1 Determinants of seaports container throughput between the Hamburg-Le-Havre range and the Mediterranean ports: Infrastructure investments and their returns on TEU Master Thesis Department Urban, Port & Transport Economics Antonios Fachouridis Student number 447351 Supervisor Mr. Martijn Streng Rotterdam, March 2018
50
Embed
thesis final copy · 1 Determinants of seaports container throughput between the Hamburg-Le-Havre range and the Mediterranean ports: Infrastructure investments and their returns on
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Determinants of seaports container throughput between the Hamburg-Le-Havre range and the Mediterranean ports:
Infrastructure investments and their returns on TEU
Master ThesisDepartment Urban, Port & Transport Economics
Antonios FachouridisStudent number 447351Supervisor Mr. Martijn Streng
“And if you find her poor, Ithaca will not have fooled you. Wise as you will have become, so full ofexperience,youwillhaveunderstoodbythenwhattheseIthacasmean.”Constan-neP.Cavafy.
Thepresentstudycons-tutesmyfinalefforttowardscomple-ngmymasterstudies attheUrban,PortandTransportEconomicsattheErasmusUniversityRoberdam.As everyjourneycomestoanend,a newchapterin the course of life opens, rich in adventures and challenges. I would like to take advantage of theopportunity of this shortsec-on to express mygra-tudeto the peoplewhosupportedme fulfilling thisobjec-ve.
Firstandforemost,all my teachers attheErasmusUniversityRoberdamandespeciallymy supervisor,mr.Mar-jnStreng,forhisindispensablefeedbackandguidancewhichallowedmetostructuremywri-ngandsharpenmythoughts.Furthermore,Iwouldliketothankthesecondreader,mr.BartKuipersforhiseffortandinterestinreviewingandchallengingmystudy.
Lastbutnot least, I would liketo thankmyparentsGeorgios andSofia,mybrotherPetrosandmysisterEfhymia,fortheirimmensesupportofeverykindandbeingconstantlybymyside.
6
Abstract
Thepresentpaperinves-gatestheeffectoftheinfrastructureinvestmentsontheportcontainerthroughputbetweentwoportranges:TheHamburg-LeHavrerangeversus theMediterraneanrange.TheHamburg-LeHavre range includes Germany, Netherlands, Belgium and France. The Mediterranean range includesPortugal, Spain, Italy, Croa-a, SloveniaandGreece. Theinfrastructure investmentsinclude infrastructureinvestments in the four modes of transport, namely Air, Rail, Road and Sea and investments insuperstructures,namelyTransportEquipment.Paneldataanalysishas beenmade,withdependedvariablethe TEU (port container throughput) and independent variables the investments in infrastructures andsuperstructures.Theresultsshowthatboth infrastructureand superstructureinvestmentshavea posi-veandsignificanteffectontheportcontainerthroughput.Therearealsosizabledifferencesinthereturns ofinvestmentsbetweentheHamburg-LeHavreandtheMediterraneanrange.Policyrecommenda-onsundertheprismofthefindingsof thispaperaremade:thecontainerthroughputasaSpecialPurposeVehicleforfinancingtransportinfrastructureprojectsanditspoten-aluseforabalancedregionalpolicyarediscussed.
Key words: TEU, port container throughput, mari%me port infrastructure investments, road transportinfrastructure investments, rail transport infrastructure investments, air transport infrastructureinvestments,transportequipmentinvestments.
7
1.Introduc4on
Containerthroughputisthemost importantanddirectfactor for evalua-ngthecompe--vestrengthof aport (Lechao Liu&Gyei-Kark PARK, 2011). Addi-onally,port container throughputfiguresare of utmostimportance for the policy making of the port and regional authori-es. The current port containerthroughputexplanatorymodelsmakelibleornouseoftheinfrastructureinvestments’influenceontheportperformance.
However,thereareseveralpapers whichunderlinethedecisiveeffectofinfrastructureinvestmentsonportperformance. For example, Grossmann et al. (2006) state that, except for port infrastructure, highlysignificant for the compe--veness of ports are the infrastructure links from theport to the hinterlandmarketbypipelines,rail,waterways,roadandair.
Furthermore,eventhoughinfrastructureinvestmentsintheothermodesoftransport(air,road,rail)playacri-cal role in the compe--veness of a port, they have not been tested for improving the predic-veperformanceofthecontainerthroughputforecas-ngmodelsyet.Thereforethecurrentforecas-ngmodelsare of limited power and the significance of the infrastructure investments, since not en-modeled andmeasured,asaresultareunderes-matedbytheportauthori-esandtherelevantpar-es(liners,shippers,regionalauthori-es,terminaloperatorsetc).
Asaresultof thisconnec-on,theaccuracyof thecurrentcontainerforecas-ngmodelsis expectedtobeimproved. Container forecas-ng is of utmost importance for the long term policy making of the portauthori-es, liners, shippers and terminaloperators. Therefore, except for its academic contribu-on, thepresentpapermightaswellassistthestrategicdecisionmakingoftheaforemen-onedagents.
Finally,thepresentedmodelscanbeusedbythegovernmentalorganiza-onsandportauthori-es,inordertopredictthe returnsof infrastructureinvestments in termsof TEU.Withan es-ma-onof theaveragetransportedgoods’valuepercontainer,theamountofrevenuesintaxes,customdu-es,clearanceetceteracanbecalculated.Thiscould bean interes-ng insight for governments,localauthori-esand investmentfundswhich couldmonetarize the returns of infrastructure investments aswell, in order to assess theabrac-venessofinfrastructureinvestments.
9
2.LiteratureReview
In this sec-on, the literature review will be presented, for both studies that have not included theinfrastructureinvestmentseffectontheportcontainerthroughput,andarecentstudywhichhasincludedtheinvestmentsinmari-meportinfrastructure.Addi-onally,theeffectoftheinfrastructureinvestmentsontheports’performanceasperthecurrentstudieswillbereviewed.
YasmineRashed(2015)concludedthat,theEU18industrialconfidenceindicatorandtheindex of industrialproduc-onareleadingthecontainerthroughputattheportofAntwerp.AtaHawaM.(2015)study,ForeignDirect Investment,Popula-on andGDPwere chosen as theprinciple components to analyze the port’scontainerthroughput.
Pi-noot Kotcharat (2016) developed a forecas-ng model, predic-ng the container throughput in theChabangPort.Heusedasexplanatoryvariablestheemployment,privateinvestmentindexandthebunkerpriceinSingapore.
Lastbutnot least,asmen-oned in theintroduc-on,onlyonestudywasfoundtoemploymari-meportinfrastructureinvestmentforpredic-ngtheportcontainerthroughput.ArjunMakhecha(2016),usedamongothers Sea Infrastructure Investments for explaining the container throughput of theHamburg-Le Havrerange.However, for someof theportsof theregion,thespecificvariablewasfoundeither insignificantorwith nega-ve coefficient. According to the author, these results indicate that the investments in portinfrastructureintheregionarenotop-mal.
Except for the aforemen-oned variables, and asmen-oned in the introduc-on, even though there is aplethora of studiesstressingthecri-caleffectof infrastructureinvestmentsontheperformanceofaport,libleresearchhasbeenmadeinordertoverifyandquan-fytheirsignificanceindeterminingportcontainerthroughput.
According toOosterhaven & Knaap (2003), investmentsin thehinterland infrastructurecan improvethecompe--veness of a port.Meersman etal. (2008) emphasizethat successful ports belong to successfulsupply chains. JoseL. Tongzon (2009) inves-gated the forwarders’ port choicecriteriaand found thatadequateinfrastructure (roads and railways) playa crucialrole to forwarders’ decisionschoosing aport.Adequateinfrastructure,decreasesport conges-onandshipwai-ng -me,allows fora quickerandsaferfreight movement and enables the ships to achieve economiesof scale, resul-ng in reduced mari-metransportcosts.
