461
1 INTRODUCTION
Constantly growing global seaborne containerised
trade leads to increasing flows of containers in
seaports.Increasingnumberofhandledcontainersis
visible at global, regional and national level.
Generally speaking, this increasing trend is
depended on the economic development, for which
themainindicatorisGrossDomesticProduct(GDP).
In thi
s article the attempt has been made to
construct hierarchical model of container ports
throughputdevelopment.Thepresentedhierarchical
approachisbasedonthreemodelsofcontainerports
throughput development at three different
geographicallevels:global(container throughput in
all seaports on the world), regional (container
throughput in the Baltic seaports) and national
(cont
ainer throughput in Polish seaports). In this
article simultaneously the relationship between
global GDP and global container throughput and
between container throughput at the different
geographical levels: global‐regional and regional‐
nationalhavebeenillustrated.
Thearticleisdividedintothreemainparts.Inthe
firstpartthegeneralcharact
eristicofcontainertraffic
at the global, Baltic and Polish level has been
provided.Inthesecondpartthegeneralassumptions
tothemodelhavebeendescribed,whileinthethird
partthemodelshavebeenpresented and evaluated
for their fit and possibility to use for predictive
purposes.
2 GENER
ALCHARACTERISTICSOF
CONTAINERTRAFFICONTHEWORLDAND
INTHEBALTICANDPOLISHSEAPORTS
Nowadays volume of cargo carried in containers
constitutes about 16% of global seaborne trade, in
2000 it was 10%, whilst in 1990 only 5.8%.
Containerised cargo is the fastest growing cargo
segment expanding at an av
erage rate of 8.2%
between1990and2010[3]. These figures show that
the role of containerization in maritime transport is
constantly growing. The main factors of the
increasing share of containerized cargo in total
seaborne trade are: the increasing demand for
Hierarchical Model of Container Ports Throughput
M.Rozmarynowska&L.Smolarek
GdyniaMaritimeUniversity,Poland
ABSTRACT: In this article the attempt has been made to construct hierarchical model of container ports
throughputdevelopment.Thepresentedhierarchicalapproachusestherelationshipsofdevelopmentofglobal
economyandcontainerflowsatdifferentgeographicallevels:global(containerthroughputinallseaportonthe
world), regional (cont
ainer throughput in the Baltic seaports) and national (container throughput in Polish
seaports).Modelhavebeenevaluatedfortheirfitandusefulnessforpredictivepurposes.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 9
Number 4
December 2015
DOI:10.12716/1001.09.04.01
462
manufacturedgoodsandgrowing”containerisation”
ofbulkandbreak‐bulkcargoes.Itisestimatedthatin
2013thetotalvolumeofcontainerizedcargoloaded
achievedalevelofabout1.52blntonnes,whichwas
5.5% more than in 2012. If we take into account
volumes in TEU, containerised trade reached 159
millionTEUin2013,indicat
ingthe4.6%increasein
comparisonto2012(Fig.1).Theglobalcontainerised
seaborne trade can be divided into seven
geographicalcategories[2]:
The Intraregional trade led by intra‐Asian trade
(28.1%),
North–Southtrade(17.0%),
Thetrans‐Pacifictrade(13.6%),
FarEast–Europetra
de(13.1%),
SecondaryEast–Westtrade(12.6%),
South‐Southtrade(11.7%),
Transatlantictrade(3.9%).
Figure1.Globalcontainerizedtrade, world containerport
throughputandworldrealGDPin2000‐2013
Figure2. World and Baltic ports container throughput in
2000‐2013
The important element in serving containerised
tradeareseaports.Itisestimatedthatthetotalworld
containerportthroughputgrewfrom232mlnTEUin
2000 to 651 mln TEU in 2013 (Fig. 2). Comparing
thesefigureswithtotalcontainerisedtrade,itcanbe
indicatedthatonaverageoneTEUisreloadedab
out
4times(loadedintheportoforigin,unloadedfrom
ocean‐goingvesselandloadedontothefeedervessel
incontainerhubportandfinallyunloadedinportof
destination.). Over a half of total world container
throughput is taking place in Asia – the
manufacturing centre of the world. The Asian and
world leader is China
, approximately 50% of total
Asian ports throughput, and 25% of total world
throughputisservedbyChineseseaports.
