18
Click here to load reader
Click here to load reader
This article was downloaded by: [Dalhousie University]On: 23 September 2013, At: 14:49Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Maritime Policy & Management: Theflagship journal of internationalshipping and port researchPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/tmpm20
Monetary values of transport serviceattributes: land versus maritime ro-ro transport. An application usingadaptive stated preferencesAngela Stefania Bergantino a & Simona Bolis ba Department of Economics, University of Bari, Italyb Faculty of Economics, University of Lugano, Via Ospedale, 13,6900 Lugano, SwitzerlandPublished online: 08 Apr 2008.
To cite this article: Angela Stefania Bergantino & Simona Bolis (2008) Monetary values of transportservice attributes: land versus maritime ro-ro transport. An application using adaptive statedpreferences, Maritime Policy & Management: The flagship journal of international shipping and portresearch, 35:2, 159-174
To link to this article: http://dx.doi.org/10.1080/03088830801956821
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 tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand 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 Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout 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
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
MARIT. POL. MGMT., APRIL 2008VOL. 35, NO. 2, 159–174
Monetary values of transport service attributes: landversus maritime ro-ro transport. An application usingadaptive stated preferences
ANGELA STEFANIA BERGANTINO*y and SIMONA BOLISz
yDepartment of Economics, University of Bari, ItalyzFaculty of Economics, University of Lugano, Via Ospedale, 13, 6900Lugano, Switzerland
The objective of this work is that of illustrating a methodology to derive, in termsof trade-offs, the relative values of transport service attributes for logisticsoperators. In particular, the analysis yields insights on the preferences for time-savings, frequency and reliability of the transport service in terms of price, of asample of freight forwarders located in the South of Italy. The data is gatheredusing an Adaptive Stated Preference application and the estimates are obtainedthrough a Tobit ML estimator on both company specific data and pooled data.The results show that frequency is the most highly rated characteristic of theservice together with reliability. The value of time is significantly lower in bothcases. The outcome is consistent across estimations and, substantially in line withthe outcome of previous studies carried out on operators active at different levelsof the transport-logistic chain. Until now, in fact, studies have been carried outuniquely on carriers or producers. The selection of freight forwarders allows toshed light on a segment of the market that accounts for more than half of thetransport-related decisions. Furthermore, the present study focuses on acomparison between all land transport and maritime ro-ro alternatives, in linewith the growing interest on the integration among transport modes. The outputsignals both the absence of any a priori preclusion for ro-ro maritime services andthe extremely important role that frequency of service assumes for thedevelopment and the establishment of maritime services as realistic alternativesto all land transport.
1. Introduction
The purpose of this paper is to determine the relative importance of transport serviceattributes for users, when a short sea shipping service alternative is available [1–8].The growing interest to move freight traffic off the road network towards less
polluting alternative modes, has, only recently, been accompanied by significantefforts to quantify, empirically, the relative importance of the determinants ofoperators’ choices. In this context, the superiority of stated preference techniquesversus revealed preference techniques is generally accepted, mainly due to the
characteristics of the data needed for the experiments [9]. More recently, a growingbody of literature has been emphasizing the advantages of combining revealedpreference and stated preference data in order to exploit the strengths of both
*To whom correspondence should be addressed. e-mail: [email protected]
Maritime Policy & Management ISSN 0308–8839 print/ISSN 1464–5254 online � 2008 Taylor & Francishttp://www.tandf.co.uk/journalsDOI: 10.1080/03088830801956821
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
[10–18]. Although our work can be placed within this body of literature, it differsfrom most of the previous analyses in three main respects.
First, we have opted for an interactive approach which allows operators’preferences to be elicited on hypothetical alternatives created on the basis of therespondent’s choice. In particular, we carried out an adaptive stated preferenceexperiment to collect data on shippers’ preferences with respect to hypotheticalalternatives to the currently (revealed) adopted service. In the present applicationrevealed preference techniques are used in gathering data on the current choices ofthe respondent and to select a ‘‘typical’’ transport for each company (the currentchoice); the rest of the data are collected through adaptive stated preferencetechniques. The hypothetical services are constructed through a particular formof ‘‘pivoting’’, defined by Train and Wilson [18] stated preference of revealedpreference approach. The attributes in the stated preference experiment are, in fact,created changing the attributes of the chosen revealed preference alternative.Furthermore, the alternatives are varied (improved or made worse off), dependingon the respondent’s choice at each iteration.
The advantage of pivoting is that it creates more realism in the stated preferenceexperiment by assuring that the alternatives are similar to the respondent experiencein a revealed preference setting. It also provides a greater specificity of the contextof the stated preference task, since the respondent can think of the stated preferencealternatives. On the other hand, as we shall see, it creates some problems in theutilization of the data collected.
Second, most of the previous studies do not focus on a specific mode but, instead,leave the choice of the type transport mode to the respondent identifying only itsattributes [19–23]. In our study, instead, we have restricted the modal choice tomaritime ro-ro services in order to obtain information on consumers’ preferencesand attitudes for short sea shipping transport, being, one of the scope of the analysis,that of identifying the potential market for this type of service.
