overlapping jurisdictions. Immediately following the formation of the ILWU, industry unions faced a series of challenges affecting the welfare of their members. The following review of the history of the labor market trends for dockworkers reveals how industry unions responded to the threat of job loss from technological advancement in ocean transportation. In addition, this review provides important insight on the effect of technological change and deregulation on the wages of dockworkers. During the 1940s and 1950s in a move to improve productivity, dockworker employers sought to introduce labor-saving devices, such as lift trucks. In response, organized labor resisted with strikes and the threat of strikes. However, the ILWU’s resistance weakened in 1955 when a U.S. Congressional committee inquired into the productivity of longshoremen. The threat of government intervention led the ILWU in 1960 to sign the Mechanization and Modernization Agreement (MMA) with the Pacific Maritime Association (PMA), which represents West Coast ports in bargaining with the ILWU.8 In the agreement, dockworker employers could implement mechanized handling procedures in return for jurisdiction over new dockside equipment and other benefits, such as pay guarantees, higher wages, early retirement schemes, pensions and job maintenance guarantees, except under attrition and during economic slowdowns (Killingsworth, 1962; Betcherman & Rebne, 1987). 3.1. Containerization The movement toward improved productivity continued following the signing of the MMA. In the late 1960s, dockworker employers introduced containerization, the foremost port labor-saving technology. Containerization radically altered cargo-handling tasks, as capital was substituted for labor.9 Whereas a gang of 20 dockworkers could load 20 tons of cargo per hour on a break-bulk (general cargo on pallets) ship, one port crane and perhaps half

Shipping Deregulation’s Wage Effect

225

as many men could load 400–500 tons of general cargo (in containers) per hour on a container ship. Given the dramatic increase in job displacement due to technological change, a major labor issue for dockworker unions was how they would react to the anticipated job losses from containerization. The ILA’s strategy was to generate benefits for its members through new work rules, a coastwide master contract and developing a Guaranteed Annual Income (GAI) systems for workers displaced by technology. In the late 1960s, the ILA negotiated ‘‘rules on containers’’ (RCs): (1) a 50-mile port zone requiring stuffing or stripping by union dockworkers of all containers that originated from, or were destined for, places less than 50 miles from a port’s dock center and (2) a prohibition on leasing containers to inland consolidators. The importance of RCs to both dockworkers and their employers is evident in that they were a key issue in the 1971 strike, the first and only U.S. nationwide dockworker strike. The ILA wanted the 50-mile rule to be included in its master contract for all ports, from Maine to Texas. The ILA negotiated with a new coast-wide association, the Council of North Altantic Steamship Associations (CONASA). The new contract and the 1973 Dublin Agreement resulted in the 50-mile rule becoming a coast-wide rule.10 Smaller Southern and Gulf Coast ports resented how the larger East Coast ports, in particular New York, dominated the negotiations in ILA coast-wide master contracts. Since their agricultural and other goods would continue to be shipped as break-bulk cargoes, they were not as threatened by containerization. However, the establishment of CONASA in 1970 gave the smaller Southern and Gulf Coast ports a greater voice in negotiating master contracts. The ILA also negotiated local GAI plans and ‘‘work preservation’’ schemes. For example, the 1965 ILA GAI agreement at the Port of New York/New Jersey provided 1,600 paid hours per year to fully registered longshoremen who qualified by working 700 or more hours in a given fiscal year. In exchange, the ILA agreed to reduced gang sizes and greater flexibility in work practices. ILA jobs have also been threatened by lower-cost nonunion dockworker labor. The ILA has faced competition, in particular, from the hiring of nonunion dockworkers at Gulf Coast ports. The region’s large number of bulk and break-bulk cargo ports and right-to-work states has been conducive to the hiring of nonunion dockworkers. Today, bulk and break-bulk cargoes at Gulf Coast ports are handled almost entirely by non-ILA labor. The ILA Maine-to-Texas master contract requires container shipping lines to use ILA labor in all East and Gulf Coast ports.11 However, break-bulk

226

JAMES PEOPLES ET AL.

and bulk shipping lines are not subject to this contract, and thus are free to use non-ILA labor.12 Following the introduction of containerization, the demand for dockworkers dramatically declined. In 1950 there were 100,000 full-time dockworkers in the U.S.; by 2003, this number had declined to 10,500 (Greenwald, 2004).13 Alternatively, the productivity of dockworkers dramatically increased. In the New York City metropolitan area, for example, there were 35,000 dockworkers in 1954 and 2,700 in 2000. However, during the same time period, the port’s cargo tonnage increased from 13.2 to 44.9 million tons – i.e., the productivity per dockworker increased 3,118 percent, while the number of dockworkers decreased 95 percent (Greenwald, 2004). Dockworker productivity at container ports has been enhanced from improvements in communication and information technologies, infrastructures, and cranes. The wages and benefits of U.S. dockworkers have dramatically increased. In 2002, the average annual salary of a full-time ILWU worker ranged between $105,278 and $167,122. The average of the additional dockworker benefit package was $70,000 (Swoboda, 2002).

3.2. Shipping Deregulation Why would dockworker employers agree to pay six-figure salaries to generally unskilled workers? Prior to shipping deregulation, dockworkers accepted automation (therefore loss of jobs), new technologies, and work-rule changes in return for significantly higher wages and benefits. A contributing factor to the high dockworker wages has also been the significant decline under containerization in labor cost as a percent of a port’s total cost. Under shipping deregulation, the increase in dockworker bargaining power – attributed to the increase in demand for dockworkers (from the significant growth in container cargo) and the reluctance of dockworker employers to chance a strike (given the rising costs from disruptions in the utilization of expensive container port infrastructures and container ships and the potential lost of cargo to port competitors) – provided opportunities for even greater increases in dockworker wages and benefits. The door-to-door rate provision of the 1984 Act not only enabled container shipping lines to develop more cost-efficient ocean transportation networks in U.S. trades but also to obtain lower rates (from volume discounts) for the inland transportation of ocean container cargo. Lower ocean and inland transport rates (from lower transportation costs) and lower

Shipping Deregulation’s Wage Effect

227

shipper inventory and other logistics costs from improvements in ocean and inland transportation services (from service contracts) stimulated the growth in U.S. container cargo and therefore the demand for dockworkers.14 The response of shipping lines and U.S. ports to the growth in container cargo was to invest billions of dollars in larger container ships and in the expansion and improvement of container marine terminals, thereby increasing the costs from disruptions in the utilization of container ships and port infrastructures. Fearing that these disruptions would also result in the lost of cargo to port competitors, dockworker (port) employers became more reluctant to chance a strike – agreeing to settle labor contract extensions much earlier (prior to contract expiration) than in the past and agreeing to higher dockworker wage demands than would have occurred under more lengthy negotiations. Between World War II and the mid-1970s the ILA struck at the end of almost every master contract.15 However, ILA employers have recently become increasingly reluctant to risk dockworker strikes. ‘‘They (employers) are willing to pay higher wages – an ever-decreasing percentage of total costs (for container ports) – in exchange for being able to keep their ships, cranes, chassis, and other equipment running’’ (Armbruster, 2000, p. 17).16 In June 2000, ILA members approved a three-year master contract extension, more than a year before the old master contract was to expire and the third time in recent years that a contract extension had been approved without full-scale negotiations between the ILA and its employers. The 1999 ILWU contract gave the union a 9 percent wage increase in the first year and expanded the union’s jurisdiction, whereby employers agreed to train longshoremen for container maintenance and repair. ‘‘Maritime industry sources agree the ILWU contract that will be in effect until July 1, 1999 was negotiated from a position of almost absolute power by the union’’ (Mongelluzzo, 1996, p. 1B).

