behavioural approaches to freight transport modal choice

25
This article was downloaded by: [Moskow State Univ Bibliote] On: 03 December 2013, At: 15:17 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Transport Reviews: A Transnational Transdisciplinary Journal Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ttrv20 Behavioural approaches to freight transport modal choice R. Gray a a Department of Shipping and Transport , Plymouth Polytechnic , England Published online: 13 Mar 2007. To cite this article: R. Gray (1982) Behavioural approaches to freight transport modal choice, Transport Reviews: A Transnational Transdisciplinary Journal, 2:2, 161-184 To link to this article: http://dx.doi.org/10.1080/01441648208716491 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/ page/terms-and-conditions

Upload: r

Post on 19-Dec-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Behavioural approaches to freight transport modal choice

This article was downloaded by: [Moskow State Univ Bibliote]On: 03 December 2013, At: 15:17Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Transport Reviews: A TransnationalTransdisciplinary JournalPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/ttrv20

Behavioural approaches to freighttransport modal choiceR. Gray aa Department of Shipping and Transport , PlymouthPolytechnic , EnglandPublished online: 13 Mar 2007.

To cite this article: R. Gray (1982) Behavioural approaches to freight transport modal choice,Transport Reviews: A Transnational Transdisciplinary Journal, 2:2, 161-184

To link to this article: http://dx.doi.org/10.1080/01441648208716491

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information(the “Content”) contained in the publications on our platform. However, Taylor& Francis, our agents, and our licensors make no representations or warrantieswhatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions andviews of the authors, and are not the views of or endorsed by Taylor & Francis. Theaccuracy of the Content should not be relied upon and should be independentlyverified with primary sources of information. Taylor and Francis shall not be liablefor any losses, actions, claims, proceedings, demands, costs, expenses, damages,and other liabilities whatsoever or howsoever caused arising directly or indirectly inconnection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden.Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Page 2: Behavioural approaches to freight transport modal choice

TRANSPORT REVIEWS, 1982, VOL. 2, No. 2, 161-184

Behavioural approaches to freight transportmodal choice

By R. GRAY

Department of Shipping and Transport, Plymouth Polytechnic, England

ABSTRACTThree main types of institution are interested in the determinants of freight

transport modal choice—the government, the carrier and the shipper.Researchers have emphasized the importance of understanding the decision-making procedures of actors involved in freight modal choice (e.g. Meyburg1979, Roberts 1971). This requires the development of behavioural models andthe paper is a review of the state of the art in such modelling.

Most models can be included under two main prevailing orthodoxies. First,there are models which assume that modal choice is based on some form of short-term cost optimization by the shipper. This approach may be called 'economicpositivism' since it assumes that modal choice is determined by economic or costvariables. Second, there are models which assume that modal choice is based onrelationships between physical aspects of the transport system (e.g. speed,frequency) and physical aspects of the product (e.g. perishability, value-weightratio): This approach may be called 'technological positivism' since it assumesthat modal choice is determined by technological variables. In addition, there is athird broad approach to the study of freight modal choice which bases itsassumptions on the perceptions of members of shipper-organizations, par-ticularly transport managers. This approach may be called^the 'perceptualapproach'.

The review examines empirical studies undertaken in Australia, Canada, theUnited Kingdom and the U.S.A. Particular areas for discussion are theassumptions of the disciplines from which the models are drawn and the relevantunit of analysis (e.g. the firm, the individual person, the consignment).

§1. INTRODUCTION

The demand for freight transport has received much less research attention thanthe demand for passenger transport. Roberts (1977) suggests two possible reasonsfor this. First, freight is much more complex than passenger transport in terms of'choice-influencing attributes' and in terms of the size of the individual unit ofmovement and, secondly, published data on freight transport is generally in-adequate. He also asserts that most freight transport demand models have not beenparticularly successful in their ability to forecast, and suggests that there is a need forapproaches at the disaggregate level which attempt to model the behaviour of theindividual decision-making unit.

Individual choice models were developed for passenger travel demand and havesubsequently been used in freight demand research. One problem which emerges isthe difficulty of determining the appropriate unit of analysis for freight modal choicemodels. Whereas the individual passenger is easily identified as the decision-makerin passenger travel models, the decision-making unit for freight modal choice isuncertain. The decision-making unit is often assumed to be the firm of neo-classicaleconomics or, with increasing frequency in recent years, the consignment. Obviously

0144 1647/82/0202 0161 S0400 Ç 1982 Taylor & Francis Ltd

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 3: Behavioural approaches to freight transport modal choice

162 R. Gray

neither of these units is a human actor and is therefore incapable of making adecision. In such cases it is assumed either than there is unitary decision-making onthe part of the firm, or that there is an identifiable discrete decision represented bythe consignment.

Modal choice is only one stage in the activity of shippers who must also makechoices regarding origins and destinations and shipment sizes. However, it is at thestage of modal choice that the interests of the three major types of institutionassociated with freight transport (the government, the carrier and the shipper)mainly coincide. Furthermore, it is the level of decision-making in freight transportdemand which is most frequently studied by researchers.

§2. T H E CLASSIFICATION OF FREIGHT MODAL CHOICE MODELS

The researcher should always have a clear idea of the objectives of his analysis.An analysis which provides a modal split at the regional level may not necessarilyprovide a meaningful interpretation of the modal choice at the level of the firm.Heggie (1978) suggests that behavioural models of passenger travel choice shouldfulfil the three main roles of helping people to understand and explain behaviour, ofassisting with policy formulation (which requires that models are intelligible to thenon-specialist), and of providing accurate predictions for design and evaluation. Thesame criteria are also appropriate for freight modal choice. A number of writers haveconsidered the implications of the design of freight modal choice models, and thissection brings together some of their main conclusions.

Bayliss (1971, 1972) considers that modal choice models can be classifiedaccording to three broad approaches. These are:

(1) Market research approach(2) Aggregate flow approach(3) Consignment approach

Bayliss is critical of many market research studies, not only for being too general(by not referring to specific commodities or activities) but also for not being genuinemodal split analyses, since shippers are not asked to state their preference for onemode compared with another, but merely to rank attributes of modes in order ofpreference. It should be stated that Bayliss adopts a more limited interpretation ofmarket research than might normally be accepted.

The aggregate flow approach establishes relationships between nationally orregionally aggregated data ranging from fully aggregated analyses (e.g. the relation-ship between gross national product and total ton-mileage) to models disaggregatedto specific commodities by mode on particular network links. The larger scalemodels generally take no account of the behaviour of shippers in the sense that theydo not use a unit of analysis which can be directly attributable to individual shippers.Indeed, they need not be concerned with the underlying reasons for using aparticular mode since they aim merely to establish quantifiable traffic flows.

The final approach, the consignment approach, was applied by Bayliss andEdwards (1970) in a study for the Ministry of Transport in the U.K., and uses as itsbasis the individual consignments shipped by firms. Bayliss claims that theadvantage of this approach is that it observes actual behaviour in terms of the flow ofconsignments, rather than being dependent on general attitudes (as in the case ofmarket research studies) or ignoring behaviour (as in the case of aggregate flow studies).

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 4: Behavioural approaches to freight transport modal choice

Freight transport modal choice 163

The U.S. Transportation Research Board (1977) considers that all freight modalchoice models attempt in some way to replicate the decision-making of shippersalthough without noticeable success. It notes a lack of consensus among modellersand suggests that the main unresolved questions include those about the degree ofaggregation, the relevant variables to be used, the level of approach (e.g. theindividual firm or all firms within a region), and the degree of homogeneity ofcommodities.

The classification of approaches to freight modal choice proposed by theTransportation Research Board is an amalgamation of theoretical approaches andstatistical techniques. In the order given by the Board, the approaches are:

(1) Macro-economic approach(2) Inventory-theoretic approach(3) Regression analysis(4) Abstract mode (using regression analysis)(5) Linear programming(6) Discriminant analysis(7) Logit analysis(8) Probit analysis

The reader of this review is assumed to be familiar with the objectives of regressionanalysis and linear programming. Discriminant analysis, logit analysis and probitanalysis are statistical techniques which have been used extensively in disaggregateand behavioural passenger travel demand modelling and to a limited extent in freightmodal choice modelling (see Hartwig and Linton (1974) for a description of theirapplication in freight modelling, and Stopher and Meyberg (1976) for passengertravel applications). Discriminant analysis is a technique which identifies two ormore modes in a population by means of a discriminating function whereas logit andprobit analyses relate the probability that a given mode will be chosen by a shipper tothe independent variables hypothesized to influence that choice. Probit models arebased on the assumption that members of a population have a response patternapproximating to the normal or Gaussian distribution, whereas logit models makeno assumption about statistical distributions. It is often claimed that the use of suchtechniques is theoretically justified since they are associated with utility theory.

