Although the world seaborne trade in 2016 recorded an increase in volume of 2.6%, up from 1.8% in 2015, the maritime transport sector still has been confronting the effects of the economic downturn of 2009. Recently, “weaker world economic growth, dwindling merchandise trade volumes and rising cost pressures continued to weigh on the performance of world seaports” (UNCTAD, review of maritime transport 2017). more ever, the trends in the world container ports such as the deployment of mega-ships, the alliances or consolidation of shipping lines are also imposing great pressures on container ports and force them continue improving their level of performance. As a country with over 3,200 km of coastline and 49 seaports (Vinamarine, 2008), Vietnam has not been outside the influenced area of maritime world. Recent years, Vietnam ports have been facing the fierce competition from both domestic ports and international ports, especially when there have been some new and modern ports being constructed in Vietnam such as VIP Green Port or Lach Huyen Port.

To have an insight about the performance efficiency of Vietnamese container ports, this study employs Data Envelopment Analysis (DEA) to evaluate the efficiency of 21 major container ports in Vietnam together with 7 majors container ports of Korea which is known as a strong maritime country with the hope that the results will bring more accurate and interesting information about container ports in Vietnam especially when comparing with Korean ports. In addition, to see how different in level of efficiency among different regions in Vietnam, Vietnamese container ports are divided into 3 groups corresponding to 3 regions in Vietnam, namely the North, the South and the Central and their average efficiency score are compared to each other.

Since the first academic journal paper of Roll and Hayuth(1993) which applied DEA to analyze the efficiency of 20 ports, there have been a large number of researches interested in port efficiency using various DEA models or combining DEA with other methods. martinez- Budria (1999) applied DEA-BCC model to analyze efficiency of 26 Spanish ports using panel data from 1993 to 1997 and he chose depreciation cost, labor cost, other expenditures as input factors and cargo throughput, rental income from leasing port facilities as outputs. Tongzon(2001) employed DEA-CCR and DEA-Additive to 16 container ports comprising 4 Australia ports and other 12 international ports. This study selected 6 input factors namely number of cranes, number of berths, number of tugs, terminal area, vessel waiting time and number of port employees and 2 output factors selected are container throughput and the number of container moved per working hour. Furthermore, Cullinane et al.(2005) employed DEA to analyze the relationship between efficiency and organizational structures using panel data from 1992 to 1999 of 30 container ports around the world. Cullinane et al.(2006) applied both DEA and SFA in examining efficiency of 57 ports and terminals of top 30 world’s biggest container ports in 2001 and compared the results obtained. Some other previous studies on DEA and port efficiency are summarized at table 1:

Data envelopment analysis (DEA) is a method for evaluating relative efficiency of the performances between homogeneous organizations such as business firms, hospitals, government agencies, etc.; such units are called decision making units (DmUs). The level of efficiency is determined by comparison each DmU to the single reference DmU or a convex combination of the DmUs located on the frontier.

One of the advantages of DEA comparing to traditional ratio which is commonly used measure of efficiency that is the former can take into account many outputs and inputs at the same time while the latter only uses single output and input. Another advantage is a specific functional form relating inputs to outputs doesn’t need to be imposed in the model so that we can apply it to multiple production process. Table 2

DmUs selected include 7 Korean container ports and 21 Vietnamese container ports. In which, Vietnam container ports comprise ports from three different regions; they are from the South, the Central and the North. Although it is better to analyze each container terminal in a port, this study treats the whole container port rather than a terminal as a DmU due to limitations in collecting data.

Banker et al.(1989) suggested the sample size should follow as (4), that is, if number of DmUs is,number of inputs is and number of outputs is ,then:

Input factors chose in this paper as proxy of capital, land and labors are container berth length (m), container yard area (m2) and number of ship-to-shore cranes. Container cargo throughput volume (TEUs) is used as the only output factor. Data collected is in 2016 through port’s websites, Google Earth, related Organizations websites such as VPA (Vietnam Seaports Association), KOSIS (Korean statistical Information Service) and directly contacting by phone. The data is statistically described as follows:

Among several types of DEA models, the constant return to scale base output oriented (CCR-O) and variables return to scale base output oriented (BCC-O) are selected to analysis container port efficiency in this study due to the fact that it is not practical to reduce the berth length and equipment quantity. Although port authorities can transfer equipment between ports, it is also really hard and need the collaboration between port operators because of the variety of port authorities. Furthermore, this paper attempt to set a reference target output for the operators of inefficient ports.

DEA-Solver software providing by Cooper, W.W., Seiford, L.m. and Tone, K. (2006) was employed to analyze the efficiency of 28 DmUs, results as follows:

This paper evaluates the efficiency of 21 major container ports in three regions in Vietnam together with some major container ports in Korea. The results point out that on average, the overall technical efficiency of Korea container ports is similar to Vietnamese container ports. However, after decomposing the inefficiency into component parts, the results indicate that while Korean container ports are operating at nearly optimal scale and their inefficiency is caused substantially by managerial inefficiency, the inefficiency of Vietnam container ports is caused not only by managerial inefficiency but also by the disadvantageous conditions of scale.

Among three different region groups of container ports in Vietnam, container ports in the Central of Vietnam are the most technical inefficient group and being influenced the most by operating scale. Besides that, this study addresses that majority of Vietnamese container ports is operating under increasing return-to scale. This conclusion is consistent with Nguyen et al.(2017) which analyzed 43 Vietnamese ports handling various types of cargo (container, general cargo or bulk cargo) even though this study only focuses on container ports and there are some major Korean container ports selected to evaluate the efficiency together with Vietnamese container ports. However, in this paper, there are some different conclusions with Nguyen et al.(2017) when the container ports are analyzed separately and in more detail. As the results indicated from this study, container ports in Southern Vietnam are slightly more efficient than in the North because of their advantage in operating scale whereas container ports in the North appear to be more efficient than in the South in term of pure technical efficiency, while Nguyen et al.(2017) justified that ports in the North including container ports are more efficient than in the South. From these results, the study implies that in applying DEA to analyze efficiency, different types of ports should be analyzed separately to have more accurate results so that port authorizes can identify which part of their port (container terminals, bulk cargo terminals or specialized cargo terminal) performs inefficiently.

In the meantime this research has some limitations in the selection of DmUs as well as input and output factors. Some DmUs in Vietnam consist of ports and container terminals. It is better to compare the same status of DmUs. This research considered input factors such as container berth length, yard areas, number of crane and output factor of container throughput. In the future research it is necessary to consider various input and output factors.