Access the full text.
Sign up today, get DeepDyve free for 14 days.
M. Razzaque, C. Sheng (1998)
Outsourcing of logistics functions: a literature surveyInternational Journal of Physical Distribution & Logistics Management, 28
F. Cruijssen (2006)
Horizontal cooperation in transport and logistics
S. Mantel, Mohan Tatikonda, Y. Liao (2006)
A behavioral study of supply manager decision-making: Factors influencing make versus buy evaluationJournal of Operations Management, 24
L. Visser, C. Ruijgrok (2007)
Measuring the thresholds in decision making on outsourcing
M. Tsai, C. Wen, Chiang-Shin Chen (2007)
Demand choices of high-tech industry for logistics service providers—an empirical case of an offshore science park in TaiwanIndustrial Marketing Management, 36
Colin Camerer (2003)
Behavioral Game Theory: Experiments in Strategic Interaction
Konstantinos Selviaridis, M. Spring (2007)
Third party logistics : a literature review and research agendaThe International Journal of Logistics Management, 18
B. Groothedde (2005)
Collaborative logistics and transportation networks : A modelling approach to hub network design
P. Green, V. Srinivasan (1978)
Conjoint Analysis in Consumer Research: Issues and OutlookJournal of Consumer Research, 5
P. Laarhoven, Magnus Berglund, M. Peters (2000)
Third‐party logistics in Europe – five years laterInternational Journal of Physical Distribution & Logistics Management, 30
M. Hannan (1984)
Structural Inertia and Organizational ChangeThe Sociological Review, 49
M. Brodbeck, H. Simon (1957)
Models of Man.Journal of the American Statistical Association, 53
Colin Camerer (1990)
Behavioral Game Theory
F. Budde, U. Felcht, Heiner Frankemölle (2008)
Today's Challenges and Strategic Choices
M. Kosfeld (1999)
Individual decision making and social interaction
R. Gulati, David Kletter (2005)
Shrinking Core, Expanding Periphery: The Relational Architecture of High-Performing OrganizationsCalifornia Management Review, 47
R. Lieb, K. Lieb (2005)
The North American Third-Party Logistics Industry in 2011: The Provider CEO PerspectiveTransportation Journal, 51
M. Ben-Akiva, M. Bradley, T. Morikawa, J. Benjamín, T. Novak, H. Oppewal, V. Rao (1994)
Combining revealed and stated preferences dataMarketing Letters, 5
C.M. Thompson, E. Pearson, L. Comrie, H. Hartley (1941)
Tables of Percentage Points of the Incomplete Beta-FunctionBiometrika, 32
M. Wardman (1988)
A COMPARISON OF REVEALED PREFERENCE AND STATED PREFERENCE MODELS OF TRAVEL BEHAVIOURJournal of Transport Economics and Policy, 22
D Kahneman, A Tversky (1979)
Prospect theory: an analysis of decision under riskEconometrica, 47
A. Falk, E. Fehr (2002)
Psychological Foundations of IncentivesERN: Behavioral Economics (Topic)
J. Mentzer, M. Myers, Mee-Shew Cheung (2004)
Global market segmentation for logistics servicesIndustrial Marketing Management, 33
J Ortuzar (2000)
Stated preference modelling techniques
G. Muilerman (2001)
Time-based logistics: An analysis of the relevance, causes and impacts
Academisch Proefschrift (2007)
Valuing Environmental Decay : Quantitative Policy-oriented Studies on Urban and Rural Environments
S. Min, A. Roath, P. Daugherty, S. Genchev, Haozhe Chen, Aaron Arndt, R. Richey (2005)
Supply chain collaboration: what's happening?The International Journal of Logistics Management, 16
L. Visser, W. Amstel (2008)
Supplier Involvement in Purchasing Logistics Services
R Lieb, BA Bentz (2005)
The North American third- party logistics industry in 2004: the provider CEO perspectiveInt J Phys Distrib, 35
R. Danielis, E. Marcucci, Lucia Rotaris (2005)
Logistics managers’ stated preferences for freight service attributesTransportation Research Part E-logistics and Transportation Review, 41
T. Simatupang, R. Sridharan (2002)
THE COLLABORATIVE SUPPLY CHAIN.The International Journal of Logistics Management, 13
C. Lindblom (1959)
THE SCIENCE OF MUDDLING THROUGHEmergence: Complexity and Organization, 19
E Bergkvist (2001)
Freight transportation: valuation of time and forecasting of flows, Umea economic studies 549
D. Kahneman, A. Tversky (1979)
Decision, probability, and utility: Prospect theory: An analysis of decision under risk
Michael Cohen, J. March, Johan Olsen (1972)
A Garbage Can Model of Organizational Choice.Administrative Science Quarterly, 17
H. Simon (1960)
The new science of management decision
C. Carter, Lutz Kaufmann, Alex Michel (2007)
Behavioral supply management: a taxonomy of judgment and decision‐making biasesInternational Journal of Physical Distribution & Logistics Management, 37
Mark Vijver (2009)
Collaboration in buyer-supplier relationships
A. McKinnon (2005)
Supply Chain Excellence in the European Chemical Industry
WJ Hopp (2004)
50th anniversary article: fifty years of management scienceManage Sci, 50
JJ Louvière (1988)
Conjoint analysis modelling of stated-preferences: a review of theory, methods, recent developments and external validityJ Transport Econom Policy, 2
M. Ben-Akiva, S. Lerman (1985)
Discrete Choice Analysis: Theory and Application to Travel Demand
M. Kouwenhoven, C. Rohr, Stephen Miller, H. Siemonsma, P. Burge, J. Laird (2007)
Isles of Scilly Travel Demand Study
Robert Mosch (2005)
The Economic Effects of Trust: Theory and Empirical Evidence
W. Hopp (2004)
Fifty Years of Management ScienceManag. Sci., 50
Dan Andersson, A. Norrman (2002)
Procurement of logistics services—a minutes work or a multi-year project?European Journal of Purchasing & Supply Management, 8
D. Pearmain, E. Kroes (1990)
STATED PREFERENCE TECHNIQUES: A GUIDE TO PRACTICE
Logist. Res. (2010) 2:165–176 DOI 10.1007/s12159-010-0036-3 OR IGINAL PAPER Lenny Visser Received: 14 September 2009 / Accepted: 28 September 2010 / Published online: 17 October 2010 Springer-Verlag 2010 Abstract Most existing models in supply chain manage- activities to partners in their network. This has led to an ment literature proving the potential of vertical logistics explosion of partnerships up and down the value chain. collaboration rely on the assumptions of neoclassical or new As a result, successful organizations discover that lever- institutional economic theory. These theories see individual aging relational capital is an important route to long-term decision makers as fully rational agents. Nevertheless, success [19]. reviewing the decision-making-behavior literature makes Logistics is one of the activities outsourced to specialized clear that individual decision makers also derive utility from providers. Simultaneously, the complexity of outsourcing behavioral factors and display inertia in their choices. This logistics services is increasing. Logistics services are paper takes up this issue and quantitatively measures the increasingly being purchased in bundles and, at the same impact of behavioral factors and inertia on vertical logistics time, more varied activities, such as value adding and IT collaboration decision between shipper and logistics service services, are included in such bundles of services [1, 4, 45]. provider. A stated preference experiment is used to reach our As a result, collaboration between logistics service providers research objective. The results of the experiment show that and shippers is becoming increasingly important. Both the behavioral factors like trust, confidentiality and com- parties need each other to decrease