Finally, although most of the studies focusing on freight transport alternativesconsider producers’ preferences for transport services attributes, we have chosen tofocus our application on freight-forwarders instead of producers [19–28]. The latterhas allowed us to gain insight on a part of the market for transport service, whichaccounts, on average, for more than half of the transport decisions. Outsourcing oftransport operations is, in fact, spreading rapidly and, as a recent study points out,freight forwarders are increasingly assuming full responsibility over the completecycle of the transfer of the freight from door-to-door [29]. Selecting freight-forwarding agents has two main advantages. On the one hand, it allows us tocomplete the analysis of transport services users’ preferences: insights can be gainedfrom a wider spectrum of consumers. On the other hand, it allows overcomingthe issue of obtaining a representative population across the different productivesectors at an acceptable cost.
Furthermore, a comparison of the outcome of our study with those of thepapers focusing merely on the producers’ behaviour helps in understandingwhether the preferences of these two set of transport service consumers tend toconverge or to diverge and, thus, whether a need exists for taking differentiatedpolicy/marketing initiatives or for preferring the analysis of one of the two groupof operators.
Our paper, analyses the individual preferences of 16 freight forwarding managers,located in the southern Italian region, and in the gravitational area of the Apulian
160 A. S. Bergantino and S. Bolis
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
ports. To the authors’ knowledge, this is the first adaptive stated preference
experiment performed with the scope of determining the preferences of operators in
terms of service attributes of sea transport and of studying the potential reallocation
of traffic from surface transport services to maritime ro-ro services.The rest of this paper is organized as follows. In section 2 we describe the
methodology used to assemble the dataset, the criteria followed in identifying the
sample and the design of both the revealed preference survey and the adaptive
stated preference experiment. Section 3 contains a detailed description of the
database and an illustration of the estimation procedure, section 4 reports the main
empirical findings and a brief comparison of the main results of other EU studies.
Section 5 briefly summarizes the main conclusions.
2. Data collection methodology
2.1. The adaptive state preferencesThe methodology falls within the broad family of conjoint analysis experiments, as
we attempt to determine the value that individuals place on any product as
equivalent to the sum of the utility they derive from all the attributes pertaining to
a specific transport service. The conjoint alternative scenario approach is a research
technique used to measure the trade-offs individuals make in choosing between
products and service providers. It was first developed in the marketing sector and
has been largely used to predict consumers’ choices for future products and services,
and now it is a well-established procedure in transport studies.In particular, given the need to avoid presenting options which are irrelevant for
the respondent, we discard traditional stated preference techniques in favour of the
adaptive stated preference (ASP). This interactive data collection technique amends
attribute levels during the experiment on the basis of the choices the respondent
makes. One significant advantage of this method is that it makes it possible to cope
with a wide range of ‘‘situations’’ which are comparable with the real world known
by the respondent, through its capacity to adapt to ‘‘personal’’ contexts (type of
commodity, time variance of attribute valuation, etc.).The ASP experiment starts from an existing freight transport option chosen by the
respondent (usually the company manager). It is the ‘‘typical’’ transport of the firm.
Starting from this option, the ASP exercise implies asking the respondent to rate
various hypothetical alternatives for performing the same transport task expressed
in terms of the relevant attributes.On the basis of the relevant literature four attributes are identified as most
significant in defining the transport service [30]:
. price (P), i.e. out-of-pocket cost of transport, including loading and unloading;
. time (T), i.e. door-to-door transit time, including loading and unloading;
. reliability (R), i.e. as% of deliveries as scheduled;
. frequency (F), i.e. as% of service per week offered by the carrier.
To our knowledge, this is the first ASP experiment performed with the scope
of determining the preferences of operators in terms of service attributes of sea
transport and of studying the potential reallocation of traffic from surface transport
services to maritime ro-ro services.
Monetary values of transport service attributes 161
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
2.2. The sampleThe present analysis has been carried out on a sample of freight-forwarders. It is, in
fact, increasingly common, especially for medium–long distance transfers, to
delegate the decision on the mode to be used outside the firm to third parties.
Choosing freight-forwarding agencies makes it possible to capture information from
a sector of the industry which accounts, on average, for more than half of the
transport decisions, as outsourcing of transport operations is spreading rapidly [31].
Recent studies have demonstrated that freight forwarders are becoming ‘‘one-stop
shop’’ specialist companies: this is part of a process that has led to the blurring
of boundaries between what were formerly distinct activities. There is a growing
body of evidence showing that ‘‘freight forwarders, from the perspective of the
shipper, assume the role of the carrier; from the point of view of the actual carrier;
they assumes the role of the shipper’’ [29, p. 1].Furthermore, the focus on freight forwarders leads to a sample which, although
small, is homogeneous with regard to respondents’ activity. Given the limited
resources available, and in the light of extending the experiment, selecting producers
would have limited the scope of the analysis to a specific productive sector or
would have excessively constrained the dimension of the dataset for each industrial
sector [32]. On the other hand, choosing transporters, given the current situation of
the Italian surface transport industry, would have probably led to interpretation
problems due to the resistance of small operators to intermodal transport [33].All in all, selecting freight-forwarding agents instead of producers on the one hand
allows insights to be gained from a wider spectrum of possible uses, and, on the other
hand, to gather a set of information on the subject of who is really behind the
decision-making process in transport attribute choices. Although the objective
function of the freight forwarder would necessarily differ from that of the producer,
it could reasonably be argued that, given the recent evolution of the market and of
the contractual agreements in force, once the organization of the transport service
has been outsourced, the real (final) decision maker, the shipper, would be the freight
forwarder herself/himself. S/he would be the residual claimant to any cost-quality
advantages obtained.Also, in line with the scope of our investigation, we have restricted the interviewed
sample to those freight forwarders who have a certain familiarity with the maritime
mode (must have bought maritime services at least once during the previous year)
and, given the purposes of this study, we have focused the empirical application on
a specific geographical context.In particular, we have analysed the preferences of freight forwarder localized in
south Italian regions with respect to the possibility of accessing maritime ro-ro
services from the port of Bari, Brindisi or Taranto. In order to present the
participating freight forwarder with comparable alternatives, we have considered
traffic-flows between origin–destination areas which are reachable both by sea and
by land.The data collection has been carried out in two phases. First a revealed preference
survey was carried out. A questionnaire was distributed to the freight forwarders to
collect information for designing the ‘‘typical’’ transport performed by the company
to be used for the ASP experiment.The second phase of the study, which followed a thorough pre-test of all
instruments, focused on the subset of freight-forwarder that accepted to conduct the
162 A. S. Bergantino and S. Bolis
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
study. It consisted in an interactive conjoint analysis interview carried out with themanagers of the companies participating to the study.
