The empirical link between export dispersion and export performance:
A contingency-based approach
Itzhak Gnizy
a,1,*, John W. Cadogan
b, João S. Oliveira
c, Asmat Nizam
daFacultyofBusinessAdministration,OnoAcademicCollege,KiryatOno55000,Israel
bSchoolofBusinessandEconomics,LoughboroughUniversity,Loughborough,UKandHonoraryProfessorandDocent,LappeenrantaUniversityofTechnology, Finland
cSchoolofBusinessandEconomics,LoughboroughUniversity,Loughborough,UK
dUniversitiUtara,Malaysia
ARTICLE INFO Articlehistory:
Received25September2015 Receivedinrevisedform15April2016 Accepted18July2016
Availableonline26July2016 Keywords:
Exportdispersion
Exportcross-functionaldispersionof influence
Exportmarketing Exportperformance
ABSTRACT
Practitionersand scholars pointout thatfirms areincreasingly dispersing theircapabilitiesacross organizationalfunctions.However,itisnotclearwhetherallformsofdispersion,ofanyfunction,resultin thesameconsequences.Thisstudyinitiatesinvestigationintothelinkbetweenthecross-functional dispersionofinfluenceonexportmarketingdecisions(exportdispersion)and exportperformance.
Drawingondatafromasampleof225UKexporters,thefindingssupporttheargumentthatactive participationofnon-exportfunctionsinexport-marketingdecisionsaffectsexportsuccess.However, those performance consequences are dependent on internal and external contingencies. Export dispersionisbeneficialforexportperformancewhentheexportcustomerenvironmentismoreturbulent and,simultaneously,theexporttechnologicalenvironmentismorestableandthefirmhaslowerlevelsof exportinformationsharing.Inallotherscenariosexaminedinthisstudy,greaterlevelsofconcentration ofexportdecision-making(i.e.lowerlevelsofexportdispersion)appeartobemorebeneficialforexport performance.Ourfindingsimplythatthemanagementofthefirm’slevelofexportdispersionisa complextask,wherebythedegreeofexportdispersionpursuedneedstomatchexternalenvironmental andinternalfirmfactors.
ã2016ElsevierLtd.Allrightsreserved.
1.Introduction
Thecurrentstudyaimstoaddressacriticalresearchgapinthe internationalmarketingliteraturebyexaminingtherelationship between cross-functional dispersion of influence on export marketingactivities(henceforth “exportdispersion”)andexport performanceforthefirsttime.Thestudyalsoseekstoexaminethe roleofexternalandinternalfirmcontingenciesinshapingthelink betweenexportdispersionandexportperformance.
Previousinvestigationspinpointcross-functionaldispersionof influence on marketing activities as critical for organizational
performance becauseit enables firms to effectively respondto changes in customers’ needs and market conditions (Krohmer, Homburg, & Workman, 2002; Krush, Sohi, & Saini, 2015).
Nonetheless,thereislackofknowledgeonofhowcross-functional dispersion of influence onmarketing activitiesaffects business successintheexport-specificcontext.Thisisunfortunatebecause itiswidelyrecognizedthatrelationshipsbetweenbusinesssuccess anditspredictorsdonotnecessarilyholdinthe“export-specific” contextofthefirmdue totheidiosyncratic natureofexporting (Boso, Cadogan, & Story, 2012; Cadogan, Diamantopoulos, &
Siguaw,2002).Thus,acorequestionremainsregardingwhether firms should engage all their business functions in export marketingdecisions(i.e.enhancetheirexportdispersionlevels), or whether,alternatively,firms shouldconcentrateexport deci- sion-makinginone(orfew)functionalunit(s).Theanswerisnot obvious:ononehand,greaterdispersionlevelsmayleadtobetter performance.Ontheotherhand,lowerdispersionlevelsmayalso bringbenefitstofirms(Krohmeretal.,2002).Therefore,thelackof
* Correspondingauthor.
E-mailaddresses:[email protected](I.Gnizy),[email protected] (J.W. Cadogan),[email protected](J.S. Oliveira),[email protected] (A.Nizam).
1 Contactauthor. Theresearchwasconductedwhiletheauthorwasapost- doctoral scholar at the School of Businessand Economics at Loughborough University,UK
http://dx.doi.org/10.1016/j.ibusrev.2016.07.002 0969-5931/ã2016ElsevierLtd.Allrightsreserved.
ContentslistsavailableatScienceDirect
International Business Review
j o u r n a l h o m e p ag e :w w w . e l s e vi e r . c o m / l o c a t e / i b u s r e v
informationaboutthenatureofexportdispersion-exportperfor- mancelinkconstitutesasignificantshortfallinknowledge.
Export dispersion decisions present the firm with key challenges, especially in terms of identifying environmental conditions and internal firm attributes that most or least appropriatefordispersionandshapeitsperformanceconsequen- ces.Priorresearchinnon-internationalbusinesscontextsindicates that the benefits of dispersion may depend on environmental contingencyfactors(e.g.,marketdynamism;Krohmeretal.,2002), andsoitisimportanttodeterminewhetherexportenvironments, which are often complex and dynamic, shape the export dispersion-exportperformance relationshipin a fashionsimilar tothatuncoveredinadomesticdecision-makingcontext.Itisalso important to know how to leverage export dispersion, which constitutes a feature pertaining to internal cross-functional interactions,for export success. Critical interfunctionalinterac- tionsareinformationsharingandgoalalignment(Jaworski&Kohli, 1993;Kahn, 2001).Thus,we investigate thoseinternal features thatmayoperatetoenhancetheperformanceoutcomesofexport dispersion.
Theremainderofthisarticleisorganizedasfollows.Wepresent anoverviewofthepertinentliteratureandthenmovetodiscuss thetheoreticalframework,presentourtheoreticalmodel,define key constructs, and develop the hypotheses. Subsequently, we describethemethodologyusedfordatacollection,theprocedures adoptedformeasurevalidationandstructuralmodeltesting,and outline our findings. Finally, we discuss the theoretical and practicalimplications,identifylimitations,andsuggestdirections forfutureresearch.Wethenpresentaconclusionofthestudy.