Lusthaku (2017),men-ons that the determinant of efficiency of port opera-ons is connected with thefactorsofcosts and-me,whicharecorrelatedwiththehinterlandinfrastructuresuchasinlandwaterwayconnec-ons,roadandraillines.Lustakuconcludesthatbothhinterlandandportinfrastructureinfluencetheportperformanceandcontainerthroughput.
Lustaku’sanalysisisqualita-veandtheeffectsofhinterlandinfrastructureonportperformanceareratherblurred. Even thought that generally admibed, the investments in infrastructure posi-vely impact thecompe--venessofports,many-mestheirreturnoninvestmentsisques-oned(TshepoKgareetal,2011).
Portefficiencyvarieswidelyfromcountry tocountryandspecificallyfromregiontoregion(T.Rajasekar&Malabika Deo, 2014).Therefore,itis crucial to testtheinfrastructureinvestmentsreturnsonports’TEUbetweengeographically andculturallydifferentregions.Asperthecurrentstudy,thechosenport regionsare within the European Union, namely the Hamburg-Le Havre and the Mediterranean region. TheHamburg-LeHavreregionincludesGermany,Netherlands,BelgiumandFrance. TheMediterranean regionIncludesPortugal,Spain,Italy,Slovenia,Croa-aandGreece.Thetworegions,competeeachothernotonlyforthemarketoftheEuropeanhinterland,butalsoforinfrastructureinvestmentfundingfromtheEuropeanUnion(TEN-Tnetwork),whichmakestheanalysisevenmoreinteres-ng.
The variables which will be used are assumed to cover the majority of a country’s infrastructureinvestments.Twogroupsof infrastructureinvestmentshavebeendis-nguished:Infrastructureinvestmentsandsuperstructureinvestments, adis-nc-onthathasbeenmadebyTheWorldBankGroup (2000). Ontheir reportfor theprivatesectorandtheinfrastructurenetwork, docksand storageyards aredefined aport’sinfrastructureandsheds,fueltanks,canesandvancarriersassuperstructure.
KazutomoAbeandJohnS.Wilson(2009)stressedtheimportanceoftradecostsandtheirnega-veeffectonthe interna-onal trade flows. Their regression analysis recommends that the expansion of portinfrastructurewouldceterisparibusreducetheimportchargesandtradecostspaidbytheimporters,whichwouldinturnreducethetransportcostsandleadtoatradeexpansionthroughtheports.
According toTshepoKgareetal.(2011), theobsoletetradeinfrastructureofSouthAfricaresultedinportandterminalconges-onwhichinturngainedthereputa-onofaninefficientport.Long portandterminalwai-nghours increasethelead-meandpipelinecosts ofasupplychain,thereforelimitthechancesoftheporttobeusedandfinallyrestrictitspoten-alcontainerthroughput.
12
NilGüler(2003),suggested thatthebenefitsof portdevelopmentprojectsaretransportsaving costs andreducedturn-round-me.“A port investmentmay, depending on thesitua-on,ease conges-on,increaseproduc-vity,reduceshipwai-ng-mecost,cargo-handlingcostandfinallyreduceoveralltransportcosts.”Therefore,portinfrastructureinvestmentsincreasethecompe--venessof aportand itsability to abractcontainerflows.
Hypothesis 1:Mari%me PortInfrastructure Investmenthas aposi%veandsignificant effecton the TEU forbothportranges.
2.4.2RoadTransportInfrastructureInvestments
Roadtransport representsthe largestshare,or75%,of thetotal inland freighttransportedwithintheEU(Eurostat, 2017). Road transport in this study refers to the transporta-on of containers through trucks.According toaresearchconductedby CharlesK.etal. (2012), 98%of theques-onnairesresponded thatroadtransportinefficienciesaffectportperformance.Poorroadnetworkisfoundtoresultinaslowuptakeofcargointothehinterland.Asaresult,highertruckturn-round-meandthereforehighcargodwell-meattheportoccurs.Thestudysuggeststhatinvestmentsinroadinfrastructuresshouldbemobilizedforamoreefficientroadnetwork.
Anotherstudy, one conductedbyStephenG. et al.(2012) foundthatroad transport improvementshavedirect effects related to transport cost savings, which are correlated with the accessibility changes.Intui-vely,roadtransportinfrastructure(suchasstreetwidening,longerroadnetwork,tunnels andbridges)canwellresultinshortertransport-mesbetweentheportandthehinterlandandincreasetheflexibilityofthesupplychain.
Comparingwithroad, theshareof railintranspor-ng freighthas remainedstableataround18.5%since2011 (Eurostat, 2017). Railway transport has advantages of high carrying capacity, lower influence byweather condi-ons, and lower energy consump-on (M. Sreenivas & T. Srinivas, 2008). AmbweneMwakibete’s(2015)study,revealedthatrailtransportplaysasuperbroleintheportperformanceofDaresSalaam. Reduc-on of port conges-on, increase of cargo traffic and lower logis-cs costs are among thecontribu-onsofrailinfrastructureonportperformance.
Moving cargoviaairlinesis-mesfasterthanbyrail,roadorsea.However,airisusuallyusedforveryhighadded value commodi-es, such as technological advancements, which are-me sensi-ve, andgoods ofstrategicimportance.Addi-onally,containersastheones loadedontheships,arenotcarriedvia airplanes.Finally,airlinesareusedmainlytotransferpassengers,ratherthangoods.
Fromthis perspec-ve,unlikelywiththeMari-mePort,RoadandRailTransportInfrastructure,thereisnotanobviouseffectofapoten-alAirTransportInfrastructureInvestmentontheperformanceoftheseaports.However,Air TransportInfrastructure Investmentshaveadirecteffectona country’sGDP.According toasurvey conducted by Invervistas (2015), a consul-ng company with extensive exper-se in avia-on,transporta-on,and tourism,theEuropeanairportscontributetotheemploymentof 12.3millionpeopleearning€356billioninincomeannually,andgenerate€675billioninGDPeachyear,equalto4.1%ofGDPofEurope.
Transport Equipment is the key elementwhich makesthevarious infrastructures func-on internally andexternally with each other. In thepaper of Grossman et al. (2006), apart from the port infrastructure,superstructures(tractor units, container gantries, cranes, et cetera) arealsoa key factor influencing thecompe--veposi-onofaportandthusthevolumeofcargohandledinthatport.AspertheOECDdefini-onglossary, intheassetsof thetransportequipment,seaport,rail,roadandairporttransportequipmentareincluded.Finally,thedefini-ons oftheinfrastructureandsuperstructureelements willbefurtherelaboratedintheEmpiricalAnalysissec-on.
Whatthepresentpaper inves-gatesis therela-onshipbetween thetransport infrastructureinvestmentsandtheportcontainerthroughput.Thereforewearelookingforcausesandtheircorrespondingeffects.Themostpopularmethodinthescien-ficliteratureinordertofindifandtowhatextenddoesone(ora group)of variables-elements explain aphenomenon is thecause and effectmodels. In ourcase, the causeandeffectmodel is to assist in theunderstanding of the rela-onshipbetween theaforemen-onedtransportinfrastructureinvestmentsontheportcontainerthroughputofthedefinedportranges.
Thereareseveraltypesof regressionanalysis:-meseries,crosssec-onal,paneldataandthepooleddata.Thechoiceof the proper one isbased on the structureof the collected data,which in turn is builtanddependsontheresearchques-ons.Theaforemen-onedtypesof regressionanalysiswillbeelaboratedinthefollowingsec-on,andtheonewhichbestdescribesourownresearchques-onsistobechosen,inordertoperformtheempiricalanalysispart.