One of the region where there is a fast
developmentofcontainerportsthroughputobserved
is Baltic Sea region (BSR). Baltic Sea region covers
nine countries with direct connection to the Balt
ic
Sea: Denmark, Estonia, Finland, Germany, Latvia,
Lithuania, Poland, Russia, Sweden. All of these
countries have at last one port that serve container
traffic. Total container throughput in Baltic ports
constituteabout1.5%oftotalworldturnover.In2013
the volumes reloaded in this region were esti
mated
at9.7mlnTEU.AsitcanbeseenintheFigure2,the
pattern of changes of world and Baltic container
turnoverissimilar.Theaveragetempoofgrowthof
worldcontainerthroughputinyears2000‐2008,was
10.6%,whilefortheBalticitwas11.6%.
The crisis year 2009, was much more severe for
Balt
icports (‐24%). However,since 2010, the higher
averageannualtempoofgrowthhasbeenobserved
for the Baltic traffic than for the total world traffic
(+14.8%versus+8.4%).
Figure 3. Baltic and Polish ports container throughput in
2000‐2013
OverthepastdozenyearstheoverallBSRmarket
structure has changed. In 2000 about 72.8% of
containershandledbyBalticseaportspassedthrough
the three Baltic countries: Finland, Sweden and
Denmark. Finland was the leader with the market
share of 30.3%, followed by Sweden (26.4%) and
Denmark (16.1%). Russian Baltic seaports handled
only10.1%ofBalt
iccontainertraffic.Polishseaports
wereonthefifthpositionhandling7.6%ofcontainer
traffic,thenextpositionswereheldbyportsofBaltic
States: Latvia, Lithuania and Estonia (6.7%) and
German Baltic seaports (2.7%) [3]. However, since
severalyearstheBalticleader isRussia,itssharein
thi
s segment of Baltic market is estimated at nearly
30%.Since2012,thesecondlargestcontainermarket
isPoland.CurrentlyPolishportshandleover20%of
total volumes served by Baltic ports. Since several
years Poland has been one of the fastest growing
container market within Baltic Sea region. In 2000
Polish ports handled 228 thou. TEU, but since then
container t
raffic in Polish ports increased 8.6 times,
while total Baltic market grew only 3 times (Fig.3).
Such spectacular increase in Poland was associated
not only with the development of domestic market
butalsowiththedevelopmentofseatra
nsit(from/to
EasternBalticregion).
463
In Poland, containers are handled in all major
ports,i.e.Gdańsk,Gdynia,SzczecinandŚwinoujście.
Today the largest Polish container port is Gdańsk,
with two terminals: Deepwater Container Terminal
(DCT)andGdańskContainerTerminal(GTK).Inthe
second largest Polish port, Gdyni a, there are two
containertermi
nals:BalticContainerTerminal(BCT),
Gdynia Container Terminal (GCT), but some
containersarealsoreloadedinBalticGeneralCargo
Terminal(BTDG),InSzczecincontainersarehandled
in DB Port Szczecin. There is also possibility to
handle containers in Port HandlowyŚwinoujście.
The total annual handling capacity of all that
terminals is esti
mated at around 2.7 mln TEU. This
means that in 2013 Polish container terminals used
around73%oftheirhandlingcapacity.Ithavetobe
underlined that majority of cargo, over 95% of
containers,ishandledinthreemainterminals:DCT,
BCT and BCT. The handling capacity of these
termi
nalsisestimatedat2.4 mln(in2013theyused
over77%oftheircapacity)
Polish ports mainly serve feeder container
services. The only Polish terminal that serve ocean
connections is DCT Gdańsk. Today DCT serve two
suchservicesoperatedbyMaerskLine.Thefirstone
(AE10service) have been launched in 2010 and
connects Gdańsk with Asian container ports. The
second one (Service CR
X) since May 2014 connects
Gdańsk with ports in Mexico, Belize, Panama and
CostaRica.
3 GENERALASSUMPTIONSTOTHEMODEL
Thehierarchicalapproachinthisarticleisbasedon
three models of development of container
throughputatthreedescribedinthepreviouschapter
geographical levels: global, regional (Balt
ic Sea
region)andnational(Poland).
In the first model (development of world
container throughput‐CP
W) the global Gross
Domestic Product (GDP
G) have been used as the
explanatory variable. GDP is used because this
parameter is commonly regarded as a good
explanatoryvariableforcontainertrafficinseaports.
IntheworldliteraturetherelationshipbetweenGDP
and seaports throughput (including container
throughput) is often used in the context of creating
predictive models. Authors of art
icles usually focus
on the relationship between national GDP and
turnover of seaports at national or regional levels.
Among the publications dealing with this issue are
[1],[6],[7],[8].Authorofthis articlealsotookonthis
subjectintwopublications:[4],[5].
