Journal of Transportation Management

Volume 26 | Issue 2 Article 4

1-1-2016

Parcel shipping: Understanding the needs ofbusiness shippersMichael S. GarverCentral Michigan University, garve1ms@cmich.edu

Zachary WilliamsCentral Michigan University, willi1zs@cmich.edu

Sean P. GoffnettCentral Michigan University, sean.goffnett@cmich.edu

Brian J. GibsonAuburn University, brian.gibson@auburn.edu

Follow this and additional works at: https://digitalcommons.wayne.edu/jotm

Part of the Operations and Supply Chain Management Commons, and the TransportationCommons

This Article is brought to you for free and open access by DigitalCommons@WayneState. It has been accepted for inclusion in Journal ofTransportation Management by an authorized editor of DigitalCommons@WayneState.

Recommended CitationGarver, Michael S., Williams, Zachary, Goffnett, Sean P., & Gibson, Brian J. (2016). Parcel shipping: Understanding the needs ofbusiness shippers. Journal of Transportation Management, 26(2), 29-46. doi: 10.22237/jotm/1451606580

Fall/Winter 201629

PARCEL SHIPPING: UNDERSTANDING THE NEEDS OF BUSINESS SHIPPERS

Michael S. GarverZachary WilliamsSean P. Goffnett

Central Michigan University

Brian J. GibsonAuburn University

ABSTRACTResearch on carrier selection addresses how shippers choose carriers. To date, this extensiveresearch stream has not adequately addressed a known and significant shipping segment: businessparcel shippers. In this research, input from 374 business parcel shippers was captured and analyzedusing Maximum Difference Scaling. The respondents were asked to evaluate the importance of 17carrier selection variables in regard to choosing a parcel carrier. The overall results indicate thatdelivery promises, transit times, rates, pick-up promises, and tracking are the most importantattributes when a parcel shipper makes a carrier selection. In addition, the results of attributeimportance were used classify the parcel shippers into four unique segments.

INTRODUCTIONThe parcel shipping market, which is commonlycharacterized as shipments of up to 150 pounds(Burks et al., 2004), has grown substantiallyover the last decade. Data from the CommodityFlow Survey, issued by the U.S. Census Bureauin conjunction with the U.S. Department ofTransportation, reveals that in the ten year periodfrom 2002 to 2012, the parcel mode oftransportation grew by 59% to nearly $1.6trillion worth of goods shipped between U.S.businesses (2013; 2010). Comparatively, overthe past 25 years, the parcel shipping industryhas greatly outperformed the less-than-truckload(LTL) industry in terms of growth (Jindel, 2010).

Three primary factors have driven the substantialgrowth of the parcel shipping industry. First,U.S. retail e-commerce sales grew by 406% to$224 billion between 2002 and 2012, greatlyexpanding the need for parcel shipping (U.S.Department of Commerce, 2013). Second,changes to manufacturing and inventoryprocesses have created increased volume ofsmaller, more frequent parcel deliveries. Finally,parcel carriers have increased their maximumshipment weights from 70 pounds to 150 pounds

and developed pricing innovations to convertLTL freight (Haber 2013; Jindel 2010).

Although the parcel shipping market has grownsubstantially, academic research on this topic hasnot. Carrier selection research is one of the mostresearched topics in logistics (e.g., McGinnis,1979; Abshire & Premeaux, 1991; Voss et al.,2006). Yet, in this wide stream of research,business parcel carrier selection has receivedalmost no attention. This dearth of research is aproblem for a number of reasons. First, the sizeand growth of the parcel shipping market issubstantial. Second, academic research hassuggested that carrier selection is specific to themode (truckload or TL, LTL, etc.), as eachmode’s customers likely have their own uniqueneeds (Kent et al., 2001). Without a betterunderstanding of the specific needs of parcelshippers, parcel carriers cannot develop the bestservice solutions for their customers.

The purpose of this research study is to examinethe preferences and characteristics of businessparcel shippers. More specifically, the study willanswer the following questions:

Journal of Transportation Management30

1. What is the relative importance of carrierselection variables that business parcelshippers consider when choosing a parcelcarrier?

2. Based on the importance of selectionvariables, can business parcel shippers besegmented according to the importanceof these variables?

To pursue answers to these research questions, abrief literature review of parcel shippers andcarriers is presented. Next is a discussion of theresearch design and method employed in thisstudy, followed by a presentation of the results.This is followed by the discussion andimplications. Finally, future research and studylimitations are presented.

LITERATURE REVIEWAt the outset of the study, a literature review ofparcel markets was undertaken to understand therespective requirements of parcel shippers andcapabilities of parcel carriers. This effortrevealed a dearth of parcel research relative tothe number of studies focusing on LTL and TLtransportation. Within the parcel sector, theresearch highlights the growing demand forparcel transportation. Less attention has beenpaid to shipper needs or carrier service offerings.

Parcel Shipping Demand DriversShipping methods are often dictated by a firm’soperational strategies and the purchasingpractices of buyers. In the case of parcelshipping, changes in the way goods and servicesare produced and distributed contribute to thegrowing importance of this method. Inparticular, the adoption of lean inventoryprinciples, the use of just-in-time (JIT)manufacturing and customized mass production,and the dramatic growth of e-commerce activityare key contributors to the growth of parcelshipping (Morlok et al., 2000).

In a lean operating environment, excessiveinventory is considered waste (Liker, 2004;Vokurka and Lummus, 2000). A major

challenge is the trade-off between decreasedinventory levels due to small batch sizes andincreased transportation costs resulting fromfrequent deliveries (Chen and Saker, 2010). Tolower total cost in a lean operation, managersmust allow for trade-offs between inventory,material handling, storage, transportation, etc.Thus, managers are likely to ship smallerbatches using parcel carriers or work with freightforwarders and consolidators (Myerson, 2012).

