[david banister, joseph berechman] transport inves(bookfi.org)

Transport Investment andEconomic Development

A major concern of all decision makers has been to ensure that there areclear benefits from transport investment proposals. The travel time savingsare clear, but the wider economic developments have presented enormousdifficulty in terms of both theoretical arguments and empirical evidence.This book reviews the history of the debate and argues that the agenda haschanged. Concerns about economic development need to be placed in thenew economy and a much wider social and environmental context. Theseissues are presented together with a major analytical investigation of macroeconomic models, evaluation in transport and micro economic approaches.The final part of the book presents a series of case studies for road, rail andairport investment schemes, particularly focusing on the economicdevelopment aspects.

This book makes a major contribution to the debate and is directed atresearchers, decision makers and students who are interested in the widereconomic development impacts of transport investment decisions.

David Banister is Professor of Transport Planning at University CollegeLondon. He has produced sixteen books on all aspects of transport policyand planning analysis.Joseph Berechman is a Professor and Chairman of the Public PolicyDepartment at Tel Aviv University, Israel. His major academic interestsinclude Transportation Economics, Transportation and Land Use SystemsPlanning and Policy Analysis. He has published extensively in these fieldsand was a faculty member and visiting scholar in a number of Americanand European universities and research institutes. Presently he is also aSenior Research Associate at the University Transportation Research Centerin New York.

Also available:

Towards an Urban RenaissanceUrban Task Force1–851121–65–X

Transport Policy and the EnvironmentEdited by David Banister0–419–23140–4

Transport and Urban DevelopmentEdited by David Banister0–419–20390–7

Transport Planning: An International PerspectiveDavid Banister0–419–18930–0

Transport, the Environment and Sustainable DevelopmentDavid Banister and Kenneth Button0–419–17870–8

Transport Investment andEconomic Development

David BanisterandJoseph Berechman

First published 2000 in the UK and the USAby UCL Press11 New Fetter Lane, London EC4P 4EE

The name of University College London (UCL) is a registeredtrade mark used by UCL Press with the consent of the owner.

This edition published in the Taylor & Francis e-Library, 2003.

UCL Press is an imprint of the Taylor & Francis Group

© 2000 David Banister and Joseph Berechman

The right of David Banister and Joseph Berechman to beidentified as the Authors of this work has been asserted by themin accordance with the Copyright, Designs and Patents Act 1988

All rights reserved. No part of this book may be reprinted orreproduced or utilized in any form or by any electronic,mechanical, or other means, now known or hereafterinvented, including photocopying and recording, or in anyinformation storage or retrieval system, without permission inwriting from the publishers.

The publisher makes no representation, express or implied, withregard to the accuracy of the information contained in thisbook and cannot accept any legal responsibility or liability forany errors or omissions that may be made.

British Library Cataloguing in Publication DataA catalogue record for this book is availablefrom the British Library

Library of Congress Cataloging in Publication Data

A catalogue record for this book has been requested

ISBN 0-203-22087-0 Master e-book ISBN

ISBN 0-203-27560-8 (Adobe eReader Format)ISBN 0419-25590-7 (hbk)ISBN 0419-25600-8 (pbk)

This book is dedicated to my late father, JacobBerechman, whose love for learning has guided me

throughout my life.

The cathedrals would never have been built if people had consideredonly the short term. It is the difference between a society that plantsconifers and poplars and one that plants oak trees.

Bertrand de Jouvenel

The sovereign has the duty of erecting and maintaining certain publicworks and certain public institutions, which it can never be for the interestof any individual, or small number of individuals, to erect and maintainbecause the profit could never repay the expense to any individual orsmall number of individuals though it may frequently do much morethan repay it to a great society.

Adam Smith, Wealth of Nations (1967 edition)

Only if overall fixed-asset investment (e.g., highways, bridges and powergrid) grows by 15 to 18 percent, can we reach 8 percent economic growth.

Zeng Peiyan, minister in charge of the State Development PlanningCommission, The New York Times (‘China Plans to Spend $1 Trillion

on Big Public Projects’), 24 September 1998

Contents

List of figures ixAcknowledgements xi

PART I

Objectives and scope 1

1 Background and objectives 3

2 Scope of analysis: definitions, approach andmethodological framework 33

PART II

Contemporary issues 57

3 Transport infrastructure investment 59

4 The evolving economy 83

5 Social, spatial and environmental effects 107

PART III

Methodology: analytical approaches and modelling 127

6 Modelling the growth effects of transport capitalinvestments: a macro level analysis 131

7 Economic evaluation of transportation projects 161

8 A model of transport infrastructure development and localeconomic growth 211

Contentsviii

PART IV

Empirical case studies 237

9 The economic impacts of roads 239

10 The economic impacts of rail 257

11 The economic impacts of airports 287

12 Interpretation of impacts and policy conclusions 317

References 335Index 363

List of figures

1.1 Links between GDP and infrastructure stock, 1990. 202.1 The basic causality paradigm of the relationships between

transport infrastructure investment and economicdevelopment. 41

Part III The complementarity of approaches. 1287.1 Traditional view of the effects of transportation

infrastructure investment. 1647.2 The new scheme for the evaluation of economic growth

benefits from transport investment. 173A7.1.1 Measurement of benefits. 203

8.1 Schematic view of the relationships between theproduction, household and transportation sectors. 224

8.2 Equilibrium solutions before and after increase intransportation infrastructure capacity. 228

8.3 Equilibrium labour in the economy as a function oftransportation infrastructure capacity. 229

9.1 The M25 London orbital motorway. 2419.2 Map of the A71 route in France. 2499.3 Map of the Amsterdam agglomeration. 252

10.1 Methodological framework of LRRT analysis. 26010.2 The Buffalo rail transit and development patterns in the

metropolitan area. 26410.3 The Japanese Shinkansen high-speed rail system. 27910.4 The French TGV high-speed rail system. 28110.5 The San Franscisco BART transit system. 28411.1 Employment impact model. 29112.1 Illustration of the necessary sets of conditions. 31912.2 Transport and economic development at the regional level. 32512.3 The role of policymaking in achieving economic growth. 333

Acknowledgements

The idea for this book arose from our common concern over the lack ofresearch and understanding of the links between transport investment andeconomic development. It seems a long way from those early discussions atthe Tinbergen Institute in 1994 to this completed text. We now understandwhy no other author(s) has really tried to tackle the complexity of theselinks. We would like to thank the staff and our colleagues at the TinbergenInstitute in Amsterdam for their continuous support and understanding inaccommodating us at Kaisergracht, in particular Elfie Bonke and MarianDuppon, but also many others. Prof. Peter Nijkamp (Department ofEconomics, The Free University of Amsterdam), who managed to lure bothof us to the Tinbergen at the same time as VSB Visiting Professors (1994–7),facilitated the whole enterprise. This opportunity to work together more thananything provided the necessary space, time and support to really initiate thelong discussions and debates, now reflected in the book. Jody Kersten fromthe Free University was also instrumental in getting us to work together byorganizing the local arrangements in Amsterdam.

Several of our academic colleagues helped with particular chapters: Prof.Robert (Buzz) Paaswell (Director, University Transportation Research Center,The City College, New York) on the Buffalo Case Study (Chapter 10); AlanMcLellan (University College London) on the Economic Development Effectsof Airports (Chapter 11); Prof. José Hulguin-Veras (University TransportationResearch Center, the City College, New York), on the Benefit Cost Analysis(Chapter 7); Prof. Jan Brueckner, (Department of Economics, University ofIllinois) on the microeconomic modelling (Chapter 8). Many others havecommented on different chapters, or on particular sections of the book thathave been present at conferences and workshops and have at various stagesencouraged us to continue with the writing of the book. Our thanks arewarmly extended to all these colleagues.

David Banister and Joseph BerechmanJune 1999

London

Objectives and scope

Main issues and structure

It has always been assumed that a high quality transport infrastructure is anessential prerequisite for economic development, yet this assumption has neverreally been investigated in depth. Over time it has become axiomatic andpart of the underlying rationale for transport investment decisions. The basicpurpose of this book is to explore in depth the arguments for and against thisassumption and to come to conclusions on the key relationships betweentransport and infrastructure investment and economic development.

We also accept that there are other important influences on economicdevelopment. In particular, we argue that political and institutional factorsprovide the broader context within which decisions are made. They alsoinfluence the means by which finance for investment can be raised and providethe organizational and legal framework for action. In this book we primarilyconcentrate on the links between transport investment and economicdevelopment, but at the end we return to the political and organizationalcontext through the case studies and concentrate on this in the concludingchapter.

The first part of the book outlines the main issues and arguments in thedebate, taking a series of different perspectives. First, we examine the historicalevidence, then the extensive literature on the productivity of infrastructureinvestments and the links between infrastructure and development. Most ofthe book is addressed at infrastructure investment in developed economies,but the very different impacts in developing countries and cities are coveredin the overview. We also outline the importance of the subject matter and aseries of key questions to be addressed in any analysis of transportinfrastructure investment and economic development.

The range of the subject matter is vast and of key concern to all nationaland local governments. To limit the scope of this book we outline a conceptualapproach in Chapter 2, together with a methodological framework withinwhich analysis can take place. A principal element in this framework is thescale at which analysis should be undertaken. Most of the recent research

Part I

2 Objectives and scope

has concentrated on the links between national economic indicators andinfrastructure investment, but causality and time are difficult elements todeal with at this scale. The regional analysis focuses mainly on accessibilitychanges within networks rather than economic impacts. Our analysis isprimarily at the microeconomic level where causality can be implied, butthere are still limitations with data.

Part I sets the scene and limits the scope of our investigations. It also actsas the framework within which the other three parts of the book are placed.We then progress on to the changing contemporary world in which decisionsare made (Part II), the review of methods and the development of a microanalytic model (Part III), and the presentation of the case study material(Part IV).

Background and objectives

1.1 Introduction

The belief that public investment in infrastructure will generate economicgrowth has often been used as a justification for the allocation of resourcesto the transport sector. Much of the road-building programme in developedand developing countries has been promoted on these grounds, yet thearguments seem far from clear.

In the USA, for example, the Clinton administration has proposed asubstantial investment in infrastructure, as it has public support (Gillen, 1993),even though Congress is against it because of the implications for the publicbudgets. It is popular among users of the road as they can see a higher qualityinfrastructure that will allow them to maintain their very mobile car-basedlifestyles. Industry has also traditionally been a strong supporter ofinfrastructure investment, arguing that it will make it more competitive, helpget the country out of recession and create jobs in the short term. These areall strong arguments but, as we shall see later in this book, the evidence isequivocal, particularly in countries and cities where there is already a highquality infrastructure.

Similar sentiments were being expressed in the UK with the publication ofRoads for Prosperity (UK Department of Transport 1989). An expandedprogramme of investment in motorways and trunk roads to relieve congestionwas announced. The programme effectively doubled the existing investmentplans and was seen as a commitment to the provision of infrastructure ‘suitedto the single market and other competitive challenges of the 1990s and beyond’(para. 2). It was also argued that the investment was necessary for industryand to improve the country’s economic geography, through increasingopportunities for less favoured regions, assisting urban regeneration andhelping more prosperous areas to cope with growth. The programme hadresulted from a substantial increase in road traffic (+35 per cent) in the 1980sand the prospects of a doubling of road traffic from 1988 to 2025. Somethinghad to be done and the government decided that road building was the mainalternative to be pursued. One of the fundamental objectives of the road

Chapter 1


programme is to assist economic growth by reducing transport costs. Yeteven in the following year when the review of the road programme was made(UK Department of Transport 1990), there was little mention of the economicgrowth arguments, only those on environment, safety and urban regeneration.

Similar arguments have been presented in Europe. A review of theproblems carried out by an international team of experts (Group Transport2000 Plus 1990) concluded that the overcrowding and saturation of newfacilities even at the moment of commission, the need for new links and thechoice of layouts, gauges and operating methods were all issues to beaddressed. But above all the people consulted by the team of experts focusedon infrastructure costs and many of them talked in terms of passing themdirectly to the users. The costs of constructing new infrastructure andreplacing existing infrastructure are considerable and the massiveinvestments in the 1950s and 1960s have been followed by lower levels ofinvestment. Since 1975, investment on inland transport has fallen in westEurope by 20 per cent in real terms and it has halved as a proportion ofgross domestic product (GDP) to 0.8 per cent. This reduction ininfrastructure investment reflected general reductions in public expenditure,the world recession in the 1970s resulting from high oil prices and thegenerally lower levels of increase in transport demand. Non-investment intransport infrastructure takes time to show an effect and, given the short-term time horizons of politicians, any delay in commitments to expensiveprojects meant savings in public budgets and lower taxes. Investmentdecisions were delayed, particularly expensive new links between countriesand those that involved tunnelling. With the economic upturn in the 1980sthere was substantial new growth in transport demand, but it also becameapparent that growth in traffic had continued throughout the 1970s aswell. Investment had not been reduced because of reductions in demand formobility, but for other macroeconomic reasons such as pressures on publicbudgets, high interest rates and industrial recession.

Underlying these arguments is the premise that there is a fundamental linkbetween growth in transport investment and economic growth. It has beenconsistently argued that there is a clear relationship over time between GDPgrowth (a measure of economic growth) and a range of measures of transportand transport-related investments. Such comparisons have generally beenmade over the last fifty years since the advent of the car, and more recentlyair travel. It has also been a period of economic growth and stability, as therehave been no major wars to reduce or redirect output. Similarly, there hasbeen a substantial increase in trade within and between countries. It is notsurprising that the demand for travel has increased in parallel with economicgrowth. Many other measures of wealth or well-being have also increased ina similar way: for example, the growth in income levels, the purchase ofconsumer goods, the numbers of people in schools and higher or continuingeducation, life expectancy, etc.

Background and objectives 5

However, as efficiency and productivity increase, the linear links with GDPmay be reduced as there is no a priori reason why transport demand shouldrise with GDP. Production and distribution processes (and individual passengertravel) could become either less transport intensive or more transport intensive.Conversely, if prices rise substantially or there is a concerted international action,then again the simple linear relationship may be broken. This has alreadyhappened with energy consumption as price rises and greater efficiency inproduction and consumption have resulted in growth levels far less than thosein GDP. These trends and relationships are important, but they are not set intabloids of stone—they are not immovable. Sustainable growth and developmenthas the basic objective of maintaining growth in the national and internationaleconomies, but with the use of less resources, particularly non-renewableresources. This means that we would expect a continued growth in GDP butwith fewer resources used in transport. This does not necessarily mean thatthere will be less transport. But it does mean that we have to become moreefficient in our use of resources. In addition, the current technological revolutionin information and communications may also weaken the links betweentransport growth and economic growth. The concentration on physical measuresignores the other forms of transactions that take place, such as movement ofinformation, finance, commerce and document handling by informatics andtechnologicalmeans. This is where real substitution is taking place.

Apart from the arguments at the national level, there have also been strongurban and regional arguments for investment in transport infrastructure. Theregional development policies in the European Union (EU) are powerful andsubstantial investment has been transferred from the centre to the periphery.This means that the larger countries of Germany, France and the UK havesupported infrastructure projects in the poorer European Union countries ofSpain, Greece, Portugal and Italy. The arguments used by the EU are thatregional development policy strengthens integration and cohesion of the EUas a whole, while at the same time reducing the disadvantages in peripheralor poorly connected locations. In the longer term it will maintain and enhancethe competitiveness of Europe. In other words, it is asserted that expandedtransport infrastructure will provide overall economic development in thelonger term.

Even here there are many unresolved questions. It is not clear whethersuch a policy actually provides the greatest benefit to these peripheral regions.Little empirical evidence is available on whether infrastructure investment inthe periphery actually strengthens the centre, as it extends market area andpermits migration of labour to the centre where opportunities are perceivedto be greater. It is unclear whether the local economy in the peripheral regionbenefits over the longer term. If competitiveness of the EU or the individualcountry in world markets is being discussed, then infrastructure investmentshould be in those locations where the greatest return is expected. This islikely to be in the regions with the most buoyant economic conditions or


where particular circumstances are likely to result in high returns, for example,where there is a particular skill or a natural resource available.

The issue here is whether a high quality transport infrastructure is anecessary condition to bring about economic growth in depressed or emergingregions. Traditional arguments (e.g. Botham 1980) and more recent reviews(e.g. Hart 1993) have all suggested that road building is not the keydeterminant for growth (Section 1.2). The Merseyside situation in the UK isinformative here. In Liverpool (the main city in the Merseyside conurbation),substantial road programmes were promoted in the 1960s and 1970s as theexpected increases in population and employment, together with risingproductivity and income, would all lead to substantial increases in the volumeof goods and passenger travel, particularly by car. It was clear even beforethe studies were completed that the Merseyside conurbation was losingpopulation and employment and that the whole of the local economy neededto be restructured. Inadequate road networks were not a key component ofthat restructuring process and investment was required in retraining, newindustries and a regeneration of the local economy. Nevertheless, the roadinvestment programme was still kept as an integral part of the strategy, onlythe argument changed. Roads were originally justified on the basis of theexpected growth in traffic, and the necessity to accommodate and direct thisgrowth. Subsequently, the same roads were being defended as a means toregenerate the local economy (Banister 1994).

From a firm’s perspective, similar conclusions can be drawn. There aremany ways in which firms can use the transport system to their own advantageso that costs can be minimized. If a road network is improved, then the firmis likely to make longer and more frequent journeys, which may minimizetheir own costs, but raise substantial environmental costs (McKinnon andWoodburn 1994). Conversely, firms could make more use of logistics systemsand new forms of management and organization to minimize transport costs(Weijers 1995). Similarly, firms might consolidate on one site or disperse toseveral sites to maintain competitiveness. In all cases, there seem to be arange of alternatives available so that profit levels are maintained or increased.Transport costs are only one part of that decision, yet there are many ways inwhich each firm can maintain its competitive position.

The evidence cited here gives a flavour of the main issues and problems tobe raised in this book. At one level the traditionally held view that there is astrong link between transport infrastructure growth and economic growthdoes seems to be supported, particularly if national statistics on trends overthe last fifty years are used. However, this aggregate view simplifies the moreinteresting political and economic arguments for investment, the regionalvariations and the actions of individual firms and people in their own decisions.

Some governments are less convinced by the strengths of these arguments.In Canada there has been a recent recommendation (cited in Gillen 1993)against any large public investment in transport infrastructure. Part of the


justification for this recommendation was the lack of any clear understandingof how such investment would lead to long-term economic growth anddevelopment. The recommendation from the Royal Commission on NationalPassenger Transportation was first to get the pricing of the infrastructureright.

This chapter aims to set the scene for the book by outlining some of themain debates in greater detail. The development of the argument is presentedthrough an historical perspective on the debate, particularly the huge interestin the subject in the 1980s and its current revival. This is followed by areview of the triggers for this renewed interest, principally through the debateon the productivity of infrastructure investments from the macroeconomicliterature. We then turn to the development literature to explore the seeminglyclearer relationships in developing countries. All of these debates have takenplace at the macro level, while most of the remainder of the book concentrateson the regional and local scales.

1.2 The debate: a historical perspective

1.2.1 Introduction

The debate over the links between transport infrastructure investment andeconomic development is not new. Ever since roads and railways were built,one of the main arguments has been the impact that the infrastructure wouldhave on production costs. Initially, as there were few links in the network,the impacts would be clearly identified and causal relationships could beinferred. Transport investment would help open up new areas for agriculturalproduction, create new markets for goods and link in isolated areas with themain towns and cities. Essentially, this is the development argument that hasbeen applied more recently to countries passing through the developmentstage (see Section 1.4). We have no fundamental disagreement with thesearguments. Our contention is to establish whether the same arguments arestill relevant in advanced economies where the infrastructure is already welldeveloped, where more complex market systems are in operation and wheretransport costs play a less important role in the total production costs. Weare also addressing the new forms of production based on post-industrialand technological developments, with high levels of car ownership andmobility, and high levels of employment in service industries. In this sectionwe ask whether the arguments used nearly two hundred years ago are stillrelevant today.

1.2.2 The early days 1800–1970: location theory

Early studies by economists opened up a debate as to whether a reduction intransport costs brought new areas and products into the market. Rostow


(1960a) argued that this was the case and that transport investment alsocontributed to a major new export sector and was instrumental in thedevelopment of the modern coal, iron and engineering industries. Conversely,Mitchell (1964) in his extensive economic history of the UK railway systemconcluded that these necessary conditions stated by Rostow were alreadymet in the UK before the railways were built. In the UK the railways wereeffectively completed in 1852 and did not have a great immediate effect onthe economy. There were substantial direct effects in the construction phasethrough the employment of unskilled labourers and stimulation of the ironand steel industries, but their major effect was in the development of thecapital market and the levels of savings. They encouraged investment inprofitable (and unprofitable) enterprises.

A fascinating study of the development of the horse-drawn barge(trekschuit) and the canal network (trekvaart) and its impact on the Dutcheconomy also illustrates this historical debate (De Vries 1981). The growthin the canal network was phenomenal, with some 658 km constructed (1632–1839), linking thirty cities so that people (and freight) could travel aroundthe country. The passenger services were used for both business and pleasure,with charges being made to travel or to walk (the charging points for walkerswere the bridges). Demand peaked in the 1670s, but decline followed whichwas attributed to poor maintenance, dishonesty by skippers, competitionfrom unregulated carriers and poor economic conditions. There was a revivalin the 1800s, but the canals were then being replaced by railways and roads.At the end of his investigations, De Vries (1981) concluded that the economicrationale for the canal network was unclear, as it may have only affected thelevel of economic performance, not the actual rate of economic growth. Butthe canal system may have contributed more to gross regional production (in1670) than the railways did two hundred years later (in 1850).

This debate was complicated by the more sophisticated economicarguments of Fogel (1964), who conducted an historical study of the impactof railroad development on the American economic growth during thenineteenth century. He concluded that railways had a primary impact on thecosts of transport and that there were social savings resulting from themovement of agricultural output by rail. The social saving was defined as thedifference between transport by rail and the second best alternative, mainlywaterways. His analysis covered the four commodities (wheat, corn, porkand beef) which accounted for over 90 per cent of agricultural regionalmovements in the USA in the nineteenth century. Fogel also examined thederived effects or those consequences that followed from the savings intransport costs. These were divided into those which would have been inducedby any innovation that lowered transport costs (disembodied) and those thatspecifically related to railways (embodied). His conclusions were very differentto those reached by Rostow. Fogel thought that ‘no single innovation wasvital for economic growth during the 19th century’. Economic growth was a


consequence of the knowledge acquired in the course of scientific revolutionand this was the basis for a multiplicity of innovations. Rail development inthe USA has helped shaping growth in a particular direction but was not aprerequisite for it.

In the UK the industrial revolution was completed before the railwayswere built. The railways were part of that process, not a precondition. Theyemerged out of an effort to apply scientific and technological knowledge tothe improvement of products and the reduction of costs. Cheap inlandtransport is a necessary condition for economic growth, but the satisfactionof this condition did not entail a specific form of transport.

The land use transport links were explicitly included in Von Thünen’sclassic (1826) study on the impact that transport has had on patterns ofagricultural development. As the quality of transport improves, the landdevoted to agricultural production is extended. This in turn allows land valuesand land uses to be reflected in the relative location advantages which thetransport system provides. Many other studies, all classics, developed fromthis starting point (e.g. Isard 1956; Wingo 1963; Alonso 1964). They werebased on concepts of urban economics, land economics and rents usingmethods that assumed optimality and equilibrium in land allocation, a singlemarket and no uncertainties.

As distance from the centre increased, the total costs of transport alsoincreased, and these factors determined the highest use value of anyparticular location. As distance from the centre increased, land valuesdecreased. This theory simplifies reality and promotes transport as the maindeterminant of land value and hence uses. Changes in the costs of transportwill influence the distribution of activities through the land market. Ascities become more complex with high quality transport, they will increasein size and residential densities will reduce. The lower transport costs andlower peripheral land costs mean that with a fixed budget more can bespent on housing. These theories have had great intuitive appeal over thelast two hundred years as their logic is simple and the mechanisms drivingthem are transparent.

The research of Christaller (1933) in southern Germany was the mostinfluential as he demonstrated the links between transport costs and the spatialdistribution of economic activity. He proposed an urban hierarchy of a numberof market towns, each with different transport costs, specializations anddifferential product values. As the towns went up the hierarchy, the range ofproducts increased and the quality of transport improved. The larger centreswere able to increase their share of the total economic activity and this inturn led to concentration of economic activity with a few centres dominatingthe region. Improvements in transport infrastructure strengthened theaccessibility and dominance of the central city—central place theory.

The economic base to these early theories was complemented by otherstudies in the USA which focused more on historical and social factors and


on cycles of growth and decline (e.g. Hurd 1924; Burgess 1925; Hoyt 1939).Rather than assuming that all activities could locate anywhere within the‘ideal’ city, additional constraints were placed on the requirements of particulartypes of activities. For example, certain industries require waterside facilitiesand their workforce would locate nearby, thus limiting other types of activities.Similarly, high quality housing might locate near the city centre, but overtime high income people would move out to the city fringe and this ‘old’housing would be cascaded to lower income households. Simple concepts ofdistance were replaced by evolutionary approaches to location. Harris andUllman (1945) concentrated on specialization, agglomeration economies,clustering and class segregation as reasons why ideal patterns do not emergein real cities. Instead of one centre emerging, several sub-centres woulddevelop. This in turn has led to concepts of hierarchies of development (Berry1967).

These studies have their limitations, but they have formed the basis ofmuch current thinking. They were essentially descriptive and usedsimplifications designed to establish causal relationships to help understandthe development of urban areas. Monocentricity of employment is perhapsthe most widely criticized assumption (Deakin 1991). But other factors, suchas the standard household structure with one worker, the dominance oftransport, the power of market forces and the limited treatment of time,space, political, institutional, legal and social factors, all mean that newapproaches are needed (Alcaly 1976).

The critique of these early approaches was both technical and conceptual.In the 1970s and 1980s questions were asked about the methods being usedand the underlying logic of the links between transport and economicdevelopment. Initial scepticism turned into full-scale criticism. For example,Wilson (1978) stated:

When we turn to transportation, the roles that improved transportationare supposed to play are numerous. For example, transportationimprovements have been cited as having important positive effects onpolitical unity, social cohesion, economic growth, specialisation, andprice stability as well as on attitudinal change. Yet…precisely oppositeeffects are equally plausible.

(Wilson 1978:102) The developmental tradition (Hart 1983) has involved a strong belief thattransport made a vital contribution to economic growth. His historical analysistraces both public and private involvement in transport infrastructureinvestment, with assumed high rates of return and substantial multiplier effects.Yet, many of the early companies, in particular those involved in railinvestments, went bankrupt. But this may have been due to other factorssuch as macroeconomic factors and conditions for competition. The conclusion


from much of the economic history has been to downgrade the impact oftransport innovation on aggregate performance. Hart (1983:15) concludesthat transport was one aspect of productivity improvement, but that thechanges in agriculture and manufacturing stimulated growth in incomes andbegan to generate substantial volumes of traffic: ‘improved transport was aluxury afforded out of economic growth’.

Location theory argues for the strengthening of the centre with aconcentration of economic activity. Yet, much of the historical evidence, evenin the early period, suggested that there was substantial variation betweendifferent cities and a weakening of the influence of the city. Regionaldevelopment policies also assumed that investment in transport would helpalleviate depressed industrial regions and open up rural areas by increasingtheir share of economic activity. Even here the concentration arguments werebeing replaced by those promoting the spread of growth from the moreprosperous regions.

1.2.3 The 1970s and 1980s: urban and regional modelling

The next twenty years have seen the development of a series of moresophisticated models for land use and transport analysis at both the urbanand regional scales. This has allowed the addition of more complex socialstructures for households, information on land availability and housing stock,as well as more complex notions of transport costs (i.e. generalized costs oftravel). More important, though, has been the recognition that cities exist, soideal forms are no longer taken as the starting point. A vast literature hasemerged (for a review see Berechman and Small 1988; Anas et al. 1998).

This new generation of urban models allocates housing and jobs withinurban regions on the basis of their relative accessibility, land availability,income levels, population and employment by category and other social andphysical characteristics. These methods have had some success in modellingthe existing urban structure and predicting changes in locations giveninvestment options (e.g. roads), new developments (e.g. housing) or changesin travel costs (e.g. pricing). Again, most of the models focus on changes intransport costs as being the principal driving mechanism. Other importantlocation factors such as quality of housing, lifestyle considerations, familyconstraints, quality of the neighbourhood, which come out as important insocial surveys, do not feature as dominant issues in the analysis.

Urban modelling approaches have also been matched by econometricanalyses, input-output models and regional analyses. In many of these studies,the focus has been on the employment impacts, not on understanding howcities or regions work. Botham (1983) tried to estimate the contribution ofthe roads programme in Great Britain to regional development. Themethodology adopted requires the counterfactual situation to be set up toanswer the question as to the nature of the spatial structure of the British


economy in the absence of the roads programme (1957–76). Using a set ofregression equations, Botham estimates the changes in the spatial distributionof employment brought about by the changes in accessibility. The tentativeconclusion reached is that the impact is marginal, if (in the absence of roadinvestment) it is assumed that transport costs remain constant over time. Ifcongestion is assumed to increase costs, the impact is increased, but it isdifficult to put a value on it. This means that the roads programme in GreatBritain has encouraged spatial concentration.

This conclusion is qualified by the recognition that other policies mayhave had a greater impact than infrastructure investment. Pricing and taxationpolicy, restrictions on drivers’ working hours and the value of work timemay all have had a greater impact on accessibility. Similarly, spatialspecialization, labour supply, wage rates and migration patterns may all haveinfluenced employment patterns more substantially than transportinfrastructure investment.

At the major link level, similar problems have been encountered. In theiranalysis of the traffic generation effects of the M62 motorway between Leedsand Manchester, Gwilliam and Judge (1978) conclude that large-scale trafficgrowth may indicate that a motorway has stimulated economic growth.Growth in traffic on the M62 route was 34.3 per cent (1970–77), and thiscompared with a national growth rate on rural roads (excluding motorways)of 24.8 per cent over the same period. The difference between the two figuresgives an estimate of the generated traffic but, as noted by Botham (1983:25),the results are very sensitive to the assumptions made, namely thecounterfactual situation. Dodgson (1974) has calculated that the greatestreduction in the total costs of manufacturing and distribution in any areabrought about by the construction of the M62 motorway was about 0.33 percent. For most areas, it was substantially less. Any attempt to measure theeffects on local employment failed as the scale of the impact was so small.

A third study (Mackie and Simon 1986) takes an individual link todetermine whether road investment benefits industry. The Humber Bridgeforms a major new link across an estuary in Yorkshire (Great Britain), and itsdirect impact seems to have been in extending the market area of companiesrather than their location. Some firms felt that the tolls had cancelled outoperating cost savings, but that timesavings could be used to increase vehicleand driver productivity (Mackie and Simon 1986:383).

This conclusion raises an important issue in the debate. In cost benefitanalysis, a clear distinction is drawn between user and non-user benefits.User benefits result from travel time savings, operating cost reductions andreductions in the numbers and severity of accidents. Non-user benefits accrueto other people and are often associated with the economic developmentbenefits. Traditionally, economists have excluded these non-user benefits fromthe cost benefit analysis, because of the difficulties in distinguishing generationeffects from redistribution effects and the associated problems of double-counting


benefits (Friedlander 1975; Dodgson 1973. See also Chapter 7). Rephann(1993) concludes, ‘benefit-cost analysis only includes development benefitsto the extent that it results in user benefits for existing and generated traffic’(p.439). It would seem that the crucial issue of benefits and costs redistributionalso needs to be considered in an overall assessment of the economicdevelopment effects of transport investment.

In addition to the methodological questions, there are important policyconsiderations. Within government there is a common assumption that newroads encourage economic growth by inducing industries to relocate to thoseregions with better quality infrastructure. Two hypotheses underlie this view(SACTRA 1977):

The first is that the direct road user benefits which result from a schemesubstantially understate the total benefit to the economy. The second isthat trunk road construction is an efficient method of attractingeconomic growth to selected areas. Thus the regional distribution ofbenefits, even if the total is accurately measured, may be important.Yet, despite the strength of the assertions about the importance of theseeffects, little actual evidence has been presented to us.

(SACTRA 1977: para 20.18). The conclusion from this influential report is that the restructuring effects ofroad construction on economic growth in developed countries are weak andat best not proven. The main benefits to industry are in travel timesavings,but its importance in terms of relocation has been overstated. The SACTRAcommittee supported the government in its current practice of excluding theindirect effects from the evaluation.

Yet the arguments for new roads continues as it produces ‘importantefficiency gains for commerce and industry’ (UK Department of Transport1993) and makes industry more competitive. The standard argumentpromoted by the Confederation of British Industry and other roadorganizations is that improved roads reduce transport costs (primary effect),while the lowering of distribution costs relative to production cost changesfirms’ optimum production and distribution strategies (secondary effect)(Mackie and Simon 1986). The strong argument on the development effectsof roads has become diluted as roads are now seen as being ‘necessary’ but‘not sufficient’ for economic development (Huddleston and Pangotra 1990).The weaker development argument suggests that if a region has all theeconomic factors present for growth, then its full potential will not be realizedwithout further transport investment. This type of argument has been centralto regional development theory (Hirschman 1958; Hansen 1965).

Hirschman promoted unbalanced growth with social overhead capital beingconcentrated in regions and industries where it can maximize development.Although this strategy increases inequalities, it was claimed that they would


only occur in the short run as the private sector would ensure stable growthoverall. It was unclear as to what mechanisms would be used to ensure stability.The approach adopted by Hansen (1965) differs as public capital is dividedinto economic overhead capital (roads, sewerage, water, utilities) and socialoverhead capital (education, health, nutrition). He also divides regions intothose that are congested, intermediate and underdeveloped. His argument isthat economic overhead capital should be concentrated in the intermediateregions and that social overhead capital should go into the underdevelopedregions. The debate is over whether, through concentration or dispersal,regional development objectives can be achieved and whether economicgrowth at a regional level can be reconciled with the distribution of thatgrowth. The spillover effects of economic growth will affect all parts of theregion. The question is whether the effects are greatest if directed towardsthe location of greatest potential (i.e. the centre) or whether it is dispersed toall parts of the region.

There seems to be little agreement here. As Rephann (1993) concludes,Hirschman would promote road investment in the developing urbanizedregions; Hansen would target the intermediate regions; while growth poleproponents (e.g. Allen and MacLennan 1970; Hansen 1971) would arguefor concentrating highways in urban areas which exhibit prior dynamism.Rephann (1993) comes to five conclusions on the empirical evidence fromthe USA: • Road investment appears to have a greater effect on economic activity in

the less industrialized regions such as the Sunbelt.• Extremely underdeveloped regions are less promising candidates for road

development than regions in an intermediate stage of development whichare experiencing low growth.

• The effects are positively correlated with urbanization levels andmetropolitan proximity.

• Other types of infrastructure (e.g. airports) may contribute to theeffectiveness of highway investments.

• Additional roads may result in diminishing marginal returns. These conclusions are consistent with those proposed by Hansen (1965).The basic implication for regional development theory is that road investmentwill be most effective in promoting development ‘when certain “indicators”,“signals”, or “triggering forces” are present’ (Rephann 1993:447).

Much of the debate during the 1960s, the 1970s and 1980s revolved aroundfreight transport. It was consistently argued that a reduction in freighttransport costs would lead to the exploitation of scale economies, and this inturn would lead to reductions in commodity prices. The underlying argumentwas that these benefits exceeded the direct reductions in transport costs(Tinbergen 1957; Bos and Kocyk 1961). However, this argument depends on


the assumption that increased use of factor inputs (such as labour, capitaland natural resources) would be attracted by the rise in factor prices whichresulted (Friedlander 1975). However, if the factor inputs were assumed tobe fixed, the reduction in transport costs would be an accurate measure ofthe total effect of the scheme, provided that the benefits to any generatedtraffic could also be captured (Dodgson 1973).

These indirect benefits would be greatest in locations where the factors ofproduction were currently unemployed. In addition, the increased demandresulting from the road itself may have multiplier effects on regional andnational income. Their effects would result in an increase in aggregate demandand should not be regarded as benefits specific to road construction (SACTRA1977). There are several weaknesses in the argument presented here,particularly as it relates to depressed areas: • A reduction in transport costs to a depressed area may make it easier to

supply other areas from the area in question, but at the same time it willmake it easier to supply that area from elsewhere. It is directly analogousto the international trade theory argument of the impacts of a reductionin the tariff.

• Regions that will benefit from investment are those that already have non-transport advantages, such as a buoyant local economy, new industries, aheavy inflow of investment, available sites and a high quality labour market.

• The problems of depressed areas can be traced to non-transport factors(e.g. the need for industrial restructuring and increases in labourproductivity). Improved communications may do more harm than good.The same applies to the effect of improved transport to and from portsand the impacts that this may have on the country’s balance of trade(SACTRA 1977).

• Where transport cost savings have led to an economic benefit, they havenot necessarily led to a parallel increase in employment. Where savingsoccur, restructuring can take place allowing more capital intensiveproduction, further savings in labour costs and an increased profit margin(Hopkins 1986 quoted in Grieco 1994).

• These arguments ignore the issue of transport financing. If taxes are used,for example, it would reduce income and hence demand.

The empirical evidence to support the arguments also reflects clearly on thethinking during the 1970s and 1980s. Some has already been mentioned withrespect to particular networks or links, but the majority of the evidence comesfrom surveys of firms about their intentions to expand or relocate as a resultof a new transport link (Table 1.1).

The same conclusions can be drawn from the surveys of firms as from themodelling and review material presented in this section. The general conclusionfrom these surveys has been that transport is a second order consideration


for location or relocation decisions, provided that there is a good qualityroad network available. Transport is a background variable that has to bepresent, but comes behind factors that more directly affect the efficiency,productivity and profitability of a firm. It is also considered a factor that isbeyond the control of an individual firm.

The scale of any impact is small whether perceived by the firms or translatedinto savings in transports costs. However, regional policy in the UK stillfavoured regions with high unemployment. The first priority was for improvedtrunk roads linking these regions with those of greater prosperity; the secondbeing for higher levels of intra-regional road spending to promote localmobility and growth (Hart 1993). Continental Europe showed less of atransport bias towards disadvantaged industrial regions and most investmentat this time went to road and rail infrastructure in the rural areas.

1.2.4 The 1990s: macroeconomic approaches

The debate seemed to diminish in its intensity over the latter part of the1980s and even into the 1990s, but it now seems to have re-emerged. Muchof the discussion follows the main conclusions from the 1970s’ and 1980s’interest in the subject, but there are several new dimensions. In this section

Table 1.1 Surveys of firms’ location decisions


we shall concentrate on some of the main new themes which have emerged,most of which are taken up later in the book in more detail (Chapter 6). 1 The lack of investment in the infrastructure of all types (e.g. schools,

water and public buildings) has had an adverse effect on efficiency andproductivity. The difference here is examining the implications of notinvesting in the infrastructure rather than identifying the impacts ofadditional investment. This debate is outlined in Sections 1.3 and 6.5and concerns much of the recent research in the USA and elsewhere (e.g.Aschauer 1989a, b; Munnell 1990a; Berechman 1995; Helling 1997).

2 The land development effects of roads have a long history, but recentdebates have tried to identify under what conditions measurabledevelopment effects can be identified. If particular situations can beisolated where rises in land values, rent levels and house prices can bemeasured, then there is a strong case to recapture some of that increasedvalue, through taxation or through contributions of developers to thefinancing of the transport infrastructure (Anas et al., 1998).

3 The crucial question of the financing of transport infrastructure throughtraditional public expenditure routes or through new forms of public-private partnership has become a central issue in the UK, Europe andUSA (Chapter 3).

4 Environmental arguments have become a more important considerationin the political agenda, particularly if it can be demonstrated that newroads lead to more and longer trips. Low density development has beenfacilitated by high quality infrastructure and the availability of the car.This in turn leads to greater land consumption, more energy consumptionand greater levels of environmental pollution. But even here the debatecontinues over the effects that new roads have on employment, new traffic,destination choice and the use of different modes of transport (SACTRA1994 and Chapter 5).

5 The view from industry (CBI 1995) is that competitiveness is beingjeopardized by a lack of investment in the infrastructure. With a congestednetwork, distribution is becoming increasingly unreliable and acompetitive time to market is not being provided. This contrasts with theheavier levels of investment being made in Europe and, as with the earlierarguments, the relative competitiveness of region has been extended tothe international level. The Confederation of British Industry argues thatfrom 1975–90 UK transport investment was little more than half theEuropean average. This has narrowed recently, but is now widening again(1998). Their case for greater investment revolves around strategy andconsensus. In France, for example, a key priority is to ‘assert the positionof France in European and world-wide competition’, so investment takesplace. Elsewhere (e.g. in Germany and the Netherlands), long-terminfrastructure plans are agreed and implemented (Chapter 4).


6 A further new element in the debate has been the concept of induced traffic(SACTRA 1994), which includes all forms of new traffic generated bynew road construction. The economic values of a road investment schemecan be overestimated by the omission of induced traffic, particularly wherea network is operating close to capacity; where traveller response to changesin travel time or costs is high; or where the implementation of a schemecauses large changes in travel costs. The SACTRA report recommendedthat ‘scheme appraisal must be carried out within the context of economicand environmental appraisals at the strategic area-wide level which takeaccount of induced traffic through variable demand methods. Much greateremphasis needs to be placed on the strategic assessment of trunk routeswithin a corridor or regional or urban context’ (1994:iv, para 17). Thesequestions are returned to in Chapter 3; in particular the links betweentransport investments and land use or development changes.

All of these more recent elements, together with the longer establishedarguments from the 1960s, 1970s and 1980s, crystallize the importance ofthe issue. Both the methodological arguments and the empirical evidence areinconclusive, yet major investment decisions are made on the basis ofinadequate theory, data and arguments concerning causality. Two possibleexplanations might be that either we know the answers but are not preparedto discuss them, or that we really do not know the nature, strength anddirection of these links. Our starting point is the latter explanation. Table 1.2summarizes how priorities for transport infrastructure investment havechanged over the last fifty years, both in Europe and the UK.

In reviewing Table 1.2 we need also to recognize the political dimensionwhich must be seen as an integral part of the rationale, as transport andeconomic development links are still maintained almost as an act of faith.Unless there are clear and measurable economic benefits from investment intransport infrastructure in the poorer areas of Europe, what other justificationcan be used? There is a history of infrastructure investment in both congestedand peripheral regions, yet the rationale is still not clear.

1.3 The debate: productivity of infrastructureinvestments

Many studies in both developed and developing countries have tried toestablish a statistical link between aggregate infrastructure investment andgrowth in GDP. The findings are staggering with rates of return of up to 60per cent. In turn, there has been an extensive debate over the validity of theanalysis and the claimed causality in the relationship. Two main criticismshave been raised. The first question is whether the simple relationship betweenoutput increase (GDP) and input (rate of investment in infrastructure) is notinfluenced by other factors not included in the analysis. The second is the


nature of the causality—whether growth leads to additional infrastructureinvestment, or whether investment leads to growth, or whether there is aninteraction effect. An example of a simple statistical causation is shown byFigure 1.1.

A more sophisticated development in the debate has been the use of timeseries data for the USA and other countries (Aschauer 1989a, b; Munnell1990a). Aschauer tests two hypotheses in trying to demonstrate whether publiccapital ‘crowds out’ private capital. The first hypothesis argues that higherpublic investment raises the national rate of capital accumulation above thelevels chosen by private sector agents. The second argues that public capital,particularly infrastructure capital (including roads, water, sewers and airports),is likely to bear a complementary relationship with private capital in theprivate production technology (Aschauer 1989b). His main conclusions arethat there is a link between non-military public capital stock and measures ofprivate sector productivity (Aschauer 1989c) and that the public sector inputsare complementary (Hypothesis 2). He further concluded that the decline inUS productivity in the 1970s had been precipitated by declining rates of publiccapital investment. This important part of the debate is given extensive

Table 1.2 Changes in policy on transport infrastructure investment


coverage in Chapter 6, where the theoretical, analytical and empirical evidencefrom the USA and elsewhere is presented.

The general conclusions reached are that public capital has an impacton economic growth, on private capital and on labour productivity, butthe magnitude and significance of these effects are not clear. The key issuein any analysis of complex interrelationships is the unravelling of theselinkages, so that it is clear what can be concluded as predictable correlationsand what is still unconnected or remains as uncertain relationships.Certainly, the results from the production function analysis may overstatethe scale of the expected impacts of public infrastructure investment, butthe links between public investment in infrastructure and economic growthand private capital productivity are important concerns for analysis(Munnell 1993).

Figure 1.1 Links Between GDP and infrastructure stock per capita, 1990.Source: World Bank 1994Note: Axes are logarithmic; infrastructure includes roads, rail, power, irrigation and

telephones.a Purchasing power parity (PPP) dollars are valued in Summers and Heston 1985

international prices


1.4 The debate: infrastructures for development

Although most of the debate and material presented in this book concentrateson infrastructure investment in developed countries, there is a substantialamount of literature on developing countries. Indeed, much of the earlyresearch carried out in the 1960s by geographers takes examples from thedeveloping world, particularly in terms of development of transport networksfrom ports into the hinterland, the opening up of new areas for developmentand the development of dominance (e.g. Taaffe et al. 1963).

In developing countries, infrastructure includes all forms of services, notonly transport. Over $200bn a year is invested in new infrastructure, some 4per cent of the national output and 50 per cent of all investment (World Bank1994). This includes public utilities (power, telecommunications, piped watersupply, sanitation and sewerage, solid waste collection and disposal and pipedgas), public works (roads and major dam and canal works for irrigation anddrainage) and other transport projects (urban and interurban railways, urbantransport, ports and waterways, and airports).

The quality of this infrastructure is alleged to determine the relative successof each country in expanding and diversifying its economic base, inaccommodating population growth, in reducing poverty and in improvingthe environment. The development argument is that good infrastructure (in ageneral sense) raises productivity and lowers production costs, but that it hasto expand fast enough both to accommodate growth and to open up newareas for development. Although the precise nature of the link is unclear(Section 1.5), at the cross-section there does appear to be a correspondencebetween the increase in the infrastructure stock and the level of GDP percapital (Figure 1.1). A 1 per cent increase in the total infrastructure stock issaid to be matched by a 1 per cent increase in GDP across all countries (WorldBank 1994). But this general conclusion may give different results if specifictypes of infrastructure are examined individually.1

Similarly, the balance between the different forms of infrastructurechanges with the level of development. High-income economies invest morein power, roads and telecommunications infrastructure, while low-incomeeconomies are more concerned with the basic infrastructure such as waterand irrigation.

The World Bank concludes that there is a high potential return frominvestment in infrastructure provided that the overall policy conditions aregood. Returns are lower by 50 per cent or more in countries with restrictivetrade policies. These conclusions were confirmed by the Brookings Institute(Kresge and Roberts 1971) some twenty-five years ago, when they concludedthat although infrastructure investments had good rates of return, successlargely depended upon general economic policy: ‘Infrastructure is a necessary,although not sufficient, precondition for growth—adequate complements ofother resources must be present as well’ (World Bank 1994:17).


The growth impacts also depend on other factors such as the timing, location ofthe additional capacity, levels of actual and forecast demand and the existing structureof the network. Most investment is promoted by the public sector and is seen as ameans to promote economic development and to provide employment, particularlyin recessions. It is seen as an investment in the future. Many of the investment techniquesused are labour intensive and this provides employment with intensive economicgrowth. However, despite these clear intentions, even in developing countries, it isoften capital expenditure on infrastructure that is reduced as the impacts are notimmediately apparent. In this respect, there is little difference between developed anddeveloping countries as the political short-term priorities and pressures on publicfinances take precedence over the longer term economic development objectives.

An examination of the rates of return from projects funded by the WorldBank gives an average return of 16 per cent (1980–90). These rates of returnhave been re-estimated after the project has been completed (including loandisbursement—Table 1.3). As can be seen from the table, returns have beenlowest (and declining) for irrigation and drainage projects. Transport investmentshave a consistently high rate of return, except for airports (small sample) andrailways. The main reason for the higher or lower than expected rates of returnis the actual growth in demand. Overestimation of demand in some cases exceeded20 per cent, but in the railway sector traffic failed to reach its projected level in29 out of 31 cases. In 10 cases, traffic actually declined. The highest rates ofreturn were in the projects that had high growth levels in demand, and this wasmost evident in roads and ports (Table 1.3).

Table 1.3 Average economic rates of return on World Bank supported projects1974–92

Source: World Bank (1994: Table 1.2)Note: 1 Rates of return are f inancial, not economic.


At the cross-section, it does seem that there is a clear relationship betweengrowth in the infrastructure stock and GDP. This conclusion is not unsurprisingas development on a worldwide scale must show variation as income levelsvary. One of the main determinants must be the level of infrastructure provisionas this again varies substantially between countries, in proportion to theirwealth levels. The more fundamental conclusion on whether this link is acausal one has not been proved, nor has the direction of that link. The WorldBank evidence also seems to suggest that the relationship is a one-to-oneratio between infrastructure and GDP growth. It would be interesting to seewhether this relationship is stable over time. Perhaps growth in GDP in high-income economies is no longer related to the increase in infrastructureprovision in the same ratio. However, it should be remembered that the WorldBank definition of infrastructure includes all the social overhead capital, notonly transport infrastructure that is the main concern here.2

Similarly, the high rates of return on infrastructure projects in developingcountries are to be expected, particularly where the infrastructure is openingup new areas for development or where the existing network is limited. Newinvestment will make a substantial impact on accessibility, but moreimportantly will open up new markets and areas for development. The ratesof return sought by the World Bank are about twice the levels expected intransport projects in developed countries—about 15 per cent as comparedwith 8 per cent—and there are numerous projects that exceed this threshold.

The main conclusion from this review of the evidence in developingcountries is that infrastructure (in the general sense of social overhead capital)can produce major benefits in economic growth, poverty alleviation andenvironmental sustainability. But this only occurs when it provides servicesthat respond to effective demand and it does so efficiently. The World Bank(1994) places the responsibility for poor performance on the individualnational governments. Improvements are required in the management ofprojects so that they are run like businesses, and in the introduction ofcompetition. The discipline of the market is required in infrastructureprovision, together with a greater involvement of the private and communitysectors. Too often in the past has the provision of infrastructure been seensolely as the prerogative of the public sector. Investment per se is not sufficient,since poor management of projects can cancel out any potential benefits. Themeasurement of the quality of infrastructure management is rather complexand largely inaccurate. In this book, we do not attempt to measure the qualityof the management in any quantitative analysis, so the problem of predictingfuture effects remains unsolved.

But in a more general sense, the World Bank’s view on transport anddevelopment is very clear:

Transport is central to development. Without physical access to jobs,health, education, and other amenities, the quality of life suffers;


without physical access to resources and markets, growth stagnates,and poverty reduction cannot be sustained. Inappropriately designedtransport strategies and programs, however, can result in networksand services that aggravate the condition of the poor, harm theenvironment, ignore the changing needs of users, and exceed thecapacity of public finances.

Macroeconomic studies by the World Bank show that investing intransport promotes growth by increasing the social return to privateinvestment without crowding out other productive investment.Microeconomic analysis confirms the high social value of transport.The estimated economic rate of return on transport projects atcompletion is 22 per cent, half as high again as the Bank average.Improvements in rural transport have lowered the costs of agriculturalproduction directly, by increasing access to markets and credit, andindirectly, by facilitating the development of the non-agricultural ruraleconomy. Improvements in urban transport have increased labourmarket efficiency and access to amenities, making changes in the scaleand form of urban agglomerations possible. Improvements in interurbantransport have facilitated domestic and international trade and spedthe movement of freight, as well as of people.

(World Bank 1996:1)

There are parallels between experiences in developed and developing countries,but the clear focus of this book is on developed economies. It is here that thedebates are less clear, yet the arguments for the links between transportinfrastructure investment and economic development are still made. Indeveloping countries, where the transport network is sparse and of a lowerquality, the development arguments seem clearer. But as all economies globallymove towards being more developed, the links between transportinfrastructure investment and economic development will become less clearin all countries.

1.5 The need for the book

1.5.1 Rationale and novelty

Transport investment and economic development have in the past been closelylinked, with substantial pressure being placed on central government to financemajor infrastructure projects on the understanding that this will lead toincreased competitiveness and economic development. From the review ofthe historical evidence cited in this chapter, the case is far from clear,particularly in developed economies where additional investment has littleimpact on the overall accessibility within the transport system.

The first aim of the book is therefore to bring together the wide-ranging


historical, economic, geographical and development literature which has beenproduced over the last thirty years on the subject. The purpose of this reviewis to present a synthesis of knowledge and to identify the main debates thatwill be elaborated on in the main parts of the book.

The second aim is to examine new methodological approaches for theanalysis of the relationships between transportation network developmentand economic growth. These are essentially microeconomic models thatdiffer from the more traditional macroeconomic approaches to economicdevelopment (see further in Chapter 2). The thinking behind the argumentfor new approaches is that only at the local level will the impacts oftransport investment actually be identified. Macro level analysis does notallow the subtlety of change and impacts of specific infrastructure projectsto be measured, nor does it allow the causality of relationships to beestablished, nor does it control adequately for the full range of impacts tobe measured.

Related to this second aim is the realization that new priorities are presentin the economy and that many of the traditional arguments relating to bothlocation theory and economic base theory no longer apply. People and firmsare much more flexible in their location decisions and patterns of interactionsand travel are more complex. Similarly, there are new priorities which influencedecisions and this book confronts some of them. There are new economicforces at work as the economy moves from one based on manufacturing toone based on service and information industries. Just as the industrialrevolution caused a fundamental shift in employment from agriculture toindustry, so is the new technological revolution causing a major change inthe employment base of most developed economies. Parallel with this changeis the new demographic profile of the population, with an increased lifeexpectancy, early retirement and new household structures. Traditionalpatterns of families and a dependent elderly population are now being replacedby a much more heterogeneous family structure with a large new active elderlypopulation. The economic forces, which have been dependent upon traditionalpatterns of population structure, are now being transformed by these newpatterns.

Similarly, where transport investment has traditionally been seen as a rolefor only the public sector, new forms of financing are being promoted throughprivate sector investments and joint ventures involving both the public andprivate sectors. Projects which have been mainly financed from nationalbudgets are now being supported by international agencies such as theEuropean Union (e.g. through the Regional Development Funds), theEuropean Investment Bank and a wide range of private sector financialinstitutions (e.g. banks and pension funds).

Finally, much analysis of infrastructure investment has concentrated solelyon the impact of the new link being proposed in terms of its specificcontribution to improving travel, principally through travel time reductions.


In terms of the economic development impacts and the contribution of asingle link to the network as a whole, there is much less knowledge. Themajor analytical and empirical parts of the book provide a range of contrastingapproaches. We tackle these through analysis at different levels and withexamples from different modes. We also interpret the economic developmentimpacts of the extensive literature of benefit-cost analysis, which lies at theheart of current debates on investment, particularly as it relates to externalitiesand network effects. We develop a microeconomic modelling approach toexplore the linkages between firms, employees and the transport network.All these elements provide new insights and interpretations to the complexrelationships being investigated.

The case studies of road, rail and airport investment decisions are basedon secondary data and exhaustive investigations into how development effectscan be isolated and measured. This has proved quite a frustrating task as theinformation available is never perfect, but even then some clear conclusionscan be drawn. Again, this complementary analysis, which carefully assemblesand reviews the available material, is new and provides more qualitativeanalysis to match the quantitative and theoretical analysis in Parts II and III.As such, we aim to provide a major synthetic and original input to the greatdebate on the links between transport infrastructure investment and economicdevelopment.

1.5.2 The ten key questions

Having presented the general rationale for the book and a basic review of thekey elements in the debate, it is now appropriate to address some of themajor unresolved issues that will be taken up later. A major contribution tothe literature will have been achieved if even some of these questions can beanswered. The ten key questions that have emerged so far are listed below intwo groups. The first group (1–6) we examine in detail. The second group(7–10) we either touch upon briefly or not at all, although the questions arestill quite important.

1 Is the growth impact of any new transport link in developed countrieslikely to be significant? Most networks in developed countries arealready well connected, so the impact of any new link is likely to besmall individually—yet the combined effects of new links may besubstantial. Accessibility can be assessed as an absolute and as a relativeconcept. How important is accessibility in regional competitiveness,both in terms of its contribution to the overall buoyancy of a regionaleconomy and in terms of its redistributional effects from one region toanother? Travel timesavings resulting from new links in most developedcountries are likely to be small. In many cases investment is in upgradingexisting links and in marginal additions to networks (e.g. bypasses).


Does investment have to be of a sufficient (unspecified) scale beforeany impact can be identified, let alone measured? Chapters 8, 9 and 10examine aspects of this issue.

2 Are transport costs a small part of total production and labour costs?Transport costs are often a small part of the total production costs of firms.Even where they form a high proportion of the total activity/participationcosts of individuals, they are relatively insensitive to marginal changes incosts or accessibility. Firms’ and individuals’ activity patterns are more likelyto be influenced by the uncertainties and unreliability of the transport system.The time and monetary costs of these factors have already been taken intothe decision to travel and have effectively been discounted. Will changes inthese factors be of a sufficient scale or importance to have any real impacton travel decisions? The effect of investment in a new link, given the network,is discussed in Chapters 3, 7, 9 and 10.

3 Are buoyant local economic conditions more important than transportinfrastructure improvements in generating growth? Buoyant localeconomies attract more investment and in-migration of labour (perhapsat the expense of less buoyant economies). The return on infrastructureinvestments may be higher in these buoyant economies, but theconstruction costs may also be higher (due to higher labour costs, landacquisition costs or environmental costs). The classical debate in theregional development literature is between those who argue for moreinvestments in the buoyant areas, which will then have spin-off orcascading benefits to other areas (Myrdal 1957), and those who suggestthat investments should be targeted directly to the less well-off regions(Rostow 1960a, b). The impacts will be felt in both the short term andthe longer term and they will be:

• direct—transport related;• indirect—land use and development related;• new—jobs, activities, etc. attracted to the region.

Measurement in each of these categories is not easy, in particular thenew effects—it is also important to consider the scale or level at whichthe impact can be measured and identified. The modelling part of thebook (Part III) examines transportation investment within the frameworkof the national or local economy. Chapters 10 and 11 briefly discussother aspects of local economic conditions when evaluating the impactsof rail and airport investment.

4 Are the unique characteristics of the area and the spatial extent overwhich the growth impacts are to be felt considered? Even with a clearlydefined before and after survey, it is necessary to control for other changeswhich are taking place in the economy. Hence a control area is oftenused to isolate changes due to other effects rather than the transportinvestment. The difficulty here is to find an appropriate corridor or area


for comparative purposes. This problem is particularly acute where theinvestment is large (and unique) or where the expected impacts are highlylocalized (e.g. in a rail investment where the impacts are expected aroundthe stations). Chapters 5, 8, 9 and 10 address the local and uniquecharacteristics, as well as the spatial extent, over which the investmentimpacts are likely to occur.

5 Can one generalize about the results from specific case studies? There isa substantial amount of information available from individual studies,but as yet little consensus or theory deriving from it. It has been difficultto generalize the results as case studies have been focused at the microlevel, and the macro level studies have been subject to much debate. Thisquestion is addressed in all of the three major parts of the book.

6 What is the role of technological change in affecting the relationshipsbetween transport investment and economic growth? Technology ispermitting new forms of production as processes are sped up, throughincreased specialization and through short production runs. The roleof transport and infrastructure in these processes is becoming morecomplex. For example, there has been a substantial increase in theamount of traffic in transit at any one point in time so that thewarehousing and stock level requirements can be reduced. Freightdistribution networks have been radically reorganized to take accountof the availability of new technology and high quality (road)infrastructures. It is also possible that changes in transport infrastructurecan promote technological change, particularly in the way in whichparticular activities are carried out. The best example here is thepossibility of teleworking which requires both a good quality transportinfrastructure and an appropriate telecommunications infrastructure.Chapter 4, 5 and 6 discuss the impact of technology on the consequencesfrom a transportation infrastructure investment.

7 How slow, long term and complex are the adjustment and readjustmentprocesses within the regional economy following a transport investment?It is very difficult to establish cause and effect, and the effects may operatein different directions. For example, the relative effects in one locationmay be positive, but it may have negative effects elsewhere—a competitioneffect. It is important to carry out a range of analyses to cover ex anteand ex post situations so that our understanding of the basic processes atwork can be strengthened. Improvements in the relative position of onepart of the regional economy might be at the expense of another part ofthe economy (this is the pie theory—namely that the total size is fixed,but the debate is over shares of the pie). The relative positions change,but there may also be some net increase in production in the economy asa whole (national or international), particularly if interregional trade ispromoted. Through these actions it is argued that the competitive positionof one region will improve (perhaps at the expense of others)—with lower


production costs and better access to markets. This in turn will generatemore output and profitability, which will lead to higher income, levels,more expenditure and increases in employment. This question is brieflytouched upon in Chapter 4.

8 Does a good transport infrastructure raise the image and the perceptionsof an area, thereby attracting additional private investment? A newtransport investment raises the profile of one area and development isoften promoted on the quality of the new transport infrastructure. Forexample, this has been a major argument by some cities for new investmentin metro and tram systems (e.g. Atlanta and Sheffield). Similarly, a lackof infrastructure may work against the attractiveness of a particularlocation. These arguments have been extensively used in justifyingdeveloper contributions to the costs of infrastructure as developers areseen as one of the main beneficiaries from the investment. The intentionis to recapture some of the benefits from infrastructure investment ratherthan allowing the benefits to remain with the developer-value capture.The increased opportunity for high rent levels often accrues only to thedeveloper, not the provider (often the public sector) of the infrastructure.Chapter 9 and 10 partly discuss this issue.

9 How can one assess the course of economic development in an area if thetransport investment was not made? This counterfactual situation hasalways presented difficulties. Although it is difficult to measure what hasactually happened, it is even harder to measure what would have happenedwith no transport investment, or with an investment in an alternativetransport project, or an investment in a different type of project altogether(e.g. a training project). The scale of the impact is likely to be small andthere will be substantial ‘noise’ in any analysis. Does this make the problemtoo difficult for analysis, so any theoretical advance will be in the formof a general theory (e.g. Myrdal), and any micro level advance will resultfrom individual case studies from which it will be difficult to generalize?Chapter 11 examines an investment option at an airport with and withouta new major facility.

10 What is the role of expectations, regarding the impacts of investment, inachieving growth? This question has two main aspects. First, as timepasses it becomes harder to measure the impacts of any particularinvestment with any degree of certainty. Second, it also affects theexpectations of investors about whether their returns will be realizedover a short or long period and this, in turn, will affect their propensityto invest. We do not really address this question within the book apartfrom a minor reference in Chapter 3.

All of these questions suggest that we need to make a careful analysis of themain issues to clarify the relationships conceptually, theoretically andempirically. The basic question addressed in this book, namely the link between


transport infrastructure investment and economic development, is likely tobecome increasingly important as the user benefits of additional investmentin a well-connected network diminish and as other non-user benefits becomemore important.

1.6 Objectives and structure of the book

The basic question addressed in this book is whether there is (or should be) aclear link between investment in transport infrastructure and economicdevelopment. If a link can be established, it is then important to determinewhether there is any causality inferred. It is also necessary to establish whetherthe relationship is linear or whether there are decreasing marginalproductivities, especially in regions that are already well connected. This againraises important questions as to what are the necessary conditions for causallinks to be established.

The book is divided into four main parts. In Part I (Chapters 1 and 2), thecontext of the debate is set out through a short historical review, discussionof the productivity of infrastructure investment and the infrastructure fordevelopment argument. It also outlines the rationale for the book, and coversthe definitions and approaches adopted and methodological framework used.Part II provides a perspective on some of the major changes taking place insociety. Even if links can (historically) be found, it is argued that thesecontextual changes must now form a central part of the debate. The primaryconcern over economic growth must be extended to include the means bywhich transport infrastructure is funded, the changing nature of the economyitself and the new priority being given to distributional and environmentalconcerns. In the past, apparently simple relationships may have beenappropriate, but the new complexity of change means that new methods arerequired.

In Part III three major analytical approaches are presented. A macro levelanalysis of the means by which the growth effects of modelling capitalinvestments in transport is presented in Chapter 6, together with someempirical results. This is followed by a comprehensive coverage of benefitcost analysis in transportation. Key elements for our debate are highlighted,including values of time, discount rates, the treatment of time, risk anduncertainty. The importance of network accessibility forms a key concernhere and, where possible, the analytical evidence is complemented by empiricaldata. The third major chapter at the heart of Part III is a microeconomicmodelling approach to transport infrastructure development and economicgrowth. It includes the theoretical framework, the modelling approach andsome results from a series of simulations and empirical studies.

As a complement to the analytical chapters, Part IV presents a series ofcase studies derived from secondary data. In each case, major examples aresupplemented by minor cases, with chapters on road investment, rail


investment and airport investment. The transport impacts are described, butmost emphasis is placed on the economic development impacts in the termsmost suitable to the project (e.g. development, employment, incomegeneration, etc.). The concluding chapter returns to the questions raised inthis chapter with a summary of whether they have been answered or not. Italso presents a new conceptual framework within which transport andeconomic development can be placed. Other avenues for research are alsohighlighted as the debate is continuing. This book provides the initial catalystto finding the answers to one of the great unresolved questions in transport.We do not pretend to be able to answer all the questions, but at least we willhave satisfied our own curiosity on the true nature of these links.

Notes

1 The same study also reports that a 1 per cent increase in per capita GDP resultsin a 0.3 per cent increase in household access to safe water, a 0.8 per cent increasein paved roads, a 1.5 per cent increase in power and a 1.7 per cent increase intelecommunications (Data for 1990, World Bank 1994). This, however, is thereverse causality to that examined in this book.

2 Most definitions of infrastructure concentrate on the economic infrastructurewhich includes services such as: public utilities: power, telecommunications, pipedwater supply, sanitation, sewerage, solid waste collection and disposal, and pipedgas; public works: roads and major dam and canal works for irrigation anddrainage; other transport sectors: urban and interurban railways, urban transport,ports and waterways, and airports. It includes the social overhead capital, whichcomprises those basic services without which all forms of productive servicescannot function. It encompasses activities that share technical features (e.g.economies of scale) and economic features (e.g. spillovers from users to non-users). Transport infrastructure comprises large-scale capital intensive naturalmonopolies that form a common element in the total costs of production andtravel. They form the basic network used for the movement of people and goods(based on World Bank 1994).

Scope of analysis

Definitions, approach andmethodological framework

2.1 Introduction

The literature on the relationships between transport infrastructure investmentand economic development provides ample evidence to suggest positivecorrelation between these variables. Anas (1995), for example, has shownhow reductions in commuting time, resulting from subway headwaysimprovements in New York City, will positively affect commercial andresidential real estate properties. Rephann (1993) has found a positive effectof road development on economic growth at the regional level. Aschauer(1989a, 1990) provides statistical evidence showing the effect of aggregateinvestment in infrastructure capital (including transport) on macroeconomicmeasures like GDP. While these and similar evidence has been criticized onanalytical and empirical grounds (Gramlich 1994), the fact remains that manyresearchers and policymakers strongly believe that transport infrastructuredevelopment will have a significant impact on economic growth at the urban,regional and countrywide level. Some of these arguments have been presentedin Chapter 1.

While we accept this view as a basis for our discussion we contend thatthere are a number of unsettled theoretical, methodological and empiricalquestions that need to be properly studied before one can conclude withcertainty that a given transport infrastructure has measurable and significanteffects on economic development. Key examples of such unsettled issues are: • How to correctly define and measure infrastructure development, given

that the level of transport services rendered by infrastructure facilities, isaffected by a number of supply-side factors such as physical, managerial(e.g. transportation systems management), economic and urban planningpolicies.

• The definition and measurement of economic development that is affectedby a large number of transport and non-transport factors.

• The nature of the causality mechanism by which infrastructure investmentis transformed into economic growth. Is there only one mechanism that

Chapter 2


can explain growth at the urban, regional and national levels? How doesit work and how does it evolve over time and space?

• Even if there are such mechanisms, are some investments more productive,in terms of economic growth, than others (e.g. rail vs. road), and if sohow can they be identified?

In this chapter we argue that a useful way to gain a better understanding ofthese issues is first to provide a conceptual and methodological frameworkwhich will enable us to examine these and related questions in a coherentand systematic way. Thus, the main objective of this chapter is to suggestsuch a framework and define and explore its various components. Then, wewill show how each of the questions like those presented above can fit intothis framework, thereby enabling a better and more productive analysis andmeasurement of the impact of transport infrastructure investment on economicdevelopment.

A major conclusion emanating from the introduction (Chapter 1) and thedevelopment of this methodological framework (Chapter 2) is theunderstanding that particular models or methods of impact measurement areonly adequate for specific problems and not for others. For example, we willargue that production function type models, used in the literature to measurethe relationships between the country’s level of infrastructure supply andGDP growth, are largely inadequate for evaluating the effect of infrastructuredevelopment on the economic growth of cities and regions. Similarly, modelsthat do not account for continuous changes in urban demography, the urbaneconomy and production processes cannot be used to predict the impact of anew infrastructure facility on urban economic development.

The structure of the chapter is as follows. In Section 2.2 we providedefinitions of the basic concepts and terms. Section 2.3 describes the keytenets of the conceptual approach which underlies the analysis in this book.In Section 2.4 we present the overall methodological framework proposedfor this book, including a detailed examination of the principal componentsof this framework relative to its structure and linkages with other components.Section 2.5 discusses the key methodological and analytical questions whichneed to be addressed in improving our understanding of the links betweentransport infrastructure investment and economic development. We end withan overview of the linkages between the contextual (Part II) and the analyticalparts of the book (Part III).

2.2 Basic definitions

The objective of this section is to clarify at the outset some of the basic conceptsand terms that will be used throughout this book.

The terms economic growth and economic development are usedinterchangeably in the literature and in the ensuing discussion when deemed

Scope of analysis 35

relevant; we will follow that practice. However, for analytical purposes it isuseful to distinguish between the two concepts with respect to their use inmodelling and measurement. The concept of economic growth is mainlyapplicable when examining the effect of the expansion of public capital onthe national economy. In this context it measures changes in the level of GDP(or GDP per capita) resulting from additional (gross) investment in the totalstock of infrastructure such as the roadway system, ports and airports, healthand educational facilities.

In contrast, the concept of economic development is used primarily whenexamining the effect of additional investment in specific types of infrastructureon the urban and regional economy. Moreover, this concept also encompassessome non-growth objectives such as changes in urban form, equity effectsand reductions in environmental quality. Thus, changes in regionalemployment (by type), adjusted for changes in location of firms andhouseholds, are used as a proper measure for assessing the effect of transportinfrastructure investment on local economic development. In general, weregard the change in economic opportunity resulting from accessibilityimprovements, which is capitalized in the form of a greater use of input factors,expanded output or enhanced welfare, as economic development.

Another key concept that needs to be clarified is transport infrastructure.Infrastructure is the durable capital of the city, region and the country and itslocation is fixed. In the transport sector it includes roads, railways, canals,ports, airports, communications (e.g. air traffic control) and terminals (orother interchanges). A key characteristic of the services rendered by theseinfrastructure facilities is their spatial dimension which is manifested by theirnetwork structure (e.g. road or rail networks). On the other hand, benefitsfrom these services may not be spatially ubiquitous and, in many cases, benefitssignificantly attenuate from the point of supply (i.e. the need to access a busor railway station).

The spatial dimension of transport infrastructure facilities also impliesthat they are natural monopolies over a large territory in that the per unitcost declines with increasing output (to a given capacity). Anothercharacteristic is their public good nature (again given capacity), since marketexclusion is not feasible in any practical sense for cost, social and legal reasons.Transport infrastructure facilities are durable, which implies significant sunkcosts as they often remain in place long after households and businesses haverelocated. Their indivisibility means that the capital investment costs are oftenvery high and the costs of additional capacity and maintenance are alsoconsiderable. Finally, there is a propensity for infrastructure facilities togenerate social and environmental externalities either through their particularlocation or through the services that they produce.

In summary, transport infrastructures have the following characteristics:they are networks; they form an indispensable part of the total productioncosts of goods; they have substantial elements of natural monopoly; their


capital costs are high, but their running costs are low and the sunk costsnecessary to establish an infrastructure are substantial (Kay 1993).

2.3 Conceptual approach

The modelling and measurement of the impact of transport infrastructureinvestment on economic development requires a specification of thetheoretical framework which underlies the potential causality links betweenthese two phenomena. Here we outline the conceptual background, followedby a more detailed methodological framework for the modelling of theselinks (Section 2.4).

The most fundamental outcome of an investment in transport infrastructureis the changes in the relative prices of accessibility of various locations. Sincethe network structure of transport systems makes accessibility spatially non-uniform, an investment in a new facility, or the improvement of an existingone, necessarily alters the present equilibrium structure of accessibility prices.This price change, in turn, implies changes in the relative advantage of spatiallylocated activities and the economic opportunities both for the productionand consumption sectors. The main reason being that the costs of inputs (e.g.labour) and the prices of outputs (e.g. housing) at alternative locations changeas a function of costs of accessibility to these locations. Furthermore, theextent and strength of various scale, scope and network economies, whichaffect the location decisions of firms, may become less pronounced as relativeaccessibility improves.

It should be understood that the degree to which an improvement intransport infrastructure affects economic development is not independent ofthe economic and demographic characteristics of the region where theimprovement takes place. A given change in accessibility will have a differenteffect on the location and consumption decisions (e.g. mode choice and use)of two-employee households than it will have on single-employee households.Similarly, retailers will react differently to accessibility improvements thanindustrial firms with regard to their location and use of labour. Hence, froma theoretical standpoint, the analysis of the impact of infrastructure investmentof economic development must consider the nature of the local economy andthe different actors that make decisions.

This argument is based on three fundamental premises. First, the investmentis an effective investment. This means that the investment has tangible effectson the performance of transport networks. Thus, investments whosecomposition, magnitude, type or location do not alter considerably theperformance of transport networks are assumed to be non-effective ones and,as a consequence, not to generate economic development. On the other hand,an investment that improves the organization and provision of transportservices in an area, even if it does not involve the construction of new facilities,is regarded as effective if it generates measurable effects.


Second, the causal linkage between infrastructure investment and economicgrowth must be manifest in changes in transport-economic behaviour. Thispremise implies that economic development ensues only if economic agentssuch as households and firms, as well as markets, react to the changes in theperformance of the transport network. In the short to medium term, thisreaction is confined to travel variables such as trip generation rates, travelvolumes or choice of routes. In the longer term, this reaction must also bemanifested in location decisions of households and firms and in changes inland and property prices. In total, changes in accessibility from infrastructureimprovements need to be accompanied by changes in economic behaviourand prices in order to constitute economic development.

Third, transport improvements which influence travel behaviour andtransport markets must eventually be transformed into measurable economicbenefits. These benefits include improved factor productivity, larger output,increased demand for inputs, increased property values and greater demandfor consumer goods.

Having stated these three important premises, it must also be stated that thedegree to which infrastructure improvements will affect economic developmentis obviously not independent of the level and performance of the inplace capitalinfrastructure. Thus, in areas where the stock of the transport infrastructure(e.g. roads, access roads, rail systems) is highly developed, even a sizeableinfrastructure investment is unlikely to affect travel behaviour and marketssignificantly and, as a consequence, economic development. In general, therefore,we can expect, ceteris paribus, a declining marginal ‘economic developmenteffect’ from additional infrastructure investment. At the extreme, when theregion’s transport infrastructure is fully developed so that any additionalinvestment will not improve accessibility, no economic gains from the investment(save for the multiplier effects) will result.

In summary, what the discussion in this section suggests is that the transportsystem can be viewed as a constraint on the attainment of economicopportunities within a region by households, commercial and industrialactivities. An additional investment in transport infrastructure facilities lessensthis constraint, thereby enabling the attainment of higher economicdevelopment1. As the region’s infrastructure becomes more developed, in termsof cost and ease of travel, the less is its binding effect on economic development.Hence, in analysing the effect of an additional investment on the local economyit is necessary first to assess its relative contribution to the region’s accessibility,through its effect on travel behaviour and markets.

An important theoretical question arising from the above discussion iswhether transport development constitutes necessary and sufficient conditionsfor local economic development. If transport can be regarded as a constrainton the attainment of economic opportunities in an area, then it can be regardedas a necessary condition. One of the main objectives of this book is to determinewhether and when it becomes a sufficient condition for economic development.


We now present the three major elements of this question: method of financing,equilibrium analysis and dynamic process.

Method of f inancing

First, transport investments require the use of real resources, such as labour,capital and land. If not financed from sources outside this region, it will benecessary to divert regional resources from other uses such as consumption,production or other forms of investments. Thus, if the region’s level oftransport infrastructure is well developed so that it does not impose a seriousbinding constraint on the local economy, any additional investment, in fact,may reduce the region’s economic well-being. For this reason, it is necessaryto account for the effect on the local economy of the way resources are raised(i.e. the method financing) for additional infrastructure investment.

The method of financing infrastructure expansion has some furtherramifications for local economic development. Consider two alternatives:first, the use of general revenues raised through local income or consumptiontaxes. Either tax will reduce households’ disposable income thereby affecting,inter alia, individuals’ propensity to travel. It will further affect the economy,as households’ demand for output will decline. Alternatively, inefficiency inthe form of dead-weight loss associated with such tax schemes can also beexpected. In addition, income tax may also affect labour to capital ratiossince it raises the cost of using a unit of labour. All of these effects have anegative impact on local economic development.

A second alternative is to finance new infrastructure investment by taxingtravel (e.g. hiking up petrol taxes or imposing road tolls). Such taxes, inevitably,affect the use of specific modes of travel (e.g. roads) thereby lowering thelevel of accessibility associated with the use of this mode.

Equilibrium analysis

In response to improvements in accessibility from infrastructure investment,firms and households can increase their demand for infrastructure facilities(the trip generation effect). They may also change their trip pattern (the tripdistribution effect); their choice of travel mode and their travel route (themodal split and trip assignment effects). They may relocate (the spatial locationeffect), or they may adopt all of these options. In turn, each of these effectswill influence the degree of use (and the quality level) of transport which is inexistence or which is being newly constructed (as well as accessibility). Thisdiscussion implies that, in modelling the effects of transport improvementson the local economy, it is necessary to carry it out within an equilibriumframework so that changes in the total demand for infrastructure facilities,resulting from changes in travel patterns and rates and in spatial location,are equilibrated with network supply.


Important factors affecting the equilibrium results from infrastructureimprovements are the interrelationships between economic development andtravel behaviour. If economic development increases per capita regional incomeand if travel demand is income elastic, then more travel will ensue from theeconomic development, induced by infrastructure investment. As aconsequence, we can expect a different state of equilibrium in comparison tothe case where travel behaviour is unaffected by economic development.Furthermore, the additional demand for travel may, in turn, generate asubsequent need for new infrastructure facilities.

Dynamic process

An important aspect of these interrelationships is that they are notinstantaneous and, in general, require considerable periods of time to transpire.The main reasons for this are the long periods necessary for investmentimplementation as well as the time needed for the demand side adjustment.Underlying these time lags are market imperfections including incompleteinformation concerning infrastructure development, uncertainty regardingthe behaviour of public authorities and private entities, high transaction costsemanating from imperfect land market and general market externalities. Allof these make the transformation of transport improvements into economicbenefits highly time dependent. The overall result is a dynamic process whoseevolution depends on the initial conditions of local transport and activitysystems and on the local transport and economic policies.

2.4 Methodological framework

On the basis of the above discussion, Figure 2.1 presents a schematic paradigmshowing the causal relationships between infrastructure investment andeconomic development. From this figure it is apparent that we conceiveinfrastructure investment as affecting network performance, transport marketsand externalities (e.g. pollution). It is, of course, a question of analyticalmodelling to show exactly the mechanism by which these effects come about.This issue is discussed at a later stage (Chapter 8). Whatever the mechanismis, the combined results (termed here as ‘accessibility effects’) are changes inrelative accessibility in terms of mode, network links, spatial location andtime of day. Accessibility effects, in turn, stimulate the so-called ‘real effects’.These include changes in factor productivity, changes in the location ofhouseholds and firms, changes in production and in consumption decisions,and changes in agglomeration economies. As explained below (Figure 2.1),the accessibility effects are further capitalized in the form of land rent andconsumer surplus. Thus, we can measure economic development (frominfrastructure investment) either through the ‘real effects’ or through the‘capitalized effects’. It is also seen that the investment generates the so-called,


‘multiplier effect’ which results from the infusion of a large sum of capitalinto the local economy. We do not regard this effect as part of the economicdevelopment from the transport investment. Lastly, in the longer run, economicdevelopment is likely to encourage further demand for travel that, in turn,will give rise to further infrastructure development.

The use of this paradigm for the analytical representation and empiricalmeasurement of the linkages between transport investment and economicdevelopment requires the inclusion of two additional factors: the geographicunit and the time span of analysis. That is, while the basic relationships shownin Figure 2.1 are assumed to hold at any geographic level and at any timescale, their actual analysis, modelling and measurement requires a differentdefinition and specification for each of the geographic levels and time periods.

Conceptually, we consider three geographical scales and two time periodsfor the analysis. These are: the urban level, the regional level and the country(or whole economy) level, and the short to medium run effects (up to 10years2) and the medium to long term effects (over ten years). This basic matrixof analysis is a 3×2 matrix of the geographic and time units, where each ofthe matrix’s cells contains the relationships expressed in Figure 2.1.

One major result from the adoption of this approach is that each of thefour main components in Figure 2.1 requires a different definition and adifferent model or method of analysis at each cell of the matrix. The analysisof the effect of infrastructure expansion on economic development at eachtime period and (in particular) at each geographic level calls for a specificdefinition of infrastructure investment and of economic growth. Similarly,the measurement of network performance and the modelling of transport-economic behaviour will vary as a function of the geographic unit and timescale. For example, at the urban level the addition of a new road or rail linkto the existing network in a specific area represents ‘infrastructure investment’while its short-to-medium run effect on the volume of retail business can beregarded as local ‘economic growth’. To measure such an effect, one needsaccessibility type models that link retail trade with travel times. Such definitionsand modelling approaches would, of course, largely be inadequate in thelong term at the regional or country level.

The distinction between the urban and regional levels of analysis issomewhat problematic, as often it is very difficult (if not outright impossible)analytically to separate the city from the region where it is located relative tothe incidence of accessibility and economic development benefits. Wenevertheless make this distinction since, in some cases, transport investmentsare being made with the explicit objective of benefiting a particular localarea such as a city or a county. Such an example is the Buffalo, New Yorksubway which was constructed with the declared intent to revitalize Buffalo’sdeclining downtown at the expense of outlying suburban areas (Berechmanand Paaswell 1983; Chapter 10).

The international level of analysis refers to the cases where an investment


links the transport networks of two (or more) countries. The Channel Tunnelis a prominent example of such an investment. While the economicdevelopment benefits from such an investment can be quite significant, it isnot a major concern of this book where the focus is primarily on the threemore local level impacts—the country, regional and local levels.

The above methodological framework permits a systematic typology ofvarious definitions, models and empirical results which have been reported inthe literature and relate to the effect that transport infrastructure investmenthas on economic growth. However, before we can proceed to discuss this inmore detail, it is necessary to elaborate further on the major factors whichunderlie the above framework, in particular, the suggested relationships between

Figure 2.1 The basic causality paradigm of the relationships between transport infrastructureinvestment and economic development.


transport infrastructure investment and economic development. Figure 2.1depicts the general structure of the methodological framework used in thisbook, together with the interrelationships between each of the main components.Next we discuss the nature of each of these four components.

2.4.1 The investment component

A fundamental concept in transport supply analysis is the network structureof all transport systems. Land transport networks can be viewed as the durablecapital component of delivery systems that connect spatially diffused activities.The formation of such durable capital facilities requires very high fixed costs(a substantial part of which is sunk costs). In the absence of significantcongestion this capital is a local public good. The provision of these capitalfacilities is normally carried out under conditions of scale economies (thusimplying the possibility of a natural monopoly) whereas their consumptionis associated with network, scope and density economies. Externalities arenormally associated with the consumption of services emanating from thesefacilities.

Adopting this view it becomes obvious that in assessing the effects of aninvestment in transport infrastructure on economic growth, several principalfactors have to be considered with regard to the in-place network. These are:the type of the investment, its relative size and its efficient provision andconsumption.

The type of the investment has two dimensions: its particular technology(e.g. rail vs. road) and its purpose. To see this within the concept of a network,consider the case of a rail investment that provides a missing link betweentwo previously disjointed rail networks. Such a project is likely to have amarkedly different (most likely a greater) effect on mobility and economicgrowth than a similar size investment in a new link in either of these twoseparate networks would have. A similar argument can be made for roadprojects or for a project that makes rail and road networks easily accessibleto each other. Therefore, it is the type of the transport technology (e.g. rail,road) and the nature of the investment (e.g. linking separated networks, amaintenance project or the addition of a new link) that matters in relation toits expected effects.

The magnitude of the investment is also crucially related to the concept ofa transport network. In general, even a large size investment (in monetary orphysical terms) usually represents only a modest expansion of the in-placenetwork and, as a consequence, will have small mobility and economic effects.Therefore, in general, when defining an investment in transport infrastructureit is necessary to assess its relative size rather than its absolute size. Definedin this way, the effect of a modest relative-size investment is likely to be quitesmall and localized. That is, in areas where the in-place network is welldeveloped, the economic growth effects from even a large investment are


likely to be confined to the urban level. On the other hand, the cumulativeeffects of such investments will be associated with regional or country leveleconomic development.

The efficiency of an investment is another major element that needs to beregarded when considering the effect of a particular infrastructure investmenton economic growth. The importance of this attribute of an infrastructureinvestment stems from the fact that it would be totally wrong to regard theeconomic benefits from an inefficient investment as the correct ones. Anillustrative example is the case when the direct transport benefits from theinvestment are rather small (present value of users’ time and operating costssavings are below the present value of the investment costs) and the only wayto justify the investment is to ascribe to it high economic development benefits.Under these circumstances it is correct to ask whether this particularinvestment should be carried out even though it can generate high economicdevelopment benefits.

In defining the efficiency of infrastructure investment we need to emphasizethat, since we speak of a public capital good, it is necessary to regard efficiencyfrom the point of view of the economy or the public sector rather than fromthat of the private sector. Thus, profit maximization is not an objective in theprovision of infrastructure whereas social welfare should be3. Given this viewwe can define the efficient provision of infrastructure facility in terms oftechnical efficiency, allocative efficiency or social optimum. The first definitionimplies an investment whose output (capacity) meets the demand for thisfacility, e.g. in terms of travel volume. The second implies optimal output atminimum costs. The third, implies that all social costs and benefits have beenaccounted for and a social discount rate has been used to discount futurebenefits and costs. With respect to the network concept, social optimumrequires that the benefits and costs of the entire relevant network be regarded,following the provision of the new investment.

An infrastructure investment can generate various production andconsumption externalities that need to be accounted for if social optimum isto be attained. A textbook definition of externalities refers to non-pricedchanges in the utility of a third party caused by production or consumptionactivities of direct users. Changes in the congestion, pollution or noise levelsfrom additional traffic, following an infrastructure investment, which accrueto non-users or to other trip-makers on other links of the network, are obviousexamples. On the other hand, economic development benefits emanating froman investment would not qualify as externalities since, under normalconditions, they carry a market price (e.g. the wage rate of additional labouror land rents from land development). In general, it would be erroneous toassess the economic effects from an investment without first considering thenon-priced externalities. For example, the optimal level of an investmentshould be computed only after the expected traffic in the expanded network(following the addition of the new capacity), has been subjected to congestion


tolls or to pollution charges which, in turn, will affect the actual level of thistraffic thus the optimal size of the investment.

