Does higher technical efficiency induce a higher service level? A paradox association in the context of port operations

(1)

ContentslistsavailableatScienceDirect

The Asian Journal of Shipping and Logistics

HOSTED BY

jou rn al h om ep a g e :w w w . e l s e v i e r . c o m / l o c a t e / a j s l

Original Article

Does higher technical efﬁciency induce a higher service level?

A paradox association in the context of port operations

Ziaul Haque Munim

FacultyofTechnology,NaturalandMaritimeSciences,UniversityofSouth-EasternNorwayandSchoolofBusinessandLaw,UniversityofAgder,Norway

a r t i c l e i n f o

Articlehistory:

Received8August2019

Receivedinrevisedform11February2020 Accepted12February2020

Keywords:

Dataenvelopmentanalysis BootstrapDEA

Portcongestion Benchmarking Mixedmethods Datatriangulation

a b s t r a c t

Researchersandpractitionersarebenchmarkingtechnicalefficiencyofportsandexploringthedriversof highefficiency.Paradoxically,thisstudyarguesthathightechnicalefficiency(TE=1)isnotalwaysessen- tial,butanoptimallevelneedstobeachievedwhilebalancingtheportservicelevel.Thisstudyapplies dataenvelopmentanalysis(DEA)andfreedisposalhull(FDH)methodstoperformefficiencyrankings of38containerterminalsfrom17differentportsin12Asiancountries.Fourterminalsaretechnically efficient(TE=1)inallfrontierapproaches,thereofoneBangladeshi,oneChinese,oneIndianandone Vietnamese.Furthermore,thisstudypresentsacasestudycombiningqualitativeandquantitativedata analysistoinvestigatethecharacteristicsofaporthostinghightechnicallyefficientcontainerterminals.

Thefindingsuggeststhatportswithgrowingthroughput,notinvestingactivelyininfrastructureand equipment,becomehightechnicallyefficientovertime,butthehighertheirtechnicalefficiency,the lowertheirservicelevel.

1. Introduction

Thecontainerisationofportshassignificantlyimprovedoverall portefficiency,asdidefficientcontainerhandlinganddedicated containerships. Because theoperational function of portsis no longerlimitedtoordinarycargohandling,portsnowcompeteto attractcargo,whichhasfosteredcompetitionamongneighbour- ingports.Inaddition,“thederiveddemandformaritimetransport hasevolvedfromademandforthepossessionofgoodstoaninte- grateddemandforthepossessionofgoodsthathavebeenadded value,timely,reliablyandcost-efficiently”(Panayides,2006,p.3).

Thecompetitiveenvironment requiresindividual terminalsand portstoremainefﬁcientinordertosustaininthemarket(Heaver, 2002; Notteboom&Winkelmans, 2001; Robinson, 2002).Thus, benchmarkingtheperformanceof individualport terminalscan providethe‘best-in-class’terminaloperationsaswellasnecessary measurestoimprovetheoverallcompetitivenessofterminalsand ports.Meanwhile,theemergenceofincreasinglyinternationalised productionpatternslinkingnationaleconomiesacrosstheglobe madethemonitoringofport’sperformanceeventougher.

E-mailaddress:[email protected]

PeerreviewunderresponsibilityoftheKoreanAssociationofShippingand Logistics,Inc.

Excessiveportcapacityisagrowingconcernintheportindus- try. AccordingtotheWTO(2016),fortheﬁrsttimesince 1990, globaltradevolumehasdroppedbelowGDPgrowth.Meanwhile, many countries are heavilyinvesting in building new portsor expandingexistingportterminals,basedonthepreviousforecast of worldtradegrowth,which isoftenbased ontheassociation oftradetoGDPgrowthratio.Asthisassociationseemstohave weakenedrecently,particularlyfordevelopedcountries(Munim

&Schramm,2018),theinvestment decisionsin expandingport capacitymayrequirere-evaluation,anditiscrucialtoinvestigate whethernationsareoverinvestinginseaports.Ontheotherhand, portinfrastructureisvitalfortheeconomiesofdevelopingcoun- tries(Munim&Schramm,2018),andtheymightnotbeinvesting enoughintoit.Tothisend, thisstudyfirstscrutinisesthetech- nicalefficiencyof38Asiancontainerterminals.Second,withthe helpof anin-depthcasestudy,this studyexplorestheassocia- tionbetweentechnical efficiencyand portservice levelof high technicallyefficientport/terminals.Nowadays,portstypicallyhave multiplecontainerterminalsand majorstrategicdecisionssuch astheselectionofportgovernance models(Munim,Saeed,and Larsen(2019)andcapacityexpansion(Wiegmans,Ubbels,Rietveld,

&Nijkamp,2002)aremadeontheterminallevelratherthanthe portlevel.Therefore,theﬁndingsofthisstudywillbehelpfulfor therespectiveportmanagersandauthoritiestoevaluatetheircon- tainerterminalscomparedtocompetingones.Atthesametime, thisstudyquestionsblindapplicationsofthefrontierapproaches

https://doi.org/10.1016/j.ajsl.2020.02.001

by-nc-nd/4.0/).

(2)

toportefﬁciencybenchmarkingwithouttakingtheportservice levelintoaccount.

Theremainderofthispaperisstructuredasfollows.Section2 presentsareviewoftheliteraturerelatedtoportefﬁciencybench- markingwithanoverviewofdifferentmodelsappliedintheﬁeld.

ThedataandmethodologyarepresentedinSection3.Section4 presentsthefindings,includingefficiencyrankingsandbootstrap DEAapplication.Anin-depthcasestudyofaporthostinghightech- nicallyefficientterminalsisexploredinSection5.Finally,Section 6concludeswiththecontributiontoliterature,policyimplications andfutureresearchdirections.

2. Literaturereview

Theacademicliteratureonportefﬁciencybenchmarkinghas beenenrichedwithnewtechniquesandtoolsoverthelastthree decades.RollandHayuth(1993),Cullinane,Song,andGray(2002), CullinaneandSong(2003),Estache,delaFe,andTrujillo(2004), Bichou and Gray (2004) and many others have evaluated port performancefromdifferentperspectivesusingdifferentmethods.

Bichou(2006)mentionedthree broadcategoriesofportperfor- mancebenchmarking:individualmetricsand indices, economic impactstudiesandfrontierapproaches. Inthefrontierconcept, theefﬁciencyrange isdenotedbytheupperandlowerlimit of aboundary.StochasticFrontierAnalysis(SFA)andDataEnvelop- mentAnalysis(DEA) arethetwo mostfrequentlyusedfrontier approachesinportefﬁciencyresearch.TheuseoftheFreeDisposal Hull(FDH)methodisalsoevident.