Arcelus and Rowcraft (1993) highlighted thelink between the JIT manufacturing movementand an increased need for parcel shipments. JITis an order pull system based on actual demandand consumption that attempts to minimizeinventory levels and shorten lead times. As aresult, smaller, more frequent orders are requiredand firms become much more reliant on rapidreplenishment and expedited delivery,capabilities that parcel carriers excel in.Similarly, one-off production of personalcomputers, footwear, and clothing drives directdelivery to end users (Andrews, 1998). Again,parcel shipping is a logical delivery solution.

The evolution of consumer buying practices hasled to significant growth in parcel shippingactivity. Christopher Jr. (2011) notes that e-commerce has been the fastest growing tradesector since 1999 and was largely unaffected bythe global economic downturn. At the height ofthe recession in 2009, e-commerce activityactually increased, allowing many parcel carriersto remain profitable (Andrews, 2011). US retaile-commerce sales reached $263 billion in 2013and will continue to increase at an annual rate of13.7% through 2017, when sales are expectedsurpass the $440 billion mark (emarketer, 2014).This growth has driven demand for parceltransportation, to the point of taxing the carriers’network capacity during peak holiday demand(Stock, 2013).

Although heavy attention has been given to therapid growth of business-to-consumer (B2C) e-commerce activity, it is a fraction of business-to-business (B2B) e-commerce activity. Laudon

Fall/Winter 2016 31

and Traver (2012) expect a $1.1 trillion increasein B2B e-commerce sales, rising from $3.3trillion in 2011 to $4.4 trillion in 2015. Thisgrowing B2B activity is further driving demandfor parcel shipping service and is leading to rateincreases in the form of higher minimum charges(Burnson, 2014).

Finally, changing retail strategies are fuelingparcel transportation’s growth. Subscriptionbased services like Amazon Prime allowconsumers and small businesses to place smallorders without incurring charges for second daydelivery (Anderson, 2014). A strategic shift tosmaller store sizes with lower in-store SKUvariety drives the need for home delivery ofSKUs that are offered only online (Gustafson,2014). And, liberal e-commerce return policieswith free shipping lead to high return rateswhich Sarkis et al. (2004) estimate at greaterthan 30%.

Parcel Shippers’ NeedsRecent research purports to show the need forcarriers to focus on shipper’s most importantneeds (e.g., Dobie 2005). Understanding shipperneeds is a key prerequisite for carriers todevelop, implement, and refine customer drivenstrategy (LeMay, 1986; Coulter et al., 1989Lambert et al., 1993). Despite the growingactivity and importance of the parcel shippingmarket, the literature review yielded only tworesearch studies that specifically focused on theneeds of parcel shippers.

Ding, et al. (2005) developed a fuzzy multi-criteria decision-making model to support theselection of suitable Taiwanese courier serviceproviders. Six primary criteria were included:speed and reliability; freight rates; safety; salesstaff; service and convenience; and, carrierconsiderations. Thirty sub-criteria of interest toparcel shippers were used by this model tosystematically appraise and rank four parcelcarriers.

Lin and Lee (2009) identified seven factors thatare important in choosing parcel carriers when

firms and consumers are selling products in anonline environment. The researchers found thatthe following factors were important whenchoosing a parcel carrier:

On-time, tracking, and quick response,Fare rate and freight loss,Security and reputation,Personnel courtesy and quality,Equipment, package, and flexible service,Diversified service,Promotion and reputation.

These studies took important steps in identifyingparcel shipping customers as a known andunique segment of the transportation market.The current research seeks to extend the priorresearch and further answer questions regardingthe needs of parcel shippers.

Parcel Carriers’ CapabilitiesMuch academic literature has been focused onvarious motor carrier markets, including LTL(Jarrah et al., 2009; Lin et al., 2009; Barcos etal., 2010; and Hernandez et al., 2011) andtruckload (TL) (Kent and Smith, 2005; Ergun etal., 2007; Liu et al., and 2010; Pai, 2011).However, many distinct differences exist formotor carriers that operate in the parcelenvironment that necessitates independent studyof this market segment.

Parcel shipping has been hailed by Morlok et al.(2000) as a major element of the U.S.transportation system that is essential to moderncommerce. From a service standpoint, theseauthors state that parcel carriers are at theforefront of modern transportation services.Parcel carriers are industry leaders due to theirdifferentiated time-definite service options,intermodal service, in-transit visibility, and dataintegration with the management systems ofcustomers.

Parcel carriers also have an order processingadvantage over other motor carriers. FedExGround receives more than 95 percent of allpackages via electronic manifest. When

Journal of Transportation Management32

manifests are communicated electronically,parcel carriers gain knowledge of shipmentsearly and create more efficient loads.Additionally, parcel carriers have advantages interms of accurate billing. Finally, parcel carrierscapture the dimensions and weight of everypackage, whereas LTL carriers typically rely oncustomer input for weight and classification(Jindel, 2010).

Given the current state of the parcel shippingliterature, additional study is warranted. Thecurrent study will extend the knowledge base byinvestigating the alignment of parcel carriercapabilities with the needs of parcel shippers.Poor alignment can result in resources beingwasted on unneeded service elements whileimportant service attributes go unfulfilled.

METHODOLOGY

Maximum difference scaling (MD) is a discretechoice survey method that asks surveyrespondents to choose the most and leastimportant items from a set of options. MDallows a large number of items to be traded offagainst each other in an efficient manner, whichis independent of any rating scale bias.Additionally, MD produces a needs basedsegmentation, allowing priorities to be estimatedfor any subgroup (Cohen, 2003). Given thesecapabilities, it is well suited to the researchobjectives of this parcel shipping study.MD is gaining attention from academicresearchers and practitioners (e.g., Cohen andOrme, 2004; Garver, 2009; and, Williams et al.,2011). Another study identified MD as themethod that delivered the most valid resultswhen conducting importance research (Chrzanand Golovashkina, 2006). Moreover, Garver etal. (2010) recommend MD as it has key distinctadvantages over other methods, particularlyrating scales. Traditional rating scales do notforce choices, thus respondents may be free toselect everything as important for example.