In summary, it is a mistake to examine the effects of a particularinfrastructure investment irrespective of its type, relative size, its social efficientlevel and the externalities it may generate. All of these factors should beconsidered when viewing the investment within the network of which it is apart. Having defined the investment component of the paradigm depicted inFigure 2.1 (trigger mechanism), we now turn to examine the networkperformance component.

2.4.2 Network performance component

Four principal determinants characterize the network performance moduleof the analytical paradigm depicted in Figure 2.1 (performance andaccessibility). These are: accessibility and travel flows; savings of vehicleoperating costs; network effectiveness and intermodality (i.e. the interactionwith other transport facilities and technologies).

Beginning with the accessibility and travel flows determinant it is obviousthat, given the particular infrastructure investment, the performance of thenetwork is measured primarily by such factors as travel time savings betweenany pair of locations, by the resultant volume of traffic on each link and bychanges in the relative accessibility of locations i and j (i, j = 1,....., N). Similarly,users’ savings in vehicle operating costs such as in fuel and maintenance costsare also measures of network performance. While the magnitude of changesin these factors cannot be divorced from the attributes of the investmentdefined above, they constitute key indicators of network performance.

Since in this chapter we regard the ‘network’ as our ‘basic unit of analysis’,it is necessary to consider further the overall effectiveness of this unit. Threefactors define network effectiveness: positive network externalities; networkconnectivity; network efficiency. The first factor relates to the fact that asmore links are added to the network additional locations become accessiblefrom all other locations thereby making all potential users better off. In thisregard, the layout of the network plays a vital role in enhancing the network’seffectiveness. Network connectivity is defined as the number of alternativeroutes available to users who wish to reach a given destination location giventheir origin location. The larger is this number, the greater is networkconnectivity. Network efficiency is defined as the capability of a transportnetwork to process the area’s volume of daily traffic in terms of the relativelength of the peak period, or the percentage of total traffic that is processedduring the peak period.

A new investment in a transport infrastructure facility usually affects theuse of other infrastructure facilities, either because of cross demand elasticitiesor because two transport technologies are complementary or because thereare scope and network economies so that the expansion of one network reduces


the costs of using another one. The generic name used to describe these effectsis intermodality. To illustrate, the construction of a new road diverts trafficfrom other links, thereby reducing travel times in these links, but in the longerrun it generates more traffic in the entire network. The construction of atransport centre where rail, bus and auto systems meet can reduce connectiontimes, but in the longer run it may produce more traffic in each of thesesystems, thereby increasing travel times. The development of truck freightfacilities (e.g. truck parks) is likely to increase truck freight movement at theexpense of rail freight traffic, whereas the development of facilities whichmake rail and truck technologies complementary may increase overall freightmovement by truck and rail due to network economies.

Of particular interest in this regard is the alleged complementarity betweentransport and telecommunication technologies. Yet present knowledge in thisarea does not permit an unequivocal prediction of the effect of an investmentin one technology on the development and use of the other. Nevertheless, theability to use telecommunications to affect travel needs careful consideration.

At the urban level and in the short-to-medium run, accessibility factorsare the main network performance measurers. If, for example, accessibility ismeasured in terms of travel times on specific routes by alternative modes,then changes in relative accessibility between spatially distributed activitieswill define the performance of the network following the investment. In thelonger run, when transport and land use markets can be assumed to be inequilibrium, changes in the relative generalized prices for commuting and forfreight hauling are the appropriate indicators of network performance.

2.4.3 Transport-economic behaviour

This is made up of two principal elements (location and real effects, Figure2.1). First, there is the structure and performance of transport markets. Second,there is the response of land-use activity systems (mainly, households andfirms) to changes in transport markets relative to their spatial location andconsumption and production decisions. The essence of the transport-economicbehaviour component is thus the analysis of the interrelationships betweenthe transport and land-use markets.

As alluded to above, the distinction between short-to-medium andmedium-to-long run periods is predicated primarily on the basis of thetimespan required for markets to achieve a state of equilibrium. Since weargued above that infrastructure investments affect the relative prices oftravel between locations, the equilibrium number of commuter trips or theequilibrium volume of goods hauled between locations, by mode type andby time of day, will also change as a result. However, these may not be theonly changes. For example, the introduction of a new freight rail link isbound to affect not only relative prices of freight movement in a given area(city or a region), but also the degree of competitiveness in freight markets.


Depending on demand and supply conditions, these markets can evolveinto becoming more competitive or more monopolistic. For example, whenlarge-scale economies are associated with rail operations. In either case, inthe long run, this effect will determine equilibrium prices and quantities intransport markets. Since economic entities like households and firms maketheir decisions on the basis of these factors, economic growth (which is theultimate manifestation of these decisions) is also influenced by theseequilibrium conditions.

Almost all of the theoretical models developed for analysing the localeconomic results from transport development have regarded the costs of travelas a key input factor in private production processes. Thus, a major premiseunderlying the voluminous literature on urban and regional development isthat changes in the equilibrium prices and quantities of commuting and haulagewill affect the spatial locations of households and firms (e.g. Henderson 1977;Muth, 1985). In particular, if the development of the transport infrastructureleads to lower generalized costs of travel then, holding all other factorsconstant, these models predict decentralization of households and firms intothe hinterland.

The response of land use activity systems in the form of classic locationtheory argues that land, labour and capital are the primary inputs to theproduction process, and that the use of land is determined by the relativeimportance of each of the inputs. Transport infrastructure determines therelative accessibility of places and hence has a major impact on the locationof industry through the implicit trade off between accessibility and land values.Some industries need to locate in the more accessible sites and so are preparedto pay more for that competitive advantage. Others can locate moreperipherally as accessibility is less important. All the models developed arebased on equilibrium theories of land allocation and optimality in decision-making (Isard, 1956; Alonso, 1964).

These arguments have been developed with respect to traditional forms ofmanufacturing where agglomeration economies and economies of scale areimportant. They may be less appropriate for the new forms of manufacturingwhich have developed over the last twenty years where industry has becomemuch more selective in where it locates. Access to suppliers and marketshave become less important than the availability of a skilled labour force,suitable land for development, a high quality environment and car basedaccessibility. All these factors have led to a growing demand for developmentat sites on the periphery of major urban areas. These are now the mostaccessible locations. Conversely, low-skilled production tasks can now becarried in inaccessible peripheral locations or in low-income economies wherethe costs of labour are much cheaper. The post-industrial organization andthe structural changes taking place in industry, together with technologicalchange, make it difficult to isolate the impact of transport infrastructure onlocation decisions of firms and people. Economies of scope have become


more important in the new manufacturing and service processes thantraditional economies of scale (Banister, 1994).

Investment decisions in transport infrastructure will also affect theconsumption and production decisions by households and firms. These canbe seen in a number of ways. First, as explained earlier, infrastructure facilitiesare regarded by private firms as a free of charge public capital. In theirproduction process, this public capital may be a substitute or a complementaryfactor for private capital as well as for other input factors like labour. Theexpansion of this public capital will affect both the level of output and theuse of other input factors including labour and private capital. A second wayby which transport infrastructure capital development affect private firms’output is through its effect on agglomeration economies. Improved accessibilitycauses firms to relocate which, in turn, affects agglomeration economies,thereby their level of output. Third, better accessibility also affects thepropensity of households to supply labour, used by firms as a key input factor.Theory maintains that households equilibrate marginal utility (earnings) fromadditional hour of labour including home-work-home time of commute, withthe marginal utility of leisure time consumption, given a 24 hours timeconstraint. Reducing commute time, by investing in a transport infrastructure,will enable households to achieve another equilibrium with more hours ofwork supplied (Berechman 1994).

As with the other components, the exact modelling of changes in transportmarkets from infrastructure investment and the interaction of these marketswith land-use activity markets will vary as a function of the geographicaland time units considered. At the urban level and in the short-to-mediumrun, lower costs of commuting will encourage households to suburbanizewhile largely not affecting the location (and output) of industrial andcommercial firms. Hence, partial equilibrium models and costbenefit analysestype models are used to examine the effect of infrastructure expansion onurban development (e.g. Nijkamp, 1986; Forkenbrock and Foster 1990;Huddleston and Pangotra 1990). At the country and long-run level of analysis,expansion of the transport infrastructure can be expected to affect equilibriumprices and quantities in transport markets and, as a result, the equilibriumuse of inputs and level of output by firms, as well as the supply of labour byhouseholds. Spatial changes are not regarded at this level of geographicalanalysis. It is for these reasons that economy-wide aggregate productionfunction models, in which the stock of public capital is an input factor, areused to explore the effect of infrastructure investment on economic growth(Aschauer 1991; Munnell 1992).

2.4.4 Economic development component

Given the theoretical and empirical models used for the analysis of theserelationships at each level of analysis (the transport-economic behaviour


module), the primary issues dealt with here are the proper definition andmeasurement of economic growth (Figure 2.1). It is possible to argue that theuse of a model like an aggregate production function, where change in totaloutput (e.g. GDP) is the explained variable, also defines economic growth.While under certain conditions economic growth can be defined and measuredas changes in total output, this is certainly not the case for the urban andregional level of analysis in the short-to-medium and in the longer run. Thisis so because of the need to separate efficiency gains from transfer effects andspatial effects from production and consumption effects4.

A significant feature of all infrastructure projects is their investmentmultiplier effect which stimulates local use of factors and demand for finalgoods and is a function of the size of the investment but not of its type. Thatis, a capital investment in an infrastructure facility of any kind (e.g. health,education or recreation), at the urban and regional level, necessarily infusesthe local economy with a substantial amount of funds. These funds, in turn,stimulate this economy in terms of firms’ demand for labour and to otherinputs and in terms of consumers’ demand for goods and services. Theseincreased demands, further stimulate economic activity—the multiplier effect.Since we are interested in the efficiency gains solely from the transportimprovements, it is necessary to differentiate these gains from the multipliereffect which, in essence, are transfer payments.

Given these observations, economic growth can be defined and measuredin a number of alternative ways, depending on the geographic and time levelsof analysis. Thus, at the urban level in short-to-medium run economic growthcan be defined as increased employment, by type (e.g. Dodgson 1974). At themedium-to-long run this change in employment should be adjusted to firms’and households’ locations to reflect changes in spatial equilibrium (Berechmanand Paaswell 1994). Econometric cost and production models were used toexplore effects of investment in public infrastructure, though changes in spatialequilibrium patterns were not accounted for (Munnell 1990b). Multiregionalinput-output models are also used to gauge the economic effects of additionalcapital expenditures, where regional output is the main indicator of economicgrowth (Rephann 1993). At the country level, long-run level of analysis, partialand full factor productivity are common growth indicators (e.g. Bajo-Rubioand Sosvilla-Rivero 1993). Social rate of return is another measurer ofeconomic growth at this level (e.g. Garcia-Milà and McGuire 1992).

A further complication is the fact that while most, if not all, of the studiesof economic growth from transport infrastructure analysis are static ones,the phenomena of infrastructure expansion and economic growth areessentially dynamic in nature. Thus, from a policy analysis viewpoint, ratherthan question what the economic development effects from a giveninfrastructure policy are, it is more useful to ask what the dynamic evolutionor economic growth path of this policy is. Very few studies actually tried totackle this question (see review by Gramlich 1994) and it remains as a


significant research question to delineate growth path from continuousinfrastructure development. Two important elements of the debate will bementioned here.

It is very difficult to generalize about the importance of transportinfrastructure, as a causal factor in bringing about economic growth indepressed or peripheral regions. Although much of the evidence is notsupported by clear analysis, it seems that transport is only one of theingredients necessary to generate growth (Botham 1983), particularly if thatgrowth is to be sustained over a period of time. Directing transport investmenttowards disadvantaged regions may improve their relative economic positionprovided that other policies are also in place to promote greater efficiencyand productivity (Hart 1993). Past studies have attempted to assess the impactof infrastructure on the cost structures and hence the competitiveness of localplants and companies. However, there is little evidence that new transportinfrastructure significantly lowers the costs of production (Diamond andSpence 1989). When a firm relocates, transport is normally only a secondorder consideration as these costs are a small part of total production costs.However, these costs may be important in sectors such as retailing and certainservices where accessibility to customers may influence a firm’s performance.A second impact may result from the reduction of transport costs and itseffect in reducing overall production costs, which in turn increases profit andoutput. However, if transport costs are a small part of total production costs,the impact here is, again, minimal. Most of the research (Parkinson 1981)has related to the road sector, but similar conclusions could be drawn fromrail and urban transport investments (e.g. metro and trams).

The second important issue is the current debate over the impact oftransport investment on levels of traffic demand. The traditional argumenthas been to build new infrastructure to promote development and to relievecongestion as both factors will, in turn, lead to economic growth. However,counter-arguments (SACTRA 1999) now acknowledge that in many situationsroads induce more traffic. This induced traffic includes generated trips, existingtrips to new destinations and longer journeys. The question here is whetherthat induced traffic actually increases economic growth, or whether its impactis neutral or even negative. If new investment in road only leads to moretraffic without any material increase in output or productivity from theeconomy, then can it be justified? Such debates have important implicationsfor investment and pricing strategies, given that in many high-incomeeconomies the elasticities of demand for travel are low and that improvementsin accessibility may only result in a new congested equilibrium being reached.Short-term relief from congestion results in new patterns of activity beingundertaken, which in turn result in more travel and further congestion.

In summary, in addition to the quite complex modelling and data issuesinvolved in economic growth analysis, the key questions in evaluating theeconomic consequences from transport infrastructure investment are: how


to separate the measured growth effects from transfer and spatial effects andhow to do so within a framework of equilibrium analysis.

2.5 Discussion and overview

Before moving into the substance of the book, it is important to emphasizecertain arguments, which seem to have become ‘accepted’ by not really beingchallenged. It is often assumed that those countries, regions or cities whichattract a high proportion of transport infrastructure investment will have acompetitive advantage over those locations which have been less successfulin obtaining investment. This argument started with the Rostow (1960a) andFogel (1964) exchange about whether transport is a prerequisite for economicgrowth or part of the process (discussed in Chapter 1).

Transport growth has been linked directly to growth in gross domesticproduct (GDP) and one of the major determinants of future levels of transportdemand is the assumed increase in GDP. Historically, a 1 per cent growth inGDP has led to a 0.93 per cent growth in freight (including a 1.74 per centgrowth in road freight) and a 1.24 per cent growth in passenger traffic(including a 1.40 per cent growth in car traffic). Although the link betweenenergy consumption and economic growth was broken in the 1970s, the linkbetween transport and economic growth still remains (Short 1993). Thecurrent environmental imperative (Chapter 5) questions whether this linkexists in practice and, if so, whether it should be broken. If global warmingreduction targets are to be achieved, then there is a strong case for reducingthe transport intensity of the economy (decoupling, Chapter 12).

Transport costs as a proportion of total production costs in many industriesare relatively small, and other factors such as availability of skilled labour,suitable sites, government grants and a quality environment may all be moreimportant than transport. Any decrease in transport costs may not be reflectedby cheaper prices to consumers. Savings may be absorbed by entrepreneursor landowners through higher profits and rents, or they may be absorbed byemployees in higher wages (Chapters 4, 7, 8). Again, there are no simpleanswers. A combination of these factors is likely, but it then becomes a verydifficult measurement problem. Even with large-scale transport infrastructureinvestment, one is expecting measurable changes across a wide range ofindustries, but to isolate the effects on profits, rents, wages and cheaper pricesis almost impossible, particularly given transport’s small size in totalproduction costs.

Most of the literature mentioned in the earlier part of this chapter (andChapter 1) relates to traditional manufacturing industry, not to the new service-based economies. Here again, the transport dimension of total productioncosts becomes obscured as many transactions are carried out remotely andthere is a high level of complementarity, even between competing networks.Transport and communications networks do not compete with each other, as


the most accessible locations are those where several of the networks actuallycome together, as can be seen, for example, in Lille (France) in rail networks(Chapters 4, 10). The same types of development at a multi-network levelcan be seen at Charles de Gaulle Airport near Paris where the road, rail andair networks all work together to promote a vast new commercial and businesscentre at Roissy. There are synergetic effects between and within networks,so the relationships are not necessarily linear. Measurement must also bebroader than just the use of the network as access it also relates to theopportunity to use it and the possibility of being linked to a potential customer.The value of telecommunications and transport networks relates both to yourown access and the number of other people who also have access to it. Inuncongested conditions, their use of the networks is not relevant, but incongested conditions the levels of use and price for access to the networksboth become more important (Chapter 8).

There are many arguments from traditional location theory and more recenttrade theory (Krugman 1991a) about where new development is likely totake place. Geographic concentration relies on the interaction betweenincreasing returns, transport costs and demand. With sufficiently strongreturns, each manufacturer will serve a national market from a single location.Krugman (1991a:41) argues that it is the interaction of increasing returnsand uncertainty which ‘makes sense of Marshall’s labour pooling argumentfor localization’. Labour market pooling, together with the supply ofintermediate goods and knowledge spillovers, ensures economies oflocalization at the regional and city levels. People can change jobs withoutmoving houses.

In a more general sense, Krugman (1991a) also argues that patterns ofdevelopment reflect the culture of Europe, not the geography. The richerregions are closer to large markets that are themselves the richer locations.There is a circularity here which may reinforce the dominance of the centre,even if access is improved to the low wage peripheral regions in Europe throughroad and rail investment. A reduction in transport costs allows firms to locatewhere it is cheapest, but it also facilitates concentration of production in onelocation to realize economies of scale. As Krugman (1991a:46) suggests, ‘itmay pay to concentrate (production) at the location with higher costs butbetter access’. Low transport costs result in production at peripheral low costlabour markets, and high transport costs mean production in many locationsserving local markets. Intermediate transport costs result in production beingcentred where labour costs are highest.

Economic theory suggests that, in urban areas, all benefits from transportimprovements are capitalized as ‘consumer surplus’ and as ‘producer surplus’,the latter referred to as rents or, in the case or urban land market, as ‘landrents’ (Anas 1984). Thus, in theory it is possible to regard the sum of consumersurplus and land rent as the total benefits from a transport project and comparethem with its costs in order to assess the projects profitability for the economy.


However, the actual measurement of these effects is quite problematic bothon theoretical and empirical grounds and in many cases we can only measureland rents which vastly underestimate total benefits from the project (Mohring1994). In a study of the effect of reducing New York subway rush-hourheadway from ten to five minutes Anas (1995), found that over 97 per centof total benefits from this change is increased in consumer surplus, while lessthan 3 per cent is capitalized into the rent of housing and commercialproperties. For this reason transport analysts typically enumerate all possibleinternal and external changes generated by a specific project and subsequentlydefine them as benefits or costs from the project. This approach leaves opentwo cardinal questions: Does the list of changes exhaust total benefits fromthe improvements? Does the division of the internal and external changesinto benefits and costs give correct solutions?

Underlying much of the argument over increased local developmentresulting from transport investment is the change in accessibility, or the easewith which people can travel to and from a particular location. Traditionally,improvements in accessibility have been viewed as a benefit to the local areaas it becomes more attractive as a location to live and work in, and as propertyand land values rise. However, the empirical evidence now suggests that theaccessibility changes are relatively small, particularly in a dense network ofroutes, and that the impacts are highly localized around the new facility (e.g.rail station or airport). A new investment may give a short-term relief tourban congestion, but the additional capacity resulting from the reallocationof travel from existing congested links will quickly be absorbed as a newcongested equilibrium is reached.

The changes in accessibility resulting from new investment in an alreadydense and congested network will not be of a sufficient scale to have a majorlong-term impact on the local economy. It is unlikely to be of a sufficientscale to attract major new employment into the city. The impact may be toencourage longer distance travel out of the city as the new investment willmake other locations more accessible. Accessibility works in both directions.There seems to be a scale element here, as the investment must be of a sufficientscale and located in an area with particularly poor accessibility to have ameasurable impact. It must have a demonstration effect as well as anaccessibility effect. In this way it may increase one location’s accessibility,relative to another area’s accessibility. But such cases may be rare in thedeveloped world where transport infrastructure is fairly ubiquitous. It is onlyin the developing countries that the changes in accessibility resulting frominvestment in new infrastructure will have a major impact on regional andlocal development.

More important is the complementarity found within networks.Accessibility tends to be viewed as the impact of one new link on the networkas a whole. However, many investments are strongly complementary and donot need to be consumed in fixed proportions as they form systems.


Competition is really taking place between systems and not between individualproducts. So accessibility should not only be viewed as the changes in oneparticular system (e.g. rail), but the new competitive position of that systemin relation to other systems (e.g. road). Lille again provides a good example,as Euralille provides an interchange between international Eurostar railservices, national TGV rail, regional rail and local VAL and tramway systems.The real value of improvements in the quality of the network is that it providesthe opportunity for people and businesses to take part in the network, even ifthey choose not to. There is an optionality benefit. The value of membershipto one user is positively affected when another user joins and enlarges thenetwork (Katz and Schapiro 1994). New concepts of networks andaccessibility are required to determine under what conditions the competitiveposition of one network will be changed as compared with another on atleast three dimensions—to influence expectations, to facilitate co-ordinationand to ensure compatibility.

Locations with poor quality transport will be significantly disadvantagedwhen compared with locations with high quality transport infrastructure.Yet, much of the evidence cited here suggests that investment which doestake place only marginally affects accessibility. It seems that changes at themargin may not be sufficiently large to result in a location change or in thatlocation becoming uncompetitive. Accessibility may be a surrogate for abroader more important issue, namely the image of the area. It is not theimpact of investment on the production costs and the competitiveness of afirm that is important. The changes will not show themselves directly in thebalance sheet.

Change and location is a much more subtle process. We would argue that theimpact of investment (or lack of it) is important in establishing the image of anarea and hence its attractiveness to new development. This in turn will have animpact on the local labour market so that high quality (and high income) labourwill be attracted. Transport investment may act as the trigger mechanism to thisprocess. The alternative explanation seems to lead to the conclusion that onlyexisting locations will ever be attractive as they have first mover advantages andwill always be more accessible than other new or peripheral locations.

There is a basic uncertainty (and perhaps confusion) over whether changesin accessibility lead to new development or a reallocation of developmentfrom other areas. There may be no absolute gain, as a fixed total of activityis reallocated as a result of the transport infrastructure investment. If this isthe case, then there may be a regional development argument for transportinfrastructure investment in peripheral regions or in local economies whereindustrial restructuring is taking place, provided that it can also bedemonstrated empirically and theoretically that industry then moves to theseinvestment locations.

These first two chapters have given a broad overview of the main debatingpoints within the rich research area of transport investment decisions and


economic development. There are no clear or simple answers, just a series ofcomplex questions. We have tried to structure the debate and give a flavourof the uncertainties and controversies that exist. We have also tried to developa conceptual framework within which analysis can take place. This has meanta simplification of the processes at work, together with a clear differentiationbetween the various scales of analysis and the role of time in those processes.

The next part of the book (Part II) elaborates on the contextual issues thatalso influence the analysis. We would argue that these additional factors, notnormally considered in analysis, have altered the main arguments, in somecases fundamentally. So even if the traditional arguments were appropriateto the situation ten or fifteen years ago, they are not so relevant today. Newforms of investment mean that the traditional roles for the public and privatesectors is changing. The move from the industrial to the post-industrial societymeans that new forms of production and consumption are evolving. There isnew concern over the spatial, social and distributional issues, together withthe environmental imperative. These changes all mean that decisions are nowinfluenced by both economic and non-economic factors. Previously, theeconomic factors had been the primary concern in the analysis of transportinvestment and economic development. These contextual issues form the basisfrom which the analytical part of the book is developed (Part III). Here, theimportance of scale is primary as we move from the macroeconomic analysisand the assessment of transport projects in the context of economicdevelopment to micro level analysis where this book makes its majorcontribution. Table 2.1 summarizes the linkages between the contextual andanalytical parts of the book.

Notes

1 If travel costs were zero everywhere, attempts by firms and households to optimizetheir location in order to maximize profits or utility would become nonsensicalsince it would be possible to carry out necessary interactions (e.g. getting fromplace of residence to place of employment) at zero cost.

2 We consider ten years as the time it takes for land use and travel markets toconverge to a state of equilibrium following an external change.

3 Facilities like roads where revenues from road charges or from developmentrights are used by these entrepreneurs to pay for the capital costs. Under sucharrangements, and given a profit maximization objective, a social welfaremaximizing road is unlikely to be built. A correct approach would be to definean efficient road, from the economy viewpoint, and then recruit private firms tobuild and operate it (Banister et al. 1995; Gomez-Ibanez and Meyer 1995).

4 Even at the aggregate country level, in many cases, it is important to separatethese effects. First, such models can be sector based models so that positive growthimpacts in one sector need to be balanced against changes in others (Deno 1991).Second, transfer effects occur at the country level as well. Third, when the country’seconomy is strongly affected by international trade, infrastructure developmentcan encourage cross-border relocation of firms as the many USA-Mexico cross-border studies have shown.

Tabl

e 2.

1 Li

nkag

es b

etw

een

the

cont

extu

al a

nd a

naly

tical

par

ts o

f the

boo

k

Contemporary issues


Part of our main argument for revisiting the key debates over transportinvestment and economic development is to incorporate some of the majorchanges that have taken place in society over the last thirty years. We havedivided these changes into three. The first relates to the nature of transportinvestment and how that has changed, particularly in western economieswhere there are already extensive high quality networks. Much of the debateis now over replacement of infrastructure, rather than new investment, andmost recently the traditional role that the public sector has played in providinginfrastructure has been questioned. All governments are concerned about theincreasing levels of public expenditure and the proportion of publicexpenditure in relation to total expenditure. The public goods argumentsrelating to transport are being questioned, as is the right to free use of theinfrastructure at the point of delivery. More fundamental is the issue of whoactually owns the transport infrastructure, what rights do individuals andcompanies have over access to them, and what are the associated propertyrights.

Secondly, the debates go much wider than the actual infrastructure andinclude the changes that are taking place in the economy itself. We are nowin a transition phase from a manufacturing based economy with a majorservice sector to one that is information based. Technological change willhave a fundamental impact on the rationale for work, the location ofworkplaces, the time available for leisure activities and the way in whichpeople organize their lives. This in turn will affect the structure of cities, thebalance between rural and urban areas and the ability of firms and industryto locate wherever they wish. Traditional locational constraints have beenbroken.

On a national and international scale, there are two complementary forcesat work. Economic activity is becoming global in its scale and location asmultinational companies increase their market share and as productionprocesses are changed with ‘flatter’ organizations and a substantial amount

Part II

58 Contemporary issues

of outsourcing. Manufacturing and assembly are carried out where appropriateskills are available at the lowest cost (including transport costs). Companiesare becoming more fragmented both structurally and spatially as the newtechnological infrastructure, together with a high quality transportinfrastructure, permit global production processes.

The second force is that of demographic change as the growth in totalpopulation within western countries stabilizes. However, the structure of thatpopulation is fundamentally altering as people live longer and traditionalfamily structures become weaker. The implications of these external factorson the demand for travel, on location and on economic development areprofound. This, in itself, would justify a review and re-analysis of the debateson the links between transport investment and economic development.

The third change has been the much greater priority given to distributionaland environmental issues. Although governments are primarily concernedwith economic growth and maintaining or increasing competitiveness, thereis also a need to make greater efforts to improve levels of equity (both socialequity and spatial equity), and to reduce the environmental impacts oftransport. Regional development and concerns over the problems of particulargroups within society have always featured highly in national (andinternational) policy statements. But with the reduction in the role of thestate in investment and the aim of reducing levels of public subsidy, a lowerpriority has been given to these activities. Macroeconomic policies of reducinglevels of taxation have taken preference over the broader welfare policiesfollowed in the past.

In addition, the environmental debate has become more important as theglobal dimensions have been added to the local pollution concerns, and asconcerns have risen over the possible links between levels of car-based mobilityand health. These debates have been widened as governments have given ahigh priority to sustainability and sustainable development. To achieveobjectives related to the environment and sustainability, actions need to betaken to increase the efficiency of travel through using less energy andproducing less emissions. This can be achieved through a combination ofpricing, technology, regulation, planning and raising levels of public awarenessand support for action.

All these issues will be discussed in detail to give a clear picture of thechanges that have taken place, together with the implications for the linksbetween transport investment and economic growth. This forms the contextwithin which the more detailed methodological, modelling and empiricalresearch can be placed. It also provides the rationale for a ‘new look’ at theseimportant relationships.

Transport infrastructureinvestment

3.1 Introduction

In Chapter 2.2 we introduced the basic definitions of transport infrastructure.Here we elaborate on these concepts, again starting with the broad definitionsgiven by Kay (1993) who states that transport and other infrastructures havethe following characteristics: 1 They are networks involving delivery systems and there are substantial

interactions in the provision of services to individual customers.2 They form a small but indispensable part of the total costs of a wide

range of products in which they are used. Thus, the losses that resultfrom service failure are often very large relative to the basic cost ofservice provision.

3 They have substantial elements of natural monopoly. Competitiveprovision of the infrastructure is costly, often prohibitively so. This neednot exclude competition in the use of infrastructure.

4 Capital costs of infrastructure are generally large relative to their runningcosts.

5 The sunk costs of establishing an infrastructure are substantial. A highproportion of the total cost of a service has already been irrevocablyincurred before that service is offered.

Several examples of infrastructure can be identified, some of which have allfive characteristics, while others only have a few elements. The distributionnetworks of public utilities and the development of road and rail systemsgenerally meet all five conditions. Activities that meet some of the criteria aresometimes not categorized as infrastructure. For example, postal services andpayment systems in financial services have several features in common withutility distribution systems. They involve networks, have substantial sunkcosts, are strategic, indispensable and have a relatively low unit cost (De Ruset al. 1997).

The success of cities and regions has always been based on the quality of their

Chapter 3


infrastructure. This in turn requires a commitment over a long period of time tocontinued new investment and replacement of existing stock. Infrastructure isthe durable capital of the city and a country, and its location is fixed: in thetransport sector it includes roads, railways, communications (e.g. air traffic control)and terminals. As the World Bank (1994) suggests, one should take a very broad-based approach to infrastructure as it covers all the social overhead capitalnecessary for development (including education, health and nutrition), ratherthan the narrower definition of economic overhead capital (including roads,sewerage, water and utilities). Often the services that are obtained from theinfrastructure have a spatial dimension (e.g. the distribution of the rail network),with the benefit from that service declining as distance from the supply pointincreases (e.g. stations). Their key characteristics are: many people benefit froma single infrastructure; they can be used over and over again; the infrastructureremains when people and businesses move in and out of an area and it providesthe means for integration and co-ordination of activities over time and space.The infrastructure forms the arteries of cities and nations and the communicationssystems are the nerves. The health and prosperity of cities and nations dependson the quality of these networks (Banister 1993b).

Infrastructure is a capital good for which users do not pay the full marketprice and is often perceived as a source of external economies (Youngson1967; Lakshmanan 1989). Provision of infrastructure leads to a high cost forthe first user and a small marginal cost for additional users (Diewert 1986).Improved transport infrastructure influences production and consumptionpatterns as it reduces transport costs and travel times. As such it canredistribute benefits among economic groups and between regions. It isinteresting to note here that it is extremely difficult to build any new transportinfrastructure in many countries as people do not think that there are anybenefits to them, either directly or indirectly. This relates to both road andrail investment. Public perceptions of the benefits seem to differ to the claimedeconomic benefits.

Infrastructure investment may reduce the need for capital and labour, asproductivity is improved. These impacts are felt at both the firm level (can beinvestigated through microeconomic analysis) and at the regional and nationallevels (can be investigated through production function models). Infrastructurecan also influence employment and private capital through changes inaccessibility and marginal transport costs and the possibility of privateinvestment in infrastructure. These issues can be investigated through regionaleconomic models and through surveys among entrepreneurs on their locationdecisions (Rietveld and Nijkamp 1993).

In their conclusions, Rietveld and Nijkamp (1993) make some generalremarks about infrastructure: 1 Infrastructure is a generic term that needs to be carefully qualified to

make it suitable for focused policy analysis. Additional infrastructure in

Transport infrastructure investment 61

regions where there are already good quality transport systems does nothave the same impact as where the existing network is sparse or of apoor quality.

2 The links between transport and land use need to be clarified: for example,large-scale concentrations of public/private activities (e.g. offices orwarehouses) may be a response to existing transport infrastructures, butit may equally be a result of patterns which would have occurred in anycase. This is a matter of cause and effect.

3 Infrastructure is subject to decreasing marginal productivities. When aregion has good links, any addition to the network will have aproportionally reduced impact. An extensive high quality network mayjust make more industry footloose and this in turn reduces the importanceof location as a decision factor by firms and individuals.

4 New types of high quality infrastructure may have a significant impact:for example, the new European high-speed rail network may revolutionizetravel around the continent as happened when the first generation ofrailways were built over 150 years ago.

5 Improvements in transport infrastructure may not be a sufficient reasonfor regional development as many intermediary factors also play a role.The impacts may also be redistributive as well as generative. The gainsin one region or city may be at the expense of another, so the overalleffect may be neutral or even negative.

6 Improvements in transport infrastructure lead to a decrease in transportcosts. This advantage may be absorbed by entrepreneurs or land ownersin the form of profits or rents, or it may be absorbed by employees inhigher wages. It could also be passed on to consumers as lower prices. Acombination of the above alternatives is likely, but there is no method toestablish the most appropriate distribution.

7 Infrastructure is a multidimensional phenomenon and there may besynergies between various types of infrastructure. This has been recognizedat a theoretical level, but little is said on exactly what these synergiesmight be.

8 Infrastructure analysis has focused mainly on firms, but not on households.These two are interrelated, but little analysis has looked at the combinedinteractions between firm’s and household’s location decisions.

From these different perspectives, the wide-ranging features of transportinfrastructure are clear, but in practice the operational definitions mainlyrelate to their form and function. In this book, we take the broad definitionof transport infrastructure to include the networks and terminal facilitiesused for the movement of people and goods. Their characteristics include: • large scale capital intensive natural monopolies;• common element in the total costs of products or travel;


• road, rail and air infrastructure, but many of the points made would beequally applicable to other forms of transport infrastructure;

• all scales of infrastructure. The scope of the analysis here explicitly deals with the national/international,regional and local scales and the analysis and case studies presented in therest of the book have been selected to explore the full range of infrastructuretypes.

As can be seen from the introductory section in this book (Part I), therehas been a history of interest and debate over the links between transportand economic development. The importance of the issue cannot be questionedas it underlies much of the rationale for investment in the transportinfrastructure. We also take as given that a minimum quantity and quality ofinfrastructure is essential as twentieth-century production, communication,employment and wealth patterns depend upon mobility and transport. Theevidence cited in Part I from the World Bank and other sources is clear onthis. Our interest lies in revisiting these old debates to establish whether theyare still valid in advanced economies and to investigate the new agenda.Consequently, most of the remaining analysis and the case studies focus onthe theoretical and empirical evidence from the developed world. In eachcase there is a variety of measures and so the impact of transport infrastructureinvestment is unlikely to be consistent either in its scale or direction. Thiscomplexity is inevitable and makes it inherently difficult to come to simpleclear conclusions. It also makes causality difficult to assess. But it does allowa clearer realism within the analysis and avoids over-simple conclusions.

3.2 Type of investment

It is often argued that most transport infrastructure investment has beentraditionally the responsibility of the public sector, with only limitedcontributions from the private sector (Chapter 3.1). This is true if the restrictivedefinition of the network is used as the basis for analysis. But even here,much of the original transport network has been built with private capital.Public involvement has mainly been seen during the twentieth century.However, many other types of investment in transport infrastructure are stillin the private domain. Many transport interchanges and terminal facilities(e.g. rail stations, car parks and airports) have been private sector investments,as have some transport vehicles and communications systems. The questionof finance and the arguments about the particular role of the public andprivate sectors are covered in Chapter 3.3. Again, it is clear that some of thetraditional criteria for the involvement of the public sector as the prime agencyfor investment finance are weakening and the possibilities of partnershipfunding are increasing.

Second, in many western countries and cities there is now little new


investment in the transport infrastructure as few new roads and railways arebeing built. Small-scale additions to the network are taking place (e.g.bypasses), but the 1980s saw the end of the first great period of motorwayconstruction in developed countries. In the UK the symbolic opening of theM25 orbital motorway around London in October 1986 marked the end ofnew construction (Chapter 9.2). The rationale for this conclusion to motorwaybuilding was threefold. First, the network had been completed and the maincentres had been joined by a high quality restricted access road network thatin many cases followed the original rail network. It is now being duplicatedin some countries by the high-speed rail network. Second, the funding fornew roads was being substantially reduced as national governments cut backon public expenditure, as the road-building programme had always beenunpopular and as public budgets have come under increasing pressure. Third,with the new arguments on environment and sustainability (Chapter 5), itwas no longer seen as desirable to build more new roads, particularly asrecent evidence (SACTRA 1994) suggested that they induced new traffic. Itis unclear how long this period of ‘no new construction’ will last.

Most investment is now being channelled into replacing and upgradingexisting infrastructure. Many of the road and rail networks were built in thelast century and even the more recent motorway networks built in the lastfifty years require substantial investment to replace worn-out sections. Thisprocess is both expensive and disruptive. Similarly, capacity on existingnetworks is being increased through adding lanes to roads and through theuse of new technology on both roads and railways. Advanced informationand communications systems, together with new signalling and controlsystems, substantially increase the capacity of existing transport infrastructure.

The new debate is over whether there is a case for the extension ofprivatization of road and rail infrastructure. The US experience (Gomez-Ibanez and Meyer 1995) suggests that ‘toll roads are unlikely to be a verypromising area for privatization’ as there are few new possibilities for viableinvestment opportunities. All the potentially profitable roads have alreadybeen built. There is greater potential for the private sector in roads built fordevelopment reasons as there is likely to be less opposition. The costs arelower and the potential for development is higher—the less difficult category.Roads built to relieve congestion require more complicated packages as costsare higher and there is likely to be greater opposition. Consequently, privatesector interest is lower—the more difficult category. The main role for theprivate sector may be as innovator; the benchmark against which theperformance of public authorities can be measured and stimulated. Exampleshere would include new methods of charging by time of day or levels ofcongestion, or by maximizing capacity through charging single occupancyvehicles for using high occupancy vehicle lanes. The US experience providesan informative reference point against which to judge road investmentdecisions in Europe.


In western Europe as a whole there are some 40,000 km of motorways, ofwhich 13,500 km are tolled (Table 3.1). About 90 per cent of the tolledmotorways are in France, Italy and Spain. In 1991, the annual revenue fromtolls ranged from £500 million in Spain to over £1bn in Italy and around £2bnin France (UK Department of Transport 1993). France has the best developedsystem of toll motorways, constructed through letting concessions to semi-public bodies (Société d’Economie Mixte—SEM) which are contracted to buildparticular sections of road. Legislation has also been passed to allow privatecompanies to build certain roads. In all cases, the state has implicitly supportedthe private sector by providing financial support through low or zero interestcash advances, guaranteed loans, or the provision of related infrastructure. Atpresent, France has seven autoroute SEMs, one private autoroute concessionaire(COFIROUTE), and two tunnel SEMs (UK Department of Transport 1993).The SEMs and COFIROUTE keep the revenues from the tolls that can then beused for road maintenance and the construction of new autoroutes.

Elsewhere, private sector funding has mainly been directed at very specific,often relatively small-scale links, such as bridges and airport roads. Thesenew links often duplicate existing routes where capacity is limited. They areconstructed for congestion relief reasons. This places them in Gomez-Ibanezand Meyer’s more difficult category. The main difference to the US situation

Table 3.1 Tolled motorways in Europe

Source: Munro-Lafon and Mussett (1994).Note: These are not congestion tolls; other European countries have extensive

motorway networks which are not tolled, for example, Germany 11,100 km, UK3,300 km, Netherlands 2,200 km, Belgium 1,700 km, Sweden 1,100 km.


is that in Europe the public sector sometimes allows the private sector to takeover the operation of the existing congested link and the new privately fundedparallel link. This ‘deal’ places the private sector in a virtual monopoly position,particularly where there are no alternative routes. If growth in demand isexpected and the scale of investment modest, then the interests of the privatesector are substantial. In the UK, the two best examples are the QueenElizabeth II Bridge across the river Thames, which duplicates the existingDartford tunnels on the M25 London orbital motorway, and the second SevernBridge across the river Severn between Bristol in England and Newport andCardiff in Wales. In both cases the private sector construction company hastaken over the existing tunnel or bridge (Table 3.2).

The private sector may not be interested in funding new links within analready dense network, but it is interested in taking over existing networksand running them. The privatization of the rail networks has already takenplace in the USA, Japan, Sweden and the UK with the new owners taking theresponsibility for replacement investment, new capital investment andupgrading existing facilities. Users are charged according to the route, timeand other criteria (e.g. length or weight of train). The possibilities for tollroads have already been mentioned here, but a debate is beginning on whetherother roads should also be privatized with shadow tolls being charged. Thearguments here are not clear-cut. Supporters of private roads would arguethat private sector management practices are more cost effective and thatpaying for roads by the user makes the cost allocation issue clear. Opponentswould say that roads should be seen as public goods and paid for through the

Table 3.2 The Dartford River Crossing on the M25 London orbital motorway


public sector as funding can be obtained more cheaply by governments thanprivate companies. We now develop these arguments further.

3.3 Investment f inancing

One of the basic political questions being addressed by decision makers is theamount of infrastructure which should be provided—whether it should meetunconstrained demand or whether demand should be constrained—and howit should be financed. Most transport infrastructure projects are funded by thepublic sector as they are large scale, involve a high risk and have long paybackperiods. As noted above it is only where the private sector has some degree ofa monopoly position (e.g. in road bridges across fjords or estuaries) that realinterest for some alternative forms of financing has been raised. Transportprojects are notorious for their cost overruns, for technical deficiencies inconstruction and consequent high maintenance costs and for optimism in theirestimates of future demand. The Channel Tunnel project illustrates all of theseproblems, as does the Oresund link between Denmark and Sweden and theGreat Belt project between Zealand and Funen (The Netherlands).

3.3.1 The historical arguments

The loss of interest in transport infrastructure investment by private investorsmay be explained by the switch in the financing of railway investment frominternational private sector capital to national public financing and theincreasing share of road transport investment from public funds, at the expenseof rail transport. In the aviation sector, in many countries both airlines andairport infrastructure have until recently been financed through the publicsector. This contrasts with the railways of the last century where privatesector investment was dominant. The growth in road transport has meantthat many firms in manufacturing operated their own transport fleets ratherthan buying in services from railway companies. The growth in own fleetoperations was in part due to restrictive licensing in road transport. In additionto these regulatory issues, there is the question of property rights andownership of the railways (Foster 1992) and more recently the ownership ofthe roads. Any uncertainty over the ownership of the (new and existing)infrastructure and time involved would increase the risk to the private sector.

The private sector has only really shown interest in terminals andinterchanges as these are seen as being closer to the other sectors where privatecapital has been directed (e.g. office and commercial sectors). Transport nodesallow the development of associated buildings that can be sold or rented outand also allow the internal space to be franchised. Airport terminalsdemonstrate this potential as some of the highest rents are charged for thisprime retail space. In the UK the fifteen largest airports will extend theirretail floor space by 40 per cent to 600,000 sq ft over the next seven years.


However, the interest of the private sector in the links on the transportinfrastructure network has been less enthusiastic.

The historical evidence outlined above emphasizes the decline of privatesector railways together with the legislative and administrative controls asthe main reasons for the fall in market investment in transport. Theinfrastructure school of thought emphasizes certain features of the transportinfrastructure, which contrast with the directly productive activities financedthrough the market economy. Hirschman (1958) contrasted ‘social overheadcapital’ and directly productive activities. He defines social overhead capitalas comprising those basic services without which primary, secondary andtertiary productive services cannot function. In the widest sense it includesall public services from law and order through education and public healthto transport, communications, power and water supply and agriculturaloverhead capital (e.g. irrigation and drainage systems). The central part ofthe concept can be restricted to transport and power. Hirschman seesinfrastructure projects as having the following characteristics: 1 They are an input to directly productive activities.2 They are typically provided by public agencies or by private agencies

under public control.3 The products are supplied free or at regulated prices.4 The products are not subject to competing imports.5 Production is characterized by ‘lumpiness’ (technical indivisibilities) and

a high capital output ratio.6 Output may not be measurable. In the Hirschman classification above, (3) would represent a major obstacleto market provision. The non-operation of the exclusion principal and thepresence of free riders preclude market investment in this sector. The othercharacteristics such as capital lumpiness and non-measurable outputs wouldpresent less formidable obstacles to market provision. The Hirschmanarguments would all favour continued involvement of the public sector intransport infrastructure provision.

Economic and technological changes since Hirschman definedinfrastructure in the terms set out above have blurred the distinction betweeninfrastructure and directly productive activity. Thus infrastructure canincreasingly be provided in the market sector and a return to the high levelsof private sector investment in transport, typical of the last century, can againoccur. There are five changes that allow this to happen: 1 Pricing mechanisms are available for roads, seaports and airports and

the exclusion principle may be applied. Smart card technology has reducedthe transaction costs of road pricing. Second-best alternatives such asshadow tolls can also be used.


2 Capital intensity and lumpiness are less of a barrier to private sectorinvestment now than when Hirschman wrote. For example, thedevelopment of airports in the immediate post-war period coincided withthe belief that only the state had the resources to undertake such largeinvestments. This contrasts with the present situation when manygovernments experience severe constraints on public expenditures whereasprivate sector financial institutions have experienced a large increase intheir supply of available funds for investment.

3 Administrative reforms have been instituted to establish agencies chargedwith devising pricing formulae for privatized utilities such as gas, water,electricity, telecommunications and airports.

4 Cross-border interconnectors for gas and electricity have made theseproducts internationally traded goods. Reverse charges have broughtcompetition in international telephone calls.

5 Output measurement techniques have been improved by research in areassuch as programme budgeting, cost benefit analysis, cost effectivenessanalysis and research on factor productivity.

The ability to charge prices for transport infrastructure, the constraintsimposed by the public sector borrowing requirement (PSBR) when privatefinance is more readily available and the development of output andproductivity measures have reduced the distinction between social overheadcapital and directly productive investment.

3.3.2 The privatization arguments

Privatization has an important demonstration effect and its perceived successis likely to lead to further actions. The arguments have included wider shareownership, income redistribution and reductions in trade union power. Theempirical evidence is less convincing and privatization may not be the mostsuitable policy instrument for achieving each of these goals (Vickers andYarrow 1988). However, the economic arguments may be more powerful.Stevens (1992) concludes:

The property rights theory of the firm suggests that public enterprisesshould perform less efficiently and less profitably than privateenterprises. In a private enterprise, both internal control—via theshareholders—and external control—through the discipline of thecapital market—provide incentives to avoid inefficiencies. By contrast,public enterprises are not subject to the discipline of the capital markets,and internal monitoring is conducted by politicians who do notnecessarily see their role as supervising the efficiency with whichmanagers allocate resources.

(Stevens 1992:12)


Much of the evidence is inconclusive on whether the private sector isnecessarily more efficient than the public sector. Many cost savings can andhave been made in the public sector prior to privatization and the empiricalevidence has not led to clear conclusions (Kay, et al. 1986; Millward 1986).

Several different strategies seem to have been developed to reduce levels ofpublic expenditure in transport. Privatization of transport companies is onemeans to reduce pressures on public sector budgets and reduce the publicsector borrowing requirement. However, when the transport infrastructureis considered, the options available are less clear. In the past, transportnetworks have been seen as a public good and a long list of problems havebeen presented as to why it would be difficult to use private sector funding,except in certain situations or where very clear guarantees were given.

The main concern to the private sector is the risk of investing large amountsof capital in projects where there are: • long periods between the start of the investment and the financial returns

to investors;• irreversibility of investment or where there are substantial sunk costs, it

is costly to start a project and even more costly to withdraw;• financial returns do not flow until the whole infrastructure is completed;• political influences on the production of goods and services;• long amortization periods when loan repayment periods are often over a

much shorter term;• uncertain impacts of concurrent investments on demand in a network

setting. Quite naturally, the conservatism of developers would mean that investmentwould be concentrated on buildings and transport terminals/interchangesrather than in the infrastructure itself. There are lower risk projects availablewith shorter payback periods and greater certainty (until recently) over therates of return.

In addition to the general questions of risk, transport projects have in thepast proved difficult to assess. They often have: • substantial cost overruns;• levels of demand which are difficult to predict accurately;• been considered as a free good at the point of delivery;• aroused considerable public opposition to their construction and this

has often delayed their implementation, sometimes after substantialinvestment has been made in scoping studies (e.g. environmental impactassessments).

High levels of risk and uncertainty mean that the private sector has beenreluctant to get involved, despite having the resources available. Investment


funds currently have large sums of capital available and it would seem naturalto direct some of this cash into transport infrastructure.

3.3.3 Role of the public and private sectors and jointprojects

The key question facing governments in many countries is to establish themeans to bridge this funding gap. Both the public and private sectors haveimportant roles to play in the construction, renewal and maintenance of theroad, rail and air infrastructure. Over the next decade it is likely that newmeans of financing will be established, yet the public sector cannot withdrawcompletely. Much of the funding will still remain the responsibility of thepublic sector, but the role for transport planning must be to facilitate privatesector and joint ventures through advice, predictions, land assembly andaccelerated public inquiry procedures. The public sector still has a key role toplay: • where the market fails and intervention takes place for accessibility,

distributional and equity reasons, and where transaction costs are high;• where there are significant externalities involving the use of non-renewable

resources, land acquisition, safety and environmental concerns;• where transport interacts with other sectors, such as the generation effects

of new developments and priorities given for regional or local developmentobjectives;

• where transport has national and international implications, such aspromoting a capital city (e.g. London or Paris) as a world city ormaintaining high quality international air and rail links.

However, all these roles are essentially passive and the more important positionfor the public sector must be to promote a partnership between the publicand private sectors.

Public sector actions

There are a series of positive actions that can be taken by the public sectorto facilitate interest from the private sector. As most transportinfrastructure investment is long term, there must be a stable policyframework and a set of planning procedures within which the privatesector can operate. In addition, the public sector could raise capital throughspecial funds that are seen as ancillary budgets remaining underparliamentary control. Only the balance would be shown in the generalbudget. Access to these funds would take the form of a competition betweenthe various government departments or there could be a competition withindepartments for the available budget.


Alternatively, earmarked charges could be raised from toll roads (as inNorway) to finance further investment. Some economists (e.g. Buchanan 1963)have favoured earmarking as a means to compartmentalize fiscal decisionsand to allow individuals to participate in public expenditure decisions. Anew impetus has been given to the hypothecation arguments with therealization that greater transparency and public acceptability of increasedtaxation is essential if substantial sums of public expenditure are to be allocatedto transport infrastructure. The acceptability of environmentally efficient taxeson motorists will depend upon part of the revenues raised being used to fundless environmentally damaging modes of transport. If long-run changes indemand are also desired, then the decisions concerning hypothecation mustalso be transparent, otherwise most motorists will continue to use the carand pay a higher price.

A second major contribution that can be made by the public sector is tofacilitate the complex processes of raising the substantial sums of capitalrequired for transport infrastructure projects. These include raising capital: • in Europe through loans from the European Investment Bank and the

European Union’s European Coal and Steel Community, the RegionalDevelopment Funds and the new European Investment Fund;

• through transport bonds and other long-term investments (such as pensionfunds) which are extensively used in Japan;

• through tax incentives to the private sector by making their capitalcontributions tax deductible. This might allow pension funds and financialinstitutions to become involved in road construction as an investmentopportunity (suggested for road building in Norway with funds set asidein development areas. This proposal was not accepted by the government);

• through employment taxes (as in Paris and other French cities) or a taxon petrol (as in Germany and the USA);

• through user charges from tolls and road pricing. More controversially, the public sector could guarantee loans to the public orthe private sectors, thus accepting a substantial part of the risk. Considerabledebate is currently taking place on this issue as the balance of the risk is stillwith the public sector and not the operator of the system, for example, wherethe government (public sector) underwrites the loans to the railways (publicsector). Much of the French TGV routes have been funded in this way. UnderFrench transport policy, the French Railways (SNCF) are responsible fordeveloping the railways and they have had a high credit rating as thegovernment guarantees the loans. This means that interest rates are lowerthan commercial rates (typically by 1 per cent) and high rates of return arenot essential (as they would be on equity risk capital). This allows SNCF torun at a loss and pay no corporation tax. A private sector package wouldhave taken much longer to set up (Gérardin 1990). In the longer term, the


TGV system is expected to make a profit and may repay a substantial part ofthe amortized debt, but in the short term SNCF has substantial debts. Asimilar company in the private sector with similar levels of debts (FFrl00bn,1987) would have been placed in liquidation.

However, these semi-public undertakings in both the road and rail sectorsin France have allowed revenue funding to switch between projects. The cashflows generated by the infrastructure for which the loans have been completelyrepaid have been used to finance further extensions to the network: whatGérardin (1990) calls the overspill principle, similar to that used in the lastcentury to finance the last great expansion of the rail network.

Private sector projects

There are particular types of infrastructure projects which the private sectorare prepared to finance and operate with minimum levels of guarantees (e.g.the Queen Elizabeth II Bridge on the M25 around London, Table 3.2). Theselow risk projects are comparatively small in scale and place the private sectoroperator in a monopoly position. Interest is particularly high where there arecongested conditions on the existing infrastructure and where anticipatedgrowth in demand is high. If a reasonably lengthy ownership period can benegotiated, then the expected payback is substantial. In this case the privatesector plans, designs, builds, operates, owns and finances the project. Therole of the public sector is secondary and limited to the promotion ofsupporting actions (e.g. enabling legislation) and negotiations on the lengthof the ownership period. This needs to be balanced against the expected growthin traffic and the levels of charges to be paid by the users. Successfulnegotiations on time, traffic growth and user charges mean that the projectwill go ahead.

In most cases, one or more of these conditions are not met and as aconsequence the private sector has been reluctant about making any firmcommitment. A further complication is the ownership of the infrastructureafter the period guaranteed to the private sector. Reversion to the publicsector may mean that the route no longer has a toll on it, but maintenancecosts and reconstruction costs are likely to be substantial (particularly ontunnel infrastructure). This means that the public sector may incur substantialcosts at some time in the future.

The most notable exception to all the above is the Channel Tunnelproject which has been funded by the private sector. This is Europe’s largesttransport construction project, costing some £10bn and the length of theownership period by the private sector is fifty-five years. It has beencompletely designed, constructed, owned, operated and financed by theprivate sector. The substantial risks are with the private sector over a verylong period of time, yet the rewards may also be substantial if predictedlevels of demand are met.


Joint projects

It is in partnership between the private and public sectors where most potentiallies. The public sector can anticipate the growth in demand that results fromthe growth in the economy, from rising income levels and from newdevelopments. It can also assist in the land assembly process and in the publicinquiry so that the time from project inception to completion is minimized.In certain situations, it can also help with the costs of construction of theactual infrastructure, but it is the private sector which will manage and runthe facility, including setting the levels of tolls or fares.

There are several different approaches that can be used. The land couldremain in public ownership with a contract between the public and privatesectors. This concession would be granted through a tender or franchise andfunding would be the responsibility of the private sector with some publiccontribution in the form of loans and loan guarantee. The private sectoroperator would set the charges, but constraints on the quality of service wouldbe set by the public authority granting the concession (e.g. minimum levels ofservice and safety standards). Maintenance of the infrastructure would alsoremain in the public sector, and public money would only be used to ensurethat the project actually takes place. This means that the risk to the privatesector has to be equivalent to other investment opportunities. One possibilityhere is to restrict the public sector contribution to those factors which reducethe negative externalities (i.e. the public sector would pay for environmentalimprovements), but even this is controversial if the polluter pays principle isused. It is often cheaper to guarantee loans than to contribute directly. Ratherthan have a toll for the use of the facility, the public sector could pay a fee orshadow toll to the private sector for every vehicle using the road (Button andRietveld 1993). Such an arrangement would allow a scheme proposed by thepublic sector to be built by the private sector and operated as a free (at thepoint of use) facility. The risk to the contractor is in the demand forecasts asthey would only be paid a fixed rate for the actual numbers of vehicles usingthe facility. At the end of the contract period, the public sector might havethe option to buy the road.

Development gains can also provide an important incentive to the privatesector. In the past, planning permission has often been granted subject tocertain conditions being met. These conditions have involved roadconstruction, particularly in locations where development pressures aresubstantial. A different option would be to give the developers the rights todevelop land around the road or rail infrastructure that the developer hadfinanced. Land values at new accessible locations, principally at roadinterchanges or rail terminals/stations, rise substantially (Stopher 1993) andthe potential for development is considerable.

In this case the public authority would acquire more land than is actuallyrequired for the construction of the infrastructure. This land could either be


sold to the developer with the profits being used to finance the constructionof the infrastructure by the public sector, or an agreement could be reachedwith the private sector that it takes the responsibility for construction andacquires the land. In this second alternative, the private sector has theassociated land development rights that are similar to the air rights beinggranted in urban areas over new station developments. The private sectorcould either carry out the full development or build the infrastructure andsell on the associated development rights.

Another approach is the auctioning of a prepared project to reduce theuncertainty regarding time and costs as construction can begin immediately.The private sector has to make a bid to complete the final design and tofinance, construct and operate the project. Fielding and Klein (1993) arguethat competition in the bidding would be enhanced and the post-contractualadministration costs would be reduced. The clearing-before-awardingapproach should ensure that: • the risk to the private sector is reduced and thus bids would be expected

to have a lower rate of return;• the costs to the public sector for inquiries, environmental impact

assessment and land acquisition may be lower than they would be to theprivate sector, again reducing risks and sunk costs;

• the whole process involves both the public and private sectors inappropriate roles. Public sector involvement in the approving andawarding phases means that the proposals may be less vulnerable topolitical tampering and all the necessary procedures are followed in full.

There do seem to be substantial opportunities for the private sector to becomemore involved in the process of planning, designing, building, financing andoperating the links on the transport network, as well as investing in andoperating the nodes on the network. It is likely that progress must be made inpartnership between the public and the private sectors. The combination ofboth sectors substantially reduces the front-end risk and the likelihood offinal cost overrun. It also clarifies the difficulties of estimating the paybackto the private sector and the necessary period over which it would accrue.The financial rates of return could be substantial and revenue flows in theshort term are possible. The public sector also needs to take a much moreactive role in promoting the project and steering it through the planning anddesign stages.

The main difficulties have been overcome, but other factors must also beresolved: • Risk sharing between the public and private sectors is necessary. This is

fine in principle, but in practice the private sector has been adept in makingsure that most of the risk remains in the public sector.


• The free rider problem has to be resolved. Again, companies andindividuals have been adept at not contributing to the benefits (directand indirect) that transport infrastructure investment brings to themthrough, for example, larger market areas or increases in property values.

• New transport infrastructure needs to be seen as part of the developmentprocess, not separate from it. Implicit in the analysis in this book is thattransport infrastructure investment is an integral part of the widerdevelopment process, but it is often thought of and analysed as independentof this.

• Consistency and stability in financing needs to be maintained as politicaland commercial horizons are often short term, while major infrastructuredecisions are long term.

With joint projects, the private sector would recover their costs through usercharges, but a balance needs to be sought between the private and publicsectors in terms of their risks and rewards.

An interesting example of a joint project is the Cross-Israel Highway project,now under construction. The main component of this undertaking is an 84km, limited access highway, running South-North to bypass major metropolitanareas in the centre of the country. The cost of construction (net of landacquisition) is estimated at $1 billion. Following an international bidding process(in which the required level of service was prespecified) the project is financedand built by a private consortium. Revenues will come from users fees. Thecontract signed by the government and the private consortium places themajority of the risk with the public sector. It includes elements such as guaranteedminimum level of demand and rate of currency exchange.

The discussion here has attempted to open up the debate and it reflects theinvestment component (Figure 2.1) in the conceptual framework developed.We have not come up with a magic solution to the problem of financing andpricing transport infrastructure, but it has raised some of the principal issuesthat must be resolved. The private sector has a strong tradition of commercial,office and residential development and has successfully moved into the transportsector to take over the development of terminals and interchanges. The challengeset here is to determine whether private sector portfolios can be further extendedto include transport links. The transport infrastructure needs substantialinvestment in new and upgraded roads and railways. Public budgets are limitedand there must be a commercial opportunity for the private sector to enter themarket. The question then moves to the identification of how commerciallyprovided infrastructure (in its broadest sense) impacts economic development.

3.4 Implications for economic development

In this chapter we have tried to illustrate the case that the traditionalarguments between transport infrastructure investment and economic


development have been weakened, if not broken. Four main reasons leadus to that conclusion. First, transport is now seen as part of the wider processof economic integration that has strong political, social and environmentalimplications, as well as the better known economic implications. It is alsopart of the new agenda of competitiveness and a renewed focus on measuresof productivity. Second, there has been a substantial decline in the levels ofinvestment in new infrastructure. Available funds are being channelled intomaintenance and upgrading the existing infrastructure. Third, the role ofthe private sector and new partnerships have given an added importance tothe alternative sources of investment capital. Finally, there is the questionof the impact of investment on transport efficiency. Here there seems to bea fundamental difference between the project evaluation which gives highlevels of consumer surplus from each alternative (from cost benefit analysis,see Chapter 7) and the macroeconomic modelling impacts which suggestrather more limited effects on GDP growth and factor productivity (Chapter6). We examine each of these four points in more detail in this concludingsection.

Transport and economic integration

This has broadened the debate from the narrower focus of economicdevelopment. Development is concerned with raising the economicperformance and condition of all parts of the state, region or locality, througha combination of policy interventions to raise growth levels and the distributionof that growth. Economic integration is concerned with achieving convergencein the economic performance by removing restrictions on the movement ofgoods, services and factors of production. As well as the removal of the physicalbarriers, it covers the harmonization of social, fiscal and environmentalbarriers and in Europe it may lead to economic and monetary union. Equallyimportant to the aims of integration is the means to achieve it. Here, theprocesses of regulation and liberalization act as powerful opposing orcomplementary forces to achieve change through microeconomic policydecisions. Yet the ability to move people and goods around countries andregions does not in itself guarantee integration. Indeed, it could be argued(Vickerman 1994; Vickerman et al. 1999) that the richer regions becomericher and the poorer regions poorer, irrespective of whether investment takesplace in the transport infrastructure or not.

An essential component in our understanding of the links between transportand economic integration must be the microeconomic conceptualization ofthe processes at work in the location decisions of firms, the impact of externalfactors on integration and their relative competitiveness over time (Chapter8). The ingredients of any new approach to analysis must incorporate thefocus on competitiveness of industry and productivity rather than thetraditional concerns with development. The argument here is that transport


may have an instrumental role in the efficient production of goods and services;it can be substituted for other factors of production; there may be economiesof scale; and increasing returns to scale, and conditions of imperfectcompetition, knowledge and technology may all exist.

In the SACTRA report (1994: para 9.30–9.32), two situations have beenidentified where real economic benefits arise in imperfect markets and whereprices do not reflect marginal resource costs. The first situation occurs wherethere is widespread underemployment of labour (i.e. the wage rate does notequal the opportunity value), and infrastructure investment could bring aboutnew economic activity at the national level. The key element here is that ofadditionality—there is a net gain. The second situation occurs where theroad diverted economic activity from a low unemployment location to a highunemployment location. The net result here is that the economy as a wholecan operate at a higher level of employment than would otherwise be thecase. It seems that for both underemployment and the unemployment, thecases argued are cautious. Even if there were change, causality would bedifficult to infer. Road investment decisions in advanced economies are seldomof sufficient scale to encourage major change in levels of underemploymentor unemployment.

Far more likely is the situation where individual firms and people makesavings through relocation in closer proximity to the new investment. Thespatial arguments may be more important than the economic arguments, yetthe spatial arguments have important economic implications. Businesses mayreduce their distribution costs through taking advantage of a high qualityroad network, thereby increasing productivity. This in turn may result infewer distribution points so that travel distances are increased, but savingsaccrue to the individual company and indirectly to the final consumer throughcheaper prices. In a well-connected network of high quality roads, companieswill act to minimize their total distribution costs, which in turn may lead tomore transport intensive operations. Economies of scale in warehousing areexploited by firms so that, although the total vehicle miles are increased, thenumber of manufacturing and distribution points can be reduced (a fulldiscussion is presented in Chapter 4).

Current rationale has switched to the need to maintain and improvecompetitiveness, particularly with respect to international competitors (UKDepartment of Transport 1996: para 4.1–4.19). Here it is argued that transportinvestment can form an important component of business operating costsand that congestion and unreliability of trips add to costs, ‘particularly forthose companies in the service sector or those businesses essentially servingurban areas’ (ibid.: para 4.12(a)). However, the same study (carried out forthe UK Department of Transport by Ernst and Young) stated that ‘the impactof new transport infrastructure on business costs is much less clear than isoften perceived, though over 60 per cent of respondents to the survey reportedbenefiting from particular transport infrastructure improvements completed


within the last five years’ (ibid.: para 4.12(c)). The conclusions reached fromthis assessment of contemporary issues in transport investment are: • The argument has moved from concern over development to one of

competitiveness.• The role of government has become less interventionist with lower levels

of commitment to financing new infrastructure.• The role of government is increasingly seen as setting the right market

framework within which others can invest.• The private sector has been reluctant to fill the gap as the risks are high,

the returns accrue only over a long time period and there are better marketopportunities for investment.

• The possibilities of a new generation of road investment is unlikely andeven where there is a commitment in Europe to invest in trans-Europeannetworks the problem of finance may prevent action taking place.

• Uncertainties and unreliability within transport networks are key concernsto firms and individuals. Congestion is usually cited as a major problem.Coping strategies include relocation decisions and improvements in theinternal efficiency of firms and individuals, through the use of logistics,technology and trip chains.

New investment and maintenance

These concerns feature highly in national priorities. In many countries, thelevels of investment in new infrastructure have been reduced as public budgetshave been squeezed and as public concerns over environmental issues haveincreased. The option of building new infrastructure to meet expected levelsof demand is seen as being financially and politically unacceptable. In addition,much of the existing infrastructure is in need of renewal and upgrading as itis ‘worn out’. It is both expensive and disruptive to rebuild existing roads asthey still have to be used during the reconstruction. Increasingly, public budgetswill be directed towards maintenance of existing roads rather than investmentin new roads. Expansion of the existing network is quicker (althoughexpensive) through adding lanes to the highways as land acquisition hasalready taken place and as the time-consuming planning procedures do nothave to take place. As public funds are reduced, and increasingly allocated tomaintenance and expansion of the existing network, new sources of fundingare required for new infrastructure.

Private sector and new partnerships

These now form an essential component for expanding the network. Twoimportant points need to be elaborated on: financing and economic benefits.The greatest potential for real progress in developing new forms of financing


for transport investment is through joint funding opportunities where the risksand returns are enjoyed by both the private and the public sectors. The principalrole for the public sector is in the planning and design stages of the road. Theselengthy procedures, including public inquiries, land acquisition and compulsorypurchase rights, can best be undertaken by the public sector. The private sectorwould then undertake the construction and operation of the road, after acompetitive tendering stage. Finance may come from the private sector or fromboth the public and private sectors. If the road is to relieve congestion, thenmost of the revenues would come from tolls. But if the road was built fordevelopment reasons, then returns would come primarily from the developmentrights associated with the land adjacent to the road. Another alternative is theuse of shadow tolls on new (and existing) motorways. The private sector wouldfinance, construct and operate the motorway and they would be paid a chargeaccording to the numbers and mix of vehicles using the route. It seems thatthere are many possible arrangements for public and private sector partnership,but few seem to have been adopted in either the USA or Europe. Yet it isthrough joint funding opportunities that most major new road projects will befunded, particularly where expected levels of demand are modest or wheredevelopment objectives are paramount.

Both the theoretical arguments and the practical experience from Europeand the USA seem to raise the main questions, but not to resolve the issues.Private sector investment is possible for well-defined, small-scale projectswhere the risks are low and the returns guaranteed. For larger scale, higherrisk projects with longer payback periods, the public sector is still the mainagent, perhaps with greater assistance from the private sector to promoteinnovation. More generally, the crucial questions facing governments are overthe appropriate transport policy for the end of the millennium and the rolethat road investment might have in that strategy. There is considerable pressureto reduce levels of public expenditure in road infrastructure and the beliefthat investment can be linked with economic growth is still unproven. Otherarguments, such as environmental impacts of new roads and whether thepricing of the infrastructure is appropriate, seem to dominate. Similarly, theclaims for public sector budget constraints and inefficiency and higher levelsof private sector productivity are also unclear. These complex issues maynever be completely resolved, but the question of how to raise new sources offinance for roads needs clear answers now. Joint ventures between the publicand private sectors with a sharing of the risks and the rewards must be seenas a major opportunity for releasing more capital for the funding of roadinfrastructure projects.

Impact of investment on transport eff iciency

In general, transport investment alters the transport costs of users and thesubsequent patterns of location and trade. Yet the presence of good transport


links does not guarantee prosperity and the absence of good transport linksdoes not necessarily act as a constraint on the economy (Vickerman 1998).This potential non-symmetry needs some explanation. Across most developedcountries, the expenditure on new road construction has declined as apercentage of GDP to very low levels (Table 3.3). Over the last twenty years,these figures have been reduced by between 25 per cent and 50 per cent. Theirony here is that growth in demand for transport is assumed to be stronglyrelated to growth in GDP, yet the proportion of GDP allocated to transportinfrastructure investment has fallen. This might suggest that there isovercapacity in the transport system (unlikely, at least at certain times of theday), or it might suggest that the transport intensity (the relationship betweenthe volume of use and the level of economic activity) is independent ofinvestment in the infrastructure.

When cost benefit analysis is carried out on new and upgraded roads, thebenefit cost ratios are high at between 2:1 for national roads and 5:1 forlocal roads (Roy 1994). The social rates of return on the European high-speed rail network have also been high (a minimum of 8 per cent, but as highas 25 per cent), but they have recently been questioned as being too optimisticfor the regional development effects. What is clear is that the financial rate ofreturn is substantially less than the social rate of return. This has meant thattransport projects have proved less attractive to governments and the privatesector than other investment projects. The key question then is whether adirect link can be established between non-investment in transportinfrastructure and economic efficiency and productivity.

It is here that the macroeconomic modelling work of Aschauer (1989a)and Munnell (1990a) have had a major impact on the debate (Chapters 1and 6). In the UK, Lynde and Richmond (1993) concluded that ‘a higher rateof infrastructure investment, sufficient to maintain the public capitalcontribution at its pre-1980 average level, could have brought about anincrease in the rate of growth of labour productivity in UK manufacturing

Table 3.3 Expenditure on new road construction,1988

Source: International Road Federation (1992).


from around 4.0 per cent to about 4.5 per cent per annum’. The estimatedproductivity effects of transport infrastructure derived from productionfunction models is high and considerably higher than those derived from ananalysis of firms’ transport costs. Rietveld (1994) has suggested that, althoughthe effect of improved infrastructure on transport costs may be limited, theimpact on profit margins might be substantial. This line of argument leadsRoy (1994) to conclude that there is a causation between infrastructureinvestment and productivity resulting from the macroeconomic research andhigh returns from cost benefit analysis. There are welfare gains to leisureusers of the transport infrastructure and gains ‘that rebound to the benefit ofthe economy-wide productivity and of the national interests ofcompetitiveness’. In Chapter 6 we examine these issues in depth.

The basic parameters of the arguments for public sector investment in thetransport infrastructure for economic development gains have fundamentallychanged. For the reasons outlined in this chapter, new financial arrangementsare now taking over from the traditional public sector provision in alldeveloped countries. This is not a short-term reaction to reductions in publicexpenditure, but a fundamental reassessment of government priorities and aswitching towards other sectors of the economy (e.g. education, health andemployment). New sources of finance are necessary and this must involvethe private sector working in partnership with the public sector. Similarly,the rather narrow economic development arguments have been broadenedto explore the processes of economic integration and competitiveness as themarkets have become regional and international. Investment is now increasingefficiency and productivity and the new agenda concerns the means as tohow these gains to companies and individuals can be recouped, at least inpart, through pricing the transport infrastructure. So there are two key newdimensions to the debate, namely how the infrastructure should be financedand how it should be priced.

The evolving economy

4.1 Introduction

Much of the debate has focused on the links between transport infrastructureinvestment and economic development within a traditional context of work,location and travel. As we have discussed in the previous chapter on investment,new forms of financing are required as public budgets become less availableand as the role of the private sector is encouraged through possibilities such asthe private finance initiative in the UK and public-private partnerships. In thischapter we give a complementary view of the evolving economy. Four mainchanges are highlighted, each of which will have a substantial effect on thedemand for travel. The underlying argument is that novel methodologicalframeworks and forms of analysis are required to investigate the links betweentransport infrastructure investment and economic development.

The changes in society brought about by information technology are likelyto be as great as those that caused the shift from an agrarian to an industrialeconomy in the last century. The best estimate is that nearly 40 per cent ofthe workforce in advanced countries is engaged in information industries(Hall 1996). This is the knowledge-based society. Yet some would argue thatthis is not a revolution, but an extension of manufacturing as all new jobscan be traced back to production processes (Cohen and Zysman 1987).

Technological innovation can be traced back 150 years to the invention ofphotography and the electric telegraph (Table 4.1). The interesting point here

4

Table 4.1 Short waves of communications and information technology

Source: Hall (1996).


is the increasing speed with which innovation is occurring in thecommunications and information industries. At the current rate, a furthermajor break-through will take place within the next five years.

The location of manufacturing is being separated from the informationcentres. Manufacturing production is being dispersed globally to locationswhere suitable labour can be obtained at a low cost. Transport has becomean important facilitator of that dispersal as it allows flexibility and dispersedproduction processes. Conversely, the information centres are beingconcentrated at the few ‘global cities’ (London, Paris, Tokyo, New York),together with the second tier ‘world cities’. These 20–25 cities account formost of the headquarters of financial services, the major production companiesand the national governments (Friedmann 1986; King 1990). As Sassens(1991) has concluded, ‘weight of economic activity over the last fifteen yearshas shifted from production places…to centers of finance and highlyspecialized services’.

4.2 Changes in work patterns

Society is changing from one based on work to one based more on leisureand recreation. This does not mean that we will no longer work, but that theneed to work will be reduced as the rewards are higher and as people inheritmore wealth. As standards of living rise, the imperative to earn is reducedand other activities become more attractive. The transport implications ofthese changes are profound.

Within the expanding market for travel, the importance of work-relatedactivities is being reduced (Table 4.2). In Great Britain, four main trip purposesnow account (equally) for about 75 per cent of all trips. Twenty years agothese same four purposes accounted for only 65 per cent of trips, with thework purpose being the dominant partner. The greatest increase has been inleisure activities, including personal business trips and social activities. Thesame pattern can be observed elsewhere in Europe (Banister 1994) and in theUSA (Table 4.2).

With the growth in non-work related activities, it is likely that the timespent travelling will increase. Traditionally, it has been argued (e.g. Zahavi1982; Stokes 1994) that people have a fixed travel time budget of about 60minutes per day. As modes of transport get faster, travel distances have increased,but the overall time budget has remained constant. The evidence on changes intravel time budgets is difficult to obtain as it is not collected in most surveysand as time perception is often not accurate. For example, in the former WestGermany, the number of activities undertaken each day per mobile person was2.1 in 1982—this involved 3.8 trips. The travel time per day was 63 minutesand the distance travelled was 26 kms (Brög 1992). In Great Britain, there hasbeen a remarkable consistency in daily travel time over the last eight years(1989–96), averaging at almost exactly 60 minutes per day.

The evolving economy 85

This situation may be changing. As journey lengths increase, as newactivities take place and as congestion increases, it is likely that travel timewill also exceed the 60-minute threshold. In addition, with the increase inlong distance travel (particularly by air), the average is again likely to increaseas it is difficult to see situations in which people actually choose to travel less.Certainly, the distribution around the mean value will change with increasingnumbers of people spending larger amounts of time travelling.

However, there is considerable evidence of change in distance travelled fordifferent social groups (Tables 4.3 and 4.4). The greatest percentage increaseover the 30–year period has come from the elderly and young, but the absoluteincreases still show the greatest growth in distance travelled for men (+130.1km) and women (+106.6 km), with the elderly (+81.7 km) and children (+53.3km) lagging behind (Table 4.4). Part of the explanation for the increase intravel distance is the growth in car ownership (Table 4.4). This is consistentwith the travel time budget argument where extra distance traveled by carhas no time penalty. If seven hours per week are available for travel, thenaverage speeds have increased from 16 km/hr to 30 km/hr over the 30–yearperiod. The average travel speed calculated here for Great Britain (1985–6)is similar to that for the former West Germany of 25 km/hr (1982).

Although the overall patterns of change are similar in Great Britain andthe USA, there are significant differences (Table 4.5). Trip rates are similar,but there are substantial differences in the average trip lengths with the USfigures being 50 per cent higher than those in Great Britain. This is notsurprising, given the geography of the two countries. The net result is that UStravel per person per year is much higher than in Great Britain, and theincreasing average trip lengths in Great Britain (+33 per cent) have not been

Table 4.2 Trip purposes from national travel survey data in Great Britain and the USA

Sources: Department of Transport (various), US Department of Transportation (1996).Note: Not all columns sum to 100 per cent and some of the definitions have changed.The

Great Britain and US definitions are not comparable—so these f igures are indicative.The starred values include day trips.


recorded in the USA (-4 per cent). The growth in travel (1970–94) relates toreal income increases (+50 per cent in real terms), increasing car ownershiplevels (+78 per cent in USA and +86 per cent in GB), as well as populationincreases (+27 per cent in US and +9 per cent in GB). In both countries, thelabour force has expanded as a result of the post-war baby boom and thegrowth in female participation rates. Women now make up nearly half thelabour force.

Table 4.3 Change in distance travelled per person per week in Great Britain

Source: Department of Transport (various).

Table 4.4 Travel distance per person per week by car ownership in Great Britain

Source: Department of Transport (various).Note: Distances in kilometers and percentage of households in parentheses.* These two distances were not separated in the 1994–6 survey.

Table 4.5 Comparison of travel patterns in Great Britain and the USA

Sources: Department of Transport (various) and US Department of Transportation (1996).Note: US figures exclude commercial driving—distance driven by taxi drivers, truckdrivers and delivery services.


However, if we examine the national travel survey data a little more closely,we find that for our same four social groups the growth has taken place innon-work travel (Table 4.6). The growth in travel for non-work relatedactivities has been 51 per cent, 26 per cent, 26 per cent and 29 per centrespectively in total distance travelled. This increase is explained by acombination of growth in journeys and average journey lengths. The changingnature of society is already beginning to manifest itself in terms of new patternsof mobility with work-related activities (commuting and business) onlyaccounting for 23 per cent of trips in the expanding market.

The second major change that is affecting work travel is the reducedimportance of the city centre as a source of employment and the developmentof much more dispersed patterns of employment. The pattern of work journeysis becoming more varied, both spatially and temporally. Commuting patternshave become more complex, with cross-commuting becoming more importantthan commuting to city centres. Households with an established residentialbase are likely to meet their career needs by longer distance commuting,particularly if there is more than one person employed within the household.There is likely to be a longer term adjustment process as jobs also decentralizeand move closer to where employees live, but these locational changes takeplace infrequently and have a differential impact on employees. A firm movingcloser to one group of employees is likely to become more remote from otheremployees at the same time.

Gordon et al. (1991) argue that there is a dynamic process at work andthat auto commuting trip times for the twenty largest US metropolitan areassuggests that average trip times have remained constant (1980–85), or havereduced by a statistically significant amount. Their explanation is simplythat the market operates spontaneously through the relocation of firms andhouseholds to achieve the balance of keeping commuting times within tolerablelimits. This spontaneity is optimistic, as the cycles for household and firmrelocation decisions takes place over a much longer period than the five years

Table 4.6 Non-work travel per person per week in Great Britain

Source: Department of Transport (various).Note: Distances are in kilometers.


covered by this study, and that longitudinal data are essential to follow theindividual movers and non-movers. Work is only one determinant of thelocation decision and we would argue that it is difficult to capture this processthrough cross-sectional analysis. Other factors, such as quality of theenvironment, local schools and other facilities and proximity to friends areall important social factors in the decision, as is the financial imperative ofthe availability of housing at an affordable price.

Life-cycle changes may also influence both the mobility patterns andlocation decisions of households (Table 4.7). Here, it is suggested that thereis a dynamic process which is related to more factors than just work. Locationdecisions of households reflect stages in the life cycle and the range of necessaryand desired activities in which a particular household needs to participate.With the increase in non-conventional households or groupings of people,this picture may again be complicated (Chapter 5). There are substantialpush and pull factors at work as the balance is sought between what a citycentre can offer in terms of employment and other opportunities and therequirements of families and others at different stages in the life cycle.

The reasons for residential mobility may have less to do with proximity towork than with finding the right type of accommodation. In times of recession,high interest rates or job insecurity, fewer residential moves will take place.This has been the situation in the UK (in the early 1990s) where the housingmarket was static and prices had been falling as people had to sell at a loss(the problem of negative equity). Even with cheaper prices, there seems to bea reluctance to move. In Paris, there are large differences in rents betweensitting tenants and new arrivals and this has resulted in housing immobility

Table 4.7 Life cycle effects on travel and location

Source: Banister (1994).


(Orfeuil 1992). The attractions of moving to new suburban residential estatesas owner occupiers is attractive, particularly if the developer is prepared tobuy your existing property and to give attractive repayment terms, at least inthe short term. The location of work may only be one consideration in bothof the decisions of whether to move and where to move.

The third dimension of change is the nature of work itself. The traditionalconcept of regular employment over a lifetime with fixed hours, often with justone employer, is changing. There are now several different types of employment,even within one organization—what Handy (1995) calls the shamrockorganization. The first leaf of the shamrock represents the core workers, madeup of qualified professionals, technicians and managers. These are the full-timeworkers who give 100 per cent to the company and are wellpaid employees. Butthese core workers have been reduced in number as a result of downsizing orrestructuring. The second leaf is the contractual fringe where non-essential workis contracted out to specialists who do a better job more cheaply. Technologyand just-in-time delivery means that the subcontractor carries the stock neededby the core company. Failure to deliver means that the work will be subcontractedelsewhere. Payment is made strictly on results and these subcontracted firmswould supply several different companies with their expertise. This is part of thespecialization process. The third leaf is the flexible labour force that is essentiallythe underclass as they are hired and fired according to need. Most of them arepart-time workers or temporary workers, often women or others returning tothe labour force, some with skills and others without. These people are employed,often at low wage rates with limited rights, to help overcome periods of peakdemand or shortage of full-time workers. Few companies are prepared to investin this flexible labour force and they have been labelled the ‘contingent worker’in the USA (Belous 1989; Section 4.4). A fourth leaf of the shamrock is the rolethat the customer plays in reducing the costs of the organization. We now do ourown shopping and make deliveries through the use of the car, fill our own carswith petrol and use self-service restaurants. A service is removed and can then bereinvented and charged as a new form of personal service.

The fourth dimension in the changing pattern of work travel is the effectthat technology, mainly telecommunications, might have on commuting(Nilles 1991). The original optimistic estimates have not been met and thelimited empirical evidence suggests that total travel (for work and otheractivities) is not reduced. However, there is no evidence that telecommutinggenerates greater separation of home and work (Pendyala et al. 1991).Regular telecommuting is uncommon and the likely impact is more subtleas increased flexibility is given to time management. Most telecommuterswork at home for one or two days each week and in the USA it is estimated(Handy and Mohktarian 1995) that about 1.5 per cent of the workforce (amaximum figure) telecommute on any given day. With the significantchanges in economic factors and the changing nature of work, it is likelythat the scale of homeworking will increase. Ironically, it may be a revisiting


of the homeworking which characterized manufacturing over a centuryago—this was a highly exploitative relationship. Homework associated withtechnology and a contingent workforce may mean that commuting distancesincrease with uncertainty, given multiple employment locations. Theprobability of changing jobs decreases the value of access to the presentjob, while moving costs determine the opportunity cost of moving (Crane1996a). Both factors lead to longer average commutes, flatter rent gradientsand greater residential decentralization. But as Giuliano (1996b) concludes,it is difficult to isolate location patterns between the different elements ofthe labour, from work patterns that are explained largely by socio-economiccharacteristics. She concludes that more precise measures of job uncertaintyand flexibility are required before clear conclusions on whether existingtrends of decentralization and growth in the popularity of high amenityareas can be reinforced.

Underlying these changes is the increased employment in service andtechnologically based industries. These new forms of employment alloworganizations to employ people in more flexible ways. Apart from changes inthe actual nature of work, there are changes in the conditions of work,payments, pensions, security of jobs and individual rights. The traditionalnotion of work has changed fundamentally over the last twenty years.

4.3 Economic changes

Integrally linked to the changing patterns of travel induced by work-relatedchanges are those resulting from the restructuring of the economy in post-industrial society. Recent theoretical arguments (e.g. Dosi et al. 1988) suggestthat technical change is forcing a transformation of the economy and thatnew processes of dynamic adjustment pose radically different challenges tothose allocative mechanisms postulated by traditional theory. Innovation isnow considered to be fundamental to economic growth as competition isbased on quality, not only price—this is the basic argument within neo-Schumpeterarian theory (Nelson and Winter 1982, summarized in Table 4.8).Although most of this analysis has been aspatial, networking (including faceto face contact) and location decisions are both significant determinants ofsuccess (Lundvall 1992; Sako 1992; Sabel 1993).

The spatial dimensions form the focus of recent research (Simmie andKirby 1996) where the historical considerations of industrial change arematched to those relating to new technology (Table 4.9). Many of thealternatives listed here are concerned with both the organization form andthe spatial distribution of innovations and high technology industry as theyin turn drive the economy. In particular the new role of the powerfulmultinational companies may be crucial in determining the spatialagglomeration (or dispersal) of new forms of economic activity. There arethe proponents of the new global economy (e.g. Henderson and Castells 1987),


Table 4.8 Long waves of development

Sources: Berry (1991); Hall (1996); Kondratieff (1935); Kuznets (1966); Mensch (1979);Schumpeter (1939).

Table 4.9 Historical and technological explanations of current industrial change

Source: Based on Simmie and Kirby (1996)Note: These choices are not mutually exclusive.


who argue that production is increasingly dominated by these multinationalcorporations with large research and development budgets who can determinehow the market will operate. On the other hand, there are those who maintainthat local places are equally important and that innovations take place insmall niche products (e.g. Storper and Christopherson 1987; Lundvall 1992).It is likely as with much innovation that the reality is variable and involvesall scales of operation and types of organizations. Whatever the explanation,it is clear that fundamental change is taking place.

At the macroeconomic level, globalization of the world economy has takenplace with capital centralizing and concentrating at the international level.The power of the multinational corporations bears witness to this process.The large enterprise can maximize profit levels by localizing its activitieswhere the labour rates are cheapest for the level of skill required. Regionsand cities become the pivotal nodes in the global network and control bothwhere goods are produced and how their dominant position in the marketcan be maintained. However, location does still seem to be important andnot all places are homogeneous as assumed here.

The concept of flexible specialization (Piore and Sabel 1984) argues thatfirms (particularly industries) are saturating the markets with a standard rangeof mass-produced goods. The demands of consumers are more differentiatedand as they become more sophisticated new products are required. Thesemore specialized products cannot be produced on the production line, sofirms have to adopt flexible strategies. This is a strategy of permanentinnovation where firms accommodate continuous change rather than tryingto control for it. Piore and Sabel (1984) conclude that flexible specializationconstitutes a shift of technological paradigm with the re-emergence of craftbased industries. These industries then develop their own networking withsubcontracting, thus reducing their complexity as organizations and allowingthe diffusion of innovation throughout the regional economy. There also seemto be agglomeration economies from firms grouping together in specificlocations.