2.Theoreticalbackground
2.1.Strategicexportdecision-makingprocessandexportdispersion Thefirm’sstrategicdecision-makingprocessisacriticaltheme instrategyandinternationalbusinessresearch(e.g.Papadakis&
Barwise, 2002; Papadakis, Lioukas, & Chambers, 1998). In this context,hierarchicaldecentralization,lateralcommunication,and politicalbehaviorarecriticaldimensionsofthedecision-making processininternationalfirms(Dimitratos,Petrou,Plakoyiannaki,&
Johnson, 2011; Francioni, Musso, & Cioppi, 2015). Hierarchical decentralizationconcernsthedegreeofdisseminationof power withintheorganizationintheprocessofdecision-making.Lateral communicationistheinvolvementlevelofallmajorbusinessunits inthedecision-makingprocess(Papadakis,Lioukas,&Chambers, 1998).Finally,politicalbehavioristheinfluenceofbothinternal actors and external parties on the decision-making process (Elbanna & Younies, 2008; Francioni, Musso, & Cioppi, 2015).
Issuesconcerningthestrategicdecision-makingprocesshavebeen coveredbypreviousexportperformance investigations.Existing studies focus on different aspects of export decision-making, includingdecision-makercharacteristics(Reid,1981),thedegree to which export decisions are taken following a systematic approach (Nemkova, Souchon, Hughes, & Micevski, 2015), or decision-making uncertainty (Raven, McCullough, & Tansuhaj, 1994).However,withtheexceptionoffewstudies(e.g.Cavusgil, Chan,&Zhang2003), researchonhierarchicaldecentralization, lateralcommunication,andpoliticalbehaviorinexportdecision- makingismissingintheliterature.Cross-functionaldispersionof influenceonexportmarketingactivities(i.e.exportdispersion)isa keymanifestationoftheextentofwhichhierarchicaldecentrali- zation,lateralcommunication,andpoliticalbehaviorareadopted by firms in the export decision-making process. Our export dispersioncoreconstructrelatestothehierarchicaldecentraliza- tiondimensionsincehigherexportdispersionlevelsentailgreater distributionofpoweronexportdecisionacrossvariousbusiness
functions(i.e.theyimplyenhancedlevelsofhierarchicaldecen- tralization).Higherexportdispersionlevelsalsobringaboutthe involvementofagreaternumberofmajorbusinessfunctions(e.g.
marketing/sales, manufacturing, and R&D) in export decision- making,thusimplyingheightenedlateralcommunication.Finally, higher export dispersion also entails greater levels of political behaviorbecausevariousbusinessfunctions(i.e.internalactors) activelyparticipateinandinfluencetheexportdecision-making process.
2.2.Theoreticalfoundationandlevelofanalysis
Ourstudyisunderpinnedbycontingencytheory(Donaldson, 2001),whichcontendsthatsuperiororganizationalperformanceis theresultofafitbetweenorganizationalattributesandthecontext inwhichthefirmoperates(e.g.Venkatraman,1989).Specifically, we follow the perspective of fit-as-moderation (Venkatraman, 1989), an approach that is increasingly used byexport perfor- manceresearchers(e.g.Cadogan,Kuivalainen,&Sundqvist,2009;
Lengler,Sousa,&Marques,2013; Navarro-García,Arenas-Gaitán, Rondán-Cataluña,& Rey-Moreno, 2015), which is rootedin the principle that nolevel of a particular organizational featureis universally superior. Rather, the impact of a certain predictor variable onperformance depends on the value(s) of particular moderator variable(s). The level of fit between the predictor variableand themoderatorvariable(s) is a key determinant of performance.Inthepresentstudy,weexaminethelinkbetween export dispersionand export performanceusing export market dynamism, export information sharing and interfunctionalgoal alignmentasmoderators.
Researchers examining the determinants of export perfor- mance typicallyadopteither theexport function or theexport venture level of analysis(Oliveira,Cadogan, & Souchon, 2012).
Studies on export function level examine the overall export performancelevelachievedbytheexportingentity(i.e.theyfocus on firm-wide export performance). Studies on venture level analyseanexportventurewithinthefirmwithanexportventure beingdefined asasingleproductorproductlineexported bya company to a specific foreign market (Cavusgil & Zou, 1994;
Morgan,Kaleka,&Katsikeas,2004;Oliveira,Cadogan,&Souchon, 2012).Afundamentalprincipleoftheorytestingisthatthelevelof analysis needs to match the level at which such a theory is developed(Klein,Dansereau,&Hall,1994).Ourstudyfocusesthe relationship between firm-wide cross-functional dispersion of influence on export marketing activities and firm-wide export performance levelsandthereforewe adopttheexport function levelofanalysis.
3.Constructdefinitionsandresearchmodel
3.1.Constructdefinitions 3.1.1.Exportdispersion
Following Krohmer et al. (2002) we conceptualize export dispersion as based on the distribution of power of different organizationalfunctionsoverdecisionsinvariousexport-market- ingdomains (e.g.,pricing, newproductdevelopment,customer satisfaction and service, and advertising). It is defined as “the degreeofcoherencewithanidenticalinfluencedistributionacross all the functional groups” (Krohmer et al., 2002; p. 454).
Importantly,as explainedaboveexport dispersionconstitutesa manifestation of crucial dimensions of decision-making in internationalfirms,namelyhierarchicaldecentralization,lateral communication,andpoliticalbehavior(c.f.Dimitratosetal.,2011;
Francionietal.,2015;Papadakisetal.,1998).
3.1.2.Exportperformance
Business unit performance is described asthe extent of the businessunit’sfinancialsuccess(Vorhies&Morgan,2005).Ample researchunderlines therelevanceof financialperformance asa proxy of export success (Wang & Lestari, 2013). We focus on bottom-line performance, i.e., export profits. The underlying reasoning is that our conceptualization of export dispersion involves a number of export-marketing decisions (e.g. product development,distribution,communication,andpricing)thatare knowntoinfluenceexportprofits(Zou,Fang,&Zhao,2003).