Research Process

VariablesTo determine the appropriate attributes for parcelcarrier selection, the carrier selection literaturewas thoroughly reviewed. Next, the researchersmet with industry experts to make sure that therelevant attributes were identified and that theseattributes were phrased appropriately. Thisprocess resulted in a final list of attributes that isaligned with the logistics academic literature, yetalso has relevance to logistics professionals.

Once the list of attributes was developed, theresearchers chose to include five attributes perMD survey question, a common MD bestpractice (Chrzan and Patterson, 2006). The nextstep in the MD experimental design stage was todetermine the overall number of MD surveyquestions that should be presented to studyrespondents. Following the guidelines put forthby Garver et al. (2010), each research participantwas asked 11 questions. The experimentaldesign plan in the current study led to eachattribute being shown approximately three timeseach to survey respondents.

The actual MD survey questions were developedafter the experimental design plan was created,with each question containing the followinginstructions:

“Please consider how important differentattributes are when selecting a parcelcarrier. Considering only these 5features, which is the Most Importantand which is the Least Important?”

For each of the 11 MD questions, the researchrespondents were asked to select the “mostimportant” and the “least important” attribute.

Data CollectionData for this study were collected from abusiness research panel. Members of the panelcame from a leading market research firm calledMarketTools. The choice to use an online panelas a data source follows numerous other supplychain and logistics researcher’s use of this

Fall/Winter 2016 33

approach (e.g., Autry et al. 2008; Jack et al. 2010;Richey et al. 2010; and, Grawe et al. 2011).

When using online panels as a data source,researchers have taken a series of additionalsteps to validate knowledge and skills ofrespondents (e.g., Autry et al., 2010) and thisstudy implemented those as well. First,MarketTools, was hired to provide the onlinepanel. Second, filter questions were added to thesurvey in order to screen out panelists who didnot fit the appropriate respondent profile. Figure1 demonstrates how these individuals wereeliminated from the respondent pool. As a result,

only logistics practitioners with extensive parcelshipping knowledge and buying influence areincluded in the final data set for analysis.

Data CleansingFour hundred twenty (420) completed surveys werecollected. However, after excluding respondentswith incomplete surveys, respondents lacking thenecessary expertise, or those respondents whoincorrectly answered embedded trap question, 374valid and complete surveys were retained foranalysis. When conducting MD research, aminimum of 100 data points are recommended

Journal of Transportation Management34

(Garver, et al., 2010). The final data set greatlyexceeds this benchmark.

Data AnalysisSawtooth software (7.0) was used to collect andanalyze the MD data. Specifically, HierarchicalBayes estimation was implemented to study theMD data. A MD study provides results whichcan be used to derive need-based segments,which is one of the objectives of the currentstudy (Orme, 2005; Orme. 2005b; Garver, 2009;Garver et al, 2010).

RESEARCH RESULTS

General properties of the sample will first bediscussed, then the MD results will be presented,followed by a discussion of the five need-based

segments identified using latent class cluster analysis.Then, results from classification trees, ANOVA, andcross-tabulation analysis will be presented todescribe the nature of each segment.

MD Parcel Selection Attribute ImportanceResultsA common practice in MD research is to rescaleHierarchical Bayes analysis results so that theimportance scores assigned to all attributes sumto 100 points, with higher scores reflectinggreater importance of the attribute. This meansthat the importance scores of one attributeshould be interpreted in relative, not absolute,terms (e.g., an importance score of 10 is greaterthan 5, but not twice as great). Table 1 contains

Fall/Winter 2016 35

the MD mean importance scores for the parcelcarrier selection attributes.

Several observations should be made aboutTable 1. First, there is discrimination among thedifferent parcel carrier selection attributes(Garver et al, 2010; Williams et al, 2011), withimportance scores ranging from 0.68 to 15.77(Table 1). Second, the scores of the six attributeshaving the greatest importance scores sum to70.6, which means these collectively account forjust over 70% of the total importance in parcelcarrier selection by customers. Third, fourattributes having greatest importance in parcelcarrier selection – Delivers shipments WhenPromised (15.77), Transit Time (speed) (12.14),Competitive Rates (11.78), and Picks-upShipments When Promised (11.46) – account for51% of the total importance of attributes thatinfluence the choice of parcel carriers byshippers.

Fourth, several attributes that have receivedmuch attention from practitioners and academicsreceived relatively low importance scores.Specifically, security practices (4.0) andsustainability (.8) were ranked 10th and 15th,respectively, in terms of their importance in theparcel carrier selection process, whileinformation sharing (.7) was the least importantto business customers.

Finally, while mean responses are of someassistance in interpreting empirical results, theycan be misleading (Garver, 2009; Garver et al.,2010; Williams et al., 2011). Garver (2010)suggested that researchers should examine need-based segments (if they exist) in order to trulyunderstand customers in the marketplace.Accordingly, this analysis was next undertaken,the results of which are reported below.

Identification of Parcel Need-Based SegmentsLatent Class Cluster Analysis (LCCA)Latent class cluster analysis (LCCA) was used todetermine whether meaningful, unique need-based segments exist in the sample used in thisstudy. Research over the last decade has shown

that LCCA has distinct advantages over moretraditional methods of cluster analysis (Vermuntand Magidson, 2005). Research has shown thatLCCA has improved predictive capabilities overmore traditional clustering techniques (Vermuntand Magidson, 2003).

Furthermore, LCCA assists researcher bysupplying researchers with fit statistics thatguide the selection of the appropriate number ofsegments. Finally, LCCA provides probabilitiesof segment membership, which is helpful indetermining how well the technique has workedin segmenting the market (Garver, et al., 2008)

The researchers employed Latent Gold 4.0 toconduct the analysis. Each of the 17 MD parcelcarrier selection attributes was entered intoLCCA as continuous attributes to develop thesegmentation results. Garver, et al. (2008)suggest that most segmentation studies examineup to five segments, since it is difficult for mostpractitioners to focus on more than fivesegments. With this in mind, the researchers ranthe following cluster analysis models forconsideration evaluation: a one cluster, a twocluster model, and so on. In total, six differentmodels were evaluated (up to a six clustersolution).