However, much of the flexible specialization seems to relate to existingfirms trying to survive rather than innovative new firms entering the market.Many new high technology industries do not follow this pattern as they arecapital intensive, requiring a high level of skill and support infrastructure.Their location constraints would not allow them to develop as craft industries.More important though are the links between flexible specialization and themacroeconomic globalization effects. Far from allowing local-based craftindustries, these economic factors are moving the world economy towardsglobal integration and centralization of command and control. Where small-scale enterprises do exist, they are dependent upon the multinationalcompanies as these large-scale enterprise control the networks on whichflexible specialization depends.

The concept of Marshallian districts attempts to address the reasons why


spatial concentration takes place (Becattini 1990). Here, the argument usedis that industrial districts re-emerge, often as a result of small-scale initiativescoming out of the local university, or the local ethnic community, or largefirms adopting a survival strategy of vertical disintegration.1 Similar criticismshave been made here as those directed at flexible specialization. There arefew example of new Marshallian districts, the globalization issues are notaccommodated and it is difficult to identify the conditions under which suchdistricts might develop.

The third approach at an explanation of the global-local economic debateis to examine the competitive advantages of particular locations (Porter andVan der Linde 1995). Porter (1990) argues:

competitive advantage is created and sustained through a highly localizedprocess. Differences in national economic structures, values, cultures,institutions, and histories contribute profoundly to competitive success.The role of the home nation seems to be as strong as ever. Whileglobalization of competition might appear to make the nation lessimportant, instead it seems to make it more so. With fewer impedimentsto trade to shelter uncompetitive firms and industries, the home nationtakes on a growing significance because it is the source of the skills andtechnology that underpin competitive advantage.

(Porter 1990:19) Although the multinational corporations are powerful, they do not have tooperate within national boundaries, with governments having a majorinfluence on whether their competitive advantage is maintained or enhanced.Within countries, the same arguments can be applied at the regional andlocal levels, particularly as they relate to factor conditions, demand, supportingindustries and a firm’s strategy, structure and rivalry.

The spatial restructuring of the economy resulting from new patternsof working and technological innovation is a complex process and noteasily encapsulated in one theory. Within the globalization process, it doesseem that location is important, but there are many factors which mightinfluence that choice. Traditional theories based on neo-classical economicsare no longer relevant (Chapter 1), but there is a range of new spatialimperatives.

One such review of technopole development has been carried out by Castellsand Hall (1994). They identify four types of new high-tech developments: • complexes of high technology industries occupying new industrial spaces

or regenerating old ones;• sciences cities—Akademgorodok (Russia) and Tsukuba (outside Tokyo);• planned high technology business areas—Hinshu (Japan), Sophia-

Antipolis (South of France), Cambridge (UK);


• innovative metropoles—London, Paris, Tokyo, Munich, southernCalifornia;

Yet, even from the largest scale synthesis of the evidence, it is difficult toexplain why investment and development take place in one particularlocation rather than another. Traditional views on location decisions andthe potential for agglomeration economies suggest that there are economic,behavioural and technological factors (Dosi et al. 1988). Economic factorsinclude cost minimization, externalities and economies of scale. Behaviouralfactors cover qualitative measures of transactions and learning.Technological factors balance flexibility against being locked into theproduction process. More recent contributions (e.g. Giersch 1995; McCann1995) seek to draw a distinction between those incurred in overcomingdistance and the costs incurred from being located at one point in space.McCann (1995) identifies four elements that are key to microeconomictheories of location.

Distance-transaction costs mean that firms will locate together assumingthat they buy from the same supplier and sell to the same markets. Thesame argument is true of his second element, namely the location-specificfactor efficiency costs. In both cases there are agglomeration economies ofproximity and factor efficiency. But he qualifies this second argument forclustering by the statement that this occurs: ‘only where it is clear that theexisting level of agglomeration is the cause of the existing factor-efficiencyprices can we rightly talk about agglomeration factor efficiencies’ (McCann1995:573).

The two other elements of possible agglomeration economies are hierarchyco-ordination costs, which relate to the nature and stability of the productionand consumption hierarchy, and the hierarchy coincidence opportunity costs,which relate to levels of sales (the sales maximization principle). These factorsrelate to the existing numbers of firms and households and involveagglomerations of scale in the traditional sense.

McCann (1995) argues that underlying spatial economic questions areissues of the nature of production hierarchies that help to explain why spatialclustering takes place. It is only when factor prices are pushed up throughhigher wages that this clustering process might break down. Even thoughfirms may locate in close proximity to one another, this does not necessarilymean that they will have links with other firms in the same industrial sector(localization economies) or with other firms or households in the same area(urbanization economies). The linkages are increasingly important astelecommunications and networking, together with business related travel,form important components in service and technological based industries.However, they need not be local links. They could be regional, national orinternational links, depending on the activity in which the firm is engaged.McCann (1995) reaches the conclusion:


Many clustering situations are wrongly characterized as localization orurbanization economies, when the cost reason for clustering has little ornothing to do with the location of other firms, but rather is due to therelationship between local factor efficiency prices and the costconsiderations dependent on the location of suppliers and customers intotally different regions. The result is that authors (unspecified) thenwrongly attempt to account for this observed spatial clustering in termsof hypothesized information economies, in situations in which this issimply not appropriate.

(McCann 1995:573). Yet the evidence seems to be accumulating to suggest that there are stillagglomeration economies, even in high technology economy, and that thenew factors of production are instrumental in bringing this about. At thetheoretical level Kutay (1988a, b) has demonstrated that with two differentcosts (one for commuting and one for information), location depends on therelative balance between the two. When information costs are sufficientlylow that all workers work at home, the land-rent gradient becomes convexat the centre. This reflects diseconomies of agglomeration and employerswould seek to disperse to the periphery.

However, such simple explanations may not be entirely appropriate asvertical disintegration means that the clustering of suppliers would leadto agglomeration economies (Storper and Christopherson 1987).Outsourcing and subcontractors’ work can be best maintained throughclose proximity and continuous contact. Indirectly, the clustering ofemployment may also facilitate the growth of short-term, temporary andflexible work patterns, as job accessibility is important to those inemployment or seeking employment (Sections 4.2, 4.4). The traditionalview (Vickrey 1977) that activities cluster geographically provided thatthe agglomeration benefits outweigh the congestion costs needs to bereviewed, as congestion is only one part of the location decision. The newstructure of the labour market and the changing structure of businesses,together with the role of technology, all need to be included in the newunderstanding of agglomeration. For example, the work of Romer (1996)suggests that the congestion cost curve flattens as technology improves,so agglomeration economies may be felt over a wider area, and this inturn would lead to footloose location. Others (e.g. Simmie 1998) arguethat local factor production costs and qualities are critical innovationinputs. These are not the traditional factors of infrastructure,telecommunications, land and buildings, but the new ingredients of theknowledge-based economy, such as skilled labour, venture and risk capital,new technology and new knowledge and information.


4.4 Technological change

The main source of profit and power in the late twentieth century is knowledgeand information, and conflicts are likely to occur over the distribution of andaccess to that knowledge. Even money is becoming less important and tangibleas transactions are carried out electronically, only to be seen in a symbolicform on a screen. This global view of the shifts in power presented by Toffler(1991) and summarized above will transcend all activities at all levels andwill create a radically different society. The technological revolution is thethird great change in modern society, following on from the agricultural andindustrial revolutions.

Apart from the transport effects which are dealt with in this section andthe more general spatial impacts (Section 4.5), technology impacts on theorganization of work. Technology has resulted in convergence within andbetween industries, with vertical disintegration, flexible production methodsand specialization (Capello 1994). It has also had important implications foremployment as it can make use of low-cost labour or specialized horizontalintegration. Flexibility means that production must be sensitive to changingmarkets. It also implies a flexible labour force with less job security and ahigh turnover of staff, with the consequent wage polarization (Appelbaumand Alpin 1990). The numbers of core workers have been minimized,workforce is being outsourced to cheaper labour, contingent workers aretypically part time, working for lower wages and fewer non-wage benefits(Golden and Appelbaum 1992), as with Handy’s shamrock (Section 4.2).They have a weak affiliation with their employer. Normally they have nocontract and have only been employed in the recent past. Belous (1989)includes the self-employed, temporary workers, part-time workers and thosein the business service sector as they have only weak employer affiliation. Inthe USA these workers account for between 30 per cent and 37 per cent ofthe total labour force (1988) and this level has increased by 25 per cent since1980.

More recently, the impacts of information and communications technologyhave pervaded much of our everyday life, but they have not been uniform;nor is there a strict cause and effect relationship. There is tremendous potentialfor further developments in technology that can affect all aspects of people’slifestyles and the way in which business and industry is carried out. Yet thereis also a reluctance among users to accept the new technology withoutquestion. Given both the potential and the constraints, there are at least fourways in which the availability of knowledge and information will radicallychange transport demand.

First, the production and distribution processes introduced by Henry Fordat his Highland Park assembly plant in 1913 are now being extended andreplaced. The conveyor belt now extends beyond the manufacturing systemto the distribution system. Technology and information allows a complete


service from the assembly of materials through the production of the car tothe testing and distribution processes, and the delivery to the final consumer.These concepts do not only apply to the manufacture and distribution ofvehicles, but to all commodities. Freight distribution systems have beenrestructured on regional and metropolitan warehousing depots, often ataccessible motorway intersections. Road transport informatics (RTI) impacton all parts of freight transport operations as well as location decisions. Withthe trend in Europe towards longer distance trucking and the increased useof multimodal combinations of vehicles and carriers, integrated approachesto freight transport are essential to ensure the optimal use of information andthe new flexibility in both production and distribution processes. The potentialis available to develop a Europe-wide integrated freight transport network,but old barriers still remain, namely who should pay the costs of pollution,the increased resource costs caused by the growth in international road freightand the compensation of individual member countries for transit traffic—‘the territoriality issue’.

The EU in its recent Transport Policy White Paper (CEC 1998) is trying toharmonize the user pays approach to replace the patchwork of chargingarrangements. These different charging systems create competitive distortionsbetween and within the different modes of transport and between EU memberstates. In the road sector, five EU countries levy road tolls, six use theEurovignette scheme, others run different systems (e.g. for bridges and tunnels),or do not charge directly for road space (Section 3.3.3).

Second, there is the belief that technology (in particular RTI) can help indelaying the inevitable gridlock when the city comes to a complete stop throughcongestion. Traffic management schemes have been very effective in squeezingmore capacity out of a given road network and the expectation here is thattechnology through intelligent highways and smart cars can continue thatprocess. Increased flexibility in work and leisure patterns together with thepossibility of telecommuting have all provided the opportunity for change.Again, it should be noted that both information and knowledge have beeninstrumental in creating the conditions for this opportunity. However, eachrevolution in the past has resulted in increases in travel and average trip lengthsand there is no reason to expect a change as a result of the current revolution.

Road users will be affected in three different ways: • Information services to the traveller that will allow decisions to be made

on the basis of the best real time information. These services would applyequally to public transport services and to route guidance informationgiven to the car driver. This is likely to increase intermodality in tripmaking and allow extra flexibility in response to real time changes.

• Control systems within the vehicle. By the year 2000, it is estimated that10–15 per cent of the costs of new cars will relate to RTI services (LexMotoring, 1992), being provided within the car.


• Control over the transport network, including demand management andtraffic control systems. These are already in common use.

Third, the infrastructure network forms one major key to an integratedtransport system. Investment in new roads and in upgrading and expandingexisting roads (e.g. through additional lanes) will be complemented by thenew high-speed rail networks (the TENs—trans European networks—inEurope) and telecommunications networks, including the new value addednetworks (VANs) and the local area networks (LANs). It is this combinationof networks that will facilitate the most fundamental changes brought aboutby knowledge advances and information technology. These include logisticsplanning, electronic data interchange, electronic route guidance, emergencytransport planning, information systems and databases for environmentalmonitoring (Hepworth and Ducatel 1992).

The fourth element is the more general impact that teleactivities willhave on the demand for travel and location decisions. The new info-networks (e.g. VANs and LANs) allow activities to be carried out remotelywithout the need to travel. Much of the current debate has been over theimpact on commuting patterns, but even greater opportunities lie inteleconferencing, teleshopping, telebanking and forms of teleleisure(including participatory game shows). In principle, there is less reason totravel in a physical sense (Section 4.3). But the simple substitutionarguments have been widely discredited (Mokhtarian 1996). Adaptationsto the availability of technology are much more subtle than the simplesubstitution argument suggests. Opportunities are increased and a newflexibility in travel arises. This means that the impact of the new technologyis much more varied as different responses are made by individuals andfirms to accommodate the opportunities within the context of their ownrequirements. It should also be noted that not all individuals and firmsare competing under these new market conditions. As with all innovations,particularly those involving technology, there are knowledge and costconstraints which limit the market, at least in the short term. The newflexibility described here impacts primarily on the wealthy, well-educatedpeople who have the knowledge and resources to use the technology. Thereare strong distributional implications resulting from both the availabilityand use of the new technology.

As a consequence, the spatial imperative may no longer apply as cities willbecome much looser spatial organizations, as the costs of urban centralityand high land prices will be balanced against the benefits of dispersal. Themovement out of cities will continue with only front office functions remaining.Growth will be concentrated in corridors of good communications and atperipheral urban locations where it is cost effective to link in with both thetransport and information networks. Peripheral areas may still remain isolatedand separate from the new infrastructure as access costs and capacity


requirements may make the installation costs of the new networks uneconomicand the costs of using the system too high.

The most attractive locations will be those where the transport andinformation networks link in with other factors such as a skilled labour force,a high quality environment and the availability of low cost land. Interchangesmay provide particularly suitable locations for logistical platforms.International airports, high-speed rail (e.g. TGV) stations, and majormotorway intersections could all provide sites of maximum accessibility whichwould minimize location and transport costs, and also be on the internationalinformation network.

There is no question that very significant changes are taking place on boththe political and technological fronts in Europe, North America and elsewhere,and technology is likely to have a profound effect on transport at all levels.However, the exact nature and scale of that impact is far from clear and itseems that implementation will not be equal across all transport sectors, butwill be selective and will take a considerable time for the full effects to becomeapparent.

More generally, the technological revolution is impacting on all forms oftransport, on location decisions and, more fundamentally, on the way in whichwe do things. The new informatics and telecommunications networks havecaused a revolution in data handling, processing and transmission. This inturn impacts on all households and firms and has substantial effects on therelative competitive advantage of particular regions and its importance islikely to increase (Capello 1994), as is its impact on the form and function ofcities.

4.5 Global cities and spatial change

As with the economic and technological factors, cities are also changing asthere is intense competition for global status. There is perhaps room for three(or four) cities in this group, distributed around the globe so that at least twoof them are functioning at any one point in time. Hall (1996) names London,New York and Tokyo as his three. Their unique position is based on theproduction of specialized services (e.g. financial services, media services,education and health services) which control the global movement of capitaland information. They also carry out important continental and nationalfunctions (e.g. government, culture, tourism, services and manufacturing).The key question at the turn of the century is whether they can maintain theirdominant position as other cities compete for these lucrative functions. Thiscompetition is particularly evident in Europe where the traditional dominantposition of London is being challenged. The two cities in Japan and the USAare less vulnerable as these countries have a major share in the global markets,both in terms of production and consumption. But the UK is being challengedby other major economies within Europe as there is no single dominant partner


here. Individual cities are challenging for parts of the information servicestraditionally located in London (e.g. Paris and Frankfurt). The EU is a majorplayer here as the tradition is to allocate key European functions around themember states. The recent decision to locate the European Central Bank inFrankfurt has significantly weakened London’s banking and financial servicesdominance, as has the previous decision to locate the European InvestmentBank in Luxembourg.

Europe has a rich array of the second tier world cities—Amsterdam,Brussels, Copenhagen, Berlin, Vienna, Prague, Rome, Dublin, Lisbon, Madrid,Stockholm, Helsinki, Athens, Paris. These are the capital cities of Europewith a rich history of government, culture, education, tourist and commercialactivity. In addition, there are major regional capitals that have particularexpertise (e.g. Zurich, and Frankfurt for banking; Barcelona and Milan forcommerce; Geneva for international agencies). Similar examples can beidentified in the USA (e.g. Washington for government; Chicago and SanFrancisco for financial services; Los Angeles for culture and entertainment)and Japan (e.g. Osaka for trade).

This global and world hierarchy is connected by the main internationalairlines, the new telecommunications systems and high-speed railwaynetworks. In the future, rather than one city within a particular locationhaving dominance in all functions, it would seem likely that cities will becomenetworked so that even the specialized information services become devolved.The great unknown in the pattern of urban development is the effect that thephenomenal growth in the emerging countries will have. The rate of economicgrowth in the Pacific Rim countries and Brazil is such that these rapidlygrowing economies with newly emerging mega cities will also be seeking alarger share of the economic benefits arising from the globalization ofeconomic activity and the technological revolution. Similarly, the transitionof the countries of central and eastern Europe, together with the CIS countries(Commonwealth of Independent States, the old Soviet Union), from commandeconomies to market economies will eventually lead to a new balance ofpower between the global cities.

One of the major limiting factors in the redistribution of power is thequality of the transport and communications infrastructure. Despite recentinvestment, the cities that are not in the global or world category are less wellconnected than the key centres. Even though they may have well-educatedand technical competent workers, the opportunities available to them aremore limited. This in turn encourages the mobility of labour to the higherorder centres which increases the dominance of the centre. Even though muchrecent investment has been made in improving the transport andcommunications infrastructure between the major countries of the world andbetween the established centres and more peripheral regions (particularly inEurope, Japan and the USA), it is unclear whether this has increased the


attractiveness of the centre or has made the peripheral regions morecompetitive (Chapter 5).

Apart from the competition between global and world cities, the natureof the cities themselves is also changing. This has been commented upon(Section 4.2) with respect to commuting patterns, but the spatial structureof cities has become much looser as residential suburbanization took place.This was followed by the decentralization of employment to more spaciousand cheaper sites, often close to the road network. As manufacturing andassembly plants were replaced by research and development opportunitiesassociated with high technology production, further decentralization tookplace. Most recently, the non-essential functions of many organizations havebeen relocated to peripheral sites, often not even in the city. With the adventof cheap and reliable technology, these ‘back office’ functions can be carriedout in locations where office space and suitably trained labour is cheap.They can even be carried out within the home or at a local telecentre wherethe overhead costs are minimal. One of the major disadvantages of theglobal cities is that rent levels and labour costs are very high. It is oftendifficult to attract suitable labour, particularly if the housing andenvironmental costs of living in the city are included (e.g. congestion, airpollution, safety and security).

The net result is that the global and world cities have vast labour marketareas as long-distance commuting becomes attractive. The time taken (andthe costs involved) of commuting by rail or car are balanced against thesubstantial benefits of lower housing costs, better educational opportunitiesand a higher quality of life. Again, the high-quality road and rail networkhas allowed new patterns of commuting to develop. The advent of the mobileoffice and mobile communications has allowed commuting time to be usedmore profitably and has also permitted more flexible work patterns (Section4.2). In Japan and Europe, the new high-speed rail networks have ‘shrunk’space so that it is often faster to reach city centres from more distant locationson the high-speed network than it is to reach the city centre from nearer(suburban) locations which are not on the high speed network. There is atime-distance inversion effect.

We seem to be moving towards what Webber has called the non-urbanrealm where communities share activities and exchange information(Webber, 1963). This communication could take place locally, at the cityscale, worldwide or through mixed contacts. Although people might live incities, there was no inherent reason why this should be so. The necessity forcontinuous face-to-face contact has passed and dispersed cities operate justas well as concentrated cities. It is not a spatial problem, except in thatlarger cities tend to have lower communication costs, but even these aredeclining.


4.6 Implications for travel demand and economicdevelopment

These four major trends in the economy will have enormous implications fortravel demand over the next twenty years. In each individual case, there couldbe substantial growth in demand, but when the effects are taken together theimpacts are likely to be even greater. There is a synergy between these work-related, socio-economic, technological and spatial factors, which in turn willhave dramatic effects on network performance and transport behaviour.

An examination of the trends in mobility over the last ten years (1986–96)gives a feel for the potential exponential growth over the next ten to twentyyears. In the fifteen countries of the EU, there has been a 30 per cent growthin the numbers of cars and taxis, with car ownership now averaging over 400vehicles per 1,000 population. Travel measured in vehicle km and passengerkm has also increased by a similar amount with the total amount of passengertravel per head of population increasing to about 9,000 km per annum.However, the overall pattern has considerable variation which cannot onlybe explained by population, size, area density or economic factors (e.g. GDP)alone.

The greatest increase in car ownership has been in the peripheral EUcountries with low levels of GDP per capita (Portugal, Greece and Spain).Growth has also been high in Luxembourg, Italy, the UK, Finland and Austria,but from a higher base. Denmark has had the lowest levels of increase, followedby France, Germany, Sweden, the Netherlands, Ireland and Belgium. Even inthe USA, the growth in car ownership has been over 32 per cent and the‘assumed’ saturation level of 650 cars per 1,000 population has now beenreached. The increase in car ownership in the USA over the decade has beengreater than that of many EU countries.

Over the next fifteen years, there will be a significant further increase incar drivers and the number of cars in many European countries, so that theoverall level will reach about 550 cars per 1,000 population in 2010. Thislevel is similar to that in the USA in 1985. There are likely to be 60 millionnew cars in the fifteen EU countries (+38 per cent) and much of the growthwill take place in households where there is already one car. It is unlikely thatthe road capacity will increase by anywhere near the same amount, socongestion will increase.

Substantial growth has taken place in car use with the dominant positionof the car being reinforced as the main mode of transport. Only in Japan,where there are long journey distances, high densities and substantialinvestment in the Shinkansen high-speed railway, has the proportion of travelby rail been increased (to 37 per cent of passenger km). Even here, it could beargued that the particular circumstances of megalopolis along the easterncoast of Japan has made such action essential. There is no urban developmenton the same scale, density or intensity anywhere in the EU or the USA.


Stability is often assumed in the relationship between car ownership andcar use. Evidence from Britain and other European countries suggests thattrip lengths have increased significantly and that the lower mileage recordedby second cars in car-owning households is outweighed by the increasedmileage recorded by households obtaining their first car. In countries wherecar ownership is still increasing and where structural changes in the economyare taking place, both the numbers of trips made and the distances travelledwill continue to increase leading to greater congestion.

Two sets of conclusions can be drawn from this chapter. It seems that thecombination of all these four major economic changes will result in an increasein travel demand, perhaps of an even greater scale than has been seen in therecent past. That growth will not be in the work purpose, but in the leisureand other purposes. Globalization effects in terms of the economic argumentsand the spatial arguments means that economies of scale and agglomerationboth continue to exist, but there is a high level of transport intensity as newforms of production and specialization take place. Both the labour marketcharacteristics (flexibility and fluidity), and the production characteristics(flexibility and specialization) are changing radically. The great unknown isthe role that technology will actually play apart from facilitating change andthe distribution of power.

In most of these macroeconomic, spatial and technological changes,transport has been seen as an essential given. It has been assumed that therewill be the means to transport goods and people, locally, nationally andinternationally, at a speed, quality and price that can be accommodated withinthese changes. In other words, the evolving economy is not transportdependent. The conclusion must be that this new agenda is independent oftransport or that there is a sufficient supply of transport to allow it to happen.

This conclusion has been in part tested through the Centre for Economicsand Business Research (CEBR 1994) model of the UK economy where differentlevels of road expenditure were tested for their impact on congestion andemployment (Table 4.10). Two basic options were tested in the model, a 50 percent increase in roads expenditure and a 50 per cent decrease in roadsexpenditure for motorways and trunk roads (i.e. those roads principallyconstructed and maintained by the national government through the HighwaysAgency). The figures given here are interesting as the argument that roadinvestment has little effect on traffic growth is clearly supported. There seemsto be little variation around the expected increase in traffic (57 per cent to2010). However, congestion is expected to rise and speeds fall irrespective ofwhether investment is kept at current levels or increased or decreased.Congestion will increase by 7.6 per cent under the increased investment option(13.4%-5.8%) and by 21.6 per cent under the reduced investment option(13.4%+8.2%). Translated in monetary costs of lost time, the numbers aresubstantial and even larger when related to the effect on GDP. The full economicimpact takes time to reveal itself, as the effect on GDP in the first five years


with the 50 per cent reduction target is only 0.24 per cent (1998), but the effecton growth and employment then increases to 1.1 per cent (2010). Accordingto the CEBR report (1994), reductions in road investment would directly affectthe construction sector in the first instance, and then its suppliers. Over time,the costs imposed by rising congestion would have an additional impact onconsumer expenditure and the costs to industry and business resulting fromhigher transport costs. This in turn would lead to lower levels of investmentand in the longer term have a negative impact on competitiveness. The increasedinvestment would have the reverse effects and both sets of figures would bemoderated by interest rates as public expenditure would be reduced or increased.

This macroeconomic modelling approach suggests that investment levels inthe road infrastructure are strongly related to economic performance, measuredby transport indicators (congestion and speed), by economic indicators (GDPand employment) and by wider indicators (inflation rates, balance of payments,levels of public expenditure). If the assumptions used in the CEBR model areaccepted, then the arguments used in this book are not supported. The evolvingeconomy is very transport dependent and the existing levels of supply are notsufficient for the maintenance of the competitiveness of the UK economy. Itseems that the evidence from the wider literature cited here is at odds with themore specific macro modelling approaches applied directly to the assessmentof the impact of road investment options on employment.

Perhaps, the reality is somewhere in between. But one also needs clearly toestablish the explanation in terms of whether it is due to the methods beingused or the actual situation. Cost benefit analysis has been the main methodused for the evaluation of public sector investment decisions in the transportsector (Chapter 7). It was developed in a time when unemployment was low

Table 4.10 Road investment, congestion and employment in the UK for alternativeinfrastructure investment levels in 2010

Source: Based on CEBR (1994).Notes: Target year (2010) changes are given in this table.Current levels of expenditure is about £2bn per annum.To stabilize congestion over the period to 2010 would mean raising petrol prices to £15 per gallon

(1994 prices or £3.33 per liter) or a 5.50 times increase in real terms.Real costs of fuel to rise by 3.54% per annum in the figures given above.All values given in 1993 prices.


and takes no account of the employment effects. The methodological argumentwas that transport investment decisions have high sunk costs and low marginalcosts, so that the prices charged should be close to zero to maximize use.Although revenues were low, the consumer surplus was large and this couldbe recycled into productive activities, principally private sector growth. Thismay in turn result in new jobs being created, but more often it led to higherproductivity (and profits). Financial rates of return from public projects couldbe increased if government tax revenue from accelerated growth leads tohigher levels of employment and company taxes being paid. The multipliereffects are substantial, depending on the output elasticities used and the impliedrates of return. However, all these additional benefits are dependent on theassumption that the consumer surplus is recycled into productive activities,in particular higher levels of employment, rather than higher levels ofcapitalization and profits.

A broader based macroeconomic approach to evaluation of transportinfrastructure investment, such as that used by the CEBR (1994), attempts toinclude these multiplier effects through a variety of methods. Feedback isexplicit as investment brings lower labour costs resulting from a wider (moreaccessible) labour catchment area and a closer matching of skills to jobs.This in turn may also lead to higher productivity. These first round effectsmay be accessible locations and raised labour costs (unless there is a surplusof labour). Again, assumptions are made about the link between investmentand employment rather than substitution effects resulting in higherproductivity and output without any additional labour.

Other effects are also important, particularly at the broader programmeor regional level as new investment may result in induced travel as thepropensity to travel may increase with the improved infrastructure quality.Similarly, barrier conditions need to be considered as actions taking place inone location may affect other locations, particularly in peripheral locationsor through the means by which projects might be financed (geographical andfunctional barriers).

The financing and taxation issues are also important. The taxationcomponent in publicly funded projects is often substantial, but it can bereduced if construction workers would have been unemployed, leading toshort-term welfare savings. It can also be reduced if economic growth isstimulated. CEBR calculations on London Underground suggest that ‘adynamic approach to fiscal analysis of projects can show them to have apositive effect on the Public Sector Borrowing Requirement within a relativelyshort period of time, largely because of their contributions to faster economicgrowth’ (quoted in EUROCASE 1996:56).

The net cost to the economy of a project is therefore unclear as it increasesdemand and this may in turn affect costs. On the supply side, highergovernment spending on public projects may lead to increases in borrowingand interest rates in the short term and levels of taxation in the longer term.


On the demand side, this may result in price and wage inflation that reducesthe real value of output gains and this in turn is inflationary. These monetary(supply) and resource (demand) effects are known as ‘crowding out’ (in aderegulated or ‘free’ market the effects maybe reduced as capital is moremobile).

In all of this discussion, the two critical factors are the links betweentransport infrastructure investment and economic growth, together with themultiplier effects on employment. Most of the modelling and evaluationapproaches make the assumption that there is a link and then establish thesize of that relationship. The other effects on taxation and public finances allflow on from that relationship. The arguments presented in this chapter onthe evolving economy would suggest that the link is weak and becomingweaker as labour markets become more flexible and fluid and as technologyhas a greater role in the production process.

In addition is the network argument which suggests that the changes inaccessibility and reduced transport costs brought about by new investmentare small in most developed countries. These changes are not likely to affectthe production processes of companies, particularly where the transport supplychain forms a relatively small part of the total production costs. All of thesequestions are crucial both to our understanding of these fundamentalrelationships and in developing appropriate methods for analysing them.

Note

1 Vertical disintegration takes place when a company restructures itself and contractsout much of its activity to outside organisations.

Social, spatial andenvironmental effects

5.1 Introduction

Although economic factors and competitiveness have been the driving forcesfor most national governments, there is a growing realization that otherpriorities are also important in balancing the dominance of the growtharguments. Principal among these complementary priorities is the changingstructure of the population itself, both in terms of its age profile and newforms of family structure. In addition, there is the fact that the currentgeneration is the first to have had almost unlimited access to the car and theopportunity for high levels of mobility.

The second great social issue is the concern over equity. This covers bothsocial equity and spatial equity. Not all society has equal access to transportand the growth arguments have to be balanced against the distribution of thebenefits. Provision has to be made for those without access to a car or publictransport and governments have always tried to provide opportunities inremote and peripheral areas, often through transport investment. Projectswith lower levels of economic return have to be balanced against the regionaldevelopment benefits and reductions in levels of isolation.

The most recent addition to the debate is the growing importance ofenvironment, particularly as it relates to sustainability. Most governmentshave made a firm commitment to stabilize levels of CO2 emissions at their1990 levels in the year 2000. This commitment was made under theFramework Convention on Climate Change (Rio Summit 1992). In thetransport sector, the level of CO2 emissions is directly related to the amountof fossil fuels used. Stabilizing emissions requires a combination of moreefficient use of fuel and less travel, but the trends in most countries are in theopposite direction. Further investment in transport may help achieve morefuel efficiency, at least in the short term, as relief of congestion allows thetransport system to operate more efficiently. But, in the longer term thosebenefits may be offset by the growth in traffic. In addition to the CO2

stabilization targets there are many other environmental costs associated withtransport. In many cases, the objectives of transport investment must be to

Chapter 5


identify situations where the economic, equity and environmental factors allpoint in the same direction. It is not a simple trade-off between these factors,but a clear policy challenge. We want investment in opportunities that leadto economic development, with a more equitable distribution andenvironmental benefits—this is the ‘win-win-win’ situation.

In terms of the more specific discussions relevant to this book, there areimportant considerations about urban form and structure and the new meansby which investment decisions should be made. The economic anddevelopment benefits have to be supplemented by distributional andenvironmental benefits. Here we discuss the issues and principles and inChapter 7 we present the analytical evidence.

5.2 Demographic changes

The patterns of mobility are similar in all fifteen European Union countriesand those in the wider European economic area (EU countries and EFTA).The levels of mobility are somewhat lower than those in the USA, but higherthan in Japan. In all cases, except for Japan, there is an overriding dependenceon the car for travel (Table 5.1). The growth in population is slow, with thetotal population being stable at around 370 million in the EU15, 260 millionin the USA and 125 million in Japan. The current low levels of fertility will bemaintained, at least into the next century.

5.2.1 A geing and changing family structures

However, within this relatively stable population, two major changes can bedetected. The most significant growth in population will take place in theelderly and non-working population as a joint effect of the increases in lifeexpectancy and the tendency to retire earlier. The proportion of elderly peoplein western Europe (over 65 years) will rise from 13 per cent (1985) to 20 percent (2020). This means that for the OECD countries (Organization forEconomic Co-operation and Development 1985), the number of elderly peoplewill increase over that same period from 85 million to 147 million. Thepopulation of most advanced economies is becoming greyer.

In addition to the ageing of the population, there are other importantchanges taking place in demographic terms: • Average household size is expected to continue to fall from current levels of

2.7 persons per household to 2.4 persons per household (2010). Householdsize reflects the lower fertility rates and births outside marriage. In the 1970s,divorce rates doubled in Belgium and France and tripled in the Netherlands.By 1986, births outside marriage accounted for nearly half the total births inDenmark and Sweden (Masser, et al. 1992). The concept of a traditionalhousehold with two adults (married) with children is no longer valid. Indeed,

Social, spatial and enviromental effects 109

it may never have been true. The structure of households is no longer basedon the family unit as there are now so many variants.

• Perhaps, at the micro level, this means that analysis should be based onthe individual rather than the household as a unit. Similarly, much of tripgeneration is based on the assumption that certain activities relate tohousehold size. There are common activities in which all householdshave to participate. However, if the unit of study, namely the household,no longer exists or exists in many different forms, then it becomes difficultto establish a clear methodological framework for analyses. Perhaps eachperson should be treated as an individual with particular characteristics,not as part of a household unit, but this weakens the concept of family orjoint activities. The question of whether the most appropriate scale foranalysis should be at the individual level or in some grouping (family orother) is unresolved, as activities are undertaken individually, as part ofa family, as part of a group or as part of some other arrangement.

Table 5.1 International comparisons, 1984–94

Source: Transport statistics Great Britain (UK Department of Transport 1997).Notes: Car includes cars and taxis. EU15 had 158 million cars and taxis in 1994 and 116 million in

1984.


• The increased participation of women in the labour force is also apparentin all European countries, particularly the growth in part-time working.There are still substantial differences between European countries in thelevel of female participation with a particular growth in the Mediterraneancountries (e.g. Italy). This southern growth is from a lower starting point,but the national differences may also reflect the different cultures andtraditions

• The implications for work-related travel are likely to be substantial asmany households have two wage earners, so the location decision maynot be optimal for even one of the workers. It has been argued (e.g.Boddy and Thrift 1990; Banister and Bayliss 1992) that households nowestablish a residential base and career needs are met by (long-distance)commuting. The broader implications of the new patterns of work havebeen discussed in Chapter 4.

5.2.2 The motorization effect

The second major change has been the motorization effect. The current cohortof elderly people in the EU are the first to have experienced the use of the car alltheir lives and they will not want to give it up. To expect that today’s elderlypopulation will adopt the travel patterns of the elderly of yesterday is unrealistic.The implication of this argument is that dynamic approaches must be developedto account for the desire of individuals to maintain the ability to drive as longas possible. A demographic analysis of car ownership and use patterns takes asits starting point the growth in licence ownership and car ownership for differentage groups of men and women (Madre and Lambert 1989; Banister 1994). Inthe USA nearly 90 per cent of the adult population have driving licences andthere are, on average, nearly two vehicles per household. The distances travelledby residents averages at over 29,000 km per household. These levels are aboutone-third higher than those in the EU.

The latest evidence from the GB national travel surveys demonstrates theincreasing dependence on the car over the last twenty years and the gradualgrowth in mobility (Table 5.2). Over that period total travel per person hasincreased by 37 per cent with the car’s share also increasing from 68 per centto 78 per cent. The proportion of travel by all other modes has declined(except van/truck) both in absolute terms and as a proportion of the totaltravel. The growth in travel is explained in equal parts by the increase in tripsmade and the increase in the average journey length.

The picture painted here is of substantial further increases in travel overthe next fifteen years, resulting from demographic changes. Two possibleinfluences might limit this growth. First, broader environmental concernsmight lead to the realization that unconstrained growth is not desirable andthat alternatives to travel by car and air must be sought. To achieve such achange may not be possible in the short term, but the birth of a new ‘green


generation’ who are prepared not to travel so much, particularly byenvironmentally damaging modes of transport, might trigger such a change.At present there seems to be no evidence of such a movement. The second isthe argument that in Europe distances between cities are relatively short andthat the road infrastructure is not so well developed as that in the USA.Consequently, the need to own a car and the ability to use it is not so dominantas in the USA and therefore saturation levels of ownership may be lower inEurope than in the USA. Again, the evidence is limited. This saturation levelmay just be wishful thinking, rather than one based on a true understandingof consumer choice and marketing by the vehicle manufacturers.

In terms of identifying future trends and the impacts on travel demand,certain conclusions can be drawn. Within the overall patterns it seems thatparticular sections of the population may travel more by car. Women and theelderly are two groups that have traditionally driven less than other people.Fundamental changes have taken place in women’s participation rates in thelabour force, their greater independence and the increase in ‘non-standard’households. These changes would all suggest that increases in their travelpatterns (including the increases in numbers of trips made, trip lengths,complexity of trips and use of the car) would be greater than average. Similarly,with the growth in life expectancy, health, aspirations and affluence of theelderly, one would expect that they would keep the car for as long as possibleand make greater use of it in their extended retirement. It is unrealistic to

Table 5.2 Distance travelled per person per year in Great Britain (km)

Source: UK Department of Transport (various).Note: Other includes London Underground, motorcycle, taxi/minibus, bicycle, other private

and other public.1 The top figure includes short walks and the bottom figure includes only trips over 1.6 km

in length.2 The top figure is the average journey length for all trips and the bottom figure is the

average journey length for all trips over 1.6 km.Figures in brackets are percentages.


expect that elderly people in the future will have the same travel patterns bymode as a similar elderly group today.

In summary, there are at least three compelling arguments that wouldstrongly suggest that trip rates by mode for particular groups would notremain stable in time: 1 Present-day expectations and travel patterns will influence aspirations in

the future. This cohort effect will be most apparent with the elderly whoare the first generation to have experienced mass car ownership and socan be expected to continue to use that mode as long as possible.

2 The growth in leisure time and the high value now being placed on thequality of life, and the importance of stage in life cycle: life-cycle changesrefer not only to the four basic conventional groups (i.e. married coupleswith no children; families with young children; families of adults; retired),but to the wide range of unconventional groups (e.g. single-parentfamilies). Changes in lifestyle and life-cycle effects have had fundamentalimpacts on the range of activities that people require, the increasingcomplexity of travel patterns and the increase in travel distances.Complementary changes have also taken place with the structural changesin the economy and changes in the distribution of industry, commerceand retailing which have tended to follow the decentralization ofpopulation.

3 The increase in levels of affluence and the unprecedented growth in carownership levels: some of this affluence has resulted from the growth inwestern economies, but the greater part has been the growth in savingsand wealth from property value increases. That new wealth is likely tobe used by the newly retired elderly or passed on to their next generation.

It seems that the demand for travel will continue to increase but the nature ofthat demand may change as a result of demographic factors. Although thechanges in population structure are important, other changes (such as theindustrial structure, technological innovation, levels of affluence and leisuretime) will also influence demand. The problem here is in unravelling thecomplexity of issues so that the effects of one group of factors can be isolated.Similarly, there is a range of policy instruments that can be used to influencelevels of demand and mediate between the different interests.

5.3 Spatial and social equity effects

These general changes brought about by the aging and motorization effectsconceal other fundamental changes within the population. Transport as withother commodities will never be available to all people equally, nor will it bedistributed equally over space. As we have already seen, there areagglomeration economies (Section 4.3), and income and age constraints will


mean that not all people will have equal access to facilities and services. Evenif it were available to all equally, it would not be ideal as different people(and businesses) have different requirements. Many of these requirementshave already been mentioned in the changing patterns of work and leisure,the changes in family structure and the changes in business organization. Inaddition to the changes in patterns of demand, there have been significantchanges in the distribution of services and facilities (the spatial factors areconsidered in Section 5.5). The basic issue here is that of accessibility tofacilities, both at the aggregate level and for particular groups of people.Accessibility relates both to the physical distribution of land uses within theurban areas and the availability of transport, and to the needs of the peopleto use the services provided. Access is a function of both travel times and thenumber and quality of nearby destinations (Handy 1993) and the valuedifferent people place on access to different destinations also varies.

The regional dimension is important as investment in infrastructure is oftenjustified on the basis of improvements in accessibility and an increase ineconomic performance. The arguments, particularly on causality, have neverbeen clear (Vickerman 1995) and the fundamental process of regionaleconomic development leading to convergence or divergence is still intenselydebated (e.g. Romer 1986; Krugman 1991a; Vickerman 1994). Traditionalarguments of regional economic growth are aspatial. Neo-classical theoriessuggest the free movement of resources are seeking higher marginal returns.Keynesian theories view regional variations in aggregate aspatial analysisand these are both at odds with microeconomic approaches to theunderstanding of location.

Recent arguments have stressed the importance of space and the existenceof increasing returns as the basis for understanding the spatial economy. Therationale here is that increasing returns explain the separation of productionand the spatial concentration of industry. However, the assumptions used inthese models have also been questioned (Krugman 1991a) as they are basedon perfect competition assumptions of free entry and common levels oftechnology. In particular, the importance allocated to transport costs in thesemodels is too great. The question here is the degree to which competitivenesswill improve from transport cost reduction. As noted throughout this book,transport costs are a small part of total production costs, yet they seem tohave been given a disproportionately large role in explaining competitiveadvantage and location decisions of firms. With the advent of a hightechnology, service-based society, with flexible labour markets and high levelsof skills, transport costs alone cannot explain which locations are mostattractive.

The ‘new growth economies’ (Romer 1986) emphasizes economic growthas endogenous to an economic system, rather than as the result of outsideforces. It is the differential quality of factors of production, including theskills and knowledge of the labour force, which are internal to the economic


system that explains the differential growth. Yet even here it is difficult todraw tight boundaries around systems as much of the development takesplace within a national, international or global context. Regional boundariesare not geographically based. The ‘new economic geography’ (Krugman1991a) argues that imperfect competition models, together with economiesof scale, can best explain location decisions. The Krugman (1991a) argumentis that a tension exists between convergence factors, such as market size(agglomeration economies) which leads to concentration and divergencefactors, such as the competitive elements which allow disadvantaged regionsto maintain production (a deconcentrating effect). From the analyticalresearch, it seems that changes in transport costs can lead to concentrationor deconcentration, depending on existing cost structures, the elasticity ofsubstitution and the initial quality of the transport infrastructure (Krugmanand Venables 1990). If transport costs are very high, there will be adecentralizing effect unless the local markets are small or the extent of scaleeconomies substantial enough to outweigh the transport costs (Vickerman1995). If transport costs are low, concentration will take place, unless thereare substantial local markets and low-scale economies that would justify alarger number of locations.

The difficulty comes then in trying to relate location and land use factorsto the regional development arguments, particularly under conditions ofincreasing returns to scale. Firms may move to locations where new scaleeconomies can be achieved through cheap raw material (and labour) inputs.But the transport costs are not only the distance-related costs, as they shouldinclude all the other quality factors related to the production process—theintegrated logistic chain (Chapter 4). As Vickerman (1995) clearly argues,the main problem is one of aggregation. It is not appropriate to have a singleapproach, as the infrastructure is important for the individual firm, but theaggregation of individual benefits, does not necessarily lead to regionalbenefits. There are many other factors at work (Gramlich 1994). Two basiceconomic arguments determine the role that transport investment decisionshave on the spatial distribution of economic development. The non-spatialarguments examine the aggregate level of economic activity in terms ofproductivity and competitiveness. Infrastructure is seen as a public good whichenhances the productivity of the private factors of production, or it combineswith private capital in an optimism ratio to raise productive potential. Sparecapacity leads to new opportunities and internal scale economies or externalagglomeration economies. Conversely, bottlenecks limit growth potential.The alternative set of spatial arguments emphasizes the differentialperformance of the different locations, together with the tension betweenforces of convergence and divergence. The actual effects are much moredisaggregate in nature and depend on the individual firm, its competitiveposition, economies of scale and imperfect competition.

These distributional questions need to be balanced against the actual


infrastructure investment decisions made over the last twenty years and theimplied government thinking on causality. Transport infrastructure investmenthas fallen as a proportion of GDP and it is not clear that firms have respondedin particular ways in particular locations. New forms of operation have beenintroduced to maintain competitive position and reduce production costs (e.g.technology, logistics and new organizational structures, Chapter 4). Thesemay be included as second or third round multiplier effects. Industry has torely upon new forms of operation to increase productivity and efficiency ifnew investment in transport infrastructure does not take place.

5.4 Environmental and sustainability effects

In 1996 the transport sector was responsible for over 25 per cent of worldprimary energy use and 23 per cent of CO2 emissions from fossil fuel use. Itforms the most rapidly growing sector with energy use in 1996 at about 70EJ1. Without action, this figure will double to 140 EJ in 2025. Industrializedcountries will contribute the majority of this figure until 2025. After thatdate, the majority of transport related emissions would come from thosecountries that are currently developing rapidly or have economies in transition.Transport activity increases with rising economic activity, disposable income,access to motorized transport and falling real vehicle and fuel costs.

Projections of transport greenhouse gas emissions follow the historic trendsas CO2 emissions are directly related to energy use in the transport sector.The assumptions made are that the relationships between transport fuelconsumption and variables such as gross domestic product (GDP), fuel pricesand vehicle energy efficiency will remain stable, at least until 2025 (Grübleret al. 1993). More recent research (e.g. Acutt and Dodgson 1998) suggeststhat the relationship between energy use and economic factors is not stableand that, in Europe, car ownership and use may saturate at lower per capitalevels than those found in the USA and Canada. In addition, technologicalinnovation may result in greater levels of mobility being achieved with lowerlevels of energy input. There are also strong political and economic argumentsfor breaking the historic links between transport demand, energy use andeconomic factors as has happened in the energy sector (Peake 1994; Banister1996).

Despite the many advantages brought about by the car and other transport,there are also serious negative consequences for society as a whole (Table 5.3 ).

The environmental costs of transport have been grouped under four mainheadings—pollution, resources, environment and development (Table 5.3).Decisions taken to improve benefits along one dimension may be likely toincrease costs along another dimension or in another sector. The complexityof decision making in environmental policy cannot be underestimated, butall governments must now face difficult choices. More detailed discussionson the elements of the environment, the role of transport, and policy measures


can be found in Johansson (1987), Banister and Button (1993), Whitelegg(1993), Maddison et al (1996), Banister (1998).

Many of the environmental costs of transport are non-linear in their effects(e.g. health effects and congestion). The crucial issue becomes not how tomeasure, but how to avoid reaching critical levels where the environmentalcosts become too high (e.g. lethal doses of pollution). Still, the measurementdifficulties are substantial and placing values (or money costs) onenvironmental factors tends to be subjective (Button 1994). It is only recentlythat these issues have become central concerns in evaluation (Chapter 7) andin decisions on infrastructure investment.

The challenge for environmental policy in transport is to improve as manyelements of this complex interrelated list of environmental costs as possible(Table 5.3) without increasing those elsewhere, or at least being aware ofthem and making an informed choice. It should also be remembered thattransport is only one (albeit important) part of the economy and so theenvironmental choices in the transport sector need to be balanced againstother priorities. It is argued that transport infrastructure investment has socialbenefits (e.g. bypasses of congested town centres) but that it destroys theenvironment (e.g. through the generation of more car travel).

More recently, the environmental arguments have been linked to those ofsustainability. This more sophisticated view links environmental concernswith those of economic development and equity. To achieve an objective ofsustainable development, at least five different sets of objectives need to beaddressed. In this section we have been concerned with the environmentalobjectives. The second objective is to maintain competitiveness througheconomic growth and development objectives (Section 5.3). Where possible,the environmental and development objectives should be working in the samedirection—this is the ‘win-win’ situation, and many transport investmentdecisions have tried to achieve these benefits. For example, as noted above,bypass schemes have been justified both by the economic benefits from reducedtravel times and by opening up new areas for development. But they havealso brought environmental benefits to town centres. In addition to thesetwo fundamental objectives, the concerns over sustainability present threenew objectives. The equity objectives (Section 5.3) are concerned with thedistribution of costs and benefits to society, both socially and spatially. Theseintragenerational effects are contrasted with the intergenerational objectives(futurity), highlighted by the most often quoted definition of sustainabledevelopment, namely—‘development that meets the needs of the presentwithout compromising the ability of future generations to meet their ownneeds’ (WCED 1987). The final objective is participation in its widest form.Too often in the past, decisions have been made without the support of theaffected parties. To achieve the objectives stated for sustainable development,we have to carry out our daily activities in different ways, using resourcesmore efficiently. Similarly, industry and the new post-industrial economy need


to be more sustainable in their operations and organisation. This requiresclear policy directions through pricing, regulation and control, but the scaleof change necessary to achieve sustainability objectives also requires politicalsupport from all affected parties. Unless this support is forthcoming, littleprogress will be made.

Underlying much of the debate over the environment and sustainability is

Table 5.3 The environmental costs of transport

Sources: Based on Banister (1993b) and Whitelegg (1993).Note: SSSI—Areas of special scientific interest, which can only be developed in very special situations.


the crucial link between the environment and competitiveness. Much of theliterature discusses the balance between these two dimensions. We wouldargue that the most productive way to actually achieve sustainability objectivesis through these two dimensions operating in the same direction—the ‘win-win’ situation. Porter and Van der Linde (1995:97) have argued that the‘struggle between ecology and the economy grows out of a static view ofenvironmental regulation, in which technology, products, processes andcustomer needs are all fixed’. In the static world, firms have made cost-minimizing choices. Environmental regulation raises costs and reduces marketshare of domestic companies on global markets. They go on to develop adynamic paradigm, based on innovation and the capacity to improvecompetitiveness through shifting the constraints. Properly designedenvironmental standards can trigger innovation, which may more than offsetthe costs of compliance. We would go further and suggest that environmentalincentives should be used to promote greater efficiency and innovation. Apositive promotion of environmental incentives is one way to achievesustainability objectives and gain public support through the demonstrationeffects of policy actions.

Transport policymakers have always accepted that transport imposesenvironmental costs. However, the scale of the problem and its nature are nowmuch greater. There is much more transport today than there has been in thepast and that trend is likely to increase, particularly with the emerging economiesof central and eastern Europe and those of the Pacific Rim. In the longer termthe greatest growth is likely to be in China and India. In addition, the continuousgrowth in air transport has added a further new element of travel. But it is therapid growth in car ownership and use that forms the most important factorin assessing the environmental costs of transport. It is clear that the simplegrowth in transport poses environmental de gradation, but there are alsosubstantial qualitative factors (Table 5.3). For many years transport policy hasbeen primarily concerned with the local problems of transport, principallycongestion, accidents and noise. The debate in the last twenty years has becomemore sophisticated and complex as the broader impacts of transport haveembraced both global and international effects, as can be seen from the followingexample.

Concern over the damaging effects of ‘acid rain’ on forests and water lifegrew in the 1970s and 1980s. The importance of NOX and other gaseousemissions from cars were recognized as major contributing factors. Concernalso emerged in the 1980s over high-level ozone depletion and its impact onthe long-term incidence of skin cancer. Transport’s role was relatively small,as it was confined to CFCs in air-conditioning units. In the 1990s globalwarming has become the key issue with its impact on raising averagetemperatures and the consequences for climate change and sea level rises.The new agenda requires the collective action of national governments andinternational agencies in limiting the growth of CO2 emissions.


The scientific evidence is powerful, but the essential catalyst for changehas been public concern over environmental issues. In part, this interest is areflection of increased affluence and being able to afford to take action onenvironmental issues. More fundamentally, there is a growing concern overthe long-term future of the planet and a commitment to sustainability.Transport has become a key element in that debate and one that is beginningto attract a disproportionate amount of attention. Transport contributes atall levels to environmental degradation (local, global and transboundary). Ithas a high profile, it is perceived as being a major intruder and it also interactswith other activities which can be seen as being environmentally harmful(e.g. tourism).

Transport is also seen as an area where governments can and haveintervened through fiscal measures, regulation and the planning system. Manyof the environmental costs imposed by transport are the consequences ofpolicy decisions made for other reasons (e.g. regional development). Actionto improve the environmental quality should also rest with governments, butgovernments will only act if there is a direct political benefit, and/or if thereis sufficient public support, and/or if there is some international agreement.This is the irony of the debate and it is reflected in the inconsistencies inpeople’s actions and the inability of governments to take effective steps.

People are aware of the environmental costs of transport and are supportiveof actions by governments to improve environmental quality, provided thatthey result in no change to their lifestyles and they can continue their use ofthe car, and provided that it does not increase costs. This is a problem withno solution. It accounts for the general resistance against higher prices intransport so that some of the environmental externalities can be internalized.It accounts for the belief that technological solutions will solve the problemthrough more efficient engines, alternative fuels (e.g. electricity or hydrogen)and add-on technologies (e.g. catalytic converters). It accounts for the focuson positive-planning policies to reduce journey lengths through higher densitiesand concentration of development in larger settlements. It accounts for theuse of public awareness campaigns and raising the social consciousness togain public support for actions that are often politically unpopular. In short,even if we could establish clear links between health quality and amount ofmotorized travel, this might not be a necessary condition to radically changepolicy direction. There will always be strong reasons to continue to keep theexternal costs of transport as externalities and to resist the strongenvironmental case for internalizing them.

5.5 Urban form and structure

As with the debate over whether road infrastructure reduces congestion andvehicle emissions or leads to a more dispersed and inefficient pattern of landdevelopment (Downs 1992), there is also uncertainty over the most efficient


urban form. Urban form covers the spatial configuration of fixed elementswithin a metropolitan area, including the pattern of land use and density andthe supporting transport and communications infrastructure. There is someagreement over the different urban form types (UK Department of theEnvironment 1993). 1 Urban infill. Here the aim is to make maximum use of urban sites to

accommodate development. Increasingly, the terms urban compactionand intensification are used.

2 Urban extensions. This is the suburbanization solution that has beenpopular, but rarely questioned.

3 Multiple village extensions. This has resulted from pragmatic approachesto planning rather than a selected solution. It is often unpopular,particularly with residents in the villages.

4 Key villages. This concept was popular, but has not been used recently. Itwas argued that key villages would help maintain rural services and publictransport.

5 New settlements. In the 1980s there were many schemes for privatelypromoted new settlements, but few have actually been started. There isnow a renewed interest in the concept as the housing market is strongerthan in the last ten years.

In addition to urban form types, settlements need to be considered in relationto one another, not solely in isolation. The compact city results from higherpopulation sizes and densities in the city, with high quality accessible publictransport. The edge city encourages development at selected peripheral pointstogether with increased investment in orbital roads to link the edge cities(Newton 1997). The corridor city focuses growth along linear corridorswhere high quality public transport is available, while the fringe cityencourages general suburban development along the road network. All thesepossibilities apply to individual cities and also to city regions (Banister etal. 1997).

Most urban areas do not conform to any one type as patterns ofdevelopment are continually changing. Overall, it is clear that there has beena flattening out of density gradients, but there is still no consensus over whatis the most desirable urban form in terms of energy efficiency andenvironmental quality. Even if there was some agreement it may not be possibleto achieve that pattern of development.

If there is little agreement over the ideal urban form which is both energyefficient and environmentally attractive, there is even less consensus on therole that transport plays. The well-publicized debate in the American literaturebetween the ‘Stalinist views’ of Newman and Kenworthy (1989) and the‘Friedman views’ of Gordon and Richardson (1997) highlighted thesubstantial differences. The much-cited Newman and Kenworthy (1989)


review of thirty-two world cities by urban density and energy use in transportacted as the focus of the debate. Questions were raised about the quality ofthe data, the importance of particular cities in shaping the curve (e.g. HongKong and Moscow), the relatively small differences between cities in thesame part of the world and the policy conclusions, particularly about therole of public transport. The contrast between Newman and Kenworthy’sconclusions and those of Gordon and Richardson (1997) could not be morestark.Gordon and Richardson concluded ‘that urban sprawl is atransportation solution, not a problem’. The argument was that there is adynamic process which is continuously at work. As urban sprawl takes place,jobs follow people so that the journey to work length remains relativelyconstant over time. Gordon and Richardson base their analysis on US journeyto work data. But the journey to work is becoming less important. Householdsoften have more than one worker and the growth in travel is taking placefor other trip purposes (Ewing 1997). Most commentators do not take theseextreme positions, but are content to focus on the intermediate issues ofreducing trip lengths, encouraging moderate concentration, specializationand mixed use (Banister 1997).

Transport seems to play an ever-decreasing role in the location decisionsof households and businesses (Giuliano 1996a), but there still seems to be anidentifiable localized link with journey lengths (Cervero and Landis 1995a),even for the journey to work. These are shorter in ‘balanced than unbalancedareas’ (Frank and Pivo 1994). Even if development took place in transit-oriented development (TOD) or in more traditional neighbourhoods, somecommentators (e.g. Calthorpe 1993; Crane 1996b) suggest that the cost oftravel by all modes would increase. Others again argue that shorter journeysmean more journeys, as travel time budgets are fixed (Gordon andRichardson 1997).

One popular element of the debate is the role that telecommuting andother forms of technological substitution might have on travel. The originaloptimistic views that we would all stay at home and communicate have beenreplaced by more sophisticated arguments (Mokhtarian 1996). In California,it is estimated that 6.1 per cent of the workforce telecommutes on averagefor 1.2 days a week. This means that about 1.5 per cent of the workforcetelecommutes on any given day and this accounts for about 1.1 per cent ofvehicle miles travelled. When considered with total household travel it amountsof 0.7 per cent of all travel. The reductions in the future may be less ascommute distances of telecommuters fall closer to average, and as thestimulation effect grows. Mokhtarian’s conclusion is that the aggregate travelimpact ‘will remain relatively flat well into the future’, even if the amount oftelecommuting increases considerably.

Perhaps there is a greater potential in other activities as firms downsize,leaving traditional city locations and have a dispersed workforce distributedin locations (even homes) where the labour and overhead costs are much


lower (Section 4.4). Similarly, telephone banking, teleshopping, cataloguemarketing and other services may offer a greater travel reduction potential.These types of transactions are not dependent on a personal contact, as withmany work-related activities, but on a purely impersonal relationship. Theevidence of the impact of telecommuting and other forms of ‘teleactivities’on residential location is not conclusive. They may encourage more dispersedlocational patterns (Hopkins et al. 1994; Lund and Mokhtarian 1994), orthey may have little direct impact (Nijkamp and Salomon 1989; Nilles 1991).As with many other innovations, ‘teleactions’ allow a wider variety of actionsand an increased flexibility in what types of actions can be carried out.

The links between land use, urban form, sustainability and transport arecomplex, and the role that infrastructure investment can have in this process isunclear. Some would argue for a balance between jobs and housing (Cervero1989) to minimize trip lengths. Others urge neo-traditional neighbourhood design(Calthorpe 1993) to bring the small scale back to cities. Others look towardstransit-oriented development (Cervero 1994) to influence mode choice. Yet theoutcomes are still unproven as the variety and scale of responses have beensubstantial. In particular, it seems that it is difficult to get the car user to leave thevehicle at home and use other forms of transport. Similarly, the complexity of thelabour market and the distribution of facilities means that journey lengths havealso become longer (Boarnet and Sarmiento 1996). New investment in transportinfrastructure will always facilitate more travel, particularly by car. Even if theinvestment is in rail or public transport, mode switchers (to public transport)make it easier for non-mode switchers (car drivers) to use their vehicles. Similarly,the new opportunities provided by technology make it easier to carry out work,shopping and business-related activities from home or a local telecentre. Againthis new flexibility provides freedom to organize everyday activities on a user-oriented basis, reducing the intended effects of land use policy interventions.

5.6 Implications for economic development

The new agenda relating to the social, spatial and environmental effects ofinfrastructure investment has been outlined in this chapter. The apparentlysimple causal relationships between investment and economic developmentdo not hold and may never have been appropriate. There is no agreementwithin the literature and there are sufficient new elements to question thevalidity of any relationships established. It is impossible to unravel the fullcomplexity of the arguments, yet it seems that many actions will haveunexpected results.

For example, the construction or expansion of the existing infrastructuremay affect the emissions levels both directly and indirectly. By reducing levelsof congestion (at least in the short term), new roads would allow traffic toflow more smoothly at a faster speed. This means that there will be lesspollution per unit distance up to optimal speeds (about 50 mph), with the


exception of NOX that increases with speed. The indirect effects reflect growingcar dependence, longer journeys and changes in land use, all resulting fromadditional road capacity. The net result is that total travel increases (Mackie1996), with additional fuel being used and higher pollution levels. In cities,these effects are likely to be even more pronounced, provided that congestionis reduced. Vehicles operate least efficiently under congested conditions witha high proportion of stop-start driving. In addition, many city journeys areshort and vehicles operating under cold start conditions also use more fueland create more emissions. Catalytic converters are ineffective when cold. Itis here that most VOCs, CO2 and NOX emissions are made.

To get round the problem of conflicting emissions factors, the US CleanAir Amendment Act (1990) required an analysis of the net impact of allinfrastructure projects in non-attainment areas. These non-attainment areasare designated in metropolitan areas in which national air quality standardsare not met. Actions are required in all non-attainment areas to reduceemissions levels and reach the preset targets within a given time frame that isset according to the severity of the problem. Ironically, the greater the severity,the longer the time given—one extreme ozone non-attainment area (LosAngeles) has been given twenty years (US Department of Transportation(DOT), Bureau of Transportation Statistics 1996:195). Similar air qualitystandards are being introduced in Europe, with local authorities having newresponsibilities to measure emissions levels. If limits are exceeded, the airquality management areas will be designated and strategies evolved to improvethe environmental quality, including the possibility of banning cars at certaintimes and under certain conditions.

This new emphasis on the environment and sustainability has also resultedin a reassessment of the road building alternative. It is now seen as beingineffective in the longer term and having undesirable environmentalconsequences in the short term. New roads generate more travel. The evidenceon induced demand in the USA is less clear than in the UK and Europe. It mayreflect the existing levels of congestion within the transport network. In Europemuch of the strategic network, together with the urban road network, iscongested at certain times of the day. New infrastructure only gives short-termrelief as existing traffic is redistributed across the network. But it also seemsthat there is a substantial latent demand which is released (Mackie 1996). Inthe USA, the growth in traffic is usually less than the capacity added, even overa longer period of time (Downs 1992). Again, there is considerable doubt overthe true nature of latent demand—whether the demand curve has actuallychanged, or whether it reflects a movement along the demand curve. The formercase would be considered new demand (e.g. changing tastes and preferences ofindividuals), but the latter case is really only a response to changing economicconditions (e.g. lower generalized travel costs).

Current methods are not sufficiently flexible to relate emissions to trafficflow and the impacts of induced travel. So it is impossible to determine whether


a new highway has a positive or negative effect on emissions (TransportationResearch Board 1995). Once a road has been built, the negative environmentaleffects in terms of emissions, land take, noise, intrusion and accidents (Table5.3) are likely to be felt for many years. It is both difficult to establish soundmethodologies for the measurement and assessment of the environmentaleffects and to establish an appropriate time span over which those effectsshould be considered. Road infrastructure investment decisions mean thatthe new link will be present for many generations.

The environmental and sustainability issues have also resulted in a similarquestioning of traditional concepts of physical accessibility. New technologyand life-styles that are increasingly leisure based have changed the importanceof certain activities, particularly work-related. Much of this change has resultedfrom the recent increases in living standards, higher real income levels anduniversal access to technology. The service and manufacturing base of theeconomy is being replaced by a knowledge—and skills-based economy. Newforms of lifestyle and production have made it possible to create much greaterflexibility in travel behaviour and in the location of industrial, commercialand residential activities.

This new dynamic is continuous and creates changes in travel demandpatterns, greater complexity with circumferential movements around citiesand reverse commuting. It has also created new demands for social and leisurebased activities both in the urban area and nationally and internationally.Isolating transport’s role in the land development process is difficult,particularly in urban settings. In the USA, much of the low-density residentialdevelopment has been facilitated by the availability of cheap land, the lowcosts of motoring and high quality of the road network. Employment hasalso shifted from the congested and high-rent central business districts to thesuburbs and the metropolitan areas have now expanded to becomemulticentred cities. Between 1982 and 1992, built-up and urban land in theUSA has increased by 14 million acres. As a result, developed land in theUSA totals 95 million acres or about 5 per cent of the total land area (excludingAlaska) (US Department of Transportation, Bureau of Transportation Statistics1996:164).

New forms of economic development have replaced the traditional notionsas we pass through a transition phase from a work-based to a leisure-basedsociety (Handy 1995), and as the new forms of production take place withassociated changes in the organization and functioning of companies. It isimpossible to present the full range of the changes that are taking place. Oneexample illustrates this well. The economic effects of developing the trans-European transport network (TENs) may not be as great as originallypredicted. The construction of the Ecu 90Bn ECU (£75 Billion) fourteenpriority projects agreed at the Essen European Council could create between130,000 and 230,000 new permanent jobs, but it now emerges that thisestimate is based on two strong assumptions: that there is no significant


economic overcapacity; and that these priority projects would not affect otherrelated projects (a crowding out effect). The net effect on jobs may even benegative if investment is simply diverted from other projects (EU DGII 1997).The conclusion is that new opportunities may be created, but it is not clearwhether they will be taken up. Economic benefits will accrue from increasedoutput and from social cohesion and environmental benefits, as most of theprojects are rail investment in core and peripheral locations (Banister et al.1998).

Note

1 EJ=Exa Joules. Joule is a measure of energy (kg m2 s-2) and Exa is 1018.

Methodology: analyticalapproaches and modelling


In the review of approaches to the analysis of the links between transportinfrastructure and economic growth, we concluded that most research hasconcentrated on the macroeconomic level. We also concluded, that althoughstatistical relationships can be established at this level, it is difficult to constructcausal relationships that support the data as the effects of external factors,time and stage in development will all influence the direction and strength ofthose relationships.

In this part we develop the analytical aspects of this book in three relatedways. Chapter 6 explores the production and consumption effects ofinvestments through a review of the macroeconomic literature on modellingpublic infrastructure and growth. Two basic types of models are reviewed,the production function models and the cost function-based models, andimpacts are assessed in terms of productivity gains and economic growth.These approaches are essentially positivistic and their success (or failure) isbased on the strength of the empirical relationships developed.

Chapter 7 examines the economic evaluation of transportation projectswhere decisions have to be made about investments in the infrastructure withthe expected outcomes being measured in terms of economic development.This is a more normative approach which takes a rational perspective of thenetwork characteristics and examines the economic impacts of investmentalternatives. Issues such as accessibility, value of time, discount rates, thetime span of projects, risk and uncertainty, as well as multiplier effects andnew forms of financing are all included here. Crucial new questions are raisedabout the method of financing infrastructure investments (see also Chapter4) and the allocation of risks between the different agencies involved withthe construction of new transport infrastructure.

However, the primary objective of this part of the book is to propose analternative analytical framework capable of showing how development ofthe transport infrastructure can affect economic growth at the micro level.While the concept of economic growth was explained earlier it should be

Part III

128 Methodology

re-emphasized that our focus is on local economic growth, namely that ofcities and regions. Therefore, in Chapter 8 we construct a simple model oflocal economy consisting of production (firms), consumption and laboursupply (households) and transport sectors. By changing a key parameter inthis system (i.e. the capacity of the transportation system) we can observehow it affects growth, mainly the equilibrium level of employment in thiseconomy (see diagram above).

Under market conditions investment in the transportation system willnecessarily affect land use and, as a consequence, a host of other variablessuch as consumption, labour demand and supply and travel behaviour.However, from a normative economics viewpoint the objective of aninvestment in transportation infrastructure is not to enhance theperformance of the transportation system per se but rather theenhancement of social welfare. Quite obviously it is much easier to defineand measure system performance than social welfare. It follows that notevery investment in infrastructure should be undertaken, even if it improvessystem’s conditions such as traffic flow, unless it can also improve economicwelfare conditions, such as the equilibrium level of employment orconsumption.1

Another important observation is that it is possible to affect the behaviourof consumers by changing relative prices rather than by a direct investment.For example, it is possible to mitigate congestion levels by increasinggasoline tax without any change in the system’s capacity. Indeed, in manycases transportation planners will obtain better results if they can setcorrectly the price of travel rather than directly regulating traffic quantitiesand use of modes.2 However, for certain infrastructure facilities (including

The complementarity of approaches.

Methodology: analytical approaches and modelling 129

transportation), investment by the state can be highly warranted, providedit is possible to show social welfare improvement.

It has been observed that a key factor shaping urban land use patterns isthe dominant mode of use during the initial stage of development. Cities thathave adopted fixed rail (above ground and underground) have developed aquite different urban form than cities that use buses, trams and private carsas their predominant modes. However, this modal effect on urban growthpattern tends to abate as cities reach maturity and rate of suburbanizationgrowth far exceeds the rate of development of this mode. In the analysis inChapter 8, we assume an in-place transportation system and focus on theeffect of an incremental development of this system on local economic growth.This system can be a highway network or fixed rail provided its capacityserves as a constraint on travel times and flow.

The types of analysis presented in this part of the book are complementary,as they examine the links between transport infrastructure investment andeconomic development at three separate contexts and scales, using differentanalytical methods and approaches. In essence, they are examining the sameproblem at the policy and project levels. The difference is that the argumentsused are based on different premises. We argue that in developed economieswhere the transport networks are already well established, any new link islikely to have a very small impact (probably immeasurable) on GDP growthand on regional accessibility. Actually to identify effects, the scale of analysismust be at the local level where impacts can be measured by location decisionsof firms, labour supply and measures of output, such as productivity.

Notes

1 We disregard here equity issues which can arise when a transportation investmentis made. These equity effects are transportation related as well as economic growthrelated. For example, in older cities the suburbanization effect of infrastructuredevelopment has led to the immigration of skilled labour to suburbs, therebyleaving behind unskilled poor households. See Chapters 5 and 8 for a discussion.

2 It has been estimated that a 10% increase in gasoline price will retard excesscommuting by 15%.

Modelling the growth effectsof transport capitalinvestments

A macro level analysis

The sovereign has the duty of erecting and maintaining certain public worksand certain public institutions, which it can never be for the interest of anyindividual, or small number of individuals, to erect and maintain becausethe profit could never repay the expense to any individual or small numberof individuals though it may frequently do much more than repay it to agreat society.

Adam Smith, Wealth of Nations, 1967 edition

6.1 Introduction

In Part IV we examine several case studies of the effect of transportationinfrastructure investment on urban and regional development.1 A majorconclusion from these studies is that such developments are difficult to measureand to substantiate. The main reasons for this are the scale of the analysis,unaccounted for spillover effects and the presence of many other interveningvariables, mainly counterproductive local policies. But what about macrolevel (national or state) analysis where these effects are either inconsequentialor can be controlled for, as can the effects of other key aggregate variables?In this chapter we examine this issue with regard to two key questions. First,does the level of the infrastructure stock (primarily transportationinfrastructure) affect national or state economy growth? Second, if it doeswhat is the marginal contribution, from additional investment in public capital,on factor productivity?

Beginning with the seminal paper by Aschauer (1989a), many empiricalstudies have established a statistical link between the level of the publicinfrastructure stock and between economic growth and productivity. Thebasic argument, as put forward by Aschauer, is that rather than crowdingout private investment, public investment stimulates it by increasing the rateof return to private capital (quoted in US, DOT 1992). Yet, from a theoreticalperspective we might ask why, in the first place, should there be such a linkage?And if it exists, what is the economic mechanism that underlies theserelationships? The importance of answering these questions emanates from

6

132 Methodology

the scientific tradition of requiring empirical models, here used to measureand estimate economic phenomena, to be founded on theoreticalconsiderations regarding the cause and effect and functional relationshipsbetween principal variables. In addition, for policy purposes, it is ratherimportant to understand the conditions under which an additional investmentwill engender growth. For this reason Section 6.2 examines theoreticalarguments in modelling the impact of public infrastructure on growth andproductivity. Section 6.3 surveys several types of analytical models found inthe literature that link infrastructure capital with economic growth. Majorempirical results from these models are presented and evaluated in Sections6.4 and 6.5. Principal conclusions are presented in Section 6.6.

At the outset we must emphasize that only relatively few studies, found inthe literature, have examined the effect of transportation capital accumulationon economic growth. Most studies have examined the impact of aggregatepublic capital outlay or a subset of it as, for example, all capital dedicated tohighways, sanitation and sewage, electric and water utilities (e.g. Munnell1990a; Holtz-Eakin 1993), or highways, water and sewers (Morrison andSchwartz 1996). Nonetheless, we survey these studies since transportation isa major component in the country’s or state’s total public capital stock (about15–20 per cent) so that expansion of the transportation infrastructure capitalis likely to affect growth in the same way as the expansion of other capitaloutlays. Furthermore, it is conceivable that the same causality mechanismwhich links changes in total capital stock with economic growth can alsoexplain the relationships between transportation capital and growth. Hencethe importance of reviewing these studies.

The studies reviewed in this chapter focus mainly on economic growthand productivity gains from government spending on infrastructure capital.2

It goes without saying that other types of government spending, such as onresearch and education programs, can also influence economic growth.Furthermore, the studies reviewed here, by and large, considered economicgains accrued to the private sector only, in particular to manufacturing, therebydisregarding possible gains to other sectors. Consumers, for example, willrealize welfare gains from government spending on infrastructure facilities,which act to abate negative environmental externalities. Hence, the resultsfrom the studies reviewed here do not cover the full scope of effects frompublic capital investments.

6.2 Theoretical considerations in modellingpublic infrastructure and growth

As the name implies, public infrastructure capital is provided by a publicagency (whether national, regional or local) which, for brevity, we refer to as‘the government’. Two fundamental questions underlie the modelling of theeffect on economic growth of public infrastructure provision. First, why should

Growth effects of transport capital investments 133

the government be engaged in this activity for which it has to set complexand costly institutions whose role is to fund, build, operate and maintaininfrastructure facilities? Second, what is the structural mechanism by whichinfrastructure development affects economic growth? These two questionsare interrelated as the analysis of the functional relationships betweeninfrastructure and growth requires first that we understand the rationale forthe government involvement in infrastructure supply.

6.2.1 Rationale of government provision ofinfrastructure3

Arguments for public supply of infrastructure facilities have a long history ineconomic thought (see the quotation from Adam Smith on p. 131). Put inmodern terms, a standard textbook argument for the provision oftransportation infrastructure by the public sector is that, left to the privatesector, these facilities would be produced at a substantially sub-optimal sociallevel or not at all. Nevertheless, these facilities contribute positively andsignificantly to the national economy and to social welfare.

Briefly stated, the generic name given in the economic literature to thisphenomenon is market failure. It refers to such phenomena as public goodand scale economies in facility construction, in the assembly of massive unitsof land necessary for building and connecting large transport networks, andin securing right of ways. The generation of externalities, positive andnegative, by the provision and use of transport facilities is another marketfailure argument put forward to explain public supply of such infrastructurefacilities. Accordingly, the internalization of these externalities, which isnecessary for the optimization of social welfare, can be achieved only if thepublic sector owns and control the capacity and level of utilization of theseinfrastructure systems.4 Private provision of such facilities would be sub-optimal, as private investors would not be able to internalize these externalitiesas the government can.

To illustrate, consider the case of a public good where exclusion ofindividuals from its consumption is rather costly or unfeasible while the long-run marginal costs of servicing an additional user, given capacity constraints,are negligible. Inter-city and inter-state highways, local streets and feederroads, as well as forms of mass transit, are typical examples. Under theseconditions, individuals have an incentive not to reveal their true preferencesregarding their desired level of consumption of these goods, thereby benefitingfrom their provision without having to bear the associated costs (the freerider phenomenon). Consequently, private enterprises would not be able toearn sufficient revenues to recover the capital and maintenance costs ofbuilding, operating and maintaining these infrastructure facilities. Therefore,these facilities would not be produced at all or will be produced at a level,which is substantially below the optimal social level. Indeed, historical records

134 Methodology

show that the provision of transport facilities like local roads, turnpikes,canals and bridges in the long run could not be supported by the privatesector. The main reasons for this are heavy losses induced by the inability toenforce excludability, recover capital costs and competition from substitutablefacilities and modes (Taylor 1951).

It might be asked, however, to what degree does the government attain theobjectives implied by the rationale outlined above when undertaking capitalinfrastructure investment projects? Given the case, providing a satisfactoryanswer to this question is a formidable task which requires detailedinformation (in many cases, unattainable) and careful welfare analysis.However, it should be observed that, by and large, the above rationale isregarded by many as a ‘maxim’ that does not require analytical examinationor empirical verification. The numerous documented cases of the so-called‘government failure’ cast a doubt on this perspective. We next turn to evaluatetheoretical arguments linking infrastructure investment with economic growth.

6.2.2 Interrelationship between infrastructure investmentand economic growth: some theoretical arguments

Conceptually, economic growth can be defined as enhanced individuals’ utilityfrom increase in the aggregate quantity of goods and services they consumeand from a larger variety of these goods and services available in the economy(Quigley 1998).5 For measurement purposes, economic growth can be definedin terms of the annual rate of increase in the per capita level of output.Alternatively, it can be defined in terms of enhanced productivity of inputfactors. Quigley (1998) notes that these consumption and productiondefinitions are analogous relative to the underlying conditions necessary forincreased utility or aggregate output.6

Given these definitions, the question is how to model the relationshipsbetween public infrastructure development and economic growth in order toascertain empirically the degree to which the former affects the latter. Thecausality interrelationships between infrastructure investment and economicgrowth, embedded in the models reviewed shortly, are based on twofundamental premises: • that infrastructure capital expansion increases the efficiency and

profitability of the business sector;• that this increase stimulates business investment in private capital

(Aschauer 1989a, b). It does so by enabling firms to exploit scale and agglomeration economies,by increasing their efficiency through market expansion and competition, bybetter utilization of inputs, by linking disconnected markets and by makingfirms and markets more receptive to innovation leading to further growth.


What are the conditions necessary for the validity of these premises? Belowwe examine some key functions, focusing on transportation infrastructure asour major illustration of public capital and its effect on economic growth.7

In order to establish causality relationships between infrastructureexpansion and economic growth we must assume that infrastructure, such astransportation, is indeed a consequential intermediate input in privateproduction processes. Its ample supply at low costs to users8 is thereforeconjectured to have a positive impact on economic growth by stimulating theproduction of a large number of final goods and services that use publiccapital as a significant input factor.

A corollary condition is that transportation capital stock must be acomplementary factor to some private input factors in order to spur furtherinvestment in private capital. If public transportation capital is a substituteto all private input factors and if it is correctly priced, the use of these factorsmay decline if additional investment in transportation infrastructure is made.To illustrate, it has been alleged that increased telecommuting, through higherprivate firms’ investment in telecommunication facilities will enable employeesto benefit from flexible working hours and locations (e.g. work at home),thereby increasing their productivity. If, however, telecommunication andtransportation facilities are substitute factors, added investment intransportation infrastructure may discourage private investment intelecommunications, thereby reducing potential labour productivity gains fromtelecommuting. It follows that in specifying an empirical model for theestimation of the effect of transportation infrastructure expansion on growth,the model’s structure should allow for the estimation of substitution andcomplementarity relationships between private inputs and public capital.

As already explained, in order to evaluate the effect of public capital ongrowth it is necessary to specify a causality mechanism and direction betweenthese variables. The models reviewed below implicitly or explicitly conjecturethat public capital positively affects the rate of return of private capital, henceprivate capital accumulation. Given the technical substitution between privatecapital and labour inputs, labour productivity rates improve as a function ofthe growth rate of the stock of private capital. These effects, in turn, spurgreater total output, thus, growth (Deno 1988; Aschauer 1989a, b, 1991;Munnell, 1990b). We call this causality linkage ‘public infrastructureaccumulation induced growth’. But what about cases where highly productivecountries, states or regions, with high growth rates, attract private capital andproductive labour which, in turn, demand higher levels of infrastructureinvestment? In such cases the causality direction is reversed as the present stateof high rate of growth stimulates infrastructure investment rather than thereverse. Disregarding such possibilities might result in problems of simultaneityin the empirical analysis which, in turn, will generate incorrect estimates.

Assuming public infrastructure accumulation induced growth typecausality, economic growth further stimulates the demand for public capital

136 Methodology

services (like transportation) which, subsequently, provide the pretext forfurther infrastructure investment. Hence, infrastructure investment decisionsshould be regarded as endogenous to the economy. For this reason the analysisof the impact of public capital on the economy should ideally be carried outwithin a general equilibrium model.

This characterization of the causality linkage further requires a dynamicmodel framework. Two alternative approaches can be found in the literature.One which embeds the aggregate output-infrastructure functionalrelationships within a growth model that specifies a time dependent growthpath for exogenous variables such as technical efficiency, size of the labourforce and public capital depreciation (see, for example, Aschauer 1991;Holtz-Eakin and Schwartz 1995). Alternatively, it is possible to linkrecursively changes in the economy, caused by infrastructure investment,with the level of infrastructure services generated by this investment. Thatis, we require that capital stock investment made in period ‘t’ will affectprivate input factor flow, thus economic growth, in subsequent periods (t+1,t+2, or t+3, etc.,). The modelling objective then is jointly to estimatelongitudinal changes in output, in private capital formation and ininfrastructure accumulation (see for example, Deno and Eberts 1991;Morrison and Schwartz 1996).

For reasons like public funds availability and lengthy time periodsrequired to construct public infrastructure facilities, we can expect thelevel of public capital stock in the economy, in general, to be suboptimal.As is the case with the relaxation of any real constraint on the economy,when the amount of an input factor (such as transportation capital) isbelow its optimal level, its marginal productivity is positive and probablyhigh relative to that of other input factors. Therefore, a well-specifiedempirical model of, say, the effect of transportation infrastructuredevelopment on the economy, is likely to reveal a positive and most likelyhigh rate of economic growth from the investment. The theoretical questionthen is, are the benefits from public capital accumulation primarily theresult of public capital scarcity or does public capital have genuineeconomic growth impacts on the economy?

The problem of optimal public capital stock is further acerbated whenconsidering the fact that the efficient management of capital infrastructurecan have a profound effect on the actual amount of public capital servicesavailable in the economy. For example, the use of improved traffic managementtechnologies such as intelligent transportation systems (ITS) can effectivelyenhance the performance of existing transportation infrastructure facilities.Moreover, the level and quality of services available to users from publicinfrastructure are affected by myriad factors. The type of the infrastructurefacility (e.g. rail vs. highway), its location, particular design, level ofmaintenance and form of governance are some examples. Therefore, thelumping together of all forms of public infrastructure facilities to compute a


single aggregate measure of public capital stock while disregarding theparticular attributes of specific types of capital is likely to generate biasedestimates of the rate of economic growth from this capital.

Incorrect estimates of economic growth from public infrastructuredevelopment can also occur if the measure of economic growth, e.g. the annualrate of change in per capita GNP, does not account for externalities (bothpositive and negative), resulting from the expansion and use of such capital.Air and noise pollution, environmental degradation and traffic accidents areexamples of such externalities, which are unaccounted for in most aggregatemeasures of economic growth.

Another important dimension that needs to be considered is the spatialarrangements of public capital infrastructure. That is, the assumption thatthe benefits from public capital accumulation can be captured in anaggregate production model does not comply with our knowledge of thespatial organization and spatial equilibrium of activities (Haughwout1996). Investment in transportation infrastructure, for example, affectsthe relative attractiveness of regions. This, in turn, affects the economicand locational behaviour of households and firms as well as the use oftechnology. Some models reviewed below do regard the impact of spaceand spatial spillover effects. They show that once spatial impacts areaccounted for, the effect of public capital on economic growth is rathersmall or even insignificant.

The last relevant issue to be mentioned here is the method of fundingused by the government for the provision of public infrastructure. The useof taxation as a means to raise public funds, in general, tends to distortresource allocation as well affect optimal rates of private consumption andinvestment. If a government (national or local) borrows money throughcapital markets it faces the possibility of ‘crowding out’ private investors.In general, the social price of public infrastructure funding, including therelevant deadweight loss, should be considered as it affects the netcontribution of infrastructure expansion to economic growth (Morrisonand Schwartz 1996). Alternative methods of financing public capital arenot inconsequential with respect to social welfare gains and optimal use ofresources.

6.3 Analytical models

In this section we review two principal model type approaches used to assessthe effect of infrastructure development on economic growth. These are:production function models (Section 6.3.1) and cost function models (Section6.3.2). Subsequently, in Section 6.3.3, we examine some analytical andmeasurement difficulties associated with the use of these models. Thefundamental question that these model types attempt to answer is: What isthe elasticity of aggregate output with respect to public capital?

138 Methodology

6.3.1 Production function based models

A prototype production function model, found in many productivity analysisstudies (e.g. Aschauer 1989a; Munnell 1990a; Durkin and Wassmer 1994;Holtz-Eakin and Schwartz 1995) has the following Cobb-Douglas structure:

(6.1)

Where Y is aggregate output (e.g. GDP),9 MFP is a measure of multi-factorproductivity (i.e. the level of technology), and L, KP, KG are, respectively,labour (e.g. aggregate hours worked by the labour force in the private sector),private capital and non-military public capital.10 The subscript ‘t’ denotestime. Some models (Jones 1998, Chapter 6) further augment the labourcomponent, L, by a factor h, which indicates the labour skill level ofindividuals, i.e., (hL)

a

t .Given the estimated parameters, a major objective of the analysis is to

compute the elasticity of output with respect to the public infrastructure stock,eG, i.e.

Typically a log form of equation (6.1) is estimated (Munnell 1993).

ln (6.2)

where g is the output elasticity with respect to public capital (i.e.