3.1.3.Moderators
Weexamine exportmarket dynamism(externalfactor),and export information sharing and interfunctional goal alignment (internalfactors)asmoderatorsof theexportdispersion-export performance link. Export market dynamism refers to the frequency/rate of export market-related changes. We focus on market dynamismasa moderatorsince prior researchin non- internationalcontextsuggeststhatthebenefitsofdispersionare contingentuponthedegreeofmarketdynamismfacedbyfirms (Krohmer et al., 2002). Furthermore, a growing body of the literature highlights the importance of market dynamism as a moderatoroftherelationshipbetweenexportperformanceandits predictors(e.g.Cadogan,Kuivalainen,&Sundqvist,2009;Lisboa, Skarmeas, & Lages, 2013). We specifically examine two key dimensions of export market dynamism, namelycustomerand technological dynamism (Cadogan, Kuivalainen, & Sundqvist, 2009; Ito& Pucik,1993). Asfor theinternal attributes, export dispersion constitutes a feature of firms’ interfunctional inter- actions and reflects the degree of participation of different functionalareasin export decision-making,a criticalactivityof internationalfirms. Interfunctional interactions that have been studied in the literature include information sharing and goal alignmentacrossbusinessfunctions(Jaworski&Kohli,1993;Kahn, 2001).Weexaminehowthosetwofactorsactascontingenciesof the dispersion-performance relationship in the export context.
Exportinformationsharingreferstothedegreeofwhichthefirm disseminatesexportinformation(e.g.,oncompetitors,customers, andmarkettrends)withinthefirmanditsavailabilitytodecision makers(e.g.Cadoganetal.,2012).Interfunctionalgoalalignment concernstheextenttowhich differentbusinessfunctionswork togethertoattaincommongoals(e.g.Cadoganetal.,2005).
3.2.Exportdispersionandexportperformance
Existingresearchdemonstratespositiveimplicationsofcross- functionalinteractionsonperformance(e.g.Luo,Slotegraaf,&Pan, 2006).Cross-functionalinteractionsarebeneficialtofirmsinthe context of decision-making since they can lead to enhanced coordination,integrationandlearning,spanningoforganizational boundaries, reduced cycle times, and improved new product development (Krohmer et al., 2002). Specifically, different marketing activities require interactions of various functions andthusdispersionofthemarketingdecisioninfluence(Krohmer etal.,2002).
Cross-functionaldispersionofmarketingislinkedtoreduced conflict and increased communication with other functions (Moorman & Rust 1999).In the export context, when multiple functionsparticipateintheexportdecision-makingprocess,they are more likely toexhibit higher commitment totheresulting decisions, thus enhancing the chances of their successful implementation.Inaddition,theparticipationofmultiplebusiness functionsintheexportdecision-makingprocessallowsmultiple points of view tobe considered and leads to thechallenge of assumptionsduetointegratedknowledgeandskills(Moorman&
Rust1999).Specifically,theintegratedexperiencesandexpertise
of employees from different functions and their all-inclusive perspectives facilitate the firm’seffectiveness in responding to changes in export environments. Furthermore,cross-functional dispersionofexportdecisionsleadstoareductionincycletimes and enhances theefficient useof resources since,for example, thereis alesserneedfor repetitivediscussions(Engelen, 2011).
Therefore,exportdispersionislikelytohaveapositiveimpacton export-relatedeffectivenessandefficiency,bothofwhichcontrib- utetoenhancethefirm’sexportprofits.
Exportdispersioncanbeseenasariskmanagementtoolinthe exportdecision-makingprocess.Whilemanaginguncertaintyisa keyobjectiveforinternationalfirms(Ghoshal,1987), behavioral uncertaintyisacriticaltypeofriskthatthesefirmsface(Miller, 1992).Forexample,managersatvariouslevelsoftheorganization are often faced with incentives to behave opportunistically to enhancetheirpersonalwelfareattheexpenseofthefirm’ssuccess (Jensen & Meckling, 1976; Miller, 1992). However, greater participation of individuals from multiple functional areas in export decision-making (i.e.greaterlevelsofexport dispersion) canreducebehavioraluncertaintyasitimplieslessrelianceona smallnumberofdecision-makers.Assuch,exportdispersioncan contributetoreducedbehavioraluncertainty withregardtothe firm’sexportactivities.
Conversely, lower levels of dispersion (i.e. greater levels of concentrationofdecision-makinginone/fewfunctions)mayresult inlesseffectivedecisionsthatarebasedon‘territorialviewpoints', reflectanarrowerworldviewandlackasuperordinateattention (Cadogan et al., 2005). Furthermore, the concentration of the responsibilityforexportmarketingdecisionsandactivitieswithin onegroupofspecialistsinthefirmisconcomitantwithproblems suchasreducedinformationsharingamongfunctions,andinter- functional conflict (Moorman & Rust 1999). In addition, lower levelsofexportdispersioncanleadtohigherlevelsofbehavioral uncertainty,astheywilllikelyimplygreaterrelianceonasmaller numberofdecision-makersforexportdecisions,therebyincreas- ingthechancesofopportunisticbehavior.Therefore,wehypothe- sizethat:
H1.Thegreaterthefirm’sdegreeofexportdispersionthegreater thefirm’sexportperformance.
3.3.Moderatingeffectsofexportdispersion-Performancelink Environmentaldynamismmoderatesthelinkbetweenexport decision-makingfactorsandexportperformance(GlennRichey&
Myers2001).Krohmeretal.(2002) findthatmarketdynamism negatively moderatestherelationshipbetweencross-functional dispersionofinfluenceonmarketingactivitiesandperformance.
Thesamemayalsoapplyintheexportcontext.Activeparticipation of multiple business functions in the export decision-making process (i.e.greaterlevelsof exportdispersion) arelikelytobe time-consuming, hence reducing the firm’s ability to adapt to changes intheexportenvironment ina timelyfashion.Greater levelsofexportmarketdynamismrequirethefirmtobequickerto respond to environmental change (Rose & Shoham 2002).
Accordingly, it is likely thatexport dispersionwillbecome less beneficialforperformanceasexportmarketdynamismrises.Thus, wehypothesizethat:
H2. Export market dynamism negatively moderates the relationship between export dispersion and export perfor- mance.
Cadogan et al. (2005) identify exporting interfunctional interactionsascriticalfactorsintermsofdeterminingthesuccess ofexportmarketingdecisionsandstrategy.Specifically,theyargue thatitisimportantforcommunicationandinformationsharingto
take place so that functional areas other than exporting can understand the context that exporting decisions are made in.
Communication and information sharing will be particularly relevantwhenthosefunctionsaremoreinvolvedinformulating those decisions. The need for information sharing is, therefore greater under conditions of high export dispersion. Hence, we hypothesizethat:
H3. Export information sharing positively moderates the relationship between export dispersion and export perfor- mance.