Utility is associated in economic demand theory with a satisfaction derived fromconsuming some quantity of a good. Owing to the difficulty of measuring utility inabsolute terms some economists have redefined utility to refer to an ordinal attribute;that is, to a consumer's preference for a particular combination of goods relative toanother combination. Economic rationality assumes utility maximization, and theconcept has been adopted by decision theorists. Luce and Raiffa (1957) have outlinedutility axioms which include the assumptions that decision-makers either prefer onealternative to another or are indifferent between them. If the decision-makers areindifferent between choices, the choices can be substituted for each other. Wherethere is more than one potential choice, each may be decomposed into separatechoices with particular probabilities, and the decision-maker will choose thealternative with the greater probability.

The somewhat misleadingly titled macro-economic approach refers to anapproach adopted by the Mathematica (1967) researchers in the North-EastCorridor (of the U.S.A.) Transportation Project and is based on the use of

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 5: Behavioural approaches to freight transport modal choice

164 R. Gray

simultaneous equations applied in a sequence of increasing disaggregation. Theinventory-theoretic and abstract mode approaches are considered in §3.1.

In another review, Roberts (1977) considers that there are seven desirablefeatures required for freight demand models. Roberts is concerned with demandmodels rather than merely with modal choice models since he considers that theshipper must make the three simultaneous choices of where to buy or sell the goodsbeing transported, how much to ship, and by what mode (or carrier) to ship. Thus,freight transport modal choice should not be considered in isolation. The validity ofassuming that a single unitary shipper makes such a wide range of simultaneouschoices in the modern complex industrial organization is questioned at several pointsin this review. Roberts' seven criteria are:

(1) The model should be at the disaggregate level—the four main types ofaggregation are by time, space, commodities and modes.

(2) The behaviour of individual decision-making units should be modelled. Aunit is equated with the firm.

(3) The shipment or consignment should be the basic unit of analysis since it cantake into account the commodity, the shipper establishment, the shipmentsize and the frequency of shipment.

(4) The model should focus on the destination rather than on the senderorganization since (in the U.S.A.) receivers or consignees generally pay thefreight and decisions about transport are more clearly identified withreceivers.

(5) Model parameters should be determined empirically, where possible usingshipping documentation. Roberts does not suggest that the variablesthemselves should be determined empirically, and it is taken for granted thatsuch variables as transit time, loss or damage in transit etc are significant inmodal choice. It is the value of time, etc. which is to be determinedempirically.

(6) The model should use generalized attributes, since by using such measuresas value-weight ratios, volume-weight ratios, and shelf-life it is possible toapply the model to most classes of commodity.

(7) The aggregation scheme should be based on forecastable data sources (e.g.Census data). Roberts does not consider that the disaggregate behaviouralmodel's form is dependent on the aggregation scheme and he assumes thatthe scheme is more likely to be determined by policy questions not associatedwith the disaggregate behavioural model.

2.1. Classification adopted in this reviewBehavioural approaches to freight modal choice can be placed under three broad

headings which allow consideration of the different theoretical assumptionsassociated with them. Many models adopt positivistic explanations which meansthat they assume knowledge to consist in the description of the relationships betweenphenomena. Such explanations may be put under two main headings. First, thateconomic value related to the firm (e.g. marginal revenue) determines the use oftransport. This approach may be called economic positivism. Second, that physicalattributes or technology determine the use of transport at the disaggregate level.This approach may be called technological positivism. A third approach assumes thatthe shipper's choice of mode may be based on known variables, the relationships

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 6: Behavioural approaches to freight transport modal choice

Freight transport modal choice 165

between which have been subjectively interpreted by the shipper and without anyassumptions on the part of the researcher. This approach may be called theperceptual approach.

Economic positivism is based on the neo-classical economic theory of the firm: thatis, the part of microeconomics concerned with the firm in terms of input and outputrelated to price and profit. In its traditional form the theory assumes that a firmmaximizes profits with full information and complete certainty and that there are noproblems of an organizational character. In this form the theory is unrealistic since itdoes not take into account many of the characteristics of modern firms, for example,the separation of ownership from management, the organizational complexity ofmany modern, large business organizations, and the imperfect knowledge of theenvironment within which firms operate (see, for example, Galbraith 1972). Theessence of the economic positivist approach is that the decision-making unit,assumed to be the firm, attempts to maximize short-term revenue or minimize short-term costs in a trading situation in which transport plays a central role.

In contrast, the essence of the technological positivist approach is that modalchoice is explained in terms of relationships between technological phenomena, orbetween the physical aspects of the commodity (weight, volume etc.) and of thetransport system (speed, frequency etc.). Within the constraints of technologicalpositivism, variables are selected for empirical reasons, mainly as a result of previousapplied research. The review of modal choice modelling by the TransportationResearch Board (1977) is to a large extent a review of this type of approach.According to the Transportation Research Board the most frequently usedindependent variables are:

(1) Weight of shipment(2) Distance shipped(3) Value per ton(4) Commodity type(5) Annual tonnage of the commodity shipped(6) Freight rate charged(7) Transit time(8) Measure of reliability

Most of these variables are concerned with the nature of the product rather thanwith the producing firm considered as a human activity system. It is in contrast to theeconomic positivist approach which is concerned with the profit or cost profile of thefirm. None of the above variables may be said to reveal such a profile, either singly orin combination. The concept of the association of modal choice primarily with thenature of the commodity is central to the technological positivist approach. If ananalogy is drawn with an industrial production process, the commodity is processed(i.e. is transferred spatially) by machinery (i.e. the transport mode) which hasperformance specifications (e.g. transit time). In the technological positivistapproach, the freight rate is not seen as data contributing to the analysis of the cost tothe shipper as in the economic positivist approach, but simply as one more variable todescribe a technological process. Thus, it has been found in some studies (e.g.Bayliss and Edwards 1970) that the freight rate is not a significant determinant ofmodal choice.

In the technological positivist approach the selection of the transport mode tendsto be the dependent variable to a number of independent variables related to the

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 7: Behavioural approaches to freight transport modal choice

166 R. Gray

transport technology. The perceptual approach differs from the technologicalpositivist approach by assuming that the independent variables are based primarilyon the transport user's subjective interpretation of the situation rather than onphysical attributes objectively measureable by the researcher. Thus the perceptionsof the shipper assume importance. The approach assumes that shippers are able toadopt reasonably lasting impressions of transport services. The elements ofselectivity and stability in perception often result in the establishment of stereotypes,that is, of biased generalizations towards particular transport modes.

The essential difference between the perceptual approach and the other twoapproaches is that the unit of analysis is a person, although often equated with thefirm. In the economic positivist approach the unit of analysis is an economic value(e.g. marginal revenue) and in the technological positivist approach a unit ofproduction (e.g. a weight of goods or a consignment). Perceptual studies do notappear to assume a divergence between interests of the individual perceiver andthose of his employing firm (or others in the firm), although such an assumption maybe implicit in using a particular subsection of the organizational membership as theunit of analysis (usually the transport manager).

§3. EXAMPLES OF THE THREE APPROACHES

This section includes descriptions of freight modal choice studies which indifferent ways claim to take into account the behaviour of shippers. They areclassified according to the three approaches developed in the preceding section—theeconomic positivist, technological positivist and perceptual approaches. Obviouslythis classification is imposed on the studies by the author and some studies combineelements of more than one of the three approaches. For example, it will be seen thatconsiderations of shippers' perceptions pervade all three approaches. However, it isalways possible to identify a particular emphasis in each study and this willdetermine under which category it is placed. Naturally, a review cannot do justice tothe complexities of most of the studies and the discussion is restricted to assumptionsabout behaviour.