). To test for capital productivity Aschauer (1989a) used

the following procedure. He has divided equation (6.2) through by KP,t,introduced a constant term, C0, a trend variable (as a proxy for MFP) and ameasure of capacity utilization by the private sector at time t.11 The objectiveof this last term is to control for the effect of business cycle. Therefore:

ln (6.3)

where the term, lt, is a trend variable, CUt measures capacity utilization andthe parameter g, is the elasticity of output (per unit of private capital) withrespect to public capital (per unit of private capital).12

To test for the possibility of constant returns to scale Egbert and de Haan(1995) introduced an additional restriction:

ln

(6.4)


In estimating (6.4) if the term (a+ß +g–1) is statistically not different than 0,we cannot reject the hypothesis of constant returns to scale.13

One major drawback of this type of analysis is that models like (6.4) donot directly link longitudinal changes in the stock of public capital with changesin private investments. That is, the model implicitly assumes that the effect ofinfrastructure investment on economic growth is conditional upon the levelof private capital in the economy and that both investments take place duringthe same time period. In other words, a model such as (6.4) assumes that theeffects of public and private capital on output occur simultaneously. It furtherassumes that these changes take place irrespective of time-related changes inother intervening variables such as the size of the labour force or technicalefficiency.14

To account for these deficiencies several extensions of the basic modelhave been proposed. One approach is to embed the model within amacroeconomic growth model by explicitly making the relationshipsbetween public infrastructure accumulation and economic growthdependent on the growth path of key exogenous variables (Holtz-Eakinand Schwartz 1995). These authors have specified a model in whichtechnical efficiency (transforming physical units into units of labour) andthe size of the labour force grow over time at constant rates J and h,respectively. In addition, public capital stock accumulates at annual fixedproportion q (which is the propensity to invest by the government) and itdepreciates at a geometric rate d. Hence, the steady state form of theirmodel (in which variables are expressed in per employee units) is givenby:

ln (6.5)

where yi, is the per employee level of output of observation i, which is thegross state product (GSP) and KP,i is the stock of private capital (of that state).The basic concept of growth implies periodical changes in output fromperiodical changes in inputs. Therefore, it may be desirable to define thebasic model (6.2), as a first-order difference equation:15

(6.6)

Durkin and Wassmer (1994) have empirically estimated such a model withthe added lag effect of public capital on output. The general structure of theirlagged model is:

(6.7)

Model (6.7) indicates that public capital investment at time ‘t’ affects thelevel of private capital and labour at subsequent time periods. If the stock ofpublic capital at time ‘t’, KG,t, is assumed to affect output at that time period

140 Methodology

but to affect private capital and labour at time period ‘t+1’, a simultaneousequation model such as (6.8)–(6.9) or (6.8)–(6.10) can be used:

ln (6.8)

ln (6.9)

or:

ln (6.10)

An important question in economic growth theory is what is the level ofinvestment in the economy necessary to achieve a steady-state growth path,defined as that growth rate at which the amount of capital per employeeremains constant.16 Thus, we may ask what should be the level of publicinvestment necessary to achieve a steady-state growth path? Aschauer (1991)proposed such a model whose empirical objective is to assess the effect oftransportation infrastructure investment on growth of different states (in theUSA). He begins by asserting that the rate of growth of the ratio of private

capital to labour, is a positive function of the capacity of the regional

transportation network,17 KG, and of a set of other factors affecting privatecapital accumulation Z. That is:

(6.11)

Given the differences between states relative to their labour productivity rates,he introduces a ‘catch up function’ which defines how the rate of growth oflabour productivity in less productive states will converge to that of the highlyproductive ones.18 Thus, at period t the level of labour productivity in a lowproductivity state is defined as:

ln (6.12)

where f represents the rate of growth of labour productivity in the high-levelproductivity states; is the stock of private capital at time t, qt is the catchup function of the low productivity rate state, and h is the base rate of growth.From (6.12) the annual growth rate of labour productivity is:

(6.13)


Equation (6.13) implies that the annual change in labour productivity (thecatch up function) at time t is negatively related to the level of labourproductivity at time t-1. It is a first order difference equation which can besolved for the average rate of labour productivity growth over the interval[0, T] where 0 and T are the base and final periods, respectively.

Let, , and is labour productivity at the base

year.19 From (6.12),

(6.14)

Jointly, equations (6.12) and (6.14) constitute a non-linear (in theparameters) dynamic model of labour productivity changes frominfrastructure capital accumulation.20 (Empirical results are discussed inSections 6.4 and 6.5 below.)

One major statistical problem with this approach is that taking first orderdifferences can amplify measurement errors. For example, the amount ofpublic capital is largely measured with errors caused by the use of alternativedefinitions of public capital and inaccurate assessment of the actualconditions of the public capital stock. Taking first order differences willexacerbate such errors, as the variance of the errors will double, resultingin biased estimates.21

6.3.2 Cost function based models

Against the analytical and statistical limitations of the production functionapproach, several studies adopted a cost function model approach. This sectionintroduces the general structure of this model and shows its use in theassessment of returns to infrastructure investment. For cost function studiessee, for example, Berndt and Hannson (1992); Lynde and Richmond (1992);Nadiri and Mamuneas (1994); Morrison and Schwartz (1996).

As with the production function model the objective of this model is toinvestigate the effect of public capital formation on national or state’seconomic growth and productivity. However, whereas the production functionmodel produces marginal product measures (e.g. the marginal productivityof an additional unit of public capital investment) the cost function modelproduces shadow value parameters that indicate the cost savings from anadditional public capital investment. Therefore, to explain growth the costfunction model must show how the expansion of public infrastructure enablesprivate firms to reduce their average costs by lessening the use of privateinputs or by increasing their productivity. These cost savings are regarded asreturns to public investment (Morrison and Schwartz 1996).

Most of the cost function studies reviewed here make the implicit

142 Methodology

assumption that the cost function indeed represents the behaviour of privatefirms with respect to their demand and use of inputs, as manifested by theircost minimization behaviour.22 The general structure of the cost function modelis given by (6.15):

(6.15)

where C is the private sectors’ variable cost function, w– is a vector of inputfactor prices and, as before, KP is private capital, KG is public capital, t is atime index (representing technological change) and Y is output. In general,the input price vector includes the prices of labour (production and non-production labour), private capital, public capital, materials and energy. Thequestion of how properly to define and measure the price of public capitalis, however, quite a convoluted one (see shortly). Nevertheless, the inclusionof public capital allows for the evaluation of its impact on the shape of thelong-run private cost function. Thus, the key element in the analysis is thecomputation of the shadow price of public capital or its cost elasticity, i.e.–¶ ln C/¶ ln KG.23

Several studies have attempted to provide a proper measure for the priceof public capital (e.g. Morrison and Schwartz 1996). While it is possible toargue that for private firms the price of public infrastructure such astransportation is zero, it is certainly not so from a social viewpoint. A furtherquestion is how to compute the price of public infrastructure in the presenceof spillover effects across metropolitan areas or even state borders. If sucheffects are, in fact, extensive they represent real alternative costs for firmsthat do not locate where they can most benefit from them.24

The computation of a measure for the price of public capital involves (atleast) three main factors: (a) the rate at which a government raises funds (e.g.through borrowing); (b) the rate of depreciation of public capital; (c) thedeadweight loss (excess burden) of taxation. Morrison and Schwartz (1996),for example, use government’s bond yields as the cost of government’s fundraising. They further use private sector rate of depreciation of capital as anupper bound of the rate of depreciation of public capital (assuming that thelatter has, on the average, a longer life span than private capital has). Theirmeasure of the price of public capital is: (1\l)(rG + dG), where l is investmentprice deflator and rG and dG are bonds’ yields and rate of depreciationrespectively.25 They further augment this measure by adding an index of thedeadweight loss of taxation. For the USA, they take this loss to be between$1.00 and $1.50 for each dollar raised through taxes.26

The cost function model enables the derivation of several measures ofproductivity change and economic growth from public infrastructuredevelopment. The total change in costs with respect to time, hC,t, whichmeasures overall changes in productivity, can be decomposed as follows(Appelbaum and Berechman 1991):


(6.16)

where Si is the share of input i in total costs and hC,Y is the elasticity of costswith respect to output. Note that hC,Y can be less, equal to or greater than 1,depending on whether the economy operates under conditions of scaleeconomies and whether the effect of public capital is to reduce total variablecosts. Morrison and Schwartz (1996) further decompose (6.16) explicitly toallow for the effects of technical change and for the level of private andpublic infrastructure capital on total variable costs.27

Two principal types of cost functions models were used in the literature.These are the translogarithmic (e.g. Nadiri and Mamuneas 1994) andgeneralized Leontief (e.g. Seitz 1993). Both functions belong to a family offunctions known as a ‘flexible form’ type. That is, these functions place veryfew a priori restrictions on the attributes of the underlying productiontechnology such as the degree of factor substitution or its homogeneity, therebypermitting substitution and complementary relationships between inputs.28

Specifying (6.15) as a flexible form function, we can further compute theeffect of public capital on the demand for private capital, i.e. ¶KP /¶KG.

Before we turn to discuss some of the difficulties involved in using theproduction and cost function models, we should point out a third approachused in the literature to estimate the relationships between public capital andeconomic growth—the profit function model approach. This model, whichis rather closely related to the cost function model, assumes that the decisionunit (e.g. the state) can choose the optimal level of output e.g., gross stateproduct, by selecting the optimal level of input factors, given their marketprices. Deno (1988) used this approach to estimate the elasticity of outputwith respect to transportation capital (see Section 6.4, Table 6.1).

6.3.3 Analytical and measurement problems

The production and cost function based models presented above constitutethe major analytical mechanism used to study the macro-level linkage betweenpublic infrastructure development and economic growth. Aside from somestatistical problems (see shortly), they all share some common deficienciesrelative to the theoretical requirements outlined above (Section 6.2). Therefore,before we present the myriad empirical results from the estimation of thesemodels, it is useful briefly to examine some of these issues.

One fundamental problem with these models is that they implicitly assumean efficient level of public capital stock in terms of the level of service thatthis stock provides. In other words, it is assumed that available public capitalis rather efficient so that better management or improved maintenance couldnot increase the level of services and, therefore, only additional investmentwill spawn farther growth. Even the dynamic models, which are formulated

144 Methodology

in a way that ensures convergence into a steady-state growth path, implicitlyassume the presence of efficient capital stock in the economy. Furthermore,the models examined above presuppose efficient and competitive marketsfor private inputs which, under variable cost minimization behaviour, willresult in their efficient (and optimal) use. How valid are these assumptionsfor the analysis of public capital impact on growth? It is rather doubtfulwhether public agencies, which operate in political environment and faceuncertain fiscal constraints, are capable of supplying and maintaining anefficient level of public infrastructure capital. As a consequence, the estimationof positive growth rates from additional investments, in fact, may constitutea faulty policy guide as better management of available infrastructure facilitiesmay yield the same economic growth effects.

A second major problem with the macroeconomic models presented aboveis the actual direction of the causality. That is, do increases in investment inpublic capital stock lead to greater output, as is assumed by these models or,alternatively, does increased aggregate output propel further public investment(Eisner 1991). The empirical evidence on this question is somewhat mixed.Some researchers argue that the causation is, in fact, reversed: risingproductivity in the private sector effectuates infrastructure capital outlays(Tatom 1993; Gramlich 1994); whereas others (Aschauer 1991) argue theopposite (see further in Section 6.4). In this regard it also questionable whetherproduction and cost function models constitute a suitable analytical basis forestablishing causal relationships.

The use of cost functions in general has been justified on the groundsthat production functions omit input prices and a priori place restrictionson the firms’ technology and behaviour (Friedlander 1990). As shown above(6.16), cost functions can also be used to disentangle the effects of technologychanges, scale economies, effect of input shares and the links between costsand outputs. Yet they leave open the question of what is the actual price ofpublic infrastructure that private businesses face. Public infrastructure, suchas the roadway system, is regarded by the private sector as a public good,which is provided to users at zero marginal costs.29 Hence, should the privatesector’s cost function reflect the costs associated with public infrastructureprovision? To the economy as a whole these costs are real and should becomputed and accounted for (as was done by Morrison and Schwartz 1996).The theoretical question still remains, however. To what extent does thecost function model reflect the costs of public capital development eventhough private sector decision-makers do not normally regard these costsas costs to their firms? In general, the contribution of public capital toproductivity cannot be equated with that of private capital using, as a basis,market prices that are generated from market transactions. Hence, variationsin the estimated elasticities of costs (and output) to public capital expansion,in part, are due to alternative approaches to the pricing of this capital(Jorgenson 1991).


From the empirical results presented below it can be concluded that, inmany cases, the correlation between trends in output and public infrastructureinvestment is spurious. This can emanate from a number of factors, a keyone being the probable lagged relationships between the public investmentand the consequent economic growth. Investment in a particular year mightlead to an increase in output three or even five years later. Statistically theproblem arises from the existence of a common trend in the time series datawhich then leads to false correlation estimates (Jorgenson 1991). The first-order difference equation approach, described above, presumably controlsfor non-stationary variables. In general, however, if the underlying commontrends are not accounted for properly, what the analysis merely demonstratesis that the patterns of productivity and public investment growth are similar(Schultze 1990).

In summary, when using econometric models to estimate themacroeconomic effects of public infrastructure development on cycles ofeconomic growth and productivity changes, a number of factors may bringabout inaccurate estimates. Key ones are the mis-specification of modelsrelative to causality and time lags, unaccounted for scale effects, incorrectmeasurement and weighting of inputs and outputs, unaccounted for truetechnological changes, and the presence of externalities including spatialspillovers. In the next two sections we survey empirical results obtained fromvarious studies on the effects of aggregate public infrastructure capital,including transportation, on productivity and economic growth.

6.4 Effects of transportation infrastructuredevelopment: empirical results

Most empirical studies have estimated the effects of total public capital oneconomic growth and productivity. Only some of these have included as aseparate category transportation capital. Very few studies, however, havedirectly examined the impacts of transportation infrastructure developmenton economic growth. Given the aim of this book we begin by examining insome detail two such studies. Subsequently we examine studies which havedecomposed total public capital into transportation infrastructure capitaland others.

Aschauer (1991) has analysed the relationships between transportationinfrastructure spending and economic growth and labour productivity usinga production function based growth model such as in equation (6.11) to(6.14)30 Annual change in output per employee was used as the dependentvariable (the left-hand side of equation 6.14). His database was composed ofobservations on highway and transit spending in 48 contiguous USA states,for the period 1969–86. The principal finding from this study is that theeffect of total transportation expenditures on the growth rate of the ratio ofprivate capital to labour is rather high (f=0.166, in equation 6.13), and is

146 Methodology

significant (R2»0.44). He interpreted this result to contemplate that a $10bincrease in transportation outlay at year ‘t’, spread over the 48 states, wouldresult in $2.05 billion increase in private capital in the same year, a quitehefty effect. Based on the estimated elasticity of output with respect to privatecapital (a=0.486, in equation 6.14), he calculated that this increase in privatecapital will further stimulate total output by $980mn during that year, andoutput per employee by $8. Furthermore, since his estimated parameters showa slow rate of convergence towards a steady state, the ultimate increase inoutput is estimated to be 28.1 times over the initial increase in outputamounting to $27b ($980mn times 28.1). By further decomposing totaltransportation spending between transit and highway expenditures, Aschauerfound that the impact of transit investment alone on the rate of growth ofcapital to labour ratio is 2.3 times that of total transportation spending(f=0.384). With highway expenditures only, the estimated parameter wassignificantly less than that of transit (f=0.231).31 It should be rememberedthough that expenditures on the physical infrastructure stock of highwaysand roads constitute the largest share of total transportation capitalinvestment, and that a good part of it represents maintenance, repair andreconstruction costs. Hence, despite the larger effect of transit spending onproductivity, from a policy perspective, the opportunities to significantlyincrease transit investment at the expense of highway are rather limited.

Lastly, Aschauer addressed the issue of the direction of causality, mentionedabove. Using state budget as an instrumental variable in this simultaneousequations growth model, he concluded that transport expenditures induceproductivity growth rather than the reverse.

Seitz (1993) investigated the effect on productivity changes from theexpansion of the highway network in Germany (formerly FRG), using ageneral Leontief cost function model. The database was composed of paneldata of 31 industries in the manufacturing sector and of the annual capitalstock of the FRG’s road network, for the years 1970–89. The database alsodistinguished between capital expenditures on motorways only and on thetotal road network. Industry output is measured in terms of net value added.Labour inputs are in units of total number of hours worked and the cost oflabour was obtained by dividing total wages paid by the number of hoursworked. Private capital is measured in terms of net capital stock and theprice of capital is an adjusted Jorgenson measure of users cost of capital.32

The stock of transportation infrastructure is measured in two alternativeways: as the length of the motorway network (in km) and as the monetaryvalue of the real net capital stock of roads and bridges (in 1980 prices inbillions of DM). The estimated system of equations contains the cost functionand the labour and capital share equations, with industry-fixed effect usingindustry-specific dummy variable.33

Under the two measures of output the estimated parameters were quitesimilar so that the same conclusions were drawn. First, these parameters


have indicated complementary relationships between private capital and publictransport infrastructure (¶KP/¶KG >0), whereas labour and public capital aresubstitute inputs (¶L/¶KG <0). In other words, additional investment intransportation infrastructure is likely to raise marginal productivity of privatecapital. This result was also obtained from studies of the effect of aggregatepublic capital on private capital (e.g. Tatom 1991) discussed below. Thesubstitution relationships between public capital and labour imply a reductionin the use of labour following additional infrastructure investment.

What is the marginal benefit from the expansion of the transportationcapital stock? To answer this question, Seitz computed the shadow price oftransportation infrastructure investment, Ei,G, (Ei,G=¶Ci/¶KG). It is definedas the change in average cost for each of industry i in the sample caused by a1 km increase in the size of the motorway or by DM1bn expansion of the netcapital stock of roads and bridges. For all industries the average decline inaverage costs, when the motorway is expanded by 1 km, is about DM32 (forevery DM1mn increase in output). For the net capital stock output measureaverage costs declined by about DM500 (which is about 5 per cent of averageprivate costs). However, these average cost declines vary considerably byindustry type. Estimates for some industries were statistically insignificant(19 out of 31 in the case, where net capital stock was used as the outputmeasure).34

The results from these two studies are not directly comparable due todifferences in the model approaches and databases used. However, bothconclude that investment in transportation infrastructure stock enhancesprivate capital profitability, thereby stimulating private capital investmentand hence economic growth. Moreover, on the average investment intransportation infrastructure stock produces a high rate of return for theeconomy with considerable marginal net benefits from an additionalinvestment. On the other hand, both studies report substitution relationshipsbetween labour and transportation infrastructure investment implying thatfurther investment may reduce employment. This conclusion should be takencautiously, however, since the estimated results were derived from partialequilibrium models. That is, if additional infrastructure investment leads toprivate capital growth (due to the complementary relationships) and the lattereffect induces output expansion, more labour may ultimately be needed,despite the substitution between public capital and labour, depending on therelative magnitude of these effects.

What can be learnt about these issues from studies that regardedtransportation capital as part of total public capital? To answer this questionwe next turn to studies which have used total public capital stock categorizedinto transportation capital and other types of public capital. They thendiscerned the effect of transportation infrastructure on employment andoutput.

Garcia-Milà and McGuire (1992) have estimated a production function

148 Methodology

model with public capital input composed of highway capital and humancapital (measured as education expenditures). A third input is private capitaland output is measured as gross state product (GSP). The estimated elasticityof GSP with respect to highway capital is 0.04, whereas that for humancapital is 0.15 (for private structures and private capital equipment theelasticities were 0.10 and 0.37, respectively).

A similar result regarding GSP elasticity with respect to transportationinfrastructure (highway stock) was obtained by Munnell (1990b). Using datafrom 48 USA states for the years 1970–86, she found GSP elasticity withrespect to highways to be 0.06, whereas for water and sewer infrastructureto be 0.12. For the category labelled ‘other public stock’ the elasticity was0.01. (For labour and for private capital the corresponding output elasticitieswere 0.55 and 0.31 respectively).

Eisner (1991) further examined Munnell’s results and the data from whichthey were derived. He found that when the database is arranged as a cross-section, public capital indeed has a small but significant effect on states’growth. However, when the data are arranged to reflect longitudinalvariations, the effect of public capital is insignificant. He also found thatstates with high public capital per capita also have high output per capita,and that investment in public capital in a given year does not increase outputthat year. What seems to be in doubt is the causality relationships betweenpublic infrastructure and growth, or the specification of the empirical model(e.g. the use of lagged variables), or both.

McGuire (1992), who studied the sensitivity and robustness of the resultsobtained by Munnell (1990b) and Garcia-Milà and McGuire (1992), obtainedsomewhat different results, mainly in order of magnitude.35 To that end, heused the same data sets as these researchers but categorized public capitalinto highways, water and sewers and others. Using a Cobb-Douglasproduction function, he estimated output elasticity with respect to highwaycapital to be in the range of 0.121 to 0.370. When controlling for state effects36 he estimated the elasticity of output (measured as GSP) with respect tohighway infrastructure capital to be within the range of 0.121 to 0.127 (andthe elasticity of output with respect to water and sewer capital to be in therange of 0.043 to 0.064).

A major problem with production function models that use state-baseddata (GSP) is that they fail to consider inter-state factor mobility, mainlylabour and capital. That is, the prime effect of transportation infrastructureinvestment is to change the relative accessibility and attractiveness of specificregions. This effect, in turn, results in the relocation of firms and labour (andhouseholds) across jurisdictions. Failure to account for these spatial changesis bound to produce biased elasticity estimates (see also below the study byMoreno et al. 1997).

Some empirical studies found that when using a lagged model (like equation6.7) the effect of public infrastructure on growth was statistically significant


(Bajo-Rubio and Sosvilla-Rivero 1993; Durkin and Wassmer 1994). Kelejianand Robinson (1997) have estimated a production function model in whichprivate output at time ‘t’ is a function of labour at ‘t’ and private and publiccapital stock at ‘t-1’, where public capital is composed of highway capitalstock, water and sewer, and others. To account for spillover effects they havealso included in their model output to labour ratio in neighbouring states.Based on their estimates they conclude that previous inferences of positivemarginal productivity of labour with respect to infrastructure are notsupported.

Keeler and Ying (1988) have used a cost function approach to assess theimpact of highway investment on the cost and productivity of private trucking.Their study examined the use of Class 1 regional trucking firms since 1950 inthe USA. Their results indicate that the expansion of highways between 1950and 1973 had a significant impact on the productivity growth of the privatetrucking sector.

Deno (1988) has used a profit function model in which profits of the privatesector are specified as a function of the prices and quantities of private capitalof labour and of the stock of public capital. The latter is disaggregated intohighway capital, and sewer and water capital. Using a state level database,he estimated output elasticity with respect to highway capital to be 0.31(with respect to sewer capital to be 0.30 and with respect to water capital tobe 0.07).

Haughwout (1996) has studied the effect of the highway stock on state’seconomy within the framework of a spatial equilibrium model. To that endhe estimated a two stage least square model in which the first equation explainsoutput (in GSP units) as a function of the state’s land area, private capital,labour and population density. In the second equation population density isdefined as a function of the public highway stock, of other public capital andstate’s debt. Using data on 48 contiguous states for the years 1977–92, heconcluded that highway investment will induce higher population densitywhich, in turn, will lead to higher output but the latter effect is rather small.Table 6.1 summarizes the main results from these studies.

As Table 6.1 shows, the range of output elasticity with respect totransportation capital is quite large. Apparently, the specific model used, theestimation method as well as the nature of the database all affect the estimatedresults. While there are still major doubts regarding the actual magnitude ofthe elasticity of output with respect to changes in transportation capital, theabove results suggest the following. First, transportation infrastructure capitalseems to have a small but significant impact on economic growth in terms ofoutput elasticity. However, other results (not shown here) also indicate thatother public inputs, such as education, have a substantially higher impact ongrowth than transportation capital.

The second major implication is that the nature of the causality betweentransportation infrastructure development with economic growth is rather

150 Methodology

equivocal with respect to direction, functional relationships and effect ofintervening variables. While the above results seem to suggest thattransportation capital expansion (as well as the expansion of other publiccapital inputs) propels economic growth and enhances productivity, it is farfrom clear what conditions must be met for this result to transpire. Forexample, high rate of population growth coupled with increasing populationdensities that, in turn, create high demand for transportation services, mightbe a prerequisite for transportation infrastructure investment to have asignificant effect on growth. Furthermore, economic growth tends to lagbehind transportation investment, as the capitalization of the investment’seffects is time dependent. It may also depend on the behaviour of the labourand private capital markets. That is, for transportation capital expansion tohave a consequential impact on economic growth it should affect labour and

Table 6.1 Selected results from studies of the impact of transportation infrastructure investmenton economic growth


private capital inputs, relative to their price, their actual use and productivity.Similarly, the impact of transportation investment on technological changeshould also be expected as, for example, is the case for the development anduse of telecommunications technology.

6.5 Effects of aggregate public infrastructuredevelopment

We have noted above that the majority of studies have estimated the effectof changes in total public capital stock on state or national economic growthand productivity changes. Since the main focus of this chapter is the effectof expanding the transportation infrastructure stock on growth andproductivity, in this section we briefly summarize major results from studiesthat have used aggregate public capital and subsequently use these resultsfor comparative purposes. We categorize these results into productivity gains(Section 6.5.1) and economic growth results (Section 6.5.2).

6.5.1 Productivity gains

Munnell (1990a, 1993) has estimated a log-linear production function (suchas equation 6.2), using USA data, which covered non-military federal, stateand local public capital for selected years (1948–87). She reported the elasticityof labour productivity with respect to public capital to range from 0.31 to0.39 (i.e. a 10 per cent increase in public capital would raise labour productivityby 3.1 to 3.9 per cent). From additional calculations of multi factorproductivity (MFP), Munnell concluded that much of the increase in MFPduring the early part of the sampled period was, in fact, due to the build-upof public capital vis-à-vis its effect on output.

Holtz-Eakin and Schwartz (1994) have computed the effect on productivitygains from raising the propensity to invest in public capital in each state by10 per cent of new spending in 1986 (about $10bn in 1982 prices). Using themaximum level of the underlying structural parameters (see equation 6.5),the results show only modest effects. Across the sampled 48 USA states theaverage increase in productivity was about 1.02 per cent for the entire sampledperiod (1970–86), implying an insignificant annual increase in the productivityrate of each state’s economy.

Morrison and Schwartz (1996) have used a cost function model to estimatethe direct effect of public capital on productivity changes. They used thestate’s GSP as their database. The range of impacts was 0.192 per cent fornorthern states and 0.622 per cent for southern states. When including negativeproductivity growth effects, the total impact of infrastructure investment onproductivity growth was (mean values): 0.245 per cent (east), 0.153 per cent(north), 0.622 per cent (south) and -0.971 per cent (west). They concludedthat the direct cost-savings impact of public capital investment is generally

152 Methodology

significant but it is rather small. The introduction of the social price of publicinfrastructure further lowered these estimates.

Nadiri and Mamuneas (1994) have used a translogarithmic cost functionmodel with three private inputs: labour, intermediate and private capital andtwo public inputs, infrastructure and R&D capital. They have applied thismodel to USA data describing manufacturing industries at the two digitsStandard Industrial Classification (SIC) level. Major results from this studyindicate that the elasticity of total costs with respect to public capital expansionranges from -0.11 to -0.22, which is lower than previous studies havedetermined. With respect to the demand for private inputs, they found thatan increase in infrastructure capital leads to a decline in the demand forlabour and private capital in each industry and to an increase in the demandfor intermediate inputs.

These results by Nadiri and Mamuneas (1991, 1994) essentially indicate anon-constant degree of substitution between public infrastructure and privateinputs. Further analysis shows that the level and changes in labour productivityduring the sampled period were mildly affected by the public sector’s provisionof capital. That is, changes in the demand for labour are the result of twomajor effects: a downward shift in the cost function of the analysed industriesinduced by investment in public capital stock and the substitution of privateinputs (i.e. labour) with public ones.

Overall, the results from these studies indicate that additional public capitalinvestment has a positive impact on productivity, mainly labour, but that thiseffect is rather small and is affected by a number of intervening factors. Inparticular, this is the case when we consider the ‘output’ and ‘substitution’ effectsof public capital increase. That is, if public capital is indeed a substitute inputfactor for labour, as the results from several studies show, additional infrastructureinvestment may have a counter effect on job creation and use. Since the outputeffect tends to foster greater aggregate demand for labour, while the substitutioneffect operates in the opposite direction, the overall impact on labour of furtherinvestment in public infrastructure is a priori not certain. The empirical resultspresented above seem to suggest that, on average, the output effect surpasses thesubstitution effect. What the above results seem to propose is that, followingpublic capital development, private output grew proportionally faster than labour,hence the increase in labour productivity. It should be emphasized, however, thatthese results are from a macro-level analysis and may not hold for an individualinvestment or for a particular region or industry. For this reason it is ratherimportant to assess the contribution to employment and labour productivity ofspecific infrastructure projects in specific geographical areas.

6.5.2 Economic growth

Aschauer (1989a, b) has attempted to estimate what would have been the impacton growth of an additional investment of $60bn in public capital in 1985 (a 6


per cent increase in the infrastructure stock). Assuming elasticity estimate of0.24 for the change in productivity from public capital investment, he concludedthat aggregate private output would have increased by 1.4 per cent, a gain of$70bn in the first year. With conservative assumptions, he further estimated totalgains to be $600bn, which translate to benefit cost ratio of 10:1.

Munnell (1990a) has re-estimated the original aggregate time-seriesparameters of Aschauer and came to similar conclusions. Yet, Munnell concludedthat these impacts were much too large to be fully credible. Her studies (1990a,1993) found that a 1 per cent increase in the stock of public capital would raiseprivate output by 0.34 per cent, still a quite large estimate. In another study shefound that while public capital has a positive impact on several measures ofstate-level economic activity (such as GSP, investment and employment growth),the order of magnitude was much less than that reported by Aschauer (seeMunnell 1990b). For example, the elasticity of output with respect to publiccapital was estimated to be 0.15. This lower figure is consistent with thosefound by other researchers in the USA.

Table 6.2 summarizes principal results from studies that estimated theeffect of total public capital stock on growth. The figures shown in this tableagain highlight how diversified are the empirical results, depending on modelspecification, the database and measures of output.

Table 6.2 Some estimates of the effect of total public capital expansiona (US data)

Source: Based, in part, US DOT (1992).Notes:a All the national and state level analyses used Cobb-Douglas production function models

except where noted;b GSP=gross state product;c This is the direct impact on productivity growth rate. Mean total impact for the western

region of the US is: -0.971.

154 Methodology

6.5.3 Results from non-USA studies

The above results pertain mainly to USA data and experience. It is, therefore,illuminating to examine the degree to which results from other countriessupport the USA findings. Berechman (1995) has applied a simultaneousequation model (such as in 6.8 and 6.9) to Israeli macroeconomic data forthe years 1964–89. His results show that a 10 per cent increase in publiccapital in year t will increase output at this year by 3.5 per cent and willincrease private capital by 5 per cent at t+1. Estimation of the second set ofequations (6.8 and 6.10) shows a much smaller effect of public capital onoutput. Specifically, a 10 per cent increase in public capital at year t willincrease labour at year t+1 by 0.33 per cent and total output by 0.73 percent.37 In general, these results imply that investment in public capital affectstotal output both directly (by increasing total factor productivity) andindirectly vis-à-vis its effect on partial productivity of labour and privatecapital. Ford and Poret (1991) have used a database composed of observationsfrom 11 OECD countries. Their results vary greatly by country and resultsthat are compatible with USA findings were obtained only for Germany,Canada, Belgium and Sweden. Easterly and Rebelo (1993), in their study of100 developing countries (1970–88), found that transportation andcommunication investments are positively correlated with growth (in the range0.59 to 0.66) and uncorrelated with levels of private investment. Theirconclusion was that there were supernormal returns on public investment ininfrastructure and that ‘it raises growth by increasing the social return toprivate investment but not by raising private investment itself.

Berndt and Hansson (1992) used a cost function approach to test the linksbetween public infrastructure capital and private sector’s costs in Sweden(1960–88). Their main conclusion was that public capital expansion improvesprivate sector productivity by reducing its costs. However, they have alsoconcluded that total public capital stock exceeds that which can be justifiedon the basis of reducing private sector costs alone, and that this excess isbeing lessened over time as less public infrastructure investment is now takingplace. Aschauer’s (1989c) study of the Group of 7 Countries (G7) suggesteda positive statistical link between the ratio of public capital investment toGDP and the growth in output per worker.

Moreno et al. (1997) have studied the effect of public capital developmentwith transportation infrastructure included on the growth of Spanish regions(1964–91). In their study they used a production function model with publiccapital which was divided between the monetary value of stock of roads,highways, railways, harbours, airports, water and sewage facilities and urbanstructures, and between what they labelled the ‘social stock’ of health andeducation. They found a slight impact of public capital on industrialproductivity in Spanish regions. However, when a spatial dimension wasintroduced into their model to account for inter-regional spillover effects they


found that public capital development has no direct effect on regional growth.Table 6.3 compares elasticity estimates from various countries.

International comparison difficulties notwithstanding, the elasticityestimates reported in Table 6.3 suggest that, in general, output elasticities fornon-USA countries and regions are far smaller than those within the USA.The explanation for this phenomenon is not quite clear, particularly fordeveloping countries. In those countries the stock of public capital per capitais much lower than in the USA and, therefore, one could expect that theimpact of public sector infrastructure investment on output would be larger.

6.6 Conclusions

No doubt the question as to what degree does public capital, in particulartransportation infrastructure, affect economic growth and factor productivityis rather complex and subject to a myriad of intervening factors. This chapterhas concentrated on the recent flood of econometric studies carried out in theUSA and elsewhere attempting empirically to assess these relationships. Ifone needs to draw a single conclusion from this literature it is that publiccapital development has some positive impacts on economic growth andprivate factors productivity, but the magnitude and significance of theestimated effects are far from being conclusive. Furthermore, when consideringthe models and the databases used, it is quite difficult unequivocally todetermine how robust these estimates are. As generally is the case, complexeconomic phenomena lie somewhere between orderly patterns of economic

Table 6.3 Estimates of public capital growth effects worldwide

Source: Based on World Bank (1994: Box 1.1).

156 Methodology

development and random processes, between predictable correlation of trendsand between unrelated outcomes.

The widespread use of production function models, as in the originalseminal research by Aschauer, has been criticized analytically andeconometrically, as well as in terms of the scale of the expected impacts. Onemajor problem cited with Aschauer’s modelling approach is that when usinglongitudinal data on output, private and public input factors and on proxyvariables representing technological changes, there is likely to be an underlyingcommon non-stationary trend. Unless properly treated, the statistical estimatesof the elasticities of aggregate output with respect to public capital areimplausible and probably wrong. In fact, it has been shown that the intervalestimated for the elasticity of output with respect to public capital containsboth negative and positive values so that further investment in publicinfrastructure stock may adversely affect economic growth.

A related attack against Aschauer’s type of analysis is that it merelydemonstrates that the patterns of productivity and public investment growthare similar and that this is what the correlation shows. In other words, thismodel does not demonstrate causality, rather, it presupposes it. There wererising patterns of productivity and public investment in the USA in the 1950sand 1960s and falling patterns in the 1970s and 1980s. Consequently, thecorrelation is likely to generate grossly inflated estimates of the returns frompublic infrastructure investment (Schultze 1990). In the words of Krugman,Aschauer’s findings are ‘more a matter of correlation than causation’(Krugman 1994: Chapter 4).

The treatment of transportation capital in the models reviewed in thischapter is a good example of the causality problem. By and large,transportation infrastructure is an unpriced input factor into privateproduction processes. As a result, do private enterprises at all regard theavailability of this factor in their decisions concerning optimal level of outputand use of private input factors? If they do not, generic production functionmodels, such as equation 6.1, may be inappropriate.38 If they do, to whatdegree? Only few macroeconomic studies have attempted to deal with thisquestion by assessing the shadow price of transportation capital and includingit in their model (e.g. Morrison and Schwartz 1996). But even these attemptslargely disregard the true social costs of using transportation infrastructure.The latter costs are the product of congestion costs, the costs of infrastructuredeterioration (e.g. highway pavement, bridges or tunnels, structural decay,caused by different vehicles, mainly trucks, Small et al. 1989) and the costsfrom other non-market externalities such as air and noise pollution.

Given that the empirical models examined above did not account for theseissues, it is plausible to contend that their elasticity estimates are rather biased.It is hard to discern the direction of the estimation bias, but it does cast adoubt on the use of these estimates for policy design. Furthermore, the modelsreviewed in this chapter failed to account for the changing economic


environment examined in Part II. In particular, the kinds of measures ofefficiency and productivity they used may be inadequate when consideringthe increasing size and role of the service sector in the economy.

The above criticism notwithstanding, it has been suggested that the maincontribution of Aschauer’s type of studies has been to draw attention to theimportance of public infrastructure in promoting economic growth and privatecapital productivity (see Aschauer 1993a; Munnell 1993). Moreover, theanalysis also indicates that with respect to the growth effect of output frompublic capital expansion, what matters is not the size of the investment inpublic capital stock but rather the annual per cent increase of this stock. Thatis, a large size investment in public infrastructure is bound to have aninsignificant impact on economic growth if it constitutes a negligible additionto the in-place public infrastructure stock. For example, a massive investmentin a new transportation link may yield insignificant growth effects if this linkconstitutes but a small portion of a well-developed network

On a more general level, the main motivation for undertaking the modellingefforts explored above is to show the need for further investment in publicinfrastructure. This, however, may be superfluous since it is plausible to regardthe present level of public infrastructure in the economy, includingtransportation, as suboptimal. Following Jorgensen (1991) and Gramlich(1994), the main reasons cited are: 1 Engineering Needs Assessment (ENA) determines the mismatch between

what is available in terms of the infrastructure stock (including bothquantity and quality) and what should be available, given somenormatively set standards. However, ENA does not make a strong casefor an overall shortfall in local and national infrastructure capital, exceptwhere there is substantial congestion. No attempt is made to link theENA with economic growth.

2 Contemporary political processes and institutions make new investmentsin public capital rather difficult to implement. Peterson (1991) suggeststhat infrastructure may be under-supplied relative to people’s preferences,but here again nothing was made of the potential economic benefits.

3 Studies of public investments’ rate of return involve a benefit-cost analysisto establish whether the rate of return on a particular project exceedsmarket rates or treasury set thresholds. Rarely these methods involveestimates of non-market benefits, in particular, the project’s contributionto national or regional growth.

Two contrasting general conclusions can be drawn from the discussion inthis chapter. First, commonly used methods at the macro level to assess theeffect of public capital expansion on economic growth suffer from significantanalytical and measurement shortcomings. This would suggest that a microlevel analysis is more appropriate for the task at hand as this is also the scale

158 Methodology

at which most forms of benefit cost analysis are carried out. Second, despitethe criticisms developed in this chapter, there does seem to be a positivestatistical link between public capital investment and output growth. If non-market benefits are also added, the case for public investment may be quitestrong.

Notes

1 We regard the infrastructure of the economy as the physical stock of publicfacilities, such as roads and highways and water and sewer systems, and not asthe rules, regulations and institutions that run the economy (see Jones 1998:Chapter 7, who regard the latter as national infrastructure).

2 According to (US, DOT 1992) total public infrastructure stock includes: highways,streets, educational and hospital buildings, sewer and water facilities, conservationand development facilities, gas, electric and transit facilities and othermiscellaneous but non-military structures and equipment.

3 In part, this brief discussion of government’s provision of transport infrastructurefacilities is based on Berechman (1994).

4 There is also the argument of equity, which essentially implies that spatial mobility,provided by transport infrastructure facilities, is a merit good that should beavailable at a reasonable level and price to all citizens, irrespective of their abilityto pay for it. In particular, people residing in rural areas, where population issparse, cannot pay for infrastructure facilities needed to make them accessible toactivity centres. Hence the need for government’s involvement in the provisionof transport infrastructure. See Chapter 3 for a further discussion.

5 Fujita (1988) has pointed out that if individuals’ utility function is Cobb-Douglas intraded goods, housing and local consumption goods, and assuming constant elasticityof substitution in the utility function between the consumption goods, then utilitywill increase with the number and type of the locally produced consumption goods.

6 If the aggregate production function is a Cobb-Douglas in private inputs (e.g.labour, capital and land) and has a constant elasticity of substitution in somespecialized input factors (e.g. shared public capital), aggregate output will increasewith the amount and with the number of types of input factors.

7 In recent years there is a growing literature on general purpose technologies (GPT)which are technologies that spawn economic growth by enabling the developmentof new opportunities relative to improved productivity and efficiency. Theintroduction of the electric motor, which took advantage of electric power, is anoticeable example of GPT (see, for example, Bresnahan and Trajtenberg 1995).Public capital, like transportation facilities, falls within the category of infrastructuregoods, which enter as an input factor most consumption and production activities,and as such cannot be regarded as GPT. The main reasons being that the growtheffect from such facilities is less continuous and less dynamic and does not necessarilygenerate compatible innovations as GPT does.

8 We do not claim here that infrastructure should be priced to users at below itsreal social costs. Our contention is that since infrastructure is regarded as apublic good, its supply by the government will ensure its ubiquitous availabilityat a price which is below the price that the free market would have set.

9 Some authors (e.g., Durkin and Wassmer, 1994), define Y to be private outputonly. Others (e.g., Aschauer 1989a) use total output which includes private andpublic outputs.

10 In this model output is assumed to be affected by the amount of public capital.


But public infrastructure facilities such as city streets and roads are in fact publicgoods (at least to the extent that their usage does not exceed their capacity),available at zero costs to users. It is therefore legitimate to ask whether suchformulation is appropriate.

11 Holtz-Eakin and Schwartz (1995) have divided equation (6.2) by labour quantityto yield the production function per effective labour unit.

12 Aschauer (1989a) estimated g to equal 0.39, i.e. a 10 percent increase in the ratioof public capital to private capital would raise productivity (output per unit ofprivate capital) by 3.9 percent—quite a substantial effect.

13 Several authors (e.g. Young 1992; Holtz-Eakin 1994) have concluded constantreturns to scale, whereas others (e.g. Moreno et al. 1997) have estimated slighteconomies of scale for total public capital.

14 The model also assumes constant elasticity of substitution between public andprivate capital and between public capital and labour.

15 As explained shortly, the use of a first order difference equation model is also forthe purpose of controlling for non-stationary variables.

16 The most well known early steady-state growth model is that by Solow (1956).17 Aschauer measures the capacity of the highway network in units of “miles paved

per square mile of land area”—a somewhat unconventional measure of networkcapacity.

18 The concept of a “catch up function” implies that labour productivity rates indifferent countries or regions tend to converge, over time, to the same rate. Thereason being that with free trade, factor mobility and technology transfer, theproductivity rates of regions with initial low levels of productivity tend to catchup with those of the high level productivity states.

19 The value of ß must be between 0 and 2, for a stable steady-state solution.20 In the empirical estimation Aschauer uses Zellner’s SURE procedure to estimate

the model’s parameters.21 Tatom (1991) and McGuire (1992) have used the annual growth rate instead of

the actual value of a particular variable, in order to avoid error problemsassociated with the use of a first-order difference equation model.

22 One important analytical implication of this assumption is that the cost functionapproach allows the derivation of input demand functions with endogenousdependent variables.

23 It shows the percent change in costs from a 1 percent change in public capital.Under proper conditions it can be shown to be the dual of the output elasticitymeasure from the production function model, ¶ ln Y/¶ ln KG.

24 Several studies (e.g. Holtz-Eakin and Schwartz 1994) of spillover effects of stateinfrastructure, mainly highways, did not find such effects to be either sizeable orstatistically significant.

25 See Munnell (1990a) for a definition of price deflator of public capital.26 This is the marginal cost of tax dollars. For example, assuming the cost to be

$1.50 for every $1 of public spending, it includes the cost of $1.0 of tax revenueand additional $0.50 of efficiency loss to the economy.

27 They point out that if hC,Y=1, equation (6.16) is equivalent to the commonproductivity index hY,t, where

.

28 For econometric analysis and exposition see, for example, McFadden (1978),Johnston (1984) and Berndt (1991).

160 Methodology

29 We disregard highway congestion pricing which, at present, is quite uncommon.30 Equations (6.12) and (6.14) were estimated jointly using the SURE technique.31 One would expect that the parameter q for the total transportation outlay would

fall between that of transit and that of highway. While Aschauer does not explainthis apparent inconsistency, one plausible explanation is that the two longitudinalsets of expenditure data are negatively correlated.

32 The formula is:

where rt is the cost of capital for industry i, b is the interest rate on a 10-yeargovernment bond, li, is the rate of depreciation and PIi, is price index of investmentcapital (Jorgenson 1963).

33 The estimation method used imposes the constraint that the parameters of thethree equations are identical across all industries in the sample, except for theindustry’s specific constant (see Hsiao 1986). Consequently, industry specificfixed effects can be tested.

34 Berechman and Paaswell (1996) have studied the effect of transportationinfrastructure investment on the labour force participation behaviour in low-income regions. They found that accessibility improvements from transportationinvestments affected the rate of labour participation in some employment sectorsbut not in others. See further in Chapter 8.

35 Based on 1992 review by the Federal Highway Administration (US, DOT, FHA).36 McGuire controlled for state fixed effects by including state dummy variables

and for state random effects by letting one component of the error term to varysystematically by state.

37 All estimated parameters were significantly different than zero at the 5% level.The Durbin-Watson statistic showed a positive serial correlation.

38 In Chapter 8 we deal with the impact of transportation on firms’ locationdecisions.

Economic evaluation oftransportation projects

7.1 Introduction

In Chapter 6 we examined models aimed at capturing the impacts of changesin the aggregate capital stock on national and regional growth. We haveshown that at the present level of knowledge it is impossible to determineunequivocally that expansion of the existing public capital stock (includingtransportation infrastructure) will encourage economic growth. Even if wecould conclude with a reasonable degree of certainty, that further investmentin the total capital stock promotes growth, it is necessary to recognize thatsuch investments are carried out incrementally, as individual projects areplanned, approved and executed one at a time. Underlying this process ofproject implementation is a project evaluation process whose main objectiveis to determine the economic and social contribution of a particular projectto societal welfare, including economic growth.

In this chapter we adopt the general definition of economic growth from atransportation improvement as ‘the continuous increase in economic activity,in the impacted area, that can be attributed to this investment’. As we shallargue later, if we consider the network aspect of transportation improvements,this definition might be dubious relative to the delineation of the area ofincidence of the transportation effects. The main objectives of this chapterare: First, to provide a theoretical explanation for the potential linkage betweenthe prime benefits from a transportation investment and local economicgrowth. Second, to examine in some detail key elements of the evaluationprocess that affect the measurement of benefits from a transportationinfrastructure project. Finally, to investigate the extent of the potentialeconomic growth effects.

The dominant approach used in project evaluation is benefit-cost analysis(BCA).1 The essence of this approach is a systematic quantification andcomparison of the various benefits and costs generated by a project. To thatend, the various effects from the project are first enumerated and classified asbenefits or costs. Subsequently, an attempt is made to quantify each effectand express it in monetary terms using appropriate conversion factors. Since

Chapter 7

162 Methodology

the benefits and costs effects accrue at different future points in time, it isnecessary to discount them in order to establish commensurability. The derivedpresent values of the streams of future benefits and costs provide the criterionfor the evaluation decision.

Even this succinct description of the BCA approach raises many fundamentalquestions regarding the actual identification, quantitative assessment andaggregation (into a single value) of the multidimensional effects from a giventransportation project. For example, infrastructure projects generate benefitsand costs whose distribution is non-uniform relative to different populations.As a result, we face the problem of how to make judgements between theincrease in the welfare of those who stand to gain from the project and thedecline in welfare of those who stand to lose from it. The BCA literature on thesingle benefit-cost index and similar issues is voluminous and it is beyond ourscope here to review it in any degree of detail.2 Our overall objective is toexamine the potential effect of transportation infrastructure investment oneconomic growth. In the following discussion, we will focus primarily on thoseaspects of BCA that are relevant to this objective.

Within the framework of BCA of transportation investment projects, weneed to determine which benefit effects should be included in the analysis. Amajor argument is that only direct travel costs savings, namely, travel timesand monetary operating costs, should be regarded as benefits in the evaluationof transportation projects (Mohring and Harwitz 1962; Mohring 1993). Asa result, the inclusion of other effects like economic growth as additionalbenefits amounts to the double counting of benefits, since economic growtheffects are in essence a manifestation of the capitalized travel costs savings.Only under very special conditions can such effects be considered as additionalbenefits. The production function analysis (Chapter 6) implicitly presupposesthis to be the case.

In response to this argument, it is possible to claim that infrastructureinvestment is associated with myriad of scale effects and externalities whichproduce more than just travel costs savings. By measuring economic growtheffects we do not necessarily double count investment benefits. Examples ofscale effects and externalities are network economies, land assemblyeconomies, congestion reduction effects and pollution generation orabatement. Furthermore, as Mohring (1993) has pointed out, in cases wherethe economy cannot be regarded as a closed one (e.g. regional economies),the import-export effects of transportation costs reduction is not directlycaused by the capitalization of the primary transportation benefits. Therefore,this growth should be counted as part of the project’s stream of benefits.

As we have seen in Chapter 6, a major objective of the macro levelproduction and cost functions literature is to determine the marginalproductivity of the infrastructure stock in terms of economic growth andimproved productivity. What can this information tell us about a newinvestment project? If we can assume that the political-economic system will

Economic evaluation of transportation projects 163

continue to allocate infrastructure projects as in the past, the information onthe marginal productivity of capital can be used as an additional criterion tofurther screen projects after each has been subjected to a benefit-cost analysis.For example, in Section 6.4 we reported that McGuire (1992) estimated outputelasticity with respect to highway capital to be in the range of 0.121 to 0.370.Assuming this estimate to be credible, we can demand that each transportationproject with a positive net present value (NPV) will also yield at least thatproductivity level.3

Another relevant aspect of project appraisal is the nature of growth. It canbe argued that an increase in labour productivity (from infrastructuredevelopment) can lead to employment decline as less workers would be neededto produce a given level of output. If economic growth is measured in termsof employment then productivity gains may lessen growth. Following thisreasoning, and assuming that transportation projects indeed can raise labourproductivity in some industries in some locations, we might ask whetherfurther investment in transportation infrastructure will lead to job losses inthese industries and to a growth decline throughout the economy. Krugman(1997) assails this kind of reasoning and labels it ‘fallacy of composition’.Briefly stated, productivity gains in one industry lower output prices fromthat industry. As a result, consumers’ disposable incomes rise and more moneyis spent on buying other products, thereby creating jobs elsewhere. Hencewhat we need to observe are changes in employment in the entire economyand not only in the industries where productivity gains were made. The samelogic should also apply to cases where infrastructure investment in a givenregion improves productivity and lowers regional employment.

In reality, there are many instances when the decision to implement atransportation infrastructure project in a given area is predicated on thepurported ability of this project to raise the level of employment in this specificarea (see Section 8.5 for a case study). Mainly for political and funding reasons,employment increases in adjacent areas are regarded by the political decision-makers as incidental. Consequently, it is important to look into the nature ofthe expected benefits from an infrastructure improvement. In this chapter weexamine two major issues, namely travel time savings and activity relocation.

Travel time savings from an infrastructure project can produce multipleeffects. However, the measurement of these effects in addition to travel timesavings is likely to amount to the double-counting of benefits. Only when wecan show value added type effects, which are over and above the ordinarycapitalization of travel time saving benefits can we argue that growth benefitsmay have occurred. Thus, one major objective of this chapter is to explorethe conditions under which such growth effects from an infrastructureinvestment project can legitimately be added to the travel time savings effectswithout performing double-counting of benefits (Section 7.2).

A second major objective of this chapter is to examine key factors thataffect the measured benefits from a transportation infrastructure project and

164 Methodology

which, therefore, affect the measurement of potential economic growth effects.Sections 7.3 and 7.4 critically examine these factors and their treatment in aBCA framework. In Section 7.5 we briefly survey other evaluation methodsof transportation projects.4

7.2 Fundamentals of transportation benefit-costanalysis

The most fundamental effect of transportation investment is to improve travelconditions, which, in turn, alter individuals’ behaviour such as mode choice,route choice, time of travel choice and destination choice. The aggregateresults of these changes in individuals’ behaviour are manifested, at thenetwork level, in traffic volumes and patterns, in travel times and costs byfacility type and in the relative accessibility of locations. These impacts furtheraffect location decisions of households and firms, thus land rent and urbanform. They also affect consumers’ behaviour (e.g. the allocation of timebetween work and leisure), and firms’ production and business decisions. Inaddition, the investment in infrastructure induces other non-direct effects ofwhich the main ones are the investment multiplier and various forms ofexternalities (e.g. environmental degradation). These relationships betweeninfrastructure investment and the direct and indirect effects are traditionallydepicted as in Figure 7.1.

We claim that this view is rather incorrect and present an alternativeapproach (Section 7.2.2). The various effects portrayed in Figure 7.1 raise

Figure 7.1 Traditional view of the effects of transportation infrastructure investment.


two questions relative to the correct evaluation of the overall benefits andcosts from a project. First, what should be regarded as the overall objectiveof the project and how should we define the evaluation criteria? Second,should the benefits be regarded (and measured) only as changes in network’stravel time and accessibility, or should changes in land rent, firms’ costs andeffects from various externalities also be included?

The answer to the first question is rather straightforward in economictheory. In normative terms, the objective of any public project is themaximization of social welfare, given a host of financial and legal constraintsand taking into account income redistribution effects. There is a rich literature,mainly in the field of public finance, on the proper interpretation of welfareimprovements and on the conditions which are necessary for their attainment(see, for example, Musgrave and Musgrave 1989).

The answer to the second question depends primarily on the presence (orabsence) of two basic conditions (see Section 7.2.1 and Chapter 8). First, canwe identify market imperfections leading to effects such as network andagglomeration economies? Second, to what degree are markets competitiveenough to distinguish between price changes from increased consumption orfrom pure transfers between individuals or groups? Given these conditions,how should benefits from a transportation project be measured?

7.2.1 The measurement of benef its from a transportationinvestment project

The most commonly used measure to assess the welfare gains from a publicinvestment such as in transportation infrastructure, is the change in consumersurplus (see, for example, Boardman et al. 1996: Chapter 3). Analytically,it is the area under the aggregate private demand curve bounded by theprices of the transportation service before and after the project wasimplemented.

Underlying this measure is the concept of willingness to pay for travelimprovements. It indicates the monetary sum that an informed consumer isready to pay for a particular improvement produced by a transportationproject. Under reasonable conditions, changes in consumer surplus from animprovement correctly measure the willingness to pay for this improvement(Bradford and Hildebrandt 1977). Since a transportation project may alsogenerate negative impacts such as environmental nuisances, the willingnessto pay principle can also be applied to indicate the amount of money consumersare willing to pay to avoid such harmful effects. These values can then beregarded as part of the project’s costs and be added, after proper quantification,to its capital and operating costs.

Before we proceed with our main subject of assessing the growth benefitsfrom a transportation project, the reader should be aware of the fact that therationale and use of consumer surplus is quite controversial in the economic

166 Methodology

literature. There are a host of difficult questions concerning the aggregationof the demand functions of individuals with diverse preferences and incomesin order to produce an aggregate demand function from which changes inconsumer surplus can be calculated. The key issue is that any aggregationprocedure requires the performance of interpersonal comparisons relative topreferences and marginal utility of income. This raises a further questionregarding the plausibility of using consumer surplus to assess welfare gainsfrom price changes.5

Here, we are making use of changes in the price and volume of travel beforeand after the implementation of a transportation investment project to assess itsprimary benefits. This approach reflects the rationale of using consumer surplusas a measure of welfare gains, where travel times and traffic volumes serve as thekey variables that define the travel demand function. Technically, thecomputations are as follows. By and large a transportation infrastructureproject implies capacity expansion of one or several links of the existingnetwork. Given the origin-destination (O-D) matrix, and using a trafficassignment procedure, link travel times are computed assuming the newcapacity. These, in turn, are aggregated into interzonal travel times. The changein total regional travel time is then computed as: where Di,j, Ci,j, are the interzonal O-D demand matrix and the interzonal traveltime (costs) matrix, respectively, (i,j) are indexes of zones. The superscripts A, Bdenote the ‘after’ and ‘before’ the project.

Notice that for it is necessary to assume inelastic demand and nochange in O-D locations, i.e. If travel demand is elastic, or if sometravellers switch to different destinations or use alternative modes, the signand magnitude of can take any value including The latter casemay imply that the network is as congested after the project as it was before,since travel demand has increased sufficiently to outweigh the effect of thereduction in travel time from the capacity improvement. In this case, othermeasures of accessibility improvements are needed (see later). Next wehighlight some issues relevant to the use of consumer surplus in the contextof transportation investment projects and economic growth.

As already indicated, under normal conditions (see shortly), a change inconsumer surplus from an infrastructure improvement project is used as theprincipal measure of total benefits from this project. Counting other potentialeffects as additional benefits, therefore, amounts to double counting of benefits.In this regard, we need to determine how to treat benefits to new users.Examples are: users that previously did not use a certain mode or a facility(e.g. a congested public mode or a highway), users that did not travel at acertain time period (e.g. at peak time), or that did not travel to a certaindestination (e.g. to city centre). All these types of users may have latent demandfor a certain trip type or a service. Prior to the transportation improvement,they choose not to participate as they perceived the price to be too high. Asexplained by Layard and Glaister (1994:6) and Small (1999), the measured


changes in consumer surplus already include the benefits accrued to thesenew users as well. Hence, adding the benefits of new users to the measuredtotal change in consumer surplus is erroneous. However, summing benefitsthat accrue to new users separately from the benefits gained by existing usersis a valid approach.6

The conditions under which the measurement of the prime benefits from atransportation project (namely accessibility) would not suffice to account forthe total benefits from the project need to be established. We maintain that thiswould be the case in presence of externalities which are significantly high toaffect resource allocation by individuals and firms. Underlying this discussionis our conjecture that transportation changes are a major cause for the generationof externalities at the urban and regional level (Section 7.2.2).

We have already alluded to the fact that transportation services are inputfactors into a host of production, consumption and locational activities.Therefore, a reduction in transportation costs, from an infrastructureimprovement, is likely to affect firms’ and individuals’ behaviour in othermarkets, some of which may be outside the price system. We will refer tothese effects as allocative externalities.7 Essentially, these effects entail twounderlying conditions. First, that actions by one party will impact the utilityor the production function of another party (whether an individual or a firm),thereby influencing their consumption or production levels. Second, that theresponsible party would not compensate the affected party (or would not becompensated by them) for the changes in utility or profit levels. The result isan inefficient allocation of resources in society as a non-optimal amount ofresources would be utilized for the production of goods and services in thesemarkets. One noticeable example is traffic congestion which affects drivers’travel times and volumes, which in turn influences their allocation of timebetween leisure and work activities and their use of infrastructure facilities.One firm’s cost reduction, caused by changes in its proximity to anotherfirm, is another example (see Chapter 8).

Another form of externalities are pecuniary externalities. They arise whena reduction in transportation costs from a project alters relative prices inother markets. Such price changes, in turn, will create benefits and costs tothird parties in these markets. Several researchers (e.g. Small 1999) notedthat if these markets were sufficiently competitive then benefits to one groupwould be fully offset by costs to another. Put alternatively, in competitivemarkets pecuniary externalities amount to transfers among economic entities.Therefore, there is no need to consider them in evaluating the project as longas we are not concerned with income distribution analysis. It also does nothold in the presence of market imperfections such as scale or agglomerationeffects. Under such conditions we need to consider benefits and costs accruedin other markets. The land market provides a good example. Followingchanges in accessibility from a transportation improvement, firms andhouseholds are likely to relocate, thereby raising land rents in some locations.

168 Methodology

Landowners in those sites stand to gain at the expense of the alternativelocations. In general, transportation improvements affect the relativeaccessibility of different locations resulting in some locations becoming moreattractive for residential, employment and commercial activities that previouslylocated elsewhere in the region. In the absence of allocative externalities,these shifts in location also represent transfer of benefits rather than thecreation of new ones.

Mohring and Williamson (1969) have identified another important formof benefits transfer and label it industrial reorganization benefits. Accordingly,reduced travel times from a transportation project affect production logisticsand shipments. Reduced inventories, just-in-time production, consolidationof car or truck loads and overnight guaranteed delivery are examples of thesebenefits. As important as these effects might be, it is necessary to demonstratehow they are capitalized into the firm’s decision-making process relative tothe use of input factors in order to consider them as additional benefits notalready captured by the reduction in transportation costs. A per unit real costreduction is a major indication of such an effect. For additional benefits totranspire it is necessary that allocative externalities will be present, for example,scale economies at the firm level or agglomeration economies at the sector’slevel. Otherwise, industrial reorganization benefits are similar in nature tobenefits accrued to new users and hence are already captured by the changein total consumer surplus.

The principal conclusion from the above discussion is that, in general, allthe benefits from a transportation project are those already represented bychanges in the consumer surplus, caused by accessibility changes. Additionalbenefits should be included if and only if it is possible to demonstrate theexistence of allocative externalities. Adding other effects, such as transfersbetween parties, will result in over-estimation of benefits from the project.

7.2.2 Allocative externalities and growth effects

In this analysis we consider four major categories of allocative externalitiesthat arise from or are affected by transportation infrastructure investmentprojects. These are industrial agglomeration, labour market imperfections,transportation network economies and environmental effects.8 We then showschematically how these effects may influence the level of economic growthfrom a transportation project.

Agglomeration economies

Agglomeration economies are benefits accrued to firms resulting from theirgeographical proximity to other firms. These economies can arise fromintrafirm scale and scope economies at a location, or from inter-firmexternalities (Chinitz 1961). In the latter case, agglomeration externalities


can ensue from accessibility to local public goods, from the use of sharedinput factors, from information spillovers, or from access to a common localpool of trained labour. Whatever the reason, the key element in the realizationof agglomeration benefits is that their level is an increasing function of spatialcloseness (see Chapter 8 for a microeconomic model with agglomerationexternalities).

Labour market imperfections

The second type of allocative externalities are labour market imperfections.Following Cogan (1980) the total supply of labour by individuals, measuredin units of ‘actual hours worked’, is a function of two main choices: individuals’willingness to participate in the labour force and their allocation of timebetween work and non-work activities. In turn, these choices are a functionof three main components: individuals’ preference and attributes includinghouseholds’ constraints (e.g. the number of pre-school age children),institutional arrangements (e.g. work rules) and market entry barriers (e.g.lack of adequate information on the location of employment opportunitiesor spatially inaccessible job sites). The last two components represent labourmarket imperfections in that they introduce discontinuities into the laboursupply function and cause sub-optimal labour market participation decisions.Travel time improvements can reduce these imperfections as, for example,there might be travel costs thresholds above which some individuals mightbe unwilling to enter the workforce. (In Chapter 8 we present some empiricalevidence that support these arguments.)9

Network economies

The third type of allocative externalities are network economies. A newtransportation facility, like a road or a rail link, is typically part of a largernetwork. Due to intrinsic non-linearities in network traffic flow, the additionof such a link can result in increased traffic flow over the entire network thatis larger than just the additional traffic over the new facility. Bottleneck andscheduling models of travel essentially make use of this network characteristic(Small 1992: Chapter 3).10

A related form of allocative externalities, resulting from networkeconomies, is the case where two disjoint networks are linked by a newlyconstructed facility. The effect of such an investment might be similar to thecase where trade began between two markets or countries that previouslywere closed off to each other.11 The overall trade activity that will ensuefrom the transportation investment might be greater than that indicated bythe travel volume over the new facility. Related possible benefits can alsoemerge when, prior to the infrastructure investment, a monopolist serveseach of these markets. As shown by Jara-Díaz (1986), the result of

170 Methodology

introducing a new link that reduces travel costs between these formerlydisjoint regions is the lowering of monopoly rents and deadweight losses byhaving greater competition. It can further lower prices of final products ineach of the individual regional economies. Hence, total users9 surplus fromthe transportation project may not capture the total benefits to theeconomy.12

Environmental effects

Environmental effects caused by traffic and auto use constitute the fourthtype of allocative externalities. As most are unpriced or otherwise unregulated,they can cause significant misallocations of resources in the economy. Ingeneral, environmental effects spur negative benefits and are regarded asindirect costs of transportation infrastructure investment that may inhibitgrowth. Table 7.1 illustrates the magnitude of these externalities.

A major caveat to the above discussion is that a correct assessment ofbenefits needs to be carried out at the point of social equilibrium, where theprice paid by users reflects the full marginal costs of supplying transportationservices. By full marginal costs we mean private costs of use plus the costs ofall allocative externalities including congestion and environmental costs.13

Failure to do so is likely to result in an erroneous estimate of the true benefitsfrom a transportation project (see example in 7.2.3). In conclusion, if theaggregate travel demand function has been properly estimated, allocativeexternalities have been accounted for and the point of equilibrium is correctlydefined, the benefits generated by a project can be accurately measured.

Table 7.1 External effects from transportation in 17 West European countries ($US per 100 users/km, or per 100 tons/km)

Source: Small and Kazimi (1995).


7.2.3 Spatial changes from transportation improvementsand economic growth

Improved accessibility from transportation infrastructure improvements, ceterisparibus, can affect the location decisions of households and firms in the impactedregion. These spatial rearrangements are bound to produce welfare gains toconsumers and producers, stemming from their capacity to relocate to whereutility level and profit-making capabilities are enhanced. The question then isdoes this activity relocation also constitute economic growth?

The common answer to this question is that improved transportationstimulates efficient spatial patterns of households and business and that thisincreased spatial efficiency will spur economic growth. This inference issupported by a vast amount of theoretical analysis, the main conclusion ofwhich is that reduced travel costs (from transportation investments) are likelyto encourage further activity decentralization and, at the same time, intensifyagglomeration and urbanization economies (see a recent review by Anas etal. 1998). Yet to argue that increased accessibility (from reduced transportationcosts) improves spatial efficiency and generates growth is dependent on twoconditions. First, that reduced travel times will indeed result in consequentialactivity relocation and, second, that we agree on what constitutes efficienturban activity patterns.

For an activity to relocate following generalized travel cost reductions, itis necessary that the value of these reduced costs will exceed the costs ofrelocation. In a well-developed network, like those of most contemporarymetropolitan areas, improved link capacity (even a major one) is unlikely tobring about significant travel cost reductions. On the other hand, relocationcosts, which include monetary and non-monetary household’s or firm’s costs,can be quite substantial. Moreover, as shown in Chapter 4, two-employeehouseholds presently make up a sizeable portion of all households. For suchhouseholds, a directional travel time improvement is unlikely to warrant amove. In general, travel time and costs are only one factor among many thatinfluence household’s decision to relocate. Even a ubiquitous change inmetropolitan travel costs,14 may not result in spatial relocation. Arnott (1998),for example, has shown that when considering trip timing decisions andassuming no toll revenue redistribution (hence no income effect), optimalcongestion tolls have no effect on urban spatial form.

Similar arguments also apply to firms’ relocation decisions. Typically,transportation costs constitute but a small fraction of total firms’ productioncosts. Firms’ relocation transpires over long time periods and is often spurredby non-transportation related factors such as tax opportunities or directsubsidies.15 As explained in Chapter 4, the location of firms in the servicesector is usually more responsive to consumers’ demand than to travel costsper se. High-tech firms, which in many cases locate in areas where land costsare already low, are largely insensitive to improved accessibility. Empirically,

172 Methodology

there is little evidence to indicate that noticeable firm relocation can beunequivocally attributed to a particular transportation improvement. (Thecase studies in Part IV document this claim).

Turning now to the second condition, the crucial question here is whencan decentralized spatial activity patterns (induced by travel costs reductions)be regarded as efficient. In a simple monocentric urban model it is rathereasy to ascertain an efficient urban form. But this is certainly not the case fora more realistic multi-centre (including the CBD and many sub-centres)contemporary urban area. Moreover, the durability of structures and oftransportation and non-transportation facilities make observed market spatialequilibrium patterns unlikely to be efficient. Put alternatively, under commonmarket conditions, several spatial equilibrium patterns can emerge, some beingmore efficient than others. History often determines the prevailing (thoughnot necessarily the efficient) one (Anas et al. 1998).

Still another key factor affecting the efficiency of urban forms is the factthat rarely do households and firms actually pay the full social costs ofrelocation, including the increased costs of public services provision and thenumerous pervasive externalities (e.g. traffic congestion). One often hearsarguments against activity diffusion, which are based on the enormous socialexternalities such as housing and job market segregation, that dispersed spatialpatterns entail. Given the durability of urban facilities and structures thereare real costs of reduced activity level at the CBD resulting from furthersuburbanization.

To summarize, in the long run accessibility improvements fromtransportation infrastructure investments tend to result in activity relocationand more dispersed urban patterns. However, considering the predominantmulti-centre urban form, the underpricing of congestion costs and the manyother unpriced urban externalities whose impact is exacerbated by furtheractivity dispersion, the reduction of travel costs via facility expansion isunlikely to result in more efficient urban spatial patterns. Therefore, it is notquite clear how economic growth benefits can emerge (let alone be measured)from spatial relocation following transportation infrastructure improvements.But even if spatial efficiency does improve, following a particulartransportation investment, one must establish the presence of underlyingallocative externalities, like external scale economies, in order to ascertaineconomic growth benefits.

7.2.4 A proposed scheme for the evaluation of economicgrowth benef its from a transport infrastructureinvestment

Based on the above discussion, we now present a modified version of Figure7.1 to demonstrate our conjectures regarding the relationships betweenallocative externalities and spatial redistribution, following a transportation


infrastructure project, and between economic growth. These relationshipsare shown schematically in Figure 7.2. Prior to that, we need to point outthat in both Figures 7.1. and 7.2 we have included investment multipliereffects. They have been placed as indirect effects (Figure 7.1) or outside theprocess (Figure 7.2). There are two reasons for this. First, multiplier effectsare not unique to transport investments. Multipliers are important in allinvestments involving the creation of additional income, consumption andemployment, as there are linkages which lead to second and further roundsof effects. Second, there are many other (non-investment) ways to achievesimilar multiplier effects, such as through the taxation system. The importance

Figure 7.2 The new scheme for the evaluation of economic growth benefits from transportationinvestment

174 Methodology

of multipliers should not be underestimated, but we have not included themas part of the economic growth component as they are considered to occur atone point in time rather than continuing to develop in the future, as impliedby the concept of growth.

Figure 7.2 underscores our contention that economic growth frominfrastructure improvement is predicated on the presence of allocativeexternalities. Spatial redistribution of activities can also lead to economicgrowth, as prior to the project, high transportation costs inhibited someactivities from locating where their marginal productivity (or marginal utility)exceeded the cost of relocation. The key point to observe about Figure 7.2 isthat if allocative externalities are not present in the local economy, then all ofthe benefits from an investment project are confined to travel or accessibilityrelated benefits. These benefits are fully captured by the measured change inconsumer surplus and displayed in Figure 7.2 as ‘welfare gains’. In that case,growth effects cannot be expected, and attempts to regard some benefits aseconomic growth amounts to double counting. In Appendix 7.1, we give anexample to demonstrate the possibility of assessing benefits from atransportation project while taking into account allocative externalities.

7.3 Primary benefits from transportationimprovements: network accessibility

Figure 7.2 highlights our assertion that changes in travel conditions, whichare the primary direct benefits from a transportation investment, are also thefundamental generators of potential growth benefits from the project ifallocative externalities are present. While we do not claim that the relationshipsbetween travel conditions and growth effects are linear, it is our contentionthat the former has a positive impact on the latter. In general, a greaterimprovement in travel conditions will bring about a larger economic growtheffect via its impact on the various allocative externalities but this effect mayabate quite rapidly (see simulation results in Chapter 8). Given thischaracterization of the causality linkage between transportation improvementsand economic growth, it is pertinent to examine the following two questions:first, how to properly define and measure changes in travel conditions; second,how to specify the functional relationships between changes in travelconditions and potential growth effects. In this section we focus on the firstquestion while in Chapter 8 we examine the second.

Overall, changes in travel conditions are commonly summarized by thegeneric concept of accessibility, which can be defined as ‘the ease of accessbetween spatial opportunities’. This definition implies that variables like traveltime and costs, associated with a trip to a location, are the key componentsthat determine accessibility. Alternatively, accessibility can be defined as ‘thepotential attainment of a set of transportation choices’. This definition impliesa subjective judgement that a user must makes among a number of travel


options in order to select one, based on the perceived attributes of each option,his own characteristics and assuming a utility maximization behaviour.

The result of adopting either definition leads to a different quantitativemeasure of accessibility. Under the first travel time (or distance) and themonetary costs of a trip, weighted by the level of activities at the origin anddestination locations, are the main variables that define accessibility. Forexample, in conventional trip distribution models travel time and monetarycosts are combined to produce a single measure of accessibility known as‘generalized costs’. Formally, generalized cost of using mode, k, Ck is computedas Ck = Pk + v . Tk, where Pk is the monetary cost, v is the value of time, and Tk

is travel time by mode k (including access time, wait time, in vehicle time andmodal penalties). Notice that travel time is also a function of travel volumewhere the latter is computed for a specific link of the network. Hence, it ispossible to compute a measure of ‘link accessibility’. Alternatively, it is possibleto aggregate link’s generalized costs over all links to arrive at a measure of‘network accessibility’. In either case, the travel benefits from a transportationproject, which aims at expanding the capacity of a certain highway or of apublic transit mode, are measured as changes in accessibility. As explainedabove such changes are equivalent to changes in consumer surplus.

The second definition of accessibility asserts that accessibility should bemeasured at the individual level since that is the kind of accessibility whichusers consider in making travel choices. It follows, therefore, that in additionto the modal or link attributes (e.g. speed and money costs), given the originand destination locations, the computation of individual accessibility shouldalso account for users’ attributes (e.g. income and demographic variables).Viewing accessibility this way leads to accessibility measures which are basedon random utility models such as multinomial logit. The accessibility ofindividual i denoted by ACCESSi, can then be measured as:

(7.1)

where Vik are observed transportation attributes of a choice alternative k,from the choice set R by individual i.16 The connection between thisaccessibility measure and consumer surplus can be expressed as the differencebetween accessibility level before the project is implemented (i.e. current levelof the attributes) and afterwards. Letting 1 and 2 denote the before and afterlevels, a change in consumer surplus is defined as:

(7.2)

Ben-Akiva and Lerman (1985) have noted that the use of this expression tomeasure accessibility creates a major difficulty, as different specifications ofthe multinomial logit model will produce accessibility terms that are expressedin different units.

176 Methodology

In a seminal paper, Small and Rosen (1981) have shown that for themultinomial logit model, in the absence of income effects, the expected changein consumer surplus can be computed as:

(7.3)

where l is the marginal utility of income (¶Vi/¶I) and Vk is the systematicutility of an individual.17 For example, in a discrete choice analysis andassuming a compensating variation type demand function, (7.3) can bewritten as:

(7.4)

where v1, v2 are the mean of the indirect utility for travel scenarios 1 and 2,respectively.

Niemeier (1997) provides an interesting example of the use of DCS as anaccessibility measure. In her study she empirically calculates the monetaryvalue of mode-destination accessibility for the morning journey to work. Thismeasure, when computed for the entire sample, is shown to be quite differentthan when computed for various market segments. Another example of theuse of a discrete choice-based accessibility measure is in Levine (1998). Hisaccessibility measure is used to assess the impact on residential locationdecisions of commute time, relative to other factors such as suburban housingregulation. He found that while at the regional scale accessibility matters,the match between housing and work places is much more responsive tohousing policies than to accessibility.

At this junction, it is important to emphasize that, whatever the accessibilitymeasure adopted, its formulation and use should reflect the network attributesand performance. As alluded to earlier, this carries consequential implicationsfor BCA of transportation projects. That is, travel time improvement on aspecific link, even if significant, does not necessarily imply a measurable changein travel behaviour on the entire network. In fact, commercially availablenetwork models are unable to properly distinguish between various temporaland spatial effects, newly generated trips (assuming travel demand elasticity,which is greater than zero in absolute terms) and between diversion effects.18

Under such conditions, in a well-developed metropolitan network even amajor investment in few links may generate only small welfare gains to warrantthe investment, as a conventional BCA would indicate. In part this realityprovides added support to the need also to examine potential growth benefitsfrom the project.

A final issue to be discussed here that also bears impact on the magnitudeof the measured changes in accessibility is the non-incremental aspect oftransportation investments. In reality transportation infrastructure investments


are indivisible and lumpy while benefit-cost rules are specified in terms ofmarginal changes. For example, the benefit-cost rule derived in Appendix 1states that under the first-best social equilibrium the marginal dollar investedin road capacity should equal the reduction in total congestion costs fromthat capacity investment. With a non-incremental lumpy investment this rulemay not hold. Under such conditions the magnitude of the overall computedconsumer surplus depends then on the size of the capacity improvement, thechange in the generalized travel costs, the initial level of demand and demandelasticity (Williams and Moore 1990).

As implied by Figure 7.2, economic growth transpires in the impactedarea from agglomeration effects, labour market choices, network economiesand from environmental improvements. How can these impacts be quantifiedand measured in monetary values to be used in BCA? In general, when wedeal with impacts for which market prices exist, these prices should be usedto compute the true economic value of these effects. For example, increasedoutput from agglomeration can be evaluated at the output’s market price.Similarly, if the labour market’s response to travel time reduction is increasedemployment, the market wage rate can be used to evaluate these changes.

In the case of network economies, infrastructure improvements can reducemore than proportionally travel time saving, traffic accidents and increasemode use. For example, for highway networks Kraus (1981) has estimatedincreasing returns to scale with respect to highway length expansion andwith respect to capacity expansion of intersections.19 Similarly, upgradingthe railroad track to allow greater travel speed and comfort is bound to promptgreater use of rail by passengers and freight and decreased highway traveltime and accidents. Unfortunately, these effects do not have direct marketprice. Yet, it is possible to use imputed prices to measure their economicvalues. Thus, through various methods such as revealed preferences in choicesituations, it is possible to assess the amount of money that potential usersare willing to pay to save travel time, to ship freight at greater speed orreduce the risk of car accidents. In environmental analysis, stated preferencetechniques are used to derive willingness to pay to reduce noxious externalities.The value to people of abating these effects can then be computed using thestated values as the conversion factors.

7.4 Key factors affecting the measurement ofbenefits

We have stressed the point that growth effects from transportationinfrastructure expansion are a function of the primary direct benefits fromthe project, namely accessibility improvements, hence the importance ofexamining key factors which bear the largest impact on the measured directbenefits. In this section we examine three such factors. These are: the valueof time; the discount rate of future benefits and costs and the time span of

178 Methodology

projects and risk and uncertainty associated with project selection and withthe assessment of future benefits and costs.

7.4.1 Value of time savings

By far the largest category of direct benefits from any given transportationproject is ‘travel time savings’ expressed in monetary values. These benefitsare a combination of the actual amount of time saved and the value of a unitof time saved, generally known as the ‘value of time’ (VOT).20 Since thelatter component is not directly observable, it needs to be empiricallyestimated, mainly from choice situations. In this section we examine theoriesproposed for assessing the VOT and then present some empirical results onVOT estimates. The reader is again referred to Figure 7.2 which shows ourbasic assertion that the fundamental generator of potential economic growtheffects are travel time savings. Hence our focus on VOT which is the factorthat converts units of travel time saved into monetary benefits.

For persons travelling to or from a location, travel creates three majortypes of costs: money costs; opportunity cost of time and disutility of travelling.The first type refers to out-of-pocket direct costs of travel, e.g. bus fare, tollsor the costs of car use. The second refers to the alternative use of the amountof time spent in travel in carrying out a utility producing activity such aswork. Obviously, the degree to which time can be used productively toaccomplish other activities varies considerably between people and trippurposes. In principal, its value lies between 0 (time saved cannot be usedproductively elsewhere) and 1 (time saved can be fully utilized). The thirdcost component refers to the inconvenience that travelling creates and whichindividuals are willing to pay in order to forgo. A full analytical explanationof the derivation of values of time is given in Appendix 7.2. A major problemwith the model framework presented in Appendix 7.2 is that, in reality, timeallocation decisions are usually carried out jointly within the household aspart of the household’s allocation of time to activities. The utility maximizationproblem may, therefore, be an unsuitable framework for such joint productiondecisions (Pollak and Wachter 1975). Another problem that arises whenassessing the value of time-savings is trip chaining. A significant proportionof all trips are chained trips where the decision to perform a given trip (e.g.shopping) is not independent of the decision to undertake another one (e.g.work trip). In this case the value of time savings for the second trip cannot beassessed as if it was a stand-alone trip which is what the model (Appendix7.2:7.17) indicates.

These and other conceptual difficulties notwithstanding, the basic logic ofbenefit-cost analysis framework necessitates the use of time-savings values.A large number of studies have been conducted in order to compute suchvalues for various modes and under various conditions pertaining to travelpurposes and household’s attributes. What is common to all of these studies


is that they derive the values of time savings from choice situations wheretrip makers are faced with several mutually exclusive alternatives eachconferring a specific travel time or travel time components (e.g., wait time,in-vehicle time and access and egress time). The choice between thesealternatives, while accounting for all other differences, seemingly reveals thevalue that individuals attach to the amount of travel time they can save bychoosing a specific alternative. In general, we can distinguish between severalcategories of choice situations. The main ones are choice of speed, route,mode, time of departure and safety.

Given these choice situations, the literature distinguishes between threemajor modelling approaches to the estimation of VOT. These are analyses ofspecific choice situations, models that consider the determinants of VOTsavings for travel purposes and discrete choice model approaches. Aninteresting example of a specific choice situation analysis, was conducted asa ‘real world’ experiment by Hauer and Greenough (1982). In this experiment,subway riders in Toronto were offered cash rewards in exchange for the lossof a specified amount of time (e.g. not getting on the next arriving train).From the results of this experiment the distribution of the value of time wasobtained as a function of the time of day, duration of the delay, trip purposeand socio-economic attributes. Horowitz (1978) used a regression analysison a stated preference database to study the effect on travel time of triplength, travel mode, time of day trip purpose and road conditions. Hensher(1989) proposed a model which considers the determinants of VOT savingsfor business travel and used it on stated preference data to assess the choicebetween a tolled and a free urban road.

Given that the VOT is estimated from choice situations it is just naturalthat discrete choice modelling, based on random utility theory, is theprevalent method used for estimating VOT.21 The objective is to define thesystematic utility component V in a way that enables empirical estimationin terms of observed socio-economic characteristics of users and tripattributes. For example, V can be defined as a linear function of travel costsand travel time, i.e. where the indices i, n indicate anindividual and a travel mode, respectively. From this definition, the marginalvalue of time saved, which is the marginal rate of substitution between timeand money, is ß1/ß2.

Small (1992:19), specified Vin as a non-linear function of travel costs, traveltime and the wage rate of an individual. He further introduces mode dummyvariables that interact with travel time to reflect choice alternatives. As aresult, the estimated value of time saved also varies across modes.22

A key observation from the review of the numerous available travel timestudies is that VOT varies considerably among individuals, locations, marketsegments and time of travel, and that these variations are a function of alarge number of variables. For practical application, therefore, it is commonto use VOT that represents an ‘average value’ for a group of trip makers with

180 Methodology

similar socio-economic and travel profiles. For example, it is conceivablethat the VOT for private auto users travelling to work at peak time issignificantly higher than that in non-work travel (Guttman 1975). Small (1992)and Waters II (1992) estimated VOT for peak-time users to be 50 per cent ofthe hourly pre-tax wage rate.23

Several general conclusions regarding actual VOTs can be drawn from theempirical literature on the subject. First, having higher flexibility, relative tothe various time constraints, tends to increase VOT. For example, the studyby MVA (1987) has shown that people with variable work hours have VOTwhich is 15–20 per cent higher than that of other workers. Similarly, theability to schedule activities tends to increase estimated VOT values (Small1982). Higher tax rates have an opposite effect (Forsyth 1980).

A second major conclusion is that non-linear relationships exist betweenincome and VOT. Thus, studies have shown that VOT for in-vehicle time isless than the hourly gross wage rate but it is an increasing function of incomesince rising income implies larger opportunity costs of time saved. Waters II(1992) has estimated the elasticity of VOT with respect to income to beabout 0.8, though this general elasticity value depends on trip length, modeuse and trip purpose. Gunn (1991) reports smaller effects of income on VOT,except for business travel.

A further conclusion is that users place a higher weight on wait and walktime relative to in-vehicle time. Usually, the value of these time component istwo to three times that of in-vehicle time. In this regard, inter-modal transfers,which implies additional wait time, entail heavy opportunity costs of timeand therefore should be weighted accordingly (Small 1992).

Finally, the common practice is to use the same VOT for small and largeblocks of time saved. This practice has been challenged on the grounds thatsmall time blocks saved (e.g. 5–10 minutes) are valued less by trip makersthan larger ones (e.g. 20–25 minutes). Following this rationale it might beargued that the total value of time saved for a large group of people whereeach saves only 3–5 minutes is negligible. Obviously, it is possible to introducecounter-arguments, for example, that for some activities small time savingsare a sufficient perquisite for undertaking them. In general, the use of averageVOT across people probably accounts for this problem mainly because asmall but a significant number of people place a very high value even onsmall time blocks.

Time savings for commercial traff ic

A major proportion of all highway traffic is commercial vehicle movement,mainly of trucks. Therefore, in assessing total time savings from an infrastructureproject we need to compute separately time savings for truck traffic as well asthe VOT factors that are applicable to trucks of different types.

Total time savings for commercial vehicles are composed of three


components: driver’s time, vehicle’s time and time savings of the cargo carried.Typically, the value of the third component is not calculated and is muchsmaller than the first two.24 Waters et al. (1995) list four alternative approachesfor estimating the value of time for commercial traffic. The main two arefirst, the ‘cost savings’ approach, which regards the costs that trucking firmscan save from reduced travel times while hauling the same level of freight.The second is the ‘incremental revenue’ approach, which assesses the valueof extra output that can be produced from reduced travel time.25

The cost saving approach essentially is equivalent to the measurement ofproducer surplus when the cost of travel declines though, under this approach,only a portion of the surplus is actually being regarded as ‘cost savings’. Thatis, given the down sloping aggregate demand function for shippers’ servicesand given that drivers’ and vehicles’ time are real inputs into the shippers’cost function, the same output can now be produced at lower costs to thetrucking firm. Yet, at lower costs (i.e. a downward movement of the supplycurve) and in a competitive market, at equilibrium, more output will bedemanded and produced so that total producer (and consumer) surplus israther larger than the measured cost savings.

The incremental revenue approach is the computation of the value ofadditional output at previous equilibrium price. This value depends on theelasticities of the aggregate demand and supply functions. It measures grossrevenue, which is subject to taxes and perhaps other non-related costs (e.g.costs of scheduling of additional traffic to haul the additional output). Interms of BCA this approach better reflects the value to society which isengendered by the infrastructure project than does the cost saving approach.The reason being that it shows the economic value of the additional outputthat the freed resources (drivers’ and vehicles’ time) can produce, where thisadditional output is evaluated at the gross market prices.

Blauwens and Van de Voorde (1988) have used a mode choice approachto determine the value of time savings in commodity transport. Their approach,which is similar to that used in commuting markets, is based on comparingthe costs of hauling freight by truck versus inland waterways. Using aregression analysis on data from 43 districts in Belgium for 15 commoditygroups, they conclude that VOT in freight is 74.3 per cent of the value of thecargo (see Table 7.2 for Belgium).

As with small blocks of time saved for private car traffic one needs to askwhether small time savings are valuable to shippers. An infrastructure project,even if large, that produces small time savings in a network setting wouldshow relative little benefits if small time savings would not have an economicvalue. Furthermore, the cumulative effect of a number of infrastructureimprovements that have a significant impact on travel time cannot be correctlymeasured if small time savings would not be valuable to firms and consumers.Taking this to be their working assumption and following the incrementalrevenue approach, Waters et al. (1995) have estimated the maximum value

182 Methodology

of time savings for trucks in British Columbia, Canada. Their estimated valueswere CDN$25.51 per hour for two-axle diesel truck hauling bulk commodity,to CDN$31.12 per hour for 7–8 axle trucks. For general freight, the figureswere CDN$31.02 and CDN$35.82, respectively. They also computed thevalue of time for small trucks and carried out an extensive sensitivity analysisto test for the effect of various assumptions (e.g. the proportion of drivers’hourly wage rate that should be regarded) on the estimated values. Table 7.2shows VOT values by trip purpose in various countries. The values for businesstravel were computed using the opportunity cost approach. All values areexpressed in 1997 European currency unit (Ecu) per vehicle hour, exceptwhere noted. Some countries do not differentiate VOT values by trip purpose.

Table 7.2 Value of travel time savings by trip purpose used in different countries (in 1997 Ecuper vehicle hour) and for commercial vehicles (in 1993 US $)

Sources: Waters et al. (1995); Haaland and Odeck (1997).