Since theinception by Charnes,Cooper, and Rhodes(1978), DEAhasbeenappliedin many contexts.Portefficiencystudies thathaveappliedDEAincludeRollandHayuth(1993),Barrosand Athanassiou(2004), Wangand Cullinane(2006), Barros(2006), Hung,Lu,andWang(2010)andTongzon(2001),amongothers.The notableapplicationsofSFAstudiesincludeCullinaneetal.(2002), CullinaneandSong(2003),Notteboom,Coeck,andVanDenBroeck (2000)andTongzonandHeng(2005).Inadditiontothese,Cheon (2009)extendtheDEAmodelintotieredDEAmodeltoshapeport efficiencyfromdifferenttypesofglobalterminaloperators(GTOs) perspective.Hungetal.(2010)ismostlikelythefirststudytoapply thebootstrap DEAtomeasuretheoperatingperformanceof 31 portsintheAsia-Pacificregion.Toincorporatetime intheDEA methodwhileanalysingtheefficiencyofcontainerportortermi- nal,Cullinane,Song,Ji,andWang(2004)applytheDEAWindows analysis.

Somestudiescomparetwodifferentfrontierapproachestoeffi- ciencybenchmarking.Forexample,Cullinane,Wang,Song,andJi (2006)comparetheDEAandSFAwhilebenchmarkingthetech- nicalefficiencyofcontainerports,andfindthatDEAmodelsare relativelyrobustcomparedwithdistributionalassumptionsunder SFA.Wang,Cullinane,andSong(2003)andCullinane,Song,and Wang(2005)comparetheDEAandFDHmodeltomeasurecon- tainerportproductionefficiencybutfindthatthesetwomodels leadtodifferentconclusions.Wangetal.(2003)criticiseFDHfor poormanagerialimplications,butDeBorger,Kerstens,Moesen,and Vanneste(1994)arguethatFDHhasbettermanagerialimplications becauseitestimatestheefficiencybasedonobservedproduction unitratherthanahypotheticalfrontier.

Afewnon-conventionalapproachesarealsoevidentinporteffi- ciencybenchmarkingliterature.GonzálezandTrujillo(2008)apply a translogdistance function toanalyse reforms and infrastruc- tureefficiencyinSpanishcontainerports.Haralambides,Hussain, Barros,andPeypoch(2010)introducetheLuenbergerIndicatorto benchmarkportefficiency.BlonigenandWilson(2008)examine theeffectofportefficiencyontradeusingthegravitymodelinthe USAcontext.Cheon,Dowall,andSong(2009)usetheMalmquist

TotalFactorProductivityIndextoanalysethetypologyoflong-term portefficiencyimprovementforworldcontainerports.Theyfind thatscaleefficiencywasoneofthemajorfactorsthatshapeport efficiency,althoughitisaffectedbytheexternaleconomicenvi- ronment.However,someoftheimpactfulstudiessuchasHung etal.(2010),WangandCullinane(2006)andHaralambidesetal.

(2010) reportthat most portsworldwide are experiencing sig- nificantinefficiency.Therefore,thereexistssubstantialroomfor improvementinportefficiencystudiesintermsofmethodology, theoreticalunderpinningsandimplicationsforportmanagement practice.

3. Dataandmethodology

Thisstudyanalysesdataof38Asiancontainerterminalsfrom portsin Bangladesh, China,HongKong, India,Indonesia, Japan, Korea,Pakistan,Philippine,SriLanka,ThailandandVietnam.The primarysourceofthedatais theContainerisationInternational Yearbook(CIY,ContainerisationalInternational,2012).Thesample of38terminalsisselectedbasedoncompletedataavailabilityon theterminallevelandthroughputof400,000TEUsormoreonthe averageof2009and2010containerthroughput(tomakethesam- pleidenticaltosomeextent).Foreachofthecontainerterminals, dataiscollectedoncontainerthroughput,numberofberths,berth length,maximumwaterdepthattheterminal,totalterminalarea, numberofyardgantrycranes,andthetotal numberofgantries (ship-shore and quay). Due to unavailability in CIY, container throughputdata of eight terminals(Dalian Container Terminal, Gateway Terminal India, JNP Container Terminal, Nhava Sheva InternationalContainerTerminal,JakartaInternationalContainer Terminal,KoreaExpressBusanContainerTerminal,GTContainer Terminal,CatLaiTerminal)areupdatedfromothersources.¹Inthe caseofChittagongPort,duetounavailabilityofterminalleveldata inCIY,relevantdataoftheircontainerterminalsarecollectedfrom thePortAuthority.Table1presentsdescriptivestatistics.

Applicationofmultiplemethodsinastudyenhancesthevalidity offindings(Ha&Yang,2017).Suchapplicationsareevidentinport benchmarkingstudies(forexample,Cullinaneetal.,2006;Wang etal.,2003).Hence,DEAandFDHareusedinthisstudytobench- markthecontainerterminal’stechnical efficiencyand compare results.Acontainerterminalistechnicallyefficientifitproduces themaximumthroughpututilisingtheminimumquantityofinputs suchasequipment,infrastructureandtechnology,whencompared

1Dalian Container Terminal: TEU 2010 estimated (3412368)from CIY port levelTEU in2010 using2009marketshare oftheterminal; GatewayTermi- nalIndia,JNPContainerTerminal, NhavaShevaInternationalContainerTerminal:

TEU2010estimated(1801840,685441,1574719respectively)fromIndianPort Association(IPA)portlevelTEUin2010(http://ipa.nic.in/index1.cshtml?lsid=159, accessed on November, 2016) using 2009 market share of the respective terminals. TEU 2009 of JNP Container Terminal (666879) was calculated by deducting combined CIY 2009 TEU of Gateway Terminal India and Nhava ShevaInternationalContainerTerminal fromIPAportlevel2009TEU;Jakarta InternationalContainerTerminal:TEU2009(1445912)fromamasterthesisavail- ableathttps://thesis.eur.nl/pub/33021/Syafaaruddin-D.S.-Evaluation-of-containerterminal-efﬁciency-performance-in-Indonesia-Future-Investment.pdf,accessedon November,2006;KoreaExpressBusanContainerTerminal:TEU2010(2682598) fromPortofBusanContainerStatistics2010availableathttp://www.busanpa.

com/eng/Board.do?mCode=MN0043, accessed on November, 2016; GT Con- tainerTerminal:TEU2010(1970254)updatedfromterminalwebsite,available at https://saigonnewport.com.vn/en/about/pages/throughput-market-share.aspx, accessedonNovember 2016;CatLaiTerminal: TEU 2010(2850000)updated fromterminalwebsite,availableathttps://saigonnewport.com.vn/en/about/pages/

throughput-market-share.aspx,accessedNovember2016.

(3)

Table1

Descriptivestaticsofterminaldata.