the total CO2 emission mitment significantly impact a logistics collaboration deci- by bundling transport flows and optimizing logistics sion. Next to this, the stated preference confirms that shippers networks [8]. leave beneficial collaboration opportunities unexploited The potential of logistics collaboration between shippers because of a certain level of inertia. It may be concluded that and LSPs, in terms of costs and service advantages, has both aspects function as a threshold to intensify collaboration been proved by several researchers [11, 17]. Nevertheless, between shipper and logistics service provider. in practice, barriers and resistance to change do exist, which prevents logistics collaboration from being initiated Keywords Collaboration Third-party logistics [31]. Shippers are reserved to increase the level of col- Decision making Stated preference laboration by transferring more responsibilities to a service provider [31]. Research shows that there are opportunities, in terms of cost and service advantages, when shippers 1 Behavioral research on logistics would be willing to share responsibilities at a tactical or strategic level [11, 17]. Nowadays, many companies are shrinking their core by Operations research models, which are used to prove the focusing on fewer activities while outsourcing non-core potential of logistics collaboration, have typically assumed that people act as fully rational agents. The impact of behavioral aspects is identified as a useful area for further L. Visser (&) research [9, 11]. Hopp [21] and Mantel et al. [28] also Department of Financial Management, conclude that the operations management literature has not Fontys University of Applied Sciences, always adequately addressed the human aspect of decision P.O. Box 347, 5600 AH Eindhoven, The Netherlands making. In addition, there are many contributions in the e-mail: l.visser@fontys.nl 123 166 Logist. Res. (2010) 2:165–176 supply chain management and logistics literature empha- ‘‘fully rational agents’’ who maximize their utility, are only sizing that such human factors as trust and commitment are guided by financial incentives and have complete knowl- important barriers to collaboration (e.g. [30, 39]), but these edge of all options and consequences [16]. Nevertheless, publications not quantify the impact of the human factors there is abundant evidence in the decision-making-behavior on logistics collaboration decisions. Quantifying and literature that decision makers often violate the rationalistic proving the significant impact of human factors will help to economic paradigm and that decisions lead to suboptimal significantly enhance the precision and rigor of the existing results [21, 22, 41]. As a result, the decision-making- models and their outcomes. As a result, understanding of behavior literature provides behavioral insights to signifi- logistics collaboration decisions will increase. Addition- cantly enrich the supply chain management literature. ally, this research will emphasize the importance of qual- In conformity with the neoclassical and new institutional itative factors in today’s business environment. A recent economics views of decision makers, the classical deci- study about the European logistics service providers mar- sion-making literature is based on the assumption that ket shows that LSPs undervalue the impact of these factors individuals are rational agents. This implies that individu- on a logistics collaboration decision [14]. als have well-defined and stable preferences, know and Historically, supply chain management and decision- understand all existing alternatives, and maximize their making-behavior literature are studied by separate preferences given the existing possibilities, irrespective of communities of scholars. In practice, the two fields are the complexity of the decision and its context [23]. This is intimately tied to one another. Therefore, we adopt the known as the ‘‘rational theory of decision making’’. The approach of Mantel et al. [28] that an integration of supply assumption of complete rationality in this theory is open to chain management and decision-making-behavior litera- several points of criticism. First, Nobel laureate Herbert ture can contribute to closing the research gap in under- Simon argued that real-world decision makers have limited standing collaboration decisions. The present research computational capabilities in searching for and evaluating attempts to quantitatively measure the impact of behavioral optimal alternatives. Simon introduced the model of factors on shipper’s decision to intensify collaboration with bounded rationality [40], a perspective in which actual a logistics service provider. In order to achieve the research choice behavior is brought by a sequence of heuristics or objective, a stated preference experiment is designed and rules of thumb that boundedly rational agents employ to conducted. The remainder of the paper is structured as make judgments and decisions in the (complex) real world. follows. The first section briefly reviews the relevant lit- Second, the assumption of full certainty was also ques- erature on decision-making behavior. The research meth- tioned, and more intuitive-based decision-making models odology and design of the stated preference experiment are were developed for situations that are not characterized by complete certainty [10, 26, 41]. Third, researchers have discussed. Subsequently, the results of the experiment are presented. The final section concludes with a summary of demonstrated that (economic) behavior is not only guided the main results and some suggested avenues for further by material incentives but also by psychological and research. sociological influences [7, 15]. Individuals also derive utility from equality, honesty, reciprocity and so on. This approach offers new views on logistics collaboration 2 Literature review on decision-making behavior that go beyond the notion that collaboration only results from calculations of expected material payoffs [32]. Over the past few years, a number of articles have dealt with Finally, following the classic economic approach, it may be collaboration decisions in different domains, including stated that a decision maker never remain a profitable sociology, psychology, marketing and supply chain man- opportunity unexploited independent of how small the gain agement [31, 46]. This paper discusses collaboration in the actually is. But in practice, individuals and organizations context of logistics and supply chain management. The lit- are usually reluctant to change, and in consequence, they erature in this area has tested a wide range of frameworks do not always respond to relative differences in a rational and optimizing models in order to demonstrate the potential manner [19, 20]. cost and service advantages of vertical logistics collabora- Based on the above literature review, we may conclude tion. Most of these models rely on the assumptions of neo- that the behavioral decision literature describes a number classical theory or of the new institutional economics. As of reasons that function as a threshold which must be indicated by Carter et al. [9], these theories have signifi- surmounted by a shipper in order to fully benefit from cantly enriched the a priori theoretical and grounded logistics collaboration. Summarized, these reasons are the frameworks that have been developed in the field of supply impact of non-rational aspects and the existence of inertia. chain management. However, the neoclassical and new A stated preference experiment is used to validate this institutional economic theories regard decision makers as a assumption empirically. 