The researchers used the random seed default inthe program, which randomly selects tendifferent starting points for each analysis. Thisprocedure overcomes the potential limitation ofLCCA models to produce a local solution asopposed to a global maximum.

Number of Segments - Evaluation andSelection - LCCA

The first goal of this analysis was to determine ifneed-based segments of parcel carrier customersexist, or whether the marketplace of parcelcarrier customers is homogeneous in terms ofthe importance attached to the parcel carrierselection attributes. If the sample ishomogeneous, then the interpretation of mean(overall) importance scores is valid. However, ifneed-based segments do exist, the first goal is to

Journal of Transportation Management36

determine the appropriate number of need-basedsegments. Selecting the appropriate number ofsegments is a critical task in LCCA.Accordingly, the latent class model evaluationstrategies identified by Garver, et al. (2008) wereadapted for LCCA in this study. Similarly, thefollowing “best practices” for determining theappropriate number of segments within LCCAwere followed (Vermunt and Magidson, 2005):

1) Goodness of fit measures2) Misclassification error3) Theoretical knowledge, expertise, and

researcher judgment.

Goodness of fit MeasuresThe BIC is the most popular goodness of fitmeasure for assessing LCCA models(Arunotayanun and Polak, 2011), especiallywhen the data are sparse, the situation for mostlogistics research studies (Garver et al., 2008).One reason for this popularity is that the BICmeasure simultaneously explains model fit whileaccounting for model parsimony. Typically, amodel with a lower BIC value is preferred overone with a higher BIC value (Guerrero, Egea,and Gonzalez 2007; Wen, et al., 2012).

The researchers first specified and analyzedseveral models, estimating a 1, 2, 3, 4, 5, and 6-

segment model, using the 1-segment model as thebaseline. If the 1-segment model has the lowest BICscore, then there is evidence that the parcel carriermarket is homogeneous with respect to theimportance placed on parcel carrier selectionattributes. Table 2 provides critical results forevaluating model fit and selecting the appropriatenumber of segments.

Based on the goodness of fit measure, the 6-segment model is most appropriate as it has thelowest BIC value (17435). In contrast, the BICmeasures for the 1, 2, and 3 segment models aresignificantly higher than that of the 6-segmentmodel, yet the BIC scores for the 4 and 5-segment model are relatively close.

The classification errors provide strong supportfor a 4-segment model. By definition, a 1-segment model will have no classification errors.However, as the number of segments increase,so does the probability of classification errors.For example, all else being equal, a 4-segmentmodel should have a higher classification errorthan a 3-segment model. However, in this study,the 4-segment model actually has fewer sucherrors relative to the 3-segment model.Additionally, the 5 and 6-segment models haverelatively high classification errors, relative to 4-

Fall/Winter 2016 37

segment model. Assessing classification errors lendstrong support for a 4-segment model.

The fit indices and the classification errors resultin a conflict concerning the appropriate numberof segments. Thus, the researchers relied uponguidelines put forth by Garver, et al. (2008) aswell as theoretical judgment to determine theappropriate number of segments.

From a practical standpoint, Garver et al. (2008)suggest limiting the number of segments to fiveor less segments. Aligned with practitionerguidelines, firms often have trouble oncomprehending, understanding, and focusing onmore than five segments.

From a theoretical standpoint, the 4-segmentmodel has clearer theoretical implications foracademic researchers and practitioners. Afterexamining the 4 and 6-segments models the 6-segment model does not provide true theoreticaldifferentiation among the segments. Morespecifically, the 6-segment does not truly showdifferent segments, and the results are redundant.In addition, the 4-segment demonstrates moreparsimony, a goal of all scientific endeavors.With this in mind, in addition to theclassification errors, the 4-segment model wasselected as most appropriate.

For the 4-segment model, each of the clusterswas of substantial size and the parameterestimates demonstrate that each cluster has aunique and meaningful nature, because thevalues are significantly different across the othersegments. The MD scores for the 4-segmentmodel will now be explained.

Parcel Need-Based Segment Results: Uniqueand Different SegmentsAt this time, differences among segments will bediscussed first, followed by the actual size ofeach segment. Finally, attribute importancescores for each segment will be discussed, whichwill demonstrate the nature of each segment.

Unique and Different SegmentsBefore the segment attribute importance scoresare discussed, it is important to demonstrate thatthe four need-based segments are unique andsignificantly different from one another. Toaccomplish this goal, the Wald statistic is usedwithin LCCA. As can be seen in Table 3, all ofthe 17 MD attributes show a significance levelfor the Wald statistic, which suggests that these17 attributes are significantly different across thefour segments and that these attributes aremeaningful predictors (p< .05) of drivingsegment membership. Essentially, each of the 17attributes has a significantly different attributeimportance score across the four segments.

In addition to the Wald statistics and related p-values, R2 values indicate the amount ofvariance that is explained by each parcel carrierselection attributes for each of the four differentsegments. The R2 values are a guide tosuggesting which attributes are most importantin determining segment membership. Forexample, the top five attributes that are the mostimportant attributes to determine segmentmembership include:

sustainability practices,transit time,financial stability,website usefulness, andinformation sharing capabilities.

Table 4 summarizes the importance scores foreach attribute for each segment.

Overall View of the SegmentsSegment 1: The Essentials SegmentSegment 1 tends to focus on those criticalattributes that are the foundation of parcelservices. Segment 1 places the most importanceon the following attributes.

Delivers shipments when promisedTransit time (speed)Competitive ratesPicks up shipments when promisedEffective tracking systemsAvailability of service

Journal of Transportation Management38

In addition, Segment 1 places significantly moreimportance on these attributes than other parcelcarrier segments. Segment 1 is the most pricesensitive segment, yet also placing the highestpriority on transit time speed.

