Introduction
In general the indicators for trade and economic relationships between Russian Federation and Republic of Korea look positive. The foreign trade turnover values increased from $2.9 billion in 2000 to $ 26.6 billion in 2014. In 2014 Russian exports to Korea reached $17.68 billion, imports from Korea to Russia reached $8.92 billion. However, the bilateral trade volumes in 2014 ($26.6 billion) were much less than Russia-China ($95.28 billion) or Russia-Japan ($30.6 billion) turnovers (source: Rusexporter).
The gap in investment volumes is even more revealing. According to the Export-Import Bank of Korea (KEXIM) in 2014 South Korea invested $113.6 million in Russian economy and $3.16 billion in Chinese projects. The total amount of investment accumulated in Russian Economy in 2014 was $22 billion - i.e. South Korean share was only 0.5% (source: Bank of Russia). Russian investment values into Korean economy reached $29 million (source: Ministry of Trade, Industry and Energy). While total volumes of Russian foreign direct investments reached $57 billion in 2014 (source: Bank of Russia).
Besides, the structure of mutual trade stays conservative and archaic: Russia supplies mainly oil and minerals, Korea - engineering products, electronics, and consumer goods. The current trade structure is quite vulnerable to all sorts of crises. In 2015, for instance, the trade turnover between two countries decreased significantly. There is a real threat of a slowdown in the development of trade and economic cooperation between Russia and South Korea, if it is not raised to a higher level, with a focus on intensive innovations and investment exchanges.
According to the "Concept of long-term socio-economic development of the Russian Federation until 2020", the cooperation with the Republic of Korea should be concentrated in hi-tech spheres. It is necessary to form strong technological and industrial alliances, implement large infrastructure projects.
The multilateral cooperation of countries of East Sea: Far Eastern part of Russia, Japan, South Korea, North Korea, and Chinese three north-eastern provinces (Liaoning, Jilin and Heilongjiang) plays a very significant role for the whole region. In 2015 the container land-sea route connecting Suifenhe (China) - port Vostochny (Russia) - port Busan (Korea) was opened. Hunchun (China) - port Zarubino (Russia) - port Busan (Korea) and Tianjin (China) - Busan (Korea) - Vostochny (Russia) - Trans-Siberian Railway (Russia) - Finland routes are already operating.
Also, a very promising project is a trilateral cooperation: a railroad trade route from Russia (Khasan) to the North Korean port of Rajin, then by sea to the ports of South Korea, further to South Korean Ports, Although, it was suspended because of U.N. Security Council's sanctions against North Korea.
Another area of cooperation is the transport corridor between the countries of the Asian-Pacific region, Central Asia, and Europe. The project of President of South Korea Park Geun-hye “Eurasian Initiative” project, in particular, assumes the creation of a single transport, logistics and cultural corridor from Korea to Europe. This project aims to enhance cooperation through the establishment of integrated logistics and energy systems. Likewise, both Russia and South Korea expressed a wish to participate actively in research of the Arctic region to find out the possibilities of using the Northeast Passage for freight transportation.
The development of these areas will provide the modernization of port infrastructure and port facilities in general. In March 2015 the Korean Minister of Maritime Affairs and Fisheries announced the intention to direct investments in the development and modernization of the Russian Far East ports to create anchor points for Korean vessels maintenance. Simultaneously, at the sixth meeting of the "Russian-Korean business dialogue" the President of Russian Federation Vladimir Putin specifically noted the development of infrastructure as an important factor. In June 2015 the Russian-Korean scientific research center for the analysis of transport and logistics complex and port infrastructure of the Far East and the Arctic was opened. The project was prepared by the Russian Maritime State University and the Korea Maritime Institute (KMI).
Of some interest are free ports and territories of priority development, already operating in the Far East of Russia. Korean special experience was taken into account when a free port of Vladivostok was created in October 2015. But in order to convert common ports into transport and logistics centers, tax privileges alone are not enough - it is necessary to create a highly developed infrastructure. Thus, it is very important to study the experience of other countries, adopt advanced technology, and use the best mechanisms to reach a new level of production. South Korea has accumulated rich experience in the construction of container terminals and ports. In January 2014 in Moscow, the Ministry of Transport of the Russian Federation and the Ministry of Maritime Affairs and Fisheries of the Republic of Korea signed a Memorandum of Understanding on the cooperation in the field of the port infrastructure development and modernization.
Competition and dynamic development of transport are the major motivating factors for ports to improve their infrastructure, berths, and unloading terminals, which allow passing a large turnover operatively. Many container ports must frequently review their capacity in order to ensure that they can provide satisfactory services and maintain their competitive edge. Sometimes, the necessity to build a new terminal or increase the existing capacity is inevitable. However, before a port implements such a plan, it is of great importance to know if it has fully used its existing facilities and the output is maximized, given the input (
Cullinane et al., 2004).
Using the frontier approach to efficiency estimation, the management of ports or terminals that are deemed inefficient can benchmark themselves against ports that are deduced to operate on the efficient frontier for the industry. Ports may also use the results from this sort of analysis to benchmark the performance of individual terminals.
The general purpose of the current research paper is to evaluate the efficiency of container handling and port industry, to understand whether the container terminals in Russia and South Korea operate efficiently or not. We are particularly interested in finding out what factors contributed to the significant growth of Korean container terminals, what strategies were adopted, and what did they brought in terms of effectiveness.