3.1. Examples of the economic positivist approachAn example of the neo-classical economic approach to freight modal choice is

that of Allen (1977) who describes individual shipper choice in terms of the 'rationalfirm'. According to Allen, modal (and market) choice is determined by maximizingmarginal revenue by selecting the most suitable combination of mode and marketaccording to the following equation

PkJ= [Pk - TkJ(Q, akJ) - Q TQk\Q, o^)] exp ( - iakj) (1)

where

P marginal revenueP market price for firm's product at kT\Q,OL) transport charge per unit of product (a function of quantity shipped

and time in transit)Q manufacturer's cost functiona days in transiti interest rate per day (opportunity cost of funds)exp — ia. since net revenue is received only after a. days, it is multiplied by

exp (— ice) to determine its present valuek marketj mode

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 8: Behavioural approaches to freight transport modal choice

Freight transport modal choice 167

The abstract mode approach developed by Baumol and others in theMathematica (1967) team of researchers and also described by Baumol and Vinod(1970) is in the microeconomic tradition and is an application of a more generaleconomic consumer demand theory described by Lancaster (1966). The discussionin this review is restricted to the abstract mode approach where it attempts todescribe the behaviour of the individual shipper. Examples of the approach havebeen applied to regional flows, a subject beyond the considerations of a review ofbehavioural approaches.

According to Lancaster, traditional economic consumer demand theory hasdifficulty in accommodating new goods. Each consumer has an individual preferencemap so that, when offered a choice between two sets of goods, he will either preferone set to the other, or will be indifferent between the two sets. If a new good isintroduced to a consumer it is necessary to discard that consumer's preference mapand to produce an entirely new one, thereby losing all the information concerningpreference for the original set of goods. Lancaster considers that consumers areinterested in combinations of characteristics or attributes of goods rather than merelyin the goods in themselves. As an illustration, a new car model is not the same as anold one and, therefore, should be treated as a new good in traditional economictheory. However, according to Lancaster a new car model possesses largely the samecharacteristics as an older model and these are merely combined in a slightlydifferent way.

Baumol and others have adopted the same approach in the concept of the abstractmode and of the abstract commodity transported by it. An abstract mode describes acarrier not in terms of rail, air, etc., but in terms of a vector of values. The values arethe attributes which the mode offers the shipper. As a result the mode should bedefined only in terms of attributes known to the shipper, so that, for example, thetype of fuel is unlikely to be an attribute whereas the transit time is a likely attribute,since only the latter is probably of interest to the shipper. The abstract modeapproach presented by Baumol is within the microeconomic theory of the firm sinceit is based on cost minimization (each attribute is given a cost) and on equatingmarginal cost with marginal revenue as an equilibrium condition.

A number of specifications of the abstract mode approach have been developed, asimple version of which is given in Baumol and Vinod (1970). This version is derivedfrom 'inventory theory' which envisages freight in transit as a moving inventory.The formulation is

C = rT-utT+a/s + wsT/2 (2)

where

C expected total annual variable cost of handlingT total amount transported per yearr shipping cost per unit of commodity (e.g. tons) (including freight rate,

insurance etc.)t average time required to complete a shipment in years5 average time between shipments in yearsu carrying cost in transit per unit per year (interest plus deterioration

plus pilferage rate)a cost of ordering and processing per shipmentw warehouse carrying cost per year

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 9: Behavioural approaches to freight transport modal choice

168 R. Gray

The equation tells us that the expected total annual variable cost of handling a givenquantity of goods is equal to the sum of a number of costs directly associated with thestorage and movement of goods. The abstract mode is denned in terms of the threecharacterizing elements of shipping cost, carrying cost and transit time. Theordering and processing cost, warehouse carrying cost and the total amounttransport are exogenously-given parameters. This simple model has been extendedby the authors to take into account such other factors as uncertainty in demandforecasts and delivery time and their effect on the level of safety (buffer) stock.

In recent years a prescriptive approach to the activity of industrial companies hasbeen developed which is often called 'business logistics'. Business logistics has beendenned (Bowersox 1974) as

the process of managing all activities required to strategically move rawmaterials, parts and finished inventory from vendors, between enterprisefacilities, and to customers.

The essence of the logistic approach is (Van Buijtenen 1976)

an integrated approach where cost savings are identified by consideringthe total costs of the system. This approach already implies the need foroverall management since the decisions about the different elements ofthe logistics system (transportation, inventories, facilities, unitisation,communications) are traditionally made within different functions ordepartments.

It is evident that this practical and normative approach has underlying concepts incommon with the economic theory of the firm. In particular, there is the assumption(in the case of economic theory) or the management prescription (in the case of thelogistics approach) of unitary control of the storage and movement of goods withinthe shipper company aimed at maximizing or optimizing profit through centralizedcontrol of costs.

Some freight modal choice models have adopted a logistics approach which takesinto account a wide range of specified costs incurred by the shipper when makingtransport modal choice. A leading advocate of this approach is Roberts (1977) whodescribes a disaggregate choice model as a multinomial logit function. The model isintended to give the probability of a receiver (consignee) obtaining goods from agiven origin in a given shipment size by a given mode. As stated in §2, Robertsconsiders that models should focus on receivers as shippers and on the total demandactivity rather than merely on modal choice. He sees the role of the freight analyst asspecifying and estimating the utility function U(T,C,M,R), where

T transport attributes (i.e. wait time, transit time, reliability, loss and damage,freight rate, special charges)

C commodity attributes (i.e. value, density, shelf life)M market attributes (i.e. relative price, ownership)R receiver attributes (annual use rate in lbs/year, mixed order, seasonal

purchase, shipment size in weight, reliability of use rate in days, %guarantee of availability)

The utility function of the receiver is based on the costs of the logistics process whichthe receiver should be controlling, and which therefore places the model within theeconomic positivist classification.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 10: Behavioural approaches to freight transport modal choice

Freight transport modal choice 169

Levin (1978) has also developed a multinomial logit model to investigate theextent of 'misallocation' (unnecessary social cost incurred by not using the optimalmode) of freight traffic between the competing modes of truck (road), boxcar (rail)and piggyback (road trailers transported on rail flat cars for most of the journey) inthe light of recent moves in the U.S.A. to deregulate freight rates, in particular theRailroad Revitalization and Regulatory Reform Act of 1976.

The model is based on the assumption of utility maximization on the part of theshipper, since Levin claims that

it seems natural to postulate that a shipper considers the utility of eachof the alternatives and selects that mode for which utility is at amaximum.

After suitable transformation the model is presented in the form

log — = fly+ X bijkXijk i, 7 = 1 , . . . , J. (3)

where

P{ market share of ?th modePj market share of 7th modeXijk difference between modes i and j in the value of the &th attribute

(i.e. differences in price, speed, reliability etc.)

The resulting regression model to be estimated is

P2_

°gPi °2i x x

(4)

where

Subscripts 1, 2, 3 represent truck, boxcar and piggyback respectivelyRt rate charged by transport mode i (in cents per hundredweight)V commodity value (in cents per hundredweight)T mean transit time for mode i (in years)<T; standard deviation of transit time for mode i (in years)

Levin considers that transit time is associated with the inventory costs of the goods intransit.

Although the model is based on the utility of the shipper, the lack of suitable data(in particular, waybills for road shipments) resulted in the use of aggregated datafrom a variety of sources. Levin is critical of the reliability of some of his sources andhis paper is recommended for the insights it provides into the difficulties ofcollecting freight transport data. The general finding of his empirical work was that acomparison of the predicted allocation of traffic based on actual freight rates withthat based on rates equal to long-run marginal costs, produces a welfare loss throughmisallocation much smaller than generally thought. This means that the socialbenefits of rate deregulation would be smaller than anticipated.

Oum (1979) derives a general freight transport demand model consistent withneo-classical economic theory where freight transport is seen as an input to the

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 11: Behavioural approaches to freight transport modal choice

170 R. Gray

shipper's production and distribution activities. Oum defines the cost function for ashipper's entire production and distribution activities as

/ L M

C( Y, I*, PT, Z, D) = min £ PfXf + £ £ PmlXml (5)x c , x T i = l l = l m = l

subject to a technology constraint

f(Xc,XT,Z,D)^Y (6)

where

Y shipper's total output for delivery to destination marketsP0 price of inputs other than transport (Pf is price of ith input)PT freight of M modes on L transport network links (Pmi is freight rate per ton-

mile of mode m on link Z)X0 production factors other than transport (e.g. labour, capital)XT quantities of M modes used on L links (Xml is ton-miles shipped by mode m

on link I)Z amount of AT quality attributes on M modes on a particular link (Zmnl is «th

quality attribute of mode m on link I)D distance (Dl is distance of link t)

The general model is also developed to take greater account of the quality of service.The objective of the general model is to present a

link-specific unit transport cost function for shippers of a particularcommodity group, as a function only of freight rates and qualityattributes of services of various modes and the distance of the link.