3.3. The ILWU and Technology In December 1999, the ILWU at the Ports of Long Beach and Los Angeles chose to retain its cumbersome manual system (dating back to the ILWU’s founding) of posting job assignments on a dispatch-hall chalkboard and manually writing work-order tickets at the start of each shift. Specifically, it rejected the implementation of a computerized job-dispatching system designed for getting dockworkers to start work on time. The ILWU views the computerized dispatch system as ‘‘an attempt by employers to eventually regain control of the dispatch hall’’ (Mongelluzzo, 1999, p. 1). ILWU

228

JAMES PEOPLES ET AL.

employers, however, continued to bargain for the implementation of information technology with resistance from the ILWU. Further, the ILWU successfully leveraged its work slowdowns to win concessions from employers. The bargaining situation between the PMA and the ILWU reached a boiling point in September 2002. The ILWU contract that was to expire on July 1, 2002 was extended on a day-to-day basis by the ILWU until September 1, 2002. ILWU work slowdowns followed with the productivity at West Coast ports dropping by 50 percent by September 30, 2002.17 The PMA responded with a lockout of the union. The lockout ended in 10 days when a federal judge, at the request of President Bush, invoked the Taft–Hartley Act that imposed an 80-day cooling-off period. Subsequently, negotiations between the PMA and the ILWU would resume under federal mediation. During the lockout, 200 vessels lay in anchorage at West Coast ports, resulting in a cargo backlog that would require six weeks to clear and in millions of dollars in losses to shippers (U.S. and Asian) and shipping lines. The reputation of the ILWU has suffered from its involvement in the negative impact of the 10-day lockout on the U.S. economy.18 As a consequence, it is fearful that the federal government will weaken its bargaining power.19 There is speculation that Congress will remove labor relations at U.S. ports from the National Labor Relations Act and place them under the 1926 Railway Labor Act, which eliminates the ability of a union to use the threat of a strike to gain the upper hand in contract negotiations. The 1926 Act was created when railroads had a monopoly on long-distance transportation, ‘‘a situation some would say bears a striking resemblance to the ILWU’s monopoly on West Coast port labor’’ (Tirschwell, 2002, p. 6). On November 23, 2002, the ILWU and the PMA agreed to a tentative sixyear contract that was subsequently ratified by ILWU members. In return for allowing for the implementation of new information technology by West Coast ports, the ILWU would receive an increase of $3 per hour in its hourly base wage of $27.68 over the six-year contract, job protection guarantees to ensure that no currently registered worker will lose a job as a result of technology and a 58 percent increase in pension benefits. The mediated contract was ‘‘the most lucrative in the union’s 70-year history’’ (Tirschwell, 2003, p. 6) and ‘‘more favorable to employers than they could have achieved on their own’’ (Mongelluzzo, 2003b, p. 26). The new contract will gradually remove ILWU marine clerks from positions where they performed redundant work that slowed cargo-handling operations, e.g., re-keying cargo information that was previously filed

Shipping Deregulation’s Wage Effect

229

electronically. Prior to the new contract, marine clerks were guaranteed at least eight hours of straight-time pay and two hours of overtime pay for working an eight-hour day. These perks, in turn, provided an incentive for longshoremen trained as skilled equipment operators to often choose available marine clerk as opposed to available equipment operator jobs at work dispatch halls, resulting in a shortage of skilled equipment operators. Under the new contract, the position of the skilled equipment operator is elevated to a Skill III level, paying $33.48 per hour and guaranteed overtime pay each day. Crane operators can be hired as steady workers (i.e., at a given marine terminal) to work five days a week for seven days pay versus three days a week for five days pay under the old contract. The new contract also classifies crane work as ‘‘four hours on, four hours off,’’ thereby defining a crane operator’s day of work to be four hours. ‘‘Employers accept the extra cost because they say it is more than offset by higher productivity from crane drivers who work regularly at the same terminal’’ (Mongelluzzo, 2003a, p. 37). The new contract also allows the ILWU to challenge the implementation of new technology when it believes its jurisdiction is threatened. An arbitration process with deadlines for decisions is established. In the past, challenges to new technology could drag on for years. Under the new contract, the power of local arbitrators is reduced. The coast arbitrator has the final and binding authority on issues related to the introduction of technology. In sum, three key observations arise from examining the regulatory history of the ocean transportation industry and the labor history of dockworkers. One key observation reveals that following deregulation union negotiation strength was much weaker in southern Gulf Coast states compared to the negotiation advantage of dockworker unions representing workers at the East and West Coast ports. Hence, it likely that the postderegulation wage premiums for dockworkers residing in the southern U.S. did not increase at East and West Coast rates. The second key observation reveals that following initial deregulation in 1984 the enhanced negotiation advantage by East and West Coast dockworker unions contributed to nontrivial wage gains for low-skilled union dockworkers in these regions. Such wage gains are not likely to be matched by low-skilled union workers in other industries. The third key observation reveals challenges in employing highly skilled dockworkers following passage of the 1998 Act. Changing workloads and nontrivial wage gains for high-skilled dockworkers should contribute to high-relative wages for this group of workers following passage of the 1998 Act.

230

JAMES PEOPLES ET AL.

4. DATA AND EMPIRICAL APPROACH 4.1. Data Three-hundred-thirty-six monthly Current Population Survey Outgoing Rotation Group (CPS-ORG) files for January 1973 through December 2001 are used to construct data on wage patterns for individual workers.20 These sources report worker characteristics, current hourly wages,21 regional residency, and industry and occupation of employment.22The sample is limited to union dockworkers and nontransportation operatives 16-year old or older, who worked 30 hours or more a week and report their hourly wages. The sample is restricted to union workers, because they represent such a large share of the dockworker work force that the sample of nonunion dockworkers is too small to provide reliable empirical results.23 Nonetheless, the resulting population of 391 dockworkers and 14,260 nontransportation operatives is large enough to separately examine wage patterns of union dockworkers across regions. Distinguishing regional residency of workers is key to examining whether strong union negotiation strength on the East and West Coasts presents their members with a wage advantage that surpasses that of dockworkers located in the south. Shortcomings associated with the use of (CPS-ORG) files to examine relative wages are the lack of information on nonwage compensation and firm level information. While information on health care and pension plans are available for March CPS files, the sample population for dockworkers is extremely small. March files also report information on firm characteristics such as firm size; however, before 1983 this information is only reported every five years. Descriptive statistics depicting mean worker profiles derived from using the CPS-ORG files are presented in Table 1. Information reported in this table is partitioned by the three major regulatory regimes for union dockworkers and nontransportation operatives. The first two columns report the prederegulation mean profiles of these two worker groups. The findings suggest that compared to union nontransportation workers, union dockworkers are older and are more likely to be nonwhite, married, college educated, and reside outside the Midwest region of the US. The pre-deregulation findings also indicate that union dockworkers are less likely than union nontransportation operatives to have attained a high school diploma. Finding low high-school diploma attainment levels is consistent with the observation that a large share of dockworker jobs such as longshoremen and marine clerks were low-skill content occupations prior to deregulation.