We address this objective from a quantitative standpoint by evaluating the relative efficiency of container terminals, as well as by examining the physical characteristics that may affect the efficiency.
Theoretical Review
Data Envelopment Analysis (DEA) is a mathematical programming technique that enables the determination of a unit’s efficiency based on its inputs and outputs, and compares it to other units involved in the analysis. The DEA can be described as data-oriented as it effects performance evaluations and other inferences directly from the observed data (
Daraio and Simar, 2007). The focus of the research is evaluating the technical efficiencies of a collection of Decision Making Units (DMUs) (e.g. ports, container terminals, bank branches, enterprises), which consume common inputs to generate common outputs (
Charnes et al., 1994). DEA determines the efficiency of each DMU by maximizing the ratio of a weighted sum of its outputs to a weighted sum of its inputs, while ensuring that the efficiencies of other units do not exceed 100%.
Within the family of DEA models, there is one initially proposed by Charnes, Cooper and Rhodes (CCR) in 1978. This model used constant returns to scale (CRS) concept to assess relative productive efficiencies of DMUs with multiple inputs and outputs. In 1984 Banker, Charnes, and Cooper (BCC) assumed variable returns to scale (VRS) for the model and evaluated technical efficiency and scale efficiency of a DMU.
Technical efficiency is defined as the ratio of the input usage of a fully efficient firm, producing the same output vector, to the input usage of the firm under consideration.
In current research, the model assumes I inputs, J outputs and N container terminals - DMUs. In addition, xi represents the amount of inputs employed, yi represents the amount of output produced by the i-th container terminal. Thus, the data in the sample are represented by J*N output matrix, Y and I*N input matrix, X. Since, there are N container terminals, the linear programming problem is solved N times, once for each container terminal in the sample.
To simplify the problem, in the DEA-CRS Technical Efficiency model, we consider that N container terminals operate under CRS and employ seven inputs (Xj, j=1,2,3,4,5,6,7) to produce single output (Y). The formal problem for technical efficiency (TE) can be conveniently expressed in following way:
Min TE, W TEis.t. Y∗wi≥yi,Xj∗wi≤TEi∗xi, j=1,2,3,4,5,6,7wi≥0,
where,
TEi is a scalar and represents the technical efficiency measure for the i-th container terminal;
w- is the
I*
N vector of the intensity weights, defining the linear combination of efficient container terminal to be compared with the
i-th container terminal. The inequality (
Y *
wi ≥
yi) implies that observed outputs must be less or equal to the linear combination of outputs of the container terminals that form the efficient frontier. The inequality (
Xj *
wi ≤
TEi *
xi) assures that the usage of inputs of efficient container terminals must be less or equal to the use of inputs of
i-th container terminal. It will satisfy:
TEi ≤ 1. According to
Farrell(1957), an index value of 1 refers to a point on the frontier and thus to technically efficient container terminals.
The CRS assumption is only appropriate when all DMUs are operating at an optimal scale. Otherwise, the CRS specification will bias the estimation of the technical efficiency by confounding scale effects.
In the VRS Technical Efficiency (DEA-VRS) model the substitution of CRS with VRS assumption brings about the estimation of the pure technical efficiency (PTE), i.e. TE devoid of scale effects.
This can be achieved by adding a convexity constraint (NI*wi=1) as demonstrated below:
Min TE, W TEis.t. Y∗wi≥yi,Xj∗wi≤TEi∗xi, j=1,2,3,4,5,6,7NI*wi=1wi≥0,
where, NI is an I*N vector of ones. The VRS frontier, obtained this way, envelops the data more tightly than the CRS frontier and, thus, generates efficiency scores which are greater than or equal to those obtained from the CRS frontier.
If there is a difference between CRS technical efficiency (CRSTE) and VRS technical efficiency (VRSTE) for a specific container terminal, then that terminal has scale efficiency. Scale efficiency for a container terminal can be computed from the difference between CRSTE and VRSTE. Since, CRSTE = VRSTE*SE, then SE = CRSTE / VRSTE (
Coelli et al., 2005).
Under DEA, the input and output data of DMUs are compared with each other for one selected time period. Therefore, the number of data sets corresponds to the number of chosen DMUs. In contrast, the Window DEA forms ‘time windows’ over several time periods (
Jahn et al, 2013). This approach allows to regard the DMUs as if it were a different DMUs in each of the periods, examined within one time window, and monitoring the efficiency changes during the selected time period.
The Tiered data envelopment analysis (TDEA) classifies the sampled units into rank-ordered peer groups. DMUs located on the same frontier have comparable levels of efficiency (
Barr et al, 1994). As DMUs on the highest efficiency frontier have DEA scores of 1- this group is “Tier 1”. Dropping the most efficient units from the set for benchmarking, DEA scores are recalculated for each of the remaining units. Those with a score of 1 are on the second frontier - “Tier 2”. The procedure continues until all DMUs are assigned to a frontier.
Recently, there has been a significant growth of interest in DEA among the specialists, who research the efficiency and performance of ports and container terminals. So, DEA became the dominant analysis technique in port research to measure the efficiency (
Woo et al., 2011).