Segment 2 – Dependability SegmentWhile Segment 2 places high priority on thebasics of parcel carrier shipping services

(delivered when promised, transit time, etc.), thissegment is different from other segmentsbecause they place more importance on thefollowing attributes:

Availability of serviceAbility to adjust to customer’s needsInvoice accuracyOverall reputation of carrierSecurity practicesDamage record

Fall/Winter 2016 39

Relative to other segments, Segment 2 places thehighest amount of importance on issues thatattest to the parcel carrier’s overalldependability: availability of service, ability toadjust to customer’s needs, invoice accuracy, andoverall reputation of the carrier. In essence,Segment 2 is defined by these differentiating

attributes that engender customer trust in thecarrier’s important capabilities.

Segment 3 – Tech SegmentSegment 3 is very similar to segment 1, yet onekey difference can be noted. Examiningsimilarities first, Segment 3 places significantly

Journal of Transportation Management40

higher importance on the following attributes, whichis consistent with segment 1:

Delivers shipments when promisedTransit time (speed)Competitive ratesPicks up shipments when promisedEffective tracking systems

In addition, Segment 3 places significantlyhigher importance on “usefulness of thewebsite” (4.7). Thus, given the significantlyhigher importance placed on tracking andwebsite, the researchers conclude that thissegment is more information driven.

Segment 4 – Balanced SegmentSegment 4 is very different from the othersegments. First, Segment 4 possesses morebalance in the importance placed on a widenumber of parcel carrier selection attributes.Second, they place significantly moreimportance than the other segments on thefollowing attributes:

Overall reputation of carrierSecurity practicesDamage recordFinancial stabilityRelationships with carrier personnelSustainability practicesInformation sharing capabilities

Three observations can be noted. First, Segment4 places much more importance on imagerelated attributes such as overall reputation,financial stability, and track record of damage.Second, this segment places much moreimportance on recent trends such assustainability and security. Finally, this segmentis more information focused, placing higherimportance on relationship with carrierpersonnel and information sharing capabilities.

DISCUSSION

The results indicate that business parcel shippersconsider the following attributes to be mostimportant when choosing a parcel carrier:delivers shipments when promised, transit time,and competitive rates. While not significantly

important to all parcel shippers, a number ofattributes were important in determiningsegment membership, such as sustainabilitypractices, information sharing capabilities, andwebsite usefulness. Latent class cluster analysisidentified four different business parcel shippersegments that were based on the importanceattribute of discernible variables. The resultingfour-segment model, with its unique nature, wasthe most theoretically sound and parsimoniousmodel of all models tested.

While there are significant differences inattribute importance, the results also indicatecommonalities across segments. For example,the six most important attributes (deliversshipments when promised, transit time,competitive rates, picks up shipments whenpromised, effective tracking systems, andavailability of service) are generally the mostimportant attributes to each segment. However,concerning the six most important attributes,there are significant differences in the level ofimportance across the segments. Hence, theparcel shipping business should not be viewed asa single homogeneous market. Certain attributesare significantly more important to varioussegments that emerged among parcel shippers.

The Essentials Segment (Segment 1) focuses onbasic performance considerations: deliveringwhen promised, transit time, competitive rates,picks ups, tracking, and service availability. It isinteresting to note that The Essentials is the mostprice sensitive segment, yet also places thehighest priority on transit time.

Relative to other segments, the DependabilitySegment (Segment 2) places the highest amountof importance on dependability concerns:availability of service, ability to adjust tocustomer’s needs, invoice accuracy, and overallreputation of the carrier. Likewise, theDependability Segment places significantly moreimportance on their shipments being secure anddamage free.

Fall/Winter 2016 41

The Tech Segment (Segment 3) resembles TheEssentials except that The Tech Segment placesgreater emphasis on website usefulness. Thus,given the significantly higher importance placedon website, the researchers conclude that thissegment might be more driven by technologyand information.

Regarding the Balanced Segment (Segment 4),this segment places much more importance onimage related attributes such as overallreputation, financial stability, and track record ofdamage. The segment membership also placesmuch more importance on recent trends such assustainability and security. Finally, the BalancedSegment is more information-focused, placinghigher importance on relationship with carrierpersonnel and information sharing capabilities.

CONCLUSIONS

The findings in this research provide severalvaluable contributions to transportationliterature. Parcel carriers transport a considerablevolume of high value goods each year. Due tothe growth and complexity of the parcel sector,carriers must have a greater understanding ofbusiness shipper needs in order to be successful.This includes the ability to objectively segmentparcel customers into logical groups.

Latent class analysis is a quantitative approachthat is useful in finding patterns of heterogeneity“related to characteristics of the choice situationand characteristics of the shipper”(Arunotayanun and Polak, 2011, p. 147) toidentify segments of shippers (i.e., customers)that share a common logistics service profile.Latent class cluster analysis results stemmingfrom this research categorized parcel shippersinto four distinct segments and identified siximportant attributes (delivers when promised,transit time, competitive rates, picks up whenpromised, effective tracking, and serviceavailability) that emerged among the differentshipper segments. Academics and practitionersusing the more common practice of treating

shippers as a homogeneous entity would haveobscured these results.

Second, the empirical findings support the viewthat a one-size-fits-all (single segment) supplychain strategy cannot adequately meet allcustomer needs and expectations (Anderson etal., 1997). In addition, the findings illustrateopportunity for carriers (managers) to movebeyond conventional service segments by takinga quantifiable need-based approach inunderstanding and managing shippers. Resultsindicate that there are segments of parcelshippers, like the Balanced Segment above, thatare not as sensitive to time as other shippersegments, so perceptive carriers would benefitby designing an efficient logistics serviceoperation that is reputable and secure andutilizes sustainable practices like consolidationto best serve customers.

These results are consistent with Barratt’s (2004)assertion that:“If customers can be segmented by way of theirbuying behaviour and service needs, thenseparate supply chains can be designed to meetthe specific needs of the various customersegments. Each supply chain will require adifferent strategy and a different culture tosupport that strategy” (Barratt, 2004, p. 34).