It is interesting to consider the relationship between the availability of data andthe assumptions about the behaviour of shippers, as stated by Oum, since it is typicalof the theoretical compromises found in some economic positivist approaches tofreight transport demand. The importance of considering the availability of databefore formulating a model is emphasized. In the case of Oum's study of Canadianinterregional freight flows for eight selected commodities the available dataaggregated by each link were:

(1) Yearly traffic volume in tons of each mode for each commodity group(2) Average freight rate (per ton) of each commodity(3) Average transit time and its variability for each mode(4) Distance of the link

For technical modelling reasons Oum makes the following assumptions aboutshippers:

The two [technical] assumptions together imply that a shipper changesthe amount of his product transported by each mode on each link inexactly the same proportion as the change in his total output if prices andquality attributes of all modes on all links remain unchanged. This maywell be an unrealistic assumption. However, in all the transport demandmodels estimated to date using data aggregated by each route, the authorshave imposed the same set of assumptions as those made in this study, butthey have not explicitly mentioned them. Whether or not these assump-

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 12: Behavioural approaches to freight transport modal choice

Freight transport modal choice 171

tions are valid is an empirical question which cannot be tested here owingto the lack of necessary data on individual shipper's production anddistribution activities held largely as confidential business information.

In the light of so much aggregated and averaged data and such unrealisticassumptions about shippers the model is unlikely to be accurate representation of thebehaviour of shippers.

The main empirical findings of Oum's study are that significant modalcompetition between rail and road transport exists only in medium and long-haulmarkets for higher value commodities and only in short-haul markets for lower-value commodities. Long-haul markets are rail dominated and short-haul marketsare road dominated.

3.2. Examples of the technological positivist approachThe approach adopted in a study by Stenger and Cunningham (1976) is to select

five variables which the researchers consider to be important in freight modal choice.The variables are

(1) Distance(2) Region of origin(3) Value of commodity(4) Susceptibility to loss and damage(5) Existence of substitute or competing commodities (defined in terms of how

late a delivery the customer will allow, based on the shipper's subjectiveassessment).

Stenger and Cunningham attempt by means of cluster analysis to minimize thenumber of clusters with shipments of more than one mode and to maximize thenumber of clusters with shipments of only one mode. Using combinations of the fivevariables it was found that the best set of clusters was derived from the combinationof the four variables: distance, region of origin, value of commodity, and suscept-ibility to loss and damage.

Within the clusters Stenger and Cunningham undertook 'constraint analysis'. Itwas found that for any given distance there is a maximum value per unit weightbeyond which rail movements are excluded, and that this value increases exponen-tially with distance. Thus, a value-distance constraint was applied to clusters, as wasan 'availability of rail sidings' constraint. Having identified the 'natural markets' foreither rail or road transport, either because they fall within a cluster which is notrail/road competitive or because they are not within the competitive constraints,Stenger and Cunningham attempt by means of discriminant analysis to identify theimportant characteristics which determine modal choice within the rail/roadcompetitive area. The set of variables which was found to have the most internalpredictive power is

(1) Shipment size(2) Type of receiver (retailer, manufacturer or wholesaler)(3) Production method (to stock or to order)(4) Substitutability of commodity (as defined above)

It was found that the only variable of any major significance in predicting modal• choice within the competitive area is shipment size.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 13: Behavioural approaches to freight transport modal choice

172 R. Gray

Bayliss and Edwards (1970), in a British study, use multiple regression in adiscriminant manner as a probability model where the regressand is treated as adummy variable, taking the value 1 if an event occurs and 0 if it does not. Thus,modal choice is expressed in the form

Y=f(XuX2...XJ (7)

where Y is equal to 1 if the consignment went by mode A, and 0 if it did not. Xx etc.are explanatory independent variables. The calculated value of Y is interpreted asthe conditional probability of the consignment using mode A.

The analysis uses more than twenty independent variables under three headings:

(1) Factors relating to consignmentslength of hauljourney timecharge (cost of freight)consignment weightregularity of shipmentcontainer usedspecial body requiredinter-establishment moveconsignment required more urgently than usualancillary service performed (in addition to transport)type of commoditydestination

(2) Factors relating to the shipper firmoriginsize of firmownership of rail sidingclass of haulage licence

(3) Subjective assessmentsknowledge of charge by alternative modecharge by alternative modespeed required to meet customer's requirementsspeed required to ensure high utilization of vehiclesspeed required to maintain low stock levelsready availability of vehiclesfreedom from lossfreedom from damage

All but four of the variables were entered as dummy variables (i.e. either as 1 = YESor 0 = NO), the exceptions being length of haul, journey time, weight and charge,which were entered as logarithmic values. The study attempts to explain thedistribution of traffic in terms of a series of binary modal choices: e.g. transport onown account (where the shipper owns his own vehicles) versus professional (hire andreward or public) transport; rail versus road transport on own account; rail versusprofessional road haulier. Length of haul was by far the most significant factor in thefirst two binary choices, 'explaining' 24% and 32% of the observed distribution oftraffic between modes. Consignment weight was the most significant factor in

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 14: Behavioural approaches to freight transport modal choice

Freight transport modal choice 173

explaining the distribution between rail and road haulier, although it explained only7%. No other factors were shown to be significant.

The justification for including the study by Bayliss and Edwards under theheading of technological positivism is that variables which directly measure theperformance of transport 'hardware' or associated processes are predominantly theassumed criteria for assessing modal choice. Bayliss and Edwards appear to classifythe subjective assessments separately because they are on the whole 'measured' bythe survey respondents (the shippers) rather than directly by the researchers. Thefactors relating to the firm are not economic factors of cost or profit but mainlytechnological factors, in particular transport-related facilities (e.g. ownership of arail siding).

Variables used in the technological positivist approach are not theoreticallyjustified but generally are included on an ad hoc basis. The main criteria for includingvariables would appear to be previous empirical results and the capacity to bemeasured.

3.3. Examples of the perceptual approachGilmour (1976) analyses the modal choice of distribution and transport managers

for freight movements between Melbourne and Sydney. On the basis that 'perceivedproduct features' may be more likely to determine purchasing behaviour than actualperformance characteristics, Gilmour examines the attitudes of shippers towardsmodal choice.

The research is in three main parts. First, respondents were asked to list factorswhich they were likely to consider when sending goods by a number of freight modesfrom Melbourne to Sydney. This part of the research was deliberately open-ended toavoid influencing the answers received from respondents. The second part of theresearch established similarity ratings for all possible pairs of the seven modes beingconsidered. The aggregated similarity ratings were subsequently analysed by non-metric multi-dimensional scaling. Finally, respondents were asked to give a degreeof importance rating to each of 41 factors gathered from the first stage and othersources. Cluster analysis was performed on the correlation coefficient matrix of theresults of this survey and the underlying dimensions were used to define theconfigurations (the so-called perceptual space) obtained from the non-metric multi-dimensional scaling analysis for each of five product groups of shippers. The resultssuggest that the direct cost of transport is not an important determinant of modalchoice, but that control over the shipment, availability of specialized equipment andreliability are important. Such characteristics favour road transport over rail, air orsea.

Cunningham and Kettlewood (1975) attempt to measure shippers' attitudes toBritish Rail and to competing road freight hauliers by means of two measures ofattitude, the dissatisfaction ratio and the relative attitude ratio. The dissatisfactionratio compares the weighted attributes scores (on a 5-point bipolar scale) for each ofroad and rail freight transport with the scores for an 'ideal mode' where maximumweighted scores are assumed for each attribute. The relative attitude ratio comparesthe weighted attribute scores of road freight and rail freight with each other. Publicroad haulage was found to be closer to the 'ideal' source of transport for all attributes,with the greatest difference between the modes shown for the attributes 'competitivein price', 'reliable deliveries' and 'faster transit times'.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 15: Behavioural approaches to freight transport modal choice

174 R. Gray

McGinnis (1979) undertook a study of the attitudes of shippers in the U.S.A. toeight topics which he considered were likely to influence their choice of transport andwhich were

(1) Freight rates(2) Speed(3) Reliability(4) Loss and damage(5) Inventories(6) Company policy(7) Shipper market conditions(8) Influence of the shipper's customers

In this study the shipper is equated with the traffic executive and the results arebased on 351 questionnaires returned from a broad cross-section of industries.Respondents were asked to state the degree of importance on a 5-point scale of eachof 30 statements associated with the above eight topics. Factor analysis of the matrixof scale responses was undertaken to identify possible underlying dimensions. Theresulting factors largely reproduced the original topics, although statementsassociated with speed and reliability loaded on a single factor suggesting that trafficmanagers do not consider them as separate constructs.