Tongzon(2001),
Valentine and Gray(2001),
Wang et al.(2003),
Song and Sin(2005),
So et al.(2007),
Park et al.(2007),
D’agostini et al.(2015),
Li et al.(2015) proposed DEA method with CCR, and BCC models for efficiency study. The output of the analyzed seaports and container terminals included annual container throughput, while the inputs could include berth length, total terminal area, number of container gantries, quay cranes, floating cranes, mobile cranes; number of straddle carriers, forklifts, reach stackers, top lifter; container freight station area.
On the contrast, in Russian economic literature the DEA method is relatively rare, especially in port industry.
Kharchenko(2013) described only general concepts of DEA method in her research paper “Benchmarking of Russian ports. Case study from ports of Vladivostok and Nakhodka“.
Kuznetsov and Kozlova(2007) showed the possibility of applying DEA for evaluating container terminals’ efficiency, using Korean container terminals’ data for 1999-2002 (Busan, Sebang, Hanjin, Hutchison and Korex) as an example. Thus, this topic is greatly underdeveloped, moreover, no one has ever tried to compare the relative efficiency of Russian container terminals by means of DEA.
Hence, Russian versus Korean container terminals’ efficiency comparison was never conducted before. This paper should be seen as an attempt to expand this topic.
DEA Empirical Analysis
In this paper, we assume seven inputs and one output. Output: Annual container throughput. Inputs: Total terminal area; Total quay length; Quay equipment; Yard equipment; Storage capacity; Depth alongside; Handling capacity. These inputs are the key factors of container terminal operations, and are closely related to container throughput of ports. To confirm the correlation between selected inputs and outputs, this paper applied analysis of Pearson correlation coefficients, which showed that the output variable highly correlates with the inputs. The p-value indicates that the correlation is significant.
We selected 12 container terminals in Russia and 19 container terminals in South Korea, see Table
1. To ensure comparability, only terminals specialized in container handling are included, thus, multipurpose terminals are excluded (
Jahn et al., 2013;
Cullinane and Wang, 2006).
Table 1
Decision making units selected for the analysis
Port |
Container terminal |
DMU |
Saint Petersburg |
First Container Terminal (FCT) |
DMU 1 |
Petrolesport (PLP) |
DMU 2 |
Container Terminal St. Petersburg (CTSP) |
DMU 3 |
Moby Dik |
DMU 4 |
Kaliningrad |
Kaliningrad Sea Commercial Port (KSCP) |
DMU 5 |
Baltic Stevedore Company (BSC) |
DMU 6 |
Novorossiysk |
Novoroslesexport |
DMU 7 |
Novorossiysk Commercial Sea Port (NCSP) |
DMU 8 |
NUTEP Container Terminal (NUTEP) |
DMU 9 |
Vladivostok |
Vladivostok Container Terminal (VCT) |
DMU 10 |
Vladivostok Sea Container Terminal (VSCT) |
DMU 11 |
Vostochny |
Vostoc hnaya Stevedoring Company (VSC) |
DMU 12 |
|
|
|
Busan North Port |
Jaseongdae Container Terminal (HBCT) |
DMU 13 |
Shinseondae Container Terminal (CJKBCT) |
DMU 14 |
Gamman Container Terminal (BIT) |
DMU 15 |
Singamman Container Terminal (DPCT) |
DMU 16 |
Busan New Port |
Phase 1-1 (New pier 1) (PNIT) |
DMU 17 |
Phase 1-2 (New pier 2) (PNC) |
DMU 18 |
Phase 2-1 (New pier 3) (HJNC) |
DMU 19 |
Phase 2-2 (New pier 4) (HPNT) |
DMU 20 |
Phase 2-3 (New pier5) |
DMU 21 |
Gwang Yang Port |
Phase 2-1 (HSGT) |
DMU 22 |
Phase 2-2 (KIT) |
DMU 23 |
Phase 3-1 (CJKE) |
DMU 24 |
Incheon Port |
ICT |
DMU 25 |
SICT |
DMU 26 |
E1CT |
DMU 27 |
CJKE |
DMU 28 |
Pyeongtaek· Dangjin Port |
East Pier |
DMU 29 |
Ulsan Port |
Ulsan New Port |
DMU 30 |
Jungil Container Terminal |
DMU 31 |
All the data were collected from annual reports for 2012-2014, Port-MIS, and ports’ official web sites. Table
2 shows the summary statistics of the data used.
Table 2
for sample of Russian container terminals |
Description |
Unit |
Average |
Min |
Max |
SD |
Annual container throughput |
TEU |
393,922 |
138,500 |
1,100,000 |
256,677 |
Total terminalarea |
m2 |
308,533 |
41,700 |
890,000 |
248,681 |
Total quaylength |
m |
677 |
168 |
1,433 |
359 |
Quay equipment |
unit |
5 |
1 |
9 |
3 |
Yard equipment |
unit |
24 |
11 |
66 |
15 |
Storage capacity |
TEU |
15,118 |
4,100 |
34,705 |
8,885 |
Depth alongside |
m |
10 |
7 |
14 |
2 |
Handling capacity |
TEU |
532,500 |
150,000 |
1,250,000 |
313,106 |
for sample of Korean container terminals
|
Description
|
Unit
|
Average
|
Min
|
Max
|
SD
|
Annual container throughput |
TEU |
1,123,205 |
136,138 |
3,895,202 |
904,208 |
Total terminalarea |
m2 |
531,627 |
7,691 |
1,202,000 |
362,823 |
Total quaylength |
m |
1,059 |
220 |
2,000 |
506 |
Quay equipment |
unit |
8 |
2 |
17 |
4 |
Yard equipment |
unit |
32 |
8 |
74 |
19 |
Storage capacity |
TEU |
37,379 |
2,200 |
112,319 |
26,474 |
Depth alongside |
m |
15 |
8 |
18 |
2 |
Handling capacity |
TEU |
1,131,930 |
100,000 |
2,730,000 |
708,453 |
In the first step the output-oriented Window DEA was performed for Russia DMUs. Terminal operators can influence the production level, but they cannot easily influence and change the production inputs. Therefore, we consider that the output-oriented model represents the maximum output that can be obtained for a given input level.
