153

1 INTRODUCTION

Issue in the operation and future development of

maritime ports, therefore relevant methods for the

prediction of container demand (or other types of

cargo or passenger traffic) need to be implemented.

When it comes to the main elements or types of

forecasting procedures, the following issues should be

noted:

− differences and relationships between primary

demand (requirements of the socio-economic

system - trade) and secondary demand (transport

flows served by ports),

− different time perspectives of forecasting (from

very short to long-term predictions),

− internal structures of container traffic – spatial

sources of demand,

− variations in approaches, methods and techniques

of forecasting.

Trade flows should be regarded as a key driver of

demand and development for container services in

seaports. In addition, specifically, intercontinental

trade relations affect the field of container technology

implementation. As a result, export and import

volume is strictly related to the specific production

and consumption characteristics of a country.

Therefore, GDP is an important factor affecting

containerisation activities. A much wider approach

could be also used, where the PESTE3 method can be

applied (Lappalainen, 2013). Identification of

secondary demands requires further actions, - a

commonly implemented approach is to use a five step

procedure (Jensen, 2014):

A method which considers the influence of Political, Economic,

Social, Technological and Environmental issues on the state and fu-

ture of a phenomenon or organisation.

A Simplified Forecasting Model for the Estimation of

Container Traffic in Seaports at a Nation

al Level – the

Case of Poland

M. Matczak

Gdynia Maritime University, Gdynia, Poland

ABSTRACT: Comprehensive forecasting of future volumes of container traffic in seaports is important when it

comes to port development, including investments, especially in relation to costly transport infrastructure (e.g.

new terminals). The aim of this article is to present a specific, simplified model of demand forecasting for

container traffic in seaports as well as to give a practical verification of the model in the Polish seaport sector.

The model consists of relevant indexes of containerisation (values, dynamics) referring to the macroeconomic

characteristics of the country of cargo origin as well as destination-predictor variables (e.g. population, foreign

trade, gros

s domestic product). This method will facilitate the evaluation of three basic segments of the

container market: foreign trade services, maritime transit flows and land transit flows. International

comparisons of indexes (benchmarking) as well as extrapolations of future changes can support this prediction

process. A practical implementation of this research has enabled us to calculate that the total container volume

in Poland will be approximately 4.69 – 4.87 million TEU by the year 2023.

http://www.tran

snav.eu

the

International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 14

Number 1

March 2020

DOI:

10.12716/1001.14.01.18

154

1 production and attraction (primarily demand

identification),

2 distribution (transport flows selection),

3 logistics (location of logistics nodes, e.g. maritime

ports),

4 modal split (modal allocation of the flows),

5 assignment (traffic per maritime port).

Regarding the time periods of predictions, the

following types of forecasts can be distinguished:

− short (e.g. one-day forecasts/ simulations/

planning (Gokkus, Sinan Yildirim and Akoglu,

2015) (KRILE, MAIOROV and FETISOV, 2018)

(Iannone et al., 2016)),

− medium (seasonal analysis based on monthly data

(Chen and Chen, 2010) (Peng and Chu, 2009)),

− long-term (Gökkuş, Yildirim and Aydin, 2017)

(Rahman, Muridan and Najib, 2015) (Rashed et al.,

2018) forecasts.

Implementations of competent methods of

forecasting are also related to specific sectors of

container traffic, which we can categorise into the

following elements:

− foreign trade traffic – international trade to the

country where the port is located

(intercontinental),

− inland transit traffic – international trade between

neighbouring countries served by land-based

modes of transport (e.g. railways, road haulage,

inland navigation)

− maritime transit traffic (transshipment) –

international trade between neighbouring

countries served by feeder vessels.

Container flow can be divided into further

different levels: separation of import and export

traffic (Chou, Chu and Liang, 2008) or full and empty

container turnover (Diaz, Talley and Tulpule, 2011).

Finally, different approaches (e.g. Shima & Siegel,

Brockwell & Davis, Schultz (Darabi and Suljevic,

2015)), models (e.g. time series, system dynamics,

regression, input-output models (Kotcharat, 2016))

and tools/techniques (e.g. generic programing,

decomposition approach, SARIMA

(Chen and Chen,

2010), input-output models (Kotcharat, 2016), neuron

networks (Gosasang, Chandraprakaikul and Kiattisin,

2011) (Gökkuş, Yildirim and Aydin, 2017)) to

forecasting can be implemented in the research.