Descriptive Statistics on Union Dockworkers and NonTransportation Operatives. Pre-Deregulation

Variable

High school (%) College (%) White (%) Black (%) Married (%) South (%) Northeast (%) West (%) Midwest (%) Age (year) Hourly wage (1982—1984 $)

Post-Deregulation

Post-Deregulation

Dockworker (1)

Nontransportation Operative (2)

Dockworker (3)

Nontransportation Operative (4)

Dockworker (5)

Nontransportation Operative (6)

30.87 21.47 59.73 38.62 79.86 36.24 32.88 26.84 4.02 43.59 10.77

52.63 12.33 86.91 11.89 75.49 18.82 22.64 11.07 47.44 38.75 9.68

49.79 23.86 60.49 32.72 81.48 32.51 24.27 41.15 2.05 45.74 14.18

53.25 29.90 85.35 11.88 78.60 19.84 22.93 11.08 46.13 42.51 9.08

50.00 35.71 85.71 7.14 71.42 28.57 10.71 60.71 0.00 45.50 14.22

54.52 34.68 81.69 13.10 69.17 17.14 22.35 18.11 42.38 44.80 9.73

Shipping Deregulation’s Wage Effect

Table 1.

 1973–1983, excluding 1982.  1984–1998. 1999–2001.

231

232

JAMES PEOPLES ET AL.

Excluding the findings on geographic region of residency, the contents in columns 3–6 indicate that the mean profile of dockworkers more closely resembles that of nontransportation operatives following deregulation, especially following the 1998 deregulation act. Dockworkers are still unlikely to reside in the Midwest region of the US, but are much more likely to reside in the West region of the U.S. following deregulation. The trend of an increasing share of dockworkers residing in the West region is indicative of the growing share of ocean shipping cargo handled at the ports of Los Angeles and Long Beach, California. Information reported in Table 1 also shows the pre- and post-deregulation mean wage levels of union dockworkers and the comparison group of nontransportation operatives. These wage findings reveal the same wage pattern of high-wage payment found when observing wage negotiation settlements for dockworkers. For instance, dockworkers enjoy an 11.26 percent prederegulation mean wage advantage over nontransportation operatives. This wage premium increases to 56.17 and 46.15 percent following the 1984 and 1998 deregulation acts, respectively.24 Additional information in Table 2 shows relative wage patterns for lowand high-wage dockworkers. Entries for low wage levels depict the mean value of wages that are one standard deviation or more below the mean wage for all workers for the respective occupation and regulatory regime. Entries for high wage levels depict the mean value of wages that are one standard deviation or more above the mean wage for all workers for the respective occupation and regulatory regime. Bifurcating the sample by Table 2. Distribution of Hourly Wages for Low- and High-Wage Union Dockworkers and NonTransportation Operatives (1982–1984 $)a. Dockworker

Pre-deregulation Post-deregulation Post-deregulation a

NonTransportation Operative

Low wage ($)

High wage ($)

Low wage ($)

High wage ($)

7.25 8.43 8.04

15.99 19.84 22.17

5.46 6.05 5.95

15.48 13.01 13.36

Entries for low wage levels depict the mean value of wages that are one standard deviation or more below the mean wage for all workers for the respective occupation and regulatory regime. Entries for high wage levels depict the mean value of wages that are one standard deviation or more above the mean wage for all workers for the respective occupation and regulatory regime.  1973–1983, excluding 1982.  1984–1998.  1999–2001.

Shipping Deregulation’s Wage Effect

233

wage levels uncovers an interesting difference in post-deregulation dockworker wage gains. The bifurcated mean wage findings show a much larger increase in the wage advantage of low-wage dockworkers compared to the increasing wage advantage of high-wage dockworkers following initial deregulation. The mean wage advantage for low-wage dockworkers over low-wage workers in other industries increases from 32.78 percent to 39.34 percent following deregulation in 1984. In contrast, the mean wage advantage for high-wage dockworkers over high-wage workers in other industries increases from 3.29 percent to 52.50 percent. The larger relative gains for low-wage dockworkers are consistent with dockworker unions using their post-deregulation negotiation advantage to attain lucrative wage settlements for relatively lower-skill content occupations. Findings in row three show that the wage advantage of low-wage dockworkers falls to 35.13 percent following the 1998 Act but remains above the pre-1984 deregulation level. In contrast, the wage advantage for high-wage dockworkers increases to 65.94 percent following the 1998 Act. This contrasting wage pattern for low- and high-wage dockworkers following the 1998 Act is consistent with employers responding to the shortage of skilled equipment workers by agreeing to labor negotiations that increase wages and reclassify a day’s work. 4.2. Empirical Approach Mean wage findings reveal the significance of observing deregulation’s separate wage influence on low- and high-wage dockworkers. A more complete empirical analysis, though, requires the use of multivariate techniques to control for differing worker characteristics. Initially, Eq. (1) is estimated for the entire sample of union dockworkers and union nontransportation operatives in other industries. Then, union wage Eq. (2) is estimated separately for each sample of workers residing in the southern, western, and eastern U.S.25 lnðWageÞi ¼ a þ b1 Xi þ b2 Docki þ b3 Deregi þ b4 ðDock  DeregÞi þ b5 Regioni þ mi

ð1Þ

lnðWageÞis ¼ a þ b1 Xis þ b2 Dockis þ b3 Deregis þ b4 ðDock  DeregÞis þ mis

ð2Þ

where i indexes individual workers, s indexes the three non-Midwest geographic regions of the U.S and the dependent variable ln(Wage) is the natural log of the inputted hourly wage in 1982–1984 in dollars.26 The matrix X

234

JAMES PEOPLES ET AL.

consists of individual worker information including age, as well as dummy variables identifying marital status, race, and educational attainment. In addition, the annual national unemployment rate is included as an occupation determinant to control for time-variant distortions such as changes in the business cycle. These macroeconomic controls are of added significance when examining regional wage patterns. For instance, dockworkers residing in the West Coast benefit from macroeconomic effects such as increasing trade with China.27 The matrix Region consists of three regional dummy variables that denote individual workers’ residency in one of the three non-Midwest geographic regions. The variable Dock is a dummy variable equaling one if the individual is employed as a dockworker and zero if the individual is employed as a nontransportation operative. The variable Dereg is a dummy variable depicting the post-deregulation observation period. The final variable in the occupational status equations is the interaction of the Dock and Dereg dummy variables. The coefficients that are of special interest to this study are b2, b3, and b3+b4. They are used to measure dockworker earnings differentials for workers by regulatory regimes. In Eq. (1), the coefficient b2 measures the log earnings differential between dockworkers and nontransportation operatives for the pre-deregulation observation period.28 The coefficient b3 measures the log earnings change for nontransportation operatives following deregulation. The sum of b3 and b4 measures the log earnings change for dockworkers following deregulation. Hence, the coefficient b4 measures the difference in the log earnings change between dockworkers and nontransportation operatives following deregulation.29 Given this study’s emphasis on deregulation’s wage influence across wage levels, a quantile regression approach is used to estimate Eqs. (1) and (2). The advantage of this approach compared to an OLS estimation approach is that the quantile regression method allows for parameter heterogeneity across different wage levels.30 The quantile regression model used to examine the wage relationship depicted by this study’s wage equation for each geographic region is presented below. The quantile method for estimating Eq. (1) mirrors this approach. The only difference is that the matrix Z in estimating Eq. (2) does not include the vector of regional dummies. lnðWageÞis ¼ bjy Zis þ myis

(3)

where j indexes the vector of parameters presented in Eq. (2), y indexes quantile groups, and Z is the matrix of the explanatory variables presented in Eq. (2), including the dockworker and deregulation dummies. All other

Shipping Deregulation’s Wage Effect

235

subscripts retain their same meaning as reported for Eq. (2). The parameter estimates for the quantile regression are derived by using a general methods of moment (GMM) approach to solve the following minimization problem for each regional sample population31 8 < X b by ¼ arg min y ln ðWageÞi bj Zi : b i:lnðWageÞi 4bj Zi 9 = X ð1 yÞ ln ðWageÞi bj Zi ð4Þ þ ; j i:lnðWageÞi 4b Zi

b is estimated for quantile values y ¼ 0.25, 0.50, and 0.75. Hence, b is estimated using symmetric quantile weights for the median case y ¼ 0.50 and asymmetric otherwise. In addition, all sample observations are used for each quantile regression estimate of b. Such an approach avoids sample selection bias that arises when partitioning data to estimate regressions for each quantile group.