Notes:a Figures are from EURET 385/1994 report and are shown 1990 Ecu.b Vehicle occupancy rate is assumed to be 2.4.Value of time per person is calculated by

dividing per capita GDP by the number of annual hours.c Values are derived from stated preferences studies.d Mean values from various states.e Five-axle diesel vehicles.f American Association of State Highway and Transportation Officials (AASHTO) 1977

Manual on User Benefit Analysis of Highway and Bus Transit Improvements (Washing-ton DC).

g For Ontario.h State highways.


7.4.2 Discount rate and time span of projects26

As emphasized earlier, the essence of BCA is to determine the degree to whichcapital and operational costs of a project are recovered over its lifetime. Thecommon evaluation criteria are net present value (NPV), internal rate ofreturn (IRR) and, to a lesser extent, multicriteria analysis (MCA). The use ofthese criteria requires the specification of a discount factor of future streamsof benefits and costs as well as the first and last year of the project. In thissection we focus on the choice of the discount rate and its impact on themeasured direct and growth benefits.

The importance of the discount factor cannot be exaggerated.International data show that by and large they are determined at the stateor national level and are obligatory for all projects initiated and implementedby public local, regional and national agencies. There is ample literature onthe ‘correct’ discount factor that ought to be used in the evaluation of publicprojects. The major issues examined in this literature are the theoreticalunderpinnings of the proper discount rate, how to actually calculate it andwhether the same rate should be applied to all types of public projects.

As highlighted by Figure 7.2, total benefits from a transportation projectinclude direct travel benefits and externalities effects that, in turn, includegrowth benefits. For the purpose of discounting future benefits we need todecide whether these benefit types should be lumped together into a singlemeasure of total benefits, or whether each should be evaluated separatelywith respect to discounting and time horizon. Using conventional criteriaof evaluation (see shortly), this decision can be rather consequential for theacceptance or rejection of the project.

The key idea behind much of the literature on the correct specification ofthe discount factor is that it should reflect the marginal rate of time preference,sometime also referred to as the rate of substitution of time preference.Succinctly stated, the overtime consumption pattern of consumers indicatestheir preferences about how to allocate their income between consumptionand investment (or savings), where investment in this period impliespostponement of consumption to later periods. Since governments raise capitalfor public projects by taxing consumers or by borrowing from them, iteffectively affects their temporal consumption and investment decisions. Inaddition, investment in risk-free assets (e.g. government’s risk-free bonds orthe sale of such assets) is a typical mechanism available to consumers tochange their consumption pattern between periods, hence, the use of theinterest rate of government’s risk-free bonds to determine consumers’ marginalrate of time preference.27

Investment funds raised by the government become unavailable forinvestment by firms that could have earned a market rate of return.28 Therefore,one plausible approach for obtaining an approximation for the ‘correct’ discountfactor is to compute the weighted average of a risk-free rate asset and the

184 Methodology

market rate on private investments. The weights are the proportions of thepublic funds obtained from consumption and from investment respectively.

There are a number of technical and theoretical problems with thisapproach, including the fact that capital funds invested in any specific publicproject generally come from a large pool of funds (e.g. the general budget).As a result, frequently it is impossible to ascertain what proportion of thisproject’s capital has been drawn from consumption or from private andcorporate investment. Moreover, since the weights depend on the project’sfinancial flows that are attributed to consumption or investment, and sincethese flows need to be discounted first, there is a problem of simultaneity incomputing the weights. If funding of a project is primarily a combination ofbonds issued specifically for this project, bank loans and dedicated taxes (e.g.gasoline tax for roadworks), we further face the issue that each source issubject to a different interest or tax rate. This fact again affects the computationof the ‘correct’ discount factor for this project.

Other problems in computing the proper discount factor relate to thevalidity of underlying assumptions, including that of a competitive financialmarket, the perfect insight of consumers with respect to future income andconsumption,29 and the neutrality of taxation relative to financial markets.The initial distribution of income and wealth presents another difficulty sinceit affects in the marginal rate of time preference of individuals.

The question as to whether all project types should be discounted usingthe same factor also needs to be examined. This becomes an important issuewhen evaluating projects whose output is a public good (e.g. adding capacityto the road network) as compared with projects whose output is a privategood (e.g. subsidized transit trips). As pointed out by Stiglitz (1982), themarginal rate of substitution of public goods between time periods issystematically different from that of private goods.

Following these arguments it is of no surprise that there is no consensus in theliterature regarding the computation of a ‘correct’ discount rate. Both Baumol(1964) and Stiglitz (1982) maintain that the proper discount rate for public projectsshould be somewhere between consumers’ rate of substitution of time preferenceand the social marginal rate of productivity of private capital (the shadow priceof capital). Sandomo and Dreze (1971) have developed a model which takes intoaccount the inflow of foreign capital, government borrowing power andgovernment’s budget constraint. They showed that in an open economy the correctdiscount rate of public projects is a function of the risk-free interest rate, the priceof capital used by private firms and the marginal rate of return on foreign capital.

The above theoretical constructs notwithstanding, in many countries thecommon practice is to use a discount factor which is dictated by an overseeingagency (e.g. the ministry of finance), and which is not necessarily anchoredin a well-founded economic theory. The rate imposed reflects other objectivessuch as the wish to promote a given type of project. For example, in Germanythe common practice in evaluating transportation projects is to employ a low


discount rate (presently 3 per cent). It depends, inter-alia, on the marketinterest rates, future interest rates and the client’s (the state or the region)time horizon. Underlying this low discount is the expressed wish of the Germanfederal government (FRG) to encourage long-term projects whose benefitswill be received in distant future times. In the Netherlands, the official rate ofreturn for low risk projects is 4 per cent. For projects with higher risk levelsa different rate is applied, which is based on the long-run net interest rate ofgovernment bonds. In Greece a project is evaluated on the basis of the firstyear rate of return only and, as a result, in road projects no discounting offuture benefits and costs is performed at all. Table 7.3 provides data on

Table 7.3 Evaluation method, evaluation period, discount rate and use of residual capital valuesof transportation projects,1995

Source: EURET/385/94 report commissioned by the European Commission DG VII.

Notes:a The project’s benefits are estimated for 30 years and assumed constant thereafter.

Benefits and costs are discounted over infinite lifetime.b Future discounted maintenance costs are added to the first year total project’s costs.c Various trial discount rates are used as a form of sensitivity analysis.d This is the official rate of return for risk-free projects. For risky projects a higher dis-

count factor is applied. This level is based on the long-run net interest rate of govern-ment bonds.

e Austroads (1996).f US Office of Management and Budget (OMB) 1992.

186 Methodology

evaluation methods, evaluation periods, discount rates and use of residualcapital value in different countries.

7.4.3 Risk and uncertainty in project evaluation

In the project evaluation literature the terms risk and uncertainty are used toimply several different meanings. In general, the term risk is used to indicatethe likelihood of selecting the wrong project or a project which is economicallynon-viable. Since public projects displace private ones, and since privateinvestors are risk averse and allow for risk in their choice of privateinvestments, BCA should also account for individuals’ risk aversion. If not,the public sector might accept projects that have been rejected by the privatesector (Webb 1973). However, it might be argued that the whole purpose ofBCA is to weed out unwarranted projects by a careful examination of theirexpected costs and benefits which, in essence, is an attempt to minimize therisk of selecting inadequate projects.

Sometimes the term risk is used to indicate the effect of a given project onthe welfare of taxpayers, relative to the distribution of the project’s indirectcosts (negative externalities). Arrow and Lind (1970) have argued that, inthe case of public projects, the costs (and benefits) of a project are dispersedamong a very large number of consumers and are inconsequential in assessingthe project’s value. Yet projects (e.g. a road expansion) expose a specific andrather small segment of the population to the risk of reduced welfare fromthe potential adverse effects of this project. In such cases, we need to includethe consumers’ costs of bearing this risk in the BCA, which in turn is likely toaffect the ensuing potential growth effects, if any.

The term uncertainty is sometimes used to indicate the degree of inaccuracyassociated with the forecast of the project’s future benefits and costs. We willreturn shortly to this issue. Still another aspect of uncertainty is that of theproject’s degree of inflexibility. This term is used to imply that irreversibilitycosts are rather prohibitive. That is, since there is always an intrinsic level ofuncertainty regarding the future state of the economy, a project that is largelyirreversible, or that cannot be stopped without rendering its costs sunk, shouldbe ranked inferior to one with same NPV but which is flexible to a reasonabledegree.

Inaccurate estimate of benefits and costs can still result in the selection ofa ‘correct’ project. Here, we focus primarily on uncertainty, while implicitlyassuming that the reduction of uncertainty by means of better models anddata will also lessen the risk of selecting a wrong project, or a project thatconfers unacceptable risk on specific segments of the population. Underlyingthis view is the notion that the whole objective of BCA is to diminish the levelof uncertainty and risk associated with a particular project.

As with the other factors (e.g. the value of time and appropriate discountfactor), the level of economic growth that can be expected from a given


infrastructure project is a function of the degree of uncertainty associated withthis project’s estimated impacts. That is, underestimation of the project’s costsor overestimation of its primary benefits (i.e. travel effects) may result insignificantly distorted economic growth estimates. As indicated by Figure 7.2,the functional relationships between the level of the primary benefits and theeconomic growth benefits from a given project, are non-linear. Hence, a specificlevel of uncertainty associated with the estimated direct transportation benefitscan be amplified by the forces (allocative externalities) that generate growthbenefits, thereby rendering them too excessive to be credible.

How severe are these problems in actual benefit-cost analysis? A numberof studies have attempted to ascertain how accurate benefits and costsprojections were, done as part of project evaluation, prior to implementation.To that end, they have carried out ex ante and ex post comparisons ofparticular infrastructure investments. Boardman et al. (1994) applied thisapproach to a 303 km highway project in British Columbia and concludedthat major differences in net benefits were not due to (what might have beenexpected) differences in estimates of benefits, but rather to underestimatedactual construction costs. Skamris and Flyvbjerg (1997) have concluded thatin the case of Danish bridge and tunnel projects, on the average, constructioncosts were consistently 50 per cent to 100 per cent undervalued, whereastraffic forecasts were about 60 per cent overestimated.

In a much cited study, Pickrell (1989) has surveyed ten US rail projects. Hefound that in all cases actual ridership was away below the ex-ante estimates,while actual capital and operating costs surpassed the projected ones by about50 per cent. In some cases average cost per rail passenger exceeded theestimated costs by as much as 188 per cent. Kain (1990) has studied the ex-ante land use and ridership projections for a $2.6bn rail transit investment inDallas, Texas (DART) and has demonstrated how inflated and even misleadingthese projections have actually been.

To balance the picture, other studies have found that in some cases theprojected benefits and costs were quite accurate (Walmsley and Pickett 1992).However, the majority of studies on this subject have concluded that, on theaverage, proposed transport systems cost 50 per cent more than their ex-anteestimates, while the ex-post demand is about 50 per cent below the estimateddemand. This conclusion seems to underlie an established maxim intransportation BCA, which says that in order to arrive at the correct benefitsand costs values of a transport infrastructure project one should halve theproject’s predicted benefits and double its estimated costs.

The reasons for the prevalent erroneous estimates of benefits and costsfrom a given project vary considerably. They range from unsubstantiatedworking assumptions, misspecification of models, inadequate data and thepursuant of incongruous objectives to careless or even deceitful analysis. In avery large study at the World Bank, a large gap has been discovered betweenthe economic rate of return of over 1,000 projects approved by the bank

188 Methodology

during the period 1968–80 and the rate of return of these same projectswhich were reestimated some years later (Little and Mirrlees 1990). Asignificant proportion of projects whose economic rate of return was judgedsatisfactory, when the original cost-benefit analysis was carried out, turnedout not to be so in the follow-up analysis.

Increasingly, transportation infrastructure projects are financed throughcapital market funding which, in turn, implies financial risk. One major sourceof risk is volatile interest rates or exchange rates in the case of internationalfunding (Haynes and Krmenec 1989). Another source is due to the fact thatuser charges are increasingly becoming the main source of income to supportprivately or even publicly financed capital projects (see, for example,Hirschman et al. 1995). However, in many cases the projection of futurerevenues from user charges is rather dubious. The main reasons are: theinherent uncertainty regarding the level of future demand and thecorresponding demand elasticities and the effect of other infrastructure projectson the temporal and spatial distribution of demand.30 A further source ofuncertainty is associated with technological changes in the transportationsector. The use of intelligent transportation systems, electronic toll systemsand automated busways, as well as the introduction of a new generation ofhigh-speed trains and guided vehicles, is likely to make present projection offuture benefits and costs quite unreliable.

What can be done to account for risk and uncertainty in the evaluation ofinfrastructure transportation projects? Above, we have mentioned the argumentby Arrow and Lind (1970) that for public projects the risk margins of costs (ornegative benefits) are quite small as they are distributed among a very largenumber of individuals. Thus, they can be regarded as certain sums and nospecial allowance for the project’s risk is necessary. However, in cases where asignificant portion of the project’s costs (or negative externalities) are borne bya small and identifiable group of individuals (e.g. noise pollution or trafficnuisance in a certain locale), their monetary value should be added to theproject’s costs. In the case of externalities, their shadow prices should becalculated as additional costs associated with their removal. Alternatively, therisk premium that the affected individuals would need to assume in order toavoid the expected decline in their welfare can be regarded as an additionalcost of the project.

With regard to uncertainty, Little and Mirrlees (1974) have shown that inorder to account for uncertainty relative to the actual value of the social netvalue (SNV) of a project it should be adjusted according to the followingformula:

(7.6)

where E(NPV), E(GNP) are the expected values of the project’s net present


value and the expected value of gross national product, respectively. Theterm cov(NPV, GNP) is the covariance of these two variables and ß is thecoefficient of relative risk aversion. The decision rule then is, if SNV is positive,the project should be undertaken.

It was mentioned above that a project’s inflexibility or its degree ofirreversibility is a major source of uncertainty. It is due to uncertain futuredemand, on the one hand, and the sunk costs property of the investment onthe other. One possible way to reduce this type of uncertainty is to defer theimplementation of the project by a certain period (say, one year). Such adelay may involve the loss of benefits for a period of that duration which, inturn, needs to be weighted against the benefits of acquiring new informationabout future demand, plus the saved opportunity costs of capital. Pindyck(1991) and others (e.g. Brennan and Schwartz 1985) have argued that in thepresence of sunk costs these benefits can be quite large and ignoring themcan lead to incorrect investment rules such as the standard NPV. These rulesof appraisal must, therefore, be modified to account for the opportunity costsof delaying the project.

In transportation infrastructure investments, a significant proportion ofthe capital costs is irreversible, which in essence makes them sunk costs. Theirreversibility of transportation infrastructure projects stems fundamentallyfrom two sources: the use of land and structures and legal and politicalcommitments. Transportation investments such as rail, highways or portsare the result of long-range planning processes that involve the acquisitionand legal enactment of rights of way. If, subsequent to these processesimplementation does not follow, the costs involved with these activities cannotbe recovered. Similarly, once construction begins, land and structures can beconverted into alternative uses only at prohibitive costs. Moreover, inanticipation for the implementation of the announced project, developmentalong the project’s planned route or site is likely to take place, thereby confiningthe alternative use of land. In addition, in the context of urban and regionalplanning, there might be significant political sunk costs if the project is notcompleted as planned.

At present, in many countries there is the political resolution and evenpublic support to undertake transportation projects in which tolls (whetheror not congestion tolls) are used as a means to mitigate traffic and to recovercapital costs. What is not known with certainty are the demand elasticities,hence the future revenues from the project. The reasons for this uncertaintycan vary, ranging from the unknown effect on demand of concurrentinvestments in complementary or substitutable transportation facilities, tothe relocation of residential, commercial or employment activities. Whateverthe reason might be, if the recovering of capital costs is a necessary conditionfor the investment, faced with uncertain future revenues and assuming sunkcosts, the possible impacts of project delay should be examined.

What then should be the economic decision rule whether to invest now or

190 Methodology

postpone the investment to a future period? Following Dixit (1989) andPindyck (1991), capital sunk costs should be spent if the monetary value ofthe project’s output, P, is greater than the project’s variable costs, c, and itsdiscounted (sunk) capital costs, k, and if the variance of future values of P,(denoted by s ) is zero. That is, the decision should be to invest today if

where r is the discount rate. However, if s > 0, (i.e. there existsuncertainty over future demands of the project’s output), there are opportunitycosts for delaying the investment since P can go up. In that case it may pay todelay the investment by one year and then invest only if P increases.31

A known characteristic of transportation infrastructure investments istheir lumpiness. When given an origin and destination location it is notpossible to construct a variable portion of a highway or railroad. Similarly,it is not possible to construct varying sizes of airport runway or seaportdock. By and large, technology, geography and intermodality requirementsdictate minimum size investments that can be augmented only by discreteunits of additional capacity. As a result, the investment function is discretein capacity. Therefore, at the time of the initial investment it is necessary todetermine how much to invest and how much capacity will be added on inthe future by incremental investments. The rule for an optimal investmentassuming a continuous investment function, demand certainty and no sunkcosts is shown in Appendix 7.1. However, if demand is uncertain andtransportation infrastructure investments are irreversible, it may be optimalto implement only an investment of a minimum level capacity, postponingadditional investments to future periods. In this case, it is necessary to includethe opportunity costs of a delay in computing the NPV of each incrementalinvestment.

What then would be the optimal investment in incremental capacity?Because of the lumpiness of infrastructure investment, analytical rules (i.e.those in Appendix 7.1) are difficult to derive. However, to the degree that theliterature on firms’ investment decisions under conditions of irreversibilityand uncertainty is germane to the transport investment problems, we canconclude that uncertainty over future demand and irreversibility may increasethe value of a marginal unit of capital.32 Pindyck (1988) has shown thatdemand uncertainty will result in delayed incremental investments, but itwill make these investments larger when they are finally made.

From a transportation policymaking viewpoint, there might be political,institutional or even legal costs for a delay. If the costs of a postponement asa function of the delay time become increasingly prohibitive (e.g. due toobjections by opposition to the project or if it may result in lengthy litigation),it is less likely that the delay option will actually be exercised. Similarly, if thealternative costs of land, presently held by the state, are an increasing functionof elapsed time (e.g. due to residential development in the vicinity of theproject), it may become uneconomical to exploit the delay option.

When examining the timing impacts of transport infrastructure


investments, an interesting question arises relative to the actual period whenlocal economic growth effects begin to transpire. That is, accessibilitychanges, which are the prime generators of growth benefits from the project(Figure 7.2), might accrue only at the conclusion of the project (or at leastof a substantial part of it). Yet it is quite plausible that in anticipation ofthese accessibility improvements, some development might begin even priorto the project’s implementation. In that case it becomes necessary to accountfor these benefits as if they accrue at the time the initial investment is made,rather than at later periods. Of course, such considerations will affect theproject’s computed NPV. Presently, within the framework of BCA, there isno established analytical procedure to predict the time when growth benefitsfrom a project actually materialize. The best we can do is to learn from casestudies about the revealed response of households and firms to plannedtransportation projects. In the empirical part of the book (Part IV) weexamine some aspects of this issue.

Several lessons can be drawn from the above discussion. First, risk anduncertainty factors are intrinsic to benefit and cost assessments oftransportation infrastructure projects. As such they can significantly affectthe actual travel and accessibility benefits from a project which, in turn, affectthe magnitude of potential economic growth effects that are attributed to theproject, hence the importance of properly appraising the levels of risk anduncertainty of a given infrastructure investment. Second, there is the questionof who actually makes the benefits and costs projections. That is, since theuncertainty associated with benefits and costs forecasts is central to the overallrisk profile of a project, one has to consider with great care projectionsprepared by bodies that are directly linked to the project or that stand tobenefit from its implementation. Rather, the use of an independentorganization is greatly recommended. Third, it is highly desirable that thefinal estimation of benefits and costs will be carried out later in the processwhen detailed information becomes available. This estimation can complementearlier forecasts prepared at the initial stage of alternatives analysis (Walmsleyand Pickett 1992). Finally, it is desirable to assess the value of postponing theproject’s implementation by a given period of, say, one year. The annualsavings in capital costs should then be compared with the foregone first yearbenefits properly discounted and adjusted for the uncertainty.

7.5 Alternative evaluation approaches

As is evident from Tables 7.3 and 7.4, the most common method used for theevaluation of highway and rail projects is traditional benefit cost analysiswhich includes benefits to costs ratios, NPV and IRR. In some countries (e.g.the Netherlands and Belgium), other methods are also used, mainly tocomplement BCA. The major differences between BCA and these otherapproaches are in their purpose and scope. Whereas BCA uses measures that

192 Methodology

relate mainly to the specific objectives of the proposed projects (e.g. traveltime reduction), other methods use a broader range of measures that relateto social and environmental objectives. Moreover, whereas BCA uses as itsevaluation criteria quantitative objective measures (e.g. increased trafficvolume), other methods also use qualitative-subjective measures like socialcongruity. It is convenient to classify all evaluation methods into four maincategories: 1 Benefit cost comparisons:

• cost-effectiveness analysis (CEA).• benefit-cost ratios.• benefit-cost analysis (BCA).• risk-benefit analysis.

2 Multi-criteria analysis (MCA): This is really a large family of methodsdesigned specifically for particular applications (e.g. regime analysis, flagmethods, discrete and continuous methods).

3 Impact statements (IS):• social impact statement.• environmental impact statement.

4 Others:• total cost analysis.• full costs and benefits analysis.• project’s life cycle analysis.

Given our focus on economic growth, it is pertinent to ask which of theseapproaches is most suitable for the evaluation of economic development effectsfrom an infrastructure project. A related issue is whether it is desirable toenforce the use of a standardized evaluation approach for all projects by allpublic decision-making bodies. While the use of a single method avoidsdisagreement among partners to the same project (e.g. various municipalities)and enables a true comparisons of results, the use of an inadequate appraisalmethod can foster the selection of the wrong projects everywhere.

A 1994 study by the European Commission has categorized all impactsfrom BCA into nine categories. Six are mandatory impacts that should beassessed in every case and three are discretionary. The mandatory impactsinclude construction costs, maintenance costs, vehicle operating costs, timesaving costs, safety and local environment. The discretionary impacts includestrategic environment, strategic planning and economic development, andstrategic policy. Thus, economic growth was relegated only a secondaryweight in the overall evaluation of a project. On the other hand, recentISTEA legislation in the US has placed economic development effects at thetop of the list of impacts that a local planning organization (MPO) mustconsider.33 How do these ‘other approaches’ handle economic developmentimpacts?


As already explained, a key difficulty with the application of BCA is itsinherent prerequisite to quantify all benefit and cost elements and then convertthem into monetary values. Obviously, in many cases such efforts can turninto highly speculative and subjective activities. For this reason the idea behindCost Effectiveness Analysis (CEA) is to select a policy alternative which isassessed to generate the greatest amount of ‘output’, given the investment.We use the term ‘output’ to denote that outcomes from a policy are notconverted into monetary value and aggregated to produce total benefits as inconventional BCA. Rather, the results from a policy are measured in realunits. The CEA approach then produces an index of this output over theavailable budget. Application of this index to different policy alternatives issubsequently used to select the ‘best’ policy option. For example, output froma policy whose objective is traffic calming can be ‘the change in the numberof vehicles that cross a certain intersection’. Notice that this approach doesnot guarantee that the project’s total benefits would exceed its total costs, asin the case of BCA, nor does it concurrently consider multiple types ofoutcomes from the project.

In contrast, the Multi Criteria Analysis (MCA) family of approaches tacklesthe problem of multiple outputs from a project that cannot be measured inmonetary values and then aggregated into a single index of total benefits. Itdoes so by placing weights on the different benefit and cost effects (or objectivesand criteria) which reflect the importance that a decision-maker attaches toeach effect. Thus, total benefits from the project, TB, are measured as

where wi is the weight given to effect type i, (whose magnitude

is Xi), and The weights can be derived by pooling the opinion of

experts or simply by asking decision-makers to define them. However, inreality when it comes to deciding on the proper weight to assign to eachcategory of benefits and costs, there is no agreement on how to derive a fairand systematic weighting scheme—the weighting systems and values obtainedare subjective. In this regard, MCA is not any better than BCA, which requiresthe quantification of all variables in monetary units. For example, if economicgrowth is an important criterion for the undertaking of an infrastructureinvestment, a policymaker will attach a larger weight factor for growth effects,thereby affecting the selection of projects.

Impact statement (IS) methods typically require the enumeration andquantification of all impacts, including the positive and negative ones.Subsequently, an impact matrix showing the effects from each policyalternative is produced. A decision regarding the choice of the ‘best’ alternativeis then made on the basis of inspecting the impact matrix. Under IS, there isno attempt made to distinguish between effects that should be regarded asbenefits and those that are costs. Thus, in terms of BCA, double counting ispossible. Moreover, since there is no aggregation of effects to compute totalbenefits and total costs, the final choice of an alternative is largely arbitrary.

194 Methodology

In terms of economic growth effects, the IS method is similar to MCA inidentifying such possible effects except that no weights are attached to them.

It is possible to find in the literature many variations of the above keyapproaches, in particular of BCA. We group these methods under the header‘others’ and notice that basically this category includes methods that can beregarded as an expanded list of costs and benefits. DeCorla-Souza et al. (1997)have proposed total cost analysis (TCA) as an alternative to BCA in theevaluation of transportation projects. Under their approach the full costs ofeach alternative or of each mode (including direct and indirect costs) areaccounted for.

Given the above discussion, which method is the most well used for theevaluation of transportation infrastructure projects? For illustration purposes,Table 7.4 summarizes the principal characteristics of the evaluation methodsused in the in European Union (EU) for the appraisal of rail projects.

Table 7.4 Summary of rail investment appraisal attributes in EU member countries


Source: Commission of the European Communities (1992).

Notes: NPV=net present value; IRR=internal rate of return; B/C=benefit/cost; N/A=notavailable.

196 Methodology

It is evident from Table 7.4 that most EU countries use the traditionalBCA approach with net present value (NPV) and internal rate of return (IRR)as their key evaluation criteria. It should be understood that in many Europeancountries the rail system is regarded as the backbone of the entiretransportation system. As a result, special efforts are made by these countriesto implement long-term and large-scale rail projects. For example, smallcountries like Denmark and the Netherlands, in carrying out BCA, placespecial emphasis on international rail links. In Germany a very low discountrate is used to encourage long-term rail projects.

7.6 Conclusions

The principal objectives of this chapter were: first, to examine the underlyingrationale that can be used to explain the linkage between the prime benefitsfrom a transportation infrastructure investment and local economic growth;second, to examine at some detail, key elements of the evaluation processthat affect the measurement of benefits from a project and, consequently, theextent of the potential economic growth effect; third, to provide evidencefrom various countries relative to their use of evaluation methods and criteria.Since benefit-cost analysis is the leading practical approach to the assessmentof transportation infrastructure projects, we cast the analysis within theframework of this methodology (Figure 7.2).

We have defined local economic growth from a transportation improvementas ‘the continuous increase in economic activity, in the impacted area, thatcan be attributed to this investment’. Given this definition the most importantconclusion from this chapter is that transportation capital investments donot necessarily generate economic growth benefits. In fact, under regularconditions they generate almost exclusively accessibility improvement benefitswith, perhaps, some environmental effects. Thus, attempts to measureadditional benefits, such as economic development, will amount to doublecounting of benefits. It is only when the analyst can demonstrate the existenceof certain allocative externalities that additional economic growth benefitscan rightly be ascribed to the investment. Such externalities include spatialagglomeration, labour market imperfections, transportation networkeconomies and environmental improvements.

Correspondingly, spatial relocation of land-use activities following aninfrastructure investment, even if shown to be engendered by the investment,cannot be regarded as economic growth. For such an effect to transpire theremust be a consequent increase in economic activity in the impacted areawhich can stem from agglomeration or labour market economies.

A related conclusion from the discussion in this chapter is that the primebenefits from a transportation infrastructure investment are accessibilityimprovements and that all other potential benefits, mainly economic growth,emanate from these principal benefits. Hence, all factors that can affect the


measured level of accessibility benefits from a project will also affect themeasured level of economic growth benefits. These factors include the valueof travel time, the discount rate used to compute the net present value of theproject, and risk and uncertainty elements which affect the magnitude offuture streams of benefits and costs. Without properly accounting for thesefactors, the measured accessibility benefits from a given project will be biasedand so will the measured economic growth effects. Furthermore, since therelationships between the prime benefits and potential growth effects arenon-linear, the erroneous measurement of accessibility benefits might produceexceedingly inflated growth benefits.

Finally, benefits from a transportation project can be correctly measuredonly if evaluated at the point of social equilibrium. For this to take place it isnecessary that the total social costs of the project will be borne by the users.Otherwise, without accurate knowledge of demand and cost functionelasticities it is not possible to ascertain whether the actually measured benefitsover—or underestimate the optimal ones.

Notes

1 The literature also uses the terms cost-benefit analysis (CBA or COBA) or socialbenefit cost analysis (SBCA).

2 For classical reviews of BCA the reader is referred to Prest and Turvey (1965)and Mishan (1969). For a treatment of BCA of transportation projects see Nash(1993) and Small (1999).

3 McGuire’s estimate is an average value for a large number of projects, some ofwhich may have generated output elasticity values that were well below or abovethis figure.

4 The Intermodal Surface Transportation Efficiency Act (ISTEA) in the USA requiresthat such analysis be carried out before projects can be approved for fundingand implementation.

5 It is beyond our scope here to examine these and related issues at any degree ofrigour. Suffice it to point out that beginning with Marshall (1920:811), Hicks(1943) and Friedman (1949) to more contemporary writers like Willig (1976)and Mishan (1988), the theoretical underpinnings of consumer surplus havebeen thoroughly examined. It should also be pointed out that interpersonal utilitycomparisons are done routinely at all levels including political decision-making,as a matter of carrying out social and economic public policy. For approachesdealing with income and price effects in the computation of consumer surplussee Freidman (1984: Chapter 5).

6 In many highway improvement cases, which aim at alleviating tight congestion,latent demand is high enough to make the facility after the improvement ascongested as it was prior to the expansion. Downs (1962) has pointed out that,this phenomenon notwithstanding, the project has generated real benefits tosociety in the form of more people performing trips that they wish to undertake.

7 Small (1999) calls them ‘technological externalities’.8 In a recent review Anas et at. (1998) have examined types of externalities that

affect urban spatial structures. They have categorized them as spatial non-homogeneities (i.e. factors affecting the uniqueness of locations), firms’ internal

198 Methodology

scale economies, external or inter-firm scale economies (regarded above asagglomeration economies), imperfect competition (e.g. spatial oligopoly).

9 Such imperfections may also prevail at the demand side where firms’ demand forlabour is constrained by the available pool of skilled labour which, in turn, isaffected by spatial accessibility.

10 This network effect can also work in the opposite direction. Known as the ‘Braess’paradox’, in some instances the addition of a new link to an existing networkmay actually increase total travel time. This phenomenon is mainly due to thefact that users minimize their private travel time rather than the system’s averageor system’s total travel time. If congestion is unpriced, the result of introducinga new link may be the diversion of traffic from a longer but uncongested link tothis new link with the overall result of slower traffic everywhere.

11 Interesting examples are the Scandinavian link and the Nordic link. TheScandinavian link is a transport corridor connecting Oslo, Gothenburg andStockholm with Malmo, Copenhagen and Hamburg by means of a four-lanehighway and a dual-track railway. The Nordic link is a transport corridor fromHamburg through the Danish peninsula to southern Norway. See Kristiansen(1993) for details.

12 Jara-Díaz (1986) and Jara-Díaz and Farah (1988) have shown that if the demandcurve for final goods in each region is linear and the two monopoly firms operateunder conditions of constant marginal costs, total transportation benefits, froman infrastructure expansion, are approximately half the size of the change intotal consumer surplus. The other half is attributable to gains in trade andreduction in prices of final goods.

13 See, for example, Small and Kazimi (1995) on air pollution costs and Viscusi(1993) on the economic costs of road accidents.

14 A quite unlikely result when considering metropolitan modal availability, networklayout, as well as political factors.

15 In fact, Gordon and Richardson (1994) have shown empirically that, as a resultof firms relocation to suburban areas and the formation of sub-centres over thelast two decades, average travel times and congestion levels have declined, eventhough no substantial investments in infrastructure facilites were made.

16 Using a multinomial logit formulation this expression, can be derived from theexpected value of utility maximization by an individual, i.e. E(Max.Uk), �k Î R,where Uk is total utility from alternative k (McFadden 1981).

17 Expression (7.3) is identical to the expected compensating variation (the amountof income a consumer must receive to leave his utility unaffected by the pricechange), or the equivalent variation (the amount of income a consumer wouldbe willing to forgo to avoid the price change).

18 For example, a widely used commercial network assignment model is ‘EMME2’.This model, once calibrated relative to parameters of the link’s volume-capacityfunction, computes minimum travel time over the network between all origin-destination pairs, given the demand matrix. Following a capacity improvementof a given link and using the previously calibrated parameters, the model cannotseparate between the additional traffic on the improved link from traffic causedby users switching between peak and off-peak periods, from newly generatedtraffic, or from traffic diverted from other destinations.

19 His estimate of overall network returns to scale is 0.84. It implies that if congestiontolls were imposed on all links of the network, revenue would recover 84 percent of total capital costs.

20 Sometimes the term value of travel time savings (VTTS) is also used.21 For a rigorous exposition the reader is referred to Ben-Akiva and Lerman (1985).


22 MVA report (1987:90–2), points out that the differences in VOT savings acrossmodes, in fact, may reflect self-selection by trip makers. Those with high valueof time will tend to select fast modes. The inclusion in the econometric modelvariables that can control for such effects may help overcome this problem.

23 For car passengers, Waters II (1992) found VOT to be only 35 per cent of thewage rate. He further recommended that VOT should increase with the trafficconditions and it would reach 100 per cent of the hourly wage rate for stalledtraffic.

24 Presumably, the equilibrium price of hauling cargo reflects cargo’s travel time. Inthat case the value of time savings to the cargo carried is already included in thetime savings to the driver and vehicle which are reflected in the price charged bythe shipper.

25 The other two approaches are, first, inference of VOT time from previous publicprojects that affect truck movement (e.g. willingness to pay tolls to travel athigher speeds); second, assessment of VOT from stated preference studies ofcarriers.

26 ‘The long run is a misleading guide to current affairs. In the long run we are alldead’ (John Maynard Keynes).

27 Since consumers are also the owners of firms, their investment in firms, whetherprivate or public, also reflects their substitution between consumption andinvestment.

28 The use of these rates assumes that competitive markets rates reflect the real netsocial return. See, for example, Boardman et al. (1996) for calculation of rate ofreturn of private investments.

29 Pigou maintained that consumers suffer from ‘defective telescopic faculty’,implying that they tend to over-emphasize present and near future periods.

30 See Engel et al. (1996) for data on the volatile distribution of the percentage ofvehicles paying tolls in selected cities in Chile between 1987 and 1995.

31 To illustrate and following Pindyck (1991), consider an example of a publicinvestment in a toll road. Assuming no operating costs, we denote the capitaloutlay by I. The investment is expected to generate annual revenues of B1 overthe project’s life span of N years. Assuming sunk capital costs, the decision toinvest today (t=0) is irreversible. We further assume that investment made thisyear will generate revenues the following year (t=1). Due to lack of informationregarding the impact of simultaneous investments in parallel and connectingroads there is uncertainty about future demand elasticities, hence over revenuesin subsequent years (t=2, 3, …, N). With probability p, revenues can rise to B2,(B2 > B1), but with probability of (1-p) they can fall to B3, (B3 < B1). Let r bethe project’s discount rate. What is the best course of action between the decisionto invest this year and the decision to delay the investment by one year?Obviously, if the difference between the NPV of the project postponed by oneyear and its NPV of investing this year is positive, the delay is economicallywarranted. Remembering that we invest next year only if revenues go up (whichwill happen in probability p), the rule is:

The term DNPV is the monetary value of the flexibility to invest at a future year ratherthan at present. Let I=$1,000; B1=$200; B2=$300; B3=$100; N=10; p=0.5; r=10 percent. It can easily be seen that for this DNPV=$154.4, illustration hence, it pays todelay the project by one year. Notice that this result will hold for every

200 Methodology

32 Assuming the marginal revenue function from the investment is convex in priceto users.

33 The new Transportation Equity Act (TEA-21), signed by President Clinton inlate 1998, includes the following as a key MPO factor: ‘Support the economicvitality of the metropolitan area, especially by enabling global competitiveness,productivity, and efficiency’ (Progress, 1998).


APPENDIX 7.1

Example: evaluation of transportation capacityimprovement benefits in the presence of anallocative externality

The objective of this example is to demonstrate the possibility of assessingbenefits from a transportation project while taking into account allocativeexternalities. To that end, consider a transportation infrastructure projectaimed at improving travel times and flow over a certain segment of the roadnetwork by expanding its capacity. From a social welfare viewpoint we ask:What is the benefit-cost rule which should be used in order to determine thelevel of optimal benefits and investment in the presence of congestion exter-nalities?1 Analytically, our objective is to determine optimal traffic volumeand capacity at the point where users pay the full marginal social costs oftravel, namely actual travel time and money costs plus the costs of congestionexternality. To facilitate the analysis we assume identical users with a fixeddemand function.

We use the following notation. Let, V be traffic volume; K be road capacity;P(V) is users’ demand function (travel price as a function of traffic volume);C(V, K) is users’ average cost function of travel; and F(K) is the capacityinvestment function. All variables are defined per one unit of road (1 km oflane road). The objective is to maximize a social welfare function W, withrespect to V and K. Notice that in this analysis we do not assume latentdemand so that P(V) is known with certainty. The social welfare function isdefined as:

(A7.1.1)

First order conditions are2

(A7.1.2)

(A7.1.3)

The expression: is the social marginal cost curve

whereas is total congestion costs as a function of K and V.

From condition (A7.1.2) we obtain the known result that at equilibrium user

202 Methodology

price P(V), should equal total social marginal costs which are composed ofaverage user’s costs and congestion costs, i.e.

(A7.1.4)

where P*, is the social optimal price that users need to pay in order tointernalize the congestion externality. From condition (A7.1.3) we obtain thecost-benefit rule for the investment. That is, the marginal dollar invested in

road capacity expansion, should generate benefits (measured in units

of travel time) that exactly equal the reduction in total congestion costs from

that investment, –

These conditions are graphically shown in Figure A7.1.1, where point E*represents social equilibrium and V* is optimal level of traffic, given optimalroad capacity. The reader should notice that the attainment of maximumsocial welfare requires these two conditions to hold simultaneously. Thus,the benefit-cost rule for capacity expansion applies when calculated for theoptimal volume of traffic V*.

The full price that users need to cover which equals social marginal costsat point V* is composed of the congestion externality, private users’ costsand the investment costs. To compute it, from (A7.1.2) and (A7.1.3) we obtain

(A7.1.5)

where l is the degree of homogeneity of the travel cost function in V and K.3

From (A7.1.5), to achieve a Pareto Optimum solution total payment by allusers, V*.P(V*) should equal the social costs (composed of total private travelcosts, plus congestion costs), V·C(V, K), plus total investment costs, F(K).Therefore, for this result to hold we must assume that l=1, and that there areno scale economies in capacity investment.4

The primary objective of the above discussion is to highlight the notionthat, from a theoretical viewpoint, a correct assessment of benefits from atransportation infrastructure improvement requires that the analysis be carriedout at the point of social equilibrium (point E* in Figure A7.1.1). The readershould notice that for equilibrium to be at point E* the value of the congestion

externality should be priced to users as a congestion toll whose

value is P*-PG per user (Figure A7.1.1). However, from a practical viewpoint,highway tolls, even if imposed, are rarely optimal relative to the time of day


and the congestion level at that period. Consequently, benefits from capacityexpansion are normally measured at a point of sub-optimal tolls or, which isthe more prevalent case, at market equilibrium where the demand functionP(V), intersects the users’ cost function C(V, K), (point V0 in Figure A7.1.1).The question of interest then is whether benefits from a given capacityexpansion, associated with market equilibrium solution are systematicallylarger than those associated with the first best (point E*)?

Some authors (e.g. Wheaton 1978) argued that the lack of road pricingwould lead to over-investment in highway capacity. The main reason beingthat without pricing more users will use the highways than is socially optimal,thus requiring greater capacity. Others, however, have shown that thisconclusion is rather contingent on the value of a number of key parameters.Wilson (1983) has shown that for discrete toll changes and under certainvalues of users’ demand elasticity with respect to traffic volume, Wheaton’sgeneral conclusion may not hold. The study by d’Ouville and McDonald (1990)has highlighted the role of the elasticity of substitution between the travel costfunction C(V, K) and capacity (denoted by s). They generally conclude thatthe smaller is s, the larger are the cost savings (benefits) from a given capacityexpansion.

An additional factor noted above is the elasticity of the investment functionwith respect to capacity (denoted by y), which can influence the results.

Given these factors, in general it is not possible to provide a definite answerto the above question (see Grunau 1994 for analytical exposition). However,

Figure A7.1.1 Measurement of benefits.

204 Methodology

if we can assume that travel demand is price inelastic, that (as estimated

by d’Ouville and McDonald 1990), that l =1 and y =0, then we can concludethat when moving from a regime of no tolls to that of sub-optimal ones tothat of optimal tolls, smaller capacity expansion investments will be required.If we further consider the fact that tolls are usually imposed on very few linksof a network, thereby causing traffic diversion to untolled links, over-investment in capacity can be expected, as was originally claimed by Wheaton(1978). It follows, therefore, that benefits from a transportation infrastructureproject are probably exaggerated if estimated under users’ market conditions.

Notes

1 See Small (1992: Chapter 2) for a similar analysis. For the derivation of suchrules at the network level see Yang and Huang (1998).

2 Second order conditions are assumed to hold, i.e.

3 The degree of homogeneity of a function f(x, y) is defined as, If, l =1, the function is said to be homogeneous of degree

one. If the average cost function C is homogeneous of degree 1, then the totalcost function is homogeneous of degree 2.

4 Analytically, y where y is the elasticity of the capacity investment

function with respect to the investment.


APPENDIX 7.2

Derivation of the value of time: a review andanalysis

Most theoretical approaches which underlie the derivation of the value of timeare descendants of Becker’s formulation (Becker 1965). In his model the utilityof a representative consumer is maximized subject to time and budgetconstraints. Subsequent developments by DeSerpa (1971, 1973), Mohring(1976), Bruzelius (1979) and Hensher and Truoung (1984) have introducedthe time spent in various activities directly into the utility function. To deriveanalytical definitions of the value of time we present a time allocation modelwhich contains two time categories: time spent at work and time spent inleisure activities (see also MVA 1987). Subsequently, we will introduce a thirdcategory, namely, time spent in travel. Formally, the time allocation problemfaced by a representative consumer can be modelled as follows (DeSerpa 1971,1973; MVA 1987; Small 1992):

max.U(tw, tl, z)tw, tl, z

(A7.2.1)

In this formulation zi is a consumption good whose price is pzi and z is avector of consumption goods. The term tw is time spent at work, and w is thewage rate (net of taxes). Hence the term: w·tw is earned income1 and y denotesunearned income. Let tl be the time spent on non-work activities. We furtherdenote by the total time endowment of individuals (normally, 24 hours,net of essential requirements like sleeping and eating),2 and by and and

the minimum amount of time allotted to work and leisure activities,respectively. In this analysis we assume that the level of consumption goods zis independent of the cost of travel and that it is exogenously defined. Theparameters. l , µ, q, h are the shadow prices of the constraints.

The first constraint in (A7.2.1) is the monetary budget whereas the secondis the time budget. The constraints: and imply that each timecomponent requires a minimum time duration3 (which may, of course, be zero).4

Essentially, in this model a consumer makes three simultaneous decisions. Hedecides about the allocation of income between consumption goods; the utilitymaximization quantity of each good; and the allocation of time between workand leisure activities. Setting up the Lagrangian of model A7.2.1 we get:

subject to:

206 Methodology

(A7.2.2) The Lagrangian multipliers l, µ, q, and h reflect, respectively, the marginalutility of income (earned and unearned), the marginal utility of total timeavailable , the marginal utility of changing the minimum working hoursconstraint and the marginal utility of additional leisure time. The first orderconditions are:

(A7.2.3)

(A7.2.4)

(A7.2.5)

By using the consumption good as a numeraire (and setting its price to equal1), we get:

(A7.2.6)

The left-hand-side of (A7.2.6) is the value of time (in undertaking a leisureactivity). If the leisure time constraint is not binding then this marginal

valuation of time , equals which is known as the resource value of

time (MVA 1987). If it is binding, , the difference, is referred to as

the value of time savings in an activity (DeSerpa 1971). It is also commonlyreferred to as the value of time, which is used to appraise transportationprojects. From (A7.2.6), the value of time-saving equals the resource value

of time , minus the marginal valuation of time spent for leisure activity

Dividing A7.2.3 by A7.2.5 and assuming that the work hoursconstraint is not binding (q=0), we obtain the marginal utility of timeused for work. That is:


(A7.2.7)

Expression A7.2.7 implies that when an individual is free to work as manyhours as he wishes, he will do so until the last unit of time spent at workapproximately equals his wage rate, minus the resource value of time.5

From the above first order conditions we get:

(A7.2.8)

If the minimum work hours and the minimum leisure time constraints arenot binding (i.e. q = 0, h = 0), then the left-hand side of equation (A7.2.8) isthe total reward for an additional unit of work time which, at equilibrium,equals the utility value from the last unit of leisure time foregone (the right-hand side). This value is, essentially, the opportunity costs of work time.

Notice that the above formulation does not include travel time as a separateelement of time. On the one hand it is possible to claim that travel time is justanother type of time spent in undertaking an activity similar to, say, leisure.The derivation of travel time value will then be identical to that of leisure

. On the other hand it might be argued that travel time differs from any

other type of time in a number of important ways. First, it confers a disutility,whereas the utility from time spent in all other activities is positive. Second,to a large extent, travel time is not fully controllable by individuals as it isaffected by exogenous conditions like road congestion, route layout, andfrequency of public transit modes and availability. Third, individuals canaffect their travel time by such transportation means as mode choice, time ofdeparture or choice of a route. Relocation is perhaps the most effective wayto control for travel time, though for many individuals it is prohibitivelycostly and in the short run non-feasible. Thus, given location, the question ishow to treat travel time in a utility maximization framework.

In the random utility literature Train and McFadden (1978) first took upthis question. Their objective was to examine how price and income canenter the direct utility function of mode choice. Truong and Hensher (1985)further investigated this issue within a framework of DeSerpa’s model. Theirobjective was to derive a simple form of the deterministic element of therandom utility model for the choice between travel alternatives (see also MVA1987). The main result is that the ratio of the shadow prices of the incomeconstraint, which include travel costs, to that of the constraint of travel timeassociated with a transportation alternative (e.g. a mode or a route) is thevalue of time saving spent in using that alternative.

It is possible to reformulate problem A7.2.1, to investigate the impact of

208 Methodology

change in accessibility on travel time where accessibility is a combination oftravel time and costs. Thus:

max . U (tw, tl, z, t(d))tw, tl, z

(A7.2.9)

In this model t(d) and F(d) are the time and costs of travelling distance d,respectively. Notice that the money cost of travel, F(d), is introduced directlyinto the budget constraint (see also Mohring 1976). Following the sameanalysis as above we obtain:

(A7.2.10)

This condition indicates that the marginal utility of travel time equals the(negative) ratio of the shadow price of the income constraint l weighted, tothe shadow price of the time constraint µ. The weight is the change in themarginal cost of travel from a marginal change in travel time, which we haveregarded as accessibility change.

Assuming that the minimum leisure time constraint is not binding, we get:

(A7.2.11)

The left-hand side of A7.2.11 represents the total value gained from anadditional travel or distance. It is composed of the change in utility fromtravel time plus the accessibility change during that time period. As in A7.2.6,the total value of the last unit of travel equals the value of the last unit ofleisure time foregone.

Notice that conditions which are internal or external to the individual canaffect his subjective valuation of time. Examples of internal conditions arepreferences structure, level of earned and unearned income and/or proximityto work location. On the other hand, the wage rate, the money cost of travel,the price of the consumption good and traffic conditions are external factors.Hence, the same individual might change his valuation of a unit of traveltime saved, depending on the work and travel environment he faces.

subject to:


Notes

1 Notice that we do not assume here that the hourly wage rate (w) depends on theamount of time spent at work (tw) as some studies do (see, for example, Ramjerdi1993). Similarly, we do not assume that the wage rate is location dependent, i.e.w(d); (see Gunn 1991). Adding such assumptions would not alter the mainconclusions arrived here in any fundamental way.

2 The choice of 24 hours implies that other variables in the system, such as earnedand unearned income, should be proportionate to this time period.

3 In reality, these time constraints might be a function of other factors. Here weassume them to be constants and exogenous.

4 This characterization of the model is fundamentally different from Becker’s (1965)household activity production model mainly in that here we explicitly distinguishbetween time as a commodity (which enters the utility function) and time as aresource which is subject to a resource constraint (see DeSerpa 1971).

5 By definition, the unit of time of 24 hours is always binding so that strictlyspeaking µ> 0.

A model of transportinfrastructure developmentand local economic growth

8.1 Introduction

The fundamental rationale of economic growth from transportationinfrastructure developments is that the primary effects from these investmentsare improved travel times and volumes over specific links of the transportnetwork (Chapter 7). These factors then affect the relative accessibility oflocations within the impacted area. These accessibility changes can potentiallyencourage economic growth, provided that market externalities, mainlyproduction and labour market economies, are present. We have alreadyexamined the effect of capital accumulation on national and state growth(Chapter 6). In this chapter we focus on local economic growth emanatingfrom transportation improvements, mainly with regard to the use of labourand labour productivity.

In general, changes in the relative accessibility from facility developmentin a particular area can generate two potential effects. They can bring aboutactivity relocation and they can cause firms and households in the impactedarea to modify their production and consumption schemes. We regard thesechanges as ‘local economic growth’, indicating the increase in local output orin output per capita, and in the use of input factors (mainly labour) and infactor productivity1.

Empirically, in assessing local economic growth arising from a particulartransport infrastructure investment, it is rather a complicated task to separatechanges in activity location from changes in the production and consumptionschemes of firms and households. Theoretically, however, we may wish toidentify the conditions for economic growth spurred by activity relocationand those generated by changes in the production and consumption behaviourof firms and households. In either case we need to define the mechanism bywhich changes in accessibility are transformed into local economic growth.Underlying this analysis is our contention that growth arises primarily fromthe ability of firms and households to exploit positive externalities, such asagglomeration economies, or lessened negative ones, such as alleviating trafficcongestion. Thus, we explicitly assert that the presence of such non-internalized

8

212 Methodology

externalities constitutes a necessary (though not a sufficient) condition forlocal economic growth emanating from infrastructure development. Themodel presented in this chapter is based on this assertion.

The structure of this chapter is as follows. In Section 8.2 we present atheoretical framework within which we investigate key structural elementsthat underlie the relationships between infrastructure investment and localeconomic growth. Subsequently, in Section 8.3, we formulate a simple modelof the spatial economy in which firms operate under conditions ofagglomeration economies, and households that supply labour input, are subjectto travel congestion. Through simulations (Section 8.4), we examine theimpacts of transportation network capacity improvements on local economicgrowth in terms of the equilibrium level of employment and labourproductivity. In Section 8.5 we briefly present results from an empirical analysisof the relationships between accessibility improvements and labour forceparticipation, using real data from the Bronx in New York. Major conclusionsare presented in Section 8.6.

8.2 Theoretical framework

8.2.1 Determinants of transport development and localeconomic growth in a spatial economy

As noted in Section 7.2.2, major explanation advanced in the literature forobserved differences in urban spatial patterns, the level of economic growthand labour productivity across cities, is agglomeration economies. The conceptof agglomeration economies and its use in the analysis of firms’ location,industrial clustering and the formation of urban activity centres, has a longtradition in regional science and urban economics (see, for example, Isard1956; Weber 1956, Chinitz 1961; Beeson 1992; Selting et al. 1994; Anas etal. 1998). In contemporary urban economies, industrial agglomerationeconomies are manifested in the forms of forward linkages (firms interactwith customers), backward linkages (firms interact with suppliers) andsideways linkages (firms interact with each other).2

For the present analysis we regard agglomeration economies as the casewhen average costs decline as more production takes place within a specificgeographical area. It basically arises from positive technological and pecuniaryexternalities that emerge when economic agents (firms) locate in close spatialproximity (Anas et al. 1998). In general, such externalities include a Smithianlabour specialization, technological innovation and diffusion and humancapital accumulation. Possible factors that underlie their emergence areinformation spillovers, the use of a common infrastructure facilities (such asenergy, communication and transportation), and access to a common pool ofspecialized labour force. Whatever the exact reason, these effects stimulateproductivity gains and bring about a reduction in production costs.3

Transport development and local economic growth 213

In attempting to explain the extent and impacts of agglomerationeconomies, some researchers have resorted to the concept of positive feedbackeffects in the local economy that give rise to spatial clustering of activitiessuch as employment and shopping centres (see, for example, Arthur 1991;Krugman 1991b). The presence of such feedbacks can amplify the impact ofchanges in the activity level of one economic entity on the production and thelocation decisions of others. These effects, in turn, induce economic activitiesto cluster in groups of varying sizes, in order to reduce interaction andproduction costs. The sizes of such clusters are affected by limiting forces,mainly land scarcity and transportation congestion.4

Four key factors need to be recognized in analysing the effect ofagglomeration economics on growth. First, that agglomeration economies,as production externalities, can be external to the economic entity (the firm)that produces it. Thus, agglomeration economies reduce the industry’smarginal costs at a given level of output, even if each firm operates atdecreasing returns. This is a rather important observation as it implies thatthe analysis is not confined to the case of declining cost industries such aspublic utilities. Second, that in order to determine the benefits fromagglomeration economies, they must be expressed in measurable units.Changes in activity density, in per capita income, in the unit cost of production,in output-to-input ratio, or in the variety of goods and services produced inthe region are examples of such quantifiable units. The third factor to recognizein assessing the effect of agglomeration economies is that their exploitationimplies some spatial proximity between firms, and therefore their levelattenuates over space. Thus, the correct analysis of agglomeration should becast within a framework of a spatial economy. Finally, the derivation of benefitsfrom agglomeration economies requires some spatial interaction, which inturn implies costs of interaction (e.g. travel costs). If these costs are ratherextensive relative to the agglomeration effect, spatial clustering and economicgrowth may, in fact, not transpire.

Given this perspective, how can we model the impact of accessibility changeson the local economy? Borukhov and Hochman (1977) have formulated amodel in which all locating activities wish to interact spatially with each other.This behaviour gives rise to agglomeration economies and the creation of activitycentres. In general, the formation of activity sub-centres is explained on thebasis of the interrelationships between the positive externalities emanating fromthe concentration of production and service activities and the negativeexternalities associated with the costs of spatial interaction. Harris and Wilson(1978) have shown that the trade-off between scale economies in location andtransportation costs can result in the emergence of multiple activity centreswhose number depends on the relative strength of these forces. Papageorgiuand Smith (1983) have further shown that the distribution of activityconcentration over space will become non-uniform when agglomeration

214 Methodology

economics (defined as positive externalities from co-location) are strongrelative to the costs of spatial interaction.

It should be recognized that the pattern and volume of inter-nodal travel,carried out over a transportation network, is the physical manifestation ofspatial interaction. As a consequence, the actual costs of spatial interactionare affected, inter alia, by the capacity and other attributes of this network. Itfollows that observed activity concentration, given the degree of agglomerationeconomies, can be affected by network development. Furthermore, since totaloutput in the economy (as well as the demand for labour) is affected byactivity location and concentration patterns (e.g. by sub-centring), theexpansion of the transportation network can affect the level of economicactivity through its effect on travel costs. Since travel costs also affecthouseholds’ disposable incomes and their allocation of time between labourand leisure, infrastructure expansion can further affect households’ supply oflabour, thereby the equilibrium amount of labour used in the economy. Insummary, when formulating a model to show the effects of infrastructuredevelopment on local economic growth, it is necessary to take account of allthese factors. To reiterate, the key ones are agglomeration economies, networkstructure (including the costs of spatial interaction) and the response of firmsand households to the presence of production, labour and transportationexternalities.

8.2.2 Measures of economic growth

Before presenting our modelling approach, it is useful to examine alternativemeasures of growth. We cluster them into four main categories: 1 Conventional firm-related real growth measures5 : these include changes

in output-to-input ratio, changes in partial and full factor productivity,changes in the amount of input factors employed (mainly labour), andchanges in the firm’s technical and cost efficiency.

2 Individual or household related measures of economic growth: these entailan increase in individuals’ utility relative to their consumption andopportunity space. They include changes in the size of the job marketarea, changes in the number of non-work related spatial opportunities(e.g. shopping outlets) and changes in the amount of time allocated toleisure activities. Under certain conditions (see below), the willingness ofhouseholds to increase their supply of labour is another indicator of localeconomic growth. Increases in consumption, following an infrastructureimprovement, ceteris paribus, can be used as another indicator of growth.This latter effect results from the capitalization of the investment’s benefitsin the form of enhanced consumer surplus (Anas 1995).

3 Technology-related growth measures: basically these reflect theincrease in the use of technologies, which are complementary to


traditional transportation effects, following infrastructure improvements.Changes in business production strategy, such as just-in-time production,increased intermodality in freight movement and improved access to majorregional facilities like airports, are all examples of such measures. In thisregard, if telecommunications are indeed a complementary technology toconventional transportation as some studies argue (Plaut 1997), then therate of proliferation of such technologies can also be used as a measure oflocal economic growth from transport infrastructure improvement.

4 Market-related growth measures: these are actually a combination ofthe above measures. They include indicators such as the level ofequilibrium employment, income per capita, the product range and thenumber of new firms coming into the market (the region or the city).

The main growth indicators used in the present analysis are the equilibriumemployment level and labour productivity. Other measures include theallocation of time between work and leisure and changes in the demand forthe consumption good. In terms of Figure 8.1 (p. 226), our primary growthvariables are changes in firms’ demand for labour, in households’ supply oflabour, and changes in their consumption patterns, following improvedaccessibility from an infrastructure capacity investment.

8.3 A model of a spatial production economywith transportation infrastructure effects

In this section, we formulate a model of a spatial economy with households,production firms and transportation infrastructure as its main components.Beginning from a state of equilibrium, by changing the capacity of thetransportation system, we can observe changes in the equilibrium levels ofthe economic growth measures outlined above.

8.3.1 Structure of model

We assume an urban economy which is comprised of three principal sectors: • a households’ sector, which supplies labour units and consumes leisure

and consumer goods;• a production sector represented by locating firms, which produce non-

homogeneous output for external markets and use labour and privatecapital as their main inputs;

• a transportation infrastructure sector, which is represented by aninterzonal transportation network.

The physical capacity of this network affects the level of congestion and, asexplained shortly, the amount of labour, which households are willing tosupply. Space is divided conventionally into locational zones.

216 Methodology

Beginning with the production sector, we assume non-homogeneous firmswith respect to their output type (i.e. they produce differentiable goods). Labour,on the other hand, is assumed to be a homogeneous input factor composed ofunits of time. We further assume that outputs are sold in export markets sothat the price of each is determined exogenously. The production and locationdecisions of firms, as well as their demand for labour and the amount of labourthat households supply, are governed by three primary forces—agglomerationeconomies, leisure and work time substitution and travel congestion. It isexplicitly being assumed that the level of output of each firm and their averagecosts are positively affected by agglomeration economies, which induce thesefirms to locate in close proximity. However, as firms move closer, the level oftravel congestion also grows, as a result of more trips being concentrated in asmaller area. As a result, households will alter their allocation of time by reducingthe amount of time spent on all activities, including labour. In this model wedisregard the costs of firms’ relocation.

The main role of the households sector is to supply labour for productionpurposes and consume leisure and non-leisure type commodities. Therefore,we assume homogeneous consumers who trade off leisure with work time.We further assume a fixed residential location relative to employment location,though in this model firms are free to move in order to optimize their locationrelative to agglomeration economies. Individuals travel between theirresidential and employment location zones and purchase a consumption goodfor which they pay using their income from work.

The third component in the model is the transportation sector, whichrepresents the product of government’s decision-making regardinginfrastructure provision, financing and regulation. In Parts I and II we discussedin detail the role of the public sector in the supply of transportationinfrastructure. Here, we merely point out the fact that by increasing thecapacity level of the fixed infrastructure through transportation investments,congestion is alleviated and a new spatial and production equilibrium isachieved. While this public supply is not cost free, we disregard investmentcosts, but we assume that the addition of new infrastructure facilities or theexpansion of existing ones, is welfare improving. We also disregard the moneycosts of travel that commuters pay which (in part) may cover the investment’scapital and operating costs.

Before presenting the analytical structure of the model, it is worth notingits limitations. First, we disregard the actual time it takes the land use andtransportation systems to achieve equilibrium. The model essentially carriesout static equilibrium comparisons, where the equilibrium level of employmentand productivity is compared with the previous level before a capacity-increasing investment took place. In reality, the effect of infrastructureexpansion on growth usually takes several years to evolve. In several placesin this book (e.g. Chapters 5 and 7, and the various case studies presented inPart IV), we examine this issue.


A second limitation of the model is that it assumes an exogenous demandfor output produced by local firms, which is independent of localconsumption. This may not hold if firms produce primarily for local marketsand are strongly affected by local demand. We treat demand and price ofoutput as completely determined by external markets. The underlyingassumption is that the local economy is open and small enough relative tothe national economy. Making firms’ output dependent on local demandwill mainly complicate the analysis without adding any significant insightto the qualitative results.

A third caveat is that in this analysis we basically disregard the actual useof land by firms, but more importantly by the household sector. We do notconsider residential location changes following changes in accessibility. Whatthese assumptions imply is that residential location choices, land consumptionand travel to non-work activities (e.g. shopping) are all unaffected by changesin travel times following infrastructure improvements. In a general equilibriumframework, changes in accessibility from transportation infrastructureimprovements should also influence residential location, land consumptionby households and firms and travel to non-work activities. Hence, this modelproduces only partial equilibrium results. However, given the scope of thisanalysis these results are judged sufficient to explain the effect of infrastructureimprovements on local economic growth.6

Finally, we do not account for the method of financing infrastructureinvestments, which can undoubtedly affect the model’s equilibrium solutionand the measured economic growth effects. If, for example, infrastructure isfinanced through consumption taxes (e.g. a sale tax), it may alter households’allocation of time between leisure and consumption. On the other hand, iftaxes are levied on firms (e.g. a payroll tax), it may affect their demand forlabour. The implicit assumption made here is that infrastructure investmentsare financed from external resources (e.g. through grants) provided by thenational government.

8.3.2 Model formulation

We consider an economy with three main sectors. A production sector, ahouseholds sector and a transportation sector. In this section we formulatethe structure and behaviour of each of these sectors, their spatial interactionand the organization of the economy.

The production sector

Let the index r or r’ (r, r’=1, 2), denote a firm type and the indexes i and j (i,j=1, …, N) denote spatial locations. Thus, for example, yr,i , denotes the outputof firm r in location i. For simplicity in this analysis we assume a two firmeconomy, i.e. r=1, 2. Let y1, y2 be non-homogeneous outputs of these firms.

218 Methodology

We assume that they are sold in competitive external markets so that the twofirms face horizontal demand functions with prices p1, p2, respectively, whichare independent of each firm’s location and output level.

The production of these outputs requires the use of three inputs: labour (l)supplied by households, private capital (k), and land (x). We assume the firms tobe price takers in factor markets. Given the prices and level of output demanded,the real wage rate and capital costs, and the land rent (at each location), the firmsdecide on their level of output and use of labour and capital inputs. Their objectivefunction is profit maximization subject to their production technology.

In this model all activities, including labour, are measured in units of time(i.e. daily man hours worked per worker). Total labour (in time units) availablein the economy, l

-, is divided between three main uses: labour used by firms 1,

labour used by firm 2, (l1 and l2, respectively), and labour used for consumptionand leisure purposes, le (i.e. the time that households allocate for these activities).

To simplify the analysis, we assume land rent and lot size to be constant,and independent of firms’ location. That is, in the simulation analysis weassign each land unit at each zone i, a fixed land price (px), which a firm takesas a given. One obvious extension of the model is the inclusion of a landmarket in which equilibrium land prices are a function of location decisionsby firms and households.

We assume production to be affected by inter-firm agglomeration economies,so that the level of output of the firm in site i affects production by the firm inj. This agglomeration effect is further assumed to be spatially dependent in thata closer proximity between the firms confers a positive effect on each firm’slevel of output. As we will see shortly, the ability of one firm to benefit fromlocating closer to the other is a major force that drives the model. Denoting byf1 and f2 the production functions of the two firms, it is assumed that

; , which is interpreted as agglomeration economies,

given the firms’ location sites in i and j. Let (yi¹j, dij), be the degree of

agglomeration between the two firms, which also depend on their relativeproximity, di,j (see shortly). Thus, the output level of each firm is given by thefollowing production function:

(8.1)

(8.2)

The constants A1, A2 represent the technology used. Notice that unlike theproduction function formulations described in Chapter 6, here the firms arenot assumed to include public infrastructure as a specific input factor in theirproduction function. Rather, the effect of infrastructure expansion enters the


analysis through its impact on the level of the inter-firm agglomerationeconomies and on households’ home-to-work travel time, which is shown toaffect their willingness to supply labour.

The overall agglomeration effect (yi¹j, dij) is comprised of two components.

First, a ‘fixed component, denoted by “g”’, which represents the particulartype of inter-firm agglomeration economies, due to the firms’ specific product

mix, and which is independent of their location. We assume 0 and

that g is symmetric.7 The second component is the spatial separation (e.g. thedistance) between the two firms, dij. Hence, the overall agglomeration effectattenuates nonlinearly, as spatial separation between the firms, increases.Thus, total agglomeration economies are written as:

(8.3)

where g is the spatial decay factor. Using Equation 8.3, each firm’s productionfunctions is:

(8.4)

where a, ß, s are, respectively, the labour, capital and land elasticityparameters. Since we do, assume no scale economies in production, a + ß =1.Notice that if g=o, (i.e. no product-mix specific agglomeration economies),the overall agglomeration effect is completely ineffective. On the other hand,if the spatial friction factor(g) is very large (and g > 0), the agglomerationeffect declines.Each firm’s profit functions is given by:

(8.5)

Each firm is assumed to maximize its profits (Equation 8.5) subject to itsproduction function (Equation 8.4), thereby deriving the optimal output level.As already explained, in this analysis we disregard the impact of land marketon production by assuming an inelastic demand for land and fixed lot sizes,irrespective of land price, which are exogenously determined. Thus, in whatfollows we set the land elasticity parameter, sr (Equation 8.4) to zero.

Still another important assumption in this model is the Cournot-Nashbehaviour of the production firms. That is, we assume that each firm takesthe location and output level of the other as fixed and, given the marketdemand for its output and the costs of input factors, chooses its location andoutput level accordingly.

In general, the labour demand function, of each firm, is:

220 Methodology

(8.6)

More specifically, maximization of Equation 8.5 with respect to the amountof labour input employed and subject to the production function (Equation8.4) yields the following demand function for labour:

(8.7)

From Equation 8.7, as expected, more labour demanded is associated withgreater output, i.e.:

. More interesting, however, is to observe the effect of

agglomeration8. Thus, labour demanded by firm r is an increasing function

of the level of output of firm’s r’. That is,

Similarly, the demand for capital,kdr is a function of the output level, the price

of capital and the agglomeration effect. Specifically:

(8.8)

The household sector

In this model each household located in each zone i is assumed to have oneworker who provides labour inputs, defined as units of time spent at thework place. Hence, total labour supply, ls, is lr=1

+ lr=2 = ls In addition, householdsconsume a consumption good, z, and non-work time. The latter is composedof leisure time, denoted by le, and time used for home-to-work travel, lT.Total time available in the economy, denoted by is:

(8.9)

Consumers, are assumed to maximize utility defined over z and le. Forsimplicity, the utility function is defined as Cobb-Douglas and it is maximizedsubject to budget and time availability constraints.9 That is:

(8.10)

where pz is the price of the consumption good and w is the wage rate (assumeduniform across all households). The taste coefficients µ1 and µ2 are assumed


identical across individuals and they measure the (constant) proportion of netincome spent on le and z. Also, we assume a utility function, which is homogeneousof degree one, i.e. µ1+µ2=1. The term U0 = euij implies constant idiosyncratic tasteof each home-work location pair {ij}. The monetary budget constraint (in Equation8.9) indicates that leisure time and travel time are purchased at the opportunitycosts of time which, in the present case, is the wage rate.

The solution to Equation 8.10 yields the demand functions for leisuretime (le) and for the consumption good (z), expressed as functions of thewage rate the price of z, and the shadow price of time. These functions havethe attributes of unitary own price and income elasticity and zero cross-priceelasticity. That is:

(8.11)

(8.12)

where U is the present utility level, and l and j are the shadow prices of thebudget and time constraints, respectively. From Equation 8.11, given the utilitylevel and the constraints, the amount of leisure time demanded is an increasingfunction of the propensity to consume leisure (µ1) and a decreasing functionof the wage rate (which is the alternative costs of work time) and the (shadow)price of time. Similarly, the amount of the consumption good demanded is anincreasing function of the its taste coefficient (µ2) and an inverse function ofits price.

From 8.11, given the amount of leisure time households wish to consume,we can compute from Equation 8.9 the amount of labour (in time units) theyare willing to supply, as a function of the amount of time used for home-to-work travel. As travel time increases (see shortly), households will substituteleisure time for labour, where the key substitution factors are the alternativecost of work time (the wage rate) and the leisure time elasticity parameter, µ1.In general, given the wage rate and the amount of leisure time demanded, theamount of labour supplied by households will decline as more time is usedfor travel to work. Conversely, as accessibility improves, households will bepredisposed to supply more labour. Moreover, additional benefits fromimproved accessibility might be capitalized in the consumer surplus andmanifested in increased consumption of the consumption good. The relativesize of µ1 and µ2 will determine the magnitude of this result.

The transportation sector

As already noted in this urban economy, the production of outputs (y1 and y2)entail travel of employees between their zones of residence and zones of

222 Methodology

employment where the firms locate. To simplify the analysis, we assume that thelevel of commute is a fixed proportion of the level of labour used for production(e.g. two trips per day per employee). We further assume a generic mode oftransport (e.g. private car)10 and an in-place transportation network with fixedcapacity. As firms locate at closer proximity so that they can benefit from enhancedagglomeration economies, more congestion will ensue. That is so for two mainreasons. First, more commuters now travel to the same area (e.g. the CBD),resulting in higher traffic volumes sharing the same fixed transport infrastructurefacilities. Second, because increased agglomeration results in higher levels ofoutputs that are then translated into increased demand for labour and travel. Ofcourse, both effects can transpire concurrently and, as noted above, the finalresult might be less time available for work activities. These relationships betweentotal travel time for work purposes between locations i and j, the level ofemployment used by firm r(r=1, 2), the capacity of the transport infrastructurefacilities and the distance between sites i and j, are given by Equation 8.13:

(8.13)

where tij measures total travel time between a residential site in i, andemployment site j. The amount of labour in site in j (j ¹ i) actually employed byfirm r, is (lr)j. The fixed carrying capacity of the transportation infrastructurebetween each ij pair is denoted by Kij. From Equation 8.13, the effect ofexpanding the capacity of the transport infrastructure is to reduce travel time,

i.e. On the other hand, travel time increases as the number of trips

increases. As explained earlier, the number of trips is a constant multiple of theemployment level, which, in turn, is a positive function of output level. Hence,the effect of increased output on home-to-work travel time is positive, i.e.

(r=1, 2). For each firm r (r=1, 2), an explicit form of Equation 8.13 is:

(8.14)

where r1 and r2 are parameters of the volume-capacity function (the term inthe square parenthesis). The variable lr,ij is the actual amount of labour usedby firm r(r=1, 2) located in site j, which travels from residential location insite i. We assume two trips per employee per time unit (e.g. a day). Thisvolume-capacity function is expressed in units of travel time per unit distance(e.g. 1 km). Hence, to compute total travel time we multiply this function bythe actual distance travelled between i and j, dij.

The reader should notice that Equation 8.14 might measure travel timesbetween an ij pair inaccurately. If some trips made by workers employed by


firms 1 and 2 share the same infrastructure facilities (e.g. use the same links),this may occur even if the two firms do not locate at the same site. To correctlymeasure tij we need to introduce a separate network sub-model, which in thepresent model framework is not done, mainly in order to avoid needlesscomputational intricacies that a network modelling entails. For this reasonin the actual computations of travel times to work we make the assumptionthat both trip types exactly share the same infrastructure facilities as if bothfirms are located at the same site j.11 That is:12

(8.15)

We define lT (equation 8.9) as: Sijtij Given this qualification, Equation 8.15 iscentral to the model. Travel times affect the allocation of time between leisureand work and, as a result, the equilibrium amount of labour and the wagerate. Travel times, on the other hand, are affected by the distance workersneed to travel from place of residence to place of employment, by the numberof work trips and by road capacity. The latter is an exogenous variable thatcan therefore be used to affect the equilibrium labour (and output) in theeconomy.

Operation of the Model

The overall structure and operation of the model is depicted by Figure 8.1.We begin from an assumed initial state of known output prices (which are

assumed exogenous to this economy), quantity of output demanded, thusdemand for input factors, and input prices depicted in Figure 8.1 as theproduction sector (I). We also assume a known distribution of householdsrelative to location and number of employees (the initial supply of labour inthe economy, l

-), titled as the household/labour sector (II). Lastly, we assume

a given transportation infrastructure sector (III), which generates an initialOD accessibility matrix. Given these components the model’s operation isdescribed by the following steps.

Step 1: Compute optimal firms’ location. From the above initial condi-tions, the model first determines each firm’s best location, defined as theone that maximizes agglomeration effect (Equation 8.3). Notice that atthis initial stage we assume dij > 0 , so that given the value of thefriction factor g the two firms do not necessarily locate at the same zone.Step 2: Compute optimal firms’ output. Given the firms’ location andlevel of agglomeration, the model now computes each firm’s optimumoutput by maximizing the profit function subject to the productionfunction (Equation 8.5 subject to Equation 8.4).Step 3: Compute demand for labour. Given each firm’s optimum level of

224 Methodology

output, next the model derives their demand for labour, (Equation 8.7),by maximizing their profit functions (Equation 8.5), subject to theirproduction function (Equation 8.4), with respect to the input labour whoseinitial price is assumed.Step 4: Compute equilibrium amount of labour. In this step the modelcomputes the equilibrium amount and price (wage rate) of labour (w),by equilibrating the amount of labour demand by firms with thesupply of labour, which in the first run of the model is the initial availabilityof labour in the economy

Figure 8.1 Schematic view of the relationships between the production, household andtransportation sectors.


Step 5: Compute interzonal travel times (level of congestion). Having thelabour equilibrium quantity, the model determines travel times tij thatthis amount of labour will generate (Equation 8.15), given thetransportation infrastructure.Step 6: Compute the optimal amount of leisure time. Having computedthe wage rate and the amount of time used for travel purposes (lT),subsequently from the consumers’ utility function (Equation 8.10), themodel obtains the optimum amount of leisure households desire toconsume, le, and the consumption good (Equations 8.11 and 8.12).Step 7: Compute general equilibrium in the economy. Rewriting Equation8.9, we obtain the labour supply function ls, expressed as a function oftotal time available, leisure time and congestion. That is:

(8.16)

where lT = Sijtij. Both the labour demand function (Equation 8.7), and thelabour supply function (Equation 8.16) are functions of labour units (l) andthe wage rate (w). Hence, we can solve for the labour market equilibrium.Let l* and w* be the solution variables. If the value of a particular {l*, w*}pair maximizes each firm’s profit function (Equation 8.6), then we regardthe economy to be in a state of equilibrium. Otherwise, the values of l* andw* become the new initial values and the process repeats itself from Step 1 toStep 7, till equilibrium is reached. (For a proof of existence of an equilibriumsolution, see Berechman 1994).13

The key question that this book sets out to examine is: Can the expansionof the transportation infrastructure influence local economic growth? Withinthe framework of the above model economic growth is expressed primarilyin terms of changes in the equilibrium amount of labour used in the economy.This amount is a function of the demand for labour inputs by firms and thesupply of labour by households, where both labour demand and supply areaffected by transportation costs. The former is affected through the impactof these costs on firms’ location and the latter is affected through the impactof travel costs on individuals’ decisions regarding their allocation of timebetween work and non-work activities. By comparing the equilibrium levelof employment before and after an infrastructure capacity improvement wecan ascertain the degree to which it has indeed enhanced employment. Wecan thus examine these relationships for successive transportation capacityimprovements to identify the consequent pattern of changes in employment.Changes in labour productivity, defined as the equilibrium level of output tolabour ratio, is another measure of growth used here. Next we describenumerical simulations with this model.

226 Methodology

8.4 Simulating transport infrastructureexpansion and economic growth

The principal objective of the simulation exercise is to assess the impact oftransportation infrastructure expansion on the equilibrium level ofemployment. We begin with a description of the various components of thesimulated system.

8.4.1 Input data

Spatial organization

We consider an urban area, which is a corridor or a line segment, dividedinto discrete i, j zones (i, j=1, …, N; N=5) as in Table 8.1. Households arelocated in i=1, while firms 1 and 2 (also denoted by their output type y1, andy2), are initially located at j=3 and j=5, respectively.

The production sector

The production function used in the simulation is similar to that given byEquation (8.4). That is:

(8.17) The parameter values used for these simulations are as follows: A1 = A2 = 0.5;ar + br = 1; ar=1 =0.7; ar=2 = 0.6 The land elasticity parameter, sr, assumedzero.14 The initial value of the agglomeration effect, g, is set to 2.5. Thedistance decay parameter g is set to 2.0. The exogenous prices of unit outputp1=1.2 and p2=1.3. The cost of capital is also assumed to be exogenously set,pk,r=1= 1.5, and pk,r=2 = 1.3.

Given these input data, the firms first determine their best locations.Afterwards, they maximize their profits with respect to output (as in Equation8.6) subject to its production function. Subsequently, we derive the demandfor input factors, labour and capital.

The household sector

We assume five households, located in zone 1, each endowed with 24 units oftime. Hence, total time available in the economy l

- = 120. The value of time

Table 8.1 Location of households and firms in the simulated area


($/hour) is set to be 60 per cent of the wage rate (v = 0.6w) which initially isset to $10.0/hour. Thus, v=6.

The utility function (Equation 8.10) contains two taste coefficients m1, m2.For the simulations we set them to equal 0.4 and 0.6, respectively (the utilityfunction is assumed homogeneous of degree one). The initial amount of theconsumption good is: z=9, and its price is pz=10.

The transportation sector

In this spatial configuration the transportation system is a linear networkwhere distances are measured between zonal centroids. The size of each cellis assumed to be 1×1 km. Thus, the distance from households’ location in i=1to the firms’ location in i=3 and i=5 are 2 km and 4 km respectively. Thus, theinterzonal travel distance matrix is symmetric and is as follows:

As already explained, in this model we do not have a formal network.Therefore, we consider the capacity of the route connecting any two zonalcentroids, Kij. We set it to equal 10 for all ij pairs and increased it incrementallyafterwards. To compute travel times we use a volume capacity function likein Equation 8.15 with the parameters r1, r2 set to 0.15 and 2.5, respectively.Thus, the explicit travel time function used for the simulations is:

(8.18)

From the data given above and from Equation 8.18 the initial travel timematrix can be computed. In order to convert travel times to travel costs weuse:

8.4.2 Simulation results

Using the above input data and parameters, we have performed a large numberof simulations to assess the effect of infrastructure developments on theequilibrium of labour in this economy. As in most simulation analyses, theactual numerical results are largely susceptible to the parameters’ valuesadopted and to the initial conditions (see for example, Anas and Kim 1996).

Table 8.2 Interzonal travel distance matrix (di,j)

228 Methodology

Figure 8.2 shows the impact of an incremental increase in the capacity of thetransportation infrastructure on the equilibrium level of

employment and output.In Figure 8.2 point E1 represents the intersection of labour supply, defined

also as a function of the travel time, with the demand for labour by firms,which is defined as a function of output and inter-firm accessibility. Sincehome-to-work travel time and inter-firm accessibility are both a function ofthe capacity of the interzonal transportation system (Kij), we can express theequilibrium amount of labour in the economy also as a function of thiscapacity. Point E2 represents a new equilibrium level following the expansionof the transportation capacity, from Kij to It can be seen that

E2 is associated with higher levels of labour l* and output yr* than the levelsrepresented by point E1.

The next question is how will continuous increase in transportation capacityaffect the equilibrium amounts of employment? To answer this question wehave systematically increased the capacity of the interzonal transportationsystem by steps of 0.1 (i.e. Kij=10, 10.1, 10.2, …14.9, 15) up to 50 percent.By rerunning the simulation model for each level, we have computed theconsequent equilibrium level of employment. The numerical results aregraphically shown by Figure 8.3.

The results of these simulations indicate that as capacity increases so does

Figure 8.2 Equilibrium solutions (E1, E

2), before and after increase in transportation infrastructure

capacity (K, K’; K’>K).


equilibrium employment. However, this effect abates quite rapidly.15 Beyonda certain level further expansion of the transportation infrastructure capacitywill have no measurable effect on equilibrium use of labour.

To test for the model’s sensitivity to basic conditions we run a simulationwith no agglomeration economies and reduced labour market effects. Thatis, we neutralized the agglomeration effect (i.e. g=0 in Equation 8.17), andthe substitution effect between leisure time and other activities including worktime (i.e.µ1=0 in Equation 8.10). Under these conditions the effect oftransportation capacity increase was rather negligible as the only effect ofthis capacity expansion was to somewhat increase labour supply ls (Equation8.16).

How do transportation capacity improvements affect labour productivity?Labour productivity was defined as output per unit of labour input. In thisanalysis both variables tend to increase in response to an increase intransportation capacity. The results from the simulations, however, showthat there is no constant pattern of labour productivity changes. In somecases the relative increase in output exceeds the increase in labour (hence,productivity rises) whereas in other cases the opposite happens (productivitydeclines). Apparently, it is the non-linear effects of the various parameters inthe model that determine productivity results.

It is interesting to compare the above results with those obtained byKilkenny (1998) who has examined the effect of reduced transport costs on

Figure 8.3 Equilbrium labour in the economy (l*) as a function of transportation infrastructurecapacity (K).

230 Methodology

rural development, using an equilibrium model of firms’ and workers’ location,with wage rates and output (industrial and agriculture) being the majorvariables. The results from this study show that as transport costs decrease,welfare rises in a pattern similar to that of Figure 8.3, first a sizeable increase,which then quickly lessens as transport costs further decrease.

Given these results from simulation studies that use hypothetical datawe next ask whether the use of real-world data will also support the abovefindings? Therefore, we present results from a study which has examinedthe potential effects of transportation improvements in an economicallydistressed region on the propensity of individuals to participate the labourforce.16

8.5 Results from an empirical study

Berechman and Paaswell (1996) have estimated the impact of transportationimprovements on labour force participation in an economically depressedarea, the South Bronx, New York, where major transportation infrastructureprojects were planned, mainly in order to promote employment. In thisstudy a 2SLS regression model was developed to assess the effect ofaccessibility changes on labour participation. Thus, the model included anaccessibility function and an employment function. Using real-world data,the model estimated the effect of travel time reductions on the propensityof individuals to enter the labour market for different job categories. Indoing so the model took into account mode type, time of departure, carownership, socio-economic attributes (e.g. income, education and age ofchildren) and the wage rate by industry type. The model also accounted foremployment and residential locations within and outside the South Bronx,though it did not account for possible relocation by households andemployment, following the accessibility improvements. Main regressionresults are presented in Table 8.3.

The results in Table 8.3 show parameter estimates for the accessibility andemployment equations for four selected job types, by various variables (emptycells in the table imply insignificant parameters). To illustrate, in Table 8.3 inthe accessibility function the parameter estimate for transit use is 0.806815,indicating that a 10 per cent decline in transit travel time will increase transituse by 8.06 per cent. 10 per cent increase in peak time departures (7:30–7:59) will increase travel time by 2.86 per cent.

Three main results emerge from this analysis. First, in terms of travel time,accessibility improvements indeed tend to encourage higher levels of labourmarket participation. Second, these effects are job specific as some job typesare more susceptible to accessibility improvements than others. For example,in the employment equation, improved accessibility (i.e. travel time reduction)of 10 per cent, will raise executive type employment by 2.37 per cent,technician by 1.87 per cent, administrative by 0.96 per cent and transport by


only 0.79 per cent. Third, other factors, most noticeably the wage rate in anindustry, have considerable impacts on employment decisions, which in manycases are much larger than the accessibility improvement effects. For example,a 10 per cent increase in the wage rate in the finance, insurance and realestate (FIRE) sector will boost executive type jobs by 4.85 per cent, techniciansby 4.47 per cent and administrative by 4.22 per cent. These levels are morethan twice the employment effect from accessibility improvements. In general,when considering all other intervening factors, accessibility enhancement hasa limited impact on the propensity of individuals to enter the labour force.These findings seem to support the results from the simulation analysis, mainlywith regard to the limited ability of accessibility improvements frominfrastructure development to promote employment.

8.6 Conclusions

In this chapter we have developed a microeconomic model to examine theimpact of increased capacity of the transportation infrastructure on localeconomic growth. The equilibrium level of employment is used as the principalgrowth measure. We have hypothesized that firm’s agglomeration andindividuals’ trade-off between leisure and work determine the level of outputand employment in the modelled economy. We further hypothesized that astravel time declines, following capacity expansion, and accounting for firms’location, the additional free time will in part result in increased supply oflabour. If the demand for labour inputs also increases, and accounting for thewage rate, employment may also increase in the economy.

The key conclusion that can be drawn from the simulation results of thismodel is that accessibility costs, which are a function of the capacity of thetransportation system, do matter in affecting labour market decisions.However, other key factors such as firms’ location decisions, external demandfor their final output, individuals preference structure and the shape of thetransportation network also play a major role in determining growth effect.Accounting for these factors the impact of transportation infrastructureexpansion on the economy tends to lessen quite rapidly as more capacity isadded, hence, the results of major case studies described in Part IV, whichshow that even a major infrastructure project can have a negligible effect ifthe additional capacity is only a small proportion of the in-place transportationcapacity.

Tabl

e 8.

3 A

2SL

S es

timat

ion

of a

cces

sibili

ty a

nd e

mpl

oym

ent

func

tions

Tabl

e 8.

3 C

ontin

ued.

Sour

ce: B

erec

hman

and

Paa

swel

l (19

96).

Not

e: P

aram

eter

s sh

own

are

adju

sted

coe

ffic

ient

s (s

ee t

ext)

, and

sig

nific

ant

at 0

.05

leve

l or

bett

er.

234 Methodology

Notes

1 More accurately, we need to examine changes in consumer surplus.2 In a large empirical study Schwartz (1992) has demonstrated how firms located

over the entire metropolitan area of New York, Los Angeles and Chicago interactmainly with firms located in the respective central cities.

3 As noted in Chapter 6, most of the theoretical and empirical work that linksproductivity gains with scale effects was carried out at the national level. See forexample Romer (1986). At the urban level most of the work associatesproductivity improvements with city size. See, for example, Fujita and Ogawa(1982), Henderson (1986) and Moomaw (1981).

4 Other forces affecting the extent of activity clusters include transaction costs,incomplete contracts and information, and flexibility in interaction. McCann(1995) attributes locational clustering to various costs and revenues associatedwith the location of firms. These include distance-transaction cost, location-specific factor efficiency costs, hierarchy-co-ordination costs and hierarchy-opportunity costs.