Furthermore, it is also widely recognizedthat alignment of departmental goals is vital in order to eliminate conflict and facilitateeffectivedecision-making.Researchindicatesthatlackof alignmentbetweendecision-makingpartnersisdysfunctionalfor decision-making,andcanleadtoopportunisticbehavior(suchas overstatingone’sowndepartment’sneeds),hostility,anddistor- tionofinformation.Theseactivitiesaremostlikelytonegatively influenceexportsuccesswhenmultipledepartmentsareinvolved in export decision making.However, goalalignment allows for healthy interactions and challenges of assumptions (Menon, Bharadwaj,& Howell,1996), and enables the firm to reapthe benefits of multiple worldviews brought by higher levels of dispersion.Accordingly,wehypothesizethat:
H4. Interfunctional goal alignment positively moderates the relationship between export dispersion and export perfor- mance.
3.4.Summary
OurconceptualmodelisdepictedinFig.1andsummarizesthe relationshipsbetweentheconstructs.Cross-functionaldispersion ofinfluenceonexportmarketingactivitieshasapositiveimpacton export performance. The dispersion-performance link is
moderatednegativelybyexportmarketdynamism,andpositively byexportinformationsharingandinterfunctionalgoalalignment.
Control variables to monitor for potential confounds are also included in the model. The literature suggests firms’ export resourcesand export experienceas criticalexport performance predictors (Morgan, Kaleka, and Katsikeas, 2004). Hence, we includethosepredictorsascontrols.
4.Researchmethodology
Weuseddatafromacross-sectionalmailsurveyofexporting UK companies for the model testing. We adopted the key informantapproach,whichisinaccordancewithmethodologies deployedincurrentexportperformancestudies(e.g.Zeritietal., 2014).
4.1.Samplinganddatacollection
Wedrewarandomsampleof1679exportingfirmswith50or more employees fromthe Kompass UK database.The choiceto focusonlyonsuchfirmsisbecauseourinternalfirmmoderators correspondtointer-functionalphenomena.Wethereforeselected firmswitha“largeenough”numberofemployees(Cadoganetal., 2001)tobeabletocapturesuchconstructsinaprecisemanner.We pre-notifiedtheselectedfirmsbytelephonetoconfirmwhether the contact details we had were accurate, to identify an appropriate key respondent, and torequest participation.Tele- phonepre-notificationledtotheidentification of918managers who agreed,inprinciple, toparticipateand metkey informant knowledgeability criteria. We received 277 responses. The discrepancy between the number of questionnaires sent after telephone pre-notification (918) and the number of responses received (277)is common in export performance studies (e.g.
Costa, Lages, & Hortinha, 2015; Nemkova, Souchon, Hughes, &
Micevski,2015;Zeritietal.,2014),andprobablyreflectsthefact
Fig.1.ConceptualModel.
that a proportion of respondents agreed to receive thesurvey without real intention to complete it. Subsequent screening revealedthat23outofthe277receivedresponsescorresponded tocompaniesthatwereineligibletoparticipate,therebyreducing thenumberofeligiblefirmsfrom918to895.Furthermore,ofthe 277 responses obtained,11 were discardedbecause of missing data,18indicatedtherespondent’srefusaltoparticipate.Ourfinal sampleconsistedthereforeof225usableresponses(i.e.277–23– 11–18),whichcorrespondedtoaresponserateof25.1%(i.e.(225/
918)x100).Ourresponserateisinaccordancewithresponserates reportedbyotherexportperformanceinvestigations(Sousaetal., 2008). A battery of t-tests that compared early and late respondentswith regard to ourstudy's constructs revealed no significantdifferencesatthe5%level,whichindicatesthatnon- responsebiasisnotlikelytobeaprobleminthisstudy(Armstrong
&Overton,1977).
4.2.Sampleprofile
Theaveragefirminthesamplehad350employees,hadbeen exportingfor36years,exportedto28countries,andgenerated39%
ofitssalesfromexporting.Intermsofscopeofexportoperations, 95%ofthefirmsexportedtotheEuropeanUnion(EU),70%toNorth America,69%toAsia,57%toEasternEurope,63%totheMiddle East, 54% to Australia/NewZealand, 49% to Africa, and 37% to South/Central America. Firms operated in a wide variety of industries such as food & beverages, chemical, cosmetics, pharmaceuticals,defense,aerospace,biotechnology,andautomo- tive. Respondents occupied top management positions suchas CEOs,chairpersons,salesdirectors,exportmanagers,andexport executives.
4.3.Measurementscales
We took the measures of the constructs essentially from previousstudies(AppendixA).Thescaleforexportdispersionwas adaptedfromKrohmeretal.’s(2002)measureofcross-functional dispersionofinfluenceonmarketingactivities.Themeasurewas constructed by first assessing the influence of five functional groups (export function, marketing/sales, finance/accounting,
manufacturing,andR&D)overeightstrategicdecisionsconcerning exportmarketingactivities,namely(1)pricing;(2)newproduct development; (3) customer satisfaction improvement; (4) cus- tomer service and support; (5) customersatisfaction measure- ment;(6)distributionstrategy;(7)expansionintonewmarkets, and; (8) market advertising decisions. 100-point constant-sum scales were used. The list of strategic decisions adopted is consistentwithpriorresearchondispersionofmarketingactivities (Krohmeretal.,2002).
Wethen computedthestandarddeviations ofeach of those influencescores.Intheextremecaseofidenticalinfluenceacross all functionalgroups regarding a certaindecision, the standard deviation of the corresponding influence score equals zero, indicating maximum dispersion of influence across functional areasregardingthedecision.Wethenaveragedtheeightstandard deviationsandmultipliedtheresultingfigureby 1.Theresulting scorewasthenadjustedtocreateourfinaldispersionindex,where 0 and 44.72 correspondto minimum and maximum of export dispersion,respectively.
We measured export performance using an adaptation of Cadoganetal.’s(2005)export profitperformance scale.Specifi- cally, using a 10-point scale (1=“very dissatisfied”; 10=“very satisfied”),weaskedmanagerstoratetheirsatisfactionwiththe firm’sexportprofitabilityoverthepreviousthreeyears,aswellas the profitability of their firm’s export operations during the previousfinancialyear.