5. WAGE RESULTS The OLS wage results from estimating Eq. (1) are presented in Table 3. A brief review of the control variables in this table reveals that excluding educational attainment the signs of their estimated coefficients are consistent with standard labor theory. For instance, experience, white racial status, and marital status are associated with statistically significant high wages. The findings on the estimated coefficient Dock indicate that dockworkers receive a statistically significant wage premium relative to nontransportation operatives prior to deregulation. The sum of this estimated coefficient and the interaction term Dockdereg suggests that premiums for dockworkers increased to 74.5 and 34 percent following regulatory reform in 1984 and 1998, respectively. The post-1984 findings are consistent with past research (Talley, 2002). The post-1998 findings, though, are new to the literature, and suggest a smaller post-deregulation premium for latter years. The quantile results presented in Table 4 reveal the wage premium distribution for pre- and post-deregulation wage levels. The estimated coefficient Dock is noticeably smaller for the 75th quantile group. This finding suggest that low-wage dockworkers experienced a larger wage advantage over low-wage nontransport workers than high-wage dockworkers experienced over high-wage nontransport workers. Findings on the estimated

236

Table 3.

JAMES PEOPLES ET AL.

OLS Hourly (LOG) Wage Results for Union Dockworkers Versus Union NonTransportation Operativesa. Estimated Coefficient (1)

College High school Age Age2 White Married South Northeast West Dock Dockdreg Dreg Unemployemnt Intercept

0.4250888 0.0565466 0.0177764 0.0002182 0.0432274 0.0881391 0.1047925 0.0945991 0.097518 0.2177091 0.3394535 0.1543105 0.0220136 1.684612

Number of observations ¼ 13433 R2 ¼ 0.0879 Adjusted R2 ¼ 0.0870

t-statistic (2)

22.21 3.58 4.33 4.47 2.21 5.17 5.72 5.45 4.42 2.88 3.67 10.27 3.83 18.56

Estimated Coefficient (3) 0.1661153 0.0981197 0.0192799 0.0002017 0.0466607 0.1055257 0.0791377 0.1084543 0.0666978 0.1535416 0.1398422 0533924 0.0035638 1.668018

t-statistic (4)

13.51 10.98 8.83 7.63 4.20 11.08 7.63 11.04 5.50 4.79 2.70 5.30 1.44 36.03

Number of observations ¼ 7584 R2 ¼ 0.1008 Adjusted R2 ¼ 0.0993

a

Results from estimating the wage equation for the regulation/1984 Act deregulation sample (years 1973–1998, excluding 1982) are presented in columns 1 and 2; Dreg ¼ 1 for 1984 and after and zero otherwise. Results from estimating the wage equation for the regulation/1998 Act deregulation sample (years 1973–1983 and 1999–2001, excluding 1982) are presented in columns 3 and 4; Dreg ¼ 1 for 1999 and after and zero otherwise.

coefficient Dockdereg suggest that low-wage dockworkers’ relative wage premium rose even higher than that for high-wage dockworkers following passage of the 1984 Act; the additional post-deregulation premium for the 25th quantile group is 58.4 percent compared to 49.9 percent for the 50th and 41.7 percent for the 75th quantile group. In contrast, this coefficient is markedly larger for the 75th quantile group following passage of the 1998 Act. For instance, the estimated coefficient Dockdereg is 0.1286 and 0.1003 for the 25th and 50th quantile group compared to 0.1895 for the 75th quantile group. These findings are consistent with this study’s two hypotheses that enhanced union negotiation strength contributed to low-skilled workers enjoying a large wage advantage following the 1984 act, and high demand for skilled workers contributed to these individuals enjoying a large wage advantage following the 1998 act.

Shipping Deregulation’s Wage Effect

237

Table 4. Quantile Hourly (LOG) Wage Results for Union Dockworkers Versus Union NonTransportation Operativesa. Estimated Coefficient (1)

t-statistic (2)

Estimated Coefficient (3)

t-statistic (4)

25th Quantile College High school Age Age2 White Married South West North east Dock Dockdreg Dreg Unemployment Intercept

0.0367005 0.1053973 0.027414 0.0002974 0.0709441 0.1211353 0.1383178 0.0178637 0.1217898 0.1997165 0.260328 0.0912782 0.0008645 1.275415

2.06 7.37 11.29 9.28 5.71 7.61 8.92 0.81 8.28 3.76 3.54 8.96 0.20 22.47

0.1760 0.1128 0.0215 0.00022 0.0427 0.1493 0.1228 0.0207 0.1348 0.1886 0.1286 0.0684 0.0020 1.3598

9.67 9.12 5.94 5.17 2.10 14.53 9.85 1.21 9.81 4.19 1.77 4.10 0.74 20.31

50th Quantile College High school Age Age2 White Married South West Northeast Dock Dockdreg Dreg Unemployemnt Intercept

0.0825518 0.0915489 0.0284777 0.0003142 0.0421934 0.0972917 0.1152518 0.0596778 0.1143289 0.2028396 0.2020154 0.0674689 0.0050966 1.571357

6.98 9.43 26.47 25.40 3.43 8.31 10.31 3.44 9.06 4.00 2.84 10.27 1.36 40.10

0.0170 0.1011 0.0217 0.00023 0.0309 0.1177 0.1000 0.0727 0.1229 0.1648 0.1003 0.054 0.00029 1.635

9.63 8.94 7.89 7.10 1.95 7.84 11.84 4.76 9.45 3.69 2.27 4.57 0.11 23.80

75th Quantile College High school Age Age2 West Married South West Northeast Dock

0.0956872 0.0708538 0.0228923 0.0002449 0.0454458 0.0671996 0.0531593 0.075418 0.0847444 0.1484503

9.76 9.31 12.35 10.82 6.61 7.92 7.51 5.33 10.31 4.04

0.1311 0.0727 0.0167 0.00017 0.0412 0.0701 0.0439 0.07768 0.0890 0.1367

10.03 7.39 8.04 6.90 3.42 6.49 4.13 6.39 8.91 2.92

238

JAMES PEOPLES ET AL.

Table 4. (Continued ) Estimated Coefficient (1) Dockdreg Dreg Unemployment Intercept

0.2003624 0.028745 0.0097844 1.899551

Number of observation ¼ 13433 0.25 Pseudo R2 ¼ 0.0330 0.50 Pseudo R2 ¼ 0.0395 0.75 Pseudo R2 ¼ 0.0394

t-statistic (2)

4.98 5.43 5.48 49.70

Estimated Coefficient (3) 0.1895 0.0048 0.0088 2.010

t-statistic (4)

3.13 0.42 3.28 41.94

Number of observation ¼ 7584 0.25 Pseudo R2 ¼ 0.0630 0.50 Pseudo R2 ¼ 0.0586 0.75 Pseudo R2 ¼ 0.4087

a

Results from estimating the wage equation for the regulation/1984 Act deregulation sample (years 1973–1998, excluding 1982) are presented in columns 1 and 2; Dreg ¼ 1 for 1984 and after and zero otherwise. Results from estimating the wage equation for the regulation/1998 Act deregulation sample (years 1973–1983 and 1999–2001, excluding 1982) are presented in columns 3 and 4; Dreg ¼ 1 for 1999 and after and zero otherwise.