5 To be distinguished from financial measures such as increase in the value offirms’ stock. While financial measures of firms’ performance may also reflectreal ones, external factors such as the market’s rate of interest or rate of inflationhave a critical impact on financial variables.

6 For a study which explicitly models residential behaviour relative to locationand non-work (i.e. shopping) trips see Anas (1995) and Anas and Kim (1996).The effects of infrastructure policy on location were studies by Haughwout (1996)who formulated a model which simultaneously considered equilibrium labourand land markets, though it does not consider changes in the transportationmarket and their effect on location.

7 That is, the effect of y1, on y2, is the same as the effect of y2 on y1.

8 We also assume decreasing marginal productivity of labur, i.e. 9 Notice that in this economy no non-labour income is available.

10 Travel time is, of course, not independent of the mode of travel used. Car userequires parking whereas travel by a bus requires access time, egress time, waittime and in-vehicle time. Commuters on a fixed route rail transit further needto reach rail stations either by walk or by the use of a car, which needs to beparked.

11 However, when we compute agglomeration economies (Equation 8.3), at leastinitially, we regard both firms as if they are located in separate sites.

12 Analytically it is not possible to ascertain whether this approach overestimatesor underestimates travel times between all ij pairs, which would have beenestimated, if we had included a formal network sub-model.

13 In the general formulation of the model (Berechman 1994), a social welfarefunction was also included. It has the form However, herewe do not conceive of a social planner who is capable of manipulating the systemin order to produce socially optimal values for firms’ output, the consumptiongood and leisure time. Therefore, in this analysis we use the firms’ profit functionas the criteria for the model convergence into equilibrium solution. At the pointof convergence we obtain the equilibrium (but not necessarily socially optimal)values for output and labour.

14 As explained above, in this analysis we do not consider the use of land byhouseholds and firms and therefore, the effect of changes in land rent on thelocation of firms and households.


15 The simulation results show that after 20 per cent capacity increase, employmentincreases by a merely 2–3 per cent.

16 While not directly related to transportation improvements, Durkin Jr. andWassmer (1994) have obtained similar results. Using a production function modelthey found that under certain conditions the positive marginal effect of increasein public capital spending on urban growth tends to increase first and then abateand become negative with further spending, given average city size.

Empirical case studies

Main issues and structurePart IV completes the analytical section by presenting a series of case studies.The intention here is to use carefully selected complementary case studiesthat illustrate the range of impacts which might be expected from transportinfrastructure investments at the local level. As can be seen from the tablebelow, these case studies cover a range of different modes of transport at thelocal and national levels. They are all taken from developed countries andcities. Some are analytical in nature, using high quality primary data, whileothers entail an extensive review of available secondary data and supportinginformation.

The selection of case studies depends on the availability of source materialfor the range of projects to be considered. The best source material has beenobtained on rail investment projects at both the programme level forinternational and regional links and for individual projects at the city level.This has allowed detailed analysis on the ex post impacts, particularly asthey relate to the labour markets and employment. The airport material isalso available. The impacts here are felt both at the local level through directand induced employment and at the regional/national level through multiplier

Part IV

Part IV scheme for case study selection

238 Empirical case studies

effects. Ironically, the most difficult case study has been the assembly ofinformation relating to new road construction at the regional and local levels.This is partly because little new construction is taking place, but is alsoexplained by the complex phasing and time taken for many road investments.Both the construction and the impacts are phased over a considerable periodof time, which in turn complicates data collection and the analysis of impacts.

The economic impacts ofroads

9.1 Introduction

Roads play a fundamental part in the development of cities and regions, asthey affect accessibility and the relative attractiveness of all locations. Recentinvestments have mainly taken place outside cities as new construction incities is often seen as reducing the attractiveness of city centres. Thisapparent contradiction between the benefits of roads (e.g. better access)and the negative aspects (e.g. environmental quality) has never really beenaddressed in analysis (see Part III) which has tended to concentrate on thephysical quantitative aspects rather than the social and environmentalqualitative aspects. Most road investment has taken place in specificcorridors where development has been encouraged, or in locations betweenand around cities to establish the inter-urban network. The consequence ofinvestment in these locations would suggest that development pressureshave also responded by moving to these corridors and to the networkoutside cities. These, so-called ‘greenfield sites’ are also attractive todevelopers as land prices are lower than in the city centre; land assembly iseasier; the development costs are lower; the sites are car accessible and thequality of the environment is perceived as being high. It is not surprisingthat investment in new roads out of city centres has generated substantialpressures (Headicar 1996).

This chapter examines the debate over the economic impacts of roadsthrough a case study of three major motorways: the M25 London orbitalmotorway, the French A71 motorway and the Amsterdam orbital motorway.Mainly for reasons of data availability, the first of these case studies will beexamined in greater detail than the other two. The basic motorway networkin Europe has been constructed over the last fifty years and there are nowsome 46,000 km in the fifteen European Union (EU15) countries. Two-thirdsof this network has been constructed since 1970 and one of the principaljustifications for investment has been the beneficial impacts on accessibility,economic development and employment (see, for example, Blum 1982).Although the focus is primarily on one case study, the conclusion widens the

9


discussion to examine impacts of roads on regional and urban developmentin other European countries.

9.2 The London M25 motorway

9.2.1 M25 history

The M25 was completed on 29 October 1986 and provides a high qualitymotorway orbital route around London. Its total length is 188 km and itruns mainly in greenbelt land1 at between 20 and 35 km from the centre ofLondon, carrying over 700,000 vehicles a day (see Figure 9.1). It is operatedat near to or above capacity over particular sections for much of the workingday. The idea of an orbital highway around London dates back to 1905when a Royal Commission on London’s traffic suggested a ring about 20 kmfrom the city centre (Greater London Council, 1970). The M25 conceptpredates the motor age.

The motorway became the M25 in November 1975, when two Londonorbital relief roads were combined to form one route. Previously, plans werebeing developed separately for an outer orbital in the north (M16) and anotherin the south (M25). Sections were opened over a 10–year period from thefirst section (near Potters Bar in 1975) to the final section (from Micklefieldto South Mimms in October 1986). The total cost of the motorway was£1,000mn (in 1985 prices) and some 200,000 vehicles a day use the busiestsections. Since 1986, over a quarter of the motorway has been widened fromthree to four lanes in each direction to accommodate the ever-increasing levelsof demand (Highways Agency 1996).

Many sections of the motorway are operating at or above capacity andstudies have been carried out on the introduction of traffic managementmeasures to maximize the operational efficiency of the existing M25 (e.g.Rendel et al. 1989). Road widening and junction improvements have takenplace, with other complementary measures to improve alternative roads,particularly for local traffic. Speed restrictions have been introduced onsome busy sections to reduce speed variation and lane switching andincrease utilization of existing capacity. Information signs are used to warnmotorists of delays on the M25 so that other routes can be taken. Studiesare taking place on the possibilities of reducing lane widths to 3.45m forall vehicles, 3.25m for non-Heavy Goods Vehicles (HGV) and 2.0m forthe hard shoulder, and for limiting access to the motorway. The M25 hasreached its capacity within three years of opening and most of this growthcame from traffic diverting to the new road, newly generated traffic andthe use of new destinations that had become accessible. There was verylittle mention in any study of the new traffic generated as a result of firmslocating near to the M25.

The economic impacts of roads 241

9.2.2 M25 development pressures

The justification for the M25 included congestion relief in central Londonas some 30 per cent of traffic using the motorway is bypassing the city,but it also provides links between the region’s four main airports(Heathrow, Gatwick, Stansted, Luton). The motorway provided relief tothe local network, but it has generated longer journeys as it provides aquicker route.

Reactions to M25 have been mixed. The official view is that the motorwayprovides an outstanding example of a road ‘that would aid economic recoveryand development’ (UK Department of Transport 1980). The unofficial viewis encapsulated in Chris Rea’s hit single ‘Road to Hell’ which was inspired bythe M25. In between these two extremes, the Standing Conference on Londonand Southeast Regional Planning (1982) is more cautious. In its comprehensiveimpact study of the M25, completed before the motorway was opened, theyidentified two main consequences of the road on the regional economy. Theremay be a conflict between the motorway’s transport function, the economicobjectives of assisting industrial and commercial development as part of the

Figure 9.1 The M25 London orbital motorway.


nation’s recovery from recession on the one hand, and the maintenance ofthe green belt on the other. Economic objectives were in conflict with openspace preservation. Second, they recognized that the main developmentpressures which the motorway is likely to accentuate will occur in the westernsectors, while the main opportunities and needs for new investment lie ininner London and the eastern sectors.

As a result of this, the regional strategy may fail in these respects,particularly if the counties surrounding London actively competed for newdevelopment. The Greater London Development Plan identifies the preferredlocations for industry and office development in London, together with thestrategic centres for major new retailing locations. The structure plansproduced by each of the surrounding counties make provision for furtherdevelopment in their own areas, usually specifying preferred locations,maximum acceptable increases in floor space and land take necessary foreach type of development.

Development plans (e.g. structure plans) in the UK are not in the form ofzoning ordinances detailing location, density and building heights, but moregeneral policies and proposals for the use and development of land. Theyprovide guidance to developers and a framework within which developmentcontrol can be exercised. Individual proposals are treated on their own merits,but should take notice of the provisions of the development plan (Headicar1996).

To counter the preference for locating in the western corridor from Londonwhich has a more buoyant economy and better levels of accessibility, theregional strategy was designed to promote the eastern sector with futuretransport investment to spread the M25 benefits in towards central London.The strategy also identified individual well-located existing centres for furtherdevelopment and explored the potential for ‘green development’ (science parksand recreation centres) in the western sector.

The process of road construction and development was seen at this time(early 1980s) as an integral part of strategic planning. New developmentopportunities brought about by increased road accessibility could be used tocreate jobs and local economic benefits in declining areas, particularly in theeastern sector of the M25. More specific studies were also carried out toidentify which sectors of the economy would benefit most from location inclose proximity to the motorway (Nathaniel Lichfield and Partners andGoldstein Leigh Associates 1981). Here it was suggested that five types ofdevelopment would compete for motorway accessible locations. 1 Warehousing activities serving national and regional markets, where

transport costs were a significant element of total costs. For example,Dagenham, Dartford, Grays and Redbridge.

2 High technology growth industries would locate in towns just beyondthe green belt.


3 Offices that did not require central London locations would move out ofthe city where costs were high and environmental quality was low. Butgood accessibility to specialized and local clerical staff is an essentialcomponent of office location decisions. For example, Romford, Croydon,Orpington, Hounslow, Uxbridge and Barnett.

4 Hypermarket and superstore developments would take place near M25junctions, but this conflicts with green belt policies.

5 Discount shopping stores would locate along the minor orbital routesaround London (e.g. North Circular Road) and along the main arterialroutes.

It was argued that accessibility was maximized at the junctions where theM25 met with the main radial routes into London. At these sites there werefragmented parcels of land (broken up by the M25) where existing uses wereno longer viable and these locations were appropriate for new uses.

9.2.3 M25 accessibility impacts

One of the principal debates over the impact of the M25 is its effect onregional accessibility, and on the patterns of employment change that it mightcause. Early research (Jones 1982) using the Regional Highway Traffic Modelsuggested that under uncongested conditions the M25 would have a majorimpact on accessibility, particularly in certain sectors. Central London alreadyhas high levels of accessibility both with and without the M25, so the effectsof the M25 on the centre of London are marginal.2 Significant improvementswere found in the peripheral areas (20–30 km from central London), with agreater impact in the east than in the west (increases of 10–20 per cent).However, the greatest improvements (up 50 per cent increases in accessibility)were found in particular corridors around London where the M25 met themain arterial routes (e.g. Redbridge and Harlow along the M11, Guildfordand Leatherhead along the A3). There were substantial savings in travel timeaveraging about 30 minutes in the morning peak and 20 minutes in the offpeakperiod. Jones (1982) concluded that because of the increases in accessibilityconferred by the M25, locations within easy reach of an M25 junction (10minutes travel time) are likely to be subject to considerable pressures fordevelopment.

A more recent study (Linneker and Spence 1992) has used a wider rangeof accessibility measures to assess the impact of the M25 on Britain as awhole, in particular the southeastern region. Market potential measures werecalculated for the ‘with-road’ (1987) and ‘without-road’ (1981) cases usingexogenously determined procedure of route minimization between regionalzones (179 in all). Time, distance and cost functions were calculated for HGVsand cars, at two points in time (1981 and 1987). They both had the form of:

where, MPi is the market potential of zone i; Pj is a measure


of market potential in zone j (employment in 1981 and 1987); Cij is a measureof transport cost from i to j (e.g. time, generalized cost or distance); and a isan exponent (assumed to be 1 in this study).

The results demonstrate that significant accessibility changes have takenplace, but not always in the direction anticipated. For HGVs there have beenclear increases in accessibility in all zones with the exception of Inner Londonwhere the M25 has, in fact, increased travel times. The greatest change forHGVs takes place in the Inner South East (Kent) areas for travel time savings(reduction of 9.6 per cent), while for the other measures of accessibility(distance and cost) there were increases in all areas, particularly the InnerSouth West area (Surrey, Kent, Berkshire). The M25 meant that the distancestravelled increased here by 4.7 per cent and the generalized travel cost increasewas 2.5 per cent.

For cars, the results of Linneker and Spence study were even morepronounced with travel time reductions of between 8.4 per cent and 12.7 percent in all areas around central London. Travel distances increased with thegreatest change (+6.2 per cent) in the Inner South West area, but generalizedtravel costs were reduced in all areas apart from Inner and Outer London.The greatest reduction was found in the Inner South Area (-4.5 per cent).

The implications of these results are not easy to interpret. The M25 is amajor investment that would be expected to have an impact on the accessibilityat the regional and national level. Given the assumptions on speeds, values oftime and employment, the scale of the accessibility impacts is small and theincrease in travel distance and generalized cost (for HGV) suggest that theeffects of the M25 are negative. It is only travel time that has been reduced asthe M25 provides a higher speed route, but given the levels of congestion andthe capacity limitations, even this saving may be diminished. One possibletest would be to estimate the speed of using the M25 that would negate anysavings. What level of congestion is needed on the M25 to cancel out allsavings (time, distance and cost)?

Turning now to the economic growth effects of the M25, in a follow-uppiece of research Linneker and Spence (1996) used a multiple regressionanalysis to examine the M25’s impact on employment growth as this relatesto changes in relative accessibility and transport cost advantages. Dodgson(1974) used a similar analysis to explore the employment implications of theM62 Trans-Pennine Motorway on 30 local areas (1960–66). Botham (1980)used the same approach to test the employment effects of the post-war roadprogramme between 1957 and 1972 on an intersectoral and inter-regionalbasis. The Linneker and Spence (1996) analysis uses employment criteria toindicate the regional economic development change (1981–7). Two measureswere used to examine regional economic development: 1 Differential employment shift (1981–7) is used to measure local economic

performance. This is the competitive element in shift-share analysis, which


measures the employment change in an area, given a particular industrialstructure.

2 Demand for labour index (1981–7), defined as the difference betweenactual employment change and the expected change, which arises fromthe natural increase in the local population. Positive values of this indexindicate in-migration of residence, in-commuting by trip makers, higherproductivity, increased levels of activity and lower levels ofunemployment.

These measures were analysed by regression and correlation analysis with arange of independent variables to reflect: 1 Industrial structure index (1981) which provides an industrial profile of

an area in relative terms.2 Congestion based on population density.3 Employment density (jobs per unit area—hectare).4 Labour availability based on the ratio between total employment in an

area and the resident working population.5 Accessibility measured by market potential as explained earlier. The results are counter-intuitive and differ from those produced by Dodgson(1974) and Botham (1980). It was found that areas with a high accessibilityrelative to other areas are losing employment. The market potentialaccessibility measures are negatively related to the demand for labour indexand the differential employment shift. This effect is modified if the dynamicaccessibility change is isolated (i.e. the pure M25 road effect on accessibility).Those areas that have shown the highest percentage increases in accessibility(only measured by time) have shown higher employment growth or reducedemployment loss.

Linneker and Spence (1996) were cautious in their explanation. Proximityto centres of high population densities (the large conurbations) producesthe poorest employment performance. The regional dimension seems toemphasize this difference as the London region performs better than otherlocations (see also Frost and Spence 1991). Accessibility from new roadconstruction facilitates the ability of local firms to expand market area andhence create more employment. But it may also allow expansion of firmsoutside the region into the newly accessible locations. The resultantcompetitive position is the combination of the two sets of factors. Physicalaccessibility, as measured in the Linneker and Spence (1996) study, is onlyone part of the competitive position and it affects different activities in avariety of ways.

Economic potential may need to be redefined (Newman and Vickerman1993), particularly as it is often given the impression of smooth, continuouschange in accessibility over the whole region (or country). The implication


has been that the improvements in infrastructure will be concentrated in thecore area and that new infrastructure can never promote regional developmentor cohesion except where it overcomes a physical barrier (Vickerman 1998).Improvements in transport infrastructure can change initial levels ofaccessibility, but they will reinforce locations with good accessibility. Relativegains may be greater in peripheral locations, but absolute gains are highest inthe locations with the highest starting points. As Vickerman (1995) concludes,accessibility is a relative concept and once certain minimum thresholds arereached we require further understanding of the way that accessibility is usedto further the interests of a region’s enterprises to achieve real comprehensionof the economic impact.

9.2.4 M25 retail development

One of the main changes anticipated in property market analysis (e.g.Nathaniel Lichfield and Partners and Goldstein Leigh Associates 1981), wasthe pressure for retail developments at accessible locations on the M25. Theseretail developments would include superstores and hypermarkets as well asretail warehousing. Prior to the completion of the M25 in 1986, the argumentsconflicted. Damesick (1986:157) concluded that ‘the role of the M25, byitself, in creating wholly new opportunities for economic growth in the South-East on an inter-regional basis is likely to be small, relative to the region’sother existing attributes’. The M25 only represents a marginal addition tothe region’s locational advantages and so is likely to enhance existing trendsrather than create new ones. The contrary view was put forward by Simmons(1985), who argued that government has only belatedly accepted the M25’sability to create development opportunities. He was critical of the absence ofany strategic view on where development should take place. Even if therewere a strategy, there is considerable doubt as to whether development pressurecould be directed in practice.

Gould (1987) has carried out one of the few empirical studies with hisinvestigation of planning applications for retail developments. The survey(1986) covered all seven county authorities, including the Greater LondonCouncil, through which the M25 passes or is in close proximity.3 A total of40 districts were affected by the M25 and 30 of these responded to the survey.In addition, a further 22 districts were surveyed which were not affected bythe M25, but were part of the same regional economy. There were 144 retailapplications in the ‘M25 belt’ (4.8 per authority) and 73 in the ‘control area’(3.3 per authority). These figures are in line with national trends and theupturn in retailing taking place in the 1980s. The proximity to the M25 wasnot seen to be important in over three-quarters of the cases, and this wasconfirmed with interviews with the retailers. Most of the retailers were notinfluenced by the policies of the local authorities and were concerned abouttheir own network of stores and their market share. As 75 per cent of their


turnover came from customers within a 10-minute offpeak drive time, it wasunlikely that many would use the motorway.

However, in terms of floor space the impact of the M25 has been substantial.For the larger developments (> 18,000 sq m) involving regional shoppingcentres and retail warehouses, proximity to the M25 was important. The 14applications that stated that proximity to the M25 was very importantaccounted for 821,226 sq m of floor space, which constitute 10 per cent ofthe applications with 45 per cent of the floor space. Smaller scale applicationswere not influenced by the M25. Details of 10 of these 14 applications aregiven in Table A9.1 in Appendix 9.1. These 10 account for 96 per cent of thefloor space of these 14 applications, and the other four are very smalldevelopment proposals (which included Iver, Colnbrook (Richings Park) andElstree Aldenham retail park). Moreover, these 10 key applications are locatedin just six sites. In 1986, land values soared as a result of this increaseddemand. For industrial use it was £500,000 per acre. For technological parksand retail warehousing it was £750,000 per acre. For food retailing it rose toabout £2 million per acre (Procter 1988).

The clear conclusion reached from this empirical analysis was that theM25 has an important role in enlarging the catchment areas for regionalshopping and warehouses, but not for those smaller supermarkets used morefrequently. Consumers making regular convenience or low value bulky goodspurchases are not prepared to travel for more than 10 minutes by car to theshop. As M25 access is limited, there are few locations that would benefitfrom the ‘10 minute’ rule. The M25 has a very limited impact on this scale ofretail development.

Despite the small impact of the M25 on local retail facilities, it has asubstantial impact on the regional scale of retail developments. For this levelof development to transpire, strong agglomeration economies must be present.This observation supports our basic contention (see Chapter 7) regarding thenecessary functional relationships between accessibility improvements andagglomeration economies, in order to engender economic growth. At thebeginning of the process (in 1986), planning applications were mainlyspeculative, but the planning regime under the Conservative government wassupportive of prestige large-scale developments. Many applications werereceived (see Table A9.1), but gradually that number was reduced as theproperty market situation became less buoyant in the late 1980s and early1990s. The property market business was also restructuring itself (Table A9.1).A second reason for the change in approach has been the strong public desireto maintain the green belt. Even the strong Thatcherite free market economicswas not able to shift public opinion, so many of the applications in the greenbelt were withdrawn at an early stage as it was unlikely that they would beapproved after a local inquiry, or even after statutory appeal to the minister.

Consequently, most development has taken place further away from theM25, outside the green belt. The districts adjacent to the M25 have increased


office space by 2.4 million sq m, resulting in some 160,000 new jobs, (1989–91; OECD/ECMT, 1995). The Blue Water Park development (Table A9.1)is the largest retail and leisure centre in Europe (1999). It is anchored bykey retailers such as John Lewis, Marks and Spencer and the House ofFraser, with 275 other retailers. There are three leisure villages, each with aseparate theme, together with a 12–screen cinema. The 100 hectares sitealso has 150 acres of parkland and 23 acres of lakes.4 It seems that once the‘anchor’ retailers are established, there is considerable leverage to win over

Table 9.1 Case study: Lakeside retail development in Thurrock, Essex

Sources: Based on Capital Shopping Centers (1997) and research by Stamp (1997).


other retailers so that agglomeration economies can take place. The Lakesidedevelopment has expanded by some 30 per cent as related developmentshave been added to the original development, mainly through linkagesbetween firms with similar or complementary markets. A key conclusionfrom this review is that political and policy decision making have a ratherconsequential impact on the ability of a major transport infrastructureinvestment such as the M25 to generate economic growth. An interestingexample of this impact is given by Table 9.1, which shows the evolution ofthe development of the largest (in 1995) retail shopping centre in the SouthEast of England—140,000 sq m.

The evolution of this particular retail project as outlined in Table 9.1 clearlydemonstrates how crucial were the impacts of the political and planningprocesses on the ability of the M25 to bring about economic growth effects.We shall return to this conclusion in subsequent chapters and provide asummary statement in Chapter 12.

9.3 The impact of the French A71

The A71 motorway in central France links the cities of Orleans and ClermontFerrand via the city of Bourges (see Figure 9.2).

An empirical case study of this project was undertaken by Zembri-Mary(1996) and had two main aims. The first was to establish whether there wereany anticipatory reactions, in the form of real estate transactions, to the A71in the corridor through a comparative study of three locations close to theautoroute (Montmarault, Gannat, Riom), as compared with one ‘control’location (Lapalisse) some 40 km away from it (see Figure 9.2). The secondobjective was to comment on the quantitative measures, as compared with

Figure 9.2 Map of the A71 route in France.


more qualitative explanations. Before we report the findings of this study it isworth reviewing the history of the A71, which is summarized in Table 9.2.

The study obtained the price of all 730 transactions over 20 years in thefour study locations (the three study locations and one control location).This information included area, the designation on use (e.g. agricultural), theplot location and the present and future usage of the site. A series ofadjustments were carried out on the data so that comparison could takeplace between the four locations. Explanations were sought after discussionswith experts on the quantitative estimation of change. Two clear types ofimpacts were observed, one related to the actions taken by the towns and theother related to proximity to the interchange with the A71.

The volume of transactions was high during the 1970s and the beginningof the 1980s, mainly resulting from the rationalization of the land distributionand the attempt to consolidate small unused or enclosed plots. The controllingadministrative agency, SAFER,5 had a key role here in land assembly. Thisconsolidation was matched by a slowing down of transactions betweenindividuals as landowners delayed selling to capitalize on the expectedincreases in land prices. The land market (1985–7) increased in value as aresult of the transactions necessary for the implementation and developmentof the A71. In the three locations close to the A71 prices rose by about threetimes, with 1987 being the peak year (when the A71 was opened). Theseincreases were even greater (more than ten times) when the land was changedfrom agricultural to improvement land (zoned for business use). Around

Table 9.2 History of the A71

Source: Based on Zembri-Mary (1996).


Montmarault, for example, agricultural land was valued at about Ffr 1.5 persq m (1980), but this figure increase to Ffr5 per sq m in 1987 withoutreclassification, and Ffr22–44 per sq m with reclassification as business use(also in 1987).

The conclusions reached in this study (Zembri-Mary 1996) reflect theimportance of actions being taken at the appropriate time and the involvementof the three main actors—the land owners, the town council and SAFER (therural development agency). Land is the major issue, with the links betweenthe owners and the town council important both in terms of the agreed priceand the timing of the acquisition. The plots of land in one case were valuedby the land registry at an average price of Ffr 13 per sq m (about $2.2), butthe owners went to the expropriation court and finally sold for Ffr22 per sqm (all compensation included). This meant that the franchising company ofthe area zoned for business had to pay nearly Ffr500,000 more than expected.Local authorities need to anticipate when to purchase and to have the resourcesavailable. SAFER has an important role to play in putting land together at alower price and then selling on to the local authority, as it does not have aprofit maximization motive. The land registry has to ratify all transactionsmade by the local authority so that the use of public money can be monitored.However, their role is both as guardian of the public interest and as facilitator,as the land registry can also pay above market price for a plot of land whichis essential to the progress of the road.

Several conclusions can be reached from this study. First, land values riseas a result of the road, mainly due to the construction process, but also due tothe zoning of business activities at interchanges. This increase in value takesplace in anticipation of the new road and continues after the road is opened.Second, the planning agencies such as the administration for rural landdevelopment play an important role in effecting the economic consequencesof road development. The third conclusion is that these relationships aredynamic in that they can be identified before, during and after the road wasconstructed.

9.4 The Amsterdam orbital motorway

This orbital motorway around Amsterdam was completed in 1990. The routeis only about 5 km from the city centre, but serves to divert through trafficfrom city streets (see Figure 9.3).

The travel impacts of the completion of the orbital motorway aroundAmsterdam have been substantial in terms of route choice and timing oftrips, but they are less noticeable in terms of modal choice and trip frequencies.Despite its close proximity to the city centre, it has led to a substantial (20per cent) reduction in congestion in the Amsterdam area. With regard toeconomic development effects, the completion of the road has reinforced theposition of locations that were already in a strong competitive position. Little


change was observed in office rents and those offices nearest to the motorwayjunctions demonstrated a negative relationship between rent levels anddistance. Yet the businesses interviewed claimed that proximity to the orbitalmotorway is important for transport related activities.

A survey of office rentals in locations where distance by road to theorbital motorway did and did not change provided counter-intuitive results(Bruinsma et al. 1996). Offices located in areas that should have benefittedfrom improved accessibility showed an insignificant increase in rent levels(1987–91). These increases were much lower than increases in those areasnot affected by the motorway. As the motorway was only opened in 1990,the after survey (1991) may not have picked up those changes which mayrequire longer periods to be fully capitalized in the real-estate market.Alternatively, the market may have already anticipated the impact andadjustments have already taken place. But this explanation is not supportedby trends in office rent levels prior to 1987. Bruinsma et al (1996) concludedthat no impact on office rent levels is attributable to the Amsterdam orbital

Figure 9.3 Map of the Amsterdam agglomeration.


motorway. Yet, when this qualitative analysis is compared with a quantitativeanalysis (using a regression model) the orbital motorway does appear as asignificant explanatory variable. It was found that prices are some Ffr23per sq m (about $12.1) higher at locations near to the motorway junctions,as compared with those locations 1 km away from the junctions (amountingto about a 10 per cent premium). Other location variables were alsosignificant in the regression analysis, as was the ‘newness’ of the officebuilding, with modern offices attracting a premium value. The conclusionreached here was that although the effects are not directly visible in locationswhere the new motorway sections were constructed, the orbital motorwayis an important location factor for office firms.

A survey of three groups of employers with businesses near to the existingorbital motorway sections, near to the new orbital motorway sections andfurther away from the motorway provided additional supporting evidence.Firms experienced traffic delays prior to the completion of the orbitalmotorway (90 per cent) and this affected business efficiency. The new routeimproved accessibility and this had in turn resulted in increased turnoverand productivity (10–20 per cent of firms). So the motorway had animportant impact on competitiveness, particularly as it related to transportactivities (e.g. the movement of goods, commuting and access to clients orcustomers).

From a comprehensive survey of road investment schemes in theNetherlands, it has been concluded (Rienstra et al. 1998; Rietveld andBruinsma 1998) that there is no evidence of a clear impact on overallemployment in the regions resulting from the new patterns of accessibility.Peripheral accessibility has increased faster than central accessibility, butthe two methodologies used (based on reference regions and comparisonand on a labour market model) failed to give any consistent results. Theconclusion reached was that more sensitive analysis is required at a lowerlevel of spatial disaggregation. (The model framework in Chapter 8 is anexample of such an analysis.)

This study has implications for the M25 and other major road investments.It seems that the impacts are not just in terms of reductions in delays andimprovements in accessibility, but also in general business confidence. Themeans by which these factors are then internalized (if at all) in the form ofcompetitive advantage, productivity gains, higher rent levels, etc. all needcareful examination through comparative analysis (with control areas toisolate change) over the appropriate length of time. Response to change orinvestment is a process which does not necessarily take place at the sametime and firms or individuals may respond in a variety of ways according totheir own constraints or requirements. The subtle nature of many of theseprocesses makes the issue of causality more difficult to explain. Certainly,more formal quantitative methods such as the model framework in Chapter8 need to be supplemented with more sensitive qualitative analysis.


9.5 Conclusions

The road case studies reviewed in this chapter seem to suggest a link betweentransport investment, economic activity and development. Yet in each of thesecases, those factors were not investigated within a unifying framework and,as a result, it is quite difficult to assess whether these investments have indeedled to any new economic development and, if so, by how much. In addition,the critical role of policymaking in affecting these variables was also notclearly identified though, as we have underscored many times in this book, itis a key factor in understanding the relationships between transportinvestments and economic development.

In general, road investment decisions such as those reviewed here do notattempt to quantify the impact on economic activity and development. Theymainly examine the transport costs and benefits, principally through changesin travel times. But as we have expounded in Chapter 7 (see Figure 7.2), anyeconomic development benefits are a function (though a non-linear) ofaccessibility benefits. These benefits are crucially dependent upon theassumptions made on traffic speeds through the network often assuming thatthere will be no new traffic generated (a fixed trip matrix). The absence ofthe explicit inclusion of redistributed traffic (to new accessible destinations)and newly generated traffic (resulting from latent demand and from newactivity) is a major limitation. The M25 amply illustrates these problems andthe high levels of demand, congestion and capacity limitations mean that thebenefits of travel time savings are substantially less than originally estimated.This does not mean that the road should not have been built. Rather it meansthat the methods used should be able to estimate both the travel time savingsto existing traffic (already done) and the travel time savings to a substantiallyincreased level of demand mainly at much lower speeds (not done). Onlythen can potential economic development benefits be correctly assessed.

Notes

1 The green belt is an area of tight planning constraint designed to contain theurban sprawl of major urban areas such as London.

2 This finding supports our assertion throughout this book that in well-developedeconomies where existing transportation networks provide high level accessibility,even a major project like the M25 will improve accessibility only marginally.

3 Defined as being within ten minutes offpeak drive time from the M25.4 One hectare contains 10,000 sq m or 2.471 acres. Each acre is 4,840 sq yd.5 Société d’Aménagement Foncier et d’Etablissement Rural (SAFER) (the

administration for rural land development).


APPENDIX 9.1

Table A9.1 Applications for major new retail development around the M25


Table A9.1 Continued

Source: Gould (1987) and updated.

The economic impacts of rail

10.1 Introduction

Investment in rail systems has often been justified not only on the basis ofstrict benefit cost analysis, but also by the broader employment anddevelopment benefits. In fact, many investment decisions would not havebeen made if only the transport benefits had been considered. For example,the Jubilee Line extension in London had a transport benefit-cost ratio of lessthan one (0.95:1 in 1991 prices assuming an 8 per cent discount rate over 30years), when benefit-cost ratios would be over 1.3 to 1 for investment to takeplace. This means that over 34 per cent of the benefits would be non-transportbased and related to new employment and local inward investment (LondonTransport 1993). These additional benefits are very difficult to measure priorto the investment decision and it is only through careful modelling orretrospective analysis that we can attempt a systematic review.

In this extensive chapter, we carry out analysis from two differentperspectives. The first is a detailed modelling study of the Buffalo Light RailRapid Transit (LRRT) system carried out before the investment was made.This ex ante analysis raises important methodological issues which have tobe resolved, particularly if the investment is to encourage local economicgrowth (as in the Buffalo case). As this investment has already been made itis now possible to carry out an ex post analysis to see whether the transportand local economic benefits have been achieved. The second perspective is toreview material from a wider range of sources in which empirical studieshave been carried out to assess the impact of rail investments. This evidencehas been divided into the large-scale interregional high-speed rail investmentsand the smaller, more local-scale urban investments. The focus at the urbanscale is the BART system in San Francisco where there are now twenty yearsof data available to assess the economic impacts of the rail system.

The bulk of this chapter focuses on the Buffalo LRRT study relative to itsmethodology and major test results. Therefore, the design of the chapter is asfollows. Section 10.2 describes the methodology and specific models used foranalysing the impacts of the LRRT project on the CBD area as well as major

10


regional economic and demographic long-run trends, which are inputs tothese models. Principal tests and results are described in Section 10.3. Section10.4 provides a postscript of the Buffalo LRRT project. Key conclusions fromthe Buffalo case are presented in Section 10.5. Section 10.6 discusses ex postresults from three other major rail projects, the Japanese Shinkansen, theFrench TGV high-speed rail systems, and the BART system in the Bay area inCalifornia. Main conclusions are in Section 10.7.

10.2 The Buffalo Light Rapid Rail Transit system

As explained in Parts I and II, transportation projects, by and large, aredesigned and implemented primarily in order to solve transportation problemssuch as alleviating traffic congestion or improving road safety. Transportationinvestments are seldom made with the explicit objective of achieving non-transportation objectives such as local economic growth. It is no wonder,therefore, that most studies of the non-transportation effects of transportationinvestments were based on projects whose main objective was an accessibilityimprovement (e.g. Boyce et al. 1972). In contrast, here we examine atransportation project whose principal goal was to encourage local economicgrowth, in particular to revitalize the declining CBD area. Studies done in thepast on the capability of transportation investments to bolster the growth ofa declining CBD were based on projects whose main objective was an improvedtransportation system (e.g. Mackett 1980; Poulton 1980; Cervero and Landis1997). Our main underlying objective here is to ascertain the degree to whicha project designed first and foremost to encourage CBD growth is able toaccomplish its stated economic goal.

The project under study is a 6.4 mile (about 10.6 km) Light Rail RapidTransit system (LRRT) in Buffalo, New York. Construction began in themid-1979 and lasted for about six years. The project was located within adeclining central city area and represented a large ($450mn, in 1978 prices)public investment for which user or transportation benefits were not the soleor even a major consideration. Rather, economic and land development andthe creation of jobs primarily in the CBD area were the major factors in thefunding decision. These growth objectives were to be obtained by providingadequate capacity to transfer fast and conveniently a large number ofpassengers from the city edge to the heart of the downtown. The availabilityof such capacity was seen as a major impetus for engendering retail,commercial, cultural and entertainment activities in the CBD (the downtown)area. In brief, the Buffalo LRRT project was viewed as a necessary conditionto stop the decline of the depressed urban core and encourage revitalizationby stimulating additional public and private investments. This rationalebecame the official raison d’être for undertaking this investment.

This analysis is based on an ex ante study carried out during the project’simplementation period in order to ascertain the potential ability of the LRRT

The economic impacts of rail 259

investment to reverse the current trend of decline of the CBD area and promoteits revitalization (Paaswell and Berechman 1981; Berechman and Paaswell1983). We also report some general observations of the present-day status ofthe project. They strongly support the main findings of this study.

10.2.1 The Buffalo LRRT: background and methodology

The Buffalo LRRT project is a massive capital investment focused on asmall well-defined area.1 The LRRT corridor is located in a region of generaleconomic decline whose major characteristics are typical of manynortheastern cities in the USA. Because of the regional out-migration ofpopulation and the intra-regional sustained trend of suburbanization, therehas been a continuous decline in population in the central city from 1960to the present.2 Another characteristic is the constant change in thecomposition of the regional labour force from traditional blue-collarindustries toward an increase in the white-collar service employment. Thesetrends are shown to have profound effects on shopping patterns, employmentlocation and travel behaviour (see Chapter 4, section 4.2 and 4.3 for ageneral discussion of such trends).

Several working assumptions were made in examining the effects ofthe project on the CBD area. These assumptions also served as a guidelinefor developing the methodology. First, it was assumed that the LRRTproject will produce a large number of interrelated impacts which foranalytical purposes may be categorized into four distinct impact groups:transportation, economic, shopping patterns and land use. Second, it hasbeen assumed that no one model can simultaneously treat all of theseimpact types and therefore a set of models and techniques should be used.The third working assumption is that the potential effect of any of theabove impact types will be enhanced or constrained by the regional trendsmentioned above. Unlike growth areas, where a facility can be built inanticipation of generated demand for its use, investment in a decliningarea must consider the amount of activity that can be supported by thelimited demographic and economic resources.

The fourth assumption is that while the various effects of the LRRT projectwill be felt region wide, their main domain of influence will be in the immediatecorridor and the CBD. Thus, the focus should be on impacts in these twoareas. Last, it was assumed that the ability of the LRRT project to revitalizethe CBD depends also upon other public and private sector policies. Theimportance of this assumption lies in the fact that development in other partsof the Buffalo region, which occur through private and public sector incentivesmay actually be conflicting with the LRRT objectives.

The methodological framework of the study was designed on the basis ofthese working assumptions. It is presented in Figure 10.1. The figure showshow major trends in the Buffalo region are combined with a set of models to


analyse the LRRT impacts. These impacts are categorized into distinct groups,which together influence the economic vitality of the CBD.

Next, we elaborate on the principal components of the methodology,underlying trends and models.

10.2.2 Principal regional trends

Above we asserted that in order to correctly examine the various impacts ofthe LRRT project on the CBD it is necessary first to analyse underlying majorregional trends. These include demographic, employment, retail patterns andtravel trends.

Population and employment trends

Table 10.1 shows the population and employment trends in Buffalometropolitan area for the period 1950–80.

From Table 10.1 it is seen that the city population decreased from a highof 580,000 in 1950 to a low of 357,000 in 1980, a reduction of over 38 percent of the total number of city residents. While many of these people have

Figure 10.1 Methodological framework of LRRT analysis.

Tabl

e 10

.1 P

opul

atio

n an

d em

ploy

men

t tr

ends

in B

uffa

lo M

etro

polit

an S

tats

tical

Are

a (M

SA),

195

0-80

a

Sour

ces:

US

Cen

sus,

Pop

ulat

ion

and

Hou

sing

197

0; 1

980;

Eri

e an

d N

iaga

ra C

ount

ies,

Reg

iona

l Pla

nnin

g Bo

ard,

197

5.

Not

es:

a M

SA p

opul

atio

n in

198

0 is

app

roxi

mat

ely

300,

000.

b Es

timat

es.

P=po

pula

tion;

TE=

tota

l em

ploy

men

t; U

E=un

empl

oyed

; BC

E=bl

ue c

olla

r em

ploy

men

t; W

CE=

whi

te c

olla

r em

ploy

men

t.


relocated to suburban areas, even on the county level there was an overallpopulation decline between 1970 and 1980 of about 100,000 residents.

Population changes are of course related to concurrent employmentchanges. Between the years 1970 and 1980 Erie County employment declinedfrom 422,000 to 403,000 employees, while in Buffalo blue-collaremployment declined from 97,000 to 80,000 employees. The netemployment figures have been affected by two phenomena: the out-migration of the labour force and the addition to the labour force ofhousehold members who previously did not work, mainly women. A furtheranalysis of the employment trend shows that in this period two major shiftsin the employment make-up have occurred. First, while blue-collar, cityemployment dropped 18 per cent between 1970 and 1980, serviceemployment in the city has increased by about 47 per cent during this 10–year period. These figures point to a trend of structural change inemployment make-up in which employment in manufacturing declines whilethat in the service sector rises. A second and related major shift is thesignificant increase in the rates of participation by women in the labourforce. In 1980, 40 per cent of the total labour force in the metropolitanarea were women compared with 35 per cent a decade ago.

The importance of these trends lies in the fact that service-related jobs,in contrast with manufacturing, are the major industry category of theCBD. While in the last decade the region has lost employment, totalemployment in the CBD has remained stable. About 50 per cent of thetotal city’s white-collar employment was in 1980 concentrated in the CBDarea (GBDF 1978).

Retail patterns

By its very nature the retail sector depends on its proximity to residential andemployment areas. Historically the CBD enjoyed central roles with regard toretail and commercial activity, even though it has increasingly had to shareits pre-eminence with the suburban areas over the decades. Over time, newconstruction of retail centres has taken place at an increasing rate at increasingdistances from the inner city area. Retail malls have become progressivelylarger, both in terms of store space as well as the parking spaces provided. Itis for these reasons that while in 1960 CBD sales were 27 per cent of citytotals and 15.7 per cent of the SMSA’s total sales, in 1977 the figures were13.1 per cent and 2.3 per cent respectively.

A factor which has mitigated this trend is that patronage of some CBDretail outlets has been rising in the 1980s mainly because of the increase inthe number of women who work at the CBD and also shop there. Nevertheless,the CBD area had over the years lost much of its retail dominance to thesuburban areas, a fact which largely explains the decline in its economicviability (GBDF 1978).


Transportation trends

The Buffalo metropolitan area has the familiar twentieth-century pattern ofcircumferential and radial expressway systems providing a high level ofaccessibility throughout the region and especially within the suburban area.With an excellent highway network surrounding and bisecting the entireregion, there are virtually no heavy congestion points in the area in peakperiods. This benefit can ironically be partially attributed to the fact that thenetwork was designed and constructed for anticipated regional populationincreases. The public transportation bus network also provides good servicemainly within the city limits. Modal split figures, depicted in Table 10.2, aretypical of most USA major cities. The overall transportation picture thatemerges is that of a high level of accessibility throughout the region by privateautomobiles coupled with a good inner-city bus transit system. The fact thatmost households live within ten minutes of auto travel time from a majorshopping centre indicates that transportation services in Buffalo were adequateat the time of the LRRT construction and provided high levels of accessibilityto all key locations in the region.

In summarizing the overall effect of these trends, three points shouldbe recognized. First, the combination of population decline, an increasinglydispersed shopping pattern and high levels of accessibility throughout theregion have focused economic development and activity location awayfrom the downtown area. Second, current shifts of employment to white-collar jobs in which women participate at high rates and the location ofthese jobs in the downtown indicate the potential of the CBD to capitalizeon these conditions through future development. Third, given the currentstate of the transportation system, it is clear that the above non-transportation trends and impacts are the crucial factors for CBDrevitalization.

Table 10.2 Per cent transit use peak and offpeak, city, suburbs

Source: Paaswell and Berechman (1981).

Notes:a City of Buffalo only (see Figure 10.2).b First ring (see Figure 10.2).


10.2.3 Models used in the impact analysis

As shown in Figure 10.1 the set of impacts of the LRRT project is dividedinto four categories for analysis. Therefore, the principal models formulatedand used in the analysis are, in part, impact specific.

The metropolitan region was subdivided into 34 zones for the purposeof the model analysis. Zonal division of the metropolitan area is depictedin Figure 10.2. (see Berechman and Paaswell 1983 for the basic data onland use, population and employment). Next we describe the four principalmodels which are used in the analysis: the economic base model, the accessability model, the shopping probability model and the urban land usemodel.

Figure 10.2 The Buffalo rail transit and development patterns in the metropolitan area.


Economic base model

In addition to nearly $450mn of federal and local investment originally allottedfor the LRRT (1978–83), another $300mn of public and private investment,which directly relates to the construction of the system, was planned. Muchof the investment in offices, a convention centre, and a transit mall tie theirorigins to a 1972 master plan for the city of Buffalo that considered thetransit system in the centre of Main Street to be the cornerstone of CBDrevitalization. The impact of these investments on the Buffalo economy islikely to be realized through changes in the labour force directly related tothe LRRT construction. These are of two types. The first is temporary andmedium-term employment created by LRRT and related constructionactivities. Essentially this the multiplier effect as shown in Figure 7.2, Chapter7. The second is long-term employment created by the LRRT system andLRRT-related new facilities like those mentioned above. Additional labourforce impacts are likely to arise in the service sector as explained below.These labour changes constitute the economic development effect from thistransport infrastructure investment. Of course they are time dependent, ascertain time lags exist between the date of an investment and the time itseconomic growth impact becomes tangible. Other impacts from the investmentrelate to regional income, which is likely to rise and, in turn, affect localexpenditures.

An economic base model was used to evaluate the multiplier impacts.3

The model asserts that new non-services employment will generate demandfor services of various types. These additional services in turn will createmore jobs, which subsequently will generate further demand and jobs. Theseadditional increments of employment, which represent the employmentmultiplier effect and were initiated by the LRRT total (direct and related)investment, are also used to compute the consequent increases in populationand in regional income.4

The accessibility model (ACCESS)

The ACCESS model was designed to measure changes in total accessibilitywithin the Buffalo area on a zonal basis following the construction of theLRRT system. Accessibility is defined here as a combination of interzonaltravel time and zonal activity level and is separately calculated for work andshopping trips. As these two trip types often occur at different times, peakand offpeak travel times by mode were used for computing changes in zonaltotal accessibility.

The model is a derivative of Davidson’s (1977) model. Total accessibilityof zone, i, Xi is assumed to be a function of three factors: intrazonalaccessibility, X

I

i , interzonal accessibility of zone i, by transit, XT

i and interzonalaccessibility of zone i by car, X

C

i Thus:5


(10.1)

The transit and car accessibility components are computed as gravity functionsof the type:

(10.2)

where aj is level of attractiveness of zone j, dij is travel time between zones iand j, and ß is an impedance parameter.

Let denote accessibility before and after LRRT, respectively. Thenchange in total accessibility for service purpose,6 DSi, for zone i, is:

(10.3)

where Oi is the number of households in zone i (denoted by Hi) times their(constant) trip rate (denoted by e), i.e. Oi=h·Hi, and g is impedance parameterfor interzonal shopping trips.

Change in total accessibility for work purposes, DWi, for zone i, is:

(10.4)

where Di, is the number of employees in zone i, (denoted by Ei), times their(constant) trip rate (denoted by e), i.e. Di=e·Ei, and l is the impedanceparameter for interzonal work trips. Total summation over all of (10.3) and(10.4) show the total region-wide changes in service and work accessibilitycaused by the LRRT project. The input data needed for the accessibility modelconsist of base year data to measure 1980 pre LRRT accessibility, X

1

i , andfuture year (1985) data, to measure post LRRT accessibility level, X

2

i .

Shopping probability model

A major indicator of the impact of the transit investment is the consequentchange in level of retail activity in the Buffalo CBD area. As explained at theoutset it was conjectured that the construction of the LRRT system connectingthe city fringe with the downtown will enhance shopping in the CBD, whichwill be manifested in higher levels of retail trips and sales.

To evaluate this hypothesized impact of the new transit system, it wasnecessary to develop and calibrate a model which simulates individuals’propensity to shop at a given shopping facility, given their socio-economiccharacteristics, the set of all available regional retail facilities and accessibilityvariables. By changing the level of explanatory variables, mainly reduced


travel times and increased retail floor space, attributed to the LRRTinvestment, it was possible to assess their impacts on shopping behaviour.

A key assumption that underlies the use of this model is that trip frequencyto a given shopping facility represents individuals’ choice probability ofselecting that facility, given all other available regional retail outlets. Anothermajor assumption is that all individuals sampled have the same choice set ofshopping facilities.

Three major determinants of retail trip frequency have been identified forthe analysis. These are: individuals’ socio-economic attributes includingincome, car availability and household size; attractiveness of a retail centreincluding its size (i.e. its floor space), number of employees and volume ofretail sales and accessibility to the shopping centre, mainly travel time.7 Theestimated statistical model is:

(10.5)

where r, s are indices of retail and residential zones, respectively. denotesthe proportion or frequency of trips to a shopping centre located in zone r oftotal shopping trips taken by individual l. Hsk is a vector of k socio-economicattributes of individual l who locates in residential zone s. Art is the level ofattractiveness of a shopping centre in zone r, as measured by variable t (e.g.retail floor space or number of retail employees). Lsr is a measure of accessibilitybetween residential locations s, and shopping centres in r. C, a, b and g areparameters to be estimated.

Notice that in equation (10.5) frequency of travel to a shopping centre inr is unaffected by the characteristics of other centres. Thus, competitionbetween centres is not directly accounted for and the possibility that CBDshopping will gain because of some changes in other centres is not consideredhere. However, by requiring that in the estimation of equation (10.5),

competition between centres is introduced indirectly. The results are indicativeof the propensity to shop at a given centre, but are not intended to displaythe complete dynamics of shopping behaviour. We are primarily concernedwith the pull that the delineated factors had on the relative magnitude ofshopping attraction at a given retail location r.

Urban land use model

As mentioned above, the large investment in the LRRT was alleged to impactland use, in addition to its impact on travel and shopping patterns. Sincechanges in the land use system are a major determinant of urban revitalizationthere was a need to evaluate such changes relative to the entire study area,particularly the LRRT corridor.

The specific model used for this purpose is a derivative of the well-known


Garin-Lowry model with a considerable number of analytical and empiricalchanges made to meet the particular needs of this study. Since the analyticalstructure and operation of this model are rather well documented in theliterature, here we will only specify the model’s allocation function used inthis analysis (see Batty 1976; Foot 1978; Berechman and Small 1988 for adetailed description).

For the purpose of allocating residential population to specific locationsthe following allocation function was used:

(10.6)

and,

(10.7)

where Tij is the number of employees travelling between work zone i andresidential zone j, Ei is level of employment at zone i, Ai is a vector of gravitybalancing factors, with Hj measuring residential attraction of zone j. Cij is theinterzonal generalized travel cost matrix and d is an impedance parameter.The service location sub-model employs a similar allocation function tocompute the number of service employees demanded by the population ofzone j who work at zone i.

The major inputs to this model included basic and non-basic employment,residential population, land availability, activity multipliers, employment andresidential density constraints and employment and shopping travel times alldisaggregated by zone (see Berechman and Paaswell 1983 for these data).

The operationalization of this model requires several phases. First there is thecalibration phase where the parameters of the allocation functions are estimated.Following calibration, the key variables within the model are altered in order totest the impact of major changes in the sub-region (LRRT corridor) and to simulatethe outcomes of policy alternatives. This is the prediction stage. Since the modelallows for the imposition of constraints on zonal activity levels to reflect physicallimitations on land availability or density policies, there is a constraint procedurewhich can be introduced at calibration or the prediction phases.

The models reviewed in this section were developed from a comprehensivepackage that provides a variety of tools for investigating the range of LRRTimpacts (Paaswell and Berechman 1981). The following section presents themajor tests performed and their results.

10.3 Tests and results

The methodology and models outlined above were used for the evaluation ofthe LRRT four impact categories, namely, investment multiplier, accessibility,shopping pattern and land use. As explained in Chapter 7 (see also Figure


7.2), the investment multiplier impact is not regarded as part of the potentialeconomic growth effect of the project. In the present analysis, however, theincrease in the labour force, which can be directly attributed to the LRRTinvestment, is computed and then introduced into the economic base modelin order to project increments of population (P) and dependent serviceemployment (S).

10.3.1 Investment multiplier results

The magnitude of the LRRT investment was $450mn (in 1978 prices), andthe LRRT-related investments, mainly in retail and commercial development,amounted to additional $300mn. Taking 65 per cent of these investments togenerate short and medium term employment, a new labour force of 5,370workers was computed. Any additional employment, if it occurred, wouldbe due to the increase in population related to this new employment which,in turn, would generate demand for additional, population-related serviceemployment.8

10.3.2 Accessibility model results

From equations (10.3) and (10.4) above, accessibility indices were computedfor each zone for the current (pre LRRT) and projected year (post LRRT).The pre-LRRT interzonal travel time matrices were computed using actualnetwork data and information on bus services including bus frequencies,location of stops and routes. The post-LRRT travel times were computed onthe assumptions that the actual light rail travel times and bus operationparameters will be as those indicated by the system’s planners. The post-LRRT information on and Oi and Dj, as well as on the parameters g and l(see equations 10.3 and 10.4) were obtained from the land-use model.

The results of measuring travel time and total accessibility before andafter LRRT, for work and service trip purposes, were used to rank the studyzones from most accessible (rank 1) to least accessible (rank 34). The resultsare given in Table 10.3.

These results elucidate a number of interesting points. First, there are majordifferences in the accessibility of a zone for work trip and service trip purposes.Not only are work trips carried out mainly at peak times while service(shopping mostly) trips are done mainly at offpeak times, but modal splitalso differs for these trips, so that more transit trips are done for work purposesat peak time.

A second point is that for both trip types there are differences in the resultantrankings between time accessibility and total accessibility. However, withineach trip type and accessibility measure there is almost no change in theranking of the zones before and after the LRRT project. The principalimplication of these results is that relative accessibility, however measured,


for a given trip type will not change significantly after the LRRT systembegins operation.

Another point worth observing is that the zones which are most accessiblefor work are suburban zones (e.g. zones 23, 24). The explanation is thatmajor highways and roads service these zones quite adequately and that thesezones already contain a relatively large number of trip-generating andattraction activities.

Table 10.3 Comparison of zone rank by time and total accessibility for work and servicetrips, before and after LRRT


The CBD on the other hand (zones 1, 2) is ranked medium for work trips,mainly because it is less accessible to car trips and is served by buses whichprovide high level service to this area, though not as good as the car does. Forservice trip purposes, the CBD zones would still have low to medium level ofaccessibility after the LRRT, mainly because most shopping trips are carriedout using the car. Note, however, the impact of activity level on totalaccessibility rank of the CBD. Whereas zones 1 and 2 are ranked 34 and 32respectively for service purposes on the time accessibility scale, they are ranked23 and 17, respectively, on the total accessibility scale. The latter scale containsthe impact of zonal activity level. Those zones in the study area with currenthigh accessibility levels gained little with the introduction of the LRRT, whilethose zones adjacent to the LRRT corridor gained the most. However, thesegains were not substantial enough to offset the high accessibility of suburbanzones already well served by an extensive highway network.

The results of the accessibility model runs lead to the following mainconclusions. First, total accessibility and time accessibility in particular willchange, but very moderately by the construction of the LRRT system.Second, given the planned route of the LRRT system, trip makers in mostzones that are not immediately adjacent to it will still use the highwaysystem or the current bus system in their daily trips. Hence it is very unlikelythat current modal split ratios will change significantly once the system isin full operation. Third, given the results regarding the actual changes inthe levels of accessibility in the zones adjacent to the LRRT route, it can beexpected that for a special portion of the public residing in those zones,there will be a significant increase in actual and perceived accessibility.Those inner city residents who do not own an automobile or who areconsidered to be transportation disadvantaged (e.g. the elderly or the young)should get the most benefit of improved accessibility first to the CBD andeventually to the region. Fourth, since travel times in the post-LRRT periodwill not be greatly improved, the only way the CBD can increase its shareof shopping and employment is if other factors which affect the level ofCBD attractiveness are improved.

10.3.3 Shopping analysis results

Recall that the shopping probability model (equation 10.5) contains variables ofthree types: households’ socio-economic characteristics, accessibility andattractiveness of a shopping facility. The socio-economic variables used were levelof household’s income and car availability. Input information for these variableswas obtained in the survey of shopping patterns in Buffalo (Paaswell et al. 1979),which included 246 observations with complete data. Location of households andtheir preferred shopping centres were also obtained from this survey. The accessibilitymeasures used in the analysis were travel times by private car and by transit. Inaddition, it was necessary to determine which areas will be included as shopping


destinations, as it was assumed that the CBD competes directly with major shoppingcentres and not with local or neighbourhood stores. Thus, for this analysis theentire Buffalo metropolitan area was divided into four ‘super’ retail areas namelyCBD, Other City, Suburban Malls and all others. The attractiveness variables selectedfor the model were: number of retail establishments in each super retail area, totalsquare feet of retail space, total number of retail employees and total retail sales.Seven principal tests were carried out.9 These were: • Test 1: the entire 246 observations were used.• Test 2: households with income categorized as low income.• Test 3: households with income categorized as high income.• Test 4: households which shop at the downtown and live inside the city.• Test 5: households which shop at the downtown, but live outside the

city.• Test 6: households which shop at suburban malls and reside in suburban

zones.• Test 7: households which travel less than 20 minutes for shopping. The results of these tests, in terms of the estimated parameters of model(10.5) are reported in Table 10.4.

Several key points should be observed about the results from these tests.In all seven tests the value for the accessibility coefficient d is insignificantand small. It should be noted that 91 per cent of the sampled households live

Table 10.4 Parameter estimates of the shopping probability model (10.5) seven tests

Notea Because of the importance of these coefficients for the analysis we report their esti-

mates and standard errors in brackets. None is significant at 0.01 or 0.05 level.

a = Household’s income coefficientb = Car availability coefficientg = Coefficient of a shopping centre attractiveness variabled = Coefficient of an accessibility variableC = Regression’s constantF = F test for regressionM = Number of observations for a given runNS = Not significant at 0.01 level


less than 20 minutes from their selected areas of shopping. Given thedistribution of shopping centres in the Buffalo region, travel time variableswere found not to affect households’ shopping choices. Improved travel times,therefore, would not alter prevailing shopping choices, holding constant allthe other independent variables.

The only variable whose coefficient is significant for all of these runs islevel of attraction of the shopping area. This result is extremely important inthe context of the present analysis. It implies that any increase in the share ofthe CBD in the regional retail activity through the LRRT is conditional uponthe degree to which the system will increase the CBD attractiveness as ashopping area. It may do so by generating more and better retail outlets butnot by affecting the CBD’s relative level of accessibility. Hence, the aboveresults from the accessibility model, which indicate that the LRRT will havelittle if any impact on the relative accessibility of the CBD, may not diminishthe importance of the LRRT system as a potential factor in improving theCBD attractiveness. We, next, elaborate on this result by presenting the resultsof the land use model.

10.3.4 Land use model results

Following calibration, the model was used to predict land use impacts ofpotential changes generated by the LRRT system. The results of two runs,which directly relate to our objectives here, are shown below. The first is a testfor the impact of the LRRT’s direct investment only. The second is a test for theoverall impact of the project including changes in zonal attractiveness, inaccessibility and taking into account the LRRT direct and related investments. 1 Prediction of LRRT’s investment effect: the direct effect of the investment

is to increase basic employment. The additional basic employment ascomputed from the economic base model was introduced into the landuse model and the resultant vectors of population and employment wereobserved. These results, in terms of per cent changes from original (baseyear) data, are reported in Table 10.5.

2 Prediction of the LRRT’s overall impact: in this prediction run the effectsof the total investment, of changes in the zonal level of attraction foractivity location (mainly services) and of changes in accessibility weresimultaneously introduced into the model. The predicted per cent changein the distribution of population and employment are reported in Table10.5.

These results indicate that the impact of the investment factor alone is toincrease service employment in the CBD (zones 1 and 2) while reducingresidential population there. However, since total population in these zonesis currently small, this latter effect can be ignored. On the other hand, base


year level of service employment in the CBD is the highest in the region(45,687 employees). Thus, the predicted 34 per cent increase implies anaddition of 15,330 employees which in itself is larger than any amount ofservice labour at any other zone, at the base year. The planned commercialand retail facilities in the CBD discussed above are the main sources for thisnew service employment.

Table 10.5 Results of the two prediction runs of the land-use model


The predicted population distribution from the total impact of the LRRTsuggests a similar decline of residents in the CBD. A more dramatic result is thelarge increase in population in major suburban zones (e.g. zones 21, 22, 23, 27).This additional population, which is generated by LRRT-related employment, isdistributed by the model to these zones for reasons similar to those which underliethe long-run trends of suburbanization. These include greater availability ofresidential land, lower densities and high levels of accessibility.

With regard to service employment, the results of the total impact testagain show that the CBD is bound to benefit most from the LRRT system.This large gain (over 100 per cent) is mainly because of the increase inattractiveness of zones 1 and 2 for services, which corroborates the resultsobtained before from the other models. The new service employment isconsistent with anticipated retail, commercial and cultural development there.In addition, the analysis shows a high per cent increase in service employmentin zone 18, which is located at the northern end of the LRRT route (seeFigure 10.2). This also indicates the positive effect of the system on zonalattractiveness, but because of the relatively small number of service employeesin the base year in this zone the implications of this result are limited.

The final set of results to be observed pertains to the impact of LRRT onzones immediately adjacent to its corridor. With regard to population, thecurrent trend of population decline is not going to be reversed. Moreover,these adjacent zones are also not going to benefit much from the LRRT withregard to service employment. In contrast to the CBD zones, where the bulkof the physical development is planned, only very limited LRRT-relateddevelopment will occur in those zones. When coupled with virtually no changein accessibility levels after the LRRT implementation, this lack of developmentexplains the results for the adjacent zones.

10.3.5 Overall downtown results of the LRRT project

From the results presented in this section the following major conclusionscan be stated. 1 The downtown’s economy: the LRRT is expected to have a positive impact

upon the economy of the downtown. This result will transpire byattracting additional service employment to the CBD area and iscontingent on the implementation of additional private investments, someof which are related to LRRT investment.

2 Accessibility of the downtown: the accessibility of the downtown area toall other zones in the region by mode and trip type will not changesignificantly after the LRRT construction.

3 Shopping at downtown: if, as projected, the LRRT will have positiveimpact on downtown attractiveness, a larger share of the regional retailtrade will be captured by the CBD, all other factors remaining the same.


10.4 The Buffalo LRRT: postscript10

In the early 1960s the Niagara Frontier Transportation Study, one of a numberof seminal computer-assisted regional transportation studies, forecast the needfor significant capacity improvements in a number of corridors serving Erieand Niagara Counties in New York State. Influenced by the strength of theauto, steel and chemical industries and grain mills that were the backbone ofthis manufacturing region, the planners believed all that was in the area’sfuture would be growth of employment, population and the services whichsupport a healthy economy. The city of Buffalo had a population in 1960 of530,000 and the two counties’ area of 1.2 million. A product of this planningprocess, although more than a decade and a half later, was the Buffalo LightRail Rapid Transit system (LRRT).

The LRRT, as opened finally in 1982, was a six-mile line serving the cityof Buffalo along the ‘Main Street corridor’. It should be noted that by 1980the city had begun a population decline, although population in the regionwas stable. The city was feeling the familiar effects of suburbanization—losing both people and jobs to the surrounding areas. Transit ridership on thebus only transit system was over 28,000,000 trips per year. The LRRT wasexpected to add 80,000 trips per day to this total. At that time, transit hadcaptured fewer than 10 per cent of regional trips. The average travel time towork by all modes was about 20 minutes and by transit 30 minutes.

During the next decade and a half Buffalo suffered a major loss inemployment. All of the heavy manufacturing—steel, tyres, automanufacturing—closed shop. Local government, the universities and otherservices picked up some of the employment slack but by 1992, the populationhad dwindled to 323,000. The thirty-eighth largest city in the country in1980 had become the fiftieth largest by 1992. Its home county, Erie, hadgone under one million for the first time in 40 years. Buffalo had lost a baseof support for its transit—the basic population. The 180,000 jobs supportedin the city in 1980 became 147,000 by 1992. In 1960, when planning begunon the transportation needs, more than one of every two jobs in the regionwere in the central city; by 1992 this had dwindled to less than one in three.

In addition to losing population and having jobs become more suburbanand dispersed, planners and decision-makers in Buffalo made a number ofcritical decisions that mitigated any potential success of the Light Rail. Theseincluded: 1 Keeping the CBD viable and attractive to cars. Parking policies were

enacted to ensure easy access of cars to activities throughout the CBDand substantial new parking capacity was added along streets thatparalleled the LRRT.

2 No retail incentives were enacted, especially to attract high-end shoppers.In fact the city was inactive as developers put in place several million sq


ft of new retail malls away from the CBD and away from the LRRT (orany transit).

3 New highway capacity was added to bring people from the suburbs tothe CBD. In fact, some of this capacity competed directly with the LightRail.

4 As a result of population declines and highway capacity increases, averagetravel time to work actually declined to less than 19 minutes.

5 To make the LRRT ‘work’, bus routes were changed to feed the LRRT.Many of these one-seat rides from suburb or outer parts of the city nowbecame two-seat rides, as passengers had to transfer.

6 The NFTA and city did adopt a downtown free fare zone. However, thisput significant pressure on the operating budget (the LRRT returnedonly 20 cents on the dollar of cost from the fare box). As a result therewere many service cuts, making the system less attractive.

Today, NFTA has 28 million annual transit trips, and the LRRT wasassumed to almost double this figure. While more than nationally promotedPortland, Oregon rail ridership, it is far from Calgary’s 140,000 dailyriders. Buffalo has replaced some of its old industries with a baseballstadium and hockey arena, but little else draws people to the CBD. Thesystem, designed for an active region and dynamic core, must now serve acar-oriented suburban region and hope for the day when the Buffalo CBDis rediscovered.

10.5 Key lessons from the Buffalo LRRT project

The ex ante asserted significant impact of the LRRT on the attractiveness ofthe downtown as a centre of retail, commercial, cultural and entertainmentactivities is mainly due to a planned massive physical development in a well-defined small area. But attractiveness of an area is also determined by featureslike variety and quality of activities, personal safety and comfort, ease ofaccess to retail outlets, upgrading of existing facilities and, most importantly,by competition from other accessible and attractive areas in the region. Allthese issues are affected by urban public policies and, as a result, it is apparentthat proper private and public sector programmes are necessary to inducethese features. The lack of regional or even citywide co-ordination of policiesto ensure the attainment of the LRRT objectives is probably the most seriousthreat to the revitalization of the downtown. Conflicting highway, parking,transit and zoning policies are examples of this problem.

Since the LRRT route does not extend to the suburbs, suburbanites whowish to use the system need to drive and park their cars at the nearestLRRT station. Given that very few parking facilities have been plannedand that the highway system provides adequate access to the heart of theCBD, no significant changes in modal split can be expected. Moreover,


express buses, which operate between suburban zones and the downtown,also reduce the effectiveness of the new system. Still another example of theconsequences of lack of policy co-ordination is the concurrent developmentof retail and commercial facilities in suburban zones. Even though thisdevelopment may not be as massive as the planned development at thedowntown, the proximity of these facilities to residential population is likelyto counterbalance the positive impacts of the LRRT on the attractivenessof the CBD.

The Buffalo transit development was consistent with a 1972 downtownplan that linked CBD access via rail to downtown renewal. However, relatedinvestment projects like new hotels, new office buildings and a pedestrianmall, which integrate travel with adjacent activities, were not begun untiltransit construction funds were secure. In this sense, more than many otherurban projects, the impact of transit investment and the stimulus that transitprovides through its interaction with other activities are seen as the maindeterminants of CBD revitalization. Thus, the key general lesson from thisstudy is that improved accessibility per se is neither a necessary nor sufficientcondition for such a goal as CBD revitalization. A corollary to this conclusionis that capital transport investment must be targeted and the implications ofthe investment, land development, employment increase and location of servicefacilities must be clearly determined and enforced through complementarypublic policies.

10.6 Empirical evidence from other rail projects

The study of the Buffalo LRRT project was an ex ante analysis. In this sectionwe review empirical evidence from existing high-speed, inter-city rail projects.While the main effect of such projects is reduced travel time between cities,here we focus on their potential benefits from enlarged labour market areasand increased economic activity around the stations. We review the evidencefrom Japan and France, where high-speed rail links have been constructed.We also examine impacts from the California BART system. The focus hereis on the local and regional impacts in terms of the development effects.

10.6.1 The Japanese Shinkansen high-speed rail system

The Japanese Shinkansen have been in operation since 1964 when the Tokaidoline was opened between Tokyo and Osaka. (See Figure 10.3.) Systematicanalysis has been carried out on population and employment changes overtime (1975–85), but with mixed results. Nakamura and Ueda (1989) foundthat higher than expected population growth took place in three of the sixprefectures where the Shinkansen stations were located. More detailed analysissuggested that the greatest impact on population growth took place in locationswhere there was both a Shinkansen station and a high-speed road. However,


the causality of the relationship was questioned as to whether the Shinkansenstation was leading growth or growth would have taken place without theShinkansen (Sands 1993). Multivariate analysis was used to unravel some ofthe relationships. The study concluded that ‘growth caused by the Shinkansencould be predicted for a region with 90 per cent accuracy’ (Sands 1993:260).It seemed that there were three principal conditions needed for growth. First,a high incidence of information exchange industries (business services, bankingservices, real estate) is required. This result seems to support the conclusionof Chapter 7 where agglomeration economies are seen as a prerequisite fordevelopment. Second, sufficient opportunities for higher education(universities) must be there. The third condition is good accessibility to aShinkansen station. Growth rates in retail, industrial, construction andwholesale sectors were 16–34 per cent higher in cities with a Shinkansen

Figure 10.3 The Japanese Shinkansen high-speed rail system.


station than in those cities without (Brotchie 1991). Yet, these high net figuresmay conceal a strong reallocation process of activity within a city from existingcentres to new station-accessible centres.

More localized studies have attempted to examine the effects of stationlocation on development (Amano and Nakagawa 1990). The originalTokaido line had twelve stations, with three completely new stations locatedat peripheral sites near to the city. The other stations were located in citycentres. The development impacts at the existing stations were minimal,but the new stations located at peripheral sites had substantial local impacts.At Shin Yokohama, some 15 miles south-west of Tokyo, a new station wasopened with little initial impact. It was only when a new underground linewas built to central Yokohama (4 miles away and cut the journey timefrom 30 minutes to 12 minutes with a high frequency service) thatdevelopment took place and the station became an important stopping pointon the Shinkansen. Perhaps this is a manifestation of the work of underlyingnetwork economies. Only limited development took place on the Yokohamaside of the railway (the south side) because local residents have resisteddevelopment. On the non-Yokohama side of the railway (the north side) anedge city has developed with new offices in a highly accessible location(Hall 1995). The impact of the Shinkansen is clear at local level as populationand employment growth rates are consistently higher in areas with aShinkansen service than in those without, particularly in the informationexchange sector and in the hotel and food service sectors (Sands 1993:268).Limited evidence has also been found on land value increases. Nakamuraand Ueda (1989) found that land values in commercial areas rose by 67 percent with a Shinkansen station.

10.6.2 The French TGV high-speed rail system

The French Le Train à Grande Vitesse (TGV) system has been in operationsince 1981 when the line between Paris and Lyon was opened. Connection tothe TGV network has been seen as a ‘boost to the local economy’ (Bertolini1998:166), but this impact may only be a redistribution as appropriate localeconomic conditions are also important (Troin 1995). On the TGV-PSE (Paris-Sud Est to Lyon), there were three new stations at Lyon Part-Dieu, Le Creusotand Maçon. It was only in Lyon that significant growth took place, as therewas a heavy demand for office space around the new Part-Dieu station. By1990, the TGV station had 40 per cent of the city’s total office space with afurther 60 per cent of new projects. This amounted to a 43 per cent increase(1983–90) to some 250,000 sq m. The explanation was that the new stationhad good local access (as well as being two hours from Paris), with a highvisibility and level of convenience for customers. In addition, there was littleavailable space in central Lyon, so relocation was an attractive option forhigh tech service industries needing good access to Paris. Business trips by


rail increased by over 50 per cent (Pieda 1991). These local advantages werenot available in Le Creusot and Mâcon so little TGV-related activity tookplace. (See Figure 10.4.)

As the TGV network has been extended, the impacts elsewhere were alsovariable. Substantial growth has taken place at Le Mans, Nantes and Vendômeon the TGV Atlantique where local conditions were buoyant. A 20 per centrent premium is typical in these new development locations. The TGV links

Figure 10.4 The French TGV high-speed rail system.


have reduced travel time substantially to Paris (e.g. Vendome to Paris isnow 42 minutes by TGV when previously it was 135 minutes). They haveraised the image of the newly connected locations and opened up newlocations for businesses to consider location (or relocation). However, asnoted above, development has been inconsistent across station locations,as impacts have been variable and highly localized. The extent ofdevelopment has depended on the overall economic strength of the localeconomy and the presence of service sector firms requiring access to Paris(Sands 1993).

The clearest example of all these elements coming together is in thetechnopole concept at Euralille.11 A new high-speed railway station hasbeen built as a hub for the TGV network in northern Europe. This, inturn, allows decentralization from Paris (60 minutes away), the provisionof a service function to Brussels (30 minutes away) and a businessbridgehead on the continent for London (at present 120 minutes away,but eventually 90 minutes away). In addition, there are plans to link therail investments with international airports at Paris-Roissy and Brussels-Zaventem and to expand the local Lille-Lesquin airport (Ampe 1995). Atthe new Lille Europe station there is a business/recreation complex,including a trade centre, congress, exhibition and concert facilities. Newhotel accommodation is also being provided, with a large retailing centre,leisure facilities and public support services. The Lille Business School isbeing relocated here, together with new housing. The novel aspect of thiscombined transport and development package is that the investments haveproceeded simultaneously. The management and finance have been carriedout through the French system of Société d’Economie Mixte (SEM), whichis based on partnerships between the public sector (mainly theinfrastructure investment) and the private sector (mainly the associatedbusiness, retailing and housing investment).

The main conclusion from this brief review of two high-speed railinvestments is that the impacts are found both at the network and local levels.The network effects relate to the substantial increase in accessibility to thekey national and international markets. The local impacts are more variableand relate to the presence of a buoyant local economy that can take advantageof the new opportunities offered by high-speed rail accessibility. Impacts areapparent mainly in the service and high-tech sectors, but also more generallyacross all business sectors. Image is important as is the availability of supportservices and facilities, such as those found in Euralille. Local road and railconnections also make a difference, particularly if the high-speed station islocated at the periphery of existing urban areas (e.g. Shin Yokohama andEuralille). Finally, the need for supporting public policies at all levels cannotbe over-emphasized.


10.6.3 The San Francisco BART system

A unique opportunity for examining the impact of an urban rail investmentis provided by the Bay Area Rapid Transit (BART) system in San Francisco,which was opened in 1973 and now has over twenty years of data to assessits influence. (See Figure 10.5.) Early studies indicated that BART had a modestinfluence on land use and urban development, directly by improvingaccessibility and indirectly by inducing policies supportive of compactdevelopment (e.g. incentive zoning and redevelopment finance). It was alsoconcluded that it was only where the necessary supporting conditions werepresent (e.g. buoyant local economy) that growth actually took place (Knightand Trygg 1977).

The main conclusion reached with twenty years of evidence is that BARThas had a very modest impact in a highly localized and rather uneven pattern.These are similar conclusions to the impacts of high-speed rail in France andJapan. The buoyant centre of San Francisco has been able to grow andmaintain its dominant position in the urban hierarchy. The centre of Oaklandhas also benefited through its attractiveness for public and private investment,part of which is due to the high levels of public transport and road accessibility.Walnut Creek has attracted office development, Pleasant Hill has 1,800apartments and condominium units within 400 m of the BART station andFremont has a mix of transit-oriented developments (Cervero and Landis1997). It should be noted that each of these last three stations is locatedtowards the end of the BART network, where little existing developmentwas in existence in 1973.

Overall, the effects of BART seem to have been to maintain the dominanceof the centre of San Francisco. Over 75 per cent of all office constructionwithin 800 m of BART stations since 1973 took place here (40 million sq ftof office development in downtown San Francisco, compared with 12 millionsq ft elsewhere in San Francisco). However, outside San Francisco, evengreater office development took place near road-accessible suburbs likePleasanton and San Ramon rather that in the rail accessible locations.Overall, 35 million sq ft of office space was built in areas unserved byBART (since 1973), compared to 9 million sq ft within 800 m of an EastBay BART station.

In the housing sector provision has taken place in the BART accessiblelocations. But again this may have resulted from a policy decision to encouragetransit-oriented housing developments (Cervero and Landis 1995b) and thisemphasizes the pivotal role that the public sector has in supporting stationarea development (Cervero and Landis 1995b). So BART is clearly not asufficient condition for station-related development to take place, but giventhe appropriate circumstances it can make an important contribution. Theempirical evidence suggests that the non-BART served corridors had a 20 percent greater increase in population than the BART served corridors.


The same picture is apparent for employment growth. BART’s locationaladvantage is clear for the FIRE (finance, insurance, real estate) and non-business service sectors, particularly for the Richmond (the north) and Fremont

Figure 10.5 The San Francisco BART transit system.


(the south) lines. This advantage is reflected in higher office rents immediatelyaround stations.

The inevitable conclusion reached is that urban rail transport and landdevelopment have only a tenuous link. Where there are impacts, it seems thatthey are highly localized and in specific sectors. Even where development isspecifically targeted at accessible rail locations, the impacts on patronage arelimited. When these rather modest changes are placed within the city-widecontext of change over a twenty-year period they become even less significant.The rather pessimistic view portrayed by Webber (1976) twenty years agoabout whether BART would ever live up to expectations, either as a successfultransport system or as a catalyst for urban development and economic growth,has been realized.

10.7 Conclusions

Many rail investments are justified both as the basis of transport benefits(principally time savings) and economic development benefits (e.g. new jobs,higher rents, growth in property prices). Transport benefits alone, in manycases, are not sufficient to justify the enormous investment. Yet the analyticaland empirical evidence suggests that the development impacts are not uniformand only occur where other economic conditions already favour development.These investments do not act as the catalyst for change, but they can act toreinforce a change that is already taking place (or is likely to take place). Ifthe development impacts of rail investment are examined in the wider contextof urban change and development related to urban road investment, thentheir impact is further reduced. The San Francisco evidence is particularlyclear on this. Employment growth and residential growth was some 20 percent higher in those corridors served only by freeways rather than those servedby BART. One possible explanation for the small and variable impact ofurban rail investment is the condition of ubiquitous accessibility found inmany urban areas, often reinforced by cheap access to the private car. Anyadditional infrastructure, particularly where the network is already welldeveloped, makes very little impact on the overall accessibility. As the roadsystem is always going to be more available than the rail system, it is notsurprising that road accessibility dictates where growth takes place.

With high-speed rail the situation is somewhat different, particularly inJapan where population densities are quite high. Rail transport dictatesdevelopment patterns in Tokyo and the other major Japanese cities asemployment is highly centralized with limited parking, making land values afunction of travel time to Tokyo station and other cities (Wegener 1995). Inaddition to the accessibility to central employment, high-speed rail stationsalso act as strong localized attractors for local employment centres. But hereagain the local conditions do not automatically lead to development andemployment opportunities taking place. The necessary economic conditions


have to be present and public sector involvement is necessary to attract interestfrom the private sector.

Finally, the Buffalo LRRT project provides a generic lesson for all railprojects. That is, all of the potential positive impacts of the investment on thetarget area (Buffalo’s CBD) at best amount to a necessary but not a sufficientcondition for economic revitalization. In this case the principal reason is thatnon-LRRT public policies may enhance or conflict with these benefits andthus reinforce or depress CBD revitalization by the LRRT. Providing a highlevel mass transit capacity such as the LRRT system, which is capable ofbringing a large number of people to the CBD in short intervals, is one thing.What these people will do there is another issue.

Notes

1 The city of Buffalo is within Erie County and Buffalo SMSA is made up of Erieand Niagara Counties. The study area is all SMSA, but major results of theanalysis discussed below pertain mainly to the metropolitan area (Erie County)and the CBD (down town) area (Figure 10.2).

2 Between 1980 and 1990 Buffalo population has declined further to the presentlevel of 330,000 inhabitants.

3 See Oppenheim (1980) for a review of this model.4 An independent input-output model for Buffalo (Dickson 1978) gave values for

the economic base model’s coefficients.5 In the Berechman and Paaswell (1983) equation 1 was erroneously presented as

a linear function of these components though it was not computed that way.6 In this analysis retail activities were used as a surrogate measure for service

activities.7 The selection of these variables emanated from a separate study (see Paaswell

and Berechman 1981).8 These calculations are based on labour productivity rates of capital expenditures

on labour and material (i.e. number of employees per $1mn investment) and theestimated yearly capital expenditures (see Dickson 1978; NFTC 1977 for data).This employment figure was then introduced into the economic base model usingthe 1980 county ratio of 2.5 residents to employee. Based on these figures it wasexpected that total net increase in population due to the LRRT would be in therange of 19,000 to 25,000 people, while the net increase in total employmentwould be 7,000 to 10,000 employees. In terms of income generated, the abovecapital expenditure was expected to generate $l,040mn in regional income overthe LRRT investment’s period (1978–85).

9 For all tests retail floor space was used as a measure of attractiveness and theminimum of auto or transit travel time as the accessibility variable.

10 Prof. Robert Paaswell, Director, University Transportation Research Center, theCity College, New York, has written this section. We thank him for hiscontribution.

11 The technopole concept relates to the concentration of transport modes andfacilities with the development of technology parks and investment in high qualitytelecommunications infrastructure all in one location.

The economic impacts ofairports

11.1 INTRODUCTION

Since its inception in the first half of the twentieth century, commercial airtravel has grown rapidly. In the 1970s and 1980s global air travel grew ataround 6 per cent per annum, far outstripping growth rates in the globaleconomy during this period. It is anticipated that this trend will continue forsome time, before falling to a lower level of between 2 and 3 per cent.

Recently, the growth rates have been further encouraged by growth inincome levels, the globalization of economies and the demand for long-distanceleisure travel. In addition there has been a fundamental restructuring of theairline industry. At a domestic and international level aviation markets havebeen or are being deregulated allowing airlines greater freedom to determinetheir own route network, while state-owned assets have been privatized,creating a more commercially driven environment. This has intensified thedevelopment of hub and spoke operations and in recent years the movetowards the creation of global airlines, through code sharing, alliances,marketing partnerships, shared equity and acquisitions.

Both growth and restructuring have had an impact upon the provision ofinfrastructure. Simultaneously, airports must handle an ever-growing numberof passengers and freight, as well as respond to different demand patterns. Involume terms this pressure is particularly intense at existing hub airports,particularly those which perform an international role. In 1993 there were29 airports in the world with more than 20 million passengers per annum. Atpresent, airports plan to spend some £25bn (over US$ 40bn) worldwide onairport development to meet the expected increase in demand. In the transportcontext, air travel is the most dynamic growth sector.

In this chapter we focus on the economic impacts of airports, principallyconcentrating on the employment effects and the impacts on the local economy.This involves the identification of the means by which these impacts can beisolated and measured and in greater detail analysed through two case studies.These case studies focus on London Heathrow, in particular on the currentdecision over the development of Terminal 5 (T5) and on Manchester where

11


construction of a second runway is now under way. The material presentedin this chapter is from secondary sources, with a clear focus on the economicdevelopment impacts of transport investments.

Andrew and Bailey (1996) identify six different types of airport by passengernumbers and traffic mix and detail their economic impacts (see Table 11.1).The table shows that airports with similar volumes of traffic can generatequite different economic impacts because of fundamental differences in theirtraffic mix and because each segment of traffic sustains a different number ofjobs. In turn an important determinant of the traffic mix will be the size anddiversity of the local economy. Heathrow’s location in the South East ofEngland, where approximately one-quarter of western Europe’s thousandlargest companies have their headquarters (Pieda 1995), may be related tothe high number of business passengers using scheduled flights from this airport(Andrew and Bailey 1994).

We begin with a general discussion of the economic impacts of airports(Section 11.2). Subsequently, in Section 11.3, we analyse the case ofTerminal 5 in London Heathrow airport, followed by analysis ofManchester’s second runway project, in Section 11.4. Major conclusions arepresented in Section 11.5.

11.2 The economic impacts of airports

Airports have a major impact on their locality (up to 15 miles) and morewidely in the regional and international contexts. IATA (1991) suggests thatthe economic benefits of airports include wealth creation, employmentgeneration, tax payments, travel and tourism and regional economic benefits.It should be noted that airports are also tourist attractions in themselves. Forexample, Gatwick is the fifth largest tourist attraction in the South East ofEngland and has plans to develop a theme park and a visitors centre. TheIATA (1991) report puts a global figure of US$710bn (1989) on the totaleconomic impact of the air transport business, of which about 30 per cent isaccounted for by direct expenditure and indirect expenditure, and theremainder is induced expenditure. The industry supports some 21.6 millionjobs worldwide (1989) and this will increase to 30 million in 2010.

Before we proceed with the appraisal of the employment impacts of airportinvestments, it is important to place this analysis within the context of theanalytical framework proposed in Part III. In Chapter 7 we have distinguishedbetween employment impacts from the investment multiplier and employmentimpacts from the primary (transportation) benefits of the project (see Figure7.2). The former is mainly the result of financial leakage into the local economyand is a function of the magnitude of the investment rather than its type (e.g.a transportation versus a health project) and is a short-termphenomenon.The latter effect, sometimes referred to as the employment multiplier, issupposedly a ‘technical summary’ of the labour market equilibrium process

The economic impacts of airports 289

(see Section 8.3), which is represented by a single number or a set of numbers,one for each labour category. In many cases employment multiplier numbersare obtained from input-output tables or from specific technical studies.

Analytically, the use of employment multipliers assumes that a linearrelationship exists between the volume of airport traffic and the consequentlevel of employment, and that this relationship holds for all airports irrespectiveof size or function (the mix of passenger and freight, national or international).This assumed relationship further presupposes the existence of two implicitbut largely implausible assumptions. First, there are no economies ordiseconomies of scale in airport operations so that labour increases in directproportion to the increase in the volume of traffic. Second, labour productivityat airports remains constant, even though there is evidence showing thatovertime labour productivity at airports, measured by passengers throughputper staff employed, has improved. An additional limitation on the use ofemployment multipliers is a lack of pertinent data to continuously updatethem. They are considered stable over time, irrespective of technological andtemporal changes in labour markets.

Given the above, it follows that employment multipliers are, at best, onlycrude measures of the employment impacts of airports’ development, whichprobably overestimate the true level of employment. Moreover, the applicationof a single multiplier number to all, or to very large categories of employment,adds to the inaccuracy of the employment results. Given these qualifications,the calculations of the expected increase in airport activity from future increasein the demand for air travel and from the construction of facilities like Terminal5 to accommodate this demand are based on the use of such employmentmultipliers.

A further step in the analysis of the labour impact of airports is to defineand classify the types of jobs generated by airports. By disaggregating overallemployment figures, attempts can then be made to understand how differenttypes of job relate to each other, to effectuate the employment function of theairport and the wider economy. By quantifying these relationships, modelscan then be developed to predict the employment implications of future airportdevelopment. Many studies, both in the UK and USA, have undertaken thisexercise as part of larger assessments of the economic importance of airports.There seems to be a broad consensus as to the types of jobs that airportssupport (CLES 1988; York Consulting 1991; Andrew and Bailey 1994). Table11.1 presents this classification of airports by traffic mix and type of economicimpacts.