123 Logist. Res. (2010) 2:165–176 167 3 Methodology U ¼ b þ b X þ e j 0 k kj j where U is the overall utility for a particular alternative j; 3.1 Stated preference methodology b is the constant term; b represents the relative utility 0 k associated with attribute k (e.g. a specific LSP selection For our study, the stated preference (SP) methodology was criteria such as costs or service); X is the independent chosen, because SP techniques are a family of market kj variable representing attribute k for alternative j; and e is research tools that enable the analysis of the decision j an error component. It is assumed that the error terms are behavior of individual respondents, by proposing (hypo- thetical) alternatives [2, 43]. SP uses carefully constructed independently and identically distributed with a Weibull distribution [3]. interviews or questionnaires, in which respondents are asked to make choices between alternative product descriptions so 3.2 Attribute selection as to reveal how respondents value different attributes. SP results can be used to determine and quantify the rela- A standard template, as defined by Pearmain et al. [35] and tive importance of attributes that are of interest to Sheldon [38], is used to design our experiment. One of the the researcher. Alternatively, observations of respondent most important activities for an SP researcher is to decide behavior, revealed preferences (RP) research, could be used which attributes to be included in the experiment. to answer the research question. However, such a direct Researchers should be selective in adding attributes approach has some disadvantages. Firstly, validity issues because SP experiments with more than seven attributes arise because the researcher cannot be sure whether the respondent is incorporating additional variables in his become excessively complex and can confuse respondents. Therefore, we follow the design rules that recommend evaluation. Moreover, in general, RP research requires lar- ger sample sizes in order to develop efficient statistical selecting attributes and accompanying levels, based on prior qualitative research, consisting of a literature review models as with SP. Each SP interview produces multiple observations per individual because a respondent is asked to and case studies. During the literature review process, different organizational theories and the existing logistics consider a number of situations. Conversely, RP data most outsourcing and collaboration literature are reviewed. This often only result in a single observation per individual. review resulted in a list of variables that possibly influence Finally, SP enables a researcher to control more precisely the a logistics collaboration decision. Subsequently, seven case choices offered to a respondent. Thus, the effects of vari- studies are used to validate the results of the literature ables of interest can be isolated from those of other factors, review in the specific context of this research project [49]. and SP techniques can ensure that data are of sufficient quality to construct good-quality statistical models [35]. Both the literature review and case studies make clear that costs, service, trust, confidentiality and commitment are the During an SP experiment, the decision makers are assumed to select the alternative that gives them the main criteria in a vertical collaboration decision between a shipper and a logistics service provider [48, 49]. Table 1 highest utility. The attractiveness (or utility) of each contains a definition of the five selected variables and the alternative consists of a systematic (observable) component expected impact of each variable on a logistics collabora- and a random error (unobservable) term. The general utility tion decision, from a shipper perspective. function can be written as follows: Table 1 Definition and expected impact of the variables Variable Definition Expected impact Costs Costs of the logistics services provided by an LSP Negative: shippers prefer collaboration alternatives with lower costs Service Service level offered by an LSP in terms of number Positive: shippers prefer collaboration of shipments delivered on time alternatives that result in a higher level of customer service Trust The belief that a partner will not harm the interests Positive: shippers are more willing to cooperate of the counterpart when the level of trust is higher Confidentiality The belief that a collaboration partner keeps shared Positive: shippers are more wiling to cooperate information confidential when the level of confidentiality is high Commitment Participating actors are loyal and tolerant and do not Negative: shippers are more willing to cooperate worry constantly about being replaced when the required commitment from their side decreases 123 168 Logist. Res. (2010) 2:165–176 The initial utility function of our choice model includes applicable, as well as by using findings from case studies one alternative constant, three latent generic variables [13, 48, 49]. The levels used in the three different parts of (trust, confidentiality and commitment) and two alterna- our experiment are summarized in ‘‘Appendix 1’’. tive-specific variables (costs and service). Linearity is In order to place the experiment in a realistic context for assumed for the two specific variables. For each of the five the specific respondent, that is, as close as possible to his or variables, two aspects need to be measured: does the spe- her daily life, levels shown for the attribute costs and cific variable significantly impact a logistics collaboration service depend on the respondent’s current costs for decision and is this impact positive or negative? physical distribution and current service level. At the beginning of each interview, a respondent is asked for his 3.3 Context of the experiment current cost level, as well as the current percentage of orders delivered on time to customers. These data are The heart of our SP experiment is to give respondents the entered into the Excel file used to program our SP exper- choice to reduce logistics costs by implementing a con- iment. Subsequently, the program automatically generates cept that requires more intensive collaboration with the customized levels for each respondent on the choice logistics service providers. To make this choice clear and cards. explicit to the respondents, a standard case description is For this experiment, an orthogonal design is used. This used. The context of this standard case description is ensures that the attributes presented to respondents are based on seven cases that are conducted prior to this varied independently from one another, thus avoiding stated preference experiment [49]. For a detailed correlation between attributes. The orthogonal plan asso- description of this case, we refer to Visser and Ruijgrok ciated with each of the three parts