Regional wage patterns presented in Tables 5–8 are used to examine whether regional union strength creates a wage advantage for union dockworkers residing in the East and West Coasts compared to union dockworkers residing in the South following deregulation. Results from using the OLS procedure are presented in Table 5. The estimated coefficient on the dummy variable Dock suggests that the pre-deregulation premium for dockworkers varies by region and is largest for workers residing in the South. These findings closely match findings of previous research on regional wage patterns for dockworkers (Talley, 2002). The estimated coefficient on the interaction term Dockdereg reveals that deregulation is associated with larger increases in the wage premium of dockworkers residing in the Northeast and West. These findings support this study’s hypothesis that regional union strength on the coast promotes a greater interindustry wage advantage for dockworkers employed in these regions compared to the interindustry wage advantage for dockworkers in the South following deregulation. Findings presented in Tables 6–8 reveal a more nuanced deregulation effect on dockworker premiums across regions by revealing dockworker premiums differing by quantile group. Wage results for union workers residing in the Southern U.S. are presented in Table 6. The estimated coefficient on the dockworker dummy suggests that dockworkers residing in the South receive a statistically significant wage premium prior to deregulation. The difference in this pre-deregulation premium is largest

Shipping Deregulation’s Wage Effect

239

Table 5. Regional OLS Hourly (LOG) Wage Results for Union Dockworkers Versus Union NonTransportation Operativesa. Estimated Coefficient (1) South College High school Age Age2 White Married Dock Dockdreg Dreg Unemployemnt Intercept

0.4182341 0.0697776 0.0244639 0.0002799 0.1166035 0.0189412 0.3755514 0.1556468 0.1662788 0.0205515 1.434612

t-statistic (2)

9.53 1.91 2.63 2.52 2.90 0.47 2.62 0.91 4.62 1.56 6.98

Number of observation ¼ 2729 R2 ¼ 0.0839 Adjusted R2 ¼ 0.0806 Northeast College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.4365906 0.0994658 0.0122877 0.0001387 0.1053585 0.0940459 0.0873943 0.3750196 0.0961015 0.0038684 1.685759

0.1752481 0.1147181 0.0206617 0.0002437 0.0075973 0.1166573 0.2157461

0.2170281 0.1071847 0.0191276 0.0001694 0.1007447 0.0655673 0.205112 0.203889 0.0828546 0.0079889 1.550112

t-statistic (4)

8.26 5.09 4.08 3.01 4.57 3.26 3.13 2.37 2.94 1.11 14.95

Number of observations ¼ 2017 R2 ¼ 0.0994 Adjusted R2 ¼ 0.0949

10.91 3.19 1.50 1.46 2.41 2.79 0.67 2.24 3.17 0.34 8.99

Number of observations ¼ 3138 R2 ¼ 0.0854 Adjusted R2 ¼ 0.0824 West College High school Age Age2 White Married Dock

Estimated Coefficient (3)

0.2253897 0.2253897 0.0228368 0.0002369 0.0699979 0.0931666 0.0967316 0.3145482 0.0727934 0.0206544 0.0206544

8.92 8.92 5.07 4.50 2.90 5.10 1.55 3.78 2.83 3.19 3.19

Number of observations ¼ 2193 R2 ¼ 0.1020 Adjusted R2 ¼ 0.0979 3.39 2.34 1.73 1.72 0.14 2.33 1.56

2910669 0.2131856 0.0275331 0.0002605 0.0574513 0.0862586 0.16249180

10.31 8.31 5.31 4.13 2.27 4.08 2.60

240

JAMES PEOPLES ET AL.

Table 5 (Continued ) Estimated Coefficient (1) Dockdreg Dreg Unemployment Intercept

0.4238906 0.2534289 0.011321 1.76022

Number of observations ¼ 1686 R2 ¼ 0.0803 Adjusted R2 ¼ 0.0748

t-Statistic (2) 2.62 5.60 0.72 6.75

Estimated Coefficient (3)

t-Statistic (4)

0.2475324 0.2115208 0.0067353 1.371865

2.97 6.60 0.79 11.29

Number of observations ¼ 1575 R2 ¼ 0.1674 Adjusted R2 ¼ 0.1621

a

Results from estimating the wage equation for the regulation/1984 Act deregulation sample (years 1973–1998, excluding 1982) are presented in columns 1 and 2; Dreg ¼ 1 for 1984 and after and zero otherwise. Results from estimating the wage equation for the regulation/1998 Act deregulation sample (years 1973–1983 and 1999–2001, excluding 1982) are presented in columns 3 and 4; Dreg ¼ 1 for 1999 and after and zero otherwise.

between the lowest (25th quantile) and highest (75th quantile) wage groups. Findings on the sum of the estimated coefficients Dock and Dockdereg uncovers an interesting post-deregulation wage premium pattern by quantile group. Post-deregulation dockworker premiums for low-wage workers are markedly lower than premiums for higher wage levels. For instance, the sum of these estimated coefficients are 0.34 and 0.31 for low-wage workers following the 1984 and 1998 acts, respectively. The sum of these coefficient estimates are 0.42 and 0.39 for the 50th and 75th quantiles following the 1984 act and are 0.48 and 0.46 for these quantile groups, respectively following the 1998 Act. Wage findings on dockworkers residing in the Northeast region of the U.S. are presented in Table 7. The estimated coefficient on the dockworker dummy suggests a significant pre-deregulation wage advantage for dockworkers. Pre-deregulation premiums were lowest for high-wage dockworkers (75th quantile). Such a finding suggests that lower-skilled union dockworkers (25th and 50th quantile) were the biggest beneficiaries of shipping regulation in this region. In addition, low-wage dockworkers (25th quantile) maintained their relatively high-wage advantage over nontransportation operatives following initial deregulation, as the estimated coefficient on the Dockdereg interaction term in column (1) suggests that the wage premium for low-wage dockworkers is larger than that of workers receiving wages at the 50th and 75th quantile level. The post-deregulation wage pattern for different quantile groups changes dramatically following passage of the 1998 Act. For the

Shipping Deregulation’s Wage Effect

241

Table 6. Regional Quantile Hourly (LOG) Wage Results for Union Dockworkers Versus Union NonTransportation Operativesa. Estimated Coefficient (1)

t-statistic (2)

Estimated Coefficient (3)

t-statistic (4)

South 25th Quantile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.0386612 0.0725144 0.0258872 0.0002833 0.1437369 0.0881707 0.3016398 0.0410212 0.1036537 0.0075923 1.120121

1.28 2.85 2.74 2.37 4.23 2.65 2.93 0.23 2.97 0.80 5.65

0.2271 0.0962 0.0249 0.0024 0.1371 0.0875 0.1921 0.1329 0.1303 0.0196 1.253

5.12 2.34 5.46 4.88 4.52 2.58 1.67 0.88 2.55 2.15 12.03

50th Quantile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.1241745 0.1010819 0.0316212 0.0003475 0.1480099 0.0776229 0.2994652 0.1204169 0.0736127 0.0065213 1.236248

4.60 4.77 5.33 4.74 5.36 2.90 3.10 1.17 2.98 0.81 8.12

0.3046 0.1428 0.0187 0.00014 0.1261 0.0818 0.2031 0.2764 0.1024 0.0014 1.427

10.10 5.05 3.13 2.07 5.90 2.31 2.82 2.16 2.05 0.11 11.61

75th Quantile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.1093983 0.083671 0.0287908 0.0003027 0.1063798 0.0507227 0.2422403 0.1462923 0.0460119 0.0004136 1.606161