Obs Mean SD Min Max

Containerthroughput2009(TEUs) 38 1,621,333 1,648,211 385,521 8,961,785

Containerthroughput2010(TEUs) 38 1,862,017 1,963,093 348,713 10,568,100

Containerthroughputavg.(TEUs) 38 1,741,675 1,792,508 406,345 9,764,943

Berth(n) 38 4.24 2.50 1 10

Berthlength(sq.m) 38 1174.53 709.81 300 3200

Depth(m) 38 13.79 2.26 8.55 17.80

Terminalarea(sq.m) 38 718,353.10 1,108,676 75,400 6,747,319

Yardgantry(n) 38 30.79 24.19 0 101

Ship-shoreand/orquaygantry(n) 38 11.24 8.52 0 36

Obs:numberofobservations,SD:standarddeviation.

toareferencecontainerterminal.Technicalefﬁciencyofacontainer terminalcanbeexpressedas:

Technicalefﬁciency= Actualproductivity

Referenceproductivity(estimatedfrontier) TheapplicationofDEAandFDHasamethodologyforbench- markingstudiesiswidelyacceptedacrossdisciplines.Bothofthe methodsarenon-parametricfrontierapproachesandarebasedon alinearprogrammingframework.InDEA,avirtualfrontierthat deﬁnesthebestinclassisestimatedbasedoninputsandoutputs ofdecision-makingunits(DMUs,inthiscase,containerterminals).

OtherDMUsarethencomparedwiththebestinclasstoestimate thetechnicalefficiency.InFDH, aDMUisefficientifthereisno otherDMUthatproducesthesameormoreoutputbutemploys lessinput.FDHdoesnotestimateahypotheticalfrontierbutismore relatedtoobservingrealinput–outputrelationship.Thus,frontier modelsrequireinputand outputtobeidentified.Toreducethe impactofabnormalfluctuationinportthroughputovertime,sim- ilartoCheon,Dowall,andSong(2010),thisstudyusestheaverage ofcontainerthroughputof2009and2010astheoutput.Theinputs arenumberofberths,berthlength,maximumwaterdepthatthe terminal,totalterminalarea,numberofyardgantrycranes,and thetotalnumberofgantrycranes.Alltheinputsandoutputare selectedbasedonacriticalreviewofinputsandoutputsusedin thepreviousstudiesincludingHungetal.(2010),Yuen,Zhang,and Cheung(2013),andDeOliveiraandCariou(2015).

Tomeasureefﬁciency,DEAcanbeappliedinaninput-oriented and an output-oriented assumption. Input-oriented DEA mea- sures the potential proportionate reduction in input quantity withoutchangingtheoutputquantity.Ontheotherhand,output- orientedDEA exploresthepotentialproportionateexpansionin outputquantitywithoutchangingtheinputquantity.Thisstudy appliesinput-orientedDEAtoﬁndareasofimprovementinport resource utilisation. Two important concepts in DEA are constant return to scale (CRS) and variable return to scale (VRS).

Charnes et al. (1978) introduced the input-oriented constant- return-to-scalemodel(DEA-CCR),andBanker,Charnes,andCooper (1984)introducedthevariable-return-to-scalemodel(DEA-BCC).

Aninput-orientedDEA-CCR,DEA-BCCandFDHcanbewrittenas thefollowingseriesoflinearprogrammingenvelopmentproblem withdifferentconstraints.

min_,ı (1)

s.t. xs−Xı≥0 (2)

Yı≥ys (3)

ı≥0 (DEA-CCR) (4)

eı=1 (DEA-BCC) (5)

ıs ∈{0,1} (FDH) (6)

where s=1, ..., S denotes the number of container termi- nalsthatusex_s=(xs1,xs2,...,xsm)∈R^m₊ inputs toproducey_s=

(y_s1,y_s2,...,ysn)∈Rⁿ₊ outputs (the superscript [] represents transposeofthematrix).Thes-thcolumnsofdatamatricesXandY areformedbycolumnvectorsxsandys,respectively.Non-negative vectorı=(ı1,ı2...ıs)∈R^S₊formsthelinearcombinationsofthe Scontainerterminals.Finally,letthesuitablydimensionedvector ofunityvaluesbee=(1,1,...,1).

FromcomputationalperspectiveDEAandFDHaresimilar.Sim- ilartoCullinaneetal.(2005),Eqs.(1) to(4),(1)to(3)plus(5), and(1)to(3)plus(6)aresolvedtoestimateefficiencyresultsof DEA-CCR,DEA-BCCandFDHmodels,respectively.Theresultsof DEA-BCCmodeldenotepuretechnicalefficiency(PTE).DEA-CCR modeldenotestheoveralltechnicalefficiency,whichconsistsof two components: scaleefficiency and puretechnical efficiency.

WhilecomparingscoresfrombothDEA-CCRandDEA-BCCmodel,if aDMUhasadifferentefficiencyscorethatmeansthattheparticu- larDMUhasscaleinefficiency.Scaleefficiencyofthes-thobserved containerterminalscanbeobtainedby:

SEs=CCRs

BCCs (7)

Thisstudyusesthe‘Benchmarking’packageoftheRsoftware foranalysispurpose.Furtherdetailsonthemethodologiescanbe foundinDeBorgeretal.(1994)andBankeretal.(1984).

4. Analysisandresults

4.1. Technicalefﬁciencybenchmarking

Consideringtheperspectiveofterminalmanagers,theefficiency technologyinthisstudyisdefinedasaminimisationoftheterminal inputslinearprogrammingmodel.Table2presentsthetechnical efficiencyscoresofallcontainerterminalsconsideredinthisstudy for input-orientedDEA-CCR,DEA-BCCand FDHaswellasscale efficiencies.FortheDEAmodels,overalltechnicalefficiency(CCR, mean=0.527)canbedecomposed intopuretechnicalefficiency (BCC,mean=0.869)andscaleefficiency(Scale,mean=0.596).The meanefficiencyscoreoftheaverageofDEA-CCR,DEA-BCCandFDH is0.791.IncontrasttoHungetal.(2010),thisstudyfindsthatthe overallterminalinefficienciesareduetoscaleinefficienciesrather thanpuretechnicalinefficiencies.Therefore,containerterminals haveroomforimprovementsbyadjustingtheirscales,exceptthe scaleefficientterminals(Scale=1).Onthisnote,avalidstrategyfor terminalswithlowscaleefficiencywouldbetoconsidercooper- atingwithotherterminalswithhightechnicalefficiencyinclose proximity.

Apartfromthat,anumberofpointsemergefromTable2.First, consideringthescaleefficiencyscore,fouroutof38Asiancontainer terminalsinthesamplearescale-efficient(10.53%).Second,despite aratherhighaveragepuretechnicalefficiency(BCC,mean=0.869), 52.63%oftheterminalsstillfallbelowtheaveragescore.Onthe otherhand,basedontheefficiencyscorederivedusingtheFDH method,86.84%oftheterminalsareefficient.However,asWang

(4)