3.1TimeSeriesData
Timeseriesdatais theobserva-onsofonevariable,TEU,throughthe-me(months,quarters,years).Timeseries datacanbeunivariateandmul-variate.A-meseriesunivariatemodelwouldabempttoexplaintheTEUvaria-onofasinglecountrythrough-me,usingpreviousobserva-ons ofthegivenvariable(TEU)ofthegivencountry(auto-regressive). The general formula can be described as:
Alterna-vely, a mul-variate -me seriesmodel, would abempt to explain theTEU varia-on of mul-plecountriesthrough-me,usingpreviousobserva-onsofthegivenvariable(TEU)ofmul-plecountries.
Ontheotherhand,crosssec-onaldatasets,areobserva-onsatasinglepoint in-me,forseveralen--es(countries).Cross sec-onalmodelsaredividedinunivariateandmul-variatemodels.Anexampleofacrosssec-onalunivariatemodelwouldhavebeenthestudyingoftheTEUvaria-onatasinglepointin-me,formul-plecountries,employing asinglevariableother than TEU (forexampletheinvestmentsinmari-meportinfrastructure).
Alterna-vely,amul-variatecrosssec-onalmodel,wouldhavebeenthestudyingof theTEUvaria-onatasingle point -me, formul-plecountries, employing mul-ple variables other than TEU (for example theInvestmentsinmari-meportinfrastructure,GDP,etc).
Even though crosssec-onalmodels are closer to a causeand effect analysis, thedimensionof -me ismissing.Therefore,itmighthappenthatarela-onshipwhichappears tobesignificantforonepointin-me,tobeinsignificantforadifferentpointin-me.
3.3PanelData
Whatdifferen-ates paneldatafromcrosssec-onaldata,isthatthesamecrosssec-onalunitsarefollowedover -me. Panel datasets (or longitudinal data) are structured by observing inmul-ple points in -me(months, quarters, years), mul-ple en--es (countries). Therefore, panel data are characterized by twodimensions,-me(t=1,...,T)anden-ty(i=1,...,N).
An advantage of the panel data comparing to the cross sec-on, is that they allow the researcher toinves-gatetheimportanceofthelageffectsof theexplanatoryvariables onthebehaviorof thedependedvariable(TEU).This informa-oncanbecrucial, sincemanyeconomicpoliciescanbeexpectedtohaveanimpactonlyaferacertainperiodof-mehaspassed(Wooldridge,2012).
Alterna-vely,a mul-variatepanelmodel,would havebeen thestudying of theTEU varia-onatmul-plepoints in -me, for mul-ple countries, employing mul-ple variables other than TEU (for example theInvestmentsinmari-meportinfrastructure,GDP,etc).
Pooleddataaremostlyusedinsurveys,whererandompeopleorspecialists (dependingontheshortofthestudy) areasked(interviewsorques-onnaires)abouttheirintui-on,regardingtheeffectofvariousfactorson the behavior of a certain phenomenon. An example of a pooled model would have been askingspecialists,suchasforwarders,liners,shippinglinesetcetera,regardingthepoten-aleffectsofInvestmentsinmari-meportinfrastructure,GDPetcontheTEUvaria-on.
Pooledmodeling is-me and resourcesdemanding. Addi-onally, thereliability of theanalysisresults arevulnerabletotheextenttheintui-onofthespecialistsiscorrect.Concluding,pooleddataarenota propermethodtoanswerourresearchques-ons.
Infrastructureinvestments,beitthesea,air,roadorrail,consumea significantperiodof-meun-ltheyarecomplete. Thedesign, planning, construc-on, comple-onandbeginning of func-on of an infrastructureobjectmight lastfrommonthstoyears.Furthermore, itmightrequire someaddi-onal -meeven to seetheireffectontheeconomyoverall.
Asmen-onedintheliteraturereview,ArjunMakhecha(2016)foundtheinvestmentsinportinfrastructureinsignificant in explaining the port container throughput. However, Arjun Makhecha did not test hisregression models for lag effects, in casethe investments in port infrastructure happens to significantlyaffecttheportcontainerthroughputonlyafersomeyears.
For example, the Channel Tunnel, which connects Folkestone, Kent (UK), with Coquelles, Pas-de-Calais(France)viarail,took20yearsinorderthereturnsof investments tomaketheprojectprofitable(OECD/ITF,2014).Engineering-wise, according toOECD(2011),major infrastructurecantake10-20years toplananddevelop.O. Pokornáand D. Mocková(2001), men-onof a construc-on period between 2-7 years of atransportinfrastructureprojecttocomplete.
An infrastructure development might take some years in order to be officially delivered for func-on.However, real life examples (Egna-a Odos/ Egna-aMotorway,Greece) show that projects are par-allydelivered for use,unofficially, as soonas that part of the project is safeand func-onal. Therefore, it isexpectedthatinfrastructureinvestmentsmightwellbeginimpac-ngtheeconomyandthecompe--venessofaport,notnecessarilytheveryfirstyearandaswellasbefore10years.
Chou etal. (2007),presentedtheimportanceof thenon-sta-onary rela-onshipbetween thevolumesofcontainers and themacroeconomicvariables. They conclude that, not taking careof thenon-sta-onaryrela-onship between the TEU (depended variable) and the explanatory variables (investments ininfrastructure)leadstoanoveres-ma-onoftheforecastedcontainerthroughputvolumes.
Various sta-s-cal tests exist in order data sta-onarity to be diagnosed. The most popular one in thecontainer throughput forecas-ng literature is theAugmentedDickey-Fuller (ADF) test. Aferapplying thetest, if any variablesare non-sta-onary on their levelbut sta-onary on their 1stdifference,haveto beturnedinto logarithmsbeforeused in theequa-on (YasmineRashed,2015). This methodology has beenappliedforthemajorityoftheportcontainerthroughputforecas-ng,withthemostrecentexamplesDraganD.etal.(2014)andYasmineRashedetal. (2015)andPi-nootKotcharat(2016).Inthepresentpaper, forprac-calissuestheIm–Pesaran–Shintestwillbeperformed2.
A major drawback of the first differencing is that, the model only considers the short-run adjustmentsrelatedtohowthedifferencein onevariablecorrelateswiththechangesintheother(M.Jansen, 2014).Hence,itignoresthelong-termrela-onshipbetweenvariables(Huietal.,2004).
In the co-integra-on the residuals are sta-onary with mean zero and there is a long run equilibriumrela-onship between Yt and Xt. Thenull hypothesis is the existence of no co-integra-on. In theno co-integra-on,therela-onshipbetweenthedependedandtheexplanatoryvariablesisvalidintheshortrunandnotinthelongrun.
To avoid thisissue,theerror correc-onmodel (ECM) canbeperformed.Theerrorcorrec-onmodelisadifferencedmodel that contains an error correc-on term,whichpredicts short-term adjustments of thedependentvariable.Themainideaof ECM isthat,apossibledisequilibriumintheshortruncorrectsitselfover-me,crea-nga paththatfluctuatesaroundthelong-runequilibrium(Huietal.,2004).Therefore,theECM isonlyvalidif thereisatruerela-onshipbetween thevariablesinthelong-run(VanDorsseretal.,2011).Aco-integra-ontestcanbeusedtotestwhethersucharela-onshipexists(Huietal.,2004).
Firstly,theportcontainerthroughputintermsofTEUhasbeenchosenasadependedvariable.Thereasonisthat,themajorityofthetransportedgoodsaretransferredwithincontainers.Asof2009,approximately90%of non- bulkcargoworldwideis movedby containersstacked ontransport shipsandthis trend isbeingconstantly increased(Ebeling,2009).It isassumedthat inthefuture,moreandmorecommodi-es willbecontainerized(Havenga&VanEeden,2011)causingthecontainershippingindustrytoexpandevenmore.Havenga&VanEeden(2011) also predict amaturing inthecontaineriza-ontrend, simplybecauseon agiven point in -me every commodity that can be shipped in containers shall be shipped in containers.Therefore,itis assumedthatcontainersarethemostrepresenta-vemeasurementfortheevalua-onofthebusynessofaport.