6.77 5.99 7.40 5.94 4.47 2.41 3.15 1.44 2.04 0.07 17.13

0.1425 0.0680 0.0155 0.00012 0.0824 0.0407 0.2283 0.2396 0.0025 0.0012 1.855

3.69 1.92 2.99 1.89 3.98 1.82 3.96 2.66 0.09 0.19 19.85

Number of observations ¼ 2729 0.25 Pseudo R2 ¼ 0.0243 0.50 Pseudo R2 ¼ 0.0338 0.75 Pseudo R2 ¼ 0.0376 a

Number of observations ¼ 2017 0.25 Pseudo R2 ¼ 0.0530 0.50 Pseudo R2 ¼ 0.0616 0.75 Pseudo R2 ¼ 0.0528

Results from estimating the wage equation for the regulation/1984 Act deregulation sample (years 1973–1998, excluding 1982) are presented in columns 1 and 2; Dreg ¼ 1 for 1984 and after and zero otherwise. Results from estimating the wage equation for the regulation/1998 Act deregulation sample (years 1973–1983 and 1999–2001, excluding 1982) are presented in columns 3 and 4; Dreg ¼ 1 for 1999 and after and zero otherwise.

242

JAMES PEOPLES ET AL.

Table 7. Regional Quantile Hourly (LOG) Wage Results for Union Dockworkers Versus Union NonTransportation Operativesa. Estimated Coefficient (1)

t-statistic (2)

Estimated Coefficient (3)

t-statistic (4)

Northeast 25th Quantile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.0130159 0.1401564 0.028216 0.0003184 0.177044 0.122067 0.1656863 0.2872097 0.0238305 0.0087977 1.084097

0.32 7.35 5.66 5.44 4.70 4.68 1.67 2.37 1.28 1.27 9.39

0.2748 0.2029 0.0208 0.00021 0.1097 0.1287 0.2224 0.1265 0.1101 0.0187 1.262

6.75 6.37 4.32 3.56 3.19 5.64 1.85 1.01 2.97 1.69 9.47

50th Quantile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.1229078 0.1346015 0.0213513 0.0002437 0.1369584 0.1198551 0.2113491 0.2181292 0.0242275 0.0158591 1.540884

6.41 6.21 3.65 3.64 6.18 0.79 2.52 2.19 1.45 2.38 10.47

0.2619 0.1737 0.0252 0.00026 0.0808 0.1167 0.1747 0.2378 0.0742 0.0172 1.440

7.74 8.74 5.66 5.09 2.45 4.95 2.11 2.01 2.94 2.77 11.44

75th Quantile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.0838901 0.0732597 0.0236936 0.000271 0.1286326 0.1128967 0.0972076 0.215469 0.0063259 0.0158735 1.75004

3.01 3.94 6.82 6.48 4.02 6.95 1.66 2.91 0.39 2.93 26.49

0.1829 0.1056 0.0198 0.00021 0.01609 0.1036 0.1210 0.3332 0.0439 0.0089 1.839

4.48 4.72 3.29 3.19 0.40 4.23 2.57 2.54 1.24 1.52 11.93

Number of observations ¼ 3138 0.25 Pseudo R2 ¼ 0.0364 0.50 Pseudo R2 ¼ 0.0370 0.75 Pseudo R2 ¼ 0.0392 a

Number of observations ¼ 2193 0.25 Pseudo R2 ¼ 0.0615 0.50 Pseudo R2 ¼ 0.0622 0.75 Pseudo R2 ¼ 0.0600

Results from estimating the wage equation for the regulation/1984 Act deregulation sample (years 1973–1998, excluding 1982) are presented in columns 1 and 2; Dreg ¼ 1 for 1984 and after and zero otherwise. Results from estimating the wage equation for the regulation/1998 Act deregulation sample (years 1973–1983 and 1999–2001, excluding 1982) are presented in columns 3 and 4; Dreg ¼ 1 for 1999 and after and zero otherwise.

Shipping Deregulation’s Wage Effect

243

Table 8. Regional Quantile Hourly (LOG) Wage Results for Union Dockworkers Versus Union NonTransportation Operativesa. Estimated Coefficient (1)

t-statistic (2)

Estimated Coefficient (3)

t-statistic (4)

West 25th Percentile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.1879529 0.1826382 0.0378534 0.000409 0.0468225 0.1427719 0.1627997 0.3832945 0.1920233 0.0035534 1.024648

3.96 4.69 3.83 3.27 1.11 3.20 2.03 3.28 4.43 0.38 4.50

0.3247 0.2395 0.0282 0.00027 0.1026 0.1013 0.1842 0.2045 0.22364 0.0124 1.0338

10.90 5.62 3.86 3.13 2.35 2.98 1.94 1.64 5.62 1.10 6.65

50th Percentile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.1245331 0.1211024 0.0373931 0.0004059 0.0247258 0.0731967 0.1854087 0.2575997 0.1136942 0.0028367 1.411603

2.88 0.31 4.47 4.14 0.97 1.60 2.56 2.91 5.04 0.40 7.20

0.2982 0.2119 0.0312 0.00029 0.0096 0.0958 0.1572 0.2594 0.2400 0.0083 1.341

9.69 6.07 3.30 2.73 0.26 2.88 2.80 3.01 5.23 0.69 7.26

75th Percentile College High school Age Age2 White Married Dock Dockdreg Dreg Unemployment Intercept

0.1768717 0.1609091 0.0148385 0.0001359 0.0415693 0.0597636 0.1889678 0.2419983 0.0569363 0.0125125 2.079436

7.32 7.94 2.38 1.73 1.58 1.95 2.89 2.97 3.11 1.63 14.71

0.2588 0.1851 0.0211 0.00019 0.0273 0.0635 0.1162 0.3188 0.173 0.0051 1.869

9.18 6.96 3.06 2.35 0.95 2.82 1.56 2.97 5.24 0.60 11.55

Number of observations ¼ 1686 0.25 Pseudo R2 ¼ 0.0603 0.50 Pseudo R2 ¼ 0.0573 0.75 Pseudo R2 ¼ 0.0726 a

Number of observations ¼ 1575 0.25 Pseudo R2 ¼ 0.1079 0.50 Pseudo R2 ¼ 0.0947 0.75 Pseudo R2 ¼ 0.0672

Results from estimating the wage equation for the regulation/1984 Act deregulation sample (years 1973–98, excluding 1982) are presented in columns 1 and 2; Dreg ¼ 1 for 1984 and after and zero otherwise. Results from estimating the wage equation for the regulation/1998 Act deregulation sample (years 1973–1983 and 1999–2001, excluding 1982) are presented in columns 3 and 4; Dreg ¼ 1 for 1999 and after and zero otherwise.

244

JAMES PEOPLES ET AL.

1999—2001 observation period, high-wage dockworkers residing in the Northeast region of the U.S. receive the largest wage premium. Table 8 presents the wage findings for dockworkers residing in the Western region of the U.S. The estimated coefficient on the dockworker dummy suggests dockworkers received a statistically significant wage premium prior to deregulation. The difference in this pre-deregulation premium is largest between the lowest (0.25th quantile) and highest (0.75th quantile) wage groups. For the Western region sample, dockworkers’ wage advantage over workers in other industries grew following both deregulation acts. However, a closer examination of this estimated coefficient reveals contrast wage patterns by quantile groups. The dockworker wage advantage is largest for lowwage workers following passage of the 1984 Act, whereas the advantage is largest for high-wage dockworkers following passage of the 1998 Act.