Currently, researchers are trying to combine

different methods and techniques to create hybrid

models (Rashed et al., 2018), where a combination of

quantitative analysis is implemented alongside expert

knowledge (Huang, Qiao and Wang, 2014) (Huang et

al., 2015).

Ultimately, issues of balance between

sophistication and generalisation in the differing

forecasting methods, especially in relation to long-

term perspectives, will need to be considered.

Sophisticated methodology with highly-enhanced

detail may result in a high level of complexity and

may be costly for forecasting procedures. However,

uncertainties about changes in the port environment

(political, economic, social, technological) may cause

SARIMA - Seasonal Autoregressive Integrated Moving Average

Model.

significant distortions in any estimated volumes.

Moreover, detailed analyses, especially of

international aspects, may require access to relevant

data and information which may not be available,

especially on the global scale.

2 METHOD

Simplified model of container traffic forecasting at a

national level, proposed by the author consists of

several steps (Fig. 1). The first challenge is the

collection of relevant data of container traffic, with

distribution on the three types of flows being

dependant on specific sources of primary demand.

Research into changes over time in the basic

parameters requires a ten-year period of coherent

data.

The identification of relevant predictor variables

constitutes the second step in the process. Three

parameters: trade volume (tons), GDP (EUR) and

population changes are the key parameters used in

the research. In the case of trade volume, specific

trade relations (intercontinental) need to be specified

(especially in relation to EU member states).

Therefore, access to a comprehensive data base is

required (e.g. Eurostat). Assessment of correlations

between container traffic and main predictors have to

be tested.

Figure 1. Proposed simplified forecasting procedure for

national container traffic

Source: Own elaboration

Calculations of the relevant indexes of

containerisation, including container traffic and

predictor variables, is the next stage in the research.

Levels of implementation of container technology in

155

the national economy and trade need to be

established. However, it should be noted that any

assessment of technology utilisation requires

comprehensive references (best practices), such as

comparisons to other countries (market leaders) or

groups of countries (e.g. EU28). Comparisons of the

levels and development of index values need to be

treated as integral growth factors.

The next step in the research is identification of

future changes in predictor variables. International

organisations (e.g. IMF, World Bank, OECD, WTO) as

well as national statistical offices are the main sources

of such information. It should be emphasized that the

range of data availability will determine the scope of

indexes in any forecast.

By combining forecasts and indexes, an estimation

of future volumes of containers could be made. The

approximation of certain outcomes necessitates the

setting of minimum and maximum scenarios for

foreign trade flows in containers.

The designation of the origin and destination

points of cargo flows, located on the foreign

hinterland of seaports is required for the next step of

the procedure. The collection of comprehensive and

coherent data (predictor variables) from neighbouring

countries, as well as the proper assignment of relevant

indexes of containerisation is necessary. At this stage,

a simplified approach, based on the size of

populations has proven to be the most suitable. This

means, however, that the results obtained should be

treated as general. This is because a relatively stable

number of citizens does not necessarily create similar

changes in the primary demand for container traffic,

so only improvements in other indexes as such could

change the forecasting of volumes.

The collected results also need to be adjusted by

including data concerning any particular countries

involved in the service of transit traffic. In these cases,

historical changes of shares need to be also included

in the analysis and extrapolated for subsequent

periods.

Finally, the total demand for container services in

ports, at the country level, is estimated. These

proposed methods could be implemented for long-

term forecasting, however the scope of any

predictions will be necessarily restricted by

availability or unavailability of data.

3 RESULTS

Initial steps in the process of implementation and

verification of the above described model of

forecasting is the identification and distribution of

container traffic in Polish maritime ports. Market

divisions between foreign trade, maritime and land

transit flows is presented in Table 1.

Table 1. Structure of container traffic in Polish maritime

ports between 2008-2018 [thou. TEU]

_______________________________________________

Polish Maritime Land Total

foreign trade transit transit

_______________________________________________

2008 854.9 1.4 3 859.3

2009 670.5 0.4 0.6 671.6

2010 823.0 226.8 0.5 1,050.3

2011 1,117.9 239 0.3 1,357.2

2012 1,250.3 405.9 0.7 1,656.9

2013 1,223.9 745.1 0.4 1,969.5

2014 1,457.0 681.8 0.8 2,139.6

2015 1,408.4 455.1 0.3 1,863.8

2016 1,447.5 583.9 1 2,032.4

2017 1,705.2 678.5 1.1 2,384.8

2018e 2,026.6 806.4 1.3 2,834.3

_______________________________________________

Source: Own elaboration [18]

3.1 Foreign trade container traffic

Significant facilitators of container flows (predictor

variables) are both the economic activity of a country

(and its neighbours) and volume of trade. Gross

domestic product constitutes the first element. In

relation to trade flows, additional assumptions

referring to the spatial pattern of cargo flows should

be made.