of our design is gener- [47, 49]. In short, the case describes two different logistics ated by using the conjoint module of the statistical program concepts of ‘‘basic collaboration’’ and ‘‘intensified col- SPSS. Based on the number of attributes and the number laboration’’. Both concepts have the same logistics of levels of each attribute included in a design, SPSS structure; products are produced in a manufacturing plant configures a matrix showing the number of choices within owned by the shipper and, via a logistics hub owned by a a complete experiment and the level of each attribute for logistics service provider, distributed to the end custom- each choice. When the number of choices per experiment ers. The starting point for the respondent is the concept of becomes high, there is a strong likelihood that respondents basic collaboration. This is a situation in which only basic will experience fatigue in carrying out the choice exercises, logistics services, such as warehousing and transportation, thus increasing the response error. The literature provides are outsourced to different LSPs. The service providers different strategies for resolving this problem [35]. In our are responsible only for running the operation of these design, we follow these guidelines. We remove the domi- activities. Coordination and sourcing are still conducted nant choices and (randomly) separate the remaining ones by the shipper. This starting point is chosen because the into sets so that the experiment is completed by groups of logistics outsourcing literature demonstrates quite clearly respondents, each responding to a different subset of that the bulk of outsourced logistics services are in the choices. In the context of our design, this results in four areas of transportation and warehousing [25, 37]. The different sets of choices, with each set proposing 21 choi- alternative proposed to the respondents is the intensified ces to a given respondent. To validate the SP design and collaboration concept, which is characterized by trans- questionnaire, two cycles of eight pilot interviews were ferring sourcing and replenishment responsibilities to the conducted before the actual SP experiment took place. An logistics service provider. With this second concept, the example of a choice card used in our experiment can be LSP functions as a lead logistics provider, which results in found in ‘‘Appendix 2’’. a higher level of collaboration between shipper and logistics service provider. 3.4 Attribute levels The following SP studies are used as reference document: [12, 24]. Although widely used in transport studies, applying SP Currently, orthogonal SP designs are recognized as inefficient research to freight transport is still fraught with difficulties Nevertheless, ‘‘efficient’’ designs require a priori assumptions about [12]. Bergkvist [5] and Tsai et al. [44] concur that defining the expected coefficient values [34]. To the best of our knowledge, no previous experiments about logistics collaboration are available. and evaluating freight transport attributes is still in its Therefore, an orthogonal design is used, and the data of this infancy, compared to passenger transport. To overcome experiment can be used to make an ‘‘efficient’’ design later on. The this difficulty, attribute levels and values for our SP orthogonal design of our experiment is improved by removing the experiment are defined by using existing SP studies where dominant choices and using a folding procedure. 123 Logist. Res. (2010) 2:165–176 169 3.5 Data collection 3.6 Data analysis and quality Face-to-face interviews are used to collect the data Various modeling approaches can be possibly used to because they enable the researcher to monitor the process analyze the data collected in an SP experiment. Based on and evaluate respondent understanding of the alternatives. the assumed Weibull distribution of the error term in the No such controls are possible with self-completed overall utility function, the binary logit approach is used. designs. In total, 47 interviews are conducted with man- The binary logit model is one of a family of discrete choice agers in 18 different companies. This conforms to the models, which are widely used to examine the choices rule of thumb from [38] that around 50 interviews are made by individuals, households or firms, in choosing one sufficient. It is also in line with recent, comparable of a set of mutually exclusive alternatives [3]. To estimate studies [12, 33, 44]. A laptop was used to assist the the binary logit model coefficients for the data, Alogit researcher during the interviews and to show the choice version 4.2 is used. All 47 questionnaires are included in cards to the respondents. the analysis because there is no reason to remove respon- All data were collected in the chemical industry for dents from the sample, because of incomplete or incon- some reasons. First, to reach our final research objective, sistent responses. All respondents answered the dominant it is important to select a sector that is not characterized question in the correct manner and indicated that they were by opportunistic choice behavior, but by well-considered able to compare the proposed choices. decision behavior. The chemical sector with its stable and Data obtained from an SP experiment can be expected to conservative character meets this criterion [13]. Besides, provide an accurate picture of decision-making behavior. the chemical sector has a broad experience in logistics Nevertheless, like all research strategies, SP research has outsourcing, and thus, enough experienced respondents are its drawbacks. A common objection to SP techniques is available from a relatively homogenous sample. Third, that they measure intended behavior and people do not although the chemical industry has broad experience in necessarily do what they say [35]. Several researchers have outsourcing logistics, the outsourced services often have an compared stated and revealed preference data. However, these experiments reveal that stated preferences perform operational and transactional character. Stronger relation- ships need to be established with LSPs to find truly inno- fairly well in predicting real-life choices [18, 27, 50]. A vative and competitive supply chain solutions. Line second potential drawback is the occurrence of non-com- organizations in the chemical industry have identified more mitment bias (i.e., the respondent provides unconstrained intensive supply chain collaboration between shippers and and unreliable answers, because the subject of the study is logistics service providers as one of the critical drivers for of no interest to her or him [33]). This drawback is avoided long-term competitiveness of the industry [28, 29]. The by using a general SP design template, thorough prepara- interviewees were not chosen on a random basis because, tion and using a realistically designed and customized in order to obtain reliable results, it is important that choice experiment [2]. individual respondents be experienced in vertical collabo- ration decisions. Therefore, the respondents were selected by the means of purposive sampling. Primary contact 4 Findings persons within each company were identified in consulta- tion with representatives from the chemical