2.3. The experimentThe ASP experiment was carried out with the support of a portable computer andof the Leeds adaptive stated preference software which presented a consistent,on-screen, series of scenarios adapting to the respondents’ choices [34]. Theinterviewing process is the following. On the first screen the respondent is asked toconfirm the information on the ‘‘typical’’ transport operation performed by thecompany (acquired through the revealed preference survey). The information is thenused to customize the ‘‘current choice’’ of the respondent which becomes the‘‘reference option’’ and does not change for the whole experiment. The currentchoice, which consists of a value for each of the four service attributes identified,is reported at each iteration on the left-hand side of the screen, column A, and isautomatically assigned a rating of 100. It is assumed that among the existingalternatives, this is the preferred one, and thus it represents the operator’s currentutility level.
From the second screen and for each subsequent iteration, two more options—Band C—appear next to column A. They report hypothetical alternatives which areautomatically generated by the software and which are characterized by differingvalues of their service attributes. Column B always refers to the same mode oftransport of the ‘‘typical’’ transport defined by the respondent (column A) whilecolumn C refers to a different mode of transport. The alternative mode of transportwe propose across all experiments is always ‘‘maritime ro-ro service’’. In ourexperiment, thus, the mode is not just another service attribute and its estimatedvalue allows us to verify whether there would be an a priori preclusion for maritimetransport.
The value of the four service attributes of each alternative are determined asfollows:
. The first time the alternatives are presented (second screen), the information istaken on the basis of the known characteristics of the firm’s original transportservice in terms of percentages (e.g.% discount or increase in price,% ofshipments currently arriving on time, etc.).
. For the subsequent iterations, on the basis of the choices reported each timeby the respondent.
In every repetition of the experiment, the hypothetical alternatives presentedin column B and C thus change: new computer generated alternatives are presentedand the respondent is asked to rank the two alternatives against option A on thebasis of the value he/she assigns to the ‘‘new’’ service.
In choosing the rating, the respondent has to use a value scale carefully illustratedby the interviewer [35]. This ranges between 0 and 200, with 100 being thereference value (the current choice). The iterations continue until, for each variable inturn—starting with price—indifference is reached. In other words, once variationsin prices as a function of the rating given by the respondent in the previousiteration do not lead to a variation in the rating, the new screen presents options inwhich the remaining attributes change values following the same procedures.The process continues until convergence is found for all attributes or at the 20thiteration [36].
Monetary values of transport service attributes 163
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
3. Data analysis and estimation methodology
The database collected is made up of 566 observations, an average of 35 observationsper respondent [37]. Table 1 contains the main characteristics of the data collectedwith regards to the ‘‘typical’’ transport services described by each respondent andused as the benchmark for the experiment (column A).
The shipment generally carried out by the ‘‘average’’ company participatingin the pilot study lasts two days, it is relatively frequent (every three days), andit is delivered at the expected time 85% of the time and costs about 1.3 eurosper kilometre. The data collected seem to be relatively coherent across the16 cases.
Table 2 shows the mean and the median values of the hypothetical offerspresented to the 16 respondents. Interestingly, although both mean and medianvalues of the variables Time, Reliability and Frequency are substantially equal orbelow the values of the current option, the mean value of the rating is always above100 (the rating of the reference alternative, the ‘‘typical’’ transport): the shippersalways prefer the new services having, on average, a rating of 107.20. The savings incost more than compensate for the reduction in the other attributes. Furthermore,there is no mode-specific preclusion. The ro-ro alternative is, on average, chosen50% of the times.
The mean and the median of the difference between the value for each attributeof current service and the hypothetical alternative are shown in table 3. Across all
Table 1. ‘‘Typical’’ transport (average values).
VariableMeasurement
unit Mean Min. Max.
Price (euro) 2,000 495 6,000Time (hours) 60 50 90Reliability (%) 85 50 100Frequency (times �month) 10 8 24Mean length (km) 1,573 800 2,600
Table 3. Hypothetical offers: mean values of the difference in service attributes.
Diff_Cost(index) Diff_Time Diff_Reliability Diff_Freq
Mean 42.53 �2.21 4.97 2.01Median 50.00 0.00 0.00 1.00
Number of observations: 566.
Table 2. Hypothetical offers: mean values of service attributes.
Cost (euro) Cost (index) Time Reliability Freq Ro-ro Rating
Mean 1149.38 57.52 62.21 80.03 7.99 0.49 107.20Median 864.00 53.00 56.00 85.00 4.00 0.00 107.50
Number of observations: 566.