Wemeasuredtwosourcesofexportmarketdynamism,namely exportcustomerdynamismandexporttechnologicaldynamism.
Weassessedtheseconstructs viatwo-itemscalesadaptedfrom JaworskiandKohli’s(1993)‘marketturbulence’and‘technological turbulence’measures,respectively.Weusedthefirm’sdegreeof exportmarketintelligencedissemination(Cadoganetal.,2001)as aproxyforfirm’slevelofexportinformationsharing.Weassessed interfunctionalgoalalignmentviaanadaptationofCadoganetal.’s (1999)‘coordinationmechanism’scale.Weusedthelogarithmic transformation of the firm’s number of employees directly involved with export matters as a proxy for export resources (Thirkell & Dau,1998). We assessed export experience via the logarithmictransformationofthenumberofyearsthefirmhad beeninvolvedinexportoperations(Bijmolt&Zwart,1994).
Table1
Modelfitindicators,correlationmatrix,andscaleproperties.
Model x2(d.f.) p-value Dx2(d.f.) RMSEA CFI NFI NNFI StandardizedRMR
Measurementmodel 108.93(79) 0.01 – 0.04 0.97 0.92 0.96 0.04
Structuralmodels
-Model1(constrainedmodel)a 25.35(14) 0.03 – 0.06 0.95 0.92 0.72 0.03
-Model2(unconstrainedmodel)b 10.57(10) 0.39 14.78(4)c 0.02 1.00 0.97 0.97 0.01
Measures 1 2 3 4 5 6 7 8
1.Exportdispersion –
2.Exportperformance 0.16* –
3.Exportcustomerdynamism 0.22* 0.11 –
4.Exporttechnologicaldynamism 0.20* 0.18* 0.59** –
5.Exportinformationsharing 0.07 0.27** 0.05 0.08 –
6.Interfunctionalgoalalignment 0.19* 0.24** 0.01 0.20* 0.27** –
7.Exportresources 0.26** 0.24* 0.19 0.23* 0.04 0.04 –
8.Exportexperience 0.17* 0.17 0.11 0.08 0.06 0.15 0.16 –
Mean 17.41 5.56 3.63 4.02 4.75 4.91 19.00 36.32
Standarddeviation 8.47 1.78 1.35 1.39 1.26 1.05 32.63 31.09
Compositereliability N.A. 0.79 0.68 0.77 0.78 0.83 N.A. N.A.
Averagevarianceextracted N.A. 0.65 0.52 0.62 0.53 0.62 N.A. N.A.
* Correlationissignificantatthe.05level.
** Correlationissignificantatthe.01level.
aSquaredmultiplecorrelationcoefficient=.17
b Squaredmultiplecorrelationcoefficient=.25.
cRelativetoModel1(constrainedmodel),Model2exhibitsasignificantimprovement(decrease)inchi-square.
4.4.Reliabilityandvalidity
We assessed measurement scale reliability and validity via confirmatoryfactoranalysis(CFA).Weenteredallreflectiveitems simultaneouslyintoaCFAusingLISREL8.80(Jöreskog&Sörbom, 2007).Weadoptedthemaximumlikelihoodestimationprocedure.
Modelfitwasassessedviatheconventionalchi-squarestatistic,as wellasotherkeyfitheuristics(Diamantopoulos&Siguaw,2009).
Table1shows themeasurementmodelstatistics aswellasthe correlationsamongtheconstructs.
AsdepictedinTable1,keyfitindicatorsofthemeasurement modelwerewithinrecommendedthresholdssuggestinggoodfit withthedata.Allaveragevariancesextracted(AVE)surpassedthe squaresof thecorrelations betweenlatentconstructs providing supportforthediscriminantvalidityofthemeasures.Inaddition, compositereliabilitieswerelarge(>0.60),aswereAVEs(0.50).
Hence,themeasuresexhibitedsufficientconvergentanddiscrimi- nantvalidityformodeltesting.
4.5.Commonmethodvarianceassessment
Following established guidelines (Chang, Witteloostuijn, &
Eden,2010),weimplementedanumberofresearchdesign-related proceduresaimedatminimizingthepotentialforcommonmethod variance(CMV),suchasvaryingthelengthofLikert-typemeasures used to assess different constructs, using reverse coding, and adoptingobjectivemeasurementinstruments.Tofurtherrule-out CMV,weranHarman’sonefactortestinCFA.Thetestyieldeda poormodel fit (chi-square=783.15, p=0.00, d.f.=104, RMSEA= 0.17, NNFI=0.42, CFI=0.50) suggesting that no single factor is responsibleformostofthevarianceinthemeasures.Furthermore, we ran Lindell and Whitney’s(2001) methodmarker test. We adoptedthemarkervariable“howsuccessfulareyourcustomersin yourkeyUK(domestic)marketsinnegotiatinglowerpricesfrom you?”, which is theoretically not linked to any of ourmodel’s constructs.Wedidnotdetectanysignificantcorrelation( 0.09 r0.15)suggestingthatCMVisnotlikelytobeanissue.Lastly, our model contains a number of relationships that are not straightforward(e.g.moderatingeffects).Assuch,itwouldhave beenarduous(ifnotimpossible)forrespondentstoformmental models and predict the relationships being examined. We thereforeconcludedthatCMVwasnotlikelytobeaproblemin thisresearch.
5.Analysesandfindings
5.1.Descriptiveresults
Table2showstheinfluencescoresofthedifferentfunctional groupsonthedifferentexportmarketingdecisionactivitiesusedto compute our export dispersion measure. The scores provide a numberof insights.First,theexportfunctionexhibitsrelatively high scores across all eight decisions. Therefore, the export functionisinapositiontocommunicateitsvaluerelevanceacross the overall spectrum of export decisions. Second, there are no exportmarketingdecisionsinwhichcross-functionaldispersionis eithermaximal(i.e.inwhichinfluenceisdistributedequallyacross allthefunctional groups)orminimal(i.e.inwhich influenceis completelyconcentratedinonefunctionalgroup).Thisindicates thatthereisnodecisiontakencompletelywithoutinputfromthe exportfunction.Third,majorfunctionalareaswithinthefirmhave somedegreeofinfluenceineachoftheexportdecisionsexamined.
It can beconcluded, hence, that export decisions tendto span functionalboundaries,beinginfluencedbothbyexport-andnon- exportdedicatedfunctionalareas.