The three factors which were found to be most important (based on meanimportance scores) are, in descending order of importance, the factors associatedwith speed and reliability, freight rates, and loss and damage. McGinnis considersthat this shows that researchers should take care in making assumption about therelative importance of different variables included in choice models. However, sucha comment raises the issue of the exact nature of the shipper. In most economic andtechnological positivist approaches the shipper is assumed to be the 'firm' whereasthe perceptual approach of McGinnis is based on the attitudes of traffic managers.The tendency to equate the attitudes of particular individuals with the actions of thetotality of the firm is a major limitation of the perceptual approach. Traffic managersare much more likely to be concerned with speed and reliability of transport modesthan with inventory control or market conditions for the shipper firm's goods.

Miklius and Casavant (1975) attempt to compare shippers' perceptions withactual performance of transport modes with the objectives of (a) quantifying andcomparing the actual reliability of delivery time for rail and truck (lorry) and (b)comparing the actual reliability of delivery time to that perceived by shippers. Thedata on actual transit times was obtained from the records of shipments of cherries inthe U.S.A. Based on a sample of nearly 2000 road and nearly 1000 rail shipments itwas found that there was no significant difference in variability of transit timebetween rail and road transport. The perceived variability of transit time wasobtained by asking 70 produce buyers for both road and rail how many of 100shipments would arrive on the promised day, one day late, two days late etc. It wasfound that the expected delay perceived by shippers was much greater than theestimated mean delay based on the survey of actual shipments. Furthermore, it wasfound that the difference between the estimated and perceived variability of transittimes was much worse for rail than for road.

Miklius and Casavant conclude that either the perceived delays are dis-proportionately influenced by a few long transit times or that perceptions byshippers have not kept up with the improvements in the reliability of rail services

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 16: Behavioural approaches to freight transport modal choice

Freight transport modal choice 175

which are widely held to have taken place. In other words, rail has becomestereotyped as offering poorer transit times.

A development of the perceptual approach may be seen in studies which arederived from organizational buyer behaviour theory. A considerable literature onorganizational purchasing has developed in recent years (see Sheth 1976), oftenderived from the behavioural theory of the firm developed by Cyert and March(1963) and the organizational theory of Simon and others (e.g. March and Simon1970). This approach suggests that a firm has a number of different goals undertakenby different organizational members (e.g. production, sales) and that the differentgoals place constraints on the activity of organizational members through thebargaining which different members must undertake with each other. Thus,organizational buyer behaviour is seen as a decision-making process undertaken byone or more individuals in buying organizations.

The significant actors in the decision-making process are considered as adecision-making unit (in contrast to the 'firm' as the decision-making unit in theeconomic positivist approach) or as a 'buying centre' (Webster and Wind 1972).However, the group is seldom taken to be the basic unit of analysis in the sense thatthe firm is a unit of analysis in the economic theory of the firm. The basic focus ofresearch is generally the individual decision-maker, usually the buyer or purchasingofficer, but, unlike the basic perceptual approach, he is placed explicitly in hisorganizational context. To a certain extent this overcomes the criticism of theperceptual approach where attitudes of individuals tend to be equated with actionswhich may be the outcome of the decisions of a large number of people.

Saleh and La Londe (1972) examined buyer behaviour in the selection of motorcarriers (road hauliers) in the U.S.A. in terms of the 'buygrid' analysis developed byRobinson and Faris (1967). A 'straight rebuy' is the purchase of the services of a roadhaulier as a frequent and recurring activity where selection is based on a limitedrange of previously determined acceptable carriers. A 'modified rebuy' occurs whenshippers wish to purchase transport services in order to reduce transit times ortransport costs as a result of changes in transport requirements (e.g. a customerchanges his location), consistently poor performance by a carrier, or contact initiatedby a competitive carrier. The research was based on personal interviews with 48traffic managers combined with a validation sample of over 400 respondentscompleting a postal questionnaire.

Results show that carrier selection decisions are on the whole made very quickly.Over half of the respondents claim to select the carrier 'instantaneously'. Whereasshippers rate highly the value of a search for alternative carriers (94% of the trafficmanagers agreeing that they should look 'across the board' when selecting a carrier),behavioural data showed that only about 12% of the respondents consider more thanfive alternative carriers in a market consisting of a large number of operatingcompanies. The results suggest that caution should be exercised when seekingattitudes in a setting where there are strong normative pressures. An extensive searchfor carriers may be widely considered as good practice but may not be actuallycarried out. Saleh and La Londe conclude that there is little planning anddeliberation by shippers when selecting road hauliers. In the general area of businessdecisions the selection of road hauliers involves a low level of perceived risk andtherefore it is considered a relatively unimportant decision-making activity.

A study by Cunningham and Kettlewood (1976) is concerned primarily with theexistence of 'source loyalty' in the freight transport market. Source loyalty is a

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 17: Behavioural approaches to freight transport modal choice

176 R. Gray

concept used in organizational buyer behaviour studies and refers to the extent towhich a buyer remains 'loyal' to a particular supplier. Cunningham and Kettlewoodexamined loyalty both to specific modes and to specific suppliers of a given mode.The results of the research were based on interviews with persons responsible fortransport decisions in 43 company establishments in Scotland. The average length oftime for which both a principal operator and a particular mode on a specific routewere maintained was about 11 years and where a change of use of mode occurred itwas mainly from rail to public road haulier.

The importance of freight transport to the shipper was examined in terms of costas a proportion of sales value, as the pressure for cost savings by organizationalsuperiors, and as the information requirements of senior management. On average,costs were only 2-4% of sales value, ranging from 1-5% for engineering goods to15-7% for mining and quarrying products. Measure of the pressure for cost savingswas based on the number of mentions by respondents of particular aspects, with 75 %of respondents mentioning the need to keep the consignee satisfied or to keepwarehouses clear of goods. Only 21% of respondents thought that cost minimizationwithout qualification was required. Over 50% of respondents provided no inform-ation to their superiors on transport expenditure.

About 75% of the freight transport buyers were allocated transport buyingresponsibility on a part-time basis as a secondary responsibility, so that they werelikely to attempt to simplify their work not only through source loyalty but also by 'establishing contracts with suppliers of transport, by having formal transportpolicies, by establishing a standard rate of charges with suppliers of transport, and byownership of their own vehicles.

§ 4. DISCUSSION

There is clearly a desire among researchers to develop behavioural approaches tothe analysis of freight modal choice. However, as this review has shown there are anumber of issues which need to be clarified before effective behavioural models canbe developed.

Such issues include

(1) The nature of the shipper' (2) Assumptions about shippers' knowledge(3) The nature of modes

4.1. The nature of the shipperThe assumption that the choice made by shippers is based on a utility assessment

is common to both the economic and the technological positivist approaches. Mostfreight modal choice models operate within a 'simple' utility assessing framework.They are 'simple' in the sense that a single decision-maker, or the single view of aconsensus of decision-makers, is assumed.

A basic difference between the economic positivist approach and the tech-nological positivist approach is that the utility in the former is equated with profit-maximization of the shipper firm, whereas the utility in the technological positivistapproach is an optimal combination of technological factors.

Many behavioural passenger travel demand models are based on utility assess-ment, whether in the form of deterministic utility maximization or as a probabilisticchoice based on utility assessment (see, for example, Stopher and Meyburg 1976).This approach is possibly acceptable for personal travel where the unit of analysis is

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 18: Behavioural approaches to freight transport modal choice

Freight transport modal choice 177

generally the individual person, but it is necessary to question whose utility isassumed in freight studies. Little research activity has been devoted to identifyingthe significant actors in freight modal choice decision-making. In fact, it isreasonable to state that complicated mathematical decision models have beendeveloped without knowing exactly who are the decision makers. Modal choice may,for example, be the outcome of the conflicting requirements of marketing, transportand finance personnel within the shipper company. This area of conflict has beenwell documented in the trade press serving the transport industry for a number ofyears.