For the purposes of this study, the data was obtained for 12 container terminals (n = 12) over a three-year period from 2012 to 2014. According to
Cooper et al.(2007) the number of data points can be determined as: w = k - p + 1, where, k = number of periods, p = length of window, w = number of windows.
Number of windows (w) = 3 - 2 + 1 = 2
Number of data points = n * p * w = 12 * 2 * 2 = 48
Thus, there are 48 different data points; the first window was formed by the 2-year period 2012-2013 and comparative analysis of 24 DMUs (n * w = 12 * 2 = 24) was applied; in the second window, in this manner, the analysis is carried out for the next 24 DMUs of the set from 2013-2014.
In the current research, we used DEA-Solver (LV 8.0) software to solve DEA-CRS model (
Cooper et al., 2007). It is worthwhile to say that, to avoid potential imbalance in data magnitudes,
Cooper et al.(2007) have recommended that within an input or output item, the ratio min/max of data should be greater than 10
-4 on average. The data set, under evaluation, for each inputs and output meets this requirement. The ratio min/max of data measures up from 0.006 till 0.389, that is more than 0.0001.
The results of Window analysis of Russian DMUs are arranged in Table
3. The average of the DEA efficiency scores per window is presented in the column denoted “mean.” The column labeled GD denotes the greatest difference in a DEA scores for the entire period, negative quantity means decline in efficiency. The DMU with efficiency score equal to 1 is considered to be efficient amongst the DMUs included in the analysis. The DMU with efficiency score less than 1.000 is deemed to be relatively inefficient. Set of efficient DMUs used as reference set (benchmarks) for each inefficient DMU.
Table 3
Window DEA-CRS model results, Russia.
DMUs |
2012 |
2013 |
2014 |
Mean per window |
Total mean (by DMU) |
GD |
DMU 1 |
1 |
0.985 |
|
0.993 |
0.963 |
-0.132 |
|
1 |
0.868 |
0.934 |
DMU 2 |
1 |
0.860 |
|
0.930 |
0.904 |
-0.155 |
|
0.913 |
0.845 |
0.879 |
DMU 3 |
0.509 |
0.618 |
|
0.563 |
0.578 |
0.077 |
|
0.599 |
0.585 |
0.592 |
DMU 4 |
0.607 |
0.588 |
|
0.598 |
0.592 |
-0.009 |
|
0.574 |
0.598 |
0.586 |
DMU 5 |
0.740 |
0.617 |
|
0.679 |
0.636 |
-0.166 |
|
0.614 |
0.574 |
0.594 |
DMU 6 |
0.510 |
0.614 |
|
0.562 |
0.593 |
0.136 |
|
0.600 |
0.646 |
0.623 |
DMU 7 |
0.881 |
1 |
|
0.940 |
0.960 |
0.080 |
|
1 |
0.961 |
0.980 |
DMU 8 |
0.736 |
1 |
|
0.868 |
0.925 |
0.264 |
|
0.964 |
1 |
0.982 |
DMU 9 |
0.624 |
0.805 |
|
0.714 |
0.747 |
0.137 |
|
0.796 |
0.761 |
0.779 |
DMU 10 |
0.788 |
0.824 |
|
0.806 |
0.824 |
0.085 |
|
0.811 |
0.874 |
0.842 |
DMU 11 |
0.965 |
1 |
|
0.983 |
0.986 |
0.019 |
|
0.995 |
0.984 |
0.990 |
DMU 12 |
0.702 |
0.840 |
|
0.771 |
0.805 |
0.137 |
|
0.839 |
0.839 |
0.839 |
The first row (with values of 1, 0.985) shows the relative technical efficiency of the DMU 1 in 2012, 2013, respectively. The second row (with values of 1, 0.868) shows the relative technical efficiency of DMU 1 in 2013, 2014, respectively, and so on. The scores in different years within the same windows show how the efficiency changes from year to another.
The result shows that none of the DMUs from Russian sample, were efficient during the entire period. The DMU 8 has the positive and highest quantity of GD (0.264), that means improving efficiency. Thus, this container terminal reached 1 by the end of the period and became efficient. DMU 7, DMU 11 have positive GD (0.080, 0.023, respectively) and were efficient in 2013 (efficiency score equals to 1), in 2014 they had a relatively high efficiency score (0.961, 0.988, respectively).
The next DMUs show minimum efficiency score by years: DMU 3 (0.509), DMU 6 (0.510) in 2012, DMU 4 - 0.574 in 2013 and DMU 5 - 0.574 in 2014.
As can be seen in the rows (by windows), 67% (GD) of all Russian DMUs show increasing efficiency trend, but during the whole period of study DMU 6, DMU 8, DMU 10, DMU 12 have showed increasing efficiency.