6. SUMMARY AND CONCLUDING REMARKS Shipping deregulation contributed to lower shipping rates and greater choice of ports of calls. An unintended consequence of this policy has been the ability of dockworker unions to take advantage of greater demand for their members to negotiate nontrivial wage gains following initial deregulation in 1984. Evidence from collective bargaining settlements on regional labor markets for dockworkers suggests that the ability of unions to negotiate relatively high wages is heavily constrained in the Southern Gulf Coast due in part to competition from nonunion LCDL. Evidence on negotiation agreements for low-skill dockworkers suggests that union negotiation strength, especially in the U.S. East and West Coasts, created a significant wage advantage for these workers, which is unparallel for lowskill workers in other industries. More recent negotiations indicate that difficulty employing sufficient amounts of skilled dockworkers helped contribute to this group of workers receiving significant wage gains following passage of the 1998 Act. Quantile regression findings from this study show that dockworkers historically received significant wage premiums over workers in other industries, and this wage advantage increased following deregulation. Excluding wage findings for dockworkers residing in the U.S. South, low-wage dockworkers enjoyed a substantially larger-wage premium increase (relative to nontransportation operatives) compared to high-wage dockworkers following initial deregulation in 1984. Wage premiums do not change significantly for low-wage dockworkers residing in the U.S. South following passage of

Shipping Deregulation’s Wage Effect

245

the 1984 Act. Findings comparing post-1998 deregulation wages to pre-1984 deregulation wages reveal that for all regions high-wage dockworkers received larger premium increases compared to the premium gains for lowwage dockworkers. This study’s post-1984 and pre-1998 wage findings support its hypothesis on regional union negotiation strength, which suggest that unparallel bargaining strength following the 1984 act created a business environment that promotes a significant wage advantage for low-wage dockworkers residing in the West and Northeast U.S. In contrast, low-cost nonunion competition in the South placed downward pressure on the wages of dockworkers in this region making it difficult for low-skilled dockworkers to enhance their postderegulation wage advantage. The post-1998 Act wage findings of this study support the hypothesis that challenges in employing high-skilled dockworkers in the late 1990s creates a business environment that pays relatively high wages to this group of workers across all regions. Even though the wage findings presented in this study suggest nontrivial post-deregulation wage premiums for dockworkers, these wages may be economically justified, at least for the group of high-wage workers. Significant productivity gains for highly skilled dockworkers (e.g., crane operators) contributed to labor costs (as a percent of total port costs) declining for these workers despite their large wage gains. A stronger argument for unjustifiably high wages could be offered for lower-wage occupations. For instance, the union’s whipsawing advantage contributed heavily to the nontrivial post-deregulation wage premium of low-wage dockworkers, given that low-skill content occupations (e.g., ILWU marine clerks) benefited from union negotiated workdays that increased wage rates after a certain number of hours of work. More recent negotiations that enhance dockworker skill content indicate a shift to enhanced job responsibilities that are more consistent with wage rates paid to this dockworker occupation. Such compensation for greater responsibility is consistent with economic theory on wage determination in increasingly competitive markets.

NOTES 1. The Transportation Act of 1940 placed inland, other than Great Lakes, waterway transportation under the jurisdiction of the ICC. 2. Liner conferences have immunity from anti-trust legislation in most OECD (Organization for Economic Cooperation and Development) countries. 3. By early 1920, the U.S. government had built or acquired 1,750 merchant ships (Frankel, 1986).

246

JAMES PEOPLES ET AL.

4. Discussion of the Shipping Act of 1984 is found in Frankel (1986), Chadwin et al. (1990), and Cassavant and Wilson (1991). 5. Within 18 months of the effective date of the Act, service-contract cargoes grew from a negligible amount to over 42 percent, with some shipping trade lanes having over 60 percent of their cargoes under service contracts (Frankel, 1986). 6. For cargo moving under a port-to-port rate, the shipper (or authorized party) is responsible for hiring inland carriers to transport cargo to and from ports. Under a door-to-door rate, the shipping line has this responsibility. 7. Some authors have distinguished among landbridging, minibridging, and microbridging: landbridging referring to cargo movement that crosses a body of land between two ocean legs; minibridging referring to cargo movement that crosses one ocean by ship and then crosses a body of land but ends at a port on another ocean; and microbridging referring to cargo movement that crosses one ocean by ship and then proceeds by rail to an inland location (Chadwin et al., 1990). For this paper, landbridging is an all encompassing term, capturing all three of these possibilities. 8. The PMA is the employers’ organization with whom the ILWU negotiates West Coast dockworker agreements. 9. For further discussion of ocean container transportation and its impacts, see Talley (2000). 10. In 1987, the FMC suspended RCs, ruling that RCs arbitrarily restrained trade under the Shipping Act of 1984; the ruling was upheld by a decision of the 1988 Federal Court of Appeals. 11. In response to the lower-cost dockworker labor in Gulf Coast ports, the ILA has agreed to a number of concessions at these ports: a lower wage scale for breakbulk and bulk cargoes, flexible starting times, a reduced entry-level wage, and a dedicated work force to specific docks. Also, multitiered wage structures were included in the 1996 ILA master contract to make ILA labor more wage competitive with non-ILA labor. 12. Lower-cost Teamster dockworkers have also threatened ILA jobs. In March 2000, the ILA and the Teamsters agreed to put aside past jurisdictional disputes and to support each other’s organizing efforts. From the perspective of the Teamsters, the agreement is intended to obtain ILA support for its bid to organize port owner– operator truck drivers. From the perspective of the ILA, the agreement is intended to prevent shipping lines and ports from playing one union against the other in contract negotiations. 13. In the United Kingdom, dock jobs fell from 80,000 in 1967 to 11,400 in 1986 (Chadwin et al., 1990). Even in recent years, significant losses have occurred. In the United Kingdom, port employment declined by 44 percent between 1989 and 1992. In France work rule reforms, introduced in 1992, led to employment declines of up to 66 percent in six major ports. In Australia, waterfront reforms introduced in 1989 led to a 42 percent two-year reduction in stevedore labor. 14. The growth in container cargo was also stimulated by globalization and reduction in trade barriers. 15. The ILA negotiates coast-wide master and local contracts with individual ports, stevedores, and marine-terminal operators that directly employ them but also compete with each other for cargo. Alternatively, the ILWU negotiates with one coast-wide employer association, the PMA.