Intercontinental trade is the main area of

implementation of containerisation, thus the total

volume of trade, should include cross-ocean flows

(the research includes: Vietnam, United States,

Taiwan, Singapore, New Zeeland, Mexico, South

Korea, Japan, India, Indonesia, Hong-Kong, China,

Chile, Canada, Brazil, Argentina, Australia). These

kinds of flows drive maritime transport with specific

effects on container traffic – TRADE (cont.).

Table 2. Macroeconomic characteristics of Poland (2008-

2017)

_______________________________________________

GDP GDP per Foreign Foreign

[billion capita trade–total trade-container

EUR] [EUR] [m tons] [m tons]

_______________________________________________

2008 366.2 9,607 196.3 9.64

2009 317.1 8,315 173.0 8.55

2010 361.8 9,515 199.9 9.43

2011 380.2 9,990 213.0 9.93

2012 389.4 10,230 211.3 9.77

2013 394.7 10,371 218.2 11.32

2014 411.2 10,815 232.6 11.78

2015 430.3 11,321 241.8 11.99

2016 426.5 11,235 250.1 13.31

2017 467.2 12,303 261.9 15.96

_______________________________________________

Source: Own elaboration based on Eurostat database

High levels of correlations between the container

turnover in Polish ports and the development of the

country’s main macroeconomic parameters (GDP

value and foreign trade turnover) were revealed. The

Pearson correlation coefficient for elaborated

parameters reached (2008-2017):

− 0,9640 for GDP,

− 0.9638 for GDP per capita,

− 0,9372 for foreign trade (intercontinental).

It could be stated, that any future changes in

macroeconomic conditions will constitute the basic

driver for further growth of maritime container

traffic.

156

The level of utilisation of container technology is

the second factor introduced into our research. Three

basic indexes of containerisation were defined in this

field:

1 TEU/TRADE (cont.),

2 TEU/GDP and additionally

3 TEU/POPULATION.

TEU/population parameters are useful when

conducting research into transit traffic, where access

to comprehensive macroeconomic data could be

difficult. An assumption of convergence in the

European dimension was adopted, so the average

value of the indexes of containerisation for the EU28

was taken as a reference point (benchmark) – Table 2.

Thanks to this, the problem of structural diversity in

container flows (trade-transit) in different countries

was eliminated.

Table 3. Containerisation measures (benchmarks) for Poland

and EU28 (2008-2017)

_______________________________________________

TEU/TRADE TEU/GDP TEU/

(cont.) [TEU/ POPULATION

[TEU/1000 tons] thou. EUR] [TEU/person]

Poland EU28 Poland UE28 Poland UE28

_______________________________________________

2008 88.69 123.26 0.0023 0.0061 0.022 0.159

2009 78.41 135.41 0.0021 0.0055 0.018 0.134

2010 87.31 135.41 0.0023 0.0059 0.022 0.150

2011 112.57 137.28 0.0029 0.0061 0.029 0.161

2012 127.94 142.48 0.0032 0.0061 0.033 0.164

2013 108.16 142.27 0.0031 0.0062 0.032 0.166

2014 123.65 148.33 0.0035 0.0063 0.038 0.176

2015 117.47 145.63 0.0033 0.0060 0.037 0.174

2016 108.79 147.56 0.0034 0.0062 0.038 0.180

2017 106.87 144.95 0.0037 0.0062 0.045 0.186

_______________________________________________

Source: Own elaboration

Statistics reveal discrepancies between the level of

the value of indexes calculated for the Polish and

European (EU28

) container sectors. In addition,

differences in growth trends of the indexes value is

also clear (Figure 1). It can therefore be concluded that

the level of technology use in Poland is still lower

than in the EU, however an increasing tempo of

growth will lead to their convergence.

European Union consists of 28 countries

Figure 2. Compound Annual Growth Rate (CAGR

) of the

value of containerisation measures in years 2008-2017 for

Poland and the EU28

Source: Own elaboration

Value of particular indexes (2008-2017) shows that

they should unified by the year 2032 (TEU/GDP

indicator). Thus, a further improvement in the

utilisation of container technology in Poland should

be expected. As a result, the future growth of

container traffic in maritime ports of Poland should

exceed the average European level during the next 20

years.