On the contrary, 33% of DMUs show decreasing efficiency trend: DMU 5 (highest negative quantity of GD), DMU 2, DMU 1, DMU 4. Among them, DMU 1 and DMU 2 have reduced their efficiency significantly - from efficiency (score equals 1) in 2012 to 0.868, 0.845, respectively, in 2014.
A DMU can have different efficiency scores for the same year in different windows. A unit that is efficient in one year, regardless of the windows, is said to be stable in its efficiency rating relative to other units (
Cooper et al., 2007). DMU 5, DMU 7, DMU 11, DMU 12 were the most stable in their efficiency. Besides, all studied DMUs have small difference between scores for the same years.
According to the analysis of the average efficiency, none of the studied DMUs have an efficiency score of 1. Five container terminals (DMU 1, DMU 2, DMU 7, DMU 8, and DMU 11) are close to the efficiency frontier, ranking average scores of above 90%. Four terminals (DMU 5, DMU 9, DMU 10, and DMU 12) showed highest scores from 60 to 89%. And three terminals (DMU 3, DMU 4, and DMU 6) showed average scores above 50%. This may indicate a shortage of about half of their respective potential throughputs.
None of the studied DMUs were highly inefficient (with average scores below 50%).
In the second step Window DEA was performed for Korean DMUs. In the above mentioned manner, the data was obtained for 19 container terminals; there are 76 different data points; two windows are formed by the 2-year periods (2012-2013 and 2013-2014); in each window the comparative analysis is applied for 38 DMUs. The results of Window analysis of Korean DMUs are arranged in Table
4.
Table 4
Window DEA-CRS model results, S.Korea.
DMUs |
2012 |
2013 |
2014 |
Mean per window |
Total mean (by DMU) |
GD |
DMU 13 |
0.638 |
0.678 |
|
0.658 |
0.660 |
0.049 |
|
0.636 |
0.688 |
0.662 |
DMU 14 |
0.954 |
0.702 |
|
0.828 |
0.772 |
-0.158 |
|
0.634 |
0.796 |
0.715 |
DMU 15 |
0.967 |
0.870 |
|
0.918 |
0.824 |
-0.330 |
|
0.821 |
0.637 |
0.729 |
DMU 16 |
1 |
0.904 |
|
0.952 |
0.944 |
0 |
|
0.871 |
1 |
0.936 |
DMU 17 |
0.660 |
0.945 |
|
0.803 |
0.839 |
0.207 |
|
0.885 |
0.867 |
0.876 |
DMU 18 |
0.994 |
1 |
|
0.997 |
0.960 |
0.006 |
|
0.847 |
1 |
0.924 |
DMU 19 |
1 |
0.973 |
|
0.986 |
0.984 |
0 |
|
0.963 |
1 |
0.981 |
DMU 20 |
0.831 |
1 |
|
0.916 |
0.942 |
0.169 |
|
0.937 |
1 |
0.969 |
DMU 21 |
0.272 |
0.649 |
|
0.460 |
0.564 |
0.452 |
|
0.610 |
0.724 |
0.667 |
DMU 22 |
0.552 |
0.531 |
|
0.542 |
0.515 |
-0.069 |
|
0.495 |
0.483 |
0.489 |
DMU 23 |
0.543 |
0.598 |
|
0.571 |
0.569 |
0.031 |
|
0.560 |
0.574 |
0.567 |
DMU 24 |
0.547 |
0.615 |
|
0.581 |
0.585 |
0.058 |
|
0.574 |
0.605 |
0.589 |
DMU 25 |
0.932 |
0.918 |
|
0.925 |
0.873 |
-0.086 |
|
0.796 |
0.845 |
0.821 |
DMU 26 |
0.917 |
1 |
|
0.959 |
0.942 |
0.083 |
|
0.850 |
1 |
0.925 |
DMU 27 |
0.627 |
1 |
|
0.814 |
0.833 |
0.373 |
|
0.704 |
1 |
0.852 |
DMU 28 |
0.980 |
0.962 |
|
0.971 |
0.801 |
-0.488 |
|
0.770 |
0.493 |
0.631 |
DMU 29 |
0.594 |
0.596 |
|
0.595 |
0.585 |
-0.006 |
|
0.559 |
0.589 |
0.574 |
DMU 30 |
0.300 |
0 |
|
0.302 |
0.293 |
-0.013 |
|
0.282 |
0.287 |
0.284 |
DMU 31 |
0.935 |
1 |
|
0.968 |
0.939 |
-0.020 |
|
0.903 |
0.915 |
0.909 |
Comparative analysis of Korean container terminals shows that none of the DMUs were efficient during the entire period. According to the analysis of the average efficiency, seven Korean container terminals: DMU 16, DMU 18, DMU 19, DMU 20, DMU 26, DMU 31 are close to the efficiency frontier, ranking average scores of above 90%. Among the terminals that showed high average scores from 60 to 89%, there are DMU 13, DMU 14, DMU 15, DMU 17, DMU 25, DMU 27, DMU 28. The DMUs 21∼24, and DMU 29 showed only half of their respective potential throughputs with scores ranging above 50%. DMU 30 is highly inefficient with average scores 0.293. As can be seen in the rows (by windows), 58 % of all Korean DMUs show increasing efficiency trend. Thus, the DMU 16, DMU 18, DMU 19, DMU 20, DMU 26 and DMU 27 have reached 1 by the end of the period. Thus, the aforementioned container terminals have become efficient. Although, DMU 27 was highly inefficient in 2012 (0.627), but growth of the efficiency was notable (the GD equals 0.373).