industry. The general questions about logistics outsourcing and Additional respondents from each company were selected collaboration at the beginning of each interview make clear by the company’s primary contact. All respondents have a that the logistics services most frequently outsourced by leadership position in supply chain management, logistics respondents are those with an operational and repetitive or purchasing. character. Traditional services such as transport and Each interview starts with general questions about warehousing are outsourced by (almost) all responding logistics outsourcing and collaboration with logistics ser- companies. These findings also confirm previous studies on vice providers. This general information is used to obtain this subject [6, 8]. Our respondents explain that the col- a general idea of the current status of logistics outsourcing laborative relationships with their LSPs focus on opera- and collaboration with the visiting company. This pro- tional management and executing the outsourced activities. vides valuable information for the interpretation of results Almost forty percent of the respondents consider closer of the SP experiment later on. The interview continues collaboration with their LSPs but remain wary of trans- with the choice experiment. Each interview lasts approx- ferring more responsibilities to a service provider and of imately 60–90 min and is conducted at the participant’s actually starting collaboration at a higher level. Fears of workplace. losing transparency and of dependency on a certain 123 170 Logist. Res. (2010) 2:165–176 provider and problems finding a reliable and capable Table 2 Estimation results partner are reasons given for not intensifying collaboration. Title Final These reasons are in line with the most commonly cited Observations 926 concerns given in the logistics outsourcing and collabora- Final logL -479,4 tion literature [36, 37]. Degrees of 12 In order to analyze the data collected during the choice freedom experiment, a model estimation procedure is followed so as to find the final model with the best overall fit. For this Parameters Coefficient T value assessment, the maximum likelihood (ML) method is used. Alternative constant -2.26 (-9.0) The closer to zero the loglikelihood value, the better the Costs SCM -17.00 (-6.1) model fits the data. The loglikelihood (LL) value is used to function B10 million compare different model specifications. This comparison is Costs SCM -26.40 (-7.6) made in a formal statistical test: the likelihood ratio test. function [10 million This test compares the negative of twice the difference of Costs purchasing -5.27 (-2.1) 2 3 the LL values (D LL) to a V value from published tables function B10 million [24]. The value in the table depends on the confidence Costs purchasing -17.70 (-6.3) function [10 million interval chosen (in this research 95%) and the degrees of Service SCM function 44.60 9.8 freedom of the model. Service purchasing function 25.40 5.5 During the first part of the model estimation procedure, Trust level 1.03 3.9 different model specifications are defined to analyze whe- 4 costs [10 million ther heterogeneity in the total sample size results in a better Trust level 0.00 * overall model. Another part of the estimation procedure 3 costs [10 million was focused on the latent variables, because in the esti- Trust level 0.00 * mation results of the initial defined utility function, not all 2 costs [10 million levels of the latent variables were valued significantly. Trust level 2 ? 3 0.77 3.7 Moreover, there was not always measured a significant ? 4 costs B10 million difference in the valuation of the different levels of the Confidentiality level 2 ? 4 0.80 3.9 latent variables. Aggregating some levels of a latent vari- Confidentiality level 3 0.46 2.3 able or putting some levels to 0 can improve the overall fit Commitment level 4 -0.32 (-2.6) of a model, because the error term of the total model is Commitment level 3 0.00 * reduced [24]. In total, fourteen different model specifica- Commitment level 2 0.00 * tions are estimated via a step-by-step procedure. These Scale part 1 1.00 * steps and estimation results of the different model speci- Scale part 2 1.00 * fications are summarized in ‘‘Appendix 3’’. The final esti- Scale part 3 1.00 * mation results are shown in Table 2. Table 2 presents the coefficient values of our final model, together with the respective t-ratios. Each coefficient rep- result, these variables have only one coefficient. The coef- resents the relative importance of an explanatory variable in ficients of the latent generic variables (trust, confidentiality our binary logit model. Coefficients are significant at a 95% and commitment) are applied to categorical variables which confidence level if the t-ratio is greater than 1.96. When reflect the total utility increase or decrease for that variable. interpreting the values in Table 2, it should be borne in mind Therefore, these variables have a coefficient for each level in that the coefficients of the quantitative and qualitative the choice experiment. The model estimation procedure variables are presented differently. The coefficients of the makes clear that there is heterogeneity in our sample size. As alternative-specific variables (cost and service) are multi- a result, some subsegments are distinguished. The first plied by continuous variables in the final utility function and subgroup is related to the cost level of the respondent. At the therefore reflect the disutility per unit of the variable. As a beginning of each interview, a respondent was asked for his current logistics cost level to place the experiment for the 3 2 We use the standard table of the V distribution published by respondent in a realistic context. The estimation procedure Thompson (1941) [42]. The degrees of freedom refer to the number of coefficients estimated in a model. Except level 1, because for n categories of a latent generic variable, For the levels of the latent generic variables that are put to 0 during the model contains n - 1 dummy variables. Level 1 is the minimum the model estimation procedure and the scale factors, no t values are level used in the experiment and therefore set to 0 for the model presented. Therefore, a* is shown in the table. estimation procedure. 