Secondly, inlandwaterwaysarenotincludedin theanalysis. Inlandwaterways,especially long rivers,arepresentonly in theHamburg-Le-Have range.Thereforetheanalysisbetweenthetworegionswouldhavebeenunequal.Aferall,inlandwaterwaysrepresentonlya6%ofthetotalinlandtransportfreight(Eurostat,2017).
What is more, we assume port efficiency, compe--veness and performance to have interchangeablemeaning,anditises-matedas theannualnumberofTEUsperport.Throughouttheanalysis,thetermsofcontainerorcargothroughputandTEUareusedinterchangeably.
Furthermore,theavailabledataseriesoftheinfrastructureinvestments refertotheperiod1987-2015.Itisassumed that this period is sufficient in order to apply a reliable panel analysis. Time-wise, this paperincorporatesthe2008economiccrisisin themodelingprocess,whichhasthebenefitof providing insightintotheimpactofthecrisis.
Theexplanatory variablesare thevariableswhich areassumed toexplainthevaria-onof thedependedvariable.Inthepresentpaper,theexplanatoryvariablesconsistofthetransportinfrastructureandtransportsuperstructure. The transport infrastructures include the Mari-me Port Infrastructure, Road TransportInfrastructure, RailTransport InfrastructureandAirTransportInfrastructure.ThetransportsuperstructureincludestheTransportEquipment.According to the OECD glossary, expenditure on new construc-on, extension of exis-ng infrastructure,includingreconstruc-on,renewalandmajorrepairsareincludedinthefollowingdataseries,exceptforthetransportequipment.
As observed from Graph 1, from 1970 un-l 1995, the country with thehighest number of containerscircula-onwastheNetherlands.Afer thatperiod, Germany becamethe country-champion in abrac-ngcontainers,followedbytheforeverincreasinggrowththeoftheSpanishports.Eversince,theNetherlandscomesinthethirdposi-on,followedbyBelgiumandItalywhichhavebeeninterchangingposi-onsthroughthe-me.
Anotherimportantno-ceisthefactthat,afer1995,thereisawideninggapbetweenthetop5countries(Germany, Spain, Netherlands, Belgium, Italy) versus the bobom 5 countries (France, Greece, Portugal,Slovenia,Croa-a).ExceptforFrance,which istheonlycountryfromtheHamburg -LeHavrerangeinthebobomgroup,therestofthecountriesbelongintheMediterraneanrange.
Interes-ngly, fromageographicalperspec-ve, thegap between the Hamburg - Le Havrerange and theMediterranean rangeintermsof containerthroughput isbeingsteadilydecreasedafer1970.Whereasinthebeginning of 1970theportcontainer throughputra-owasalmost5fortheHamburg-LeHavrerangecountriesto1fortheMediterraneanrangecountriesonaverage,in2015thera-owasalmost1to1(sub-Graph1).
Finally,theeconomiccrisisof2008isalsodepicted.As seenintheredrectangularinsideGraph1,thecrisisinfluenceddeeperthetop5countriescompara-velywiththebobom5countries.Thetop5countries lostapproximately1-2millionTEUsperyear,whereasthebobom5werealmostunaffected,exceptforGreece.Thecountriesofbothrangesneededapproximately2yearstorecoverfromthecrisis,withallofthehavingfullyrecoveredtheircontainerthroughputsafer2011.NetherlandsandBelgiumwerethecountrieswhichfaster recovered, whereasGreecewas thecountrywherethe impactof the crisis lasted for the longestperiodof-me.
4.3.2Mari4mePortInfrastructureInvestments
According toGraph2,from1985un-l2010, themari-meportinfrastructurehasbeenincreasing.A smalldropisobservedaferthe2010,whichmightbeabributedtothe2008economiccrisis andtherestric-vespendingbyeachcountry.
Spainis thecountrywiththehighestspendinginmari-meportinfrastructureinvestments,followedbyItaly.GermanyfollowsandaferthatBelgiumandNetherlands.Therestofthecountries barelyappearinGraph2.It isworth no-cing thattheHamburg- LeHavrecountries, thatis Germany,Belgium,NetherlandsandFrance, seem to follow a more steady spending policy for port infrastructure that in theMediterraneanrangecountries,especiallyItalyandSpain.
Finally,asitcan beseenin thesub-Graph 2 themari-meport infrastructure investmentsra-obetweenHamburg-LeHavreandtheMediterraneanrangehasbeenfluctua-ng.In1995theHamburg-LeHavrerangespent almost twice as much on average than theMediterranean range, whereas un-l 2000 the trendreversedandinvestmentsinportinfrastructureintheMediterraneanrangebecamesignificantlyhigherthanintheHamburg-LeHavrerange.
Contrary to mari-me port infrastructure investments, Spain comes only forth in the road transportinfrastructure. The first place is possessed by Germany, which is nevertheless renowned for the longhighways(autobahn).ThesecondandthirdcountriesareFranceandItalycorrespondingly.Therestofthe
countriesbarelyappear inGraph3. It isworthno-cing that,thebiggest thecon-nentalsizeof acountry(Germany,France,Spain,Italy),thehighertheinfrastructureinvestmentsinroadtransport.
Finally,as itcanbeseeninthesub-Graph3,theroadtransport infrastructure investmentsra-obetweenHamburg-LeHavreandtheMediterraneanrangehasbeenfluctua-ng.In1995theHamburg-LeHavrerangespentalmostthree-mesasmuchonaveragethantheMediterraneanrange,whereasun-l2007thetrendreversed and investments in road transport infrastructure in theMediterranean range became slightlyhigherthanintheHamburg-LeHavrerange.Afer2010, thera-o turnedbackhigheron theHamburg-LeHavrerange.
4.3.4RailTransportInfrastructureInvestments
According toGraph4,from1985un-l2010, therailtransportinfrastructurehasbeenincreasing.A smalldrop isobserved afer 2010,whichmight be abributed to the2008economic crisis and the restric-vespendingbyeachcountry.Thistrendisconsistentwiththemari-meportandroadinfrastructureaswell.
Francewastheonlycountrywhichdidnotdecreasethespending in railinfrastructureduring2011-2015,butalmostdoubledit, compara-vely with thepreviousperiodof 2006-2010. Following France,Germanycomessecond, followedby ItalyandSpain.Itis worthno-cing that,BelgiumandNetherlandshavebeenconstantly inves-ng more in rail transport infrastructure. Thismight be reflec-ng the effortof the twocountries,whichaccommodatetwoof thebiggestEuropeancontainerports(RoberdamandAntwerp),totackleroadtrafficconges-on,beberhinterlandconnec-vityandlessCO2emissions.
Finally, as it can beseen in thesub-Graph4therail transport infrastructureinvestments ra-obetweenHamburg-LeHavreandtheMediterraneanrangehasbeenfluctua-ng.In1995theHamburg-LeHavrerangespentalmostfour-mesasmuchonaveragethantheMediterraneanrange,whereas un-l2007thetrendreversedand investmentsinrail transport infrastructure in theMediterraneanrangebecamesignificantlyhigherthanintheHamburg-LeHavrerange.Afer2010, thera-o turnedbackhigheron theHamburg-LeHavrerange.
According toGraph5, from1990un-l2010, theair transport infrastructurehas been increasing.A smalldropisobservedaferthe2010,whichmightbeabributedtothe2008economiccrisis andtherestric-vespending by each country.Thistrendis consistentwith themari-meport, roadand rail infrastructureaswell.
Spainis thecountrywiththehighestfluctua-onswhenitcomestoairtransportinfrastructureinvestments.Betweeneitherdecade2001- 2005,or 2006 - 2010, Spainspentmorethan the rest of theother yearscombined. Germany on the other hand, is the country which spends almost equal amount of eurosthroughoutthe-me.SimilartoGermanyintermsoftheamountinvestedinairtransportpolicies,isFrance.