Given this characterization of airport types we can identify four classes ofemployment that they generate: • Direct employment is created through the expenditure required to

maintain the transport function (and ancillary services) of an airport.These jobs are entirely dependent upon the presence of the airport and

Tabl

e 11

.1 A

cla

ssifi

catio

n of

Eur

opea

n ai

rpor

ts b

y tr

affic

mix

, siz

e an

d ec

onom

ic im

pact

s

Sour

ce: A

ndre

w a

nd B

aile

y (1

996)

.


are supported by revenues produced by the airport operator, airlines,passengers and freight. Direct employment is normally divided into thosejobs which are on the airport site and those located outside the airportarea.

• Indirect employment is generated by the subsequent airport-relatedpurchases made by organizations in the above category. There may befurther ‘rounds’ of expenditure from first level to second or third levelsuppliers.

• Induced employment is supported when further non-airport relatedpurchases are made in the wider economy by those employed in the firsttwo categories.

• Attracted employment is present when activities unrelated to the operationof the airport, nevertheless locate nearby to take advantage of increasedaccessibility and of agglomeration economies. This includes inwardinvestment to the region around the airport and tourist developments.

The relationships between these categories can be summarized in anemployment impact model, that links airport and airline revenues to thediffusion of expenditure through various types of interrelated organizations,which in turn creates direct, indirect and induced employment. Positionedalongside this expenditure-employment flow is the attracted employmentcategory, where employment levels are not determined by input levels ofexpenditure (Figure 11.1).

Source: Derived from IATA (1991) The Economic Benef its of Air TransportFigure 11.1 Employment impact model.


Indirect and induced employment are normally calculated with the use ofmultipliers. The attracted employment is analyzed more generally throughqualitative interviews with firms in the main sectors involved. In calculatingairport-induced employment, we should consider the ‘crowding out’phenomenon. That is, employment can occur through the backward linkages(indirect and induced employment) where jobs generated by the airportinvestment lead to pressures on the labour and capital markets, resulting inadjustment and possible reductions in employment or investment in othersectors. Similarly, the forward effects (attracted employment) result inrelocation of new firms near the airport and greater than expected growthrates in those firms, resulting in cost reductions and spatially differentiatedgrowth rates. Again, crowding out may take place as existing economicactivities are forced to locate away from the airport. The difficulty here is inidentifying the dynamics of these various processes, which are all taking placesimultaneously and over a period of time (Rietveld et al., 1998).

Turning now to employment and non-employment related economicimpacts, Andrew and Bailey (1996) suggest analysis at four levels to identifythe precise contribution of airport development to a region: 1 Analysis of the structure of the region’s economy, together with the role

that the airport plays within it. This links in with the classification andfunction of the airport (Table 11.1).

2 Isolating and analysing the significance of each airport infrastructurevariable that supports the regional economy.

3 Assessing the economic impact of the airport business in terms of thedifferent types of employment and purchasing effects identified above.Usually this entails application of employment and income multipliersand building in allowances for industry productivity.

4 Adding the social dimension, given the airport’s potential effects on thewider labour market and the cultural activities within the region.

Based on this framework, we now examine the economic development impactsof a major form of airport infrastructure investment—a new terminal.

11.3 The economic impact of terminal 5 atHeathrow airport

11.3.1 Introduction

The 1985 Airports White Paper (UK Department of Transport 1985) statedthat aviation in the UK contributes 85,000 jobs and some £500mn (about US$900mn) to the exchequer (the treasury). Nearly two-thirds of foreign touristscome to the UK by air and this is a major growth sector. Expenditure by airtravellers visiting the UK together with civil aviation earnings amounted to


over £6bn in 1984. UK export efforts rely heavily on air transport with some£27bn of visible trade passing through UK airports (1984), three-quarters ofit through Heathrow.

Over the last ten years, air passenger traffic in the UK has grown by over60 per cent from 70 million passengers (1984) to 112 million passengers(1993). Within this market, Heathrow is in a dominant position, accountingfor some 42 per cent of all passengers and over 60 per cent of freight (Tables11.2 and 11.3). Gatwick comes second on both counts, with Manchester in adistant third position. Heathrow is also in a leading position within Europeand is the world’s largest international airport (Table 11.3). The airport hastwo major impacts, one as an employer and the other as a facilitator ofincreased activity within the regional and national economies. In this sectionwe concentrate on the impact of the proposed new terminal (T5) onemployment, but we also cover some of the wider impacts (Section 11.3.4).

11.3.2 London Heathrow airpor t

Heathrow airport is located 25 km west of London on the fringe of the built-up area. Originally conceived as a military airfield, it has been London’s pre-eminent civilian airport ever since opening as London airport on 25 February1946, because of its close proximity to the city and relatively good road andlater public transport links to surrounding areas. However, this location has

Table 11.2 Heathrow airport in the UK context, 1993

Table 11.3 Heathrow airport in the EU context, 1993


also made new development at Heathrow controversial over the years. It isbuilt on prime agricultural land and lies within the protected strip of openspace (the Green Belt that encircles London) where development hastraditionally been restrained. Perhaps more critically, its main flight pathsare directly above the urban mass of London to the east because of prevailingwesterly winds. Thus, large numbers of people are exposed to the increasednoise nuisance that may result from airport expansion.

Aware of these problems from an early stage, consecutive governmentshave commissioned a plethora of reports over the years to investigate theissues surrounding airport development in London and South East Englandas a whole (Tables 11.4 and 11.5). In many cases, these have advocated therestraint of Heathrow, with growth directed elsewhere. Most notable of thesehas been the desire to distribute growth throughout the South East byimplementing a structured airport expansion policy, underpinned by thecreation of a third London airport (ultimately at Stansted), thereby limiting

Table 11.4 London area airports: traffic, employment, productivity andaverage passengers per aircraft

Source: From various published statistics.

Table 11.5 Employment in the South East of the UK for standard industrial categories(SICs). in 1991

Source: NOMIS.

pax/

emp=

Pass

enge

rs p

er e

mpl

oyee

.?P

=Per

cent

age

chan

ge in

pro

duct

ivity

.So

urce

: And

rew

and

Bai

ley

(199

6).

Tabl

e 11

.6 D

irect

on-

airp

ort

empl

oym

ent

at H

eath

row

, Gat

wic

k, S

tans

ted

and

Luto

n ai

rpor

ts b

etw

een

1980

–93,

incl

udin

g la

bour

prod

uctiv

ity c

hang

es (

1980

-91)


development at London airport (ultimately at Stansted), thereby limitingdevelopment at Heathrow and London’s second airport, Gatwick.

However, development at Heathrow has endured the test of time. Despitethis ongoing debate to limit its growth, the airport has expanded in its firsthalf century to four terminals and its operations are entrenched in the physicaland economic fabric of west London and beyond. It is the main domestic,European and intercontinental hub for British Airways, which alone hasdelivered much of the growth in passengers and freight during this time. Itsattractiveness to other airlines operating in the South East is demonstratedby the decision of international operators such as New Zealand Air and VirginAtlantic to transfer operations from Gatwick to Heathrow when theliberalization of airport policy allowed them to do so in the 1980s (Tables11.6 and 11.7).

Table 11.7 Passengers and jobs at UK airports

Source: From various published statistics.

Table 11.8 Total employment impact of the four London airports based onestimates of off-Airport and induced employment for 1993

Source: Based on Andrew and Bailey (1996).


As shown by Table 11.6, there have been substantial increases in labourproductivity in recent years, particularly at Heathrow and Gatwick, but mostnoticeably at Stansted. The decline in productivity at Luton is difficult toexplain, but may relate to its higher on-airport employment to passengerratio (Table 11.7). It is here that the difference between the three major UKairports (Heathrow, Gatwick and Manchester) is apparent on all criteria-passengers, employment and air transport movements. There seems to be aconsistent ratio of about, 1000 jobs (direct) for every 1 million passengersfor these three large airports.

In the four London airports (Heathrow, Gatwick, Stansted, Luton), thereare about 82,000 people employed on the airports and some 72.2 millionpassengers (880 jobs per 1 million passengers). The multipliers calculated byAndrew and Bailey (1996) are 1.186 for indirect employment and 1.24 forinduced employment, giving an overall level of 1.47 for total employment(Table 11.8). These are considerably higher than those multipliers developedfor Heathrow by Pieda (1995), where figures of 1.08 and 1.24 were used,giving an overall multiplier of 1.34 (Section 11.3.3).

11.3.3 London Heathrow terminal 5

The growth and restructuring of domestic and international aviation marketscreates a potential opportunity for development at hub airports such asHeathrow. Within this context, Heathrow’s owners (BAA Plc) believe that itwould be commercially advantageous to expand the capacity of Heathrowto around 80 million passengers in the early years of the next century,consolidating its position as an important international hub. In order to achievethis, the organization would like to construct a new terminal alongside thefour existing terminals at the airport and have submitted a planning applicationto this effect.

The scale of the Terminal 5 (T5) proposal has led to a planning inquiryinto its merits, which is examining the wider social, environmental andeconomic impacts. Because of the issues at stake, this has already become thelongest UK planning inquiry, running for over three years. To aid this decision-making process it is therefore important that the role of the airport as agenerator of economic activity is understood.1

At Heathrow airport attempts have been made to quantify the numberof jobs directly and indirectly related to the operation of the airport, andthe number of induced jobs it supports in the wider economy (Pieda 1994).This has formed the basis of a model, which has then been used to predictthe employment implications of future growth and development at theairport up to 2016, when there will be 80 million passengers if T5 is built(Pieda, 1995). The analysis has been undertaken for a twenty-five year period(1991–2016) with and without T5. It is clearly acknowledged that the


accuracy of forecasts of this type are subject to internal assumptions andexternal changes.

In 1991, Heathrow employed 52,300 people on site and a further 6,500direct off site, giving a total workforce of 58,800 people. In addition to thisdirect employment, there was a 4,500 indirect employment and 15,200induced employment (see Tables 11.8 and 11.9), giving an implied multiplierof 1.08 and 1.24 respectively for the additional employment from these twosources. The overall multiplier is 1.34.

The employment impact has been evaluated within two spatial realms:the local labour market area and the UK as a whole. No attempt has beenmade to estimate the numbers of jobs created through attracted employment,although the importance of Heathrow to the national and internationaleconomy is reviewed shortly (Section 11.3.4).

Direct impacts

The direct impacts have been measured by identifying activities solely relatedto the operation of the airport. This has been supplemented by surveying theorganizations providing these services in order to establish how manyadditional jobs are supported. These organizations included the airportoperating company itself, 90 airlines, six in-flight catering companies, severalhotels, numerous freight forwarding agents and a large number of retailersand other concessionaires.

A key component of the Heathrow analysis has been the increase inproductivity measured by number of passengers per employee or by air trafficmovements per employee. Passenger throughput has increased at a muchfaster rate than air traffic movements per employee, partly because of thegrowth in aircraft size. Overall, productivity has increased by 4.7 per centper annum (1971–91) as measured by passengers per employee. This level ofincrease is not expected to continue as it relates to an exceptional period

Table 11.9 On-airport employment assumptions, 1991–2016

Source: BAA (1994).Note: Traffic support staff are 17 per cent of direct traffic related staff. Percentages giveannual changes in productivity % p.a.=per cent labour productivity change per annum.


when BAA was privatized. However, it is assumed that productivity willincrease at a substantially higher rate with T5 than without (Table 11.9).The argument used here is that productivity grows faster with faster trafficgrowth. This has been true in the past at Heathrow. The net effect is thatdirect on-airport employment will decline by 7,500 (14 per cent) if T5 is notbuilt, and increase by 2,100 (4 per cent) if T5 is built (Table 11.10). Thesechanges must be placed against a background of growth in passengers to 50million (without T5) or 80 million (with T5) in 2016. Over a twenty-fiveyear period, direct on-airport employment will not change by a significantamount, despite a possible doubling in the numbers of passengers throughthe airport. This is in contrast with the high levels of employment generationcreated by the second runway at Manchester airport (see Section 11.4.3).

Direct off-site employment will increase from the current level of 6,470employed in hotels, air freight and in-flight catering, growing to 7,030 (withoutT5) and 9,230 (with T5). The figure for local indirect employment (withinthe study area) is 4,520. This is assumed to grow in proportion to trafficgrowth but is moderated by increases in productivity. Using forecasts fromCambridge Econometrics, Pieda (1994) extended the trends from 1991 to2005 to 2016, giving substantial productivity gains. Indirect employmentdecreases in 2016 if T5 is not built (to 3,105) and increases only marginallyif T5 is built (to 4,590). Finally, the induced employment is calculated throughthe use of multipliers (Table 11.10). A review of multipliers used elsewheresuggested a value of 1.3, but as 20 per cent of Heathrow jobs are outside thestudy area, the effective multiplier is scaled down by that amount to 1.24.The total effect of the new terminal on all employment is limited. Withoutthe new terminal, employment will fall by 10,300 from the present (1991)78,400 employees (a decline of 13.1 per cent), and even with the new terminal,the growth is limited to 6,200 (+7.9 per cent) (Table 11.10).

The Pieda (1994) research acknowledges that forecasting over a twenty-five-year period, particularly in a rapid growth sector, is problematic. Anextensive sensitivity analysis was carried out on the productivity

Table 11.10 Heathrow employment impacts

Source: Based on Pieda (1994).Note: These figures relate to local employment impacts in the study area.


assumptions and wider changes in the labour market in the South East.The overall conclusion reached is that the estimates are sound, but thatassumptions made about changes in the regional economy, employmentand the structure of the labour market all limit the robustness of theanalysis. Previous attempts to predict change have all been unreliable,even over a much shorter period of time.

Indirect impacts

Indirect impacts were evaluated using input-output analysis. Hereexpenditure linkages were explored between the organizations listed aboveand their suppliers. Because this required a detailed breakdown ofpurchases for a large number of goods and services, data were onlycollected from a sample of major employers such as British Airways, withresults adjusted to estimate total impact. In 1991 it was estimated that4,500 local people were indirectly dependent upon the operation of theairport.

Induced impacts

Induced impacts were measured using multipliers. These tools help toquantify the relationship between direct and indirect employment levelsand the number of jobs these support in the broader economy throughspending and taxation, so that estimates can be made where no otherdata exist. In the case of Heathrow, multipliers have been used to estimatelocally induced jobs. Locally, a multiplier of 1.24 was chosen, which meansthat it was assumed that for every 100 employees living in the local area,a further twenty-four jobs were supported, giving a total of 15,190 (seeTables 11.8 and 11.10).

In many airport studies multipliers have been used to estimate direct,indirect and induced employment. Although this may be the best tool to usein the absence of suitable data, they should be treated with some caution. Inthe Heathrow study, for instance, the multipliers have been developed fromsecondary sources, namely general economic assessment tools and otherairport studies. In the former, the UK multiplier has been based upon researchexamining the UK economic system in the 1970s. Since then the propensityto import goods and services from overseas has increased, which means thatthe multiplier used may overstate the benefits of airport expenditures uponinduced employment in the UK. Yet the values of the multipliers are assumedto increase over time (see Table 11.10).

The multipliers used in the other airport studies are themselves based onsimilar reviews, so the methodologies used to determine them can appearsomewhat opaque and some circularity is implicit in the process (Twomey


and Tomkins 1995). Airports are not generic organizations but complexpoints of interface between the air transport system and spatial economies.In this respect, the role that they perform varies significantly dependingupon their position in the global aviation network and the economic contextthat they operate within. We return to this issue of multipliers in the conclusion(Section 11.5).

The Pieda report estimated the construction employment effects of T5 atLondon Heathrow in terms of the direct jobs created over a five-year period(1997–2002) and the indirect and induced jobs created. Over theconstruction period (phase one of T5), it was estimated that 11,135 personyears of work would be generated and which would in turn result in afurther 3,787 jobs. The multiplier used (1.34) came from another review ofstudies carried out by Pieda (1994) and a decision was taken to use themid-point of the range of multipliers from the most detailed study examined(McGuire 1983). These levels were much lower than those used in theManchester study.

11.3.4 The national signif icance of Heathrow

In addition to its role as a major employer in the South East, LondonHeathrow has an important national and international position as theleading UK and European hub.2 The employment impacts of Heathrowextend beyond the local effects (see Section 11.3.3) to its wider impactsin the national economy. Direct employment in 1993 was 58,700. Inaddition to this, some 44,100 people were employed nationally (indirectemployment), resulting from an analysis of the amount and nature ofnon-wage and salary expenditures by firms that use the airport.3 Thisfigure includes local employment (4,520) in the study area (the 18 localauthorities) outlined in Section 11.3.3, and the wider indirect employmentin the country as a whole (39,580). To calculate the induced employment,an empirically derived consumption multiplier was used to produce afigure for the actual numbers of people employed (88,730). The totalnational impact is 191,572 (1992–3), a figure which is over twice theemployment supported in the Heathrow T5 study area (128,310 ascompared with 63,220).

Other evidence of Heathrow’s national and international role is morequalitative in nature. Accessibility to international markets means thatHeathrow is important for multinational firms, particularly with respect toheadquarters functions. The UK has the European headquarters of 37 percent of the 1,000 largest industrial and financial companies. With the openingup of international markets, foreign direct investment has increased, with theUK being particularly attractive to inward investment from Japan and theUSA. Heathrow is the main UK port in terms of value of trade for both


imports and exports. In addition, the airport has a major role as a touristdestination for London and the UK. The majority of tourists (69 per cent)arrive in the UK by air, with Heathrow accounting for over 60 per cent ofthat figure (1990).4

11.4 The economic impact of the second runwayat Manchester airport

11.4.1 Introduction

There has been considerable debate in the UK over whether new airportsshould be built, whether new capacity should be concentrated in the SouthEast, or whether the regional airports should be encouraged (see Table 11.7).Policy has urged the maximum use of regional airports to reduce time-consuming journeys to central London along with the heavy road trafficcongestion associated with the South East. It has also been argued thatregional employment can be enhanced by the development of airports outsideLondon.

Over the last decade passenger traffic at Manchester airport has morethan doubled to around 12.8 million passengers and freight has tripledto 89,000 tonnes (1993). It is the third UK airport after Heathrow andGatwick. The airport has expanded with new terminal buildings in 1949and 1962, further developments in 1973 and a dedicated domesticterminal in 1989. The second international terminal was operational in1993, with a second phase being designed to cope with some 12 millioninternational passengers. The most recent development has been theagreement to build a second runway so that international and new hubopportunities can be combined with a growing range of interliningactivities. In 1990, over 80 airlines were operating services fromManchester to over 160 destinations, with scheduled international trafficaccounting for 25 per cent of all passenger movements and a further 16per cent related to domestic scheduled services, the remaining 59 percent being charter services (Twomey and Tomkins 1995). Table 11.11shows the estimated breakdown of passenger movements in Manchester.The charter market is expected to decline, but Manchester will expandas an international hub airport.

Much of the discussion on the second runway revolved around the economicbenefits of expanding air services at Manchester on employment in the region,particularly as it has suffered high unemployment, resulting from economicrestructuring and the move towards post-industrial employment patterns(Chapter 4). The numbers of staff working at the airport will increase from10,000 (1990) to 30,000 (2005) when the new runway is in operation (YorkConsulting 1994).


11.4.2 The second runway at Manchester airpor t

As noted in Part I, it is often stated that the development of high qualitytransport infrastructure is an essential component for local and regionaleconomies. The importance of a regional airport is that it supposedly performsa ‘double act’ in this process. It improves overall accessibility within the regionand provides a major source of employment. Its indirect effects are also saidto be significant in terms of indirect employment and the additional income(investment multiplier) that it generates within the regional economy. Theremay also be other less tangible effects as the airport makes a statement aboutthe importance of the region and its image.

However, not all the arguments for airport development are positive. Thecosts are largely environmental, with extensive land take requirements for theairport, plus additional support services and related industries. Airports mayalso act as an attractor, bringing new industry and services to the area, particularlyfrom multinational businesses which need airport access. Airports are a majorsource of noise and airborne pollution (emissions) and generate and attract traffic,causing congestion on the surrounding road network. Much of the recent debateover the second runway proposal at Manchester airport revolved around theissues of land take, pollution, traffic and noise. In addition, the new runway willbe built in ‘green belt’ land which has a high amenity and recreational value.

Our main interest here is to establish the nature and scale of the employmentarising from the airport. Research by the Ecole Nationale de l’Aviation Civile(quoted in Twomey and Tomkins 1995) demonstrates that about 1,000 newjobs are created on-site at an airport for every additional 1 million passengers,similar to figures for Heathrow, Gatwick and Manchester (see Table 11.7).Manchester airport conforms to this ‘rule of thumb’ with 10 million passengersand 10,000 workforce (1990). Forecasts to 2005 suggest about 30 millionpassengers and a workforce of 30,000. Manchester Airport Plc, which isresponsible for running the airport operation, employs about 20 per cent ofthe 10,000, with the remaining 80 per cent accounted for by shops, hotels,air freight forwarders, passenger agencies, handling services, maintenance

Table 11.11 Forecasts of passengers through Manchester airport 1995–2005

mppa=million of passengers per annum.Source: Twomey and Tomkins (1995).


and support facilities. A study by the Centre for Local Economic Strategies(CLES 1988) showed that in addition to the 8,400 staff employed on site atthe airport, a further 12,000 jobs were sustained in the North West region ofEngland, mainly in the Manchester conurbation—a multiplier of 2.43.5 Thisstudy also commented that the airport was the single most important factorin the attraction of inward investment to the region, with over 13,000 newjobs being provided in the last five years (1983–8).

More recent studies have been carried out by York Consulting on the economicimpact of the airport (York Consulting 1991, 1994). The numbers given aredifferent (Table 11.12), with a multiplier of 1.33 if direct and indirect employmentare used, or 1.53 if induced employment is added to indirect employment. YorkConsulting (1991) also examined the broader impacts of the new investment(the transportation impact), using figures from the USA where these additionaljob estimates are calculated as two times of site employment or some 5 per centof US GDP, though these figures seem high for the European situation.6

The analysis of Twomey and Tomkins (1995) of Manchester airport’seconomic role within the North West region is based on establishing economiclinkages through surveys covering the sourcing of supply requirements to theairport (Phelps 1993; Turok 1993).7 Their objective is ‘to establish an averagereference pattern of linkage between industry sectors at a broad spatial leveland then to examine the extent of divergence from that average at a smallerspatial scale, the region’ (Twaney and Tomkins 1991:201). From this linkageanalysis it is possible to determine the number of jobs in any one sectorsupported by demand from another identified sector. The two sectors ofinterest are air transport (SIC 7500)8 and air transport support services (SIC7640) in the North West region, and the linkages are with all other sectors inthe local and regional economy.The data used come from the Census of Employment (1991) and Twomey andTomkins (1995) acknowledge the limitations in terms of the classification used.For example, all miscellaneous transport services and storage jobs are groupedtogether (SIC 7700), so it is difficult to identify the airport-related jobs. Second,many airport jobs (e.g. airport catering is placed under catering not airports) areallocated by function rather than location and this may lead to underestimates.

Table 11.12 Estimated regional employment impact of Manchesterairport

Source: York Consulting Ltd (1991).


Third, the nature of the linkage data combines all support services for transporttogether. There is no difference made between other forms of transport (e.g.inland and sea transport), so apportionment has to be made which in turn canlead to inaccuracies. The category of miscellaneous transport services has beenomitted, as it is not possible to identify the air-related activities. Given these datalimitations, it is also important to be aware of key implicit assumptions made:‘that there is no spatial variation in either the productivity with which air transportservices are provided to industry or the unit intensity with which the latterpurchases goods and services from industry’ (Twomey and Tomkins 1995:202).

This analysis is intended to be restrictive as it only calculates the linkagesbetween direct on-site employment, defined as employment of the air servicesectors, and the local and regional economy. There is no attempt to measure thetotal employment attributable to the airport through direct off-site, indirect andinduced jobs. The air transport industry employment (SIC 7500) in the NorthWest Region9 (Table 11.13) covers those jobs which make purchases from the airtransport sector. Business services account for the largest share at about 20 percent (464 jobs). The air transport support services (Table 11.13) are moreimportant than the air transport industry, with the service sector again dominating(75 per cent). Distribution, hotels and catering account for 35 per cent of airtransport support employment, with major contributions from transport andcommunication (23 per cent) and business services (16 per cent).

Looking at the employment linkages the other way round, namely the impactof Manchester airport in terms of North West Region industry jobs that aredirectly supported by the existence of the air infrastructure, a detailed sectoralprofile can be built up. The 5,487 jobs (see Table 11.13) relate to fuel inputs,engineering and allied industries, and food and drink on the manufacturingside (about 30 per cent of the total). Again it is the service sector that dominates,including distribution companies, hotels and catering, transport andcommunications and business services (i.e. insurance and computing). The jobssustained by air services are relatively small (6.2 per cent in Table 11.13). Thetotal direct impact is about 12,000 jobs in the North West Region.

Table 11.13 Direct employment linkages at Manchester airport, 1991

Source: Based onTwomey andTomkins (1995).


Other studies have tried to be more comprehensive in their estimates ofthe regional economic impact by extending the analysis to five categories(York Consulting 1994), to include indirect, induced and spin-off effects (Table11.14). This analysis is based on estimating the total spending power in theNorth West Region (1993) resulting from the operation of Manchester airport.Through the use of multiplier analysis10, further rounds of expenditure wereestimated so that a picture of the total economic impact of the airport couldbe established. The total effect can be defined as income or the correspondingnumber of jobs. Some discussion is presented about the assumptions madeabout leakage where expenditure takes place outside the economy (the NorthWest region), and additionality which relates to new expenditurerather thanexpenditure that would have occurred anyway. With respect to the newrunway at Manchester airport, York Consulting (1994) estimates that thenet impact will be about 93 per cent of the gross impact. A spin-off effect is

Table 11.14 Regional economic impact of Manchester airport

Source: York Consulting Ltd (1991).Note: Data for this study is based on survey work, information from Mancester Airport Plc,

York consultancy Ltd and other companies.


identified (Table 11.14) which relates to the inward investment in the regionresulting from the transport infrastructure investment.

The magic number relating to the injection of spending power is £620mnnet of all taxes (1993) which was generated by the operation of Manchesterairport (Table 11.15), with a further £257mn of direct income generated inthe North West region. Manchester airport would then support about12,200jobs within the airport and a further 6,900 jobs off—site (within a 20-minute drive). This is a total of 19,100 jobs, which would translate to about16,600 full time equivalent direct jobs. The figure of 12,200 is not dissimilarto that arrived at by Twomey and Tomkins (12,183 jobs, in Table 11.13).Most of these jobs were in the North West region (97 per cent), with 67 percent coming from Greater Manchester and a further 26 per cent from Cheshire.The airport company employs about 13 per cent of the 12,200 people, withairlines (30 per cent), handling agents (15 per cent) and concessionaires (25per cent) accounting for a further 70 per cent of the total. In addition to thisdirect employment, York Consulting (1994) estimates a further 3,800 netfull-time equivalent indirect jobs and a further 6,200 net full-time equivalentinduced jobs. The total employment created by Manchester airport would be26,000 full-time equivalent jobs, or about 30,700 jobs in total. The neteconomic impact of the operation of the Airport was around £378mn ofincome (1993). Table 11.15 depicts the estimated impacts for future years.

The York Consulting analysis (1994) then made two estimates for theyear 2005, one without the new runway, the other with the second runway:

• Scenario R1: the level of operation assuming that only the current singlerunway is available—passenger traffic will increase to 22.8 million perannum by 2005.

• Scenario R2: the second runway will be constructed and ready for use in1998. Traffic will increase to 29.3 million passengers per annum in 2005.

Table 11.15 Economic impact of Manchester airport

Source: Based on York Consulting (1994).Notes: The direct income (£mn) is made up of the following components: wages and salaries

earned by residents in the North West region £150mn; Profits earned by companies basedin the North West region £36mn; Expenditure on goods and services supplied by companieslocated within the NorthWest region £241mn; Leakages to residents and companies outsidethe North West Region and expenditure by companies located in other regions £193mn.


Again, there has been a series of assumptions made about the relationshipbetween passenger numbers, income from the airport and employment in theregion. In addition, other factors are assumed to have an impact on airportoperations. Productivity growth will result in a reduction in the number ofon-site jobs per million passengers per annum. Customer care and regulationis likely to increase the number of staff per passenger, as there are increasedrequirements for high quality services and changes in other safety and securityregulations. Economies of agglomeration may also exist as there is aconcentration of services at the larger airports, which may result in moreemployment at the major hub locations. The Manchester airport studyassumed productivity growth (including the impact of any changes in customercare and regulations) of 0.5 per cent per annum (1993–2005). This figure ismuch lower than that assumed at Heathrow for Terminal 5 and is contraryto the historic growth in productivity at Heathrow and Gatwick airports ofabout 3 per cent per annum (see Table 11.6).

‘Scale’ facilities (e.g. airline headquarters, maintenance facilities andconference centres) will support about 1,800 on-site jobs in scenario R2 in2005, with a reduced figure of 1,000 on-site jobs in scenario Rl in 2005. Withdirect off-site employment, the same productivity figures have been used. Forthe indirect and induced employment, the productivity growth is higher at 2per cent per annum (1993–2005) which is in line with changes expected in theregional economy. This difference is important as it suggests that the scope forproductivity savings in the airport is much less than the economy as a whole.No explanation is given of this important assumption. Given these assumptions,Table 11.16 makes a summary of the expected impacts. Note that the neteffect on the regional economy of the second runway is estimated to be £204mnand 12,200 full-time equivalent jobs in the North West region.

In addition to these permanent jobs, there will be short-term constructionemployment as the airport company (Manchester Airport Plc) and otherairport-related firms invest in the new runway and terminal buildings. The

Table 11.16 Summary of economic impact of Manchester airport: Operational effects

Source: Based on York Consulting (1994).


estimates here are £518mn (1993 prices) for 1995–6 to 1999–2000 periodwith scenario R2 to include the new runway (£146mn) and other investment(£372mn) from Manchester Airport Plc and a further £192mn by othercompanies. The corresponding figures for scenario Rl (no runway) are £322mn(1993 prices) for the 1995–6 to 1999–2000 period from Manchester AirportPlc and a further £ 164mn from other companies. The arithmetic used hereassumes a standard relationship between expenditure, income andemployment. Each £1mn of expenditure would generate around £300,000of regional income and around 22 person years of employment (no sourcegiven). As airport construction work is capital intensive, the actual figuresused on the construction of the runway are significantly lower as £1mn ofexpenditure generates £210,000 of regional income and around 15 personyears of employment. The multiplier effects must also be taken into account.For each £1mn of expenditure, there will be £100,000 of regional incomeand seven indirect jobs, with slightly higher effects for the induced regionalincome and jobs (calculated by multipliers; see notes to Table 11.17).

The above analysis covers the operational and construction effects of theproposed airport expansion. In addition to these two major effects, there areother spin-off effects, principally inward investment to the region and tourism.

Table 11.17 Regional economic impact of investment at Manchester airport, 1993–2005:construction effects

Source: York Consulting (1994).Notes:1 Calculated by £486 million * £300,000 and £486 million * 22 person years per £1

expenditure-£486 million=£322 million from Manchester Airport Plc+£164million from other companies.

2 Calculated by £564 million * £300,000+£146 million * £210,000 and £564 million* 22 person years+£146 million * 15 person years per £1 expenditure-£564million=£372 million from Manchester Airport Plc+£192 million from othercompanies.

3 Calculated by £486 million * £100,000 and £486 million * 7 person years per £1expenditure.

4 Calculated by £710 million * £ 100,000 and £710 million * 7 person years per £ 1expenditure-£710 million=£564 million as above+£146 million as theinvestment costs of R2.

5 Calculated by multipliers of 1.36 and 1.40.


It is difficult to be precise on the employment generation here, as the methodof analysis is heavily dependent on qualitative assessments of the competitiveposition of the region, inward investment and tourism in general in the UK,and a sensitive investigation of a few sample firms.

York Consulting (1994) estimates that currently there are between 20,000and 30,000 spin-off job opportunities which can be attributed to Manchesterairport. Under scenario Rl this will increase to between 29,000 and 39,000by 2005, with the corresponding range under scenario R2 being 36,000 to46,000 jobs. These numbers are substantial and reflect the difficulty ofunravelling the complex impacts of major investments, particularly types ofinward investment (e.g. international companies) and particular sectors ofthe tourism market (e.g. high income air-based travellers). Although rangesare given to the figures, it may be equally important to test the sensitivity ofthe assumptions being made as the job opportunities created by these spin-off effects are nearly 50 per cent of the permanent operational jobs beinggenerated. The multiplier is substantial and the numbers of jobs for Manchesterairport are substantially higher than those for Heathrow Terminal 5 (seeSection 11.3.3).

The conclusion reached in this comprehensive study (York Consulting 1994)of the impact of Manchester airport and the second runway is that the airportwill act as the dynamo for the regional economy of the North West over thenext ten years.

11.5 Conclusion

11.5.1 Differences between the Heathrow andManchester analyses

The employment effects for Heathrow Terminal 5 and Manchester’s secondrunway were computed with different methodologies and results (Table11.18). The T5 studies (Pieda 1994, 1995) examined the local study areaimpact and the national impact of Heathrow Terminal 5. The multipliersused (actual and implied) range between 1.24 and 3.26. The Manchesterstudy used a standard relationship between expenditure, regional incomeand the amount of employment created. The implied multiplier for indirectto direct jobs is 1.33, but adding the induced figures raises the Manchestermultiplier to 1.53, with the more recent studies having higher figures (YorkConsulting 1994).

The crucial question seems to be the area over which the impact will befelt. The Heathrow study suggests that the national impacts are much greaterthan the local impacts, while the Manchester studies concentrate on regionalimpacts with only a limited recognition of the international attraction of theairport. The basic inconsistency in the studies is the claim that each seems tostart by reviewing all the other studies. This process should lead to reduced


variability, not enhanced variability in the multipliers as found here. Table11.18 compares results from various studies reported in the literature, relativeto the multipliers used and nature of measured effects.

11.5.2 Major airport employment impacts

Although this chapter has concentrated on two UK airports, it is alsoinformative to place these detailed case studies into the international situation.Pieda (1995) undertook a review of all airports with more than 10 millionpassengers per annum to determine the employment effects (Table 11.19).Comparison is difficult as the study areas vary in size. The larger the studyarea, the greater the potential impact. Two measures have been developed toallow direct comparison: passengers per direct job and the combined multiplier(the relationship between direct employment and total employment). Theproductivity of the US airports is substantially higher than those elsewhere(e.g. Heathrow, Manchester, Schiphol). All the US airports have values inexcess of 1,000 passengers per direct job, except Los Angeles (910) and JohnF Kennedy (650). The European figures are between 560 (Schiphol) and 700(Heathrow). There is even greater variation in the multipliers, which rangefrom 1.33 to 3.76. This again depends on the area of study, yet the nationalimportance of Heathrow puts its figure in excess of 3.00 along with JFK andLa Guardia. Most of the other airports are in the 2.00 to 3.00 range.

Table 11.18 Multipliers used to calculate employment at airports

Source: From various references.Note: Heathrow studies: Pieda (1994) quantifies local impacts as measured in the study area

(18 local authority areas which contain at least 1 per cent of the airport’s workforce)and Pieda (1995) examines the national impact.

Tabl

e 11

.19

Inte

rnat

iona

l air

port

s em

ploy

men

t im

pact

s

Sour

ces:

Base

d on

Pie

da (1

995)

Tab

les

5. 1

and

5.2

and

oth

er s

ourc

es, i

nclu

ding

Bal

timor

e W

ashi

ngto

n In

tern

atio

nal a

irpor

t (19

89),

Cal

iforn

ia D

epar

tmen

t of

Tran

spor

tatio

n (1

989)

, City

and

Cou

nty

of D

enve

r (1

987)

, Col

orad

o N

atio

nal B

anks

(198

9) a

nd P

ort A

utho

rity

of N

ew Y

ork

and

New

Jers

ey, (

1990

).

Not

es:

1C

hica

go O

’Har

e an

d M

idw

ay a

irpo

rts.

2D

alla

s/Fo

rt W

orth

air

port

.3

John

F.K

enne

dy a

irpo

rt.

4To

ront

o LB

Pea

rson

air

port

.5

Dir

ect

incl

udes

bot

h on

-air

port

and

dir

ect

off-a

irpo

rt e

mpl

oym

ent.

6C

ombi

ned

indi

rect

/in

duce

d m

ultip

lier.

7O

nly

dire

ct a

nd t

otal

em

ploy

men

t es

timat

es g

iven

. Ind

irec

t an

d in

duce

d em

ploy

men

t ha

s be

en in

ferr

ed.

8D

eriv

ed b

y ap

plyi

ng t

he in

ferr

ed in

duce

d m

ultip

lier

in t

he s

tudy

to

the

dire

ct a

nd in

dire

ct e

mpl

oym

ent.

Thi

s w

as d

one

to r

emov

e th

e in

duce

dem

ploy

men

t as

soci

ated

with

vis

itor

empl

oym

ent.

9Ex

act

defin

ition

of t

otal

em

ploy

men

t no

t gi

ven,

but

ass

umed

to

be c

onsi

sten

t.10

Onl

y di

rect

em

ploy

men

t es

timat

es g

iven

.11

To

tal e

mpl

oym

ent

not

avai

labl

e on

a c

onsi

sten

t ba

sis.

12H

eath

row

1 is

the

stu

dy a

rea

(18

loca

l aut

hori

ties)

and

Hea

thro

w 2

is t

he n

atio

nal i

mpa

ct.


11.5.3 Comments

Two main conclusions arise from this analysis of a dynamic sector wheredemand is increasing at 6 per cent per annum and where substantialinvestments are taking place. First, there is a tremendous variability in theuse of multipliers, the main instrument to assess the employment impacts ofinvestments. Not only does the value vary, but so does the nature of theemployment effect (direct, indirect and induced), as does the area over whichthe impact supposedly will be felt. There seems to be no consistency (seeTable 11.19), even though most studies start by taking (weighted) averagesof previous studies. As noted at the outset, the use of multipliers results in acrude analysis and this is particularly surprising for such a dynamic sector ofthe economy. More comprehensive analysis would be expected as the growthin air travel is likely to continue, given the substantial changes taking place incontemporary society (outlined in Chapters 3 to 5).

Second, the assumptions used on productivity are crucial both to futureemployment within airports and to direct (and indirect) off-airportemployment. Recent productivity gains (4–5 per cent per annum) areconsidered to be exceptional and have been attributed mainly to the increasesin aircraft size—a one-off change. The forecasted gains in productivity aremuch lower (under 1 per cent per annum), and substantially below theexpected changes for the regional economy as a whole. Small changes inthese figures have substantial impacts on employment.

The calculation of direct on-airport employment and the changes in levelsinfluences all the subsequent related employment as the multipliers are appliedto the on-airport employment figures. Any error in these is furthercompounded. It would seem that a more sophisticated approach is requiredthat acknowledges economies of scale and of agglomeration in airportoperations (see Tables 11.8 and 11.19), and that there are critical sizethresholds which require more flexible assumptions on productivity. Thenature of airport business (passenger or freight; business or charter; nationalor international; interlining or terminating) will also influence the ratios ofpassengers to employees. The composition of the airport workforce is alsobecoming more flexible (in line with other macroeconomic changes), as shiftworking and part-time employment replace traditional forms of employment.

In addition to these two major limitations of current studies, there arebroader questions raised about the real impacts of airports on local, regionaland national economies. The development effects are taken for granted,even though they are difficult to quantify and require the existence ofnecessary conditions in germane labour markets. Although we can say whatwe mean by indirect and induced employment, it is extremely difficult toquantify them, particularly the latter. The question here is whether it is bestnot to quantify them at all until we can develop more sophisticated measures,or to accept that the crude measures we have should continue to be used


with the appropriate caveats. One underlying issue is whether these indirectand induced jobs are wholly attributed to the existence or development ofthe airport or not. Are they new jobs or merely the redistribution ofemployment from one location to another (regionally, nationally orinternationally)? Airport development may help to increase thecompetitiveness of one location over another, but is there a net benefit tothe national economy as a whole?

More importantly though, airport investments are of a substantial scaleand one would expect identifiable employment impacts. Yet, even here theemployment impacts of both Heathrow Terminal 5 and Manchester Airport’sSecond Runway are small in relation to the scale of the labour market.Moreover, the consultants have argued that employment in airport relatedactivities would be reduced (through productivity gains) if the new investmentdoes not take place. Transport infrastructure investment is being justified inpart on the basis of maintaining existing employment, not creating newemployment.

Acknowledgement

Alan McLellan, a researcher at the Bartlett School of Planning in UniversityCollege London, helped with the data collection and drafting of this chapter.We thank him for his contribution.

Notes

1 The study area around Heathrow consisted of 18 local authority areas, whichcontain at least 1 per cent of the airport’s workforce. This accounted for about80 per cent of the direct on-site employment (1992).

2 Nearly 30 per cent of the total passengers (47.6 million in 1993) are transferringthrough the airport. Given its primary position in the UK and Europe, there is aclear preference for airlines to use the airport, but the number of slots availableis limited and valued at a premium, at least by the airlines that “own” them.Even though airport user charges are some 20 per cent higher at Heathrow thanat Gatwick, Heathrow is still more congested, emphasizing the preference for itby international airlines.

3 Pieda (1995) used input-output analysis to calculate the purchases of the fullrange of goods and services required by companies operating at the airport.

4 This position is being eroded by the recent introduction of the Eurostar railservices between London and European cities, principally Paris and Brussels.

5 Note that the multiplier relates to all direct employment at the airport onlydisregarding all direct employment off the airport. So values are not directlycomparable with those for Heathrow. Comparisons can be made with thosefigures in Table 11.12 (York Consulting Ltd 1991).

6 Twomey and Tomkins (1995) concluded that the size of this effect is 2.9 timesonsite employment at Los Angeles international airport.

7 An alternative approach is to use area-specific input-output tables to indicatelevels of economic connectedness (Szyrmer 1985, 1986).

8 SIC is the Standard Industrial Classification used in the UK to group jobs in thecensus of employment.


9 The North West region includes the counties of Cheshire, Greater Manchester,Lancashire and Merseyside.

10 The ratio of total income (direct plus indirect plus induced) to the initial injectionof spending power is termed the multiplier.

APPENDIX 11.1 METHODS USED IN ASSESSING THEECONOMIC IMPACTS OF AIRPORT INVESTMENT

There are three main groups of models that have been used to measure theeconomic development impacts of airports and these are summarized heretogether with illustrations of their use: • Economic base models: relate changes in the goods and services sold

within the region to goods and services sold outside the region.• Econometric models: test relationships between the key economic

variables through regression and other multivariate statistical methods.• Input-output models: establish dependency relationships between

economic sectors and make estimates of the induced impact sector bysector, and after aggregation for the region as a whole. Table A11.1summarizes several airport studies.

Table A11.1 Summary of studies

12

Interpretation of impacts andpolicy conclusions

There is an unquestionable relationship between economic developmentand a liberal democracy…. The exact nature of that relationship is morecomplicated… And not adequately explained.

(Fukuyama 1994:125)

There is a high degree of empirical correlation between stable democracyand economic development.

(Lipset 1959:72)

12.1 Introduction

We started out in this book with the intention of answering a single question.Does transport infrastructure investment in well-developed economiespromote economic growth primarily at the urban level? This book has takenus on a long journey and a considerable way towards finding the answer. Butwhat have we learned? We have tried first to develop a conceptual frameworkto encompass the many issues involved. Second, we have examined how localand regional growth issues are affected by new trends in economics andlifestyle, as well as by changes in behaviour. At the heart of the book wereview existing analytical methods, including the development of amicroeconomic modelling approach, to examine the underlying forces thatcontribute to economic growth and the complex interrelationships at work.A series of empirical studies complement the analytical research where otherpolicy impacts and location factors, which also have an important influenceon the economic state of urban areas, are assessed. In this final chapter, wesummarize our perspective on this important fundamental question and thinkthrough the implications for policymaking and for further research. Ourintention is to be controversial and thought provoking, as this question isone of the key determinants of decisions on transport infrastructure investmentand one of the principal unresolved challenges to transport researchers.

Before we proceed we must re-emphasize a key point made throughoutthe book, namely that we do not question the potential ability of transportation

318 Interpretation of impacts and policy conclusion

infrastructure investments to produce transportation benefits such as traveltime reductions. What we question, however, is whether there are additionalbenefits from these investments, generically referred to as economicdevelopment benefits, and how to measure them. Thus, it is the additionalityand measurability of these assumed benefits that we are concerned about.Failure to properly identify and measure these alleged development benefitsis bound to result in double counting of benefits, thereby running the risk ofimplementing the wrong projects.

12.2 Transport and economic development

From the wealth of information, data and analysis assembled in this book,we come to the basic conclusion that in developed countries where there isalready a well-connected transportation infrastructure network of a highquality, further investment in that infrastructure will not on its own result ineconomic growth. Transport infrastructure investment acts as a complementto other more important underlying conditions, which must also be met iffurther economic development is to take place. Additional transportinvestment is not a necessary condition, but acts in a supporting role whenother factors are at work.

There are three sets of necessary conditions: 1 The first, economic conditions, include the presence of underlying

positive economic externalities, such as agglomeration and labourmarket economies, the availability of a good quality (well-trained andhighly skilled) labour force and underlying dynamics in the localeconomy. This is a fundamental condition, as only when all these factorsare positive and the local economy is buoyant will new transportinvestment, in conjunction with the other necessary conditions, havean economic development impact.

2 Second, there are investment conditions that relate to the availability offunds for the investment, the scale of the investment and its location, thenetwork effects (e.g. are there missing links in the network), and theactual timing of the investment. Transport infrastructure investmentdecisions are not made in isolation, so the nature of the investment,including its ‘place’ in the network, is also one of the necessary conditionsthat needs to be considered. These factors on their own are again notsufficient and this particular focus has been a limitation on much of theanalysis found in the literature, which has tended to examine the spatialfactors in isolation as the main focus of evaluation.

3 The third set constitutes political and institutional conditions that arerelated to the broader policy environment within which transport decisionsmust be taken. To achieve economic development, complementarydecisions and a facilitating environment must be in place; otherwise the


impacts may be counterproductive. Included in this group of factors arethe sources and method of finance, the level of investment (local, regionalor national), the supporting legal, organizational and institutional policiesand processes, and any necessary complementary policy actions (e.g.grants, tax breaks and training programmes). Again, on its own, even afavourable political environment will not result in economic growth unlessthe other necessary conditions are also present.

These three basic sets of necessary conditions are illustrated in Figure 12.1.

As we have argued, individually, the necessary conditions will have little orno impact on development. Even if they are combined on a pair-wise basis,their effect will be limited. For example, in Figure 12.1, if only the investmentand political conditions are present (box 2+3), we can expect accessibilitychanges, but since the necessary economic conditions are not present, economicgrowth impacts will not transpire. In that case, relative attractiveness ofparticular locations may change, but this is merely a redistribution of existingeconomic development rather than additional growth.

Figure 12.1 Illustration of the necessary sets of conditions.


Similarly, if only the investment and economic conditions are present (box1+2), economic development effects from the investment may not follow forthe lack of supportive policies, or because of the presence of conflictingtransportation and landuse policies (see Chapter 10 on the Buffalo Light Railinvestment). It is only when all three necessary sets of conditions are presentand operating together that economic growth will ensue.

Two important further conclusions follow on from this analysis.Transportation infrastructure investments are location related and havepotential growth effects on local economies. Hence, actually to identify andmeasure the economic growth resulting from such investment, analysis musttake place at the local level. It is at this scale that the impacts on local economicdevelopment, income levels, accessibility and employment should be assessed.On the practical level, as analysis becomes more aggregate, many of theimpacts are ‘lost’ (Section 12.4). Second, if the economic growth effects areto be measured, then analysis has to move away from the primary concernwith user benefits (conventionally travel time savings) to a much widerassessment of costs and benefits (Section 12.5).

12.3 Key questions and answers

In the introduction (Chapter 1), we raised ten key questions which now alsoneed to be answered. This is done in Table 12.1. As demonstrated in Table12.1, we do not claim to have obtained definitive answers to all (or any) ofthem, but we hope that some progress has been made. Inevitably, as we makeprogress in one direction, new directions emerge so that we may be left witheven more questions than answers. But that is the nature of research. Inretrospect, the questions we raised in the introduction demonstrate theproblems that we have faced, particularly in terms of the complexity of theprocesses involved and the new agenda.

In the next sections, we further develop some of the more interestingdebating points as a commentary on the key questions in Table 12.1 and weadd some new dimensions to the discussion.

12.4 Dimensions of analysis

In Chapter 2 (Section 2.4) we have discussed our methodological framework,within which we have structured the book. It consists of three dimensions:the scale of the analysis, the type of variables used to assess the investmentand its impacts, and the time duration of the effects. The latter dimension isthe most difficult to assess and in this book we examine it primarily throughthe case studies. Table 12.2 shows the relationships between the types ofmeasured impacts and the level at which they are likely to be felt. Althoughwe have presented extensive discussions about the national and regional levelsof impacts (see Chapters 6 and 7), the main focus of the book is directed atthe local impacts of transport infrastructure investments.

Tabl

e 12

.1 Q

uest

ions

and

ans

wer

s

Tabl

e 12

.1 C

ontin

ued


We now illustrate the arguments regarding the presence of the necessaryconditions for economic development to ensue and the ideas presented in thissection regarding the scale of the analysis. Consider the following twoexamples of national and regional transportation investments, andimplications for the local level (Section 12.4.3).

12.4.1 National level

It is clear that all countries need a well-developed transport infrastructure tocompete internationally in the new global markets. As trade barriers are reducedand new markets are opened up, it is essential to have high levels of accessibility.However, it is not only the quantity of the physical infrastructure that is important,but also the quality of the infrastructure in a much wider context (the politicaland institutional conditions in Figure 12.1). The physical infrastructure needs tobe extended, not just to the links in the network, but also to the terminals andinterchanges. These nodes are often the points of congestion which cause mostdelay. The control systems used on the network are crucial to ensure that maximumefficiency is gained from its use and decisions are based on high quality real timeinformation. The management, information and control systems are alsoimportant, in many cases even more important than the physical infrastructureitself. In addition, the financial and organizational context within which theinfrastructure is placed can again influence decisions on construction, maintenance,ownership, charging and responsibility. These elements are also the key to theeffective use of any infrastructure expansion. Part of this responsibility relates tothe externalities created by infrastructure use (e.g. noise, pollution, accidents andconsumption of resources), and who is to pay for these (substantial) costs, whetherit is society as a whole, the user or both.

Table 12.2 Conceptual framework of analysis


The most important ‘new’ dimension in this process is the use that is madeof the network by the user. The choices available have increased substantially,particularly if the physical infrastructure is placed in the wider context of thechanging technology, employment, substitution of time between work andleisure activities and firms’ production, and location decisions (the economicset of conditions in Figure 12.1). A high quality ubiquitous infrastructureallows an increased flexibility in the operations of firms and greater variabilityin individual travel patterns. Company structures have become much flatterwith production being outsourced, efficient supply chains, lower stock levelsand shorter production runs. Manufacturing processes are increasingly beingcustomer driven, with the detailed specifications being provided by thecustomer. Efficient transport systems are required to facilitate these newproduction processes, but there also has to be the supporting managerial,organizational and technological infrastructures to allow this to workefficiently (the political and institutional set of conditions in Figure 12.1).

12.4.2 Regional level

Here the concept of accessibility is central to the debate. It has been arguedthat changes in accessibility resulting from transport infrastructure investmentcause a redistribution of employment between regions. It is unclear whetherthe changes in accessibility also create new activities, which we would calleconomic growth (see Figure 12.1, box 2+3). The conclusions reached heresuggest that at the regional level redistribution will take place, often to thefurther advantage of the already accessible core parts of the country. Buttransport accessibility must be seen as part of a much wider concept ofaccessibility that includes availability of skilled labour, good quality locations,the necessary supporting infrastructure and local road and rail networks(Figure 12.1, economic conditions). It seems that the actors need to be ableto manage the changing internal and external relationships to achieve thebest economic output. This is what some (e.g. Storper 1993) have called thecreative learning capacity. Changes in transport accessibility resulting frominfrastructure investment form one element in that process. Figure 12.2illustrates the regional arguments.

In Figure 12.2 movement along the accessibility axis will not achieveeconomic growth on its own. Only where there are open dynamic systems(when the economic and political sets of conditions are present, see Figure12.1) will it have a real impact, particularly where the existing infrastructureprovides only poor levels of accessibility (quadrant 1). In dynamic systemswith high levels of accessibility it will support growth (quadrant 2), but inthe lower part of the figure (quadrants 3 and 4) improving transportaccessibility is not a sufficient condition for economic growth (for the lack ofthe economic, investment and political sets of conditions).

The capacity of the system has to be enhanced so that it can respond, i.e.


move to the top part of the figure. This means that the skills and knowledgebase have to be raised (i.e. introducing economic conditions). Strong linkswith local research and the informal contact networks have to be establishedand dynamic transactional relationships between manufacturers, localsuppliers and customers are required. These are the key elements of the creativelearning capacity that is a necessary prerequisite for economic growth at theregional level (see Figure 12.1).

12.4.3 Local level

In addition, there is the need to examine the costs and benefits of new roadinvestment in terms of the opportunities for economic growth anddevelopment. These effects should be examined at the strategic (county)and regional level so that the spatial consequences can be matched to thelocal traffic impacts. These important considerations are rarely included inthe decision process. Even at planning inquiries where all materialconsiderations should be raised, the local authority has to follow the strategyset out in the development plan. Privately, there may be great concern overthe impact of a new road on the local economy and environment, but publiclythe proposal is supported if it conforms to the general strategy set out inthe development plan. If all local concerns were taken into consideration,the costs of the road would be increased and the benefits reduced, thuslowering its net value, possibly to a point where it ceased to be viable(Headicar 1996). The conclusion here is that decisions have to reflect the

Figure 12.2 Transport and economic development at the regional level.


concerns at all levels and that they must also be placed within the politicaland institutional context.

12.5 A new proposal for project appraisal

A major criticism we had of the national level analysis is that finding a rate ofGDP growth from a certain increase of total capital accumulation actually tellsus little about the contribution of the next infrastructure project (see Chapter6, Section 6.2.2). That is, some projects have a high and positive impact onGDP growth whereas others may not. The national level analysis only tells usabout the average contribution of capital accumulation. But since infrastructureinvestments are carried out project by project we need to analyse each separatelywhere the general rate of return (or rate of GDP growth) from total publiccapital formation may at best serve as a kind of a guideline.

Another issue to contemplate is that analysis of all planned infrastructureinvestment projects is done ex ante with little certain information about futuretrends and development. Hence the need to consider specific issues that mayaffect the potential impact of this investment with economic growth. In thisbook we have highlighted some of these factors.

First is the need to consider each transport project within the frameworkof a local, regional or even national network. Secondly, for growth to occurit must be the result of improvements at the network level and not at thesingle project level. The effect of a given investment on growth can bedramatically different when it links, for example, two disjoint networks ratherthan being an additional link in an established network.

Prioritizing objectives and criteria is a third factor. 1 Majority of benefits need to be transport related, since otherwise why

invest in transportation facilities in the first place.2 Need to avoid double counting in measuring non-transport benefits.3 Need to show functional linkage between primary transportation benefits

(e.g. accessibility improvements) and potential economic growth effects. If transportation investment is to take place, we would suggest a twinapproach where conventional cost benefit analysis is carried out on theproject to determine the primary (transportation related) user benefits andcosts of investment. To achieve a given rate of return, this analysis mayaccount for some or all the necessary returns. If there is a shortfall, then acomplementary analysis needs to take place that has a wider view of theinvestment proposal. As itemized in Table 12.3, this would include thecontribution of the project to the transport network as a whole throughnetwork analysis. In addition the value added of the project would beassessed through its contribution to local employment, the potential forincreases in productivity and the environmental impacts. Finally, it would


also investigate the distributional impacts in terms of the spatial effects onthe regional and local distribution of services and facilities, and the socialimpacts.

This complementary analysis would use some of the methods and modelsproposed in this book. The contribution that the components outlined inTable 12.3 would make to the overall project assessment could be measured.If additional benefits can be shown and are sufficient to raise the cost-benefitanalysis above the crucial level for the rates of return, then investment couldtake place. This more complex type of analysis seems to be increasinglyimportant, as the conventional benefits from any proposed transportinvestment may be providing an ever-decreasing proportion of the total returns.

12.6 Decoupling transport from economicgrowth

Historically, there has been a close relationship between the growth in demandfor freight and passenger traffic and economic growth, as measured by GDP.The question raised in Chapters 1 and 2 related to the underlying rationalefor this statistical relationship and whether it should (or would) continueinto the future. Our conclusion is that there is no reason why we should havetransport growth in line with economic growth. Indeed, there are strongefficiency and environmental arguments for breaking that link. We should beseeking to reduce the transport intensity of activities while at the same timemaintaining economic growth: this is the decoupling argument.

Travel can be broken down into three component parts—volume, distanceand efficiency. The first two components are usually combined to give measuresof performance (i.e. passenger km or ton km), but the third element is equallyimportant as it relates to modes, travel time and price, the use of resources,technology and organizational factors. For example, in the freight sector

Table 12.3 A suggested twin approach to project appraisal


efficiency can be increased through the use of logistics, flat organizationalstructures, new forms of handling, minimization of warehouse requirementsand spatial organization to reduce distribution costs. All these measures canincrease efficiency and reduce the volumes and distances travelled. However,they can also work to increase efficiency and the volumes/distances travelled;the arguments work in both directions.

Dematerialization can make a more fundamental impact. As productionbecomes more service based, lighter products are moved about andminiaturization takes place with a much greater emphasis on quality anddesign. The net effect is that less volume is moved around so less travel isrequired. But as the material intensity of products decreases, materialconsumption may still increase as the economy is growing and demand isalso expanding. It has been estimated that dematerialization could result in a15 to 20 per cent reduction in freight volumes between 1995 and 2020(Schleicher-Tappeser et al. 1998). This reduction must be placed against theexpected increase in freight traffic of 80 per cent over this same period. Furtherreductions could be achieved through raising the durability of products sothat they last longer, but here there is a trade-off between lasting quality andthe need to take advantage of the latest technological innovations.

The second major reorganization change could be the move from the globalto the ‘glocal’1. Traditional arguments strongly favour concentration ofproduction to take advantage of agglomeration economies. However, thedevelopment of flexible specialization (Piore and Sabel 1984, 1988; Chapter4 ) has allowed a new complementarity between global networks and regionalproduction for regional markets without the multinationals losing control.The multinationals still have the knowledge and control of information, butproduce for local markets in a dispersed manner through outsourcing.Production units are downsized so that the new lean production methods canbe introduced. For example, in the state of Baden-Wurttemberg in southGermany most suppliers for the car manufacturer Mercedes are based in andaround Stuttgart. Certain components are ordered to the hour and must beavailable within a radius of 100 km (Schleicher-Tappeser et al. 1998). Localproduction networks provide the opportunity for short travel distances anda reduction in freight traffic. This same study estimated that the potentialreduction in freight traffic through the use of regional markets, regionalproduction and the reduction in international flow of goods (that have beenproduced for local markets) could amount to between 20–30 per cent (1995–2020).

There is a substantial potential for decoupling in the freight sector withthe new forms of production outlined above. The scale economy arguments,together with economies related to specialization and the comparative costadvantages of producing large quantities for large markets, are now beingquestioned. Customer-driven requirements mean that products are nowtailored to individual specifications so that smaller scale production units for


regional markets become possible, provided that the knowledge and skillsare available.

However, there is still a long way to go as global cultures andinternationalization have been instrumental in producing similar values, tastesand lifestyles, with the consequent loss of community and locality. Similarly,most of the changes in policy have tended to encourage greaterinternationalization through trade liberalization, market-based strategies,subsidies to farmers and other groups, deregulation in transport, privatization,and even markets for environmental and consumer protection. The mostoptimistic view is that in the freight sector transport demand could remainconstant over the next twenty years (to 2020), with the strong implementationof decoupling strategies.

In the passenger sector, the opportunities for stabilization in demand seemharder to envisage, particularly within the context of increased affluence andleisure time. Decoupling must again be seen as a combination of strategies toreduce the volume of traffic, the distance travelled and measures to increaseefficiency, but at the same time maintaining economic growth. The decouplingarguments follow the same structure as in the freight sector withdematerialization of travel through less travel or travel by more efficientmodes, and through establishing local travel patterns through thereorganization of the production and consumption patterns based on localand regional networks.

In the context of this book, the important conclusion is that, even if therehas in the past been a link between transport use and economic growth, thereis no reason why this link should continue. The strong sustainability argumentwould suggest that if the link does exist it must be broken as it is unsustainable.We should be actively seeking to break the link (if it exists) between transportdevelopment and economic growth.

12.7 Complexity and causality

There may be a Catch-22 situation in the analysis of the links betweentransport infrastructure investment and economic development. To makeanalysis tractable requires a series of simplifying assumptions to be made,but this in turn reduces the usefulness of the analysis. Even if more complexformulations are possible, then the limitations of the available data are quicklyreached. This situation results in uncertainty and prevents clear conclusions.Even if it is possible to demonstrate that economic development is present intheory, it may be impossible to measure it in practice.

This complexity is increasing, particularly when placed against new fundingpriorities and mechanisms, the changing industrial base, the balancing ofeconomic priorities with social and environmental concerns, etc. One of themain themes that has been argued in this book is that even thoughsimplification and strong assumptions may have been valid in the past (see,


for example, Figure 7.1), that position has become less tenable. We mustaccommodate complexity in analysis and recognize the multi-dimensionalnature of the links between transport, location, development and the manyother new factors relevant to our understanding of these processes (see, forexample, Figure 7.2).

In our microeconomic analysis (Chapter 8), some of these factors havebeen included in the analysis with several different strategies available. Buteven here we have concentrated on work-related activities, single workers(not multiple workers), standard notions of values of time, and only twofirms. Perhaps the complexity issue needs to be approached through acombination of methods that link together more formal modelling approacheswith descriptive qualitative analysis that can make links between methods.Such an approach would begin to tackle the complexity in a robust and flexiblemanner so that a clear picture can be achieved.

Complexity relates to one of the underlying problems investigated in thisbook, namely causality. One of the key elements in the basic economicargument is that there must be a set of causal relations between transportinvestment and economic development since otherwise there would no basisfor the claims that transport infrastructure investments also promote economicgrowth. Public infrastructure investment is regarded as the trigger mechanismto the increasing of private capital rates of return through increases in privatecapital stock and labour productivity, which in turn result in higher totaloutput and economic growth.

We have argued that this causality argument is one of many that can beused and that the empirical evidence is weak. It seems doubtful that publicinfrastructure investment will lead to substantial increases in newemployment, as most of the potential savings will be realized throughincreases in productivity with the existing labour force. Changes inaccessibility may induce relocation, but only if those changes are above keythresholds and other factors are working strongly in the same direction,will these relocations further induce economic growth. In the microeconomicmodel developed in Chapter 8, we have explored the impacts ofinfrastructure investment on the choices that individuals make in how theychoose to use their time. Whether additional time (resulting from reducedtravel times following an infrastructure investment) is taken as leisure oradditional work is a major question. The proliferation of leisure timeactivities in well-developed economies may mean that additional time isnot used to increase labour supply. If that is the case, the causality linkagebetween accessibility improvements and economic growth may be ratherweak and getting weaker over time.

The conclusion reached is that the causality argument is weakening asthere seem to be decreasing returns on public physical capital. Although wehave not explored this issue in detail, the arguments for increasing returns on


knowledge and technology may be seen as the new engines of economic growthrather than the transport infrastructure (Romer 1986).

12.8 Accessibility and proximity

The evidence cited in this book is mixed on the role that accessibility changeshave in generating economic development. It should be remembered thataccessibility is a relative concept and infrastructure investment in one locationmay help that location, but at the expense of a competing location. The neteffect may be marginal. More important though is the argument that in mostadvanced economies levels of accessibility by road (and rail) are already high.This means that most infrastructure investments will only affect theaccessibility in the system as a whole in a marginal way. It is only where thereis a major change in the system-wide accessibility (e.g. a new link joining twopreviously disjoint networks, or the opening up of a previously inaccessiblelocation) that major relocation will take place. This suggests that in the futureaccessibility should not be investigated only as a relative concept, but that weshould be searching for minimum thresholds above which change takes place.These thresholds are important if accessibility changes are seen as the meansto improve the relative positions of regions, by increasing inward investmentand employment.

A second more subtle implication of changes in accessibility is that theyenhance existing trends rather than creating new ones. If the conditions areadvantageous for firms to relocate or set up a new business in some areas (forexample, where the labour force has appropriate skills or where there arefinancial incentives), then improvements in the transport infrastructure maygive one location preference over another. On its own, transport infrastructureis a second order location variable where there is a well-developed network,but in conjunction with other factors it may ‘tip the balance’ in favour of the(marginally) more accessible location.

This debate relates to the complementarity found within networks.Accessibility has tended to be viewed as the impact of one new link on thenetwork as a whole. But many investments are complementary, so competitionis really taking place between systems, not individual links. Accessibility shouldnot only be viewed as the changes taking place in one system (e.g. rail), butalso the new competitive position of that system in relation to other systems(e.g. road). There is a strong optionality value in accessibility to a particularsystem, even if no use is actually made of it.2 New concepts of networks andaccessibility are required to determine under which conditions the competitiveposition of one network will be changed as compared with another. This is aquestion of value added to influence expectations, to facilitate co-ordinationand to ensure compatibility.

Proximity is closely related to accessibility, but may also be important in thecontext of development. There are two conflicting processes at work, one of


which leads towards proximity through agglomeration factors and the otherwhich suggests that in a well-connected society non-proximity or diseconomiesof agglomeration may operate. Many services can (and are) provided remotely.For example, in a telephone enquiry service, the person actually answeringyour question need not be in close proximity to you or to others. Similarly,many products can be outsourced to locations where labour and otherproduction factors are cheaper. Provided that the charge to the user is low (i.e.a local rate telephone call or cheap products), proximity is not important.High quality communications and transport infrastructure facilities have alloweda greater flexibility in the location of many services and firms.

Economies of scale play an important role in this process and locations thatpossess specialization tendencies may benefit more from a reduction in transportcosts than other locations. Krugman (1991a) shows an interesting example ofthe impact of a reduction in transportation costs from infrastructureimprovements on manufacturing location in the presence of scale economies inproduction. With high transport costs, production will take place in dispersedlocation (e.g. core and periphery). When transport costs fall there will be ashift towards production in one location (either the core or the periphery),which can switch if transport costs keep falling (see also Rietveld and Bruinsma1998). All these factors may provide important insights into patterns ofdevelopment at the national and international levels and may even help in ourunderstanding of urban spatial structure (Anas et al. 1998).

In the situation where transport costs are decreasing (or already low) andwhere there is a ubiquitous infrastructure, non-material flows may becomemore important (Burmeister and Colletis-Wahl 1997). The traditionalframework, which treats transport as a cost, is becoming less relevant whentrying to understand the spatial dynamics of industry. Recent research proposesa richer more complex set of relationships that are based on circulatorycapacity. This concept includes the infrastructure and the organization andmanagement of flows of goods, information and people (i.e. the services thatcan be produced). The links between infrastructure and the utilization of theproduction process are not deterministic. The role of transport infrastructurein a network economy is no more than a generic resource for circulation(Burmeister and Colletis-Wahl 1997:239). Particular forms of use and theconsequent impacts depend on the strategies adopted by the actors in theproduction process. Proximity extends far beyond geographical accessibilityto include the organization and degree of control of the flows of goods,information and people.

12.9 Transport investment and economicdevelopment: the role of policy design

If there is one key lesson to be learnt from the analysis in this book, it is thatof the crucial role that policy design can play in influencing and strengthening


the potential impact of transportation infrastructure investment on localeconomic development. In Figure 12.1 we have highlighted the notion thatfor economic development to follow from transportation infrastructureinvestments, it is essential that three sets of necessary conditions be met. Thisrepresentation, however, conveys the impression that all of these sets areequally important in affecting the economic development outcomes ofinfrastructure investments. Our aim here is to argue that this, in fact, is anincorrect portrayal of reality since political and policy decision making actuallyaffects, directly or indirectly, the other two sets of necessary conditions.Diagrammatically we show this in Figure 12.3.

In Figure 12.3, the circle labelled investment type refers to the particularnature of the investment: that is, the mode (e.g. in highway, rail, freight orairport); the investment’s scale, its location and whether it is a new link inan existing network, an expansion of existing links in an existing networkor a new link that unites two disjoints networks. The circle labelled economicconditions refers to such conditions as agglomeration externalities in firms’location, labour market externalities, the presence of network economies,and the presence of inefficiencies in spatial structures. Finally, the circlelabelled policymaking, refers to key non-economic factors that influenceeconomic growth. These include such elements as the organizationalstructure and range of responsibility of decision making and overseeingagencies, the nature of the legal system, to the government level at whichdecisions are made and, most importantly, to the political involvement ofpolitical organs. The shaded area in Figure 12.3 is where all three sets of

Figure 12.3 The role of policymaking in achieving economic growth.


conditions are met and where economic development will emerge (compareit with the box 1+2+3, in Figure 12.1).

What this diagram essentially shows is that the relative importance ofeach set of conditions is not equal. Policymaking, which affects both theeconomic conditions and, more importantly, the investment type, is the crucialfactor in realizing economic growth benefits from a transportationinfrastructure investment: hence, the special attention we have paid throughoutthis book to policymaking facets of transport infrastructure investmentprojects. In democracies, the ability of governments to co-ordinate theiractivities and design complementary policies to gain maximum economicdevelopment effects from their capital investments many times seems ratherlimited. Perhaps paradoxically, however, and to which the quotations fromFukuyama and Lipset at the beginning of this chapter attest, it is only indemocratic societies that economic development reaches its maximumpotential.

Notes

1 Glocal refers to a combination of global and local, where production is stillcontrolled by the large multinational companies, but produced locally for localmarkets under franchising and other arrangements.

2 We should note here the concept of positive network externalities developedmainly for communication and computer networks. It implies that the value ofmembership to one user is positively affected when another user joins and enlargesthe network (Katz and Schapiro 1994), hence the rationale for having moreusers joining and using the network. In transportation this argument may bevalid in developing economies where new transport investment opens upundeveloped areas. In the context of this book, which focuses on transportationnetworks in highly developed economies, this argument may be invalid as thereare strong congestion externalities present when another user joins the network.

References

Aaron, H.J. (1990) ‘Where is infrastructure importance?’, in A.Munnell (ed.) Is Therea Shortfall in Public Capital Investment?, Boston: Federal Reserve Bank of Boston,Conference Series 34, pp. 51–63.

Acutt, M. and Dodgson, J. (1998) ‘Transport and global warming: modelling theimpacts of alternative policies’, in D.Banister (ed.) Transport Policy and theEnvironment, London: Spon, pp. 20–37.

Airports Council International (ACI) (1993) The Economic Impact Study Kit, Brussels:ACI Europe, p. 23.

Alcaly, R.E. (1976) ‘Transportation and urban land values: a review of the theoreticalliterature’, Land Economics 52 (1):42–53.

Allen, K. and MacLennan, D. (1970) Regional Problems and Policy in Italy andFrance, Beverly Hills, CA: Sage.

Alonso, W. (1964) Location and Land Use: Towards a General Theory of LandRents, Cambridge, MA: Harvard University Press.

Amano, K. and Nakagawa, D. (1990) ‘Study of urbanization impacts of new stationof high-speed railway’, Paper given at conference of the Korean TransportationAssociation, Dejeon City.

Ampe, F. (1995) ‘Technopole development in Euralille’, in D.Banister (ed.) Transportand Urban Development, London: Spon, pp. 128–35.

Anas, A. (1984) ‘Principles and parables of transportation/land use interaction’,in Land Use Impacts of Highway Projects, Proceedings of the WisconsinSymposium on Land-Use Impacts of Highway Projects, 9–10 April, Milwaukee,WI.

Anas, A. (1992) ‘On the birth and growth of cities: laissez-faire and planningcompared’, Regional Science and Urban Economics 22 (2):243–58.

Anas, A. (1995) ‘Capitalization of urban travel improvements into residential andcommercial real estate: simulation with a unified model of housing, travel modeand shopping choices’, Journal of Regional Science 35 (3):351–76.

Anas, A. and Kim, I. (1994) ‘General equilibrium models of polycentric urban landuse with endogenous congestion and job agglomeration’, Paper given at the 41thNorth American Meetings of the Regional Science Association, 17–29 November,Niagara Falls, Ontario.

Anas, A. and Kim, I. (1996) ‘General equilibrium models of polycentric urban landuse and endogenous congestion and job agglomeration’, Journal of UrbanEconomics 40 (2):232–56.

336 References

Anas, A., Arnott, R. and Small, K.A. (1998) ‘Urban spatial structure’, Journal ofEconomic Literature 36:1426–64.

Anderstig, C. and Matsson, L.-G. (1991) ‘Appraising large scale investments in ametropolitan transportation system’, Paper given at the Regional ScienceAssociation European Congress, 27–30 August, Lisbon, Portugal.

Andrew, R. and Bailey, R. (1994) Flight Path to Prosperity? The Impact of Airportsin the South East on their Local Economies, Harlow: SEEDS.

Andrew, R. and Bailey, R. (1996) ‘The contribution of airports to regional economicdevelopment’, in PTRC, Proceedings of Seminar B: Airport Planning Issues, Papergiven at the 24th European Transport Forum.

Appelbaum, E. and Alpin, P. (1990) ‘Differential characteristics of employmentgrowth in service industries’, in E.Appelbaum and R.Schettkat (eds) LabourMarket Adjustments to Structural Change and Technological Progress, NewYork: Praeger.

Appelbaum, E. and Berechman, J. (1991) ‘Demand conditions, regulation andthe measurement of factor productivity’, Journal of Econometrics 47 (2/3):379–400.

Arnott, R. (1998) ‘Congestion tolling and urban spatial structures’, Journal of RegionalScience 38 (3):495–504.

Arnott, R., de Palma, A. and Lindsey, R. (1992) ‘Route choice with heterogeneousdrivers and group specific congestion costs’, Regional Science and Urban Economics22:71–102.

Arnott, R., de Palma, A. and Lindsey, R. (1994) The welfare effects of congestiontolls with heterogeneous commuters’, Journal of Transport Economics and Policy28 (2):139–61.

Arrow, K.J. and Lind, R.C. (1970) ‘Uncertainty and the evaluation of public investmentdecisions’, American Economic Review 60 (3):364–78.

Arrow, K.J. and Lind, R.C. (1994) ‘Uncertainty and the evaluation of public investmentdecisions’, in R.Layard, and S.Glaister, (eds) Cost-Benefit Analysis, 2nd edn,Cambridge: Cambridge University Press.

Arthur, W.B. (1991) ‘Positive feedbacks in the economy’, Scientific American,February.

Aschauer, A.D. (1989a) ‘Is public expenditure productive?’, Journal of MonetaryEconomics 23 (2):177–200.

Aschauer, A.D. (1989b) ‘Does public capital crowd out private capital?’, Journal ofMonetary Economics 24 (2):171–88.

Aschauer, A.D. (1989c) ‘Public investment and productivity growth in the group ofseven’, Economic Perspectives 13 (5):17–25.

Aschauer, A.D. (1990) ‘Highway capacity and economic growth’, EconomicPerspectives 14 (1):14–24.

Aschauer, A.D. (1991) ‘Transportation spending and economic growth, Paper preparedfor the American Public Transit Association, September.

Aschauer, A.D. (1993a) ‘Public capital and economic growth, in Public InfrastructureInvestment: A Bridge to Productivity Growth? A Public Policy Brief, The JeromeLevy Economics Institute of Bard College, No. 4.

Aschauer, A.D. (1993b) ‘Genuine economic returns to infrastructure investment’,Policy Studies Journal 21 (4):380–90.

Atkins, S.T. (1983) ‘The value of travel time: an empirical study using route choice’,PTRC Summer Annual Meeting, University of Sussex, UK.

References 337

Austroads (1996) Benefit Cost Analysis Manual, Sydney: Austroads, p10.Baffes, J. and Shah, A. (1993) ‘Productivity of public spending, sectoral allocation

choices, and economic growth’, Policy Research Working Paper 1178, World Bank,Policy Research Department, Washington DC.

Bajo-Rubio, O. and Sosvilla-Rivero, S. (1993) ‘Does public capital affect privatesector performance? An analysis of the Spanish case’, Economic Modelling 10(3): 179–86.

Balassa, B. (1981) The Newly Industrialized Countries, New York: Pergamon Press.Balduini, G. (1974) ‘Effects de localisation industrielle des autoroutes Italiennes de

l’IRI’, Paris: ECMT.Baltimore Washington International Airport (1989) Summary of the Economic Impact

of Baltimore Washington International Airport on the State of Maryland,Baltimore, MD: Baltimore Washington International Airport.

Banister, D. (1992) ‘The British experience of bus deregulation in urbantransport: Lessons for Europe’, Paper given at the Spanish Regional ScienceAssociation’s Seminar on Urban Transport Problems, Madrid, June. WorkingPaper 5, Planning and Development Research Centre, University CollegeLondon, p. 29.

Banister, D. (1993a) ‘Policy responses in the UK’, in D.Banister and K.Button (eds)Transport, the Environment and Sustainable Development, London: Chapmanand Hall, pp. 53–78.

Banister, D. (1993b) ‘Charging systems for the use of urban infrastructure:possibilities and realities’, paper prepared for the ECMT 97th Round Tableon Charging Systems for the Use of the Urban Infrastructure, Paris, 4–5November, p. 34.

Banister, D. (1994) Transport Planning, London: Spon.Banister, D. (1996) ‘Energy, quality of life and the environment’, Transport Reviews

16 (1):23–35.Banister, D. (1997) ‘Reducing the need to travel’, Environment and Planning B24

(3): 437–49.Banister, D. (ed.) (1998) Transport Policy and the Environment, London: Spon.Banister, D. and Bayliss, D. (1992) ‘Structural changes in population and impact on

passenger transport’, European Conference of Ministers of Transport, Round Table88, Paris, pp. 103–43.

Banister, D. and Button, K. (eds) (1993) Transport, the Environment and SustainableDevelopment, London: Chapman and Hall.

Banister, D. and Edwards, M. (1995) ‘Measuring the development and socialimpacts from transport infrastructure investment: the case of the Jubilee lineextension in London’, mimeo available from authors at University CollegeLondon.

Banister, D., Andersen, B. and Barrett, S. (1995) ‘Private sector investment in transportinfrastructure in Europe’, in D.Banister, R.Capello and P.Nijkamp (eds) EuropeanTransport and Communications Networks: Policy Evolution and Change, London:Belhaven, pp. 191–220.

Banister, D.Watson, S. and Wood, C. (1997) ‘Sustainable cities: transport, energyand urban form’, Environment and Planning B 24 (1):125–43.

Banister, D.Gérardin, B. and Viegas, J. (1998) ‘Partnerships and responsibilities intransport: European and urban policy priorities’, in K.Button P.Nijkamp and H.Priemus (eds) Transport Networks in Europe; Concepts, Analysis and Policy,London: Edward Elgar, pp. 202–32.

Barrett, S. (1993) ‘Air transport markets’, in D.Banister and J.Berechman (eds)

338 References

Transport in a Unified Europe: Policies and Challenges, Amsterdam: NorthHolland, pp. 91–124.

Bates, J.J.Roberts, M.Gwilliam, K. and Goodwin, P. (1987) The Value of TravelTime Savings, Newbury: Policy Journals.

Batey, P.W.Madden, M. and Scholefield, G. (1993) ‘Socio-economic impact assessmentof large scale projects using input-output analysis: a case study of an airport’,Regional Studies 27 (3):179–92.

Batty, M. (1976) Urban Modelling, Cambridge: Cambridge University Press.Baumol, W. (1964) Welfare Economics and the Theory of the State, London: Bell.Becattini, G. (1990) ‘The Marshallian industrial district as a socio-economic notion’,

in F.Pyke, G.Becattini and W.Sengenberger (eds) Industrial Districts and Inter-Firm Co-operation in Italy, Geneva: International Institute for Labour Studies,pp. 37–51.

Becker, G. (1965) ‘A theory of the allocation of time’, Economic Journal 75 (3):493–717.

Beesley, M. (1965) ‘The value of time spent in traveling: some new evidence’,Economica 32 (2):174–85.

Beeson, P.E. (1992) ‘Agglomeration economies and productivity growth’, in E.Millsand J.McDonald (eds) Sources of Metropolitan Growth, New Brunswick Centerfor Urban Policy Research, Rutgers, the State University of New Jersey.

Beggs, J. (1984) ‘The value of travel to recreational facilities: the uncertainty issue’,International Journal of Transport Economics 11 (1):53–9.

Belous, R. (1989) ‘The contingent economy: the growth of the temporary, part-timeand subcontracted workforce’, Washington DC: National Planning Association,Report No 239.

Ben-Akiva, M. and Lerman, S. (1985) Discrete Choice Analysis: Theory andApplication to Travel Demand, Cambridge: MIT Press.

Benell, D.W. and Prentice, B.E. (1993) ‘A regression model for predicting theeconomic impacts of Canadian airports’, Logistics and Transportation Review29 (2): 139–58.

Berechman, J. (1989) ‘A model of activity location with agglomeration economiesand congestion effects’, Paper given at the 36th Annual Meeting of the RegionalScience Association.

Berechman, J. (1994) ‘Urban and regional economic impacts of transportationinvestment: a critical assessment and proposed methodology’, TransportationResearch A 28 (4):351–62.

Berechman, J. (1995) ‘Transport infrastructure investment and economicdevelopment’, in D.Banister (ed.) Transport and Urban Development, London:Spon, pp. 17–35.

Berechman, J. and Paaswell, R. (1983) ‘Rail rapid transit investment and CBDrevitalization: methodology and results’, Urban Studies 20 (4):471–86.

Berechman, J. and Paaswell, R. (1994) ‘The Bronx-Center Project: travel behavior ofthe Bronx’s population’, Working Paper WP93–BC–3, University TransportationResearch Center, City University of New York.

Berechman, J. and Paaswell, R. (1996) ‘Does accessibility improvements affect localemployment? The case of the South Bronx’, TRED Conference on Land Use andTransportation, Lincoln Institute, Cambridge, MA, 11–12 October.

Berechman, J. and Paaswell, R. (1997) ‘The implications of travel profiles fortransportation investment: the Bronx Center Project’, Transportation 24(1):51–77.

References 339

Berechman, J. and Small, K. (1988) ‘Modeling land use and transportation: aninterpretive review for growth areas’, Environment and Planning A 20 (10):1285–1310.

Bergman, A. and Marom, A. (1993) ‘Growth factor in the business sector in Israel(1958–1988)’, Discussion Paper 93.02, Research Department, Bank of Israel, June.

Berndt, E. (1991) The Practice of Econometrics, Classic and Contemporary, NewYork: Addison-Wesley.

Berndt, E.R. and Hansson, B. (1992) ‘Measuring the contribution of publicinfrastructure capital in Sweden’, Scandinavian Journal of Economics 94(supplement): 151–68.

Berry, B. (1967) Geography of Market Centers and Retail Distribution, EnglewoodCliffs NJ: Prentice Hall.

Berry, B. (1991) Long-Wave Rhythms in Economic Development and PoliticalBehaviour, Baltimore, MD: Johns Hopkins University Press.

Bertolini, L. (1998) ‘Station area redevelopment in five European countries: aninternational perspective on a complex planning challenge’, International PlanningStudies 3 (2):163–84.

Blauwens, G. and Van de Voorde, E. (1988) ‘The valuation of time savings incommodity transport’, International Journal of Transport Economics 15(1):77–87.

Blum, U. (1982) ‘Effects of transportation investment on regional growth: a theoreticaland empirical investigation’, Papers and Proceedings of the Regional ScienceAssociation 49 (2):169–84.

Boardman, A.E., Greenberg, D.H., Vining, A.R. and Weimer, D. (1996) Cost-BenefitAnalysis: Concepts and Practice, New Jersey: Prentice Hall.

Boardman, A.E., Mallery, W.L. and Vining, A.R. (1994) ‘Learning from ex ante/expost cost-benefit comparisons: the Coquihalla Highway example’, Socio-EconomicPlanning Sciences 28 (2):69–84.

Boarnet, M.G. (1994a) ‘Highways and inter-metropolitan employment location’,Working Paper No. 1994–18, Department of Urban and Regional Planning,University of California, Irvine.

Boarnet, M.G. (1994b) ‘Transportation infrastructure, economic productivity andgeographic scale: aggregate growth versus spatial redistribution’, Paper given atthe 41st North American Meetings of the Regional Science Association, 17–29November, Niagara Falls, Ontario.

Boarnet, M.G. (1994c) ‘The mono-centric model and employment location’, Journalof Urban Economics 36 (1):79–97.

Boarnet, M.G. and Sarmiento, S. (1996) ‘Can land use policy really affect travelbehavior? A study of the link between non-work travel and land use characteristics’,Paper prepared for the 1996 Lincoln Land Institute TRED Conference onTransportation and Land Use, October, Cambridge, MA.

Boddy, M. and Thrift, N. (1990) ‘Socio-economic restructuring and changes in patternsof long distance commuting in the M4 corridor’, Final report on the UK Economicand Research Council, June.

Bollinger, C.R. and Ihlanfeldt, K.R. (1997) ‘The impact of rapid transit on economicdevelopment: the case of Altanta’s MARTA’, Journal of Urban Economics 42 (2):179–204.

Bonnafous, A. (1979) ‘Underdeveloped regions and structural aspects of transportinfrastructure’, in W.A.G. Blonk (ed.) Transport and Regional Development,Farnborough: Saxon House, pp. 45–62.

340 References

Borukhov, E. and Hochman, O. (1977) ‘Optimum and market equilibrium in a modelof a city without a predetermined center’ , Environment and Planning A 9 (8):849–56.

Bos, H.C. and Koyck, L.M. (1961) ‘The appraisal of road construction projects: apractical example’, Review of Economics and Statistics 43 (1):13–20.

Botham, R. (1980) ‘Regional development effects of road investment’, TransportPlanning and Technology 6 (1):97–108.

Botham, R. (1983)’The road programme and regional development: the problem ofthe counterfactual’, in K.Button and D.Gillingwater (eds) Transport Location andSpatial Policy, Aldershot: Avebury, pp. 23–56.

Boyce, D.Allen, W., Mudge, R.Slater, P. and Isserman, A. (1972) ‘Impact of rapidtransit on suburban residential property values and land development: analysis ofthe Philadelphia-Lindewold high speed line’, unpublished paper; Philadelphia:University of Philadelphia.

Bradford, D. and Hildebrandt, G. (1977) ‘Observable public good preferences’, Journalof Public Economics 8 (2):110–31.

Brennan, M. and Schwartz, E. (1985) ‘Evaluating natural resources investments’,Journal of Business 58 (2):135–57.

Bresnahan, T.E. and Trajtenberg, M. (1995) ‘General purpose technologies “enginesof growth”?’, Journal of Econometrics 65 (1):83–108.

Brög, W. (1992) ‘Structural changes in population and impact on passengertransport’, European Conference of Ministers of Transport, Round Table 88,Paris, pp. 6–42

Brotchie, J. (1991) ‘Fast rail networks and socio economic impacts’, in J.Brotchie, M.Batty, P.Hall and J.Newton (eds) Cities of the 21st Century: New Technologiesand Spatial Systems, New York: Longman Cheshire, pp. 25–37.

Bruinsma, F. and Rietveld, P. (1993) ‘Accessibility of cities in European infrastructurenetworks: a comparison of approaches’, Research-Memorandum 1993–18, FreeUniversity of Amsterdam.