164 A. S. Bergantino and S. Bolis
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
experiments, the hypothetical services offered a mean discount of more than 40%,
a variation in travel time of about two hours, a mean decrease in reliability of 5%
and, finally, a mean reduction of frequency corresponding to a service supplied
about two times less per month.Figure 1 reports the rating sequence, ordered from the smallest rating to the
largest, of the respondents. It can be seen that the options presented have been
considered unacceptable 37 times (values between 1 and 10), while the status quo has
been reached 74 times (ratings equal to 100). About 61% of the time respondents
are either satisfied with the prospected alternative (48%) or indifferent between the
current alternative and the new service offered (13%). Considering the status quo
bias, in general respondents are willing to switch service.From figure 2, where the ratings are ordered per firm, it is possible to see that the
zero values are concentrated in the first, sixth and eighth experiments.
−50
0
50
100
150
200
250
1 19 37 55 73 91 109 127 145 163 181 199 217 235 253 271 289 307 325 343 361 379 397 415 433 451 469 487 505 523 541 559
Figure 1. Ordered sequence of ratings.
−50
0
50
100
150
200
250
1 49 73 97 121 145 169 193 217 241 265 289 313 337 361 385 409 433 457 481 505 529 55325
Figure 2. Ratings per firm.
Monetary values of transport service attributes 165
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
Analysing the raw data, it is possible to see that for firm one these values arerelated to changes in the level of reliability: the respondent considered the levelsof reliability of the hypothetical offers to be too low, notwithstanding anycompensatory decrease in price. The respondent considered reliability an essentialattribute of the transport service and any alternative transport service which impliedsignificant changes in the level currently guaranteed represented, for her, anon-viable option. For firm six and eight, variations in quality attributes did notcompensate adequately for cost variations.
Figure 3 reports, only for illustrative purposes, the cleaned database (excludingobservations from firm 6 and 8 and part of those of firm 1). As it can be seen, theremaining ratings indicate that convergence was generally found relatively easily.
4. Empirical results
The procedure chosen to estimate the empirical model is the Tobit ML estimator[2–7]. The dataset, in fact, contains a number of zero values corresponding to thosealternatives which, given the value of their attributes, have received a rating of zero.Since we can assume that those zero values correspond, in principle, to cases in whichthe latent variable—the indirect utility—might take negative values (i.e. unacceptablelevels of reliability which would compromise the respondent activity), we can treatthe zeros as a result of censoring and non-observability and thus apply the Tobitestimator. Weighting procedures have been carried out in order to account for thefact that the respondent might be more precise in her valuation when choosinga rating near 100: respondents would know whether the rating should be 95 asopposed to 105 much better than they would in when the choice is between 20 and 25(both completely unacceptable) [38].
4.1. Case by case approachIn order to avoid estimation problems linked to the well-known problem of repeatedand non independent observations, the estimation has been carried out separately foreach company [23]. The results of the estimations are shown in table 4.
All coefficients (�i) refer to the effect of a change in the respective variable (i) onthe respondents’ utility (transformation of the rating). The coefficients of cost andtime are expected to have a negative sign, those of reliability and frequency shouldshow a positive sign. Intuitively, in fact, an increase in cost and time associatedwith a transport alternative generates a decrease in the respondents’ utility, while an
0
50
100
150
200
250
1 16 31 46 61 76 91 106 121 136 151 166 181 196 211 226 241 256 271 286 301 316 331 346 361 376 391 406 421 436 451 466
Figure 3. Ratings per firm (selected observations).
166 A. S. Bergantino and S. Bolis
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
increase in frequency and reliability have, in general, a positive impact on overallutility. These exceptions find, in general, confirmation in the estimated coefficients.Exceptions are the time coefficient of firm 12—which, however, is significant onlyat the 10% level—and the reliability coefficients for firms 11 and 15. In the lattercases, however, the average level of the variable is negative.
The performance of the model is acceptable for most firms, given the limitednumber of observations. On average the variables explain 28% of the variation inutility and, thus, of the decision to change mode. Cost and reliability have, ingeneral, a coefficient generally significantly different from zero. In particular, thecoefficient of the variable cost is significant at least at the 10% level with anexpected negative coefficient in 12 cases out of 16, while the reliability coefficient issignificantly different from zero in 13 cases out of 16.
Given the focus of our study, of particular relevance is the dummy ro-ro, whichshould pick up the valuation of the willingness to use themaritimemode. Although thecoefficient takes quite differing values among the various case studies, it is generallynot significantly different from zero. Except for case 12, for which the parameter ispositive and significantly different from zero. It is possible to infer that there is no apriori reluctance of the respondents to use ro-ro services. In particular, freightforwarders in the pilot study do not seem to have strong preferences either way.
From table 4 it can be noted also that, in general, the coefficients have very lowvalues for all the variables and for all case studies. This implies that the marginalimpact of a change in a variable on the propensity to change from the currentsolution to a hypothetical one is small. The respective elasticity’s would thus be smallas well [19].
In table 5 we report the monetary valuations of tradeoffs (MVT) betweenattributes. For each attribute (i) the values are obtained as the ratio of its parameterestimates (�i) to the cost parameter estimate (�c).
MVTic ¼�i�c
ð1Þ
Table 4. Estimation results on ASP data—disaggregated data.