Taking into consideration predicted increases in

growth of Polish trade (World Economic Outlook

Database, October 2018, 2018), GDP development

(World Economic Outlook Database, October 2018, 2018) ,

changes of population (Population projection at national

level (2015-2080), 2019) as well as changes over time in

the values of indexes, calculations of future container

traffic can be made in the field of foreign trade, in

Polish ports (Figure 2).

Figure 3. Container traffic forecast for Polish maritime ports

(Polish foreign trade flows) up to 2023 [thou. TEU].

Source: Own elaboration

CAGR - Compound Annual Growth Rate

157

These collected results (maximum and minimum

values) can only be considered marginal scenarios in

the prediction of future changes in container traffic.

Taking this into account, however, the volume of

foreign trade containers should reach the level of

2.699 million TEU to 2.879 million TEU by 2023.

3.2 Transit container traffic

One of the indexes of containerisation also

implemented into this research focuses on current and

future demands in transit traffic, both land and

maritime. As regards Polish seaports, the following

countries could be regarded as potential centres of

demand growth:

− maritime transit: Russia, Finland, Sweden, Latvia,

Lithuania, Estonia,

− land transit: Czech Republic, Slovakia, Hungary,

Belarus, Ukraine.

Because of limited access to comprehensive and

coherent data about trade and economic

developments in non-EU countries, estimations were

based on population factors. Assuming that the non-

EU neighbouring countries achieve a level of

containerisation already specified for Poland (0,045),

and EU member states are characterised by the

average European level of the index (0,186), the

demand for particular types of transit traffic can be

theoretically estimated.

As regards the demand for maritime transit, the

total demand can be calculated as 10,461 thou. TEU

(data for year 2017), with a strong contribution of that

from Russia (6,451 thou. TEU).

The total demand from land transit potentially

served by Polish container ports can be estimated to

be about 7,120 thou. TEU (with a share of 67.3% from

EU countries).

These values are fully theoretical, because in the

case of individual countries the current levels of

implementation of container technology is so

different. Such a phenomenon can be observed in

non-EU countries: Russia, Ukraine, Belarus. It can,

however, be expected that technological structures

will change, and higher volumes of containers will

flow through the seaports in the future. On the other

hand, parts of the analysed countries have no access

to the sea, so foreign trade in those places is served

only by foreign ports. In these cases, theoretical

volumes of demand were estimated.

Calculations of the contribution of regional transit

traffic to Polish maritime ports was the next step in

the forecasting process. Currently, the share could be

calculated to be 6,54% and 0,02% for maritime and

land transit respectively. The share of Polish ports has

been growing by 13.02% and 19.54% annually

respectively (CAGR) in the period 2011-2017,

therefore further improvements in the market

position of Polish ports can be assumed. This means

looking at the year 2023, the shares of 13.64% and

0.04% could be implemented into our calculations. As

a result, the transit traffic served in Polish ports could

reach 1.993 thou. TEU by 2023.

3.3 Total demand

Summing up, the total forecasted traffic of containers

in the ports of Poland would reach levels from 4.692

thou. TEU up to 4.872 thou. TEU by 2023. Obviously,

numerous factors, both internal (sector) and external

(economy & trade) will have a direct influence on the

final results. However, these estimated values could

be treated as a starting point to more detailed analysis

of future growth.

4 CONCLUSIONS

The above presented methodology towards the

development of container demand forecasting should

be regarded as relatively simple but useful. Dividing

container flows into three key parts, helps facilitate

the application of different methods into the

prediction exercise. Future changes of predictor

variables (trade, GDP) as well as the development of

indexes of containerisation, constitute elementary

drivers in the further growth of container traffic. This

method, in addition to quantitative analysis, also

requires an expert opinion, because the choice of

extrapolation techniques or implementation of

specific factors, require logical verification and

sectorial knowledge. The best confirmation of the

usefulness of this method have been the preliminary

results for 2018 (2.834 m TEU), which are coherent

with the results obtained in our calculation (2.653 -

2.770 m TEU).

REFERENCES

[1] Chen, S. H. and Chen, J. N. (2010) ‘Forecasting container

throughputs at ports using genetic programming’,

Expert Systems with Applications. Elsevier Ltd, 37(3), pp.

2054–2058. doi: 10.1016/j.eswa.2009.06.054.