Table2

Technicalefﬁciencyof38Asiancontainerterminals^a

No. Containerterminal Port Country DEA-CCR DEA-BCC FDH AVG Scale

1 DalianContainerTerminal PortofDalianAuthority China 0.533 0.798 1.000 0.777 0.668

2 DalianPortContainerTerminal PortofDalianAuthority China 0.463 0.622 1.000 0.695 0.744

3 NanshaTerminal GaungzhouPortGroupCo.Ltd China 0.620 0.923 1.000 0.847 0.672

4 NanshaTerminalPhase2 GaungzhouPortGroupCo.Ltd China 0.450 0.712 1.000 0.720 0.632

5 XingangTerminal GaungzhouPortGroupCo.Ltd China 0.453 0.839 1.000 0.764 0.540

6 XinshaTerminal GaungzhouPortGroupCo.Ltd China 0.357 0.838 1.000 0.732 0.426

7 KwaiTsingContainerPortTerminals1/2/5&9 (south)

HongKongPortDevelopment Council

HongKong 0.616 0.858 1.000 0.824 0.718

8 Cosco-HITTerminals(HongKong)Ltd HongKongPortDevelopment Council

HongKong 0.593 0.871 1.000 0.822 0.681

9 QianwanContainerTerminal QingdaoPort(group)CoLtd China 1.000 1.000 1.000 1.000 1.000

10 ShanghaiContainerTerminalsLtd ShanghaiPort China 0.765 1.000 1.000 0.922 0.765

11 ShanghaiPudongInternationalContainer Terminals

ShanghaiPort China 0.743 1.000 1.000 0.914 0.743

12 GatewayTerminalIndia JawaharlalNehruPort India 0.728 1.000 1.000 0.909 0.728

13 JNPContainerTerminal JawaharlalNehruPort India 0.259 0.829 1.000 0.696 0.312

14 NhavaShevaInternationalContainerTerminal JawaharlalNehruPort India 0.663 1.000 1.000 0.888 0.663

15 PTContainerTerminal TanjungPerakPort Indonesia 0.336 0.858 1.000 0.731 0.392

16 JakartaInternationalContainerTerminal TanjungPriokPort Indonesia 0.252 0.677 0.750 0.559 0.372

17 NabetaPierTerminal NagoyaPortAuthority Japan 0.320 0.860 1.000 0.727 0.372

18 JasungdaeContainerTerminal BusanPort SouthKorea 0.246 0.693 0.912 0.617 0.355

19 KoreaExpressBusanContainerTerminal BusanPort SouthKorea 0.728 0.810 1.000 0.846 0.898

20 PusanNewPortTerminal BusanPort SouthKorea 0.670 0.707 1.000 0.792 0.947

21 U-amContainerTerminal BusanPort SouthKorea 0.249 1.000 1.000 0.750 0.249

22 GwangyangInternationalContainerTerminal GwangyangPort SouthKorea 0.351 0.830 0.933 0.705 0.423

23 KoreaInternationalTerminals GwangyangPort SouthKorea 0.372 0.705 1.000 0.692 0.528

24 KarachiInternationalContainerTerminal KarachiPort Pakistan 0.351 0.851 1.000 0.734 0.413

25 PakistanInternationalContainerTerminal KarachiPort Pakistan 0.329 0.887 1.000 0.739 0.371

26 ManilaInternationalContainerTerminal ManilaPort Philippines 0.716 1.000 1.000 0.905 0.716

27 GTContainerTerminal ColomboPort SriLanka 0.641 0.863 1.000 0.835 0.743

28 EGCTTerminalB2 LaemChabangPort Thailand 0.523 1.000 1.000 0.841 0.523

29 ESCOTerminalB3 LaemChabangPort Thailand 0.490 1.000 1.000 0.830 0.490

30 HLTTerminalC1/C2 LaemChabangPort Thailand 0.178 0.578 0.781 0.512 0.309

31 LCB1TerminalB1&A0 LaemChabangPort Thailand 0.570 0.929 1.000 0.833 0.614

32 LCITTerminalB5&C3 LaemChabangPort Thailand 0.410 0.812 1.000 0.741 0.505

33 TIPSTerminalB4 LaemChabangPort Thailand 0.523 1.000 1.000 0.841 0.523

34 TLTTerminalA2/A3 LaemChabangPort Thailand 0.168 0.687 0.750 0.535 0.244

35 CatLaiTerminal HoChiMinh Vietnam 1.000 1.000 1.000 1.000 1.000

36 CCT-ChittagongContainerTerminal ChittagongPortAuthority Bangladesh 0.366 1.000 1.000 0.789 0.366 37 GCB–Jetty6,9,10,11,12,13 ChittagongPortAuthority Bangladesh 1.000 1.000 1.000 1.000 1.000

38 MundraInternationalContainerTerminal MundraPort India 1.000 1.000 1.000 1.000 1.000

Average 0.527 0.869 0.977 0.791 0.596

Boldindicatestechnicallyefﬁcientcontainerterminal.

aForrobustnesscheckoftheefficiencyranking,technicalefficiencyof30containerterminalsexcludingtheeightwithupdateddata(seefootnote1)wereestimatedusing DEA-CCR,DEA-BCCandFDH(availableuponrequest).Asexpected,efficiencyscoreschangeslightly,butrankingremainsthesamewithaverageDEA-CCR,DEA-BCCandFDH estimateof0.514,0.871,and0.980,respectively.

etal.(2003)mentioned,anFDH-efficientterminalisnotnecessar- ilybetterthanitscounterpartswithlowertechnicalefficiencyand maynotidentifythepotentialforimprovementastheyarealready efficientinFDH.Finally,inlinewithBarros(2006),alltheefficient terminalsinDEA-CCRarealsoefficientinDEA-BCC,indicatingthat scaleefficiencyrepresentstheultimatetechnicalefficiencyscore.

4.2. BootstrapDEA

Beingadeterministicmethod,DEAdoesnotexplicitlyconsider randomerrorandoveralldeviationfromthetechnologyfrontier, which meansthat DEA estimatesmay beaffected bysampling variations.However,onlyafewtechnicalefﬁciencybenchmarking studies in the port industry have taken this into account (for example,Hung et al.,2010; Nguyen, Nguyen, Chang,Chin, and Tongzon, 2016). Simar and Wilson (1998) suggest employing bootstrapping in any DEA application as a standard practice to enhance the reliability of the estimates. Bootstrapping is a computer-basedstatisticalmethodforchecking theaccuracy of statisticalestimates.Thebasicideaofbootstrappingistoreplicate thesamplebymimickingthedatagenerationprocesstorepeatedly estimateparameters,whichinthiscaseisestimatingtheefﬁciency

of container terminals. For detail procedure of bootstrap DEA, see Simar and Wilson (1998). The result of the bias-corrected bootstrap of container terminal technical efﬁciency with 3000 bootstrapreplicatesat95%conﬁdenceintervalisshowninFig.1.

Fig.1showsthattheefficiencyscoreofa containerterminal changeswhencorrectedforbias.Scale-efficientcontainertermi- nalsbecomelessefficientwhencorrectedforbias.Forinstance, QuianwanContainerTerminal(China),MundraInternationalCon- tainer Terminal (India), Cat Lai Terminal (Vietnam) and GCB Terminal(Bangladesh)arescale-efficientcontainerterminals(see Table2),buttheirefficiencylevelsdropduringthebootstrapping process(seeFig.1).However,therankingofterminalsremainsthe sameinTable2andFig.1.Fig.1alsopresentstheupper-bound(2.5%

CI)andlower-bound(97.5%CI)confidenceintervalsfortheesti- matedefficiencyscores,wheretheupper-boundalmostcoincides withtheoriginaltechnicalefficiencyestimates.