Firm Diffindex sig. Difftime sig. Diffreliab sig. Diffreq sig. RORO sig. Adj-Sigma
1 �0.0072 �� �0.0195 0.0532 ��� 0.0466 � 0.0232 32.7%2 �0.0157 ��� �0.0158 0.0073 �0.0041 �0.0584 26.2%3 �0.0103 ��� �0.0738 �� 0.0802 � 0.0944 �� �0.0447 33.3%4 �0.0211 ��� �0.0113 �0.0096 0.1955 ��� 0.0026 26.8%5 �0.0128 ��� 0.0007 0.0081 0.1960 ��� �0.0537 37.2%6 �0.0116 ��� �0.0308 0.0008 0.3222 ��� �0.0136 35.3%7 �0.0142 ��� �0.1076 ��� 0.0680 ��� 0.1878 ��� 0.0220 35.5%8 �0.0113 ��� �0.0433 ��� 0.0843 0.4699 �0.0252 33.7%9 �0.0011 �0.0096 �0.0030 0.1869 �0.0332 21.8%
10 �0.0016 �0.0039 �0.0078 0.2283 ��� �0.0084 18.5%11 �0.0146 �� 0.0020 �0.0220 � 0.5095 ��� �0.1580 24.4%12 �0.0121 ��� 0.0409 � �0.0031 1.5404 ��� 0.4750 �� 28.1%13 �0.0010 �0.0077 �� 0.0100 � 0.1834 ��� 0.0251 21.4%14 �0.0173 ��� �0.0046 0.0340 0.8224 ��� 0.2287 30.7%15 �0.0043 �0.0009 �0.0521 ��� 0.7367 ��� �0.0120 18.9%16 �0.0071 ��� �0.0035 �0.0258 0.3039 ��� 0.1640 15.4%
Note: Asterisks represent levels of significance: ***¼ 1%, **¼ 5% and *¼ 10%.
Monetary values of transport service attributes 167
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
The significance levels are reported next to the values. The relevant estimations
are those relative to coefficients which are both significantly different from zero.
Their values are in bold. The corrected average is calculated excluding values which
are not significantly different from zero.Each column of table 5 reports the amount of money that the respondent would
be willing to pay (in case of a positive value) or to receive as compensation (in case
of a negative value) for a one-unit variation in the specific service attribute (MVT).
The ratio of the service attributes to the cost coefficient yields, in fact, the monetary
values of an attribute at the margin and hence gives an idea of how changes in
attributes are traded off against a monetary change in transport costs. In the case of
time this is the value of time (VOT), in the case of reliability and frequency this is
value of reliability (VOR) and of frequency (VOF), respectively. As can be seen, an
hour reduction of journey time is on average valued 4 euros per ton, while a 1%
reduction in reliability would require a compensation of almost 5 euros per ton.
A one step reduction in the frequency supplied would require more than 33 euros
per ton [39].While the values of both VOR and VOF are relatively high for most cases, in
general the VOT is comparatively low. Overall, it seems that, for the sample analysed,
frequency is the most precious attribute of the service required: this is true for the last
eight cases. For these operators, the willingness to pay for an increase of frequency is
significantly greater than for changes in any of the other variables. In particular, in
case 12, the willingness to pay for frequency is significant and particularly large.Although the sample considered is extremely small and not representative of the
category, it is interesting to note that, as expected, freight forwarders tend to give
a higher value to factors which enlarge their freedom of choice and the regularity
Table 5. Trade-off ratios of transport service attributes to cost (absolute
values—in euro per ton).
Firm VOT sig. VOR sig. VOF sig.
1 2.72 7.44 ��� �6.52 �
2 1.01 0.46 0.263 7.16 �� 7.78 � �9.15 ��
4 0.54 0.45 �9.26 ���
5 �0.06 0.63 �15.29 ���
6 2.65 �0.07 �27.69 ���
7 7.56 ��� �4.78 ��� �13.18 ���
8 3.84 ��� �7.48 �41.679 8.76 2.71 �170.4510 2.46 4.88 �142.39 ���
11 0.14 1.50 � �34.81 ���
12 3.38 � 0.26 127.41 ���
13 7.37 �� 9.65 � �176.49 ���
14 0.27 1.97 �47.60 ���
15 0.21 12.19 ��� �172.45 ���
16 0.49 3.64 �42.89 ���
Corr_averagea 3.79 Z4.6 Z33.4
aThe corrected average includes only the values of the trade-off relative tocoefficients which are significantly different from zero, at least at the 10%confidence level.Note: Asterisks represent levels of significance: ***¼ 1%, **¼ 5% and *¼ 10%.
168 A. S. Bergantino and S. Bolis
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
of service than to those elements, like time of journey, which are more easily takeninto account in planning their activity.
4.2. Overall sample: an indicationFor sake of exposition, taking into account the problems of heteroskedasticityimplicit in the pooled database, we report in table 6 the results of the regressionestimated on the 566 observations together.
Notwithstanding the issues of aggregation of preferences across the sample, it isinteresting to note, from the results reported in table 7, that the relative ranking ofthe trade-offs is fully respected, with frequency and reliability remaining the mostrelevant factors in the decision. As it has been the case with the case-by-caseestimation, no a priori conditioning versus ro-ro is picked up by the estimation.
4.3. Focus on ro-ro servicesIn order to test any mode-specific differences in the values of attributes, the samplehas been segmented according to the rating assigned to the ro-ro service option.Considering only the observations corresponding to the highest rating beingassigned to ro-ro services, we have re-estimated the model. The results are reportedin table 8.