For example, an article in Freight Management (1975) suggests that many firmshave failed to take total distribution costs into account when choosing between airand surface transport. It advocates that costs other than pure freight costs should beconsidered by shippers when selecting transport (e.g. capital tied up during transit).The article suggests that the failure is the result of inadequate links between thestrategic decisions of senior management and the tactical decisions of managersresponsible for transport.

This conclusion is supported by the results of a survey by Davies and Gray (1980)of 392 shipping managers (managers responsible for international transport) in theU.K. It was found that although 43% of the respondents were each responsible for£100000 or more of export freight expenditure in a year, 67% of respondentsresponsible for freight expenditure did not have to work within a defined budget forfreight. This is an indication that senior management does not recognize theimportance of international transport as a major item of expenditure. Such resultsare supported by an attitude survey where Davies and Gray found that 73% ofshipping managers agree that their senior management believed shipping to be arelatively unimportant activity, and 62% agreed that they had insufficient contactwith top management. Similar findings about the inadequate links betweentransport decision makers and senior management have been claimed in studies ofdomestic transport in the U.K. (Buxton and Lee 1975), domestic transport in theU.S.A. (Johnson and Borger 1977), and in both the U.S.A. and Canada (La Londeand Lambert 1977).

The movement of goods is quite different from the movement of people. Unlikepersonal travel the decision-maker does not undertake the actual movement andthere is likely to be a number of decision-makers, each with different interests in themovement. Therefore, in most cases it is unrealistic to assume that freight modalchoice can be described in terms of individual choice models, particularly when the'individual' is equated with a firm, or, more probably, with a researcher's theoreticalassumption of how a firm 'decides'. An approach is required which identifiesrelevant socio-organizational groups and explicitly examines their relationship to thetechnology of the transport system and the cost functions of the firms involved.

4.2. Assumptions about shippers' knowledgeIn the neo-classical economic theory of the firm there is an assumption of perfect

knowledge on the part of the firm which is not evident in empirical studies. In fact,there has been little research into the extent of knowledge of shippers in the area offreight modal choice, and this should be a prerequisite to the establishment ofbehavioural models.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 19: Behavioural approaches to freight transport modal choice

178 R. Gray

There is some evidence of the limitations of the knowledge of shippers. As statedearlier, Miklius and Casavant (1975) identified a difference between perceived andactual transit times. Bayliss and Edwards (1970) found that less than half of the firmsin their survey knew the costs of transport by alternative modes for some or all oftheir consignments, and a smaller proportion of firms were able to providealternative charges for actual consignments. They suggest that the shipper is notnecessarily acting irrationally when he does not seek knowledge of the cost ofalternatives since he may base his actions on a general idea of alternative charges.

In many models there is an assumption that the researcher knows what theshipper will take into account, so that Baumol and Vinod (1970) claim that

the choice of mode by passenger is ultimately a psychic matter ofconsumer desires in which the economist is able to say only that travellersprefer whatever they happen to prefer. Since no pleasure is ordinarilyderived from the means chosen for freight transportation, the selection ofa carrier is likely to be based on economic considerations that areamenable to formal analysis.

Baumol and Vinod assume that a shipper views freight movements in terms ofinventory theoretic cost functions. There is certainly considerable prescriptivemanagement teaching which states that firms ought to behave in this manner, but theauthor is unaware of any empirical evidence for or against this assumption.

Throughout modal choice studies one finds reference to the difficulties of datacollection. The U.S. Transportation Research Board (1977) claims that

the inconclusiveness of results obtained is usually attributed to problemsof data availability.

Roberts (1971), describing the application of a model to the individual shipper, says,

data could present a problem with this model since a great deal moreinformation is required than for most inventory models.

As shown in §3.1, owing to a lack of data Levin (1978) is forced to compromise hismodel and use aggregated data, and Oum (1979) is forced to make unrealistic (in hisown terms) assumptions.

There is sometimes too great a tendency to use published data (often from thegovernment) and a reluctance to obtain valid empirical data which is moreappropriate to the shipper's own knowledge and decision-making. This may ofcourse reflect the difficulties in establishing contact with shippers owing tocommercial confidentiality or even a lack of interest on the part of shippers.However, in some instances researchers appear to have a greater interest in the formof their models than in the reality of shippers' behaviour.

4.3. The nature of modesIn many studies of freight modal choice there is little said about the nature of

modes. Although justification is often given for the 'choice' model adopted in thestudy, the selection of modes sometimes appears arbitrary; possibly justified in anapplied research area by the requirements of the agency sponsoring the research.However, it is possible that the selection of modes may influence the results of thestudy. The research by Bayliss and Edwards (1970) includes three binary modalchoice analyses which refer to the distribution of transport between vehicles owned

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 20: Behavioural approaches to freight transport modal choice

Freight transport modal choice 179

by the shipper and hire and reward vehicles operated by professional hauliers.Starkie (1971) has criticised the results which suggest that 'length of haul' is theoverwhelming determinant of modal choice because the use of own transportdecreases sharply with distance. Starkie claims that it is almost certainly a surrogatefor a pattern of ownership of vehicle based on specific legal restrictions on own-account operators in existence in the U.K. at the time of the study.

Stenger and Cunningham (1976) found that the only variable of any majorsignificance in predicting modal choice within the competitive area (see §3.2) wasshipment size. This result should not be unexpected since the types of mode whichwere considered are to a large extent defined by their capacity. They are rail carload,trailer-on-flat-car (piggyback), truckload (i.e. lorry load), less-than-truckload andfreight forwarder/shippers' association. Even the last-mentioned mode may beclassified by shipment size since the type of operator mentioned undertakesconsolidation of smaller shipments.

Many studies adopt a plurality of bases for selecting modes. For exampleGilmour (1976) examines the modal choice between own road transport, normalroad service with a freight forwarder, overnight road service with a freightforwarder, unit train, normal rail freight, sea and air. This selection is based partly onownership patterns (own road transport compared with hire and reward roadtransport), partly on a form of abstract mode using time differentials of a singlephysical mode (overnight road compared with normal road transport), and partly ona fundamental physical difference between modes (air compared with sea).

It is possible to identify four major classifications of mode in empirical studies,and it is possible that results of studies would be more successful if modal choicewere examined separately according to each classification. The classifications are:

(1) Evident physical differences (e.g. air compared with road transport).(2) Pattern of ownership of the transport mode (ownership by the shipper or

ownership by the transport operator). The organizational relationshipsbetween the shipper and the transport world would appear to be under-estimated as an explanation of modal choice. The movement of goods isseldom perceived by the researcher as an interorganizational social system,although it is often a principal factor (possibly unwittingly) in the definitionof modes in some studies. For example, socio-organizational factors wouldappear to be relevant to the choice between using the shipper's own vehiclesand using those of professional hauliers, or between using British Rail (amonopoly operator) and road transport (a highly competitive market in theU.K.).

(3) Potential size of consignment. This refers to the constraints placed on theconsignment size for each mode. Weight or volume constraints may arisefrom the physical limitations of the transport equipment or from the nature ofthe commercial operating system. The most common example of the latter isa service which offers consolidation (sometimes called groupage or less-than-full load).

(4) Abstract mode. This approach, described in §3.1, treats all modes as variantsof a single prototype with different combinations or values of attributes(Baumol and Vinod 1970). The concept of continuity of measurement isuseful in the area of research into freight transport. The often limited datawhen divided by modal or commodity categories may become a richer source

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 21: Behavioural approaches to freight transport modal choice

180 R. Gray

of information when considered as continuous data. However, it is notalways easy to establish continuity. For example, Baumöl and Vinod explainthe difficulty of establishing continuity of data between full loads and less-than-full loads (treated as different modes) owing to the step-wise relation-ship of freight rates.

§5. GENERAL CONCLUSIONS

Behavioural approaches to freight transport modal choice have difficulty inidentifying the appropriate units of behaviour. The dominant tendency is to assumethat the 'firm' is the sole unit of behavour. This tendency is found in all threeapproaches—economic positivist, technological positivist, and perceptualapproaches—described in this review. Research into freight modal choice at thebehavioural level has tended to disregard the nature of human choice and assumethat modal choice can be explained in terms of cost relationships or technologicalphenomena, or to assume that the preferences of particular organizational memberscan be equated with the use made of modes by shipper firms. There is probablyconsiderable scope for investigation of the relationships between different organiz-ational members involved in the modal choice and associated decisions.