123 Logist. Res. (2010) 2:165–176 171 shows that respondents with logistics costs of more than 10 the three latent variables also significantly impact the million Euros value the attributes costs and trust differently respondents’ collaboration decision. The respondents not from respondents with logistics costs less or equal to 10 only derive utility from service and cost benefits but also million. The cost and trust element are more important in the from ‘‘non-rational’’ aspects. logistics collaboration decision for the first group. The sec- Finally, we use the results of our experiment to analyze ond subgroup distinguished in the sample size was related to the respondents’ inertia level. Table 2 shows the value of the the responsibility of the respondent. All our respondents constant term in our utility function. The t-value of this have a leadership position in supply chain management/ constant term is highly significant at a 95% confidence level. logistics or purchasing. These two different responsibilities This means that the respondents have a preference for the are also distinguished during the model estimation proce- alternative, which was called basic collaboration despite the dure. Table 2 shows that the coefficients for the attributes fact that the alternative of intensified collaboration results in costs and service representing respondents working in significant cost savings. The constant term represents the logistics and supply chain management have a significant initial resistance to switch to a situation of more intensive higher value than the coefficients for these variables repre- collaboration and thus refers to the respondents’ inertia senting respondents with a purchasing responsibility. As a level. The respondents are reluctant to change, and in con- result, it may be concluded that respondents with a function sequence, they do not always respond to relative differences in logistics or supply chain management attach more value in a rational manner. Inertia functions as a threshold, and as a to these two variables in their logistics collaboration deci- result, collaboration opportunities that are beneficial from an sion. The non-rational attributes were not valued differently economic point of view are unexploited. by the two subgroups of respondents. Table 3 contains a more detailed analysis of the constant In addition, Table 2 shows that the variables of service, term. In this table, the constant term is expressed as per- trust and confidentiality have a positive sign, as expected centage of the current logistics costs level for each subgroup (see Table 1). The positive sign of these variables indicates distinguished in our sample. This table shows that respon- that the larger the perceived satisfaction of these variables dents with a purchasing function and logistics costs less than on a particular collaboration alternative, the greater the or equal to 10 million have the highest resistance to change. likelihood that the collaboration will be chosen. On the They need more than a 35% cost saving to switch to a more other hand, the coefficients of the variables price and intensive form of logistics collaboration. At the other side, commitment have a negative sign. In line with our respondents with a logistics or supply chain management expectations, the greater the value of these variables in a responsibility and a cost level of more than 10 million have collaboration alternative, the lower the likelihood that the the lowest inertia level of the four subgroups in our sample. They are willing to switch to a different form of logistics specific collaboration alternative is chosen. Furthermore, the estimation results in Table 2 demon- collaboration when the cost saving is almost 9%. strate that service and costs are perceived as highly sig- nificant. The respondents found service followed by costs as the most important variables in their collaboration 5 Conclusions decision. At this aspect, there is no difference between the respondents with a purchasing or supply chain management The experiment proves that costs and service are the most responsibility. These findings are also consistent with important variables when shippers taking a vertical logis- previous research on decision criteria used in logistics tics collaboration decision but also shows that trust, com- collaboration decisions [25, 48]. The estimation results for mitment and confidentiality are significant aspects in such a the three latent variables demonstrate that some levels of decision. In addition, the results of the experiment show Table 3 ASC expressed in terms of costs Title subsample SCM function and SCM function Purchasing function Purchasing function costs B10 million and costs [10 million and costs B10 million and costs [10 million Observations 297 255 155 219 Final logL -148.2 -114.6 -89.0 -116.2 Degrees of freedom 7 7 7 7 Alternative constant 0.133 (-4.3) 0.0861 (-4.8) 0.354 (-4.3) 0.125 (-4.5) Cost scale 1.00 (-5.0) 1.00 (-6.3) 1.00 (-2.6) 1.00 (-5.6) 123 172 Logist. Res. (2010) 2:165–176 that shippers have a certain level of inertia, which con- collection and analyses. A comparable study at the LSP side strains the intensification of logistics collaboration. Based could make clear what the impact of behavioral factors is at on these findings, it is concluded that the conducted SP the logistics service providers’ side and show whether there experiment confirms our theoretical findings that both is any difference in how shippers and LSPs value the vari- aspects function as a threshold for a shipper to intensify ables in a logistics collaboration decision. Including the logistics collaboration with a logistics service provider. service provider side may, however, yield additional This conclusion contributes to the debate on logistics col- insights into logistics collaboration decisions between laboration in several ways. First, the empirical research on shippers and LSPs. These could be useful to provide rec- shipper’s choice behavior increases our understanding ommendations for removing the existing thresholds to fully about why the potential of logistics collaboration is not exploit the benefits of logistics collaboration. Furthermore, exploited fully in practice, because we have quantified the our data collection was limited to companies in the chemical shipper’s resistance to change. As a result, to remove or sector. Therefore, our findings cannot necessarily be gen- relieve the thresholds, the debate about logistics collabo- eralized to other industries. This should be investigated ration should not only improve cost–benefit considerations further through additional empirical verification; more and analysis but also shift the attention to the immaterial companies and industries should be examined. In addition, side of the discussion. In addition, better incorporating the results of the stated preference experiment identify areas human behavior into the existing models will enhance the for further research. The data show that respondents from precision and rigor of these models and will yield more different subgroups value some of the choice variables dif- realistic results. Furthermore, the findings of this research ferently and have different levels of inertia. Nevertheless, stress the importance of relationship management and our research does not explain why these differences exist. change management in the logistics field. Especially, LSPs Further work could focus on understanding these differ- should be aware of this and pay more attention to such ences. Next to this, our results do not show whether there is aspects. Emphasizing only the possible cost savings of a any relation between the differences in inertia level and the certain logistics solution is insufficient to convince the differences in the valuation of the choice variables by the other party to change its routines. subgroups. Also, this relation can be an area for further research. Finally, further research could focus on the inte- gration of the immaterial side of logistics collaboration 6 Limitations and further research decisions into the existing models that are used to prove the potential of logistics collaboration. The paper adds to the existing literature on supply chain Acknowledgments