5. Technicalefﬁciencyandservicelevel—acasestudy

Theconceptoftechnicalefﬁciencyasadriverofportcompeti- tivenesshasbeenwell-examined(Tongzon&Heng,2005).Existing maritime literatureviews high technicalefﬁciency (TE=1) of a

(5)

Fig.1.BootstrapDEAtechnicalefﬁciencyestimates.

terminalorportproductionasapositivefeature.Whileresearchers arebusy investigating thedriversof port efﬁciency(e.g.Chang

&Tovar,2014;Serebriskyetal.,2016;Tongzon&Heng,2005),it maybethecasethatthehighlyefﬁcientterminalsarenotthemost competitiveonesin termsofservice level,particularly,because theydo not invest activelyin resource expansion (Cullinane &

Wang,2010).Inthesamevein,MerkelandHolmgren(2017)argue thattheusersideofportproductionhasbeenlargelyoverlooked inportefﬁciencystudies.

Fromtheusers’perspective,themajorcomponentsoftheport servicelevelaretheberth-timeofshipsandtheport-timeofcar- goes(Sha&Huang,2010).Similarly,Li,Yu,Tang,Li,andZhang

(6)

(2017)deﬁneportservicelevelastheratiobetweenaveragewait- ingtimeandaverageservicetimeofavesselatport.Anyreduction intimeperiodsoftheseportoperationprocessleadstomoresat- isﬁedcustomers.Moreover,accordingtoYeo,Roe,andDinwoodie (2008),portservicelevelcomprisesofaport’sabilitytorespond tocustomerneedspromptly,a24/7serviceopeninghourandzero waitingtime.Thus,werefertotheportservicelevelastheavail- abilityofberthforvesselsonarrivalattheportandtheabilityto servevesselswithoutanywaitingtime.

5.1. Contextofthecase

Intheprevioussection,wefindthefourtechnicallyefficientcon- tainerterminalsonalloftheappliedfrontierapproaches.Hence, wedesignanin-depthcasestudyfocusingonChittagongPort— oneoftheportsinthesamplehostingtwoofthehighlyefficient containerterminals(GCBandCCTterminals).In2019,Chittagong Portwasrankedthe64thbusiestportintheworld,thatis,one-rank aheadofCartagena(Colombia)andone-rankbehindKobe(Japan).² Givenvariousapproachestocaseselection,weconsiderthethree maincriteria:(1)thematchbetweenthecontextofthecaseand thephenomenonunderinvestigation;(2)availabilityofdata;and (3)accessibilityofinformation.TheChittagongPortofBangladesh certainlymeetsthesecriteria.Suchsingle-casestudiesareappropri- ate“...whenacaseisrevelatory.Thismeansthatwecanobserve andstudyaphenomenonwhichwaspreviouslynotaccessibleand whichcanprovideusefulinsights”(Ghauri&Firth,2009,p.32).

TheChittagongPort(CP)istheprincipalportofBangladesh,han- dlingover90%ofthecountry’simportandexport.Althoughthe portgovernancebody,ChittagongPortAuthority,wasestablished in1976,theexistenceofCPdatesbacktothefourthcenturyBC(see Munim,Saeed,andLarsen(2019fordetailaboutCP).Also,CPhas enormouspotentialtobecomeanintermodalcontainertranship- menthubintheregion(Munim&Haralambides,2018).Similarto thecurrentstudy,CPhasbeenfoundtobehighlytechnicallyeffi- cientbyWuandGoh(2010).However,portservicelevelissuesof CP,suchaslongvesselturnaroundtime,havealsobeenidentifiedas crucialbyDappeandSuárez-Alemán(2016).Thus,CPprovidesan excellentcontextforinvestigatingtheassociationbetweentechni- calefficiencyandportservicelevel.

Ontheinput–outputbasedefficiencyrankingmethodssuchas DEA,FDHandSFA,thecapacity(portinfrastructureandavailabil- ityofequipment)oftheterminaland/orportplaysamajorrole indeterminingefficiencyranks.Whileinadequatecapacityleads tocongestion,abundantbutidlecapacityindicateinefficiency.To achievetheoptimum trade-offbetweeninadequateand surplus capacity,itischallengingtomakecapacityexpansiondecisions.The purposeofthiscasestudyisnottoscrutinisecapacityexpansion decisions,buttoshedlightontheassociationbetweentechnical efficiencyandservicelevel,throughanin-depthcasestudyfollow- ingastep-wiseprocesssuggestedbyEisenhardt(1989).

5.2. Datacollection

Datatriangulation, thatis, collectingdata onthesamephe- nomenonfrommultiplesources,improvesthevalidityofastudy andreducesthelikelihood ofmisinterpretation(Ghauri&Firth, 2009).Thus, we collectdatafromthree sources:(1) two semi- structuredinterviews,(2)10onlinemediasources,and(3)archival documentsfromCPA.Atheoreticalsamplingapproachisemployed toselecttheintervieweesandtocurtainonlinemediadata(see AppendixAfordetail).Thesampleofthepersonalinterviewees

2 Accessed from ‘https://lloydslist.maritimeintelligence.informa.com/one- hundred-container-ports-2019’onFebruary10,2020.

consistsofanemployeeofChittagongPortwithover20yearsof experienceandanemployeeofaninternationalshippinglinecom- panyagentinChittagongwithoversixyearsofexperience.Both intervieweesarecontactedmultipletimesduringthequalitative analysisprocess.Asasecondsourceofdata,10 relevantonline mediasourcesareretrieved.Forreliabilitypurposes,onlinesources offourdifferentstakeholdercategoryoftheport(shippingline, domesticnewsportal,internationalnewsportalandaninterna- tionallogisticsserviceprovider)areselected.Thethirddatasource, archivaldata,consistedofyearlyoverviewreportsandcomprehen- siveyearbooks,publishedbytheportauthority.

5.3. Caseanalysisandﬁndings

WeusetheNVivo11softwareforqualitativeanalysis.Quali- tativetextdataaboutcommontopics arecodedintotwomajor nodes:technicalefficiencyandportservicelevel.Portthroughput andfacilitiesarecodedasachildnodeoftechnicalefficiency,as thesearethemaindecidingvariablesoftechnicalefficienciesin frontierapproachessuchasDEAandFDH. Therestofthechild nodesemergedduringthecontentanalysisofdata,followingan iterativeprocess(seeanoverviewofnodecodinginAppendixB).

DataofCPoneach oftheinputvariables consideredfortechni- calefﬁciencyrankingarepresentedagainsttheoutputvariablein Fig.2,whichshowsthatinvestmentintoportresourcesusually occurinphasesratherthancontinuously.Forexample,CPhad11 yardgantriesduring2007–2012and15in2013.Insteadofbuying onegantrycarneeachyear,CPboughtfourafteranintervalofﬁve years.