5.2.Structuralmodelestimation
Followingestablishedprocedure(Songetal.,2005)westarted by mean centering the raw scores of predictor variables. Such technique contributesto diminishpotential problems ofmulti- collinearityassociatedwiththeinclusionofinteractiontermsin themodel(Aiken&West,1991).Weusedtraditionalproduct-term analysistotestformoderationeffects(Ping, 1995).Ourmoderation hypothesesarguedthattheexportdispersion-performancelinkis moderatedbyexportmarketdynamism(H2),byexportinforma- tionsharing(H3),andbyinterfunctionalgoalalignment(H4).Thus, wecomputedtherequiredmultiplicativetermsandenteredthem intothemodelequation.Toensuremodelparsimonyweentered direct effects (namely, of export dispersion, export customer dynamism, export technological dynamism, export information sharing,andinterfunctionalgoalalignment)ascontrols(Aiken&
West 1991). We also included export resources and export experienceascontrols.
We used single indicants to estimate interactions between latentconstructs(Ping,1995).Thisprocedureisrecommendedto lessen model complexity (Jaccard &Wan, 1996).We computed singleindicantsforexportmarketdynamism(bothforcustomer dynamism and for technological dynamism), for export
Table2
InfluenceofDifferentFunctionalGroupsonExportMarketingDecisions.a
ExportMarketingActivities Influenceoffunctionalgroup Issue-specificcross-functionaldispersion
ofinfluenceb Export
Function
Marketing/
Sales
Finance/
Accounting
Manufacturing R&D Sumof Influence Newproductdevelopmentfor
exportmarkets
24 32 8 13 23 100 50.5
Exportcustomerserviceandsupport 39 33 6 15 7 100 44.4
Exportpricingdecisions 33 37 15 10 5 100 44.0
Exportcustomersatisfaction improvement
39 32 5 17 7 100 43.5
Exportdistributionstrategy 43 38 8 8 3 100 34.6
Expansionintonewexportmarkets 39 46 7 3 5 100 33.2
Exportcustomersatisfaction measurement
43 40 5 9 3 100 31.0
Exportmarketadvertising 34 52 9 2 3 100 30.3
aDecisionsonexportmarketingactivitiesaresortedbythelevelofcross-functionaldispersionofinfluenceonexportmarketingactivities;inthequestionnaire,adifferent orderwasgiven.
b Thismeasureisbasedonthemeanofthestandarddeviationsofinfluenceacrossthefivefunctionalgroupswhichwerecomputedforeachofthesampledfirms.We subsequentlyrescaleditsothat0equalsminimalissue-specificcross-functionaldispersionofinfluenceand100equalsmaximaldispersion.
information sharing,and for interfunctional goalalignment via averagingthecorrespondingmeasurementitems.Exportresour- cesandexportexperiencewerealreadymeasuredviasingle-item scales.Export performance was modeled asa first-orderlatent variableofitstwoitems.Wesettheerrorvariancesofeachsingle indicantlatentvariableat[(1
a
)s
2],wherea
correspondsto theconstructreliability ands
tothestandard deviationof thesingleindicant(Jöreskog&Sörbom,1993).
In order to set the loadings and error variances of the interaction terms we followed guidelines established by Ping (1995).Specifically,weranaCFAmodelinwhichthedependent latent variable and all the latent variables involved in the interactions (namely, export dispersion, export customerdyna- mism,exporttechnologicaldynamism, exportinformationshar- ing,andinterfunctionalgoalalignmentwereincluded:however,in thisinstance,thelattervariablescontainedonlyasingleindicant, sincethisgreatlysimplifiestheestimationprocedure.Theloadings ofthesingleindicantsweresetat1 andtheerrorvariancesat [(1
a
)s
2] (Jöreskog & Sörbom, 1993). We recorded thestandardizedestimates fromthisCFA and pluggedthesevalues intotheequationsprovidedbyPing(1995)toproduceestimatesof theloadingsfortheinteractionterms’loadingsanderrorvariances.
WethenrantwostructuralmodelsinLISREL8.80(Jöreskog&
Sörbom,2007),namelyaconstrainedmodelandanunconstrained model. In the constrained model we permittedonly thedirect effectstobeestimatedfreelyandfixedinteractiontermsatzero.In the unconstrained model (Model 2, Table 1) all effects were estimatedfreely.We estimatedthefollowing equationfor both models.
Exportprofitperformance=ɣ1exportdispersion+ɣ2exportdisper- sionxexportcustomerdynamism+ɣ3exportdispersionxexport technologicaldynamism+ɣ4exportdispersionxexportinforma- tionsharing+ɣ5 export dispersion xinterfunctional goalalign- ment+ɣ6 export customer dynamism+ɣ7 export technological dynamism+ɣ8 export information sharing+ɣ9 interfunctional goalalignment+ɣ10exportresources+ɣ11exportexperience(1) Asoutlinedabove,intheconstrainedmodelweallowedonly thedirecteffectstobeestimatedfreelyandfixedinteractionterms atzero.Accordingly,intheconstrainedmodelwesetɣ2,ɣ3,ɣ4,and ɣ5 equal to zero and allowed the package to estimateall the remainingcoefficients.Intheunconstrainedmodel,allcoefficients ofEq.(1)wereestimatedfreely.AsshowninTable1,thereduction inchi-squarewhichresultedfrommovingfromtheconstrainedto the unconstrained model was statistically significant (
D
c2 (d.f.)=14.78(4), p<0.05). Theunconstrained modelalsoexplained
approximately 47% more variance in the dependent variable relativetotheconstrainedmodel(the R-squarestatistics ofthe constrained and unconstrained models were 17% and 25%, respectively). Hence, theunconstrained model fitted betterthe data than the constrained model. In addition, the chi-square statistic of the unconstrained model was non-significant (
x
2=10.57, p>0.05), and its key fit indicators were within recommended thresholds (RMSEA=0.02; CFI=1.00; NFI=0.97;NNFI=0.97), suggesting an excellent fit with the data (c.f.
Diamantopoulos & Siguaw, 2009). Thus, we decided that the unconstrained model was suitable for purposes of hypothesis testing.