The author gratefully acknowledges the support management and third-party logistics, because the empirical received during designing and analyzing the SP experiment from evidence quantitatively proves the significant impact of Marco Kouwenhoven of Significance, a Dutch research institute behavioral factors on logistics collaboration decisions specialized in transport studies. between shipper and logistics service provider. However, there are some limitations and areas for further research. First of all, data were collected only on the shipper side. Data Appendix 1 collection on the LSP side as well would dramatically See Table 4. increase the size of the study in terms of design, data Table 4 Attribute levels in SP experiment Attribute Levels basic collaboration Levels intensified collaboration Card A Card B Costs 4% below current costs 12% below costs card A 1% below current costs 6% below costs card A Equal to current costs 2% below costs card A 3% above current costs Equal to costs card A Service 4% below current service 4% below current service 1.5% below current service 1.5% below current service Equal to current service Equal to current service 2% above current service level 2% above current service level 123 Logist. Res. (2010) 2:165–176 173 Table 4 continued Attribute Levels basic collaboration Levels intensified collaboration Card A Card B Trust (1) You are not absolutely certain LSP will not harm your interests. This is not contractually agreed (2) You are not absolutely certain LSP will not harm your interests, but this is contractually agreed (3) You are convinced LSP will not harm your interests. This is not contractually agreed (4) You are convinced LSP will not harm your interests. This is contractually agreed Confidentiality (1) You are not absolutely certain information is kept confidential. This is not contractually agreed (2) You are not absolutely certain information is kept confidential, but this is contractually agreed (3) You are convinced information is kept confidential. This is not contractually agreed (4) You are convinced information is kept confidential. This is contractually agreed Commitment 1 year 1 year 2 years 2 years 3 years 3 years 5 years 5 years Appendix 2 See Table 5. Table 5 Choice card example 123 174 Logist. Res. (2010) 2:165–176 Appendix 3 See Table 6. Table 6 Model estimation procedure Model Description Observations Final log D.O.F D (LL) Interim conclusion Remark (LL) 1 Initial utility function 926 -497.0 12 Not applicable 2 Identify the optimal 926 -519.6 12 -23 D loglikelihood value =-23. specification of costs: Model 1 and 2 have the same absolute or relative D.O.F. The LL of model 1 is values for the costs closer to 0, and coefficients are attribute significant at 95% confidence level. Model 1 better fits the data than model 2 3 Subsample logistics costs: 926 -491.8 13 5.2 D loglikelihood value = 5.2. Costs, trust and This is a significant confidentiality attributes (1) B10 million Euro improvement at a 95% valued differently by (2) [10 million Euro confidence level of a likelihood respondents with costs level ratio test . B of 10 million Model 3 better fits the data than and respondents with a cost model 1 level of [10 million. These attributes need further verification 4 Subsample logistics costs: 926 -493.6 14 -1.8 D loglikelihood value =-1.8. This is a not significant (1) B10 million Euro improvement at a 95% (2) \100 million Euro confidence level of a (3) [100 million Euro Likelihood ratio test. Model 3 better fits the data than model 4 5 Subsample logistics costs 926 -489.9 12 1.9 D loglikelihood value = 1.9. and latent variable trust This is a significant improvement at a 95% Costs B10 million: trust confidence level of a levels 2, 3 and 4 Likelihood ratio test. aggregated Model 5 better fits the data than Costs [10 million: trust model 3 levels 2, 3 and 4 put to 0 6 Subsample logistics costs 926 -489.8 12 0.1 D loglikelihood value = 0.1. and latent variable Model 5 and 6 have same confidentiality D.O.F. The LL value of model 6 is closer to 0, but not all Costs B10 million: coefficients are significant at a confidentiality levels 2 95% confidence level. Model 5 and 3 aggregated better fits the data than model 6 Costs [10 million: confidentiality levels 2 and 4 aggregated; level 3 put to 0 7 Subsample logistics costs 926 -491.9 10 -2.0 D loglikelihood value =-2.0. and latent variable This is a significant confidentiality. improvement at a 95% confidence level of a All costs levels: Likelihood ratio test. confidentiality levels 2 Model 7 better fits the data than and 4 aggregated; level model 5 3 put to 0 8 Subsample logistics costs 926 -489.9 11 2.0 D loglikelihood value = 2.0. and latent variable This is a significant confidentiality. improvement at a 95% confidence level of a All costs levels: Likelihood ratio test. confidentiality levels 2 Model 8 better fits the data than and 4 aggregated model 7 123 Logist. Res. (2010) 2:165–176 175 Table 6 continued Model Description Observations Final log D.O.F D (LL) Interim conclusion Remark (LL) 9 Subsample chemical 926 -496.8 13 0.2 D loglikelihood value = 0.2. No significant difference in industry: This is not a significant the valuation of the improvement at a 95% attributes between the two (1) Base chemicals confidence level of a identified subgroups: fine (2) Specialty chemicals Likelihood ratio test. and base chemicals Model 1 better fits the data than model 9 10 Subsample responsibility 926 -487.3 14 9.7 D loglikelihood value = 9.2. Costs and service attribute are respondents: This is a significant valued differently by improvement at a 95% respondents with a supply (1) Supply chain confidence level of a chain/logistics function and management Likelihood ratio test. Model 10 respondents with a (2) Purchasing better fits the data than model 1 purchasing function 11 Subsample responsibility 926 -480.7 16 6.6 D loglikelihood value = 6.6. and logistics cost This is a significant improvement at a 95% (1) supply chain confidence level of a management Likelihood ratio test. (2) purchasing Model 11 better fits the data (3) Costs B10 million than model 11 (4) Costs [10 million 12 Subsample responsibility 926 -480.7 15 0.0 D loglikelihood value = 0.0. and logistics costs This is a significant improvement at a 95% Latent variable confidence level of a confidentiality as in Likelihood ratio test. model 8 Model 12 better fits the data than model 11 13 Subsample responsibility 926 -479.2 14 1.5 D loglikelihood value = 1.5. and logistics costs This is a significant improvement at a 95% Latent variable confidence level of a confidentiality as in Likelihood ratio test. model 9 Model 13 better fits the data Latent variable trust as in than model 12 model 5 14 Subsample responsibility 926 -479.4 12 -0.2 D loglikelihood value =-0.2. and logistics costs This is a significant improvement at a 95% Latent variable confidence level of a confidentiality as in Likelihood ratio test. Model 14 model 9 better fits the data than model Latent variable trust as in model 5 Latent variable commitment levels 2 and 3 put to 0 D.O.F. degrees of freedom in the model b 2 To determine whether the delta of the Loglikelihood value is a significant improvement or not, a standard table of the X distribution function is used [42] 4. Berglund