The idea for the two remaining models was to
explore the relat
ionship between container
throughput on the different geographical levels:
global‐regional and regional‐national. In the second
presented in this article model (development of
containerthroughputinBalticport‐CT
B)theworld
container throughput has been used as the
explanatory variable and in the last model
(developmentofcontainerthroughputinPolishports
‐CT
P),containerthroughputinBalticportshasbeen
usedasanexplanatoryvariable(Fig.4).
Figure4.Hierarchicalmodelofportscontainerthroughput
Real world GDP is derived from statistical
databaseofUNCTAD[11],andis expressed inmln
USD. Data of world container port throughput are
derived from statistical database of UNCTAD [11]
and different numbers of publication [2], and are
expressedinTEU.InthecaseofBalticportscontainer
throughput,dataused in the models arecumulated
data,whichwerefoundforeachBalt
iccountry.Data
forRussianportsarederivedfromdifferentsources
[10],[13],[15].DataforDanishportsarederivedfrom
Official Danish Statistics website [18], data for
Swedish port are derived from the website of
SwedishPortsAssociation[19],dataforFinishports
are derived from the website of Finish Port
Association [12]. In the case of Germa
ny, data are
only for port of Lubeck and are sourced from the
port’s website [14]. Data for Lithuanian ports are
obtained from the website of port of Klaipeda [15],
dataforLatviaarederivedfromthewebsiteofport
ofRiga[17],dataforPolishportsarederiveddirectly
formcontainertermi
nals(DCT,BCT,GTK,GCT)and
Port of Szczecin‐Świnoujście. All data of container
throughputareannualdataforyears2000‐2013and
areexpressedinTEU.
4 HIERARCHICALMODELOFCONTAINER
PORTSTHROUGHPUT
Thefirststageoftheanalysisistheesti
mationofthe
trendmodelforglobalGDP.Thisisimportantfrom
the viewpoint of creating predictive models using
GDPasanexplanatoryvariable.Futurevaluesofthe
explanatoryvariablecanbepredictedonthebasisof
the trend model describing the development of the
va
riableovertime.GlobalGDPdevelopmenttrendis
describedbythesquared‐Y logarithmic‐X model,of
thefollowingequation:
tGDP
G
ln*238519.081139.1*
9
10 (1)
where: GDP
G = global Gross Domestic Product in
yeart
464
Table1. Statistical evaluation of global GDP development
trendfit
_______________________________________________
coefficientofadjustedcoefficient coefficientof
correlation ofdetermination random
(ρ)(R
2
)variation(V)
_______________________________________________
GDPG 0.990441 97.9388%1.5%
_______________________________________________
Theprobabilitytestislessthan0.05.Thisfactoris
highly statistically significant, which means that in
the studied phenomenon (growth of global GDP)
thereisaclear,significantdevelopmenttrend.
To evaluate the goodness of fit of the model
adjusted coefficient of determination has been used
(R
2
). In this case coefficient of determination shows
that the model explains 97.9388% of global GDP
volatility (Tab. 1), this value indicates a good fit of
themodel.AscanbeseenintheFigure5,themodel
pretty well describes the development of
phenomenonovertime.Inadditiontothe
cleartrend,
cyclical fluctuations are visible, as well as the
collapseoftheworldeconomy,especiallyfeltinthe
year2009.Whenitcomestoassessingthesuitability
ofthemodelasaprognostictoolonthebasisofthe
levelof randomvariation coefficient(which tellsus
of how
many percent on average, the theoretical
values obtained from the model are different from
theactualvalues),themodelcanbeconsideredasa
very useful. The coefficient of random variation
reached a low level of 1.5%, which reflects small
differencesbetweenthetheoreticalandactualvalues.
Figure5.TrendmodelofglobalGDP[mlnUSD]
Below the models of development of container
throughput are presented at three levels: global,
Baltic and Polish. In this part the relationship
between global GDP and world container
throughput, between world container throughput
and Baltic container throughput and between Baltic
container throughput and Polish container
throughput has been examined. All these
relationships are described by the following
equations:
G
GDPCT
W
*3974.2635419.8*
8
10 (2)
w
CT
B
CT
6431.73
72195.6*
9
10
1
(3)
B
CT
P
CT
9321.17
32584.1*
6
10
1
(4)
where: CT
W = world container throughput; CTB =
container throughput in Baltic seaports; CT
P =
containerthroughputinPolishseaports
Table2.Statisticalevaluationofmodelsfit
_______________________________________________
coefficientof adjustedcoefficientcoefficientof
correlation ofdetermination random
(ρ)(R
2
)variation(V)
_______________________________________________
GDPG‐CTW 0.999029 99.7896%1.5%
CTW‐CTB 0.992884 98.4637%6.0%
CTB–CTP 0.994293 98.7669%9.4%
_______________________________________________
In all cases, the P value is less than 0.05, which
means that there is a statistically significant
relationship between the variables at the 95.0%
confidence level. The correlation coefficient in all
cases exceeds 0.99, indicating a strong relationship
betweenthevariables(Tab.2).