Technical efficiency scores of ports and terminals help the respectiveportauthoritiestomakestrategicdecisionsregarding governancemodelchoice,capacityexpansion,marketpenetration etc.InthefrontierapproachessuchasDEA,technicalefficiencyis estimatedbasedonportthroughput(output)andavailableport resources(input).Theratioofinputandoutputisusuallylower inportswithhighertechnicalefficiencythaninmoreinefficient ports.Meanwhile,Merkel and Holmgren(2017) findanegative associationbetweenportproductionefficiencyandpercapitaGDP.

Thisindicatesthatportsfromthelower-incomecountriesaremore technicallyefﬁcient.Beingamajorgatewayportinadeveloping country,thecaseofChittagongPortisnodifferent.

“About90%of thecountry’sexportandimportisdonethrough Chittagongport.[...]Chittagongportisexperiencing16%to17%

growthincargoandcontainerhandlingforthepastfewyears.”

[Hussain,2017,July20]

“In2017,thecontainerthroughputcrossed2.4million.[...]Being asmallport,wearehandlingacoupleofmillionsofTEUs.[...] SuchhightechnicalefﬁciencyisobviousforCPAaswehandlea largevolumeofcontainerswithacomparativelylowernumberof equipment[...]”

[RespondentA,2018,January]

However,theservicelevelattheChittagongPortisquestion- able.ChittagongPorthasfacedcapacityissuesleadingtosevere congestioninhandlingtherapidlygrowingthroughputdemand.

Based onthecombined analysisof qualitativeand quantitative data,portthroughputhasgrowndramaticallyovertheyears,but theportauthoritydidnotundertakecontinuouscapacityexpan- sioninitiatives(seeFig.3).AccordingtoRespondentB,thequality ofequipmentisnotuptostandard,whileaccidentsandunskilled employeesmakethesituationworse.

“InadequateinfrastructureatChittagongporthascreatedsevere congestionofcargovesselsattheouteranchorage.”[Milad,2017, July18]

(7)

Fig.2. Input–outputmagnitudesofcontainerproductivityinChittagongPort.

Source:CPA(n.d.)andHPC(2014).

“Tomakethingsworse,thetwogantrycraneswhichweredamaged followinganaccidentonJune25hassubstantiallydisruptedthe containerhandlingoperationsoftheport....[...]Theporthasbeen facinghugevesselcongestionforthelast twomonths,delaying berthingschedulestomanyshipswaitingattheouteranchorage.”

[Hussain,2017,July20]

Duetothepoorservicelevelattheport,theend-users–the shippers–sufferthemost.Whilecongestionpeakedin2017,evi- denceofchargingsurchargeduetocongestionatCPwasexistentin 2010(EmiratesShippingLine,2010,May21).Themajorcontribut- ingindustrytothecountry’seconomy,theready-madegarments (RMG)industry,wasabouttocollapsein2017.Theestimatedtotal

(8)

Fig.3.Relationshipbetweentechnicalefﬁciencyandservicelevel.

monthlylossofthelocalbusinessescouldbeuptoUSD9.64million duetosurchargesimposedbytheshippinglines(Milad,2017,July 18).

Despitethepoorservicelevelattheport,thestakeholdersdo expresshopeof improvement in thefuture. Recently, the port authorityhastakensomeinitiativestoimprovetheservicelevel, eventhoughtheseshouldhavebeeninitiatedearlier.Theseinitia- tivesinvolvetheprocurementofnewequipmentforbuildingnew terminals.Whiletheportmaybeimprovinginservicelevelfollow- ingsuchinitiatives,therelativetechnicalefﬁciencyoftheportmay dropduetoincreasedcapacityintheshort-term.

“...Inaround1994–95,thecontainerizationstartedattheportbut wasnotﬂourishinguntiltheearly2000s.In2007,theBangladesh Armycreated manyoff-docksto easethepressure ontheport.

[...]During2010–2011,thecongestionsituationwasalittlebit better.Butstartingfrom2013–2014,thesituationhaskeptwors- eninguntilnow.However,manyinitiativesweretakenrecently— expansionofexistinganddevelopmentofnewportterminals... thesituationwillbemuchbetterinthefuture...around2025.”

[RespondentA,2018,January]

“TheChittagongport...received46piecesofcontainer-handling equipment,whichshippershavelongdemandedtoboostcapacity attheoft-congestedportthatisoperatingwellbeyonditsdesigned TEUcapacity.”[Islam,2017,August11]

Asstatedabove,theBangladeshArmyhadtocreateoff-docks tohandlethecontainerthroughputpressureattheportin2007.

Thetechnicalefﬁciencyoftheportatthattimewouldcertainly havebeenhigh.AftertheNewmooringContainerTerminal(NCT) ofChittagongPortstartedoperatingattheendof2007,thesitu- ationimproved,butnonewinitiativeshavebeentakenforafew years,andtheservicelevelstartedtoworsenagainin2013–2014.

Due to the recent initiatives, all of the port stakeholders are expectingbetterservicelevelby2025.Thus,assumingcontinuous portthroughputgrowthovertime,thefollowingpropositionsare derived:

Proposition1. Forportsthatdonotactivelyinvestinresources,the highertheirtechnicalefﬁciency,thelowertheirservicelevel.

Proposition2. Forportsthatactivelyinvestinresources,theasso- ciationbetween theirtechnical efﬁciency andservicelevel remain constant.

5.4. Discussiononthepropositions

Toelaborateonthepropositions,Fig.3presentsthreephasesin portdevelopmentthatexplainstheproposedassociationbetween technicalefﬁciencyandservicelevel.InPhase1,aterminalstarts itsoperationattimePtandexcelsinservicelevelduetothenew moderninfrastructureandsuperstructure, butexperiencesrela- tivelylowtechnicalefﬁciencyascontainerthroughputmaybelow duetothetimeittakesforaterminaltogettheattentionofitsusers.

ThisphasereferstotheconditionofChittagongPortin1994–1995, whentheportstartedhandlingcontainerswithnewlydeveloped berthsandnewly procuredcontainerhandlingequipment(CPA Yearbook,1996–1998).

Phase2referstothesituationatthetimeP_t+iinFig.3,usually 5–10yearsfromthestartofportoperation, whenthecontainer throughputincreasesgradually.Theportincreasesitstechnicalefﬁ- ciency,buttheservicelevelstartstosufferintermsofincreased waitingtimesforberthingcomparedtoPhase1(assumingthatno furtherinvestmenthasbeenmadesincethebeginningofterminal operation).Congestionwithincontaineryardsoftheportmayalso beobservedinthisphase.InthecaseofCP,containerthroughput almostdoubledbetween1995and2000(CPAYearbook,2002).Ves- selturnaroundtimeincreasedfrom4.69daysin1997to5.9days in2000andevenreached7.11daysduring1998(CPAYearbook, 2002).