Carriers that accurately identify shippersegments can provide a “portfolio of services”that correctly meets the specific needs of eachsegment (Anderson et al., 1997). By predictingshipper desires and behaviors and placingshippers into optimal segments, carriers canadjust their marketing strategy, clarify theirmarketing message, and align their logisticsoperations to better target and serve eachsegment. Better aligned services have thepotential to reduce operating costs and increaseprofit margins.

Third, recent research in logistics/supply chainmanagement has called for using innovative,advanced research methods and statistical

Journal of Transportation Management42

methods. This study attempted to answer that call inseveral ways. First, maximum difference scaling (MD)was used to advance our understanding of theimportance of a broad set of variables in terms ofcarrier selection. These results were then subjected tolatent class cluster analysis and then to decision treeanalysis. As a result of this multi-method analysis, thestory that emerges from the data is different from priorresearch in this topic area. This represents animportant step forward in understanding how shippersselect motor carriers. Future research shouldexamine logistics service models using MD attributeimportance scores and latent class analysis to moreaccurately identify and address the unique needs ofcritical customer segments.

Future research is also needed to corroborate thedifferent segments that manifested in this research.Furthermore, identifying additional attributes anddescriptors for the different segments wouldprovide better understanding of parcel shippersegments. The segment descriptors are key parcelcarrier marketers being able to target differentmarketing mixes to each target segment, so furtherresearch is needed to better describe thedemographic characteristics of each of thesebusiness segments. Other sectors of transportationservice, namely truckload and LTL, might alsoconsist of need-based shipper (customer)segments. Previous research has generallyassumed that these sectors are homogenous,whereas this research and others like it (e.g.,Arunotayanun and Polak, 2011) that examineshipper preferences suggest further investigationinto possible heterogeneity.

In conclusion, while it is still of practicalimportance to pay close attention to shipment type(letter, packets, parcels, freight), volume, weight,route (e.g., residential, rural), haul length, andtransit time; some shippers are more profitablethan others as they are generally more willing topay for high customer service that fulfills specificneeds. This study has illustrated that parcelshippers are not homogenous. Rather, four distinctparcel shipper segments emerged based on specificneeds expressed by the shippers.

Identifying and understanding these customersegments may provide carriers with anadvantage in negotiations with shippers whovalue service characteristics beyond cost.Furthermore, shipper needs may change overtime, just as the business environment canchange (e.g., JIT, Hours of Service, and homedelivery), causing carriers to adjust theirstrategy and approach (Meixell et al, 2008).Consequently, supply chain executives andleaders must understand shipper segments toprovide optimum customer service thatcontinues to meet if not exceed shipper needsand expectations.

REFERENCES

Abshire, R. D. & Premeaux, S. R. (1991),“Motor Carrier Selection Criteria: PerceptualDifferences Between Shippers and Carriers,”Transportation Journal, 31(1): 31-35.

Anderson, M. (2014), “Changing ShoppingHabits Affect Wal-Mart, Amazon,” USAToday, Retrieved March 6, 2015 from: http://www.usatoday.com/story/money/business/2014/02/01/ walmart-amazon-changing-shopping-habits/5085679/

Anderson, D. L., Britt, F. F., & Favre, D. J.(1997), “The 7 Principles of Supply ChainManagement,” Supply Chain ManagementReview, Retrieved March 6, 2015 from http://www.scmr.com/images/01.SevenPrinciples_.pdf

Andrews, K. J. (1998), “Giving The PeopleWhat They Want: How Mass Customization IsTaking Advantage of the CustomerRevolution,” Entrepreneurial Edge, 3,Retrieved March 6, 2015 from http://edwardlowe.org/digital-library/giving-the-people-what-they-want-how-mass-customization-is-taking-advantage-of-the-customer-revolution/

Andrews, R. J. (2011), IBISWorld IndustryReport 49222: Couriers & Local DeliveryServices in the US.

Fall/Winter 2016 43

Arcelus, F. J., & Rowcroft, J. E. (1993), “FreightRates for Small Shipments,” International journalof Production Economics, 30: 571-577.

Arunotayanun, K., & Polak, J. W. (2011), “TasteHeterogeneity and Market Segmentation inFreight Shippers’ Mode Choice Behavior,”Transportation Research Part E: Logistics andTransportation Review, 47(2): 138-148.

Autry, C. W., Skinner, L. R., & Lamb, C. W.(2008), “Interorganizational CitizenshipBehaviors: An Empirical Study,” Journal ofBusiness Logistics, 29(2): 53-74.

utry, C. W., Grawe, S. J., Daugherty, P. J., &Richey, R. G. (2010), “The Effects ofTechnological Turbulence and Breadth onSupply Chain Technology Acceptance andAdoption,” Journal of Operations Management,28(6): 522-536.

Barcos, L., Rodríguez, V., Álvarez, M. J., &Robusté, F. (2010), “Routing Design for Less-Than-Truckload Motor Carriers Using AntColony Optimization,” Transportation ResearchPart E: Logistics and Transportation Review,46(3): 367-383.

Barratt, M. (2004), “Understanding TheMeaning of Collaboration in The Supply Chain,”Supply Chain Management: An InternationalJournal, 9(1): 30-42.

Burks, S. V., Monaco, K., & Myers-Kuykindall,J. (2004), How Many Trucks, How Many Miles?Trends in the Use of Heavy Freight Vehicles inthe US, 1977 to 1997, Atlanta, GA: TruckingIndustry Program, Georgia Institute ofTechnology, Working Paper.

Burnson, P. (2014), “2014 Rate Outlook: Readyto Corral Costs?,” Logistics Management,Retrieved March 6, 2015 from: http://www.logisticsmgmt.com/article/2014_rate_outlook_ready_to_corral_costs

Chen, Z. X., & Sarker, B. R. (2010), “Multi-Vendor Integrated Procurement-Production SystemUnder Shared Transportation and Just-In-TimeDelivery System,” Journal of the OperationalResearch Society, 61(11): 1654-1666.