Finally, as it can beseen in thesub-Graph 5, the air transport infrastructureinvestments ra-obetweenHamburg-LeHavreandtheMediterraneanrangehasahigherfluctua-oncomparingwithmari-meport,roadandrail.Between1990and2000,theHamburg-LeHavrerangespentalmosttwice-mesasmuchonaverage than theMediterranean range, whereas un-l 2004 the trend reversed and investments in airtransportinfrastructureintheMediterraneanrangebecameslightlyhigher than intheHamburg-LeHavrerange.Afer2005,thera-oturnedbackhigherontheHamburg-LeHavrerange.
4.3.6TransportEquipmentInvestments
According to Graph 6, from 1985 un-l 2010, the transport equipment investments have been steadilyincreasing.A smalldropisobservedaferthe2010,whichmightbeabributedtothe2008economiccrisisandtherestric-vespendingbyeachcountry.This trendisconsistentwiththemari-meport,road,railandairinfrastructureaswell.
Finally,asitcanbeseeninthesub-Graph6thetransportequipmentinvestmentsra-obetweenHamburg-LeHavreandtheMediterraneanrangehasahigherfluctua-ngcomparingwithmari-meport,roadandrail.Between1980and1990,theHamburg-LeHavrerangespentalmostfour-mesasmuchonaveragethantheMediterranean range, whereas afer 1993 this trend decreased. Unlike the transport infrastructureinvestments (sea, road, rail, air), superstructure investments (transport equipment investments) in theMediterranean range never surpassed the transport equipment investments in the Hamburg - Le Havrerange.
Overall,themajorityoftheinvestmentsareplacedinthesuperstructureinvestments,namelythetransportequipment. Between thetwo ranges, during 2006 - 2010almost € 200 billion were spent on transportequipment.Thisamountshouldnotbeof asurprise,since,asmen-onedintheDataDefini%onssec-on,itincludestransportequipmentelementsfromallthetransportmodes(sea,road,rail,air).
Regarding the infrastructure investments, road transport infrastructureinvestments reached €50 billion,followedbyrailwith€26billion,€5.5billionmari-meportandairwith€4.5billionforthesameperiod.Addi-onally,thelast decade, theinfrastructureinvestment gap betweenthetworangeshassignificantlydecreased. Concluding the data descrip-on, the port container throughput (TEU) moves in the samedirec-onasthetransportinfrastructureinvestments,showingafirstposi-verela-onship.
4.4Sta4onarityTest
Asmen-oned in theEs%ma%onConcernssec-on,our variableshavetobeexamined forsta-onarity.Onetest which can beperformed in order to test for sta-onarity is the Im-Pesaran-Shin (IPS) test. The zerohypothesisof theIm–Pesaran–Shintestistheexistenceof aunit-rootmeaningthatourvariablesarenon-sta-onary.Thenullhypothesis isrejectedwhenthep-valueislessthanthe5%significancelevel.InTable3,theIPStestresultsispresentedforboththelevelandfirstdifferenceofourvariables.
AsitcanbeobservedfromTable3,allofourvariables arenon-sta-onaryonthelevel,sincealloftheirp-valuesarehigherthan5%significancelevelandthereforethenullhypothesiscannotberejected. Ontheotherhand,all of ourvariables aresta-onaryontheirfirstdifference,sinceallof theirp-valuesarelowerthan5%significancelevelandthereforetheirnullhypothesisisrejected.Themethodologicaldirec-onsoftheportcontainerthroughputliteraturehaveso-farbeenconfirmed.
Therefore,thepanelregressionsaretobebasedonthefirstdifferenceofthevariables.Asmen-onedintheEs%ma%onConcernssec-on,amajordrawbackofthefirstdifferencingis that,themodelonlyconsiderstheshort-runadjustmentsrelatedtohowthedifferenceinonevariablecorrelateswiththechangesintheother(M.Jansen,2014).Therefore,aferrunningtheregressions,itisnecessarytoperformco-integra-ontest,inordertotest if theequilibrium rela-onshipsbetween thedepended(TEU)andtheindependentvariables(infrastructureandsuperstructureinvestments)arevalidonlyintheshortrun,butonthelongrunaswell.Theresultsoftheco-integra-ontestsarereportedattheendoftheregressionresultstables.
31
5.RegressionResults
Inthepresent sec-on theregressionresultswillbepresented. ItbeginswiththeFixedEffectsregressionresultsof theHamburg-LeHavrerangeandcon-nuewiththeMediterraneanrange.Similarly,theRandomEffects follow. Various combina-ons have been tried, regarding different lagged effects. The presentedmodelsaretheoneswherethecombina-onofthelaggedeffectsyieldthehighestR-squared.
Ingeneral,alltheinvestmentsintransportinfrastructureandsuperstructurehaveaposi-veandsignificanteffect on the port container throughput of both ranges. However, in some examples (Road and RailTransport InfrastructureInvestments) theeffect ontheTEU is insignificantor the signs arenega-ve. Yet,theirbehaviorchangesasthemodelisenriched(H5orH7).As men-onedintheLiteratureReviewsec-on,importantexplanatoryvariables(kindlyrefer to Table1) arenotincludedinthemodel. This resultsinanomibedvariablesbias,leadingtounexpectedsignsorsignificantvariablestoappearasinsignificant.
Inregressionanalysis ithappensthat,avariablethatwasnotsignificanttobecomesignificantaferaddingrelevantvariablestothemodel.Theoriginallynotsignificantvariablewassignificantlyassociatedwiththeomibedvariableandreflectstheeffectoftheomibedvariableinaddi-ontoitsowneffect(plussomeotherunobservedvariables).When theomibedvariables(Table1) areaddedintothemodel, theoriginally notsignificantvariablenolongercapturesthepar-aleffectoftheomibedvariablebutnowreflectsthe"true"effect of that variable. It turns out to be sta-s-cally significantly associated with the port containerthroughput.Concluding,ourindependentvariablesdonotpresentunexpectedbehaviorintermsofsignsorsignificance.
5.1FixedEffects
5.1.1Hamburg-LeHavreRange
AsperTable4,ingeneral,all theinvestmentsintransportinfrastructureandsuperstructurehaveaposi-veand significanteffecton theportcontainer throughputof theHamburg - LeHavrerangeof ports. SincemodelH7is themostcompletemodelamongH1-H7,theH7modelwillbeinterpreted.Forprac-calreasonsand in order to avoid repe--on, the regression output of themodels H1-H6 will be not be described.However,themodelsH1-H6areinterpretedinthesameway.
Morespecifically,Mari-mePort InfrastructureInvestments haveaposi-veandsignificanteffectat the1%level on the port container throughput of theHamburg - LeHavre range as per the H7 model. If theMari-mePortInfrastructuresintheHamburg-LeHavreregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby3,620TEUonaverage,afer5yearsceterisparibus.
RailTransportInfrastructureInvestmentshaveaposi-veandsignificanteffectatthe5%level on theportcontainer throughput of the Hamburg - Le Havre range as per the H7 model. If the Rail TransportInfrastructuresintheHamburg -LeHavreregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby382TEUonaverage,afer7yearsceterisparibus.
AirTransportInfrastructureInvestmentshavea posi-veand significant effectat the5% levelontheportcontainer throughput of the Hamburg - Le Havre range as per the H7 model. If the Air TransportInfrastructuresintheHamburg -LeHavreregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby1,210TEUonaverage,afer6yearsceterisparibus.
Road Transport Infrastructure Investments haveaposi-veand insignificant effect on the port containerthroughput of the Hamburg - Le Havre range as per the H7 model. However, the Road TransportInfrastructureInvestments haveaposi-veandsignificanteffectatthe10%levelaspertheH5model.Iftheaforemen-onedinvestmentsintheHamburg-LeHavreregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby375TEUonaverage,afer4yearsceterisparibus.