Withthecontinuousincreaseincontainerthroughput,aport entersintothePhase3,atthetime P_t+i+j inFig.3,usually5–10 yearsfromthePhase2(again,assumingnosubstantialportinvest- mentinrecentyears),wheretheportishighlytechnicallyefﬁcient butsuffersgreatlyintermsofservicelevel.Toremaincompetitive andmaintainhighservicelevel,itisimportantforaporttoinvest inimprovementandexpansionafterreachingP_t+i,andcontinue activeinvestmenttokeeptheassociationbetweentechnicalefﬁ- ciencyandservicelevelconstant.InDecember2006(11yearssince 1995)thevesselturnaroundtimeatCPreachedanewextremeto

(9)

11days(CPAYearbook,2010).Higherturnaroundtimemeansloss ofmoneytoallportusersandmayalsoleadtodamageofgoodsdue tothechaosincontainerhandling.Thereasonbehindsuchlowser- vicelevelisthatCPAdidnotinvestactivelyintheimprovementand expansionoftheportevenafterenteringthePhase3.In2007,the portauthoritymadealargeinvestmentandanewcontainertermi- nalnamedNCTstartedoperation.Thus,thesituationwasexpected toimprovesoonandmovebacktoasimilarlevelasinP_t+i.Indeed, thiswasreﬂectedinareduced turnaroundtimeof2.42daysin June2008(CPAYearbook,2010).Suchexpansiondecisionsshould betakenaheadoftime,beforereachingattheintersectionbetween Phase1and2,basedontheforecastoffutureportthroughput,as buildingterminalsorinstallingnewequipmentinterminalscan typicallytaketwotoﬁveyears.

“Thecurrentportinfrastructureandsuperstructureareexactlythe sameasitwasin2012–2013.The portauthorityhasthe fore- castdataofgraduallyincreasingyearlyportthroughput,andthey weretoincreasecapacityaccordingly.Iwouldsaythatthecapac- itywasalreadylowin2012,andnowin2017,portthroughput increasedalotfrom2012.Itisnomorepossibletohandlethecur- rentthroughputwiththesamecapacityof2012.”[RespondentB, 2018,January]

In recent years of port operations, themanagerial learnings fromthetechnicalefficiencyandservicelevelparadoxduringthe 1995–2007periodwerenotreflectedintheactionsoftheCPA.They hardlyprocuredanynewequipmentanddidnotimplementany substantialexpansionprojectaftertheinaugurationofNCTin2007 (seeFig.3).ThisisalsoreflectedinthestatementbyRespondent Babove.Again,afterfacingasituationlikeinP_t+i+jin2017,CPA ordered10newship-shoregantrycranesfortheNCTterminal,six ofwhicharetobeinstalledinOctober2018(RespondentA,2018, February).WiththecapacityexpansioninitiativestakenbyCPAat theendof2017,itisexpectedthatservicelevelmightimprovein theupcomingyears,butatthesametime,technicalefficiencymay drop.

The purpose of an in-depth case study is not to generalise but toreveala newphenomenon forfurtherinvestigation.The phenomenoninProposition1hasbeenobservedinotherhightech- nicallyefficientports,too,forexample,inQingdao(Li,2009)and HoChiMinh(Boyd,2018).Thereexistinefficientportsintheworld, forexample,thePortofOslo(TE<1)(Schøyen&Odeck,2013)that providesexcellentservice levelwithzerowaiting timefor con- tainervesselscallingattheport.Anderson,Fornell,andLehmann (1994)alsoobservedthesamephenomenonasstatedinProposition 1,forotherbusinessorganisations.Proposition2fitwithportsthat yieldaratherconstantdegreeofservicelevelandtechnicaleffi- ciencyovertime.Forexample,thePortofSingaporemaintainsa constantbuthighservicelevelandtechnicalefficiencyovertime.

Suchabalancebetweentechnicalefﬁciencyandservicelevelistyp- icallyachievedbyactivelyinvestinginportresourcesbasedonport throughputforecast.

6. Conclusionandfutureresearch

This study benchmarks the technical efficiency of 38 Asian containerterminalsusingDEAandFDHmethods.Bothmethods employtheinput-orientedefficiencymeasurement,andforDEA, technicalefficiencyestimatesareunderbothconstantandvariable return-to-scale assumptions.Mostof theportefficiencystudies (e.g.,DeOliveira&Cariou,2015;Hungetal.,2010;Cheonetal., 2010)consideredportsasDMUs,which limitsthepotentialfor strategicdecision-makingintermsofadjustingscalesamongport terminalstoachieveahigherlevelofefficiency.Therefore,inthis study,containerterminalsaretakenasDMUs.Also,thebootstrap

method is usedto validatethe accuracy of theDEA estimates.

Finally,anin-depthcasestudyexploresthecharacteristicsofthe highlytechnicalefﬁcientport,particularlyinvestigatestheassoci- ationbetweenservicelevelandtechnicalefﬁciency.

Onlyfourofthe38containerterminals(oneBangladeshi,one Chinese,oneIndianandoneVietnamese)meetthebest-in-class scaleefficiency.Althoughthetechnicalefficiencyscoresofallter- minalsdropslightlyinthebootstrapDEAestimates,therankingof terminalsremainedthesame.Amongthetopfour,ChittagongPort wasfoundhighlyefficientbyWuandGoh(2010),too.Thus,an in-depthcasestudyfocusingonChittagongPort,combiningboth quantitativeand qualitativedataderivetoa propositiononthe associationbetweenservicelevelandtechnicalefficiencyinthe contextofportoperations.

Basedonthefindings,highertechnicalefficiencyofaDMUdoes notmeanthatithasthehighestservicelevel.Forinstance,termi- nalsfromChittagongPortarehighlytechnicallyefficient(CCTand GCBbothscoredoneinDEA-BCCandFDH),butfacecapacityprob- lemsleadingtoseverecongestion.GCBisamultipurposeterminal withsixdedicatedberthsforcontainerhandlingandcanonlyserve gearedvessels.Thisrequireslessresourcefromaportproduction pointofview,whichmakestheterminalmoreefficient,inapurely technicalsense,thanitspeerswithsimilaroutput.

Technicalefficiencyrankingfromfrontierapproachescanvary dependingonthenumberofinputandoutputvariablesusedinthe study.Forinstance,theGCBterminalofCPdoesnothaveanyyard gantrycranesbutusesstraddlecarrierstomovecontainerwithin theterminal(RespondentA,2018,February).Theinclusionofstrad- dlecarriersasaninputvariablemayinfluencetheefficiencyscoreof GCB,andthiscanbetrueformanyothercontainerterminalsofthe world.Also,considerationofregionalorcountry-specificfactors suchasportgovernancemodelcanhaveanimpactonefficiency ranking(Nguyen,Nghiem,&Chang,2018).Furthermore,asargued inthisstudyandinlinewithSuárez-Alemán,Trujillo,andCullinane (2014),takingservicelevelintoaccountbymeansofwaitingtime, canchangetheefficiencyrankingscompletely.