Shipping Deregulation’s Wage Effect

247

16. The substitution of capital for labor under containerization increased (reduced) capital’s (labor’s) share of port costs. Bulk and break-bulk cargoes are more labor intensive than container cargo; their labor costs represent a higher percentage of port costs than those of container ports. Thus, U.S. Gulf Coast ports that specialize in bulk and break-bulk cargoes have had an incentive to switch to non-ILA labor. 17. In its 1996 and 1999 contract negotiations, the ILWU successfully leveraged its slowdowns to obtain concessions from employers. 18. ‘‘The West Coast docks are crucial to the commerce of the nation, handling 52 percent of all waterborne imports to the United States – equal to about 7 percent of the nation’s gross domestic product’’ (Swoboda, 2002, p. E01). 19. Some evidence of this fear may be found in a recent reaction by the ILWU to a change in hiring at the ports of Los Angeles and Long Beach. On September 8, 2003, the ports of Los Angeles and Long Beach ordered one fewer dockworker to serve as ground workers for each crane. Two days of ILWU slowdowns followed, but were abruptly ended by an arbitration ruling against the union; no subsequent ILWU slowdowns in response to this change in hiring have occurred. 20. The data set excludes information from 1982, because information on the union status of individual workers is not provided for that year. 21. The potential for duplicate entries exist when using CPS outgoing rotation files, because individuals are surveyed when they initially report information and then eight to twelve months after the initial interview. However, limiting the sample to workers reporting hourly wages avoids this problem because hourly wages are only reported once during the survey per individual. 22. The CPS codes hourly wages such that individuals receiving wages above $99.99 an hour are designated a wage value of $99.99. Hence, estimation inaccuracies arise if a large proportion of the sample receives wages above the top-coding. This data shortcoming is not an issue for this study since the maximum wage reported is $72.12. 23. Longshore equipment operators (census code 845) and stevedores (census code 876) denote dockworker occupations. Lathe and turning machine operators (census code 704), welders and cutters (census code 783), and production inspectors and examiners (census code 796) denote the group of nontransportation operatives used in this study. Wage information on this set of workers is commonly used as the benchmark comparison for transportation workers (Hirsch, 1988; Talley, 2002). 24. Mean wage differentials are calculated by taking the differential of dockworker and nontransportation operative mean wages and dividing by the mean wage of transportation operatives. 25. The small sample of dockworkers residing in the Midwest prohibits the estimation of the wage equation for that region. 26. Pooling the sample population across dockworkers and nontransportation workers allow for directly estimating dockworker’s wage differential. However, the possibility of worker heterogeneity may distort the wage results when using pooled data. For instance, compared to nontransportation operatives, dockworkers may receive higher returns on their observed attributes as compensation for working in a more risky work environment. Indeed, an F-test comparing the restricted regression with separate regressions for dockworkers and nontransportation operatives gives an F-statistic distributed as F (11, 14,629) ¼ 13.81.

248

JAMES PEOPLES ET AL.

27. West Coast dockworkers also benefit from the width restriction of the Panama Canal that prevents passage of larger containerships through the canal. Freight on larger containerships destined for the U.S. East coast from the Pacific rim are likely to choose a West Coast port of call and then use rail to transport cargo to the East Coast. 28. A positive value on the sum of these coefficients would support the wagebargaining patterns depicting successful wage negotiation for dockworkers following deregulation. Such gains may be due to a firm’s enhanced ability to pay during a period of increased U.S. demand for foreign product. Inclusion of industry performance measures might capture this ability. However, the emphasis of this study is on the distribution of wage gains following deregulation. Examining the contribution of industry performance to high dockworker wages, though, represents a path for future research. 29. Percentage wage differentials are computed using the following formula [exponential (b) 1]  100. 30. The quantile regression method provides an approach for estimating the wage equation across the distribution of wage levels while allowing covariates to differ by quantile group. Hence, wage results are not as sensitive to sample population outliers that could significantly affect estimation results when using OLS. The quantile approach also takes advantage of information from the entire sample population when estimating wage patterns. See Koenker and Basset (1978) for more on the quantile regression approach. 31. Standard errors are derived using the design matrix bootstrap approach to inference testing (Buchinsky, 1995). The choice of approach toward drawing inferences is more efficient in small samples and is robust to dependence between regressors and the regression errors (Barnes & Hughes, 2002). A pseudo R2 is used to denote a measure of goodness of fit for the quantile model. The pseudo R2 is derived by subtracting the ratio of the minimum sum of weighted deviations and raw sum of weighted deviations from one. This measure though is only an approximation of the model’s goodness and fit and may provide inaccurate results.

ACKNOWLEDGMENT The authors are grateful for comments and suggestions provided by Scott Adams and two anonymous referees.

REFERENCES Armbruster, W. (2000). Four more years: ILA, management extend master contract into 2004. JOC Week, July 3–9, 17–18. Barnes, M., & Hughes, A. (2002). A quantile regression analysis of the cross section of stock market returns. Working paper. Betcherman, G., & Rebne, D. (1987). Technology and control of the labor process: Fifty years of longshoring on the U.S. West Coast. In: D. B. Cornfield (Ed.), Workers, managers, and technological change: Emerging patterns of labor relations (pp. 73–89). New York: Plenum Press.

Shipping Deregulation’s Wage Effect

249

Buchinsky, M. (1995). Estimating the asymmetric covariance matrix for quantile regression models: A Monte Carlo study. Journal of Econometrics, 68, 303–338. Cassavant, K. L., & Wilson, W. W. (1991). Shipper perspectives of the shipping act of 1984. Transportation Quarterly, 45, 109–120. Chadwin, M. L., Pope, J. A., & Talley, W. K. (1990). Ocean container transportation: An operational perspective. New York: Taylor & Francis. Frankel, E. G. (1986). Economic and commercial implications of the U.S. shipping act of 1984. Logistics and Transportation Review, 22, 99–114. Greenwald, R. A. (2004). Working the docks: Labor, management and the new waterfront. Review of Business, 25, 16–22. Hirsch, B. (1988). Trucking regulation, unionization and labor earnings 1973–1985. Journal of Human Resources, 23, 296–319. Hirsch, B., & Macpherson, D. A. (1998). Earnings and employment in trucking: Deregulating a naturally competitive industry. In: J. Peoples (Ed.), Regulatory reform and labor markets (pp. 61–112). Boston, MA: Kluwer Academic Publishers. Killingsworth, C. C. (1962). The modernization of West Coast Longshore work rules. Industrial and Labor Relations Review, 15, 295–306. Koenker, R., & Basset, G. (1978). Regression quantiles. Econometrica, 46, 33–50. Leach, P. (2005). Horse trading. Journal of Commerce, February 28, p. 1. Lewis, I., & Vellenga, D. B. (2000). Ocean shipping reform act. Transportation Journal, 39, 27–34. Locklin, D. P. (1972). Economics of transportation. Homewood, IL: Richard D. Irwin, Inc. Mongelluzzo, B. (1996). Union steered contract talks. Journal of Commerce, July 22, pp. 1B, 4B. Mongelluzzo, B. (1999). Longshoremen reject system to automate job dispatching. Journal of Commerce, December 14, pp. 1, 4. Mongelluzzo, B. (2003a). Productivity at a price: ILWU–PMA contract uses financial incentives to attack shortages of skilled equipment operators. Journal of Commerce, February 3–9, pp. 36–37. Mongelluzzo, B. (2003b). PMA’s new face: Miniace’s team will be assertive in dealing with the ILWU. Journal of Commerce, October 13–19, 26–27. Swoboda, F. (2002). On the waterfront, it’s union vs. tech. Washington Post, January 5, p. E01. Talley, W. K. (2000). Ocean container shipping: Impacts of a technological improvement. Journal of Economic Issues, 34, 933–948. Talley, W. K. (2002). Dockworker earnings, containerization and shipping deregulation. Journal of Transport Economics and Policy, 36, 447–467. Talley, W. K. (2004). Wage differentials of intermodal transportation carriers and ports: Deregulation versus regulation. In: W. K. Talley (Ed.), The industrial organization of shipping and ports. A special issue of the journal The Review of Network Economics, 3, 207–227. Talley, W. K., & Schwarz-Miller, A. V. (1998). Railroad deregulation and union labor earnings. In: J. Peoples (Ed.), Regulatory reform and labor markets (pp. 125–153). Boston, MA: Kluwer Academic Publishers. Tirschwell, P. (2002). Looking forward. Journal of Commerce, October 21–27, p. 6. Tirschwell, P. (2003). Has the ILWU been tamed? Journal of Commerce, October 29–November 2, p. 6.

This page intentionally left blank

250