It is interesting to see that; in general, the importance assigned to the costdifference is greater, implying a greater awareness towards the price of services forthose operators choosing ro-ro service. Although only the value of the coefficientsrelating to cost and frequency are significantly different from zero, the results arequite interesting in comparative terms. From table 9, it can be seen that frequencyof service is confirmed to be the most relevant parameter in operators’ preferences.A reduction in frequency of service is evaluated, for operators choosing the ro-ro
Table 8. Estimated results on ASP data—pooled sample, ro-ro only.
Firm Diffindex sig. Difftime Diffreliab Diffreq sig. Sigma
All �0.012 �� �0.0001 0.039 0.097 �� 21.53%
Note: Asterisks represent levels of significance: **¼ 5%.
Table 7. Trade-off ratios for pooled sample (absolute values—in euro
per ton).
Firm VOT VOR sig. VOF sig.
All �0.04 �4.64 ��� �6.02 ���
Note: Asterisks represent levels of significance: ***¼ 1%.
Table 6. Estimation results on ASP data—pooled sample.
Firm Diffindex sig. Difftime sig. Diffreliab sig. Diffreq sig. Ro-ro Sigma
All �0.0040 ��� 0.0002 0.0185 ��� 0.0240 ��� 0.0019 43.87%
Note: Asterisks represent levels of significance: ***¼ 1%.
Monetary values of transport service attributes 169
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
option, about 8 euro per ton, while both the VOT and the VOR are not significantlydifferent from zero.
A comparison with the results reported in table 7 highlights the greaterimportance that operators place on frequency of service when using ro-ro servicesinstead than surface transport. These considerations should lead to someconsideration when new ro-ro services are set up or when decisions are taken asto the size of the fleet and its capacity.
4.4. A brief cross-studies comparisonAlthough a comparison between the absolute values assigned to the single serviceattributes by the respondents of different studies might be quite difficult to carry outgiven both the different approaches used and the different sectors analysed,a comparison of the relative ranking yields interesting insights. Comparing ourresults with those obtained by other studies [21, 23, 29], some common traits, in fact,emerge. In particular, there seems to be agreement in assigning the lowest value to thevariable time among the three main service attributes considered. Freight forwarders,just as producers, are more concerned with reliability and frequency than with theduration of the trip. However, in our study, in contrast with the other studies, therelative importance of these two attributes is reversed, with the former assuming agreater weight. It appears that when ro-ro alternatives are presented, availability ofservice becomes a strong decisional factor for firms which have to abandon thecurrently selected mode of transport in favour of the new alternative. Moreover, beinga freight forwarder an intermediary in the transport process, she/he sees thepossibility of responding to the needs of the client by granting the availability of theservices when required as a relevant aspect of the quality of the service supplied. Inthis context, reliability assumes a relevant but secondary role: a decrease in reliabilityis easily taken into account by price changes and by contractual agreements.
5. Concluding remarks
In this paper we have presented some evidence on the monetary values assigned toservice attributes by a sample of 16 freight-forwarder firms located in south Italy.The primary objective of analysing operators’ preferences when switching fromcurrent on-land transport services to a hypothetical maritime ro-ro alternative hasbeen that of testing for mode-specific differences in the estimated values. Secondaryobjectives, although not less important, have been to obtain a preliminary ratingof the transport attributes included in the stated preference experiment and a firston-the-field test of the soundness of the selection carried out with respect to theanalysis of the maritime ro-ro context. Finally, the analysis has been aimed at testingthe appropriateness of selecting as respondent’s freight forwarders in their vest oftransport service intermediaries.
Table 9. Estimated results on ASP data—pooled sample, ro-ro only.
Firm VOT VOR VOF sig.
All 0.008 �3.23 �8.08 ��
Note: Asterisks represent levels of significance: **¼ 5%.
170 A. S. Bergantino and S. Bolis
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
Freight forwarders have been preferred to producers in order to gain insights froma wider spectrum of possible uses and, at the same time, to verify weather thepreferences of this set of users would differ widely from those of the producers.Although the objective function of the freight forwarder would necessarily differfrom that of the producer, the recent evolution of the market and of the contractualagreements in force would generally place the freight forwarder in a position to bethe residual claimant to any cost–quality advantages obtained. In any case, this istrue for the sample interviewed, given the structure of contractual relationshipsdescribed in the first part of the interview.
Although the results presented in this paper relate only to a very limited group ofpotential users with obvious consequences on sample representativeness, somepreliminary considerations on the outcome can be drawn, waiting for the outcomesof a follow-up research project with larger geographical scope (a number ofinterviews have already been carried out in other Italian regions). Overall, in fact,initial evidence is encouraging and offers some understanding of the determinants ofthe maritime transport choice.
First of all, there seems to be no a priori preclusion for the maritime alternativeand, in particular, for the ro-ro services. The values of the service attributes aregenerally consistent: freight rates are not the main determinant of modal choice norare the time of travel. In choosing the transport service operators tend to rank otherfactors more highly. In line with the results of other studies, reliability and frequencyof service are the two most important parameters. This is the more so for themaritime alternative. For the latter, frequency seems to be in absolute terms the mostrelevant parameter with valuations being up to almost three times higher than forreliability. Considering the estimations obtained through the case by case approach,which are certainly sounder than the ones carried out on the pooled database, therelative importance of frequency is strengthen: the VOF is more than 7 times higherthan VOR.