5.3.Hypothesestesting
Table 3 showsthestandardizedcoefficientestimates forthe constrainedmodelandfortheunconstrainedmodel,aswellasthe corresponding t-values. Because our hypothesesare directional (i.e.,theyanticipatepositive/negativedirect/moderating effects) weusedone-tailedteststoassessthemagnitudeandsignificance level of the coefficient estimates (Diamantopoulos & Schlegel- milch,2000).Inone-tailedt-tests,t-valuesaredeemedsignificant atthe5%and1%levelsiftheirabsolutevaluessurpass1.65and2.33, respectively.Asmentionedabove,theestimatesassociatedwith the unconstrainedmodel are theonespertinentfor hypothesis testing.InspectionofTable3revealsthatthecoefficientlinkedto exportdispersionisnotsignificant(ɣ1= 0.05,t= 0.66,p>0.05), indicating the lack of a “main” impact of export dispersionon exportperformance.Nonetheless,thecoefficientsassociatedwith threeoutofthefourproducttermscorrespondingtomoderating effectsontheexportdispersion-exportperformancelink(namely, ɣ2,ɣ3,andɣ4)weresignificant(asdiscussedbelow).SinceH1is nested within suchproduct terms, thelatter will be the“final verdict” regarding of the effectof export dispersion onexport performance.(c.f.Kam&Franzese,2007).Wediscusstheeffectof exportperformanceunderhigh/lowvaluesofthemoderatorsin the Post Hoc Analyses section. The interactionbetweenexport dispersionandexportmarketdynamismissignificantandpositive forcustomerdynamism(ɣ2=0.16,t=2.13,p<0.05),andsignificant and negativefor technological dynamism (ɣ3= 0.16, t= 2.22, p<0.05).H2ishencepartiallysupported.Theinteractionbetween exportdispersionandexportinformationsharingissignificantand negative (ɣ4= 0.16, t= 2.11, p<0.05). H3 is, thus, rejected.
Finally,theinteractionbetweenexportdispersionandinterfunc- tionalgoalalignmentisnotsignificant(ɣ5=0.05,t=0.72,p>0.05).
Hence,H4isnotsupported.Asexpected,thepathrelatingtoexport
Table3
PathCoefficientsandT-valuesofConstrainedandUnconstrainedmodel.
Hypothesis Coefficient ParameterEstimatesandt-Valuesa
Constrainedmodel Unconstrainedmodel
StandardizedEstimates t-Values StandardizedEstimates t-Values
H1 ɣ1 Exportdispersion 0.02 0.25 0.05 0.66
H2 ɣ2 Exportdispersionxexportcustomerdynamism – – 0.16 2.13
ɣ3 Exportdispersionxexporttechnologicaldynamism – – 0.16 2.22
H3 ɣ4 Exportdispersionxexportinformationsharing – – 0.16 2.11
H4 ɣ5 Exportdispersionxinterfunctionalgoalalignment – – 0.05 0.72
Controls
ɣ6 Exportcustomerdynamism 0.03 0.29 0.02 0.24
ɣ7 Exporttechnologicaldynamism 0.10 1.16 0.07 0.77
ɣ8 Exportinformationsharing 0.21 2.57 0.24 3.00
ɣ9 Interfunctionalgoalalignment 0.09 1.03 0.13 1.43
ɣ10 Exportresources 0.21 2.19 0.27 2.83
ɣ11 Exportexperience 0.17 1.83 0.16 1.77
aCriticalt-value(5%,one-tailed)=1.65;criticalt-value(1%,one-tailed)=2.33.
resources(our first controlvariable) is positive and significant (ɣ10=0.27,t=2.83, p<0.01).Contrarilytowhat weanticipated, the path associated to export experience (our second control variable) is significant and negative (ɣ11= 0.16, t= 1.77, p<0.05).
5.4.Posthocanalyses
Werananumberofpost-hocanalysestoshedadditionallight onourmainfindings.First,weconductedadeeperexaminationof the unsupported “main” effect of export dispersion on export performance(H1).Onecouldarguethattherelationshipbetween exportdispersionandperformance mightbenon-linear.Higher levelsofdispersionimply theparticipationof manyemployees fromvariousfunctions in the export decision-making, possibly leadingtosituationswhere“toomanycooksspoilthebroth”.Such employees have different perspectives and interests, they lack export-marketingcompetences,andexportingtasksmaynotbea priorityforthem. Suchfactorsmayleadtounsophisticatedand ineffectivecompromises in theexport decision-making process and,thustolowerqualitydecisions.Inaddition,higherlevelsof
dispersionmayimplyexcessivecoordinationandmanagerialeffort to come to export decisions, which can be costly and time- consuming. As a result, the export dispersion-performance relationshipcouldpotentiallybecomelesspositiveundergreater levelsofdispersion,indicating anon-linearrelationship.Totest this argument we ran a parsimonious model in which export performanceispredictedbydispersionanddispersion-squared.To assure comparability with our main model, we included the controlvariablesusedinthemainmodel(cf.Aiken&West,1991).
Both dispersionand dispersion-squared yielded non-significant coefficientswithexportperformanceindicating thatthereisno
“main”effectofexportdispersion(neitherlinearnornon-linear) onexportperformance.
Second, we further explored the effects of the moderating variables that came significant. Specifically, we analyzed the exportdispersion-exportperformancerelationshiplinkunderlow andhighvaluesofexportcustomerdynamism,exporttechnologi- caldynamism,andexportinformationsharing(cf.AikenanWest 1991).ResultsappearinFig.2.
Inspection of Fig. 2 reveals that export dispersion is only beneficialfor exportperformance when firms’export customer
Fig.2.Exportdispersion-exportperformancelinkunderlowandhighvaluesofmoderatingvariables.
environmentsare dynamic and, at the same time, both export technologicaldynamismandexportinformationsharingarelow (thissituationisdepictedbythefulllineinquadrantBofFig.2).In alltheothercases,exportdispersionappearstobedetrimentalto exportperformance.
6.Discussion
6.1.Theoreticalimplications
Thisstudysetouttoexaminethecomplexorganizationaland export-marketingphenomenonofcross-functional dispersionof influenceaswellasitsimplicationsonexportperformance.While otherscholarshavestudiedtheconceptofdispersion,ourstudy focusesonexportperformanceoutcomesofexportdispersion,a topicthathasnotbeenaddressedinpriorresearch.Furthermore, we build on existing literature that examines the external environmentasamoderatorofthedispersion-performancelink (e.g.,Krohmeretal.,2002)toexamine,forthefirsttime,theroleof keyinternalfactorsascontingenciesofsuchrelationship.