DMU 21 and DMU 17 showed highest positive quantity of GD (0.452 and 0.207, respectively) and increased its efficiency significantly - DMU 21 from 0.272 in 2012 to 0.724 in 2014, DMU 17 from 0.660 in 2012 to 0.867 in 2014. On the contrary, 42% of all Korean DMUs show decreasing efficiency trend. DMU 15, DMU 22, and DMU 28 have shown a decreasing efficiency trend during the entire period. Among them, DMU 28 reduced its efficiency significantly - from 0.980 in 2012 and 0.866 in 2013 (mean) to 0.493 in 2014 (the GD equals -0.488). DMU 30 showed continuous low efficiency scores, in 2013 and 2014 years it showed the lowest score among all Korean DMUs (0.282 and 0.287, respectively); the GD value of this DMU is also negative that means a decline in efficiency.
None of the Korean DMUs were stable in their efficiency - no one was efficient in one year, regardless of the windows. DMU 19 has the smallest difference in a DEA scores - 0.0099, the DMU 27 has the largest difference in a DEA scores - 0.296. This variation in DEA scores of each unit reflects both the performance of that unit over time as well as that of the other unit.
In the third step Window DEA was performed for both Russian and Korean DMUs. It should be mentioned that all the data that are used to evaluate the model, are assumed to be homogeneous. That means that DMUs: (1) perform the same tasks, with similar objectives, (2) perform under the same set of market conditions (3) are identical, except for differences in intensity or magnitude (Sarkis, 2000). Therefore, we decided to compare Far eastern Russian and small Korean container terminals.
In the above mentioned manner, the data set was obtained for 13 container terminals. There are 52 different data points, two windows have been formed by the 2-year period (2012-2013 and 2013-2014). In each window the comparative analysis was applied for 26 DMUs.
The results of Window analysis of Russian and Korean DMUs are arranged in Table
5. Among Korean DMUs, six container terminals showed the highest efficiency scores - average scores of above 90% - DMU 23, DMU 24, DMU 25, DMU 26, DMU 29, DMU 31. None of the Russian DMUs showed the highest efficiency scores. Other Korean terminals - DMU 22, DMU 27, DMU 28, and DMU 29 - showed an average score above 70%. Among Russian DMUs, two container terminals - DMU 10 and DMU 12 - showed such score. Russian DMU 11 and Korean DMU 30 show only half of its respective potential - average scores above 50%.
Table 5
Window DEA-CRS model results, Russian and Korean container terminals.
DMUs |
2012 |
2013 |
2014 |
Mean per window |
Total mean (by DMU) |
GD |
DMU 10 |
0.768 |
0.803 |
|
0.79 |
0.77 |
0.004 |
|
0.717 |
0.772 |
0.74 |
DMU 11 |
0.574 |
0.595 |
|
0.58 |
0.538 |
-0.086 |
|
0.493 |
0.488 |
0.49 |
DMU 12 |
0.650 |
0.777 |
|
0.71 |
0.73 |
0.093 |
|
0.742 |
0.742 |
0.74 |
|
|
|
|
|
|
|
DMU 22 |
0.845 |
0.814 |
|
0.83 |
0.788 |
-0.11 |
|
0.756 |
0.737 |
0.75 |
DMU 23 |
0.909 |
1 |
|
0.95 |
0.97 |
0.091 |
|
0.975 |
1 |
0.99 |
DMU 24 |
0.890 |
1 |
|
0.94 |
0.96 |
0.110 |
|
0.949 |
1 |
0.97 |
DMU 25 |
1 |
0.985 |
|
0.99 |
0.98 |
0.000 |
|
0.942 |
1 |
0.97 |
DMU 26 |
0.917 |
1 |
|
0.96 |
0.94 |
0.083 |
|
0.850 |
1 |
0.93 |
DMU 27 |
0.627 |
1 |
|
0.81 |
0.83 |
0.373 |
|
0.704 |
1 |
0.85 |
DMU 28 |
0.980 |
0.962 |
|
0.97 |
0.82 |
-0.432 |
|
0.770 |
0.549 |
0.66 |
DMU 29 |
0.997 |
1 |
|
1.00 |
0.981 |
-0.01 |
|
0.939 |
0.989 |
0.96 |
DMU 30 |
0.487 |
0.493 |
|
0.49 |
0.467 |
-0.04 |
|
0.440 |
0.448 |
0.44 |
DMU 31 |
0.935 |
1 |
|
0.97 |
0.939 |
-0.02 |
|
0.903 |
0.915 |
0.91 |
As can be seen in the rows (by windows), most of the Korean container terminals have reached 1 by the end of the period and have become efficient, unlike Russian terminals.
Among Korean DMUs, there are two terminals that displayed relatively erratic behavior - one showed great improvement in efficiency (DMU 27), the other - great decline in efficiency (DMU 28) from 2012 to 2014.
Among Russian DMUs, the container terminals display more stable behavior over the entire period. Most of Korean DMUs show increasing efficiency trend, as well as Russian DMUs.