Thisstudyoffersimportantimplicationsfortheliteratureand policymakers.In anearlierstudy,an informalinspectionof the highlyefficientportsrevealedthatsomeoftheportswereinef- ficientbecausetheywereinvestingactively“[...]ineitherport equipmentorinfrastructurewiththeobjectiveofbeingremain- ingorbecomingcompetitiveinthelong-term”(Cullinane&Wang, 2010,pp.735–736).Throughtheexplorationofanin-depthcase studyofChittagongPort–ahightechnicallyefficientport–this studyprovidesevidenceofthesamephenomenon.Terminalsand portsthatdonotactivelyinvestinportresourceeventuallybecome hightechnicallyefficientintheshort-runbutsufferintermsof servicelevelinthelong-run.However,portsthatinvestactively inresourcesbasedonthelong-termforecastofportthroughput, maintainaconstanttechnicalefficiencyandservicelevel.

From theport management perspective, theimplications of theﬁndingsarethreefold.First,strategicdecisionstakenbyport authoritiesshouldnotonlybebasedontechnicalefﬁciencyscores butalsotheservicelevel.Furthermore,poorservicelevel(e.g.long waitingtimeforberthing)hassevereenvironmentalimpacts,as Poulsen,Ponte,andSornn-Friese(2018)statedthat“withaguar- anteedberthuponarrivaltheshipcanslowsteam,avoididleanchor time,andreduceGHGandairpollutionatsea”(p.85).Second,port authoritiesmustperformlong-termforecastofcontainerthrough- puteveryyearandadjusttheirexpansiondecisionsaccordingly.

Finally,terminalswithbelow-averagescaleefficiencymayconsider cooperatingwithotherhighlyefficientterminalsofthesameport orneighbouringports,andviceversa.Toimprovethescaleeffi- ciencylevels,someterminalsmayformulateaclusteroflogistics networkwiththeefficientorinefficientterminalsdependingon theircurrentreturns-to-scale.

(10)

Futureresearchshouldtestthepropositionusingquantitative data.The use of simulation techniques mayserve thepurpose oftestingtheproposition(Davis,Eisenhardt,&Bingham,2007).

Another approach would be to adopt a multiple case study approach.Thesampleofcasestudiescouldbeexpandedwithmore containerterminalsfromEuropeanandAmericanportsthatmay provideevidenceinsupportofthepropositionofthisstudyoroffer otherinterestinginsightsregardingtheassociationbetweenser- vicelevel and technicalefﬁciency.In thesame lineofthought, itwould beinteresting toinvestigatewhetherthehightechni- callyefﬁcientterminalsarethemostresilientonestodisruptions.

ThisstudyalsoobservedthatcongestionatChittagongPortleads toshipping lines stockingup containers in transhipment ports (suchasinSingapore).Thus,futureresearchshouldinvestigatethe

impactofcongestioninaperipheralportonitsrespectivetranship- mentports.

Conﬂictofinterest Nonedeclared.

Acknowledgement

TheauthorwouldliketothankProfessorKevinCullinaneand AssociateProfessorMeifengLuoforusefulsuggestionsinanearlier versionofthismanuscript.

AppendixA. Casestudydatasource Datasource1:Twopersonalinterviews

No. Interviewee^a Role Position Yearsofexperience Timescontacted

1 RespondentA PortAuthority Topmanagement 25 4

2 RespondentB Carrier Mid-levelmanager 6 3

aNameshavebeenanonymised.

Datasource2:Onlinenewssearch No. Reference/source

1 ANL(2017,July6)ChittagongPortCongestionSurcharge.RetrievedonFebruary14,2018fromhttps://www.anl.com.au/news/501/chittagong-portcongestion-surcharge

2 Barua,Dwaipayan(2017,September7)Ctgportfacesfreshcongestion:Lackoftransportturnscontainercongestionacute.TheDailyStar.Retrievedon February14,2018fromhttp://www.thedailystar.net/business/ctg-port-faces-fresh-congestion-1458688

3 Islam,Syful(2017,August11)CongestedChittagongportfurtherrestrictsshipcalls.JournalofCommerce.RetrievedonFebruary14,2018fromhttps://www.

joc.com/port-news/asian-ports/port-chittagong/chittagong-port-further-restricts-ship-calls-ﬁght-congestion20170811.html

4 Islam,Syful(2017,September11)NewequipmenttohelpeaseChittagongportcongestion.JournalofCommerce.RetrievedonFebruary14,2018fromhttps://

www.joc.com/port-news/port-equipment/new-equipment-help-ease-chittagong-port-congestion20170911.html

5 Hussain,Anwar(2017,July20)Chittagongportcongestionmayresultinhugelossesforbusinesses.DhakaTribune.RetrievedonFebruary14,2018from http://www.dhakatribune.com/business/2017/07/20/businesses-fear-loss-vessel-congestion-hits-chittagong-port

6 Huang,Ethan(2017,August23)MajordelaysatChittagongPort.MoreThanShipping.RetrievedonFebruary14,2018fromhttps://www.morethanshipping.

com/major-delays-chittagong-port/

7 COSCOShippingLine(2017,August27)ThecontinuousportcongestionatChittagongPort.RetrievedonFebruary14,2018fromhttp://lines.coscoshipping.

com/home/News/detail/15010621062040435553/50000000000000231

8 EmiratesShippingLine(2010,May21)ChittagongPortCongestionSurcharge.RetrievedonFebruary14,2018fromhttp://www.emiratesline.com/

chittagong-port-congestion-surcharge/

9 Gateway-Group(2017,August28)DemurragefeesaredoubledatChittagongPorttoeasecongestion.RetrievedonFebruary14,2018fromhttps://gateway- group.com/demurrage-fees-doubled-chittagong-port-ease-congestion

10 Milad,Masud(2017,July18)SeverecongestionslowsdownCtgport.ProthomAlo.RetrievedonFebruary14,2018fromhttp://en.prothomalo.com/

bangladesh/news/153999/Severe-congestion-slows-down-Ctg-port

Note:OnFeb14,2018,aGooglesearchwasconductedusingtheterm“congestioninChittagongport”andthe10websitelinkswereassessedbasedonrelevance,among whichthreearefromshippingcompanies’websites,threefromBangladeshinewschannels,threefrominternationalnewschannelsandonefromthewebsiteof

aninternationallogisticsserviceprovider.Thelistofnewstitlesandtheirrespectiveelectronicsourcelinksarepresentedinthetableabove.

Datasource3:Otherarchivaldocuments

No. Reference/source^a

1 Overview:ChittagongPortAuthority(1997,1998,2006,2007,2014,2015).

2 YearBook:ChittagongPortAuthority(1996–1998,2002,2010)

aYearsinparenthesis.

AppendixB. Nodescomparedbynumberofcoding references

Nodes ChildNodeLevel1 ChildNodeLevel2 ChildNodeLevel3 #codingreferences Aggregate# codingreferences

Aggregate# itemscoded

Portservicelevel 17 71 11

Portservicelevel Accidents 6 6 5

Portservicelevel Congestion 10 21 8

Portservicelevel Congestion Delays 11 11 6

Portservicelevel Damages 2 26 7

Portservicelevel Damages Competitiveadvantage 7 7 2

Portservicelevel Damages Monetaryloss 17 17 7

Portservicelevel PortStaff 1 1 1

Technicalefﬁciency 2 46 10

Technicalefﬁciency Portfacilities 11 35 9