All in all, according to our estimation—which, however, due to the limitedextension of our database should be taken with the appropriate precaution—freightforwarders seem to value a 1% improvement in reliability at about 5 euro per tonwhile a variation in frequency at about 33 euro. A reduction in one hour travel timereceives an average valuation of less than 4 euro per ton. If the results are limited tothose cases when the respondent selected the ro-ro option, the results point evenmore strongly to the relative importance of frequency of service, compared to theother service characteristics in influencing the choice of operators.
Acknowledgements
The authors are indebted to Michel Beuthe (Mons) and Romeo Danielis (Trieste) fortheir helpful comments on an earlier version of this paper. The authors would like tothank the participants to the WCTR conference held in Berkley in June 2007, inparticular Enrico Musso and Eddy Van de Voorde, and the participants to the SIETheld in Naples, in October 2007. The usual disclaimer applies for all remainingerrors. While the paper can be considered the outcome of joint work resulting froma line of research that the authors are carrying out together, systematically,since 2004, Angela S. Bergantino is mainly responsible for sections 2.2, 3, 4.1–4.3while Simona Bolis is mainly responsible for sections 2.1, 2.2 and 4.4. The remainingparts are joint work.
Monetary values of transport service attributes 171
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
References and notes1. This paper is based on an on-going research project that the authors are carrying out,
systematically, since 2002. For more details of previous studies see [2–7].2. BERGANTINO, A. S. and BOLIS, S., 2002,Motorways of the Sea: Is this a Real Alternative to
All-land Transport? Some Preliminary valuations on the North Tyrrenian Sea Routes,Mimeo, Paper presented to the 2nd STRC Conference, Ascona, Switzerland, 2–22March.
3. BERGANTINO, A. S. and BOLIS, S., 2004, An Adaptive Conjoint Analysis of Freight ServiceAlternatives: Evaluating the Maritime Option, Working Paper No. 04-10, Facolta discienze economiche, Universita della Svizzera Italiana – USI, 1–18.
4. BERGANTINO, A. S. and BOLIS, S., 2004, An analysis of maritime ro-ro freight transportservice attributes through adaptive stated preference: an application to a sample offreight forwarders. European Transport, Special Issues: Freight Transport Analysis andIntermodality, 25/26, 33–51.
5. BERGANTINO, A. S. and BOLIS, S., 2005, An adaptive conjoint analysis of freight servicealternatives: evaluating the maritime option, capitolo 10. In: Methods and Models inTransport and Telecommunications: Cross-Atlantic Perspectives, edited by A. Reggianiand L. Schintler (Berlin: Springer-Verlag), pp. 181–198.
6. BERGANTINO, A. S., BOLIS, S. and CANALI, C., 2006, A methodological framework toanalyse the market opportunities of short sea shipping: the adaptive stated preferenceapproach. In: Towards Better Performing European Transportation Systems, edited byB. Jourquin, K. Westin and P. Rietveld (London: T&F Routledge), pp. 285–304.
7. BERGANTINO, A. S. and BOLIS, S., 2006, La domanda di servizi di short sea shipping:un’indagine sulle preferenze di un campione di operatori della logistica dell’Italianord-occidentale. In: I trasporti e l’Europa. Politiche, infrastrutture, concorrenza, editedby E. Musso, E. Marcucci and G. Polidori (Milan: FrancoAngeli), pp. 336–355.
8. BERGANTINO, A. S., 2007, Il valore dei servizi di trasporto: un confronto tra il nord e ilsud del Paese. Rassegna Economica – Rivista Internazionale di Economia e Territorio,LXX, 63–80.
9. Application of revealed preference methods based on observed behaviour is, generally,not feasible since the data on real choices are usually commercially very sensitive andhence are not usually disclosed (in a liberalized environment, freight rates areindividually negotiated and held commercially confidential) and the complexity of thefreight transport decision requires the collection of large dataset on a number ofvariables and the observation of a great number of firms’ decisions in order to take intoaccount the heterogeneity of the context. Revealed preference datasets, in fact, arebased on the observation of actual choices; need a large number of observations; mayinclude only existing alternatives; and require the choice set to be defined and the level ofservice information for the discarded option to be calculated. Moreover, importantcharacterizing variables (such as, for instance, time and cost) are often correlated and,due to possible measurement error, there might be bias in forecasting. Finally, the limiteduse of the maritime alternative, especially for certain routes and products, hampersfurther the utility of revealed preferences in this context. Stated preference data, onthe other hand, overcome these problems, although questionnaire design and choice ofthe relevant attributes plays a major role in their efficacy. It has the advantage, withrespect to standard revealed preference approaches, of allowing analysis in contexts inwhich it is not possible to ‘‘observe’’ the real behaviour of operators either for lack ofdata or because the alternative to be analysed is not yet used or available for use.
10. BEN AKIVA, M. and MORIKAWA, T., 1990, Estimation of switching models fromrevealed preferences and stated intentions. Transportation Research—Part A, 22,485–495.
11. SWAIT, J., LOUVIERE, J. and WILLIAMS, M., 1994, A sequential approach to exploiting thecombined strengths of SP and RP data: application to freight shipper choice.Transportation, 21, 135–152.
12. ADAMOWICZM, W., SWAITM, J., BOXALL, P., LOUVIERE, J. and WILLIAMS, M., 1997,Perceptions versus objective measures of environmental quality in combined revealedand stated preferences models of environmental valuation. Journal of EnvironmentalEconomics and Management, 32, 65–84.