The efficiency measures of DEA-VRS model (Table
6) are higher than those of CRS, which can be evident from the definition of VRS. DEA model with CRS assumption provides information on pure technical efficiency and scale efficiency (SE) taken together, while DEA model with VRS assumption identifies technical efficiency alone. Two Russian terminals scored gently better in the BCC analysis. For DMU 10 score of 0.765 in the CCR analysis (avg. for all years) became as high as 0.827 (avg. for all years) in the BCC analysis.
Table 6
Window DEA-VRS model results and SE, Russian and Korean container terminals.
DMUs |
DEA-VRS model |
Scale efficiency |
2012 |
2013 |
2014 |
2012 |
2013 |
2014 |
DMU 10 |
0.826 |
0.863 |
|
0.930 |
0.930 |
|
|
0.779 |
0.839 |
|
0.921 |
0.921 |
DMU 11 |
0.584 |
0.605 |
|
0.983 |
0.984 |
|
|
0.554 |
0.547 |
|
0.890 |
0.891 |
DMU 12 |
0.667 |
0.798 |
|
0.975 |
0.975 |
|
|
0.744 |
0.744 |
|
0.997 |
0.997 |
|
|
|
|
|
|
|
DMU 22 |
0.878 |
0.850 |
|
0.962 |
0.958 |
|
|
0.800 |
0.774 |
|
0.945 |
0.952 |
DMU 23 |
0.909 |
1 |
|
1 |
1 |
|
|
0.975 |
1 |
|
1 |
1 |
DMU 24 |
0.890 |
1 |
|
1 |
1 |
|
|
0.949 |
1 |
|
1 |
1 |
DMU 25 |
1 |
0.985 |
|
1 |
1 |
|
|
0.942 |
1 |
|
1 |
1 |
DMU 26 |
0.917 |
1 |
|
1 |
1 |
|
|
0.850 |
1 |
|
1 |
1 |
DMU 27 |
0.627 |
1 |
|
1 |
1 |
|
|
0.704 |
1 |
|
1 |
1 |
DMU 28 |
1 |
0.981 |
|
0.980 |
0.980 |
|
|
1 |
0.846 |
|
0.770 |
0.649 |
DMU 29 |
0.997 |
1 |
|
1 |
1 |
|
|
0.939 |
0.989 |
|
1 |
1 |
DMU 30 |
0.494 |
0.501 |
|
0.986 |
0.986 |
|
|
0.446 |
0.454 |
|
0.986 |
0.986 |
DMU31 |
0.935 |
1 |
|
1 |
1 |
|
|
0.987 |
1 |
|
0.915 |
0.915 |
As well, differences in DEA scores were found with the Korean container terminals between the CCR and BCC model applications. For DMU 28 and DMU 30 (in the second window), which became efficient under VRS assumption, but have been found to be inefficient under CRS, we can infer that the CRS inefficiency in container terminals is not caused by poor input utilization (managerial inefficiency), but is rather caused by inappropriate scale size.
The scale efficiency (SE) (Table
6) indicates how close the production size of a DMU is to the most productive scale. The fact that VRS efficiency scores are higher than SE scores implies that the inefficiency takes place primarily due to scale inefficiency - DMU 28 and DMU 31 in second window.
As for the rest of the DMUs, their VRS inefficiency scores are lower than SE scores; it indicates that inefficiency takes place mainly due to technical factors rather than scale factors.
The Tiered data envelopment analysis (TDEA) allows ranking of DMUs without having the subjectivity of the classification. By adopting this approach, a group of units can be compared with other peer groups, instead of having unit-by-unit comparisons.
Table
7 illustrates how TDEA procedures generate three Tier groups of container terminals efficiency.
Table 7
Container terminals efficiency peer groups according to TDEA
Tier |
DMUs |
Reference set |
Tier 1 Most efficient N= 7 |
DMU 23 |
|
DMU 24 |
DMU 25 |
DMU 26 |
DMU 27 |
DMU 29 |
DMU 31 |
Tier 2 N= 5 |
DMU 10 |
DMU 24, DMU 25, DMU 26 |
DMU 11 |
DMU 26 |
DMU 12 |
DMU 24, DMU 25 |
DMU 22 |
DMU 23, DMU 24, DMU 25, DMU 26 |
DMU 28 |
DMU 25, DMU 26 |
Tier 3 Most inefficient N= 1 |
DMU 30 |
DMU 10, DMU 22, DMU 28 |
The most efficient group (Tier 1) consists of Korean container terminals and this result is in line with previous conclusions. The terminals in the first Tier include: the port of GwangYang - Phase 2-2 (KIT) and Phase 3-1 (CJKE), the port of Incheon - ICT, SICT, E1CT, the port of Pyeongtaek Port - East Pier, the port of Ulsan - Jungil Container Terminal.
The DMUs in Tier 2 have not been evaluated as the best performers in the previous study. All Russian container terminals are categorized in this Tier 2.
If we compare DMUs with their reference sets we can notice the relative excess of total terminal area. The Russian DMUs from Tier 2 have low “Annual container throughput”-to-“Total terminal area” ratio: DMU 10 has 2 TEU/m2, DMU 11 - 1.34 TEU/m2, DMU 12 - 0.85 TEU/m2. As well as Korean DMUs from Tier 2: DMU 22 - 1.10 TEU/m2, DMU 28 - 1.97 TEU/m2. Whereas DMUs from Tier 1 have the significantly higher values: DMU 26 - 6.24 TEU/m2, DMU 27 - 3.30 TEU/m2, DMU 26 - 6.2 TEU/m2, DMU 25 - 2.53 TEU/m2.