33
TransportEquipmentInvestments haveaposi-veandsignificanteffectontheportcontainerthroughputoftheHamburg-LeHavrerangeatthe1% in theH7model.If theTransportEquipmentInvestments intheHamburg-LeHavreregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby80TEUonaverage,effec-vethesameyear,ceterisparibus.
5.1.2MediterraneanRange
AsperTable5,ingeneral,all theinvestmentsintransportinfrastructureandsuperstructurehaveaposi-veandsignificanteffectontheportcontainerthroughputoftheMediterraneanrangeports.SincemodelH7isthemost completemodelamong H1-H7,theH7modelwill be interpreted.For prac-cal reasons and inordertoavoidrepe--on,theregressionoutputofthemodels H1-H6will benotbedescribed.However,themodelsH1-H6areinterpretedinthesameway.
Tobeginwith,Mari-mePortInfrastructureInvestmentshaveaposi-veandsignificanteffectatthe1%levelontheportcontainerthroughput of theMediterraneanrangeasper theH7model. If theMari-mePortInfrastructures intheMediterraneanregionincreasebyEUR1million, theportcontainerthroughputwillincreaseby725TEUonaverage,afer2yearsceterisparibus.
RailTransportInfrastructureInvestmentshaveanega-veandsignificanteffectatthe1%levelontheportcontainerthroughputof theMediterraneanrangeaspertheH7model.If theRailTransportInfrastructuresInvestments in theMediterranean region increaseby EUR 1million, theport container throughputwilldecreaseby284TEUonaverage,afer3yearsceterisparibus.
RoadTransportInfrastructureInvestmentshaveaposi-veandsignificanteffectatthe1%levelontheportcontainerthroughputof theMediterraneanrangeaspertheH7model. If theaforemen-onedinvestmentsintheMediterraneanregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby110TEUonaverage,afer3yearsceterisparibus.
TransportEquipmentInvestments haveaposi-veandsignificanteffectontheportcontainerthroughputofthe Mediterranean range at the 5% in the H7 model. If the Transport Equipment Investments in theMediterraneanregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby47TEUonaverage,effec-vethesameyear,ceterisparibus.
Rail Transport Infrastructure Investments have a posi-ve and insignificant effect on the port containerthroughputoftheHamburg-LeHavrerangeas pertheH7model.However,theRail TransportInfrastructureInvestmentshaveaposi-veandsignificanteffectatthe1%levelaspertheH5model.Iftheaforemen-onedinvestmentsintheHamburg- LeHavreregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby565TEUonaverage,afer7yearsceterisparibus.
AirTransportInfrastructureInvestmentshaveaposi-veandsignificanteffectatthe10% levelontheportcontainer throughput of the Hamburg - Le Havre range as per the H7 model. If the Air TransportInfrastructuresintheHamburg -LeHavreregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby1,000TEUonaverage,afer6yearsceterisparibus.
Road Transport Infrastructure Investments haveaposi-ve yet insignificant effect on the port containerthroughputoftheHamburg- LeHavrerange.However,this variablehasyieldedsignificantcorrela-onwiththeTEUintherestoftheregressions.
TransportEquipmentInvestments haveaposi-veandsignificanteffectontheportcontainerthroughputoftheHamburg-LeHavrerangeatthe1% in theH7model.If theTransportEquipmentInvestments intheHamburg-LeHavreregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby91TEUonaverage,effec-vethesameyear,ceterisparibus.
36
5.2.2MediterraneanRange
AsperTable7,ingeneral,all theinvestmentsintransportinfrastructureandsuperstructurehaveaposi-veandsignificanteffectontheportcontainerthroughputoftheMediterraneanrangeports.SincemodelH7isthemost completemodelamong H1-H7,theH7modelwill be interpreted.For prac-cal reasons and inordertoavoidrepe--on,theregressionoutputofthemodels H1-H6will benotbedescribed.However,themodelsH1-H6areinterpretedinthesameway.
Tobeginwith,Mari-mePortInfrastructureInvestmentshaveaposi-veandsignificanteffectatthe1%levelontheportcontainerthroughput of theMediterraneanrangeasper theH7model. If theMari-mePortInfrastructures intheMediterraneanregionincreasebyEUR1million, theportcontainerthroughputwillincreaseby748TEUonaverage,afer2yearsceterisparibus.
RoadTransportInfrastructureInvestmentshaveaposi-veandsignificanteffectatthe1%levelontheportcontainerthroughputof theMediterraneanrangeaspertheH7model. If theaforemen-onedinvestmentsintheMediterraneanregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby101TEUonaverage,afer3yearsceterisparibus.
TransportEquipmentInvestments haveaposi-veandsignificanteffectontheportcontainerthroughputofthe Mediterranean range at the 1% in the H7 model. If the Transport Equipment Investments in theMediterraneanregionincreasebyEUR1million,theportcontainerthroughputwillincreaseby55TEUonaverage,effec-vethesameyear,ceterisparibus.
In every case, be it Fixed Effects or Random Effects, the differences between the coefficientsand theirsignificancesareveryslight,withoutharminggeneraliza-on.
39
6.DiscussionoftheResults
Exceptfor thecoefficients’ interpreta-on,con-nuingthefindingsdiscussion,furtherpointscanbemade.AccordingtotheR-squaredinthemostcompleteversion,intheH7modelisaround50%.Thismeans thatapproximately50%of theportcontainer throughputvaria-onisexplainedbyourmodel.Considering thattherearemorevariables thaninfrastructureinvestmentsthatpoten-alinfluenceportcontainerthroughput,the current models are significantly improving the predic-ng performanceof the container forecas-ngmodels.
Addi-onally, as per the co-integra-on tests and considering the omibed variable bias weakening therobustnessofthemodels,forthemajorityofthemitseemsthatthereisalongrunequilibriumrela-onshipbetweentheportcontainerthroughputandtheinfrastructureandsuperstructureinvestments.Therefore,theinterac-onamongthemiscausal.
Hypothesis2 RoadTransport Infrastructure Investments have aposi%ve andsignificanteffectontheTEUforbothportranges.
Confirmed
Hypothesis3 Rail Transport Infrastructure Investments have a posi%ve and significanteffectontheTEUforbothportranges.
Confirmed
Hypothesis4 Air Transport Infrastructure Investments have a posi%ve and significanteffectontheTEUforbothportranges.
Confirmed
Hypothesis5 Infrastructure Investments have jointlyaposi%ve and significanteffectontheTEUforbothportranges.
Confirmed
Hypothesis6 Transport Equipment Investments haveaposi%veand significanteffectontheTEUforbothportranges.
Confirmed
Hypothesis7 Infrastructure Investments and Superstructure Investments have jointly aposi%veandsignificanteffectontheTEUforbothportranges.
Confirmed
An interes-ng noteisthat, although transport infrastructureinvestmentshaveaposi-ve and significanteffect on the port container throughput for both ranges, significant differences exist between them.InfrastructureandsuperstructureinvestmentsyieldhigherTEU returns in theHamburg - LeHavrerangethan in theMediterranean range, something that is well observed in Table 7, calculated based on theRandomEffectses-ma-ons. Forexample,EUR 1million invested inMari-mePort Infrastructure, yieldsbetween4to6-mesmorecontainersintheHamburg-LeHavrerangecomparingwiththeMediterraneanrange.
Comparingtheimpactof theinvestmentswitheachother,according to theregressionresults,thelargestimpactof theinfrastructureinvestmentsontheportcontainerthroughputcomesfromtheMari-mePortInfrastructures.Eventhough theMari-mePortInfrastructureInvestments makeupalmosta tenthof theRoadTransport InfrastructureInvestments, their impactontheportcontainerthroughputis muchhigher.Finally,theAirTransportInfrastructureInvestments,yieldmoreorlessthesamereturnsonTEUforthetworanges.