Christopher, Jr., C. G. (2011), “The EconomicImpact of E-Commerce,” CSCMP’s SupplyChain Quarterly, Q2: 22-23.

Chrzan K. & N. Golovashkina, (2006),“Empirical Test of Six Stated ImportanceMeasures,” International Journal of MarketingResearch, 48(6): 717-740.

Chrzan, K. & Patterson, M. (2006), Testing forthe Optimal Number of Attributes in MaxDiffQuestions, Proceedings of the 12th annualSawtooth Software Conference, Delray Beach,Florida, March 29-31: 63-68.

Cohen, S. H. (2003), Maximum DifferenceScaling: Improved Measures of Importance andPreference for Segmentation, Sequim, WA:Sawtooth Software Inc.

Cohen, S. H. & Orme, B. (2004), “What’s YourPreference?,” Marketing Research, 16(Summer): 32-37.

Coulter, R. L., Darden, W. R., Coulter, M. K., &Brown, G. (1989), “Freight TransportationCarrier Selection Criteria: Identification ofService Dimensions for CompetitivePositioning,” Journal of Business Research,19(1): 51-66.

Ding, J. F., Liang, G. S., Yeh, C. H., & Yeh, Y. C.(2005), “A Fuzzy Multi-Criteria Decision-Making Model for the Selection of CourierService Providers: An Empirical Study fromShippers’ Perspective in Taiwan,” MaritimeEconomics & Logistics, 7(3): 250-261.

Dobie, K. (2005), “The Core Shipper Concept: AProactive Strategy for Motor Freight Carriers,Transportation Journal, 44 (2): 37-53.

Journal of Transportation Management44

eMarketer (2014), “Total US Retail Sales Top $4.5Trillion, Outpace GDP Growth.” Retrieved March6, 2015 from: http://www.emarketer.com/Article/Total-US-Retail-Sales-Top-3645-Trillion-2013-Outpace-GDP-Growth/1010756

Ergun, O., Kuyzu, G., & Savelsbergh, M. (2007),“Reducing Truckload Transportation CostsThrough Collaboration,” TransportationScience, 41(2): 206-221.

Garver, M. S. (2009), “A Maximum DifferenceScaling Application for Customer SatisfactionResearchers,” International Journal of MarketResearch, 51(4): 481-500.

Garver, M. S., Williams, Z. and Taylor G. S.(2008), “Employing Latent Class RegressionAnalysis to Examine Logistics Theory: AnApplication of Truck Driver Retention,” Journalof Business Logistics, 29(2): 233-257.

Garver, M. S., Williams, Z. & LeMay, S. A.(2010), “Measuring the Importance of Attributesin Logistics Research,” International Journal ofLogistics Management, 21(1): 22-44.

Grawe, S. J., Daugherty, P. J., & Roath, A. S.(2011), “Knowledge Synthesis and InnovativeLogistics Processes: Enhancing OperationalFlexibility and Performance,” Journal ofBusiness Logistics, 32(1), 69-80.

Guerrero, M. M., Egea, J. M. O., & González,M. V. R. (2007), “Application of the LatentClass Regression Methodology to the Analysisof Internet Use for Banking Transactions in theEuropean Union,” Journal of Business Research,60(2): 137-145.

Gustafson, K. (2014), “A ‘Tsunami’ of StoreClosings Expected to Hit Retail,” CNBC,Retrieved March 6, 2015 from: http://www.cnbc.com/id/101353168

Haber, J. (2013), “Mode Optimization: Are YouMaking the Right Choice on LTL Shipments?,”Parcel. Retrieved March 6, 2015 from:http://www.parcelindustry.com/ME2/

Hernández, S., Peeta, S., & Kalafatas, G. (2011),“A Less-Than-Truckload Carrier CollaborationPlanning Problem Under Dynamic Capacities,”Transportation Research Part E: Logistics andTransportation Review, 47(6): 933-946.

Jack, E. P., Powers, T. L., & Skinner, L. (2010),“Reverse Logistics Capabilities: Antecedentsand Cost Savings,” International Journal ofPhysical Distribution & Logistics Management,40(3): 228-246.

Jarrah, A. I., Johnson, E., & Neubert, L. C.(2009), “Large-Scale, Less-Than-TruckloadService Network Design,” Operations Research,57(3): 609-625.

Jindel, S. (2010), “Parcel vs. LTL,” Journal ofCommerce, Retrieved March 6, 2015 from: http://www.jindel.com/newsroom/articles/PDFArticles/Parcel%20vs%20LTL_JOC_9.27.2010.pdf

Kent, J. L., Parker, R. S., & Luke, R. H. (2001),“An Empirical Examination of ShipperPerceptions of Service-Selection Attributes inFive Truckload Industry Segments,”Transportation Journal, 41(1): 27-36.

Kent, J. L., & Smith, C. D. (2005), “CarrierSelection Criteria: Differences AmongTruckload Motor Carrier Offerings,” Journal ofTransportation Management, 16(2): 48-63.

Lambert, D. M., Lewis, M. C., & Stock, J. R.(1993), “How Shippers Select and EvaluateGeneral Commodities LTL Motor Carriers,”Journal of Business Logistics, 14 (1): 131-143.

Laudon, K. C., & Traver, C. G. (2012), E-Commerce 2012: Business, Technology, Society.Pearson Prentice Hall.

Lemay, S. A. (1986), “Strategic InformationSystems in LTL Carriers of General Commodities,”Transportation Journal, 25-32.

Fall/Winter 201645

Liker, J. K. (2004), The Toyota Way: FourteenManagement Secrets from the World’s GreatestManufacturer, New York: McGraw-Hill.

Lin, C. C., Lin, D. Y., & Young, M. M. (2009),“Price Planning for Time-Definite Less-Than-Truckload Freight Services,” TransportationResearch Part E: Logistics and TransportationReview, 45(4): 525-537.