The relationship between GDP
G and CTw is
described by a linear model. The model can be
considered as well matched. For this model, a very
highcoefficient of determinationhasbeenobtained,
whichindicatesthatthemodelexplains99.7896%of
the variability in CTw. There is also a very
satisfactoryvalueofcoefficientofrandomvariation,
which indicates that the theoretical values obtained
fromthemodeldifferfromtheactualvaluesbyonly
1.5%ataverage,which is avery good result. It can
thereforebeconcludedthatdescribedmodelmaybe
useful for predicting the development of world
containerthroughput.
The second analysed relationship (global
container
throughput and container throughput in
Balticseaports)isdescribedbythedoublereciprocal
model.Inthiscase,thereisalsoquitehighvalueof
the coefficient of determination, indicating that
model explains 98.463% of the variation in CT
B.
However, for this model, the worse coefficient of
randomvariationhavebeenobtained,thetheoretical
values differ from the empirical by 6% on average.
Nevertheless this level can be regarded as
admissible. The analysis of ex‐post errors indicates
thatintheanalysedperiod(2000‐2013)thetheoretical
values deviate
from the actual values from 0.3% to
6%,exceptfromcrisisyear2009,whenthedifference
wassignificant(datafromthemodelaresignificantly
overvalued in relation to empirical data), and
exceeded15%.
The last analyzed model shows relationship
between container turnover in Baltic ports and
containerturnoverinPolish
ports.This relationship
is described by a double reciprocal model. The
adjusted coefficient of determination indicates that
the model as fitted explains 98.8618% of the
variability in CT
P, which is a satisfactory value.
However, not very satisfactory is the level of
coefficientofrandomvariation,whichreached9.4%.
Itcan be indicated that the theoretical results differ
fromtheempiricalespeciallyinyears2006‐2010.The
ex‐posterrorsintheseyearsareatthelevelof10.8%,
465
11.07%,13.5%15.8%, 13.2%, respectively. The lower
errors are in the years 2011‐2013: 0.2%, 3.1%, 3.5%,
respectively.Asitcanbeindicated,themodeliswell‐
fittedespeciallyforthelastthreeyears.Inlastthree
yearsthetempoofgrowthinPolishportswasmuch
higherthanintota
lBalticsearegion,by1.7,2.8,5.9,
respectively.
ThehightempoofgrowthinPolishportsinyears
2010‐2013 was most likely connected with the
development of transhipment traffic in port of
Gdańsk, the largest Polish container port. By 2010,
containercargohandledinPolishportswasdestined
toorcomealm
ostexclusivelyfromdomesticmarket.
However, since 2010, when the first direct ocean
service from Asia to Port of Gdańsk have been
launched, we have been dealing with a new
phenomenon.Since2010,oneofthePolishporthave
been becoming growing hub port for containerized
cargo going to or coming from the East‐Balt
ic
markets (mostly to / from Russia). This means that
containersdeliveredtothehubportbyoceanvessel
fromtheexportingcountryarereloaded inthe port
to smaller ship and transported to the importing
country. Unfortunately, the data on the volume of
containershandledinseatra
nsitarenotdisclosed,so
it cannot be fully specified what is the share of sea
transitcargointhetotalcontainertraffichandledin
Polish ports. However, according to the opinion of
some representatives from maritime sector, such
traffic increases. And, generally, it can be assumed
tha
tthehubroleofGdańskwillevolveinthenearest
future.
All three models described in previous
paragraphs have been illustrated in the Figure 6.
Fromthatfigurewecansimultaneouslyreadhowthe
global economy influence the world container
throughputandwhatisrelationshipbetweenglobal,
Balt
ic and Polish container throughput. The
presented hierarchical model is quite useful to
analysethepasttrendsandrelationshipsandcanbe
useasaprognostictool.
Figure6. Hierarchical model of container throughput
(global,BalticandPolishlevel)
However,ithavetobeunderlinedthatthethird
presented model (relationship between Baltic and
Polishthroughput)willnotbeadequateinlongterm.