Bruinsma, F. and Rietveld, P. (1997) ‘A stated preference approach to measure therelative importance of location factors: a case study of the eastern part of theNetherlands’, International Journal of Development Planning Literature 12 (1/2): 125–40.

Bruinsma, F., Pepping, G. and Rietveld, P. (1996) ‘Infrastructure and urbandevelopment: the case of the Amsterdam orbital motorway, in D.F.Batten andC.Karlsson (eds) Infrastructure and the Complexity of Economic Development,Berlin: Springer Verlag, pp. 231–49.

Bruzelius, N. (1979) The Value of Travel Time, London: Croom Helm.Buchanan, J.M. (1963) ‘The economies of earmarked taxes’, Journal of Political

Economy 71 (5):457–69.Burgess, E.W. (1925) ‘The growth of the city: an introduction to a research project’,

in R.E.Park, E.W.Burgess and R.D.McKenzie (eds) The City, Chicago: ChicagoUniversity Press, pp. 47–62.

Burmeister, A. and Colletis-Wahl, K. (1997) ‘Proximity in production networks:the circulatory dimension’, European Urban and Regional Studies 4 (3):231–241.

Button, K. (1994) ‘Overview of internalising the social costs of transport’, in OECDInternalising the Social Costs of Transport, Paris: OECD, pp. 7–30.

Button, K. (1996) ‘Air transport in the 1990s’, Built Environment 22 (3):161–6.

References 341

Button, K. and Rietveld, P. (1993) ‘Financing urban transport projects in Europe’,Transportation 20 (3), 251–265.

California Department of Transportation (1989) User Manual for the State ofCalifornia Airport Economic Impact Model, Sacramento: California Dept ofTransportation.

Calthorpe, P. (1993) The Next American Metropolis, New York: PrincetonArchitectural Press.

Canning, D. and Fay, M. (1993) ‘The effect of transportation networks on economicgrowth’, Columbia University Working Paper, New York.

Capello, R. (1994) Spatial Economic Analysis of Telecommunications NetworkExternalitites, Aldershot: Avebury.

Capello, R. and Gillespie, A. (1993) ‘Transport, communications and spatialorganisation: future trends and conceptual frameworks’, in P.Nijkamp, (ed.) Europeon the Move, Aldershot: Avebury, pp. 43–66.

Capital Shopping Centres (1997) Background Information: Lakeside Shopping Centre,Thurrock: Capital Shopping Centres.

Carlile, J.L. (1994) ‘Private funding of public highway projects’, Proceedings of theInstitution of Civil Engineers 105:53–63.

Castells, M. and Hall, P. (1994) Technopoles of the World: The Making of 21stCentury Industrial Complexes, London: Routledge.

CBI (1995) ‘Missing links: settling nation transport priorities’, a CBI DiscussionDocument, Confederation of British Industry, London, January.

CEBR (1994) Roads and Jobs: The Economic Impact of Different Levels ofExpenditure on the Roads Programme, London: CEBR.

Centre for Local Economic Strategies (CLES) (1988) The Impact on the LocalEconomy of Past and Likely Future Development at Manchester Airport,Manchester: CLES.

Cervero, R. (1989) ‘Job-housing balancing and regional mobility’, Journal of theAmerican Planning Association 55 (2):136–50.

Cervero, R. (1994) ‘Transit based housing in California: evidence on ridership impacts’,Transport Policy 1 (3):174–83.

Cervero, R. and Landis, J. (1995a) ‘Development impacts of urban transport: a USperspective’, in D.Banister (ed.) Transport and Urban Development, London: Spon,pp. 136–56.

Cervero, R. and Landis, J. (1995b) ‘The transportation-land use connection stillmatters?’ Access 7:2–10.

Cervero, R. and Landis, J. (1997) ‘Twenty years of the Bay Area Rapid TransitSystem: land use and development impacts’, Transportation Research A 31 (4):309–33.

Chandler, A.D. (1977) The Invisible Hand: The Managerial Revolution in AmericanBusiness, Cambridge MA: Harvard University.

Cheshire, P. and Giussani, B. (1994) ‘The determinants of urban economic growth:testing theory and policy prescriptions’, Paper given at the 41 st North AmericanMeetings of the Regional Science Association, 17–29 November, Niagara Falls,Ontario.

Chinitz, B. (1961) ‘Contrasts in agglomeration: New York and Pittsburgh’, AmericanEconomic Review, Papers and Proceedings 51 (2):279–89.

Chisholm, M. and O’Sullivan, P. (1973) Freight Flows and Spatial Aspects of theBritish Economy, Cambridge: Cambridge University Press.

342 References

Christaller, W. (1933) Die Zentralen One in Suddeutschland (Central Places inSouthern Germany), trans. C.W.Baskin, Englewood Cliffs: Prentice Hall.

Chui, M.K. and McFarland, R. (1987) ‘The value of travel time: new estimates usingspeed choice model’, Transportation Research Record 1116”:15–21.

City and County of Denver (1987) The Regional Economic Impact of StapletonInternational Airport and Future Airport Development, Denver: City andCountyof Denver.

Cogan, J. (1980) ‘Labor supply with costs of labor market entry’, in J.P.Smith (ed.)Female Labor Supply: Theory and Estimation, Princeton, NJ: Princeton UniversityPress, pp. 327–59.

Cohen, S. and Zysman, J. (1987) Manufacturing Matters: The Myth of the Post-Industrial Economy, New York: Basic Books.

Cole, S. (1990) ‘Attitudinal survey: trade-off analysis’, a report prepared for VIARail, Canada.

Colorado National Banks (1989) Ready for Take-Off: The Business Impact of ThreeRecent Airport Developments in the US, Denver: Colorado National Banks.

Commission of the European Communities (CEC) (1992) The Impact of Transporton the Environment, Green Paper, Brussels: CEC.

Commission of the European Communities (CEC) (1993) Trans European Networks—Towards a Masterplan for the Road Network and Road Traffic, Brussels: CECDirectorate General for Transport, Report of the Motorway Working Group.

Commission of the European Communities (CEC) (1998) Fair Payment forInfrastructure Use: A Phased Approach to a Common Transport InfrastructureCharging Framework for the EU, White Paper, com/98/466, Final, 22 July, Brussels:CEC.

Confederation of British Industry (CBI) (1995) ‘Missing links: settling nation transportpriorities’, a CBI Discussion Document, Confederation of British Industry, London,January.

Costa, J.da S.Ellson, R.W. and Martin, R.C. (1987) ‘Public capital, regional outputand development: some empirical evidence’, Journal of Regional Science 27 (3):419–37.

Crane, R. (1996a) ‘The influence of uncertain job location in urban form and thejourney to work’, Journal of Urban Economics 39 (3):342–56.

Crane, R. (1996b) ‘Cars and drivers in the new suburbs, linking access to travel inneo-traditional planning’, Journal of the American Institute of Planners 62 (1):51–65.

Cunning, D. and Fay, F. (1993) ‘The effect of transportation networks on economicgrowth’, Discussion Paper, Columbia University, Department of Economics.

Damesick, P. (1986) ‘The M25—a new geography of development—the issues’,Geographical Journal 152 (2):155–60.

Danielson, M.N. and Wolpert, J. (1991) ‘Distributing the benefits of economic growth’,Urban Studies 28 (3):393–412.

Danielson, M.N. and Wolpert, J. (1993) ‘Transportation and the distribution ofmetropolitan economic growth’, Working Paper 93–5, Woodrow Wilson Schoolof Public and International Affairs, Princeton University.

Davidson, K.B. (1977) ‘Accessibility in transport/land-use modeling and assessment’,Environment and Planning A 9 (12):1401–6.

Dawson, R.F.F. and Everall, P.P. (1972) ‘The Value of Motorists’ Time’, Report LR426, Crowthorne: Transport and Road Research Laboratory,.

Deacon, R. and Sonstelie, J. (1985) ‘Rationing by waiting time: results from a uniqueexperiment’, Journal of Political Economy 93 (4):627–47.

References 343

Deakin, E.A. (1991) ‘Jobs, housing and transportation: theory and evidence oninteractions between land use and transportation’, Transportation, Urban Formand the Environment, Transportation Research Board SR 231, pp. 25–42.

DeCorla-Souza, P., Everett, J., Gardner, B. and Culp, M. (1997) ‘Total cost analysis:an alternative to benefit-cost analysis in evaluating transportation investment’,Transportation 24 (2):107–23.

Deno, K.T. (1988) ‘The effect of public capital on US manufacturing activity: 1970–1978’, Southern Economic Journal 2.

Deno, K.T. (1991) ‘Public capital and the factor intensity of the manufacturing sector’,Urban Studies 28 (1):3–14.

Deno, K.T. and Eberts, R. (1991) ‘Public infrastructure and regional economicdevelopment: a simultaneous equation approach’, Journal of Urban Economics30 (3):329–43.

De Rus, G., Roman, C. and Trujillo, L. (1995) ‘Economic significance of airports:the case of Gran Canaria’, in PTRC, Proceedings of Seminar B: Airport PlanningIssues, presented at the 23rd European Transport Forum.

De Rus, G., Roman, C. and Trujillo, L. (1997) ‘Convergence and transportinfrastructure in the European Union’, International Journal of DevelopmentPlanning Literature 12 (1/2):141–162.

DeSerpa, A.C. (1971) “A theory of the economics of time’, Economic Journal 81 (4):828–46.

DeSerpa, A.C. (1973) ‘Microeconomic theory and the valuation of travel time: someclarification’, Regional and Urban Economics 2 (4):401–10.

De Vries, J. (1981) Barges and Capitalism: Passenger Transportation in the DutchEconomy, 1632–1839, Utrecht: HES Publishers.

Diamond, D. and Spence, N. (1989) Infrastructural and Industrial Costs in BritishIndustry, the Department of Enterprise, London: HMSO.

Dickens, I. (1992) ‘Transport investment, economic development and strategicplanning: the example of light rail transit’, Planning Research 7 (2):9–12.

Dickson, P. (1978) ‘The Buffalo Erie County input-output study’, Buffalo: Departmentof Community Development.

Diewert, W.E. (1986) ‘The measurement of the economic benefits of infrastructureservices’, Lecture Notes in Economics 278, Berlin: Springer.

Dixit, A. (1989) ‘Entry and exit decisions under uncertainty’, Journal of PoliticalEconomy 97 (3):620–38.

Dodgson, J. (1973) ‘External effects and secondary benefits in road investmentappraisal’, Journal of Transport Economics and Policy 7 (2):169–85.

Dodgson, J. (1974) ‘Motorway investment, industrial transport costs, and sub-regionalgrowth: a case study of the M62’, Regional Studies 8 (1):75–91.

Dodgson, J. (1984) ‘Benefits to generated freight transport as a measure of the‘development benefits’ of highway investment’, Paper prepared for by the PTRCAnnual Conference, Warwick, July.

Dosi, G., Freeman, C., Nelson R., Silverberg, G. and Soete, L. (1988) Technical Changeand Economic Theory, London: Pinter.

D’Ouville, E.L. and McDonald, J.F. (1990) ‘Optimal road capacity with a suboptimalcongestion toll’, Journal of Urban Economics 28 (1):34–49.

Downs, A. (1962) ‘The law of peak-hour expressway congestion’, Traffic Quarterly16 (3):393–409.

344 References

Downs, A. (1992) Stuck in Traffic, Coping with Peak Hour Congestion, WashingtonDC: Brookings Institution.

Durkin, Jr, J.T. and Wassmer, R.W. (1994) ‘Public infrastructure spending and privateincome generation in large US Cities’, Working Paper, Lincoln Institute of LandPolicy, Cambridge, MA.

Dyett, M.V. and Escudero, E. (1978) ‘Effects of BART of urban development’,Transportation Engineering Journal 104 (3):239–51.

Easterly, W. and Rebelo, S. (1993) ‘Fiscal policy and economic growth: an empiricalinvestigation’, Journal of Monetary Economics 32 (2):417–58.

Eberts, R.W. (1986) ‘Estimating the contribution of urban public infrastructure toregional economic growth’, Working Paper 8610, Federal Reserve Bank ofCleveland, December.

Eberts, R.W. (1990) ‘Public infrastructure and regional economic development’,Economic Review, Federal Reserve Bank of Cleveland, 26 (1):15–27.

Edwards, S.C. and Bayliss, B.T. (1970) Industrial Demand for Transport, London:HMSO.

Egbert, J. and de Haan, J. (1995) ‘Is public expenditure really productive? Newevidence for the USA and the Netherlands’, Economic Modeling 12 (1):60–72.

Eisner, R. (1991) ‘Infrastructure and regional economic performance’, New EnglandEconomic Review, September–October:47–58.

Engel, E., Fischer, R. and Galetovic, A. (1996) ‘A new mechanism to auction highwayfranchises’, Economic Series, No. 13, Center for Applied Economics, Departmentof Industrial Engineering, University of Chile.

EUROCASE (1996) Mobility, Transport and Traffic in the Perspective of Growth,Competitiveness and Employment, Paris: European Council of Applied Sciencesand Engineering.

European Union DGII (1997) ‘The likely macro economic and employment impactsof investments in the trans-European transport networks’, CEC Brussels reportfor DGII the Directorate of Economic and Financial Affairs, March.

Evans, P. and Karras, G. (1992) ‘Are government activities productive? Evidencefrom a panel of US states’, Working Paper, University of Illinois.

Ewing, R. (1994) ‘Beyond density, mode choice, and single purpose trips’, Joint Centerfor Environmental and Urban Problems, Florida Atlantic University/FloridaInternational University.

Ewing, R. (1997) ‘Is Los Angeles style sprawl desirable?’, Journal of the AmericanPlanning Association 63 (1):107–26.

Ferguson, M. (1988) ‘From the people who brought you Voodoo economics’, HarvardBusiness Review May–June:55–62.

Fielding, G.J. and Klein, D.B. (1993) ‘How to franchise highways?’, Journal ofTransport Economics and Policy 27 (2):113–30.

Fogel, R.W. (1964) Railroads and American Economic Growth: Essays in EconometricHistory, Baltimore: Johns Hopkins.

Foot, D. (1978) ‘Urban models 1: a computer program for the Garin-Lowry model’,Geographical Papers 65, Department of Geography, University of Reading,England.

Ford, R. and Poret, P. (1991) ‘Infrastructure and private sector productivity’, OECDEconomic Studies 16 (1):79–131.

Forkenbrock, D.J and Foster, N.J.S. (1990) ‘Economic benefits of a corridor highwayinvestment’, Transportation Research A 24 (4):303–12.

References 345

Forsyth, P.J. (1980) ‘The value of time in an economy with taxation’, Journal ofTransport Economics and Policy 14 (3):pp. 337–62.

Foster, C. (1992) Privatisation, Public Ownership and Regulation of NaturalMonopoly, Oxford: Blackwell.

Frank, L. and Pivo, G. (1994) Relationships between Land Use and Travel Behaviourin the Puget Sound Region, Seattle: Washington State Transportation Center(TRAC).

Freidman, L.S. (1984) Microeconomic Policy Analysis, New York: McGraw-Hill.Frerick, J.Helms, E. and Kreutec, H. (1972) ‘Die Erfassung und quantifizie rung des

Wachsums—und Strukturoffekte von Autobahnen’.Friedlander, A. (1975) The Interstate Highway System, Amsterdam: North Holland.Friedlander, A. (1990) ‘How does public infrastructure affect regional economic

performance?’, in A.Munnell (ed.) Is There a Shortfall in Public CapitalInvestment, Conference Series 34, Boston: Federal Reserve Bank of Boston, pp.108–12.

Friedman, M. (1949) ‘The Marshallian demand curve’, Journal of Political Economy57 (6):463–74.

Friedmann, J. (1986) ‘The world city hypothesis’, Development Change 4 (1):12–50.Frost, M.E. and Spence, N.A. (1991) ‘British employment in the eighties: the spatial,

structural and compositional change in the workforce’, Progress in Planning 35(2): 75–168.

Fujita, M. (1988) ‘A monopolistic competition model of spatial agglomeration:differentiated product approach’, Regional Science and Urban Economics 18 (2):87–124.

Fujita, M. and Ogawa, H. (1982) ‘Multiple equilibria and structural transition ofnon-monocentric urban configuration, Regional Science and Urban Economics12 (2):161–96.

Fukuyama, F. (1994) The End of History and the Last Man, Harmondsworth: Penguin.Garcia-Milà, T. and McGuire, T. (1992) ‘The contribution of publicly provided inputs

to state’s economies’, Regional Science and Urban Economics 22 (2):229–42.Garcia-Milà, T. and McGuire, T. (1998) ‘A note on the shift to a service based economy

and the consequences for regional growth’, Journal of Regional Science 38 (2):353–63.

Garin, R.A. (1966) ‘A matrix formulation of the Lowry Model for intra-metropolitanactivity location’ , Journal of the American Institute of Planners 32 (6):361–4.

Gérardin, B. (1990) ‘Private and public investment in transport: possibilities andcosts’, Proceedings of the ECMT, Round Table 81, Paris, pp. 5–32.

Giersch, H. (1995) Urban Agglomeration and Economic Growth, Berlin: SpringerVerlag.

Gillen, D. (1993) ‘Investing in infrastructure: will it really yield a more competitivenation?’, ITS Review May: 2–3, University of California, Berkeley.

Giuliano, G. (1996a) ‘The weakening transportation-land use connection’, Access 6:3–11, University of California Transportation Center.

Giuliano, G. (1996b) ‘Information technology, work patterns and intra-metropolitanlocation: a case study’, Paper presented at the TRED Conference onTransportation and Land Use, Lincoln Institute of Land Policy, Cambridge MA,October.

Glaeser, E.L., Kallal, H.D., Scheinkman, J.A. and Shleifer, A. (1992) ‘Growth in cities’,Journal of Political Economy 100 (6):1126–52.

346 References

Glazer, A. and Niskanen, E. (1992) ‘Parking and congestion’, Regional Science andUrban Economics 22 (2):122–32.

Goldberg, M. (1984) ‘Assessing land-use impacts of transportation improvments’, inLand Use Impacts of Highway Projects, Proceedings of the Wisconsin Symposiumon Land-Use impacts of Highway Projects, 9–10 April, Milwaukee, Wisconsin.

Golden, L. and Appelbaum, E. (1992) ‘What was driving the 1982–88 boom intemporary employment?’, American Journal of Economics and Sociology 51(4):473–93.

Gomez-Ibanez, J.A. and Meyer, J.R. (1995) ‘Private toll roads in the United States:recent experiences and prospects’, in D.Banister, (ed.) Transport and UrbanDevelopment, London: Chapman and Hall, pp. 248–71.

Gordon, P. and Richardson, H. (1994) ‘Congestion trends in metropolitan areas’, inCurbing Gridlock: Peak Period Fees to Relieve Traffic Congestion, volume 2,Transportation Research Board, Special Report 242, pp. 1–31.

Gordon, P. and Richardson, H. (1997) ‘Are compact cities a desirable planning goal?’,Journal of the American Planning Association 63 (1):95–106.

Gordon, P., Richardson, H. and Jun, M.J. (1991) ‘The community paradox: evidencefrom the top twenty’, Journal of the American Planning Association 57 (4):416–20.

Gould,G. (1987) ‘The Impact of the M25 on innovative forms of retailing—myth orreality?’, unpublished MPhil thesis, Bartlett School of Planning, University College,London.

Gramlich, E.M. (1994), ‘Infrastructure investment: a review essay’, Journal ofEconomic Literature 32 (3):1176–96.

Greater London Council (GLC) (1970) London Road Plans 1900–1970, GreaterLondon Research, Research Report II, London: HMSO.

Grieco, M. (1994) The Impact of Transport Investment Projects upon the Inner City,Aldershot: Avebury.

Group Transport 2000 Plus (1990) ‘Transport in a fast changing Europe’, Papercommissioned by the Transport Commissioner of the European Commission (Karelvan Miert), December.

Grübler, A., Messner, S., Schrattenholzer, L. and Schäfler, A. (1993) ‘Emission reductionat the global level’, Energy 18 (5):539–81.

Grunau, R. (1994) ‘Optimal road capacity with a suboptimal congestion toll’, Journalof Urban Economics 36 (1):1–7.

Guest, A.M. (1976) ‘Workplace and residential location: a push-pull model’, Journalof Regional Science 16 (1):27–32.

Gunn, H. (1991) ‘Research into the value of time savings and losses: the Netherlands1985 to 1991’, Hague Consulting Group, Final Report.

Guttman, J. (1975) ‘Avoiding specification errors in estimating the value of time’,Transportation 1 (1):7–24.

Gwilliam, K.M. and Judge, E.J. (1978) ‘The M62 and transport movement 1970–1977’, Paper given at the Regional Studies Conference on Transport and theRegions, London, March, and Institute for Transport Studies, Leeds University,Working Paper 41.

Haaland, K. and Odeck, J. (1997) ‘Economic evaluation methods for road projectsin member countries’, Draft, Prepared for C9 PIARC Committee.

Hall, P. (1995) ‘A European perspective on the spatial links between land use,development and transport’, in D.Banister (ed.) Transport and Urban Development,London: Spon, pp. 65–88.

References 347

Hall, P. (1996) ‘The geography of the information economy’, Interdisciplinary ScienceReviews 21 (3):199–208.

Handy, C. (1995) The Age of Unreason, London: Arrow.Handy, S. (1993) ‘Regional versus local accessibility: implications for non work travel’,

Transportation Research Record 1400:58–66.Handy, S. and Mohktarian, P. (1995) ‘Planning for telecommuting: measurement

and policy issues’, Journal of the American Planning Association 61 (1):99–112.Hansen, N.M. (1965) ‘Unbalanced growth and regional development’, Western

Economic Journal 4 (1):3–14.Hansen, N.M. (1971) Intermediate-size Cities as Growth Centers, New York: Praeger.Harris, B. and Wilson, A.G. (1978) ‘Equilibrium values and dynamics of attractiveness

terms in production-constrained spatial interaction models’, Environment andPlanning A 10 (4):371–88.

Harris, C.D. and Ullman, E.J. (1945) ‘The nature of cities’, Annals of the AmericanAcademy of Political and Social Sciences, 242:7–17.

Hart, T. (1983) ‘Transport and the economic development: the historical dimension’,in K.Button, and D.Gillingwater (eds) Transport Location and Spatial Policy,Aldershot: Avebury, pp. 12–22.

Hart, T. (1993) ‘Transport investment and disadvantaged regions: UK and Europeanpolicies since the 1950s’ , Urban Studies 30 (2):417–36.

Hauer E. and Greenough J.C. (1982), ‘A direct method for value of time estimation’,Transportation Research 16 (3):163–72.

Haughwout, F.A. (1996) Infrastructure, wages, and land prices’, unpublished paper,Woodrow Wilson School, Princeton University.

Haynes, K.F. and Krmenec, A. (1989) ‘Sensitivity analysis of uncertainty ininfrastructure expansion’, Annals of Regional Science 31 (2):301–15.

Headicar, P. (1996) ‘The local development effects of major new roads: M40 casestudy’, Transportation 23 (10):55–69.

Helling, A. (1997) ‘Transportation and economic development’, Public WorksManagement and Policy 2 (1):79–93.

Henderson, J.V. (1977) Economic Theory and the Cities, New York: Academic Press.Henderson, J.V. (1986) ‘Efficiency of resource usage and city size’, Journal of Urban

Economics 19 (1):47–70.Henderson, J.V. (1988) Urban Development: Theory Fact and Illusion, New York:

Oxford University Press.Henderson, J.V. (1992) ‘Peak shifting and cost-benefit miscalculations’, Regional

Science and Urban Economics 22 (2):112–21.Henderson, J.V. and Castells, M. (eds) (1987) Global Restructuring and Territorial

Development, London: Sage.Hensher, D. (1989) ‘Behavioral and resource values of travel time savings: a

bicentennial update’, Australian Road Research 19 (3):223–9.Hensher, D. and Truoung, T.P. (1984) ‘Valuation of travel time savings from a

direct experimental approach’, Journal of Transport Economics and Policy 19(3): 237–61.

Hepworth, M. and Ducatel, K. (1992) Transport in the Information Age: Wheelsand Wires, London: Belhaven.

Hicks, J.R. (1943) ‘The four consumer surpluses’, Review of Economics Studies 11(1):31–41.

348 References

Higgins, B. (1971) ‘The Montreal Airport site: the spatial multiplier and other factorsaffecting its selection’, Growth and Change 2 (1):14–22.

Highways Agency (1996) ‘M25 opening—tenth anniversary 29 October 1996’,Highways Agency Public Relations Fact Sheet, October.

Hirschman, A. (1958) The Strategy of Economic Development, New York: YaleUniversity Press.

Hirschman, I., McKnight, C, Pucher, J., Paaswell, R. and Berechman, J. (1995) ‘Bridgeand tunnel toll elasticities in New York: some recent evidence’, Transportation 22(2):97–113.

Hirst, P. and Zeitlin, J. (eds) (1989) Reversing Industrial Decline?, Oxford: OxfordUniversity Press.

Hoffman, K. and Kaplinsky, R. (1988) Driving Force—The Global Restructuring ofTechnology, Boulder: Westview Press.

Holland, C. (1993) ‘Paying for roads’, Transport May–June: 11.Holtz-Eakin, D. (1988) ‘Private output, government capital and the infrastructure

crisis’, Discussion Paper 394, Columbia University, New York, May.Holtz-Eakin, D. (1993) ‘State specific estimates of state and local government capital’,

Regional Science and Urban Economics 23:185–209.Holtz-Eakin, D. (1994) ‘Public sector capital and the productivity puzzle’, Review of

Economics and Statistics 76 (1):12–21.Holtz-Eakin, D. and Lovely, M.E. (1996) ‘Scale economies returns to variety, and the

productivity of public infrastructure’, Regional Science and Urban Economics 26(2):105–23.

Holtz-Eakin, D. and Schwartz, A.E. (1994) ‘Spatial productivity spillovers from publicinfrastructure: evidence from state highways’, International Public Finance 2 (3):59–68.

Holtz-Eakin, D. and Schwartz, A.E. (1995) ‘Infrastructure in a structural model ofeconomic growth’, Regional Science and Urban Economics 25 (2):131–51.

Hopkins, J.B., O’Donnell, J. and Ritter, G.T. (1994) Telecommuting: How much?How soon? Moving Toward Deployment, Proceedings of the IVHS America AnnualMeeting, Atlanta, Georgia, pp. 518–27.

Hopkins, J.R. (1986) ‘Transport infrastructure investment and economic development:an interim review’, Department of Transport, Transport Policy Division.

Horowitz, A. (1978) ‘The subjective value of time spent in travel’, TransportationResearch 12:388–93.

House of Commons Committee of Public Accounts (1991) Sale of the National BusCompany, 9th Report of the House of Commons Committee of Public Accounts,Session 1990–91, No. 119, London: HMSO.

Hoyt, H. (1939) The Structure and Growth of Residential Neighborhoods in AmericanCities, Washington DC: Government Printing Office.

Hsiao, C. (1986) Analysis of Panel Data, Cambridge: Cambridge University Press.Huallachain, B.O. (1991) ‘Sectoral clustering and growth in American metropolitan

areas’, Regional Studies 25 (5):411–26.Huddleston, J.R. and Pangotra, P.P. (1990) ‘Regional and local economic impacts of

transportation investments’, Transportation Quarterly 44 (4):579–94.Hulten, C.R. and Schwab, R.M. (1991a) ‘Public capital formation and the growth of

regional manufacturing industries’, National Tax Journal 44 (4):121–34.Hulten, C.R. and Schwab, R.M. (1991b) ‘Is there too little public capital?

References 349

Infrastructure and economic growth’, Paper given at the Conference onInfrastructure Needs and Policy Options for the Nineties, American EnterpriseInstitute, Washington DC, February.

Hurd, R.M. (1924) Principles of City Land Values, New York: Real Estate RecordAssociation.

International Air Transport Association (1991) The Economic Benefits of AirTransport, Geneva: IATA.

International Road Federation (1992) World Transport Statistics 1988–1992, Geneva:IRF.

Isard, W. (1956) Location and Space-Economy, Cambridge MA: MIT Press.Jara-Diaz, S.R. (1986) ‘On the relationships between users’ benefits and the economic

effects of transportation activities’, Journal of Regional Science 26 (2):379–91.Jara-Diaz, S.R. and Farah, M. (1988) ‘Valuation of users’ benefits in transport systems’,

Transport Reviews 8 (3):197–218.Johansson, B. (1992) ‘Infrastructure, accessibility and economic growth’, paper given

at the 1992 Regional Science Association Annual Meeting, Chicago, Illinois.Johansson, B. and Karlsson, B. (1994) ‘Transportation infrastructure for the Malar

region of Sweden’, Regional Studies 28 (2):169–86.Johansson, P.-O. (1987) The Economic Theory and Measurement of Environmental

Externalities, Cambridge: Cambridge University Press.Johnston, J. (1984) Econometric Methods, New York: McGraw-HillJones, C.I. (1998) Introduction to Economic Growth, New York: Norton.Jones, S.R. (1982) An accessibility analysis of the impact of the M25 motorway’,

Transport and Road Research Laboratory, Report LR 1055.Jorgensen, D.W. (1963) ‘Capital theory and investment behavior’, Papers and

Proceedings of the American Economic Review 53:247–59.Jorgensen, D.W. (1991) ‘Fragile statistical foundations’, Paper given at the Conference

on Infrastructure Needs and Policy Options for the 1990s, American EnterpriseInstitute of Public Policy Research.

Kain, J. (1990) ‘Deception Dallas. Strategic misrepresentation in rail transit promotionand evaluation’, Journal of American Planning Association 56 (2):184–96.

Katz, M.L. and Schapiro, C. (1994) ‘Systems competition and network effects’, Journalof Economic Perspectives 23 (2):177–200.

Kay, J.A. (1993) ‘Efficiency and private capital in the provision of infrastructure’, inOECD Infrastructure Policies for the 1990s, Paris: OECD.

Kay, J.A., Mayer, C. and Thompson, D. (eds) (1986) Privatisation and Regulation:The UK Experience, Oxford: Clarendon Press.

Keating, G. (1992) ‘Toll tales on the highway to prosperity’, Independent, 16December: 21.

Keeler, T.E. and Ying, J. (1988) ‘Measuring the benefits of a large public investment:the case of US federal-aid highway system’, Journal of Public Economics 36 (1):64–86.

Kelejian, H.H. and Robinson, D.P. (1997) ‘Infrastructure productivity estimationand its underlying econometric specifications: a sensitivity analysis’, Papers inRegional Science 76 (1):115–31.

Kelsey, A. (1986) ‘How the stock market sees transport’, Journal of the CharteredInstitute of Transport February: 9–11.

Kenamotot, Y. and Mera, K. (1985) ‘General equilibrium analysis of the benefits of

350 References

large transportation improvements’, Regional Science and Urban Economics 15(3): 343–63.

Kennedy, G. (1991) ‘Glasgow International Airport—an economic pole’, Paper givenat the International Conference on Regional Airports, Rotterdam, December.

Kilkenny, M. (1998) ‘Transport costs and rural development’, Journal of RegionalScience 38 (2):293–312.

King, A. (1990) Global Cities: Post-Imperialism and the Internationalisation ofLondon, London: Routledge.

Knight, R. and Trygg, L. (1977) ‘Urban mass transit and land use impacts’,Transportation 5 (1):12–24.

Knight, R. and Trygg, L. (1978) ‘Evidence of land use impacts of rapid transit systems’,Transportation 6 (3):231–47.

Kondratieff, N.D. (1935) ‘The long waves in economic life’, Review XVII 6; 105–15.Kraus, M. (1981) ‘Scale economies analysis in urban highway networks’, Journal of

Urban Economics 9 (1):1–22.Kresge, D.T. and Roberts, P.O. (1971) ‘Systems analysis and simulation models’, in

J.R. Meyer (ed.) Techniques of Transport Planning, volume 2, Washington DC:Brookings Institute.

Kristiansen, J. (1993) ‘Regional transport infrastructure policies’, in D. Banister andJ.Berechman (eds.) Transport in Unified Europe, Amsterdam: North-Holland, pp.221–48.

Krugman, P. (1991a) ‘Increasing returns and economic geography’, Journal of PoliticalEconomy 99 (3):483–99.

Krugman, P. (1991b) Geography and Trade, Leuven: Leuven University Press andCambridge MA: MIT Press, p. 142.

Krugman, P. (1994) Peddling Prosperity, New York: Norton.Krugman, P. (1997) ‘Is capitalism too productive?’, Foreign Affairs 76 (5):79–94.Krugman, P. and Venables, A.J. (1990) ‘Integration and the competitiveness of

peripheral industry’, in C.J. Bliss and J. Braga de Macedo, (eds) Unity with Diversityin the European Community: The Community’s Southern Frontier, Cambridge:Cambridge University Press.

Kutay, A. (1988a) ‘Technological change and spatial transformation in an informationeconomy: 1. A structural model of transition in the urban system’, Environmentand Planning A 20 (5):569–93.

Kutay, A. (1988b) ‘Technological change and spatial transformation in an informationeconomy: 2. The influence of new information technology on the urban system’,Environment and Planning A 20 (6):707–18.

Kuznets, S. (1966) Modern Economic Growth, Cambridge MA: Yale University PressLakshmanan, T.R. (1989) ‘Infrastructure and economic transformation’, in A.E.

Andersson, D.Batten, B.Johansson and P.Nijkamp (eds) Advances in Spatial Theoryand Dynamics, Amsterdam: North Holland, pp. 241–61.

Landau, R. (1988) ‘US economic growth’, Scientific American 258 (6):44–52.Landau, R. (1990) ‘Capital investment: key to competitiveness and growth, Brookings

Review Summer: 52–6.Layard, R. and Glaister, S. (1994) Cost-Benefit Analysis, 2nd edn, New York:

Cambridge University Press.Levine, J. (1998) ‘Rethinking accessibility and jobs-housing balance’, Journal of

American Planning Association 64 (1):12–25.

References 351

Lex Motoring (1992) The Lex Report on Motoring 1992, Report produced by MORIfor Lex Motoring, London.

Linneker, B.J. and Spence, N.A. (1992) ‘An accessibility analysis of the impact of theM25, London Orbital Motorway on Britain’, Regional Studies 26 (1):31–47.

Linneker, B.J and Spence, N.A (1996) ‘Road transport infrastructure and regionaleconomic development: the regional development effects of the M25 London orbitalmotorway’, Journal of Transport Geography 4 (2):77–92.

Lipset, S.M. (1959) ‘Some social requisites of democracy: economic developmentand political legitimacy’, Review of American Political Science 53 (1):69–105.

Little, I.M. and Mirrlees, J.A. (1974) Project Evaluation and Planning, London:Heinemann.

Little, I.M. and Mirrlees, J.A. (1990) ‘The costs and benefits of analysis: projectappraisal and planning twenty years on’, Proceedings of the World Bank AnnualConference on Development, World Bank, Washington DC.

London Transport (1993) ‘Benefits of the Jubilee line extension’, London TransportPublic Relations Office, October, p. 8.

Lund, J.R. and Mokhtarian, P. (1994) ‘Telecommuting and residential location: theoryand implications for commute travel in the monocentric metropolis’, TransportationResearch Record 1463, Washington DC.

Lundvall, B.A. (ed.) (1992) National Systems of Innovation: Towards a Theory ofInnovation and Interactive Learning, London: Pinter.

Lynde, C. and Richmond, J. (1992) ‘The role of public capital in production’, Reviewof Economics and Statistics 74 (1):37–44.

McCann, J. (1995) ‘Rethinking the economics of location and agglomeration’, UrbanStudies 32 (3): pp. 563–79.

McFadden, D. (1978) “Cost, revenue and profit function’, in M. Fuss and D.McFadden (eds) Production Economies: A Dual Approach to Theory andApplication, New York: North Holland.

McGuire, A. (1983) ‘The regional income and employment impacts of nuclear powerstations’, Scottish Journal of Political Economy 30 (3):264–74.

McGuire, T. (1992) ‘Highways and macroeconomic productivity: phase two’, FinalReport, Federal Highway Administration, Washington DC.

Mackett, R.L. (1980) ‘The relationship between transport and the viability of centraland inner urban areas’, Journal of Transport Economics and Policy 14 (3):267–94.

Mackie, P.J. (1996) ‘Induced traffic and economic appraisal’, Transportation 23 (1):103–19.

Mackie, P.J. and Simon, D. (1986) ‘Do road projects benefit industry?’ Journal ofTransport Economics and Policy 20 (3):377–84.

McKinnon, A. and Woodburn, A. (1994) ‘The consolidation of retail deliveries: itseffects on CO2 emissions’, Transport Policy 1 (2):125–36.

Maddison, D., Pearce, D., Johansson, O., Calthrop, E., Litman, T. and Verhoef, E.(1996) The True Cost of Road Transport, Blueprint 5, London: Earthscan.

Madre, J.-L. and Lambert, T. (1989) Prévisions à long terms du trafic automobile,Collection of reports by the CREDOC, Paris.

Majd, S. and Pindyck, R. (1987) ‘Time to build, option value, and investmentdecisions’, Journal of Financial Economics 18 (1):7–27.

Marquand, D. (1980) ‘Measuring the effects and costs of regional incentives’,Government Economic Services, Paper No. 32.

352 References

Marshall, A. (1920) Principles of Economics, 8th edn, London: Macmillan.Martellato, D. and Nijkamp, P. (1996) ‘The concept of accessibility revisited’, Paper

given at the World Congress of Regional Science Association, Tokyo, May.Martellato, D., Nijkamp, P. and Reggiani, A. (1995) ‘Measurement and measures of

network accessibility: economic perspectives’, Draft.Masser, I., Sviden, O. and Wegener, M. (1992) The Geography of Europe’s Futures,

London: Belhaven.Mensch, G. (1979) Stalemate in Technology: Innovations Overcome the Depression,

Cambridge MA: Ballinger Press.Mera, K. (1973) ‘Regional production functions and social overhead capital: an

analysis of the Japanese case’, Regional and Urban Economics 3 (2):157–85.Mills, E. and McDonald, J. (eds) (1992) Sources of Metropolitan Growth, New

Brunswick: Center for Urban Policy Research, Rutgers, the State University ofNew Jersey.

Millward, R. (1986) ‘The comparative performance of public and private enterprises’,in J.A.Kay, C.Mayer and D.Thompson (eds) Privatisation and Regulation: TheUK Experience, Oxford: Clarendon Press.

Mishan, E.J. (1969) Welfare Economics: An Assessment, Amsterdam: North Holland.Mishan, E.J. (1988) Cost-Benefit Analysis, 4th edn, London: Allen & Unwin.Mitchell, B.R. (1964) ‘The coming of the railway and UK economic growth’, Journal

of Economic History 24 (3):315–36.Mohring, H. (1976) Transportation Economics, Cambridge MA: Ballinger Press.Mohring, H. (1993) ‘Maximizing, measuring, and not double counting transportation-

improvement benefits: a primer on closed- and open-economy cost-benefit analysis’,Transportation Research B 27(6), 413–424.

Mohring, H. (1994) ‘Land rents and transport improvements: some urban parables’,Transportation 20 (3):297–84.

Mohring, H. and Harwitz, M. (1962) Highway Benefits: An Analytical Framework,Evanston IL: Northwestern University.

Mohring, H. and Williamson, Jr, H.F. (1969) ‘Scale and industrial organization -economies of transport investment’, Journal of Transport Economics and Policy3 (3):251–71.

Mokhtarian, P. (1996) ‘A synthetic approach to estimating the impacts oftelecommuting on travel’, Paper given at the TMIP Conference on Urban Design,Telecommunications and Travel Behaviour, Williamsburg, October.

Moomaw, R. (1981) ‘Productivity and city size: a critique of the evidence’, QuarterlyJournal of Economics 94 (4):675–88.

Moreno, R., Artis, M., López-Bazo, E. and Suriñach, J. (1997) ‘Evidence of the complexlink between infrastructure and regional growth’, International Journal ofDevelopment Planning Literature 12 (1/2):81–108.

Morrison, C.J. and Schwartz, A.E. (1996) ‘State infrastructure and productiveperformance’, American Economic Review 86 (5):1095–1111.

Munnell, A.H. (1990a) ‘Why has productivity growth declined? Productivity andpublic investment’, New England Economic Journal January–February:4–22.

Munnell, A.H. (1990b) (with Cook, L.M.) ‘How does public infrastructure affectregional economic performance’, New England Economic Review September–October:11–32.

Munnell, A.H. (1992) ‘Infrastructure investment and economic growth’, EconomicPerspective 6 (4):189–98.

References 353

Munnell, A.H. (1993) ‘An assessment of trends in and economic impacts ofinfrastructure investment, in OECD’ Infrastructure Policies for the 1990s, Paris:OECD, pp. 21–54.

Munro-Lafon, J.P. and Mussett, J.W. (1994) ‘European inter-urban toll roads’, Papergiven at the PTRC Annual Conference, London, September.

Musgrave, R.A. and Musgrave, P.B. (1989) Public Finance in Theory and Practice,5th edn, New York: McGraw-Hill.

Muth, R.F. (1985) ‘Models of land-use, housing, and rent: an evaluation’, Journal ofRegional Sciences 25 (4):593–606.

MVA Consultancy (1987) ‘The value of travel time savings: a report of researchundertaken for the Department of Transport’, Institute of Transport StudiesUniversity of Leeds, Transport Studies Unit of Oxford, Newbury, England.

Myrdal, G. (1957) Economic Theory and Under-Developed Regions, London:Duckworth.

Nadiri, I.M. and Mamuneas, T.P. (1991) ‘The effect of public infrastructure andR&D capital on the cost structure and performance of US manufacturingindustries’, Cambridge MA: National Bureau of Economic Research, WorkingPaper No. 3887.

Nadiri, I.M. and Mamuneas, T.P. (1994) ‘The effect of public infrastructure andR&D capital on the cost structure and performance of US manufacturing industries’, Review of Economics and Statistics 76 (1):22–37.

Nakamura, H. and Ueda, T. (1989) ‘The impacts of the Shinkansen on regionaldevelopment’, Proceedings of the 5th World Conference on Transport Research,Yokohama, Volume III, Ventura, California: Western Periodicals.

Nash, C. (1993) ‘Cost-benefit analysis of transport projects’, in A.Williams and E.Giardina, (eds) Efficiency in the Public Sector: The Theory and Practice of Cost-Benefit Analysis, Aldershot: Edward Elgar, pp. 83–105.

Nathaniel Lichfield and Partners and Goldstein Leigh Associates (1981) The PropertyMarket Effects of the M25, January.

National Economic Development Council (1985) A Fairer and Faster Route to MajorRoad Construction, London: NEDC.

National Economic Development Council (1991) A Road User Charge? Londoners’Views, report prepared by the Harris Research Centre for the National EconomicDevelopment Office, the London Planning Advisory Committee and theAutomobile Association, London.

Nelson, R. and Winter, S.G. (1982) An Evolutionary Theory of Economic Changes,Cambridge MA: Harvard University Press.

Newman, P. (1997) ‘Urban infrastructure—reshaping cities for a more sustainablefuture’, Inquiry by the Australian Academy of Technological Sciences andEngineering, Support Document 5 from Task Group 6 on Air Pollution in Australia,Melbourne.

Newman, P. and Kenworthy, J. (1989) Cities and Automobile Dependence, Aldershot:Gower.

Newman, P. and Vickerman, R.W. (1993) ‘Infrastructure indicators and regionaldevelopment: a critique of accessibility and potential measures’, Paper in theRegional Science Association Annual Conference, Nottingham, September.

Newton, P.C. (1997) Urban Infrastructure ‘Reshaping Cities for a more SustainableFuture’. An inquiry by the Australian Academy of Technological Sciences andEngineering on urban air pollution in Australia, Report No. 5.

354 References

Niagara Frontier Transportation Commission (NFTC) (1977) Technical Memo,No. 7.

Niemeier, D. (1997) ‘Accessibility: an evaluation using consumer welfare’,Transportation 24 (4):377–96.

Nijkamp, P. (1986) ‘Infrastructure and regional development: a multi-dimensionalpolicy analysis’, Empirical Economics 11 (1):1–21.

Nijkamp, P. and Salomon, I. (1989) ‘Future spatial impacts of telecommunications’,Transportation Planning and Technology 13 (3):275–87.

Nijkamp, P., Peping, G. and Banister, D. (1996) Telematics and Transport Behaviour,Berlin: Springer Verlag.

Nilles, J.M. (1991) ‘Telecommuting and urban spread: mitigator or incitor?’,Transportation 18 (4):411–31.

Oppenheim, N. (1980) Applied Models in Urban and Regional Analysis, New York:Prentice Hall.

Orfeuil, J.P. (1992) ‘Structural changes in population and impact on passengertransport’, European Conference of Ministers of Transport, Round Table 88, Paris,pp. 45–102.

Organisation for Economic Co-operation and Development (OECD) (1989)‘Comparative socio demographic trends: initial data on ageing populations’, OECDGroup Urban Affairs UR/TSDA (89) 1, April.

Organisation for Economic Co-operation and Development/ECMT (1995) UrbanTravel and Sustainable Development, Paris: OECD.

Paaswell, R.E. and Berechman, J. (1981) ‘An analysis of rapid transit investments’,Final Report, US Department of Transportation, Urban Mass TransportationAdministration, UMTA-NY-11–0022.

Paaswell, R.E., Cirrincione, M., McNally, M. and Parker-Simon, K. (1979) ‘Profileof attitudes toward downtown and the viability of central and inner urban areas’,Working Paper No. 6, Analysis of Joint Development Projects, Department ofEnvironmental Design and Planning, SUNY Buffalo.

Paddock, J., Siegel, D. and Smith, J. (1988) ‘Option valuation of claims on real assets:the case of offshore petroleum leases’, Quarterly Journal of Economics 103 (3):479–508.

Papageorgiu, Y. and Smith, T. (1983) ‘Agglomeration as a local instability of uniformsteady states’, Econometrica 51 (4):1109–19.

Parker-Simon, K. and Paaswell, R. (1980) ‘Population and economic data in transitanalysis’, Journal of Urban Planning Division (ASCE) 106 (2):43–58.

Parkinson, M. (1981) ‘The effect of road investment on economic development inthe UK’, Department of Transport, Government Economic Service Working PaperNo. 43.

Parsons, D. (1984) ‘Employment stimulation and the local labour market’, RegionalStudies 18 (5):423–8.

Peake, S. (1994) Transport in Transition: Lessons from the History of Energy, London:Earthscan.

Peaker, A. (1976) ‘New primary roads and sub-regional economic growth’, RegionalStudies 10 (1):11–13.

Pendyala, R., Goulias, K. and Kitamura, R. (1991) ‘Impact of telecommuting onspatial and temporal patterns of household travel’, Transportation 18 (4):383–409.

Peterson, G.E. (1991) ‘Historical perspective on infrastructure investment: how didwe get where we are?’, American Enterprise Institute, Discussion Paper, February.

References 355

Phelps, N. (1993) ‘Branch plants and the evolving spatial division of labour: a studyof material linkage change in the northern region of England’, Regional Studies27 (2):87–101.

Pickrell, D. (1989) ‘Urban rail transit projects: forecast versus actual ridership andcosts’, US Department of Transportation, Transportation System Center,Cambridge, MA.

Pieda (1991) ‘Rail link project: a comparative appraisal of socio economic anddevelopment impacts of alternative routes’, Reading: Pieda.

Pieda (1994) Heathrow Terminal 5: Assessment of Employment Impact, 1991–2016,Reading: Pieda, BAA/1201.

Pieda (1995) The Economic Significance of Heathrow Airport, Reading: Pieda, BAA/1204.

Pindyck, R. (1988) ‘Irreversible investment, capacity choice, and the value of thefirm’, American Economic Review 78 (5):969–85.

Pindyck, R. (1991) ‘Irreversibility, uncertainty, and investment’, Journal of EconomicLiterature 29 (3):1110–48.

Pinnoi, N. (1991) ‘Public capital stock and state productivity growth: further evidencefrom error component model’, Paper given at the North American Meetings ofthe Regional Science Association, New Orleans, 7–10 November.

Piore, M.J. and Sabel, C. (1984) The Second Industrial Divide, New York: BasicBooks.

Plaut, R (1997) Transportation-communications relationships in industry,Transportation Research A 31 (6):419–29.

Pollak, R. and Wacher, M.L. (1975) ‘The relevance of household production functionand its implications for the allocation of time’, Journal of Political Economy April:255–77.

Port Authority of New York and New Jersey (1990) The Economic Impact of theAviation Industry on the New York/New Jersey Economic Region, New York:Port Authority.

Porter, M.E. (1990) The Competitive Advantage of Nations, New York: Free Press.Porter, M.E. and Van der Linde, C. (1995) ‘Towards a new concept of the environment

competitiveness relationship’, Journal of Economic Perspectives 9 (4):97–118.Poulton, M.C. (1980) ‘The relationship between transport and the viability of central

and inner urban areas’, Journal of Transport Economics and Policy 14 (3):249–66.

Prest, A.R. and Turvey, R. (1965) ‘Cost-benefit analysis: a survey’, Economic Journal75 (4):683–735.

Procter, S. (1988) ‘M25—mixed blessing? Catalyst for regional planning’, Proceedingsof the Town and Country Planning Summer School, University of Lancaster,September, pp. 71–2.

Progress (1998) ‘Surface transportation policy’, Progress 7 (4):5.Prud’homme, R. (1993) Assessing the Role of Infrastructure in France by means of

Regionally Estimated Production Functions, Paris: Observatoire de l’Economie etdes Institutions Locales.

Quiggin, J. (1996) ‘Private sector involvement in infrastructure projects’, AustralianEconomic Review 113 (1):51–64.

Quigley, J. (1998) ‘Urban density and economic growth’, Journal of EconomicPerspectives 12 (2):127–38.

Quinn, D.J. (1986) ‘Accessibility and job search: a study of unemployed school leavers’,Regional Studies 20 (2):163–73.

356 References

Ramjerdi, F. (1993) ‘Value of time savings; theories and empirical evidence’, TOIReport, 213/1993, Institute of Transport Economics, Norwegian Center forTransport Research, Oslo.

Reid, E. (1994) ‘Beyond density, mode choice, and single-purpose trips’, Draft, JointCenter for Environmental and Urban Problems, Florida Atlantic University/FloridaInternational University.

Rendel, Palmer and Tritton (RPT) (1989) ‘M25 review—summary report’, preparedfor the Department of Transport, London: HMSO.

Rephann, T.J. (1993) ‘Highway investment and regional economic development:decision methods’ and empirical foundations’, Urban Studies 30 (2):437–50.

Rephann, T.J. and Isserman, A. (1994) ‘New highway as economic developmenttools: an evaluation using quasi-experimental matching methods’ Regional ScienceResearch Institute, West Virginia University, Research Paper 9313.

Rienstra, S., Rietveld, P., Hilferink, M. and Bruinsma, F. (1998) ‘Road infrastructureand corridor development’, in L.Lundqvist, L.-G.Mattsson and T.J Kim (eds)Network Infrastructure and the Urban Environment: Advances in Spatial SystemsModelling, Berlin: Springer Verlag, pp. 395–414.

Rietveld, P. (1989) ‘Infrastructure and regional development’, Annals of RegionalScience 23 (2):255–74.

Rietveld, P. (1994) ‘Spatial economic impacts of transport infrastructure supply’,Transportation Research A 28 (4):329–34.

Rietveld, P. and Bruinsma, F. (1998) Is Transport Infrastructure Effective? TransportInfrastructure and Accessibility: Impacts on the Space Economy, Berlin: SpringerVerlag .

Rietveld, P. and Nijkamp, P. (1993) ‘Transport and regional development’, in J.Polak and A.Heertje (eds) European Transport Economics, Paris: ECMT, pp.130–51.

Rietveld, P., Van de Velde, R. and Hilferink, M. (1998) ‘Spatial economic effects ofairports: an ex ante analysis for the Netherlands’, Paper given at the 4th NECTAREuroConference, Israel, April.

Robertson, J.A.W. (1994) ‘Airports and economic regeneration’, in PTRC, Proceedingsof Seminar B: Airport Planning Issues, presented at the 22nd European TransportForum.

Romer, P.M. (1986) ‘Increasing returns and long-run growth’, Journal of PoliticalEconomy 94 (5):1002–38.

Romer, P.M. (1996) ‘Why, indeed, in America? Theory, history and the origins ofmodern economic growth’, American Economic Review 86 (2):202–6.

Rostow, W.W. (1960a) The Stages of Economic Growth, Cambridge: CambridgeUniversity Press.

Rostow, W.W. (1960b) The Process of Economic Growth, Oxford: Clarendon Press,pp. 302–3 .

Roy, R. (1994) ‘Investment in transport infrastructure—the recovery of Europe’,European Centre for Infrastructure Studies (ECIS) Report, Rotterdam,November.

Sabel, C.F. (1993) ‘Studied trust: building new forms of co-operation in a volatileeconomy’, in D.Foray and C.Freeman (eds) Technology and the Wealth of Nations,London: Pinter.

SACTRA (Standing Advisory Committee on Trunk Road Assessment) (1977) ‘Reportof the Advisory Committee on Trunk Road Appraisal, Chairman Sir George Leitch’,

References 357

Standing Advisory Committee on Trunk Road Assessment, London: HMSO,October.

SACTRA (Standing Advisory Committee on Trunk Road Assessment) (1994) ‘Trunkroads and the generation of traffic, Report of the Standing Advisory Committeeon Trunk Road Assessment, Chairman D.Wood’, London: HMSO, December.

SACTRA (Standing Advisory Committee on Trunk Road Assessment) (1999)‘Transport and the economy, Report of the Standing Advisory Committee on TrunkRoad Assessment, Chairman E.Mackay’, London: The Stationary Office.

Sako, M. (1992) Prices, Quality and Trust: Inter-firm Relations in Britain and Japan,Cambridge: Cambridge University Press.

Sandomo, A. and Dreze, J. (1971) ‘Discount rates for public investment in closed andopen economies’, Economica November:395–412.

Sands, B. (1993) ‘The development effects of high-speed rail stations and implicationsfor California’ , Built Environment 19 (3/4):257–84.

Sassen, S. (1991) The Global City: New York, London, Tokyo, Princeton NJ: PrincetonUniversity Press.

Schleicher-Tappeser, R., Hey, C. and Steen, P. (1998) ‘Policy approaches for decouplingfreight transport from economic growth’, Paper given at the World Conference onTransport Research, Antwerp, July.

Schultze, C. (1990) ‘The federal budget and the nation’s economic health’, in H.Aaron (ed.) Setting National Priorities: Policies for the Nineties, Washington DC:Brookings Institute.

Schumpeter, J.A. (1939) Business Cycles, A Theoretical, Historical and StatisticalAnalysis of the Capitalist Process, New York: McGraw-Hill.

Schwartz, A. (1992) ‘Corporate service linkages in large metropolitan areas: a studyof New York, Los Angeles and Chicago’, Urban Affairs Quarterly 28 (2):276–96.

Scott, A.J. (1983) ‘Industrial organization and the logic of intra-metropolitan locationI: theoretical considerations’, Economic Geography 59 (2):133–250.

Scott, A.J. (1990) New Industrial Spaces, London: Pion.Secretary of State for Transport (1996) Transport—the Way Ahead, London: HMSO,

Cmnd 3234, April.Seitz, H. (1993) ‘A dual economic analysis of the benefits of the public road network’,

Annals of Regional Science 27 (2):223–39.Selling, A., Allanach, C. and Loveridge, C. (1994) ‘The role of agglomeration

economies in firm location: a review of the literature’, Staff Paper P94–14,Department of Agriculture and Applied Economics, University of Minnesota.

Shah, A. (1992) ‘Dynamics of public infrastructure, industrial productivity andprofitability’, Review of Economics and Statistics 74 (1):28–36.

Sheate, W. (1992) ‘Strategic environmental impact assessment in the transport sector’,Project Appraisal 7 (3):170–4.

Short, J. (1993) ‘Environment, global and local effects’, in Proceedings of the 12thInternational Symposium on Theory and Practice in Transport Economics—Transport Growth in Question, Paris: ECMT, pp. 579–608.

Simmie, J. (1998) ‘Innovate or stagnate: economic planning choices for localproduction nodes in the global economy’, Planning Practice and Research 13(1):35–51.

Simmie, J. and Kirby, M. (1996) ‘Innovation and the theoretical bases of technopoleplanning’, University College London, Planning and Development Research Centre,Working Paper 14 , January.

358 References

Simmons, M. (1985) ‘Orbital motorway that reinforces the South East’, Town andCountry Planning 54 (2):132–4.

Simon, H. (1993) ‘Data collection on indirect impacts’, Seminar on the EconomicImpact of Airports, Airports Council International, Munich, March.

Skamris, M. and Flyvbjerg, B. (1997) ‘Accuracy of traffic forecast and costestimates on large transportation projects’, Transportation Research Record1518:65–9.

Small K. (1982) ‘The scheduling of consumer activities: work trips’, AmericanEconomic Review 72 (3):467–79.

Small, K. (1992) Urban Transportation Economics, Chur: Harwood AcademicPublishers.

Small, K. (1999) ‘Project evaluation’, in J.Gomez-Ibanez and W.Tye (eds)Transportation Policy and Analysis: A Handbook in Honour of John R.Meyer,Cambridge: Brookings Institution Press.

Small, K. and Kazimi, C. (1995) ‘On the cost of air pollution from motor vehicle’,Journal of Transport Economics and Policy 29 (1):7–32.

Small, K. and Rosen, H. (1981) ‘Applied welfare economics with discrete choicemodels’, Econometrica 49 (1):105–30.

Small, K., Winston, C. and Evans, C. (1989) Road Work: A New Highway Price andInvestment Policy, Washington DC: Brookings Institute.

Smith, A. (1967) Wealth of Nations, Chicago: University of Chicago Press, Volume2, Book V, Part 3, p. 244.

Smyth, A. and Klavinskis, C. (1993) ‘Peripherality, accessibility and transport relatedcosts—empirical evidence and policy implications’, Warwick: PTRC.

SNCF (Société Nationale des Chemins de Per) (1999) ‘SNCF’s high-speed air-raillink’, Japan Railway and Transport Review 19:30.

Soet, M.E.de and Stevers, R.A. (1994) ‘Possible scenarios for mobility in a sustainablesociety. A conceptual experiment’, Ministry of Transport and Public Works—DGfor Public Works and Water Management, the Netherlands.

Solow, R.M. (1956) ‘A contribution to the theory of economic growth’, QuarterlyJournal of Economics 70 (1):65–94.

Solow, R. and Vickrey, W. (1971) ‘Land use in a long narrow city’, Journal of EconomicTheory 3 (4):430–47.

Stamp, J. (1997) ‘Case study: Lakeside retail development’, unpublished paper forMPhil Town Planning, University College London, December.

Standing Conference on London and South East Regional Planning (1982) ‘The impactof the M25’, Report by the Industry and Commerce Working Party of the RegionalMonitoring Group, SC1706, December.

Stevens, B. (1992) ‘Prospects for privatisation in OECD countries’, NationalWestminster Bank Quarterly Review August: 2–22.

Stiglitz, J. (1982) ‘The rate of discount for benefit-cost analysis and the theory ofsecond best’, in R.Layard and S.Glaister (eds) Cost-Benefit Analysis, Cambridge:Cambridge University Press.

Stokes, G. (1994) ‘Travel time budgets and their relevance for forecasting the futureamount of travel’, Warwick: PTRC.

Stopher, P. (1993) ‘Financing urban rail projects: the case of Los Angeles’,Transportation 20 (3):229–50.

Storper, M. (1993) ‘ “Regional worlds” of production: learning and innovation in thetechnology districts of France, Italy and the USA’, Regional Studies 27 (5):433–55.

References 359

Storper, M. and Christopherson, S. (1987) ‘Flexible specialisation and regionalindustrial agglomeration: the case of the US motion picture industry’, Annals ofthe Association of American Geographers 77 (1):104–17.

Streeter, W. (1993) ‘The French train á grand vitesse’, Built Environment 19 (3/4):184–202.

Summers, R. and Hestan, A. (1991) ‘ThePennWorldTable (MK5): anexpanded setof international comparisons l950–1988’,QuarterlyJournalofEconomics 106(2):327–68.

Szyrmer, J.M. (1985) ‘Measuring connectedness of input-output models: 1. Survey ofthe measures’, Environment and Planning A 17 (12):1591–612.

Szyrmer, J.M. (1986) ‘Measuring connectedness of input-output models: 1. Totalflow concept’, Environment and Planning A 18 (1):107–21.

Taaffe, E.J., Morrill, R.C. and Gould, P.R. (1963) ‘Transport expansion inunderdeveloped countries: a comparative analysis’, Geographical Review 53(4):503–21.

Taniguchi, M. (1993) ‘The Japanese Shinkansen’, Built Environment 19 (3/4):215–21.

Tatom, J.A. (1991) ‘Public capital and private sector performance’, Federal ReserveBank of St Louis Review 73 (3):3–15.

Tatom, J.A. (1993) ‘Shifting perspectives on the role of public capital formation’,Federal Reserve Bank of St Louis Review 75.

Taylor, R.G. (1951) The Transportation Revolution 1815–1860, New York: Sharpe,Chaps 2, 3.

Tengström, E. (1992) The Use of the Automobile: Its Implications for Man, Societyand the Environment, Stockholm: Swedish Transport Research Board.

Tinbergen, J. (1957) ‘The appraisal of road constructions: two calculation schemes’,Review of Economic Studies 39 (3):241–9.

Toffler, A. (1991) Power Shift: Knowledge, Wealth and Violence at the Edge of the21st Century, London: Bantam.

Train, K. and McFadden, D. (1978) ‘The goods/leisure tradeoff and disaggregatework trip mode choice models’, Transportation Research 12 (4):349–53.

Transportation Research Board (1995) ‘Expanding metropolitan highways:implications for air quality and energy use’, Special Report 245, Washington DC.

Troin, J.F. (1995) Rail et amènagement du territoire. Des héritage aux nouveauxdéfis, Aix-en-Provence: Edisud.

Truong, T.K. and Hensher, D.A. (1985) ‘Measurement of travel time values and theopportunity cost from a discrete choice model’, Economic Journal 95 (4):438–51.

Turok, I. (1993) ‘Inward investment and local linkages: how deeply embedded is“silicon glen” ?’ Regional Studies 27 (5):401–17.

Twomey, J. and Tomkins, J. (1995) ‘Development effects at airports: a case study ofManchester Airport’, in D.Banister (ed.) Transport and Urban Development,London: Spon, pp. 187–211.

Uchimura, K. and Gao, H. (1993) ‘The importance of infrastructure on economicdevelopment’, World Bank, Latin American and Caribbean Regional Office,Washington DC.

UK Department of the Environment (1993) Alternative Settlement Patterns, London:HMSO.

UK Department of Transport (1980) Policy for Roads: England 1980, London:HMSO, Cmnd 7980.

UK Department of Transport (1983) Policy for Roads in England 1983, London:HMSO, Cmnd 9059.

360 References

UK Department of Transport (1985) Airports Policy, White Paper, London: HMSO,Cmnd 9542.

UK Department of Transport (1985) Policy for Roads, England 1985, London:HMSO, Cmnd 7908.

UK Department of Transport (1989) Roads for Prosperity, London: HMSO, Cmnd693.

UK Department of Transport (1990) Trunk Roads, England into the 1990s, Reportprepared by Road Programme and Resources Division, Department of Transport,London: HMSO.

UK Department of Transport (1993) Paying for Better Motorways: Issues forDiscussion, London: HMSO.

UK Department of Transport (1996) Transport—The Way Forward. TheGovernment’s Response to the Great Transport Debate, London: HMSO, Cmnd3234.

UK Department of Transport (1997) Transport Statistics Great Britain 1986–1996,London: HMSO.

UK Department of Transport (various) National Travel Survey, Government StatisticalOffice, London: HMSO.

US Department of Commerce, Bureau of the Census (1990) Census of Populationand Housing, Washington D.C.

US Department of Transportation (DOT), Federal Highway Administration (1992)‘Assessing the relationships between transportation infrastructure investment andproductivity’, a Policy Discussion Paper, No. 4, August.

US Department of Transportation (DOT) (1993) ‘1990 NPTS Databook: nationalpersonal transportation survey’, prepared by the Oakridge National Laboratory,FHWA-PL-94–010A, Washington DC, November.

US Department of Transportation (DOT) Bureau of Transportation Statistics (1996)‘Transportation statistics annual report—transportation and the environment’,US Department of Transportation, Washington DC.

US Office of Management and Budget (OMB) (1992) ‘Guideline and discount ratesfor benefit-cost analysis of federal programs’, Circular No. A-94, Revised (29October), Section 8.

Van den Berg, L., Van Klink, H.A. and Pol, P.M.J. (1994) ‘From airport to growthpole: regional economic effects, airport exploitation and strategic cooperation’,Paper given at the PTRC Annual Conference, London, September.

Vickerman, R. (1994) ‘Transport infrastructure and region building in the EuropeanCommunity’, Journal of Common Market Studies 32 (1):1–24.

Vickerman, R. (1995) ‘Location, accessibility and regional development: the appraisalof trans-European networks’, Transport Policy 2 (4):225–34.

Vickerman, R.W. (1998) ‘Transport provision and regional development in Europe:towards a framework for appraisal’, in D. Banister (ed,) Transport Policy and theEnvironment, London: Spon, pp. 128–57.

Vickerman, R., Spiekermann, K. and Wegener, M. (1999) ‘Accessibility and economicdevelopment in Europe’, Regional Studies 33 (1):1–16.

Vickers, J. and Yarrow, G. (1988) Privatisation: An Economic Analysis, CambridgeMA: MIT Press.

Vickrey, W. (1977) ‘The city as a firm’, in M.S.Feldstein and R.P.Inman (eds) Economicsof Public Services, New York: Macmillan, pp. 334–43.

Viscusi, K. (1993) ‘The value of risks to life and death’, Journal of Economic Literature31 (4):1912–46.

References 361

Viton, P. (1992) ‘Consolidations of scale and scope in urban transit’, Regional Scienceand Urban Economics 22 (1):24–49.

Von Thünen, J.H. (1826) Der isolierte Staat in Beziehung auf Landwirtschaft undNationalokonomie, trans C.M.Wortenberg (1966), Oxford: Pergamon Press.

Vreckem, D.van (1993) ‘European Community policy on taxes and charges in theroad transport sector’, Paper given at the Joint OECD/ECMT Seminar onInternalising the Social Costs of Transport, Working Document 4, Paris, November.

Walmsley, D.A .and Pickett, M.W. (1992) ‘The costs and patronage of rapid transitsystems compared with forecasts’, Research Report 352, Transport ResearchLaboratory, Department of Transport, UK.

Waters II, W.G. (1992) ‘The value of travel time savings for the economic evaluationof highway investments in British Columbia’, Center for Transportation Studies,University of British Columbia’, Vancouver, BC.

Waters II, W.G., Wong, C. and Megale, K. (1995) ‘The value of commercial vehicletime savings for the evaluation of highway investment: a resource savingsapproach’, Journal of the Transportation Research Form 35 (1):97–113.

Webb, M.G. (1973) The Economics of Nationalized Industries: A TheoreticalApproach. London: Nelson.

Webber, M. (1963) ‘Order and diversity: community without propinquity’, in L.Wingo, Jun. (ed.) Cities and Space: The Future Use of Urban Land, Baltimore,MD: Johns Hopkins University Press, pp. 23–54.

Webber, M. (1976) ‘The BART experience—what have we learned?’ The Public Interest46 (3):79–108.

Weber, A. (1956) Theory of the Location of Industry, (trans) C.J.Friedrich, Chicago:University of Chicago Press.

Wegener, M. (1995) ‘Accessibility and development impacts’, in D.Banister (ed.)Transport and Urban Development, London: Spon, pp. 157–61.

Weijers, S. (1995) ‘Future positions and strategies of Dutch shippers and carriers’, inD.Banister, R.Capello and P.Nijkamp (eds) European Transport andCommunications Networks: Policy Evolution and Change, Chichester: John Wiley,pp. 247–64.

Wellard, I. (1993) ‘The Paris case study’ Seminar on the Economic Impact of Airports’,Munich, Airports Council International, March.

Wheaton, C.W. (1978) ‘Price-induced distortions in urban highway investment’, BellJournal of Economics 9 (2):622–32.

Whitelegg, J. (1993) Transport for a Sustainable Future: The Case for Europe, London:Belhaven.

Williams, H.C.W.L. and Moore, L.A.R. (1990) ‘The appraisal of highway investmentunder fixed and variable demand’, Journal of Transport Economics and Policy 24(1): 61–81.

Williamson, O.E. (1985) The Economic Institutions of Capitalism, New York: FreePress.

Williamson, O.E. (1986) Economic Organisation: Firms, Markets and Policy Control,Brighton: Wheatsheaf.

Willig, R. (1976) ‘Consumer’s surplus without apology’, American Economic Review66:589–97.

Willson, R.W. (1992) ‘Estimating the travel and parking demand effects of employer-paid parking’, Regional Science and Urban Economics 22:

Wilson, G.W. (1978) ‘The role of transportation in regional economic growth’, in

362 References

N.H.Lithwick (ed.) Regional Economic Policy: The Canadian Experience, Toronto:McGraw-Hill.

Wilson, J. (1983) ‘Optimal road capacity in the presence of unpriced congestion’,Journal of Urban Economics 13 (3):337–57.

Wingo, L. (ed.) (1963) Cities and Space: The Future Use of Urban Land, Resourcesfor the Future, Baltimore, MD: Johns Hopkins Press.

Winston, C. (1991), ‘Efficient transportation infrastructure policy, Journal ofEconomic Perspectives 5 (1):113–27.

World Bank (1994) World Development Report 1994: Infrastructure andDevelopment, Oxford: Oxford University Press.

World Bank (1996) Sustainable Transport. Priorities for Policy Reform, WashingtonDC: World Bank.

World Commission on Environment and Development (1987) Our Common Future,Chair: Gro Harlem Brundtland, Oxford: Oxford University Press.

Wylie, P.J. (1996) ‘Infrastructure and Canadian economic growth 1946–1991’,Canadian Journal of Economics (Special Issue Part 1) 29:350–5.

Yang, H. and Huang, H. (1998) ‘Principle of marginal-cost pricing: how does itwork in a general road network?’, Transportation Research A 32 (1):45–54.

York Consulting (1991) Economic Development Potential at Manchester Airport,York: York Consulting Group.

York Consulting (1994) Manchester Airport Second Runway, Proof of Evidence onEconomic Impact, York: York Consulting Group, Proof of Evidence given byNigel Mason, May.

Young, A. (1992) ‘A tale of two cities’, in O. Blanchard and S. Fischeer (eds) NBERMacroeconomics Annual, Cambridge, MA: MIT Press.

Youngson, A.J. (1967) Overhead Capital, Edinburgh: Edinburgh University Press.Zahavi, Y. (1982) ‘Travel transferability between four cities’, Traffic Engineering

and Control 23 (4):203–8.Zembri-Mary, G. (1996) ‘Relationship between transport and regional development.

The case of motorway effects on land values and policies in France: markets andactors during the construction of the A71 (Bourges-Clermont Ferrand), Papergiven at the European Regional Science Association 36th European Congress,ETH Zurich, Switzerland, August.

Index

Accessibility 11, 12, 23, 26, 36, 44, 45,

46, 47, 49, 52–53, 113, 124, 148,168, 171, 174, 176, 191, 196,208, 211, 217, 221, 227, 230–233,239, 242, 247, 252, 258, 263,266, 267, 271, 275, 278, 279,282, 301, 302, 321, 324, 330,331–332

benefits 197effects 39model 265, 269–271

Acid rain 119Activity

centres 213–214relocation 163, 211

Additional growth 319Additionality 77, 305Agglomeration economies 46, 47, 92,

94–95, 114, 134, 168–169, 212–214,216, 218, 219, 222, 228, 247, 249,279, 328

Aggregate output 134Air quality standards 123–124Air transport 119, 287–316Allocation function 268Allocative externalities 167, 169–170,

172, 174, 187, 201Amsterdam 251

orbital motorway 239, 251–253Attracted employment 291 Backword linkages 212BART system 257 see San FranciscoBehaviour 45–47 see Travel, TransportBelgium 181–182, 191Benefit-cost analysis 26, 157, 161, -197

see Cost benefit analysisBenefit cost comparisons 192Benefit cost ratio 153

Blue Water Park development 248British Columbia 182, 187Buffalo 286

Buffalo Light Rail Rapid Transit 257,258–277

Bus 276network 263

Business confidence 253 Calgary 277Canada 182Canal 8Capacity investment function 201

limitations 244Capital accumulation 326

productivity 138sunk costs 190

Car ownership 85–87, 110, 112, 119,Cars 244Case studies 28Causal linkage 37Causality 36, 62, 115, 134, 135, 144,

146, 148, 149, –150, 156, 174,253, 329–331

mechanism 33paradigm 41

CBD 259, 262, 267, 271, 272, 273–274,275–276, 276, 278

Central place theory 9Channel Tunnel 72CO2 emissions 115, 119Combined multiplier 311 see MultiplierCommercial traffic 180–182 see FreightCommunications 96 see

Telecommunications networks 321Compact city 121 see Sustainability

developmentdevelopment 283

364 Index

Company structure 324Comparative analysis 253Competition effect 28Competitiveness 17, 76, 119, 253Complement 318Complementarity 52, 331

analysis 326–327factor 135

Complexity 329–331Congestion 63, 78, 103, 104–105, 123,

251 see Toll roadscost 95, 97, 201–202

Connectivity 44Construction activities 265Construction employment 300, 308Consumer surplus 51–52, 165–168

174, 175Consumers’ behaviour 164Consumption multiplier 301Corridor 259

city 121Cost benefit analysis 12, 81, 105, 326 see

also Benefit-cost analysisCost effectiveness analysis 193Cost elasticity 142 see ElasticityCost function 149, 154, 243

model 141–143, 151Cost overrun 74Cost savings approach 181Cost of travel 46Creative learning capacity 324–325Crowding out 24, 106, 125, 31, 137 Decentralisation 101–102Decision-makers 193Decoupling 327–329Delay option 191Demand elasticities 188, 189, 203 see

ElasticityDemand for labour 223, 245Demand function 220Dematerialization 328Democracies 334Demographic change 58, 108–112Denmark 187Density 120, 121Developing countries 21–24Development argument 7, 21

effects 13, 313plans 242pressures 239, 242

Differential employment shift 244

Direct employment 289, 291, 297–299,307 see Employment

Direct impacts 297–299Discount rate 183–186Discount shopping stores 243 see

RetailingDiscrete choice analysis 176, 179Disutility of travelling 178Double counting 166, 174, 193, 196

see Cost benefit analysisDurability 328Dynamic model 136, 141

process 39systems 324–325

Econometric analyses 11

model 48, 315Economic base model 265, 269, 273, 315

base theory 25conditions 315, 333development 35evaluation 127growth 35, 47–50, 132integration 75, 76–78linkages 304–307overhead capital 14, 60performance 105potential 245rate of return 188

Economies of scale 46, 51, 76, 77, 114,289, 307, 313, 332

Edge city 121, 280Efficient management 136

urban form 172Elasticity 180, 219

of costs 143of output 138, 148, 153, 156of substitution 203

Elderly 109, 111–112Emissions 123Employment 215, 230–233, 297, 303,

304, 310, 311impact model 291location 216multiplier 288–289taxes 71

Energy use 115, 121Engineering needs assessment 157England 288Environmental 17, 116–117

costs 116, 119effects 115–120, 170impacts 58regulation 118

Index 365

Equilibrium analysis 38–39conditions 46employment 228

Equity 58, 118, 158effects 112–115

Euralille 282Europe 84, 111, 124European Commission 192European Union 5, 25, 102–103, 108,

110, 194, 196Evaluation see Cost-benefit analysis,

Benefit-cost analysisEx ante analysis 257, 258Ex post analysis 237, 257Expectations of investors 29, 322Expenditure leakage 316Externalities 43, 120, 133, 137, 188, 196,

212 Financial services 84, 100Firm related growth measures 214Firm’s location 16, 223Flexibility 99Flexible specialisation 92, 328Floor space 247Forward linkages 212France 64Free fare zone 277Free rider 67, 75, 133Freight distribution 14, 97, 329,Fremont 285French A71 motorway 239, 249–251French railways 71 see Railways, TGVFunding 137 Garin-Lowry model 268Gatwick 293GDP 4, 5, 18, 23, 35, 48, 50, 76, 80, 115,

326General equilibrium model 136General purpose technologies 158Generalized Leontief models 143 see

also Leonief cost function modelsGeneralized travel cost 171, 175, see

also travel, Transport costsGeographical scales 40Germany 146, 184–185Global cities 84, 100–102Global production 58Globalization 92, 101

effects 104Glocal 328Government failure 134

Government provision of infrastructure133–134

Gravity functions 266Great Britain 84–87, 110Greece 185Green belt 240–249, 242, 247, 303Greenfield sites 239Gross state product 148, 151 Headquarters functions 301Heathrow 292–301, 310 see also

London, Terminal 5Heavy Goods Vehicles 244 see Freight

distributionHigh tech service industries 280High technology 242Higher education 279High-speed rail 61, 63, 80, 98, 101,

102, 103, 257, 278–282, 285Highway expenditures 146Household attributes 178

sector 215–217, 220–221, 223, 226size 109

Hub 301, 302airports 287, 297 See Heathrow,

Manchesterlocations 307

Hypermarkets 243 see RetailingHypothecation 71 Image 29, 53, 282, 322Impact analysis 264–268

matrix 193statement 192, 193

Increasing returns 113, 114Incremental revenue 181Indirect employment 291, 296, 297–

299, 302 see also Employmentimpacts 299jobs 307

Induced demand 124Induced employment 291, 296, 297–299

impacts 300traffic 18, 49

Industrial reorganisation benefits 168Inflexibility 186Information exchange industries 279Information technology 83 see also

TelecommunicationInfrastructure sector 223Innovation 90, 119Input prices 144Input-output analysis 11, 48, 299, 315

366 Index

Interchanges 99, 323Intermediate input 135Intermodiality 45, 98Internal rate of return 183Internal trade theory 15Investment 42–44

conditions 318effect 273–275financing 66–75multiplier 164, 268–269, 302 see also

Multiplier analysistype 333

Inward investment 301, 303, 308Irreversibility costs 186Israel 153ISTEA 192 Japan 103, 109

Shinkansen see High-speed railJob uncertainty and flexibility 90Joint projects 70, 73–75Journey length 112Jubilee Line Extension 25, see also

London Labour 216

force 259force participation 230market effects 24, 169, 196, 225, 228,

321productivity 140, 145, 163, 211, 215

225, 229, 289, 294–296 see alsoProductivity

Lagged relationships 145Lakeside development 249Land market 250

owners 251rent 51–52use model 273–275values 9, 17, 46, 247, 251

Latest demand 124, 166, 254Leakage 305Leisure 84, 112

time 220, 221, 224, 229Leontief cost function model 146 see

also Generalized Leontief modelsLevel of attraction 273Life cycle 88, 112Lifestyles 120Lille 53 see EuralilleLinkage analysis 304Liverpool 6Local economic conditions 27

economic growth 211, 212–214,258

economy 128, 282impacts 320inquiry 247labour market 297level 320

Location decisions 37theory 7–11, 25, 46, 51, 94

Logistics systems 6Log-linear production function 151London 63, 99–100, 240–249, 257,

292–301Long-distance commuting 102Long-term employment 265Lumpy investment 177, 190Luton 294Lyon 280 M25 London orbital motorway 63,

239, 240–249accessibility 243–246

Macroeconomic approaches 16–18, 25,105

Maintenance 78Manchester 310

airport 301–309 see also HubMarginal social costs 170, 201–202Market equilibrium 203

failure 133potential measures 243–244related growth measures 215

Marshallian districts 93Measurement of benefits 165–168Method of financing 38Mode choice approach 181Model see Accessibility model, Cost

function model, Dynamic model,Econometric model, Economicbase model, Garin-Lowry model,Generalized equilibrium model,Generalized Leontief model, Landuse model, Multinomial Logitmodel, Production function basedmodel, Shopping probabilitymodel, Time allocation model,Urban and regional model

Monopoly 66Motorization effect 110–112Motorway 12, 63 see Amsterdam

Orbital, French A71, M25 accessiblelocations 242

Multicriteria analysis 183, 192, 193

Index 367

Multinational businesses 303corporations 92, 93firms 301

Multinomial logit model 175–176Multiple regression analysis 244 see also

Regression analysisMultiplier analysis 296, 297, 299, 300,

305, 310, 311, 315effect 10, 15, 39, 48, 105, 115,

173–174, 265, 308 Necessary condition 212, 318–320Neo-Schumpeterarian theory 90Net present value 163, 183Netherlands 185, 191Network 324

accessibility 174–177capacity 159development 214economics 169, 177, 333performance 40, 44–45, 102–107see also Accessibility

New economic geography 114growth economics 113settlements 121

New York 99, 212, 230, 258New York State 276Niagara Frontier Transportation Study

276Noise nuisance 293Non-transportation trends 263Non-work travel 87Normative economics 128 Oakland 283Offices 243, 252Opportunity costs 178, 180, 190, 207, 221Optimal level of traffic 202

public capital 136traffic volume 201

Optionality value 331Osaka 278–280Out-migration 262Out elasticity 149, 150, 153, 155 see also

ElasticityOutsourcing 328Overall impact 273–275Overspill principle 72 Pareto optimum solution 202Paris 88, 280Parking policies 276–277Partial equilibrium 217

models 147

Participation 118by women 262

Partnership 70, 75, 78–79, 83funding 62

Pecuniary externalities 167Perceived accessibility 271 see also

AccessibilityPlanning agencies 251

applications 246, 247inquiries 325

Pleasanton 283Policy design 332–335Policymaking 333, 334Political and institutional conditions

318Pollution 116–117, 303Population and employment 260–262Portland 277Positive externalities 211

feedback 213Post-industrial society 90Price inelastic 204Private capital investment 147 see also

Investmentcapital productivity 20goods 184output 153sector investments 25

Privatization 65, 68–70Production 127, 218

costs 7function 145, 218, 219, 226function based model 20, 138–14,

149, 154, 156, 234network 328sector 215–216, 217–220, 223

Productivity 17, 18–20, 60, 68, 76, 77,80, 106, 115, 132, 135, 147, 151–152, 156, 298, 304, 307–309,311, 316

change 142gains 212

Profit function 234model 143, 149

Profit maximization 218Project appraisal 326–327 see Cost

benefit analysis, Benefit costanalysis

evaluation process 161Property market analysis 246

rights 57, 68Proximity 95, 245, 331–332

368 Index

Public acceptability 71awareness 120capital 132, 142, 144, 147, 148, 151,

154, 157 see also Investmentcapital stock 143–144expenditure 4, 57, 63goods 184infrastructure investment 20, 135,

140141–143, 156, 330 see also Investmentinquiry 70perceptions 60sector borrowing requirement 68, 69utilities 59

Railways 8

accessible 283investment appraisal 194projects 196see also High speed rail

Random utility theory 179Rates of return 22Real estate transactions 249Redistributed traffic 254Redistribution of employment 313Regional airport 302Regional development 5, 11, 14, 53, 61,

244economic impact 305income 310shopping centres 247strategy 242trends 259, 260–264

Regression analysis 230, 253 see alsoMultiple regression analysis

Relocation costs 171decisions 171

Rent levels 252 see also Land valuesResidential location 88, 217, 222Resources 116–117Retail activity 266

centres 262 see also Regionalshopping centres

development 246–249floor space 267malls 277patterns 262–263warehouses 247

Returns to public investment 141Richmond 284Risk 66, 69, 71, 72, 73, 74, 186–191

aversion 189free bonds 183

Road accessibility 283, 285 see alsoAccessibility

pricing 67, 71, 203traffic 3transport informatics 97–97

Rural development agency 251 SAFER 251San Francisco 257

BART 283–285Saturation level 111Scale economies 219 see also Economies

of scaleService employment 262, 275, 305Shadow price 147, 156, 188, 205, 207,

208toll 65, 67, 73, 79value parameters 141–143

Shamrock 96organization 89

Shinkansen stations 278–280 see alsoJapan

Shopping analysis 271–273probability model 266trips 265see also Retail

Sideways linkages 212Simulation 225–230Simultaneous equation model 153Small time savings 180, 181Social congruity 192

equilibrium 197, 202groups 85net value 188overhead capital 14, 23, 60, 67–69rate of return 48welfare 133, 165, 234

Société d’ Economie Mixte 282South Bronx 230–233Spain 154Spatial agglomeration 90

changes 171–172efficiency 171equilibrium 149, 172organization 226proximity 213redistribution 172relocation 196

Stansted 294Stated preference data 179

preference techniques 177Static equilibrium 216Steady state 139

Index 369

Structure plans 242Suboptimal 136Suburban areas 262

malls 272 see Retail, Shoppingzones 275

Suburbanization 259, 276Sunk costs 189Superstore 243 See RetailSustainability 115–120

development 58, 117–118growth 5

Sweden 154 Taste coefficient 221Technological change 28, 57, 96–100,

188, 322innovation 83, 116related growth measured 214revolution 5, 25, 100solutions 120substitution 122

Technopole development 93 see alsoEuralille

Teleactivities 98, 122Telecommunication 45 see also

Communicationsnetworks 98, 100systems 101

Telecommuting 89, 122, 135Terminals 5 297–302 see London,

Heathrow, Air transportTerminals 323TGV 280 see also High speed railThresholds 331Time accessibility 269–271 see also

Accessibilityallocation model 205based analyses 321preference 183related changes 139scale 40, 183–186spent at work 205spent in leisure 205spent in travel 205

Tokyo 99, 278–280, 285Toll roads 63, 65, 71, 97Toronto 179Total accessibility 269–271 see also

Accessibilitycost analysis 195employment 307production costs 27

Tourism 301, 308

Trade theory 51Traffic calming 193

diversion 204growth 104

Trans-European transport network 125see High speed rail

Transfer payments 48Transit investment 146

oriented development 122Translogarithmic cost function 143, 152Transport behaviour 102–107

bonds 71capacity 201capital accumulation 132costs 7, 49, 50, 61, 113, 229, 321 seealso Generalized travel costscost advantages 244efficiency 79infrastructure 35, 215–216intensity 50, 80, 104, 327sector 228trends 263

Travel behaviour 37congestion 216costs 214costs savings 162demand 103impacts 251time 171, 176, 207, 201, 211, 224, 227,

243, 254, 263, 265, 267, 269, 273 seealso Time accessibility

time budgets 122time savings 44, 163

Treatment of time 322Trip chaining 178Trip distribution models 175Trip lengths 85–87, 103 Ubiquitous accessibility 285 see also

AccessibilityUK 104–105Uncertainty 186–191Urban and regional modelling 11–16 see

also Models, Land use modelscongestion 52form 120–123hierarchy 9land use model 267regeneration 3

Urbanization economies 171USA 84–87, 97, 103, 109, 110, 111, 124,

125, 145–146rail 187

370 Index

Use 324Utility function 205, 221, 226

maximization 178 Value added 163, 326

capture 29of time 178–182, 205–208 see also

Time, Travel timeVolume capacity function 222, 227 Wage rate 220, 230Walnut Creek 283

Warehousing 242Weighting systems 193Welfare 186Willingness to pay 165 see also Tolls,

Time costs, CongestionWomen 110, 111Work 265

journeys 87patterns 84–89time 216, 228

Yokohama 280

[david banister, joseph berechman] transport inves(bookfi.org)

Documents