Lin, P. and Lee, C. (2009), “How OnlineVendors Select Parcel Delivery Carriers,”Transportation Journal, 48(3): 20-31.

Liu, R., Jiang, Z., Liu, X., & Chen, F. (2010),“Task Selection and Routing Problems inCollaborative Truckload Transportation,”Transportation Research Part E: Logistics andTransportation Review, 46(6): 1071-1085.

McGinnis, M. A. (1979), “Shipper Attitudestoward Freight Transportation Choice: A FactorAnalytic Study,” International Journal ofPhysical Distribution and MaterialsManagement, 10(1): 25-34.

Meixell, M. J., & Norbis, M. (2008), “A Reviewof the Transportation Mode Choice and CarrierSelection Literature,” International Journal ofLogistics Management, 19(2): 183-211.

Morlok, E. K., Nitzberg, B. F. &Balasubramanian, K. (2000), The Parcel ServiceIndustry in the U.S.: Its Size and Role inCommerce, University of Pennsylvania Schoolof Engineering and Applied Science TechnicalPaper. Retrieved March 6, 2015 from: http://www.seas.upenn.edu/~morlok/morlokpage/parcelfullrpt.pdf

Myerson, P. (2012), Lean Supply Chain andLogistics Management, New York: McGraw-Hill.

Orme, B. (2005), “MaxDiff System TechnicalPaper,” Sawtooth Software Research Paper Series,available at www.sawtoothsoftware.com.

Orme, B. (2005b), “Accuracy of HB Estimation inMaxDiff Experiments,” Sawtooth SoftwareResearch Paper Series, available atwww.sawtoothsoftware.com.

Pai, D. R. (2011), “Study of Efficiency Trendsand Productivity Changes in the US TL Carriers:2000–2009,” International Journal of LogisticsEconomics and Globalization,” 3(4): 234-267.

Richey Jr, R. G., Tokman, M., & Dalela, V.(2010), “Examining Collaborative Supply ChainService Technologies: A Study of Intensity,Relationships, and Resources,” Journal of theAcademy of Marketing Science, 38(1): 71-89.

Sarkis, J., Meade, L. M., & Talluri, S. (2004),“E-logistics and the Natural Environment,”Supply Chain Management: An InternationalJournal, 9(4): 303-312.

Stock, Kyle (2013), “The Ample EvidenceBehind a Christmas Shipping Surge,” BloombergBusinessweek, Retrieved March 6, 2015 from:http://www.businessweek.com/articles/2013-12-26/the-ample-evidence-behind-a-christmas-shipping-surge

U.S. Department of Commerce (2013), Retail E-Commerce Sales Report, Retrieved March 6,2015 from http://www.census.gov/retail/index.html#ecommerce

U.S. Department of Transportation and U.S.Department of Commerce (2013), 2012Commodity Flow Survey – Table 1, RetrievedMarch 6, 2015 from http://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/publications/commodity_flow_survey/2012/united_states/table1.html

U.S. Department of Transportation and U.S.Department of Commerce (2010). 2007 EconomicCensus: Transportation – Commodity Flow Survey,Publication EC07TCF-US. Washington, D.C.:Bureau of Transportation Statistics and U.S. CensusBureau, 3.

Journal of Transportation Management46

Vokurka, R. J., & Lummus, R. R. (2000), “TheRole of Just-In-Time in Supply ChainManagement,” International Journal of LogisticsManagement, 11(1): 89-98.

Wen, C. H., Wang, W. C., & Fu, C. (2012),“Latent Class Nested Logit Model for AnalyzingHigh-Speed Rail,” Access Mode Choice,Transportation Research Part E: Logistics andTransportation Review, 48(2): 545-554.

Vermunt, Jeroen K. and Magidson, Jay (2003),“Latent Class Models for Clustering: A comparisonwith K-means,” Canadian Journal of MarketingResearch, 20: 37-44.

Vermunt, Jeroen K. and Magidson, Jay (2005),Technical Guide for Latent GOLD 4.0:Basic and Advanced, Belmont Massachusetts:Statistical Innovations Inc.

Voss, M. D., Page, Jr., T.J., Keller, S. B. &Ozment, J. (2006). “Determining ImportantCarrier Attributes: A Fresh Perspective Using theTheory of Reasoned Action,” TransportationJournal 45(3): 7-19.

Williams, Z., Garver, M. S., & Stephen Taylor,G. (2011), “Understanding Truck DriverNeedBased Segments: Creating a Strategy forRetention,” Journal of Business Logistics, 32(2):194-208.

AUTHOR BIOGRAPHIES

Michael S. Garver is Professor of Marketing at Central Michigan University, garve1ms@cmich.edu. Heearned his Ph.D. from the University of Tennessee, Knoxville. Dr. Garver stays active with the businesscommunity through speaking, consulting, and conducting best practice research. His interests include usingleading edge methods for research in marketing and logistics. Email: garve1ms@cmich.edu

Zachary Williams is Associate Professor of Marketing and Logistics at Central MichiganUniversity, willi1zs@cmich.edu. He received his Ph.D. at Mississippi State University. His primaryresearch interests are security issues, motor carrier selection, and training, development, and careerdevelopment. Email: zac.williams@cmich.edu

Sean P. Goffnett is an Associate Professor of Marketing and Logistics at Central MichiganUniversity, sean.goffnett@cmich.edu. He received his Ph.D. from Eastern Michigan University. Hisresearch interests include humanitarian logistics, leadership-followership, quality, processimprovement, and talent management. E-Mail: Goffn1sp@cmich.edu

Brian J. Gibson is Professor of Supply Chain Management at Auburn University,brian.gibson@auburn.edu. He received a Ph.D. in Logistics and Transportation from the Universityof Tennessee. His primary research interests are in the area of supply chain training & development,performance analysis, and retail logistics. Email: bgibson@business.auburn.edu.