OurresultsdifferfromtheonesofKrohmeretal.(2002)who foundpositive direct effect of marketing dispersion on perfor- mance.Ourfindingsseemtosupportthenotionthathasbeenput forwardintheliterature(e.g.,Cadogan,Diamantopoulos,&Siguaw, 2002) that the link between overall business success and its predictorsdoesnotnecessarilyholdautomaticallyintheidiosyn- craticcontextof firms’exportactivities.Interestingly,wefound that the role of export market dynamism on the relationship betweenexportdispersionandexportperformanceismixed.On theonehand,exportcustomerdynamismpositivelymoderatesthe exportdispersion-exportperformance link.Hence,whenexport customers’needsandwantschangerapidly,thereisagreaterneed for export dispersionas a mechanism for reduced conflict and enhanced coordinationlevels among businessfunctions (Moor- man & Rust, 1999). In turn, diminished conflict and greater coordinationarecriticalfactorsforthefirmtocopesuccessfully withfast-changingcustomerenvironments.On theother hand, technologicaldynamismnegativelymoderatestheexportdisper- sion-exportperformancelink.Sucharesultmaybeexplainedby thefactthatgreaterlevelsofexportdispersionarelikelytoimply theparticipationoflesstechnologicallyknowledgeablemanagers intheexportdecision-makingprocess.Thisislikelytoslowdown thedecision-making(Krohmeretal.,2002)and,assuchreducethe firm’sabilitytoadapttotechnologicalchangesinatimelyfashion, thelatterbeingacriticalingredientforsuccessintechnologically turbulentexportenvironments.
Surprisingly, we found that export information sharing negativelymoderatestheexport dispersion-exportperformance link. A potential explanation couldbe that the combination of greaterdegreesofexportdispersionwithhigherexportinforma- tionsharinglevelsgeneratesaproblemofinformationoverload,as peoplefrommultiplebusiness functionswillneedtofilterand assimilate greater amounts of export-related information. This mayslowdowntheexportdecision-makingprocess,reducingthe firm’sabilitytoadapttochangesintheexportenvironment.As such,thebenefitsofexportdispersionmaybereduced.
Furthermore, we found no empirical support for a positive moderatingeffectofinterfunctionalgoalalignmentontheexport dispersion-exportperformancerelationship.Theremaybe,thus, additionalvariablesthatwerenotincludedinthisresearchwhich affecttheroleofinterfunctionalgoalalignmentinmoderatingthe export dispersion-exportperformance link.The investigationof suchvariables constitutes a potentially fruitful future research direction.Intermsofthecontrolvariables,thestrongandpositive coefficientwefoundfortheeffectofexportresourcesonexport performance furthervalidates previousfindings (e.g.Thirkell &
Dau,1998).Againstourexpectations,wedidnotdetectapositive impactofexportexperienceonexportperformance.Itmaybethat, althoughgreaterlevelsofexportexperiencemakethefirmwiserin its export activities, they also entail higher levels of rigidity, renderingthefirmlesscapableofadaptingtochangesinexport environments.
6.2.Managerialimplications
Ourresultssuggestthatexportdispersionisakeypredictorof export performance. Nonetheless, the degree of usefulness of exportdispersiononexportperformanceiscontingentuponthe degree of export market dynamism and onthe firm’slevel of export information sharing.Specifically, asillustrated in Fig.2, export dispersionhasa positive impactonexportperformance whenexportcustomerdynamismishighand,simultaneously,the firmoperateswithmorestabletechnology(s)anddisplayslower levelsofexportinformationsharing.Hence,whilegreaterexport dispersion levelscan bring benefits tothe firmin the form of increased effectiveness and/or efficiency, and of reduction in behavioralriskinthedecision-makingprocess,suchfactorswill only produce positive returnsin the above-described scenario.
Under morestableexport customerenvironmentsand/orwhen thefirmexhibitsgreaterlevelsofinformationsharing,itmaybe preferabletopursuegreaterlevelsofconcentrationintheexport decision-makingprocess.
Thus,itiscriticalthatmanagersassessthedegreeofcustomer andtechnologicaldynamismintheirfirms’exportmarketswhen decidingonhowmuchexportdispersiontopursue.Inaddition,as showninFig.2,theusefulnessofgreaterlevelsofexportdispersion forenhancingexportperformanceishinderedbyexportinforma- tionsharingregardlessofthelevelsofexport customer/techno- logical dynamism. Hence,managersshouldnotcombinehigher levelsofexportdispersionwithinvestmentsinexportinformation.
Wealsorecommendthatmanagersevaluatetheirfirms’levelsof exportinformationsharingwhendecidingonhowmuchexport dispersiontoundertake.Itiscriticaltoemphasizethatthedecision of how much dispersion to undertake in the export decision- making process also needs to account for key industry- and market-related factors not examined in the present study.
Examplesofsuchfactorsincludethespecificindustry(s)inwhich the firmoperates, the percentage of the firm’sexport revenue derived fromBusiness-to-Business(B2B) and from Business-to- Consumer(B2C)activities,orculturaldifferencesfacedbythefirm acrossitsforeignmarkets.
6.3.Limitationsandfurtherresearchdirections
First, the dispersion of marketing activities across multiple businessfunctionshasbeenassociatedwithmarketing’sinfluence within the firm (Krush et al., 2015). Accordingly, a potentially fruitful research avenue could be the examination of the relationship betweenfirms’levelsof export dispersionand the degreeofinfluenceofthemarketingfunctionwithinthefirm,as well as the interplay between those two variables in shaping exportperformance.
Second,theliteraturedoesnotprovideacomprehensivelistof thecriticalmarketingdecisionsundertakenbyfirms.Ourmeasure of export dispersionwas adapted fromKrohmer etal.’s(2002) scale of cross-functional dispersion of influence on marketing activities and comprises thedecisionsthatare mostcommonly examinedinstudiesthatconsiderissuesrelatingtothedispersion of decision-making within the firm with regard to marketing activities(Homburgetal.,2015;Verhoef&Leeflang,2009).Thus, whileourmeasureisconsistentwithpriorresearch,itisfarfrom being exhaustive, and future studies may incorporate other