Overall, the transport infrastructure investments appear to systema-cally yield higher returns in theHamburg - LeHavrerangethan intheMediterranean range. This might bedueto variousreasons.Onereasoncanbethehigherlevelsofcorrup-onintheMediterraneancountries.Thismeansthat,foreacheurospend in infrastructureand superstructure investmentsin theMediterranean countries, lesserpartof itfinally reaches an infrastructure project, than in Northern Europe where corrup-on rates are lower(TransparencyInterna-onal,2017).
ThedifferenceismoreapparentinArea2.ThesubgroupofBelgiumandNetherlandsspentalmost2billioneurosintransportinfrastructure,which isalmostequaltotheamountspentintheMediterraneanrange.Yet,inscalefrom0to5(5asthetopqualityof infrastructure),BelgiumandNetherlandsyieldmorethan1unit higher quality of infrastructures, which is very significant. Assuming that the technologicaladvancementsaresimilar between the two ranges, theunderlying reasons for this difference could beabributedtothemismanagementofthefundsheadedtoinfrastructureinvestments.
41
Another possible reason for which transport infrastructure investments yield more containers in theHamburg-LeHavrerange,isthefactthattheportsinthisregionarecloserwitheachother,incomparisontotheMediterraneanports.Withinadistanceofabout850kilometers,11portsarelocatedwithmorethan1,224,300,000 tons throughput in 2015 (Port of Roberdam Authority, 2016a). Therefore, this spa-alproximity might lead to infrastructure synerge-c effects (Theo Nobeboom, 2012). On top of that, theNortherngovernmentalauthori-es paymoreaben-ontotheeconomicdevelopmentandplanningoftheircountries,resul-nginamorecarefulandaimedspendingoftheinvestments.
6.3ComparisontotheLiterature
It is also interes-ng to make a few notes regarding the relevance of our results with the containerforecas-ng literature. Asmen-oned intheliterature,Makhecka(2016) is theonlyonewhohasusedtheMari-me Port Infrastructure Investments. Makhecka found insignificant effect of the Mari-me PortInfrastructureInvestments ontheportcontainer throughputof theHamburg - LeHavrerangeportsandabributedthistoasubop-mallevelofinvestment.
As per the results of the current paper, Mari-me Port Infrastructure Investments have a posi-ve andsignificant effect on the port container throughput of both the Hamburg - Le Havre range and theMediterraneanrange.However,theeffectoftheinvestmentscanbeseenonlyaferacoupleofyears.
Addi-onally, in the same paper,Makhecka found inland transport in length in kilometers (motorways,railways,waterways) tohaveaninsignificanteffectontheportcontainer throughput.Inorder the inlandtransportnetwork(inlength)effecttobecaptured,therelevantvariableneeds tohaveacertainminimumdegreeofvaria-on.Makheckaimpliesthatforsomeportrangesthesevariablesremainedconstant,makingithard for theeconometric program tocaptureany significance.Makhecka suggests differentmeasuringtechniquestobeused.
Onethoughtsupportedbythefindingsofthepresentstudy,istomeasuretheamountofmoneyinvestedintheextension of theinlandnetwork instead of its length,sincethefirst hasahigher degreeof varia-onthrough-methanthelaber.Inthisway,theeffectof thechangesintheinlandtransportnetworkcanbecapturedbytheeconometricprogram.
However, the anemic economic growth and the long las-ng effects of the 2008 economic crisis, haveconstrained the governmental budgets. Simultaneously, the private sector par-cipa-on in transportinfrastructure projects is modest. Therefore, the pool of the EUR 1,5 trillion transport infrastructureinvestmentneeds,donotseemtobefulfilledupto2030.
6.4.1InfrastructureInvestmentGaps
Evidently,thereisafinancinginfrastructureissue.Under-financingofthetransportinfrastructurescanharmeconomicac-vity andemployment,compe--venessandincreasing externalcosts suchasaccidentsandenvironmental degrada-on. As a solu-on, a plethora of studies supports the idea the infrastructureinvestmentfinancingwillneedtocomeincreasinglyfromtheprivatesector(TorstenEhlers,2014).
Complex legal arrangements inorder to ensurefair distribu-onof the returns to investments and risksbetween the par-es involved, crea-on of cash flows afer many years and unabrac-ve returns toinvestmentsareusuallyfactorsthatdiscourageprivatepar-cipa-onintransport infrastructureprojects(K.Bodewig & C. Secchy, 2014). Therefore, the poten-al financing deficit is rephrasedas to, how tomakeinfrastructure financing returns to investments more abrac-ve to private investors. The current studycontributesinansweringthisproblemasfollows.
Inthepresentstudy,therela-onshipbetweenvarioustransportinfrastructureinvestmentsandtheirreturntocontainer throughput hasbeenquan-fied.Undercertainassump-ons andwithin specificboundaries,onecanpredicttheaddi-onalcontainerthroughputaspertheaddi-onalinfrastructureinvestments.Theaddi-onal portrevenues (fromcontainer handling charges, feesand port dues) and governmental taxes(collectedfromimporttariffsandtaxa-on)whichoccurduetotheaddi-onalcontainerthroughputcanbepredictedaswell.
Therefore, it can be agreed among the par-es which are influenced and/or interested in transportinfrastructure investments, a share from the created port and governmental revenues as per addi-onalcontainergenerated.Acontainerunit(TEU),isthereforeapoten-alSpecialPurposeVehicle,whichconnectsthefinancialinvestmentsoftherelatedinfrastructureandtheircorrespondingrevenues,assis-ngintoafairdistribu-onoftheprofitsgeneratedbytheinvestments.
A container unit as a Special Purpose Vehiclehas several poten-al benefits over the current transportinfrastructurefinancing:
Asmen-oned in theregression results, transport infrastructure investments in the Hamburg - Le Havrerangeyieldhighercontainer throughputreturnsthanintheMediterraneanrange.Thisimpliesthat inthelongrun,conges-oninthesea(vesseltraffic),rail(traintraffic),road(trucktraffic) andair(avia-ontraffic)transportmodesmightbecomemoreacuteintheNorth,thanintheSouthof Europe.OntheotherhandinfrastructuresintheSouthmightbefunc-oningsub-op-mally.Therefore,thepresentstudyshedslightintothe distribu-on of the (at least on crossborder infrastructureprojects) European funds, in a way thatbalancedgrowthiseasiermonitoredandachievedalongtheEuropeanperiphery.
Another similar infrastructureini-a-veregardingmostly theMediterraneanEurope is theOneBeltOneRoad ini-a-ve from the Chinese. Chinese envisage to “embrace” the European hinterland market byconstruc-ngatranscon-nental railwayfromChinatoLondonandaseamotorwayfromChinatoPiraeusviatheSuezCanal.Assuch,thepresentstudyisappropriateintodeepeningtheunderstandingofthelong-termeffectsonthetransportsystemandtheeconomicgrowthofsimilarmega-projects.
6.4.4Macro-Construc4ngRequiresMacro-Financing
Overall, observing the vision of world’s biggest players, such as Russia, China, Middle East and Europeconcerning thefutureof theworldtransportinfrastructure,oneobservesthatmega-projectsaregoing tobecome the new reality. Mega infrastructural projects implymacro-construc-ng and the laber requiresmacro-financing.Thepresentmodelshowsthatpredic-ngthecontainerthroughputbasedon aggregate-macrofiguresisnotonlypossible,butsufficientlypreciseaswell.
Concluding thePolicyImplica%ons sec-on,itis suggestedthattransportinfrastructureinvestmentplanningtobe implemented in a macro-scale(for example on regional scale), as it isnot onlymore efficient tophysicallycoordinate,butalsomonitoringitseffectsthrough-mesaferforabalancedregionalgrowth.