There is a tendency among some researchers to adopt prescriptive approachessuch as the logistics approach, and assume that they are descriptive approaches in theformulation of their models. It is possible that better results would be achieved ifgreater effort was devoted to understanding how shippers and other interestedparties behave, rather than developing more sophisticated mathematical models.

FOREIGN SUMMARIES

Les déterminants du choix modal en transport de marchandises font l'objet despréoccupations de trois types principaux d'institutions: les administrations, les chargeurs etles transporteurs. De nombreux chercheurs ont souligné l'importance d'une bonne com-préhension du processus de décision chez les agents intervenant dans le choix modal, tels queMeyburg dans l'ouvrage de Hensher et Stopher (1979), Roberts (1971),.. .Cette approcheimplique le développement de modèles de comportement et l'article fait le point de la situationà cet égard.

La plupart des modèles existant peuvent être rattachés à deux écoles principales. Unepremière famille de modèles fait l'hypothèse selon laquelle le choix modal est fait par lechargeur en fonction d'une optimisation des coûts à court terme. On peut appeler ceci du'positivisme économique' puisqu'on suppose que le choix résulte de variables de coût oud'autres données économiques. Pour une deuxième famille de modèles, le choix résulte deconsidérations sur des caractéristiques physiques du système de transport, telles que lavitesse, la fréquence,... et sur des caractéristiques physiques du produit, telles que soncaractère périssable ou son prix à la tonne. Il s'agit cette fois de 'positivisme technologique',puisque le choix modal y est fonction de variables technologiques. Il y a en plus une troisièmeapproche dont les hypothèses ont trait aux perceptions des agents des organisations liées autransport, et en particulier celles des chefs de service transports. On pourrait parler ici d'une'approche par la perception'.

L'article présente des études empiriques réalisées en Australie, au Canada, au RoyaumeUni et aux Etats Unis. Un aspect particulièrement mis en évidence est la liaison entre leshypothèses des disciplines dans lesquelles se situent les modèles et la taille des unités d'analysecorrespondantes: la firme, une personne individuelle, l'envoi,...

Es sind hauptsächlich drei Institutionen, die an den Determinanten derTransportmittelwahl im Güterverkehr Interesse zeigen: die Regierung, dasTransportunternehmen und der Spediteur. Wissenschaftler haben die Bedeutung herausges-tellt, den Entscheidungsprozeß der Beteiligten bei der Transportmittelwahl im Güterverkehr

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 22: Behavioural approaches to freight transport modal choice

Freight transport modal choice 181

nachzuvollziehen (z.B. Meyburg in Hesher & Stopher (1979), Roberts (1971). Hieraus leitetsich die Entwicklung verhaltensorientierter Modelle ab. Dieses Papier gibt den augenblick-lichen Entwicklungsstand dieser Modellansätze wieder.

Die meisten Modelle können unter zwei hauptsächlich maßgebenden Leitsätzen sub-sumiert werden. Erstens gibt es Modelle, bei denen unterstellt wird, daß dieTransportmittelwahl nach einer gewissen kurzfristigen Kosten-Optimierung durch denSpediteur getroffen wird. Dieser Ansatz mag 'ökonomisch positivistisch' gennant werden, daer davon ausgeht, daß Transportmittelwahl durch ökonomische oder Kostenvariablenbestimmt wird. Zweitens gibt es Modelle, bei denen unterstellt wird, daß dieTransportmittelwahl von Verhältnissen zwischen physikalischen Aspekten desTransportsystems (z.B. Geschwindigkeit, Frequenz) und physikalischen Aspekten des zutransportierenden Produkts (z.B. Verderblichkeit der Ware, Wert-Gewichts-Relation)abhängt. Dieser Ansatz mag 'technologisch positivistisch' gennant werden, da er davonausgeht, daß Transportmittelwahl von technologischen Variablen bestimmt wird. Zusätzlichgibt es einen dritten allgemeinen Ansatz für die Transportmittelwahl im Güterverkehr.Dieser stützt sich in seinen Annahmen auf die Kenntnisse oder Vorstellungen der Mitgliedervon Speditions-Organisationen, speziell auch auf die Kenntnisse von Disponenten. DieserAnsatz mag 'perzeptiv' genannt werden.

Der Artikel untersucht empirische Studien aus Australien, Canada, Großbritannien undden USA. Besonderes Augenmerk wird auf die unterschiedlichen Annahmen derverschiedenen Disziplinen geworfen von denen die Modelle stammen, sowie auf dieBezugseinheit der Analyse (z.B. Firma, Einzelperson, Kommission).

Los tres tipos de instituciones interesadas en saber que determina la elección de un mediode transporte de carga sobre otros son el gobierno, los transportistas y los embarcadores.Investigaciones al respecto han enfatizado la importancia de entender el mecanismo de etomade decisiones de quienes participan en el proceso de elección de medio de transporte de carga(ej. Meyburg, en Hensher y Stopher, 1979 y Roberts 1971). Este artículo revisa el estado delarte de los modelos de comportamiento que se requiere desarrollar para esto y tambiénevidencia empírea de estudios realizados en Australia, Canadá, Gran Bretaña, y los EE.UU.Se discute en forma especial las suposiciones de cada una de las disciplinas usadas para generarlos modelos, y las unidades relevantes de análisis (ej. la empresa, el individuo y el envío).

La mayorí de los modelos se puede clasificar de acuerdo a dos doctrinas dominantes. Por unlado están los modelos que suponen que la partición modal se basa en alguna forma deoptimización de corto plazo por parte del embarcador, y por otro están los modelos quesuponen que la partición modal se basa en relaciones existentes entre aspectos físicos delsistema de transporte (ej. velocidad, frecuencia) y aspectos físicos del producto (ej. razón,valor-peso, si el producto es o no perecible).

El primer enfoque se puede denominar 'positivismo económico', ya que supone que lapartición modal está determinada por variables económicas o de costo. El segundo puededenominarse 'positivismo technológico' ya que supone que la partición modal está de-terminada por variables tecnológicas. Además de éstos, existe un tercer enfoque grueso quebasa sus suposiciones en las percepciones de miembros de organizaciones de transporte, enparticular, gerentes de empresas. Este enfoque se puede denominar 'enfoque de perception'.

REFERENCES

ALLEN, W. S., 1977, The demand for freight transportation: a micro approach. TransportationResearch, 11, 1.

BAUMÖL, W. J., and VINOD, H. D., 1970, An inventory theoretic model of freight transportdemand. Management Science, 16, 413.

BAYLISS, B. T., 1971, Modal-split in freight transport—(a micro approach based on ananalysis of consignments), Freight Traffic Models (Planning and Transport Researchand Computation Symposium Proceedings, London), pp. 121-123.

BAYLISS, B. T., 1972, Demand for freight transport—practical results of studies on marketoperation, Report of 20th Round Table on Transport Economics, European Conference ofMinisters of Transport.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 23: Behavioural approaches to freight transport modal choice

182 R. Gray

BAYLISS, B. T., and EDWARDS, S. L., 1970, Industrial Demand for Transport (Ministry ofTransport).

BOWERSOX, D. J., 1974, Logistical Management (Macmillan).BUXTON, G., and LEE, L., 1975, A profile of the UK distribution executive and his

organisational responsibilities. International Journal of Physical Distribution, 5, 280.CUNNINGHAM, M. T., and KETTLEWOOD, K., 1975, The influence of the image of suppliers on

buyer behaviour in the freight transport market. International Journal of PhysicalDistribution, 5, 238.

CUNNINGHAM, M. T., and KETTLEWOOD, K., 1976, Source loyalty in the freight transportmarket. European Journal of Marketing, 10, 60.