[2] Chou, C. C., Chu, C. W. and Liang, G. S. (2008) ‘A

modified regression model for forecasting the volumes

of Taiwan’s import containers’, Mathematical and

Computer Modelling, 47(9–10), pp. 797–807. doi:

10.1016/j.mcm.2007.05.005.

[3] Darabi, S. and Suljevic, M. (2015) ‘Forecasting Process for

Predicting Container Volumes in the Shipping Industry’.

[4] Diaz, R., Talley, W. and Tulpule, M. (2011) ‘Forecasting

empty container volumes’, Asian Journal of Shipping and

Logistics, 27(2), pp. 217–236. doi: 10.1016/S2092-

5212(11)80010-6.

[5] Gokkus, Ü., Sinan Yildirim, M. and Akoglu, K. (2015)

‘Prediction of the Container Traffic in a Seaport

Stockyard Using Genetic Algorithm’, 7(03), pp. 9–15.

[6] Gökkuş, Ü., Yildirim, M. S. and Aydin, M. M. (2017)

‘Estimation of Container Traffic at Seaports by Using

Several Soft Computing Methods: A Case of Turkish

Seaports’, Discrete Dynamics in Nature and Society, 2017.

doi: 10.1155/2017/2984853.

[7] Gosasang, V., Chandraprakaikul, W. and Kiattisin, S.

(2011) ‘A comparison of traditional and neural networks

forecasting techniques for container throughput at

bangkok port’, Asian Journal of Shipping and Logistics,

27(3), pp. 463–482. doi: 10.1016/S2092-5212(11)80022-2.

[8] Huang, A. et al. (2015) ‘An interval knowledge based

forecasting paradigm for container throughput

prediction’, Procedia Computer Science. Elsevier Masson

SAS, 55(Itqm), pp. 1381–1389. doi:

10.1016/j.procs.2015.07.126.

158

[9] Huang, A., Qiao, H. and Wang, S. (2014) ‘Forecasting

container throughputs with domain knowledge’,

Procedia Computer Science. Elsevier Masson SAS,

31(Itqm), pp. 648–655. doi: 10.1016/j.procs.2014.05.312.

[10] Iannone, R. et al. (2016) ‘Proposal for a flexible discrete

event simulation model for assessing the daily operation

decisions in a Ro-Ro terminal’, Simulation Modelling

Practice and Theory. Elsevier B.V., 61, pp. 28–46. doi:

10.1016/j.simpat.2015.11.005.

[11] Jensen, M. (2014) Forecasting Container Cargo Throughput

in Ports, Erasmus University Rotterdam.

[12] Kotcharat, P. (2016) ‘The Maritime Commons: Digital

Repository of the World A forecasting model for

container throughput: empirical research for Laem

Chabang Port , Thailand Kingdom of Thailand’.

[13] KRILE, S., MAIOROV, N. and FETISOV, V. (2018)

‘Forecasting the Operational Activities of the Sea

Passenger Terminal Using Intelligent Technologies’,

Transport Problems, 13(1), pp. 27–36. doi:

10.21307/tp.2018.13.1.3.

[14] Lappalainen, A. (2013) ‘Scenario-based traffic forecast

for routes between the penta ports in 2020’, Publication

from the Centre for Maritime Studies, University of Turku,

A65.

[15] Peng, W. Y. and Chu, C. W. (2009) ‘A comparison of

univariate methods for forecasting container throughput

volumes’, Mathematical and Computer Modelling. Elsevier

Ltd, 50(7–8), pp. 1045–1057. doi:

10.1016/j.mcm.2009.05.027.

[16] Population projection at national level (2015-2080) (2019)

Eurostat, https://ec.europa.eu/eurostat/data/database.

[17] Rahman, N. S. F. A., Muridan, M. and Najib, A. F. A.

(2015) ‘A Maritime Forecasting Method for Analysing

the Total Cargo Handling at Johor Port Berhad from

2013 to 2020’, 6(3), pp. 187–193.

[18] Rashed, Y. et al. (2018) ‘A combined approach to

forecast container throughput demand: Scenarios for the

Hamburg-Le Havre range of ports’, Transportation

Research Part A: Policy and Practice. Elsevier, 117(July

2016), pp. 127–141. doi: 10.1016/j.tra.2018.08.010.

[19] Statistical Yearbook of Maritime Economy (2018) Statistic

Poland. Statistical Office in Szczecin, Warsaw/Szczecin.

[20] World Economic Outlook Database, October 2018 (2018)

IMF,

https://www.imf.org/external/pubs/ft/weo/2018/02/weodata/in

dex.aspx.