45
7.Conclusions
The present study inves-gated the effect of transport infrastructure investments on port containerthroughputbetweenHamburg-LeHavreandtheMediterraneanrange.Theregressionresultsshowedthatindeed,aspertheobjec-veofthepresentpaper,thereisasizableandquan-fiableconnec-onbetweenthetransportinfrastructuresandsuperstructureinvestmentsandtheportcontainerthroughput.Mari-mePort,Road,Rail,Air transportinvestmentsandTransportEquipmentinvestmentshavea posi-veandsignificanteffectontheportcontainerthroughputonbothHamburg-LeHavreandMediterraneanrange.
Aspertheliterature,theeffectoftheaforemen-onedinfrastructureinvestmentsissignificantonlyaferacoupleof years, since theconstruc-onperiodneedstobeaccountedfor.Onthecontrary, theTransportEquipmentsuperstructureinvestmentshaveaposi-veandsignificanteffecton theportcontainerbythefirstyear,sincetransportequipmentisabuy-to-direct-useasset.
The presented models explainmorethan 50%of theport container throughputvaria-on, even thoughcri-cal explanatory variables such as GDP, popula-on and infla-on have been omibed. Asper the co-integra-ontestsandconsideringtheomibedvariablebiasweakeningtherobustnessofourmodels,forthemajority of our variables, it seems that there is a long run equilibrium rela-onship between the portcontainerthroughputandtheinfrastructureandsuperstructureinvestments.
Compara-velybetweenthetworegions,infrastructureandsuperstructureinvestmentsyieldhighernumberof TEUs in the Hamburg - Le Havre region than in theMediterranean range. Corrup-on and synerge-ceffectsduetospa-alproximitymightexplainthisvaria-on.
Finally,thecausalrela-onshipbetweenportcontainerthroughputandinfrastructurefundingandtheabilityof the laber to predict the volumesof the first, creates room for crea-ve proposals. The idea of thecontainerunitasaSpecialPurposeVehicleservingtheworldwideincreasinginfrastructureneedsshouldbelookedatdeeperbothinacademiaandthebusinessworld.
7.1Recommenda4onsforFurtherResearch
Although the findingsof thepresentpaper areinsigh�ul regarding the forecas-ng of theport containerthroughput,severalideasshouldberesearchedatinthefuture.Firstandforemost,thetwomainlimita-onsasperthesec-onAssump%onsandLimita%onsshouldbetakencareof.
To begin with, theeffectof theinfrastructuremaintenance on theport container throughput should beexamined. On theonehand, properlymaintaining thetransport infrastructureallowsforanefficientandlong term lifeof the infrastructure.However,thereareconcerns whether thecurrentinfrastructures areproperlymaintained.“As thestockof infrastructuregrows,andinmanycasesages,moreeffortis requiredtomaintain the quan-ty and thequality of the infrastructure. In spite of this shif, observers in manycountries haveraised concernsabout underfunding of infrastructuremaintenance. Roadmaintenance isofen postponed on the expecta-on that it will bemade up for in the future and there is no risk ofimmediateassetfailure.”(OECD,2017).
AconcludingroomforfurtherstudycouldbetheuseofthecontainerunitasaSpecialPurposeVehicleinthe financingof transport infrastructure projects.Next to it, itwillbeinteres-ng to inves-gatetowhichextend can monetary revenue streams (port fees, tariffs, taxes, etc) per container be used in order toimprovetheassessmentoftransportinfrastructureprojects(freightrelated)returnsoninvestments.
47
8.Bibliography
Abe,KazutomoandWilson,JohnS.,WeatheringtheStorm:Inves-ng inPortInfrastructuretoLowerTradeCosts in EastAsia (April 1, 2009).World Bank Policy ResearchWorkingPaper, Series, Vol. , pp. -, 2009.AvailableatSSRN:hbps://ssrn.com/abstract=1401217.Inves-nginPortInfrastructuretoLowerTradeCostsinEastAsia.
Ang,G.andV.Marchal(2013),“MobilisingPrivateInvestment inSustainableTransport:TheCaseofLand-Based Passenger Transport Infrastructure”,OECD EnvironmentWorking Papers, No. 56, OECDPublishing.hbp://doi.org/10.1787/5k46hjm8jpmv-en
Be-mLushtaku, (2017). Thecompe--venessof ports in theHamburg - LeHavrerangeversusnorthernAdria-cportsinthecontestedhinterland,ErasmusUniversityRoberdam
CharlesKipkoechKotut,Dr.FredMwirigiMugambi,“TheInfluenceofHinterlandTransportInefficiencies onthe Performance of Ports-A Case Studyof KenyaPortsAuthority.” Interna%onal Journal of Science andResearch(IJSR)ISSN(Online):2319-7064,ImpactFactor(2012):3.358
Erick Leal, GabrielPerez,Port-rail integra-on: challengesand opportuni-es for La-nAmerica. NU CEPAL.DivisiondeRecursosNaturaleseInfraestructura,Dateissued:2012-06,Serie:FALBulle-n,No.310.
Forecas-ng Container Throughput at the Doraleh Port in Djibou- through Time Series Analysis. .10.1142/9789814733878_0049.
Freight Transport Sta-s-cs, Modal Split, European Commission (2017) hbp://ec.europa.eu/eurostat/sta-s-cs-explained/index.php/Freight_transport_sta-s-cs_-_modal_split#Modal_split_in_the_EU
Gosasang, Veerachai & Chandraprakaikul, Watcharavee& Kia�sin, Supaporn. (2011). A Comparison ofTradi-onal andNeural NetworksForecas-ng Techniques for Container Throughputat Bangkok Port.TheAsianJournalofShippingandLogis%cs.27.463–482.10.1016/S2092-5212(11)80022-2.
Kotcharat,Pi-noot, "A forecas-ngmodelfor container throughput: empiricalresearch for LaemChabangPort,Thailand"(2016).WorldMari%meUniversityDisserta%ons.538.
Liu, Lechao &Park, Gyei-Kark. (2011).EmpiricalAnalysisof InfluenceFactorstoContainerThroughput inKorea and China Ports *. The Asian Journal of Shipping and Logis%cs. 27. 279–303. 10.1016/S2092-5212(11)80013-1.
Lourdes Trujillo and Gustavo Nombela (2010),Mul-service Infrastructure: Priva-zing Port Services, TheWorld Bank, l ink: hbp://siteresources.worldbank.org /EXTFINANCIALSECTOR/Resources/282884-1303327122200/222Truji-10-24.pdf
Makhecha,A. (Arjun).(2016,September12).Analysisof thedeterminantsof containerthroughputof themajor ports in theHamburg Le Havre range.Mari%me Economics and Logis%cs. Retrieved from hbp://hdl.handle.net/2105/37236
Rashed Mohamed Mohamed, Yasmine, Container Throughput Modelling and Forecas-ng: An EmpiricalDynamic Econometric Time Series Approach, University of Antwerp, Faculty of Applied Economics,DepartmentofTransportandRegionalEconomics,2016.
Tongzon, Jose. (2009). Port choice and freight forwarders. Transporta%onResearch Part E: Logis%cs andTransporta%onReview.45.186-195.10.1016/j.tre.2008.02.004.
Torsten Ehlers (2014), Understanding the challenges for infrastructurefinancing. Bank For Interna-onalSeblements,MonetaryandEconomicDepartment.
Villaverde, Jose&Millán,Pablo&Baños-Pino, José&VillaverdeCastro,José. (2005).Determinantsof thedemand for mari-me importsandexports. Transporta%onResearch Part E: Logis%cs and Transporta%onReview.41.357-372.10.1016/j.tre.2004.05.002.
VonckIndra,Prof.dr.TheoNobeboom,Dr.FrancescoParola,Dr.GiovanniSaba,Dr.LucaPersico,(2015).PortTraffic Forecas-ng Tool. 7th Framework Programme SST.2013.6-2. Towards a compe--ve and resourceefficientporttransportsystemCollabora-veProjectGrantAgreementno.605176