According to this model (see Fig. 6) if the whole
Baltic container market will continue to grow the
share of the Polish ports in total Baltic container
ma
rketincreaseandfinallyexceed100%whichisnot
true, as the Polish ports constitute the part of the
wholeBalticmarket.Toconclude,thismodelmaybe
adequatemaximumforseveralyearsandmostlikely
for2‐3years.Thepresentedmodelhasbeenadjusted
totheincreasingshareofPolishportsintota
lBaltic
container throughput. Such situation had place
during the last years. However, this will not be
everlasting trend. The share will reach some level
andprobablywillbemoreorlessstableorinsome
future will decline (such situation may occur for
example if there will be some change in Balt
ic
containermarketandtheotherhubintheBalticwill
appear, that will compete with Polish ports). The
presented model shows the accelerated tempo of
increase of Polish ports share in total Baltic
throughput, however this many not be true. The
increasing role of Polish ports is much likely, but
proba
blynotinsuchrapidpeaceasitisindicatedby
model.
On the other hand, if the tempo of growth of
containerthroughputinPolishportswillremainata
highlevelandassumingthattherewillbeasimilar
pattern to this observed toda
y or there will be a
growth in development in other Baltic container
marketsandadditionallythepatternofdevelopment
of world container traffic will be consistent with a
first presented model (relationship between GDP
G
and CT
W), it can occur that second model
(relationship between (CT
W and CTB) will not be
adequateanymore, andwill have to be adaptedto
thechangingBalticmarket.
5 CONCLUSION
The presented hierarchical model illustrates
simultaneously the relationships of container
throughput development at different geographical
levels:global, Baltic and Polish.It is quite useful to
analysethepasttrendsandrelationships.However,
someerrorsofesti
matesarevisible, whichcouldbe
corrected by some correction factors. The whole
model can also fulfil some prognostic role, and
deliver the forecast of the container throughput at
global,BalticandPolishlevel.Butmostprobablythe
hierarchicalmodelasawhole,will not beadequate
forlongterm,itcould fulfil it
sprognosticroleonly
for the next several years at most. Then its
componentswillhavetobeadaptedtothechanging
circumstancesespeciallyintheBalticsearegion.
REFERENCES
[1]Chou Ch., Chu Ch., Liang G. A mod ified regression
modelforforecasting the volumes of Taiwan’s import
containers, Mathematical and Computer Modelling Tom,
47,Issue9–10,2008,p.797–807.
[2]Review of maritime transport, Issues 20012014,
UNCTAD.
[3]RozmarynowskaM.,BalticContainerOutlook2012,Actia
Forum,Gdynia,2012
[4]Rozmarynowska M., Smolarek L. Modelowanie
zależności pomiędzy wielkością przeładunków w
polskich portach mors
kich a gospodarką polską i
światową,Logistyka,3/2014,CD1,p.5434‐5443.
466
[5]Rozmarynowska M., Smolarek L. Modelowanie
zależności pomiędzy wielkością przeładunków
kontenerowych w polskich portach morskich a
zmiennymi ekonomicznymi, Logistyka, 6/2014, CD2,
p.1088‐1097
[6]Sun H., Chen X. Analysis on Differences of Economic
FactorsAffectingtheCargoThroughputofThreeMain
PortsinChina,
Logistics,2009,p.3409‐3415.
[7]Van Dorsse C., Wolters M., Van Wee B. A Very Long
TermForecastofthePortThroughputintheLeHavre–
HamburgRangeup to2100, EJTIR,12(1), 2012,p. 88‐
110.
[8]Vanoutrive T. Exploring the link between port
throughput
andeconomicactivity:somecommentson
space‐andtime‐relatedissues,50thEuropeancongressof
the Regional Science Association International (ERSA
Congress 2010): Sustainable regional growth and
developmentinthecreativeknowledgeeconomy,2010.
[9]http://en.portnews.ru/
[10]http://nmtp.info/en/
[11]http://unctadstat.unctad.org/ReportFolders/reportFold
ers.aspx?sCS_referer=&sCS_ChosenLang=en
[12]http://www.finnports.com/eng/statistics/?stats=yearly
&T=4&year=2013
[13]http://www.kscport.ru/?p=&lang=en
[14]http://www.lhg.com/index.php?id=153&L=1
[15]http://www.portofklaipeda.lt/
[16]http://www.portoftallinn.com/
[17]http://www.rop.lv/en/
[18]http://www.statbank.dk/statbank5a/default.asp?w=128
0
[19]http://www.transportgruppen.se/InEnglish/Associatio
n‐ports‐of‐Sweden/