M (2000) Strategic positioning of emerging third-party References logistics providers, Thesis No. 45, Linkoping University, Sweden 1. Andersson D, Norrman A (2002) Procurement of logistics ser- 5. Bergkvist E (2001) Freight transportation: valuation of time and vices—a minute work or multi-year project? Eur J Purch Suppl forecasting of flows, Umea economic studies 549. Umea Manage 8:3–14 University, Sweden 2. Ben-Akiva M, Bradley M, Morikawa T, Benjamin J, Novak T, 6. Budde F, Felcht UH, Franke Mo ¨ lle H (2006) Today’s challenges Oppewal H, Rao V (1994) Combining revealed and stated pref- and strategic choices. In: Budde F, Felcht UH, Frankemo ¨ lle H erence data. Market Lett 5(4):335–350 (eds) Value creation, strategies for the chemical industry, 2nd 3. Ben-Akiva M, Lerman S (1985) Discrete choice analysis: theory edn. Wiley, Weinheim and application to travel demand. MIT Press, Cambridge 123 176 Logist. Res. (2010) 2:165–176 7. Camerer CF (2003) Behavioral game theory. Experiments in 30. Mentzer J, Myers M, Cheung MS (2004) Global market seg- strategic interaction. Princeton University Press, Princeton mentation for logistics services. Ind Mark Manage 33(1):15–20 8. Capgemini, Georgia Institute of Technology, SAP, DHL (2008), 31. Min S, Roath AS, Daugherty PJ, Genchev SE, Chen H, Arndt 13th Annual Third-Party Logistics Study 2008, http://www. AD, Richey RG (2005) Supply chain collaboration: what’s hap- 3plstudy.com pening? Int J Logistics Manage 16(2):237–256 9. Carter CR, Kaufmann L, Michel A (2007) Behavioral supply 32. Mosch R (2004) The economic effects of trust. Theory and management: a taxonomy of judgment and decision-making empirical evidence, Dissertation Free University Amsterdam biasis. Int J Distrib Logistics Manage 37(8):631–669 33. Muilerman GJ (2001) Time-based logistics. An analysis of the 10. Cohen MD, March JG, Olsen JP (1972) A garbage can model of relevance, causes and impacts, Dissertation Technical University organizational choice. Adm Sci Q 17(1):1–25 Delft 11. Cruijssen F (2006) Horizontal cooperation in transport and 34. Ortuzar J (2000) Stated preference modelling techniques. PTRC logistics, Center Dissertation Serie No. 176, Tilburg University Education and Research Services, London 12. Danielis R, Marcucci E, Rotaris L (2005) Logistics managers’ 35. Pearmain D, Swanson J, Kroes E, Bradley M (1991) Stated stated preferences for freight service attributes. Transp Res Part E preference techniques: a guide to practice. Steer Davies Gleave 41:201–215 and Hague Consulting Group, 2nd edn 13. Eutralog (2004) New Issues in Intermodality, Deliverable D4.2 36. Razzaque MA, Sheng CC (1998) Outsourcing of logistics func- 14. EyeForTransport (2009) The European 3PL Market. A brief analysis tions: a literature survey. Int J Phys Distrib Logistics Manage of Eyefortransport’s recent survey, EyeforTransport, Aug 2009 28(2):89–107 15. Fehr E, Falk A (2002) Psychological foundations of incentives. 37. Selviaridis K, Spring M (2007) Third party logistics: a literature Eur Econ Rev 46:687–724 review and research agenda. Int J Logistics Manage 18(1):365–378 16. Folmer H (2007) Why economists make blunders so often, 38. Sheldon RJ (2007) Stated preference methods short course. Inaugural Lecture, 16 Oct 2007, University of Groningen Institute For Transport Studies University of Leeds 17. Groothedde B (2005) Collaborative logistics and transportation 39. Simatupang T, Sridharan R (2002) The collaborative supply networks: a modeling approach to hub network design, Trail- chain. Int J Logistics Manage 13(1):15–30 Thesis Series T2005/15, Delft, Trail 40. Simon HA (1957) Models of man. Wiley, New York 18. Green PE, Srinivason V (1978) Conjoint analysis in consumer 41. Simon HA (1960) The new science of management decision. research issues and outlook. J Consum Res 5:103–121 Prentice Hall, New Jersey 19. Gulati R, Kletter D (2005) Shrinking core, expanding periphery: 42. Thompson CM (1941) Tables of percentage points of the the relational architecture of high-performing organizations. Calif incomplete beta function and of the chi-square distribution. Manage Rev 47(3):77–104 Biometrika Vol. 32 20. Hannan MT, Freeman J (1984) Structural inertia and organiza- 43. Travisi C (2007) Valuing environmental decay, quantitative tional change. Am Sociol Rev 49(2):149–164 policy-oriented studies on urban and rural environments, Dis- 21. Hopp WJ (2004) 50th anniversary article: fifty years of man- sertation, Free University, Amsterdam agement science. Manage Sci 50(1):1–7 44. Tsai M, Wen C, Chen C (2007) Demand choices of high-tech 22. Kahneman D, Tversky A (1979) Prospect theory: an analysis of industry for logistics service providers—an empirical case of decision under risk. Econometrica 47(2):263–291 offshore science park in Taiwan. Ind Mark Manage 26:617–626 23. Kosfeld M (1999) Individual decision-making and social inter- 45. Van Laarhoven P, Berglund M, Peters M (2000) Third-party action, Dissertation, Tilburg University logistics in Europe—five years later. Int J Phys Distrib Logistics 24. Kouwenhoven M, Rohr C, Miller S, Siemonsma H, Burge P, Manage 30(5):425–442 Laird J (2007) Isles of scilli. Travel Demand Study, Rand Europe 46. Vijver M (2009) Collaboration in buyer-supplier relationships, 25. Lieb R, Bentz BA (2005) The North American third- party Dissertation, Tilburg University logistics industry in 2004: the provider CEO perspective. Int J 47. Visser LJ (2010) Thresholds in logistics collaboration decisions, Phys Distrib 35(8):595–611 Center Dissertation Serie No. 252, Tilburg University (in press) 26. Lindblom CE (1959) The science of ‘‘Muddling Through’’, Public 48. Visser LJ, Ploos van Amstel (2008) Supplier involvement in Administration Review, vol 19. Spring, New York, pp 79–88 purchasing logistics services. In: Proceedings NOFOMA con- 27. Louvie `re JJ (1988) Conjoint analysis modelling of stated-pref- ference 2008, pp 631–645 erences: a review of theory, methods, recent developments and 49. Visser LJ, Ruigrok CJ (2007) Measuring the thresholds in deci- external validity. J Transport Econom Policy 2:93–121 sion making on outsourcing. In: Proceedings Vervoerslogistieke 28. Mantel SP, Tatikonda MV, Liao Y (2006) A behavioral study of Werkdagen 2007, pp 433–446 supply manager decision-making: factors influencing make ver- 50. Wardman M (1988) A comparison of revealed preference and sus buy evaluation. J Operat Manage 24:822–838 stated preference models of travel behavior. J Transport Econom 29. McKinnon A (2004) Supply chain excellence in the European Policy 22(1):71–91 chemical industry, the European Petrochemical Association (EPCA)
Logistics Research – Springer Journals
Published: Oct 17, 2010
You can share this free article with as many people as you like with the url below! We hope you enjoy this feature!
Read and print from thousands of top scholarly journals.
Already have an account? Log in
Bookmark this article. You can see your Bookmarks on your DeepDyve Library.
To save an article, log in first, or sign up for a DeepDyve account if you don’t already have one.
Copy and paste the desired citation format or use the link below to download a file formatted for EndNote
Access the full text.
Sign up today, get DeepDyve free for 14 days.
All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.