172 A. S. Bergantino and S. Bolis
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
13. BRADLEY, M. and DALY, A., 1997, Estimation of logit choice models using mixed statedpreference and revealed preference information. In: Understanding Travel Behaviour inan Era of Change, edited by P. Stopher and M. Lee-Gosselin (London: Pergamon Press).
14. STOPHER, P., 1998, A review of separate and joint strategies for the use of data onrevealed and stated choices. Transportation, 25, 187–205.
15. WARDMAN, M., 1998, A comparison of revealed preference and stated preference modelsof travel behaviour. Journal of Transport Economics and Policy, 22, 71–91.
16. BROWNSTON, D., BUNCH, D. and TRAIN, K., 1999, Joint mixed logit models of statedand revealed preferences for alternative fuel vehicles. Transportation Research—Part B,34, 315–338.
17. LOUVIERE, J., HENSHER, D. and SWAIT, J., 2000, Stated Choice Methods (Cambridge:Cambridge University Press).
18. TRAIN, K. and WILSON, W., 2006, Econometric Analysis of Stated-preference ExperimentsConstructed from Revealed-preference Choices, Mimeo, March.
19. BOLIS, S. and MAGGI, R., 1999, Adaptive stated preference analysis of shippers’ transportand logistics choice. In: Proceedings of the 8th World Conference of Transport Research(The Netherlands: Pergamon, Elsevier Science).
20. BOLIS, S. and MAGGI, R., 2002, Stated preference—evidence on shippers’ transport andlogistics choice. In: Domanda di Trasporto Merci e Preferenze Dichiarate, edited byR. Danielis (Milan: FrancoAngeli).
21. DANIELIS, R. and ROTARIS, L., 2002, Characteristics of freight transport demand in theFriuli Venezia Giulia region: a summary. In: Domanda di Trasporto Merci e PreferenzeDichiarate, edited by R. Danielis (Milan: FrancoAngeli).
22. DANIELIS, R. (ed.), 2002, Domanda di trasporto merci e preferenze dichiarate (Milan:FrancoAngeli).
23. MAIER, G. and BERGMAN, E. M., 2002, Conjoint analysis of transport options in Austrianregions and industrial clusters. In: Domanda di Trasporto Merci e Preferenze Dichiarate,edited by R. Danielis (Milan: FrancoAngeli).
24. FOWKES, A. S. and TWEEDLE, G., 1996, Modelling and Forecasting Freight TransportDemand, Mimeo, ITS-University of Leeds.
25. FOWKES, A. S. and TWEDDLE, G., 1997, Validation of Stated Preference Forecasting:A Case Study Involving Anglo-continental Freight. European Transport Forum – 25thAnnual Meeting, Proceedings of Seminar F (London: PTRC).
26. DE JONG, G., 2000, Value of freight travel-time savings. In: Handbook of TransportModeling, edited by D. A. Hensher and K. J. Button (London: Pergamon).
27. FOWKES, T. and SHINGHAL, N., 2002, The Leeds Adaptive Stated Preference Methodology,Institute of Transport Studies, University of Leeds, Working Paper, 558.
28. BOLIS, S. and MAGGI, R., 2003, Logistic strategy and transport service choices. Anadaptive stated preferences experiment. In: Growth and Change—A Journal of Urban andRegional Policy, Special Issue STELLA, 34(4), 492–504.
29. Unescap, 2002, Major Issues in Transport, Communications, Tourism and InfrastructureDevelopment: Developments in Multimodal Transport and Logistics (Bangkok: UnescapPublications).
30. Extremely interesting is the ranking of the attributes most commonly used inanalysing transport demand and users’ preferences contained in the detailed surveyof CULLINANE, K. and TOY, N., 2000, Identifying influential attributes in freightroute/mode choice decisions: a content analysis. Transportation Research—Part E, 36,41–53.
31. Recent surveys on the evolution of the freight forwarders’ business and type of servicesoffered are contained in: KNP, 2002, Freight Forwarding Market Report 2002 (London:Key Note Publications Ltd and Unescap [29]).
32. A number of studies focusing on producers, has, often, only a few operators perindustrial sector. Given the peculiarities in the transport mode decision in relation to thetype of good, we consider it essential to be able to confront an independent operator.
33. Recent studies indicate that transport and logistics intermediaries have, in general, amore in-depth knowledge of possible alternatives. In particular, when producersexternalize the logistic and/or transport function they tend to be less concerned withthe actual characteristics of the transport service chosen as long as terms and conditions
Monetary values of transport service attributes 173
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
of the contract are respected. TSAMBOULAS, D. and KAPROS, S., 2000, The decisionmaking process in intermodal transport, Transportation Research Record no. 1707,Washington, DC.
34. For a detailed description of the data collection methodology see also [2–7].35. It is extremely important that the respondents rank options in their desired order, having
a clear understanding of the scaling, so to indicate as accurately as possible their strengthof preference.
36. Each iteration generates two responses, therefore, in the case that all 20 iterations arerun, we would obtain 40 observations by each respondent.
37. Each response, during the experiment, is interpreted as a separate observation.38. The weighting technique is the same proposed and adopted by Fowkes and Shinghal [27].39. The reduction in frequency of services varies between twice daily (upper value) and once
every two weeks (lower value).
174 A. S. Bergantino and S. Bolis
Dow
nloa
ded
by [
Dal
hous
ie U
nive
rsity
] at
14:
49 2
3 Se
ptem
ber
2013