CYERT, R. M., and MARCH, J. G., 1963, A Behavioural Theory of the Firm (Prentice-Hall).DAVIES, G. J., and GRAY, R., 1980, Who Buys International Freight Services? A survey of UK

Shipping Managers (Maclean Hunter).Freight Management, 1975, Freighting by air—whose decision? (Anon), February, p. 24.GALBRAITH, J. K., 1972, The New Industrial Estate (Andre Deutsch).GILMOUR, P., 1976, Some policy implications of subjective factors in the modal choice for

freight movements. The Logistics and Transportation Review, 12, 39.HARTWIG, J. C, and LINTON, W. E., 1974, Disaggregregate Mode Choice Models of Intercity

Freight Movement, Transportation Center Research Report (Northwestern University).HEGGIE, I. G., 1978, Putting behaviour into behavioural models of travel choice. Journal of the

Operational Research Society, 29, 541.JOHNSON, J. C, and BORGER, D. L., 1977, Physical distribution: has it reached maturity?

International Journal of Physical Distribution, 7, 283.LA LONDE, B. J., and LAMBERT, D. M., 1977, A comparative profile of the United States and

Canadian distribution managers. International Journal of Physical Distribution, 7, 264.LANCASTER, K.J., 1966, A new approach to consumer theory, Journal of Political Economy ,74,

132.LEVIN, R. C, 1978, Allocation in surface freight transportation: does rate regulation matter?

Bell Journal of Economics, 9, 18.LUCE, R., and RAIFFA, H., 1957, Games and Decisions (John Wiley).MARCH, J. G., and SIMON, H. A., 1970, Decision-making Theory, The Sociology of

Organisations, edited by O. Grusky and G. A. Miller (The Free Press).MATHEMATICA, 1967, Studies on the Demand for Freight Transportation, Vol. 1, prepared for

the Northeast Corridor Transport Project (U.S. Dept. of Transportation).MCGINNIS, M. A., 1979, Shipper attitudes toward freight transportation choice: a factor

analytic study. International Journal of Physical Distribution and MaterialsManagement, 10, 25.

MEYBURG, A. H., 1979, The applicability of behavioural modelling to the analysis of goodsmovements. Behavioural Travel Modelling, edited by D. A. Hensher and P. R. Stopher(Croom Helm).

MIKLIUS, W., and CASAVANT, K. L., 1975, Estimated and perceived variability of transit time.Transportation Journal, 15, 47.

OUM, T. H., 1979, A cross sectional study of freight transport demand and rail-truckcompetition in Canada. Bell Journal of Economics, 10, 463.

ROBERTS, P. O., 1971, The logistics management process as a model of freight traffic demand,Freight Traffic Models (Planning and Transport Research and ComputationSymposium Proceedings, London), pp. 141-146.

ROBERTS, P. O., 1977, Proceedings of the 3rd World Conference on Transport Research,Rotterdam 26-28 April 1977 (Mārtiņus Nijhoff).

ROBINSON, P. J., and FARIS, C. W., 1967, Industrial Buying and Creative Marketing (Allyn andBacon).

SALEH, F., and LA LONDE, B. J., 1972, Industrial buying behaviour and the motor carrierselection decision. Journal of Purchasing, 8, 18.

SHETH, J. N., 1976, Recent developments in organisational buying behaviour, FacultyWorking Paper no. 317, College of Commerce and Business Administration(University of Illinois at Urbana-Champaign).

STARKIE, D. N. M., 1971, Review oí Industrial Demand for Transport, Journal of TransportEconomics and Policy, 5, 224.

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 24: Behavioural approaches to freight transport modal choice

Freight transport modal choice 183

STENGER, A. J., and CUNNINGHAM, W. H. J., 1976, An Experiment in Freight Modal ChoiceDelineating the Rail-truck Interface. Report No. DOT-TST-76-83. (U.S. Dept. ofTransportation).

STOPHER, P. R., and MEYBURG, A. H. (eds), 1976, Behavioural Travel-Demand Models(Lexington Books).

TRANSPORTATION RESEARCH BOARD, 1977, Freight data requirements for statewide transpor-tation systems planning, National Cooperative Highway Research Program Report 177(National Research Council, Washington, D.C.), pp. 65-86.

VAN BUIJTENEN, P. M. (ed), 1976, Business Logistics (Mārtiņus Nijhoff).WEBSTER, F. E., and WIND, Y., 1972, Organisational Buying Behaviour (Prentice-Hall).

EDITORIAL SUGGESTIONS FOR FURTHER READING

BLAUWENS, G. J., 1980, The spatial theory of the demand for freight transport: asurvey. Changes in the Field of Transport Studies, edited by J. B. Polak, and Vander Kamp (Mārtiņus Nijhoff, The Hague), pp. 14-25.

Many spatial models of the demand for freight transport have been developed inthe economic literature and this paper surveys the scene. The first part attempts aclassification of the various models with a focus on general spatial equilibriummodels. The following two sections present special cases of equilibrium models incontinuous and discrete space in both a theoretical and applied context. Theconclusions emphasize the essentially theoretical nature of most research andsuggests the reasons for their limited application.

(D.B.)

OUM, T. H., 1979, Derived demand for freight transport and intermodalcompetition in Canada, Journal of Transport Economics and Policy, 13 (2), pp.149-168.

Demands for railway, highway and waterways modes are derived from produc-tion and distribution technology and estimated from time-series data of Canadianintercity freight transport, 1945-75. These are, in turn, used to measure priceelasticities of demand. Both own-price elasticities and possibilities of competitionbetween modes have increased steadily over time. Highway modes essentiallycomplemented railway services until the late 1950s, and thereafter became increas-ingly competitive. Shippers are found to respond to price changes with systematiclag, with mean of eight months and variance of thirteen months.

(Author)

ROBERTS, P. O., and KULLMAN, B. C, 1979, Urban goods movement: behaviouraldemand-forecasting procedures, Behavioural Travel Modelling, edited by D. A.Hensher and P. R. Stopher (Croom Helm), pp. 553-76.

This paper has three broad goals. First, it develops a methodology that can beused to establish the industry/commodity data base that is essential to allcommodity-based freight-transport analysis. Second, it develops procedures thatcan be used to specify and estimate disaggregate demand models of freight modechoice, using the industry/commodity data developed above as well as carrierattributes, i.e. transport levels of service for different modes. Third, it describes howthe industry/commodity data base can be combined with probabilistic choice modelsto create a powerful tool for analysis.

(Authors)

Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013

Page 25: Behavioural approaches to freight transport modal choice

184 Freight transport modal choice

WIGAN, M., 1979, Indicators for urban commodity movements, Behavioural TravelModelling, edited by D. A. Hensher and P. R. Stopher (Croom Helm),pp! 577-611.

Behavioural or modelling analyses of freight issues require considerable care inspecifications and testing, as the complexity of freight issues is such that the linksbetween the observable consequences and the policy or operational changes requirestringent control of the model or behavioural analysis ab initio.

The number of different dimensions of urban commodity movements andimpacts is such that cross-sectional surveys are inadequate to keep track ofgeographical, spatial and demand shifts, and the employment, environmental,economic and social impacts. Links between data sources, policy issues andindicators for monitoring purposes are set up.

The limitations of cross-sectional methods are covered in detail, and land-useinfluences and changes in operational use are also pinpointed through the com-prehensive and diverse data available for the Greater London Council area.

(Author)

ZANDI, I., WARNER, J. A., and MERSKY, R., 1979, Freight pipeline, rail, and truckcost comparison, ASCE Journal of Transportation Engineering, 105 (4),pp. 411-25.

This paper describes the use of four models for comparing the cost oftransporting manufactured products via freight pipeline, truck, truck on flat car(TOFC), and rail carload. It was found that cost comparison is sensitive to thetransportation configuration, i.e. location, transport distance, and the number ofterminals, annual tonnage, the size of shipment, the design of pipeline system, accessconditions, and accounting procedure. The study shows that, based on theannualized cost methodology, the transport cost per ton-mile via freight pipeline iscost competitive with truck, rail, and TOFC for four of the five configurations withinthe Philadelphia-Chicago transportation corridor if the annual tonnage is high. Forthe fifth configuration, truck is clearly a less expensive mode. However, based on theproject's present value methodology, pipeline is cost competitive in all fiveconfigurations. Based on the annualized cost methodology, pipeline becomesreasonably attractive for three configurations and for one configuration it comparesvery favourably (2-7 cents/ton-mile for pipe versus 2-8 cents/ton-mile for T-Rcombination).

(Authors)Dow

nloa

ded

by [

Mos

kow

Sta

te U

niv

Bib

liote

] at

15:

17 0

3 D

ecem

ber

2013