Figure
4 illustrates the correlation between throughput and terminal area. The majority of DMUs from Tier 2 are below the trend line, indicating a sub optimal use of available terminal area.
Fig. 4
Container throughput per square meter
The same situation can be observed with the number of quay cranes (QCs) and yard equipment (eq). The DMUs from Tier 1 occupy the five highest positions of “Annual container throughput”-to-“Number of QCs (eq.)” ratio ranking. As a example we may compare DMU 27 (168’819 TEU/QCs and 37’515 TEU/eq) with DMU 28 (68’495 TEU/QCs and 17’123 TEU/eq). All the Russian terminals are low enough: DMU 11 (24’697 TEU/QCs and 12’348 TEU/eq), DMU 12 (59’375 TEU/QCs and 13’970 TEU/eq).
This is likely to mean that potential throughput of DMUs from Tier 2 has not yet been attained and they have resources for throughput increase.
Again, if we compare “Annual container throughput”-to-“Total quay length” ratio, we can see the insufficient quay length utilization. DMU 27, DMU 26, DMU 25 from Tier 1 has 1304, 1162, 1048 TEU/m of quay length, respectively.
In contrast, Korean DMU 22 from Tier 2 have 539 TEU/m of quay length, and DMU 28 - 608 TEU/m of quay length. But, the Russian DMU 11 have only 247 TEU/m of quay length, and DMU 12 have 370 TEU/m of quay length.
Moreover, the major part of deep-water ports (15-18 m) is found among Korean container terminals. The average depth alongside of all Russian ports is 10 m. Typical Panamax-class container vessel, handling 3,000-5,000 TEUs, requires a berth depth of more than 12 meters; under such circumstances, many Russian ports are inaccessible to Panamax container ships. Furthermore, nowadays the Panamax containerships have no longer meet the requirements of modern market. Vessels with 10,000 TEU capacity save 37% of operating costs per container, in comparison with 4,000 TEU vessels (
Ilnitski, 2008). Post-Panamax container ships, handling above 10,000 TEUs, require a berth depth of more 15 meters.
So, there is no escape from the conclusion that depth restrictions can directly linked with aforementioned insufficient quay length utilization. In modern circumstances, container ports must overcome the disadvantages of their small depth alongside in order to stay effective and competitive.
Ulsan New Port container terminal in Tier 3 has one of the smallest annual throughput, but its terminal area, storage capacity, quay cranes and yard equipment ranks relatively high among the studied DMUs. This terminal have a necessary and sufficient condition for efficient operation, but is characterized by proximity to main hubs of this region - Busan container terminals.
Summary and Conclusions
Data Envelopment Analysis does not make accommodation for statistical noise effects such as measurement error, force majeure and other events, which are beyond control of ports. However, DEA provides a suitable method for measurement of container terminal operating efficiency.
According to the research observations, Russian Vladivostok Sea Container Terminal, Novoroslesexport and Novorossiysk Commercial Sea Port showed the best results. An efficient container terminal in modern Russia is relatively small-scaled without excess of equipment and with low annual container throughput. However it cannot satisfy the demands of current state of global container market.
The development of multilateral cooperation between the East Sea countries and transport corridors to Europe/Central Asia would result in load increase. Since Russian container terminals, according to analysis results, show a shortage of about half of their respective potential throughputs, they have got plenty of potential area, equipment and storage capacity. Given such a situation, it is important to improve existing terminals by investing in quay wall, depth alongside, modern communication technologies and information systems.
The following Korean terminals demonstrated the best performance: Singamman Container Terminal in Busan North Port, New pier 2, 3, 4 in Busan New Port and SICT in Incheon Port. These results prove that Busan Port is highly ranked among the largest container ports in the world.
The comparison of Far-eastern Russian and small Korean container terminals showed the supremacy of Korean DMUs, which was expected, granting these terminals were planned and built for container specialization. The results partially prove that the development vector of Korean ports has brought positive results in the last few decades. Korean container terminals have significantly evolved, thanks to their container hub status achieving strategies. These terminals, over the years, invested heavily in expensive and advanced equipment in order to attract new container shipping lines, provide modern and complex services, be able to handle large vessels, etc., which enhanced the efficiency of their operations.
The inefficiency of Russian terminals can partly result from out-of-date production cycle and non-container specialization. Besides, Russian container terminals are motivated to increase the scale of their operations. Since a larger scale of operation invariably means greater network connectivity (mainline and feeder services) and attaining hub status (China-Europe and China-CIS routes). Russian ports need to consider establishing new large specialized container terminals, adopt advanced technology, and use the best mechanisms to reach a new level of production. Thus, the rich experience in the construction of container terminals of South Korea and common interest in the development of transport corridors are the major motivating factors for a big teamwork.
In conducting this research, we had several limitations: this study included only terminals from Russian Federation and Republic of Korea; therefore, the models’ results do not reflect the actual position of studied DMUs in global industry and economic environment. The study focused mainly on measuring the relative efficiency of container terminals. The operating environment of each terminal such as governance, institutional factors and public policy, market characteristics, and physical location (access to the railroad, highway, etc.) were not taken into consideration.